From 8429abe0c2f8ee2ef6ed3904c393a4182c4589fc Mon Sep 17 00:00:00 2001 From: Renato Westphal Date: Tue, 1 Mar 2016 15:27:36 -0300 Subject: [PATCH] ldpd: copy original sources from OpenBSD (14/09/2016) Signed-off-by: Renato Westphal --- ldpd/accept.c | 135 +++++ ldpd/address.c | 297 ++++++++++ ldpd/adjacency.c | 346 +++++++++++ ldpd/control.c | 308 ++++++++++ ldpd/control.h | 38 ++ ldpd/hello.c | 574 +++++++++++++++++++ ldpd/init.c | 166 ++++++ ldpd/interface.c | 531 +++++++++++++++++ ldpd/keepalive.c | 58 ++ ldpd/l2vpn.c | 510 +++++++++++++++++ ldpd/labelmapping.c | 764 +++++++++++++++++++++++++ ldpd/lde.c | 1335 +++++++++++++++++++++++++++++++++++++++++++ ldpd/lde.h | 202 +++++++ ldpd/lde_lib.c | 784 +++++++++++++++++++++++++ ldpd/ldp.h | 304 ++++++++++ ldpd/ldpd.c | 1227 +++++++++++++++++++++++++++++++++++++++ ldpd/ldpd.h | 606 ++++++++++++++++++++ ldpd/ldpe.c | 808 ++++++++++++++++++++++++++ ldpd/ldpe.h | 282 +++++++++ ldpd/log.c | 600 +++++++++++++++++++ ldpd/log.h | 63 ++ ldpd/neighbor.c | 827 +++++++++++++++++++++++++++ ldpd/notification.c | 239 ++++++++ ldpd/packet.c | 788 +++++++++++++++++++++++++ ldpd/pfkey.c | 466 +++++++++++++++ ldpd/socket.c | 391 +++++++++++++ ldpd/util.c | 356 ++++++++++++ lib/imsg-buffer.c | 309 ++++++++++ lib/imsg.c | 302 ++++++++++ lib/imsg.h | 112 ++++ lib/openbsd-queue.h | 533 +++++++++++++++++ lib/openbsd-tree.h | 748 ++++++++++++++++++++++++ 32 files changed, 15009 insertions(+) create mode 100644 ldpd/accept.c create mode 100644 ldpd/address.c create mode 100644 ldpd/adjacency.c create mode 100644 ldpd/control.c create mode 100644 ldpd/control.h create mode 100644 ldpd/hello.c create mode 100644 ldpd/init.c create mode 100644 ldpd/interface.c create mode 100644 ldpd/keepalive.c create mode 100644 ldpd/l2vpn.c create mode 100644 ldpd/labelmapping.c create mode 100644 ldpd/lde.c create mode 100644 ldpd/lde.h create mode 100644 ldpd/lde_lib.c create mode 100644 ldpd/ldp.h create mode 100644 ldpd/ldpd.c create mode 100644 ldpd/ldpd.h create mode 100644 ldpd/ldpe.c create mode 100644 ldpd/ldpe.h create mode 100644 ldpd/log.c create mode 100644 ldpd/log.h create mode 100644 ldpd/neighbor.c create mode 100644 ldpd/notification.c create mode 100644 ldpd/packet.c create mode 100644 ldpd/pfkey.c create mode 100644 ldpd/socket.c create mode 100644 ldpd/util.c create mode 100644 lib/imsg-buffer.c create mode 100644 lib/imsg.c create mode 100644 lib/imsg.h create mode 100644 lib/openbsd-queue.h create mode 100644 lib/openbsd-tree.h diff --git a/ldpd/accept.c b/ldpd/accept.c new file mode 100644 index 000000000..bc13ad49e --- /dev/null +++ b/ldpd/accept.c @@ -0,0 +1,135 @@ +/* $OpenBSD$ */ + +/* + * Copyright (c) 2012 Claudio Jeker + * + * 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 "ldpd.h" +#include "ldpe.h" +#include "log.h" + +struct accept_ev { + LIST_ENTRY(accept_ev) entry; + struct event ev; + void (*accept_cb)(int, short, void *); + void *arg; + int fd; +}; + +struct { + LIST_HEAD(, accept_ev) queue; + struct event evt; +} accept_queue; + +static void accept_arm(void); +static void accept_unarm(void); +static void accept_cb(int, short, void *); +static void accept_timeout(int, short, void *); + +void +accept_init(void) +{ + LIST_INIT(&accept_queue.queue); + evtimer_set(&accept_queue.evt, accept_timeout, NULL); +} + +int +accept_add(int fd, void (*cb)(int, short, void *), void *arg) +{ + struct accept_ev *av; + + if ((av = calloc(1, sizeof(*av))) == NULL) + return (-1); + av->fd = fd; + av->accept_cb = cb; + av->arg = arg; + LIST_INSERT_HEAD(&accept_queue.queue, av, entry); + + event_set(&av->ev, av->fd, EV_READ, accept_cb, av); + event_add(&av->ev, NULL); + + log_debug("%s: accepting on fd %d", __func__, fd); + + return (0); +} + +void +accept_del(int fd) +{ + struct accept_ev *av; + + LIST_FOREACH(av, &accept_queue.queue, entry) + if (av->fd == fd) { + log_debug("%s: %d removed from queue", __func__, fd); + event_del(&av->ev); + LIST_REMOVE(av, entry); + free(av); + return; + } +} + +void +accept_pause(void) +{ + struct timeval evtpause = { 1, 0 }; + + log_debug(__func__); + accept_unarm(); + evtimer_add(&accept_queue.evt, &evtpause); +} + +void +accept_unpause(void) +{ + if (evtimer_pending(&accept_queue.evt, NULL)) { + log_debug(__func__); + evtimer_del(&accept_queue.evt); + accept_arm(); + } +} + +static void +accept_arm(void) +{ + struct accept_ev *av; + LIST_FOREACH(av, &accept_queue.queue, entry) + event_add(&av->ev, NULL); +} + +static void +accept_unarm(void) +{ + struct accept_ev *av; + LIST_FOREACH(av, &accept_queue.queue, entry) + event_del(&av->ev); +} + +static void +accept_cb(int fd, short event, void *arg) +{ + struct accept_ev *av = arg; + event_add(&av->ev, NULL); + av->accept_cb(fd, event, av->arg); +} + +static void +accept_timeout(int fd, short event, void *bula) +{ + log_debug(__func__); + accept_arm(); +} diff --git a/ldpd/address.c b/ldpd/address.c new file mode 100644 index 000000000..5e95fcc27 --- /dev/null +++ b/ldpd/address.c @@ -0,0 +1,297 @@ +/* $OpenBSD$ */ + +/* + * Copyright (c) 2009 Michele Marchetto + * + * 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 "ldpd.h" +#include "ldpe.h" +#include "lde.h" +#include "log.h" + +static void send_address(struct nbr *, int, struct if_addr_head *, + unsigned int, int); +static int gen_address_list_tlv(struct ibuf *, uint16_t, int, + struct if_addr_head *, unsigned int); +static void address_list_add(struct if_addr_head *, struct if_addr *); +static void address_list_clr(struct if_addr_head *); + +static void +send_address(struct nbr *nbr, int af, struct if_addr_head *addr_list, + unsigned int addr_count, int withdraw) +{ + struct ibuf *buf; + uint16_t msg_type; + uint8_t addr_size; + struct if_addr *if_addr; + uint16_t size; + unsigned int tlv_addr_count = 0; + int err = 0; + + /* nothing to send */ + if (LIST_EMPTY(addr_list)) + return; + + if (!withdraw) + msg_type = MSG_TYPE_ADDR; + else + msg_type = MSG_TYPE_ADDRWITHDRAW; + + switch (af) { + case AF_INET: + addr_size = sizeof(struct in_addr); + break; + case AF_INET6: + addr_size = sizeof(struct in6_addr); + break; + default: + fatalx("send_address: unknown af"); + } + + while ((if_addr = LIST_FIRST(addr_list)) != NULL) { + /* + * Send as many addresses as possible - respect the session's + * negotiated maximum pdu length. + */ + size = LDP_HDR_SIZE + LDP_MSG_SIZE + ADDR_LIST_SIZE; + if (size + addr_count * addr_size <= nbr->max_pdu_len) + tlv_addr_count = addr_count; + else + tlv_addr_count = (nbr->max_pdu_len - size) / addr_size; + size += tlv_addr_count * addr_size; + addr_count -= tlv_addr_count; + + if ((buf = ibuf_open(size)) == NULL) + fatal(__func__); + + err |= gen_ldp_hdr(buf, size); + size -= LDP_HDR_SIZE; + err |= gen_msg_hdr(buf, msg_type, size); + size -= LDP_MSG_SIZE; + err |= gen_address_list_tlv(buf, size, af, addr_list, + tlv_addr_count); + if (err) { + address_list_clr(addr_list); + ibuf_free(buf); + return; + } + + while ((if_addr = LIST_FIRST(addr_list)) != NULL) { + log_debug("msg-out: %s: lsr-id %s, address %s", + msg_name(msg_type), inet_ntoa(nbr->id), + log_addr(af, &if_addr->addr)); + + LIST_REMOVE(if_addr, entry); + free(if_addr); + if (--tlv_addr_count == 0) + break; + } + + evbuf_enqueue(&nbr->tcp->wbuf, buf); + } + + nbr_fsm(nbr, NBR_EVT_PDU_SENT); +} + +void +send_address_single(struct nbr *nbr, struct if_addr *if_addr, int withdraw) +{ + struct if_addr_head addr_list; + + LIST_INIT(&addr_list); + address_list_add(&addr_list, if_addr); + send_address(nbr, if_addr->af, &addr_list, 1, withdraw); +} + +void +send_address_all(struct nbr *nbr, int af) +{ + struct if_addr_head addr_list; + struct if_addr *if_addr; + unsigned int addr_count = 0; + + LIST_INIT(&addr_list); + LIST_FOREACH(if_addr, &global.addr_list, entry) { + if (if_addr->af != af) + continue; + + address_list_add(&addr_list, if_addr); + addr_count++; + } + + send_address(nbr, af, &addr_list, addr_count, 0); +} + +int +recv_address(struct nbr *nbr, char *buf, uint16_t len) +{ + struct ldp_msg msg; + uint16_t msg_type; + struct address_list_tlv alt; + enum imsg_type type; + struct lde_addr lde_addr; + + memcpy(&msg, buf, sizeof(msg)); + buf += LDP_MSG_SIZE; + len -= LDP_MSG_SIZE; + + /* Address List TLV */ + if (len < ADDR_LIST_SIZE) { + session_shutdown(nbr, S_BAD_MSG_LEN, msg.id, msg.type); + return (-1); + } + + memcpy(&alt, buf, sizeof(alt)); + if (ntohs(alt.length) != len - TLV_HDR_SIZE) { + session_shutdown(nbr, S_BAD_TLV_LEN, msg.id, msg.type); + return (-1); + } + if (ntohs(alt.type) != TLV_TYPE_ADDRLIST) { + session_shutdown(nbr, S_UNKNOWN_TLV, msg.id, msg.type); + return (-1); + } + switch (ntohs(alt.family)) { + case AF_IPV4: + if (!nbr->v4_enabled) + /* just ignore the message */ + return (0); + break; + case AF_IPV6: + if (!nbr->v6_enabled) + /* just ignore the message */ + return (0); + break; + default: + send_notification_nbr(nbr, S_UNSUP_ADDR, msg.id, msg.type); + return (-1); + } + buf += sizeof(alt); + len -= sizeof(alt); + + msg_type = ntohs(msg.type); + if (msg_type == MSG_TYPE_ADDR) + type = IMSG_ADDRESS_ADD; + else + type = IMSG_ADDRESS_DEL; + + while (len > 0) { + switch (ntohs(alt.family)) { + case AF_IPV4: + if (len < sizeof(struct in_addr)) { + session_shutdown(nbr, S_BAD_TLV_LEN, msg.id, + msg.type); + return (-1); + } + + memset(&lde_addr, 0, sizeof(lde_addr)); + lde_addr.af = AF_INET; + memcpy(&lde_addr.addr, buf, sizeof(struct in_addr)); + + buf += sizeof(struct in_addr); + len -= sizeof(struct in_addr); + break; + case AF_IPV6: + if (len < sizeof(struct in6_addr)) { + session_shutdown(nbr, S_BAD_TLV_LEN, msg.id, + msg.type); + return (-1); + } + + memset(&lde_addr, 0, sizeof(lde_addr)); + lde_addr.af = AF_INET6; + memcpy(&lde_addr.addr, buf, sizeof(struct in6_addr)); + + buf += sizeof(struct in6_addr); + len -= sizeof(struct in6_addr); + break; + default: + fatalx("recv_address: unknown af"); + } + + log_debug("msg-in: %s: lsr-id %s, address %s", + msg_name(msg_type), inet_ntoa(nbr->id), + log_addr(lde_addr.af, &lde_addr.addr)); + + ldpe_imsg_compose_lde(type, nbr->peerid, 0, &lde_addr, + sizeof(lde_addr)); + } + + return (0); +} + +static int +gen_address_list_tlv(struct ibuf *buf, uint16_t size, int af, + struct if_addr_head *addr_list, unsigned int tlv_addr_count) +{ + struct address_list_tlv alt; + uint16_t addr_size; + struct if_addr *if_addr; + int err = 0; + + memset(&alt, 0, sizeof(alt)); + alt.type = TLV_TYPE_ADDRLIST; + alt.length = htons(size - TLV_HDR_SIZE); + + switch (af) { + case AF_INET: + alt.family = htons(AF_IPV4); + addr_size = sizeof(struct in_addr); + break; + case AF_INET6: + alt.family = htons(AF_IPV6); + addr_size = sizeof(struct in6_addr); + break; + default: + fatalx("gen_address_list_tlv: unknown af"); + } + + err |= ibuf_add(buf, &alt, sizeof(alt)); + LIST_FOREACH(if_addr, addr_list, entry) { + err |= ibuf_add(buf, &if_addr->addr, addr_size); + if (--tlv_addr_count == 0) + break; + } + + return (err); +} + +static void +address_list_add(struct if_addr_head *addr_list, struct if_addr *if_addr) +{ + struct if_addr *new; + + new = malloc(sizeof(*new)); + if (new == NULL) + fatal(__func__); + *new = *if_addr; + + LIST_INSERT_HEAD(addr_list, new, entry); +} + +static void +address_list_clr(struct if_addr_head *addr_list) +{ + struct if_addr *if_addr; + + while ((if_addr = LIST_FIRST(addr_list)) != NULL) { + LIST_REMOVE(if_addr, entry); + free(if_addr); + } +} diff --git a/ldpd/adjacency.c b/ldpd/adjacency.c new file mode 100644 index 000000000..266717729 --- /dev/null +++ b/ldpd/adjacency.c @@ -0,0 +1,346 @@ +/* $OpenBSD$ */ + +/* + * Copyright (c) 2013, 2015 Renato Westphal + * Copyright (c) 2009 Michele Marchetto + * Copyright (c) 2005 Claudio Jeker + * Copyright (c) 2004, 2005, 2008 Esben Norby + * + * 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 "ldpd.h" +#include "ldpe.h" +#include "log.h" + +static void adj_del_single(struct adj *); +static void adj_itimer(int, short, void *); +static void tnbr_del(struct tnbr *); +static void tnbr_hello_timer(int, short, void *); +static void tnbr_start_hello_timer(struct tnbr *); +static void tnbr_stop_hello_timer(struct tnbr *); + +struct adj * +adj_new(struct in_addr lsr_id, struct hello_source *source, + union ldpd_addr *addr) +{ + struct adj *adj; + + log_debug("%s: lsr-id %s, %s", __func__, inet_ntoa(lsr_id), + log_hello_src(source)); + + if ((adj = calloc(1, sizeof(*adj))) == NULL) + fatal(__func__); + + adj->lsr_id = lsr_id; + adj->nbr = NULL; + adj->source = *source; + adj->trans_addr = *addr; + + evtimer_set(&adj->inactivity_timer, adj_itimer, adj); + + LIST_INSERT_HEAD(&global.adj_list, adj, global_entry); + + switch (source->type) { + case HELLO_LINK: + LIST_INSERT_HEAD(&source->link.ia->adj_list, adj, ia_entry); + break; + case HELLO_TARGETED: + source->target->adj = adj; + break; + } + + return (adj); +} + +static void +adj_del_single(struct adj *adj) +{ + log_debug("%s: lsr-id %s, %s (%s)", __func__, inet_ntoa(adj->lsr_id), + log_hello_src(&adj->source), af_name(adj_get_af(adj))); + + adj_stop_itimer(adj); + + LIST_REMOVE(adj, global_entry); + if (adj->nbr) + LIST_REMOVE(adj, nbr_entry); + switch (adj->source.type) { + case HELLO_LINK: + LIST_REMOVE(adj, ia_entry); + break; + case HELLO_TARGETED: + adj->source.target->adj = NULL; + break; + } + + free(adj); +} + +void +adj_del(struct adj *adj, uint32_t notif_status) +{ + struct nbr *nbr = adj->nbr; + struct adj *atmp; + + adj_del_single(adj); + + /* + * If the neighbor still exists but none of its remaining + * adjacencies (if any) are from the preferred address-family, + * then delete it. + */ + if (nbr && nbr_adj_count(nbr, nbr->af) == 0) { + LIST_FOREACH_SAFE(adj, &nbr->adj_list, nbr_entry, atmp) + adj_del_single(adj); + session_shutdown(nbr, notif_status, 0, 0); + nbr_del(nbr); + } +} + +struct adj * +adj_find(struct hello_source *source) +{ + struct adj *adj; + + LIST_FOREACH(adj, &global.adj_list, global_entry) { + if (adj->source.type != source->type) + continue; + + switch (source->type) { + case HELLO_LINK: + if (ldp_addrcmp(source->link.ia->af, + &adj->source.link.src_addr, + &source->link.src_addr) == 0) + return (adj); + break; + case HELLO_TARGETED: + if (adj->source.target == source->target) + return (adj); + break; + } + } + + return (NULL); +} + +int +adj_get_af(struct adj *adj) +{ + switch (adj->source.type) { + case HELLO_LINK: + return (adj->source.link.ia->af); + case HELLO_TARGETED: + return (adj->source.target->af); + default: + fatalx("adj_get_af: unknown hello type"); + } +} + +/* adjacency timers */ + +/* ARGSUSED */ +static void +adj_itimer(int fd, short event, void *arg) +{ + struct adj *adj = arg; + + log_debug("%s: lsr-id %s", __func__, inet_ntoa(adj->lsr_id)); + + if (adj->source.type == HELLO_TARGETED) { + if (!(adj->source.target->flags & F_TNBR_CONFIGURED) && + adj->source.target->pw_count == 0) { + /* remove dynamic targeted neighbor */ + tnbr_del(adj->source.target); + return; + } + adj->source.target->adj = NULL; + } + + adj_del(adj, S_HOLDTIME_EXP); +} + +void +adj_start_itimer(struct adj *adj) +{ + struct timeval tv; + + timerclear(&tv); + tv.tv_sec = adj->holdtime; + if (evtimer_add(&adj->inactivity_timer, &tv) == -1) + fatal(__func__); +} + +void +adj_stop_itimer(struct adj *adj) +{ + if (evtimer_pending(&adj->inactivity_timer, NULL) && + evtimer_del(&adj->inactivity_timer) == -1) + fatal(__func__); +} + +/* targeted neighbors */ + +struct tnbr * +tnbr_new(struct ldpd_conf *xconf, int af, union ldpd_addr *addr) +{ + struct tnbr *tnbr; + + if ((tnbr = calloc(1, sizeof(*tnbr))) == NULL) + fatal(__func__); + + tnbr->af = af; + tnbr->addr = *addr; + tnbr->state = TNBR_STA_DOWN; + tnbr->hello_holdtime = (ldp_af_conf_get(xconf, af))->thello_holdtime; + tnbr->hello_interval = (ldp_af_conf_get(xconf, af))->thello_interval; + + return (tnbr); +} + +static void +tnbr_del(struct tnbr *tnbr) +{ + tnbr_stop_hello_timer(tnbr); + if (tnbr->adj) + adj_del(tnbr->adj, S_SHUTDOWN); + LIST_REMOVE(tnbr, entry); + free(tnbr); +} + +struct tnbr * +tnbr_find(struct ldpd_conf *xconf, int af, union ldpd_addr *addr) +{ + struct tnbr *tnbr; + + LIST_FOREACH(tnbr, &xconf->tnbr_list, entry) + if (af == tnbr->af && + ldp_addrcmp(af, addr, &tnbr->addr) == 0) + return (tnbr); + + return (NULL); +} + +struct tnbr * +tnbr_check(struct tnbr *tnbr) +{ + if (!(tnbr->flags & (F_TNBR_CONFIGURED|F_TNBR_DYNAMIC)) && + tnbr->pw_count == 0) { + tnbr_del(tnbr); + return (NULL); + } + + return (tnbr); +} + +void +tnbr_update(struct tnbr *tnbr) +{ + int socket_ok, rtr_id_ok; + + if ((ldp_af_global_get(&global, tnbr->af))->ldp_edisc_socket != -1) + socket_ok = 1; + else + socket_ok = 0; + + if (leconf->rtr_id.s_addr != INADDR_ANY) + rtr_id_ok = 1; + else + rtr_id_ok = 0; + + if (tnbr->state == TNBR_STA_DOWN) { + if (!socket_ok || !rtr_id_ok) + return; + + tnbr->state = TNBR_STA_ACTIVE; + send_hello(HELLO_TARGETED, NULL, tnbr); + + evtimer_set(&tnbr->hello_timer, tnbr_hello_timer, tnbr); + tnbr_start_hello_timer(tnbr); + } else if (tnbr->state == TNBR_STA_ACTIVE) { + if (socket_ok && rtr_id_ok) + return; + + tnbr->state = TNBR_STA_DOWN; + tnbr_stop_hello_timer(tnbr); + } +} + +void +tnbr_update_all(int af) +{ + struct tnbr *tnbr; + + /* update targeted neighbors */ + LIST_FOREACH(tnbr, &leconf->tnbr_list, entry) + if (tnbr->af == af || af == AF_UNSPEC) + tnbr_update(tnbr); +} + +/* target neighbors timers */ + +/* ARGSUSED */ +static void +tnbr_hello_timer(int fd, short event, void *arg) +{ + struct tnbr *tnbr = arg; + + send_hello(HELLO_TARGETED, NULL, tnbr); + tnbr_start_hello_timer(tnbr); +} + +static void +tnbr_start_hello_timer(struct tnbr *tnbr) +{ + struct timeval tv; + + timerclear(&tv); + tv.tv_sec = tnbr->hello_interval; + if (evtimer_add(&tnbr->hello_timer, &tv) == -1) + fatal(__func__); +} + +static void +tnbr_stop_hello_timer(struct tnbr *tnbr) +{ + if (evtimer_pending(&tnbr->hello_timer, NULL) && + evtimer_del(&tnbr->hello_timer) == -1) + fatal(__func__); +} + +struct ctl_adj * +adj_to_ctl(struct adj *adj) +{ + static struct ctl_adj actl; + + actl.af = adj_get_af(adj); + actl.id = adj->lsr_id; + actl.type = adj->source.type; + switch (adj->source.type) { + case HELLO_LINK: + memcpy(actl.ifname, adj->source.link.ia->iface->name, + sizeof(actl.ifname)); + break; + case HELLO_TARGETED: + actl.src_addr = adj->source.target->addr; + break; + } + actl.holdtime = adj->holdtime; + actl.trans_addr = adj->trans_addr; + + return (&actl); +} diff --git a/ldpd/control.c b/ldpd/control.c new file mode 100644 index 000000000..42322eb08 --- /dev/null +++ b/ldpd/control.c @@ -0,0 +1,308 @@ +/* $OpenBSD$ */ + +/* + * 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 "ldpd.h" +#include "ldpe.h" +#include "log.h" +#include "control.h" + +#define CONTROL_BACKLOG 5 + +static void control_accept(int, short, void *); +static struct ctl_conn *control_connbyfd(int); +static struct ctl_conn *control_connbypid(pid_t); +static void control_close(int); +static void control_dispatch_imsg(int, short, void *); + +struct ctl_conns ctl_conns; + +static int control_fd; + +int +control_init(void) +{ + struct sockaddr_un sun; + int fd; + mode_t old_umask; + + if ((fd = socket(AF_UNIX, SOCK_STREAM | SOCK_CLOEXEC | SOCK_NONBLOCK, + 0)) == -1) { + log_warn("%s: socket", __func__); + return (-1); + } + + memset(&sun, 0, sizeof(sun)); + sun.sun_family = AF_UNIX; + strlcpy(sun.sun_path, LDPD_SOCKET, sizeof(sun.sun_path)); + + if (unlink(LDPD_SOCKET) == -1) + if (errno != ENOENT) { + log_warn("%s: unlink %s", __func__, LDPD_SOCKET); + close(fd); + return (-1); + } + + old_umask = umask(S_IXUSR|S_IXGRP|S_IWOTH|S_IROTH|S_IXOTH); + if (bind(fd, (struct sockaddr *)&sun, sizeof(sun)) == -1) { + log_warn("%s: bind: %s", __func__, LDPD_SOCKET); + close(fd); + umask(old_umask); + return (-1); + } + umask(old_umask); + + if (chmod(LDPD_SOCKET, S_IRUSR|S_IWUSR|S_IRGRP|S_IWGRP) == -1) { + log_warn("%s: chmod", __func__); + close(fd); + (void)unlink(LDPD_SOCKET); + return (-1); + } + + control_fd = fd; + + return (0); +} + +int +control_listen(void) +{ + if (listen(control_fd, CONTROL_BACKLOG) == -1) { + log_warn("%s: listen", __func__); + return (-1); + } + + return (accept_add(control_fd, control_accept, NULL)); +} + +void +control_cleanup(void) +{ + accept_del(control_fd); + close(control_fd); + unlink(LDPD_SOCKET); +} + +/* ARGSUSED */ +static void +control_accept(int listenfd, short event, void *bula) +{ + int connfd; + socklen_t len; + struct sockaddr_un sun; + struct ctl_conn *c; + + len = sizeof(sun); + if ((connfd = accept4(listenfd, (struct sockaddr *)&sun, &len, + SOCK_NONBLOCK | SOCK_CLOEXEC)) == -1) { + /* + * Pause accept if we are out of file descriptors, or + * libevent will haunt us here too. + */ + if (errno == ENFILE || errno == EMFILE) + accept_pause(); + else if (errno != EWOULDBLOCK && errno != EINTR && + errno != ECONNABORTED) + log_warn("%s: accept4", __func__); + return; + } + + if ((c = calloc(1, sizeof(struct ctl_conn))) == NULL) { + log_warn(__func__); + close(connfd); + return; + } + + imsg_init(&c->iev.ibuf, connfd); + c->iev.handler = control_dispatch_imsg; + c->iev.events = EV_READ; + event_set(&c->iev.ev, c->iev.ibuf.fd, c->iev.events, + c->iev.handler, &c->iev); + event_add(&c->iev.ev, NULL); + + TAILQ_INSERT_TAIL(&ctl_conns, c, entry); +} + +static struct ctl_conn * +control_connbyfd(int fd) +{ + struct ctl_conn *c; + + for (c = TAILQ_FIRST(&ctl_conns); c != NULL && c->iev.ibuf.fd != fd; + c = TAILQ_NEXT(c, entry)) + ; /* nothing */ + + return (c); +} + +static struct ctl_conn * +control_connbypid(pid_t pid) +{ + struct ctl_conn *c; + + for (c = TAILQ_FIRST(&ctl_conns); c != NULL && c->iev.ibuf.pid != pid; + c = TAILQ_NEXT(c, entry)) + ; /* nothing */ + + return (c); +} + +static void +control_close(int fd) +{ + struct ctl_conn *c; + + if ((c = control_connbyfd(fd)) == NULL) { + log_warnx("%s: fd %d: not found", __func__, fd); + return; + } + + msgbuf_clear(&c->iev.ibuf.w); + TAILQ_REMOVE(&ctl_conns, c, entry); + + event_del(&c->iev.ev); + close(c->iev.ibuf.fd); + accept_unpause(); + free(c); +} + +/* ARGSUSED */ +static void +control_dispatch_imsg(int fd, short event, void *bula) +{ + struct ctl_conn *c; + struct imsg imsg; + ssize_t n; + unsigned int ifidx; + int verbose; + + if ((c = control_connbyfd(fd)) == NULL) { + log_warnx("%s: fd %d: not found", __func__, fd); + return; + } + + if (event & EV_READ) { + if (((n = imsg_read(&c->iev.ibuf)) == -1 && errno != EAGAIN) || + n == 0) { + control_close(fd); + return; + } + } + if (event & EV_WRITE) { + if (msgbuf_write(&c->iev.ibuf.w) <= 0 && errno != EAGAIN) { + control_close(fd); + return; + } + } + + for (;;) { + if ((n = imsg_get(&c->iev.ibuf, &imsg)) == -1) { + control_close(fd); + return; + } + + if (n == 0) + break; + + switch (imsg.hdr.type) { + case IMSG_CTL_FIB_COUPLE: + case IMSG_CTL_FIB_DECOUPLE: + case IMSG_CTL_RELOAD: + c->iev.ibuf.pid = imsg.hdr.pid; + ldpe_imsg_compose_parent(imsg.hdr.type, 0, NULL, 0); + break; + case IMSG_CTL_KROUTE: + case IMSG_CTL_KROUTE_ADDR: + case IMSG_CTL_IFINFO: + c->iev.ibuf.pid = imsg.hdr.pid; + ldpe_imsg_compose_parent(imsg.hdr.type, + imsg.hdr.pid, imsg.data, + imsg.hdr.len - IMSG_HEADER_SIZE); + break; + case IMSG_CTL_SHOW_INTERFACE: + if (imsg.hdr.len == IMSG_HEADER_SIZE + + sizeof(ifidx)) { + memcpy(&ifidx, imsg.data, sizeof(ifidx)); + ldpe_iface_ctl(c, ifidx); + imsg_compose_event(&c->iev, IMSG_CTL_END, 0, + 0, -1, NULL, 0); + } + break; + case IMSG_CTL_SHOW_DISCOVERY: + ldpe_adj_ctl(c); + break; + case IMSG_CTL_SHOW_LIB: + case IMSG_CTL_SHOW_L2VPN_PW: + case IMSG_CTL_SHOW_L2VPN_BINDING: + c->iev.ibuf.pid = imsg.hdr.pid; + ldpe_imsg_compose_lde(imsg.hdr.type, 0, imsg.hdr.pid, + imsg.data, imsg.hdr.len - IMSG_HEADER_SIZE); + break; + case IMSG_CTL_SHOW_NBR: + ldpe_nbr_ctl(c); + break; + case IMSG_CTL_CLEAR_NBR: + if (imsg.hdr.len != IMSG_HEADER_SIZE + + sizeof(struct ctl_nbr)) + break; + + nbr_clear_ctl(imsg.data); + break; + case IMSG_CTL_LOG_VERBOSE: + if (imsg.hdr.len != IMSG_HEADER_SIZE + + sizeof(verbose)) + break; + + /* forward to other processes */ + ldpe_imsg_compose_parent(imsg.hdr.type, imsg.hdr.pid, + imsg.data, imsg.hdr.len - IMSG_HEADER_SIZE); + ldpe_imsg_compose_lde(imsg.hdr.type, 0, imsg.hdr.pid, + imsg.data, imsg.hdr.len - IMSG_HEADER_SIZE); + + memcpy(&verbose, imsg.data, sizeof(verbose)); + log_verbose(verbose); + break; + default: + log_debug("%s: error handling imsg %d", __func__, + imsg.hdr.type); + break; + } + imsg_free(&imsg); + } + + imsg_event_add(&c->iev); +} + +int +control_imsg_relay(struct imsg *imsg) +{ + struct ctl_conn *c; + + if ((c = control_connbypid(imsg->hdr.pid)) == NULL) + return (0); + + return (imsg_compose_event(&c->iev, imsg->hdr.type, 0, imsg->hdr.pid, + -1, imsg->data, imsg->hdr.len - IMSG_HEADER_SIZE)); +} diff --git a/ldpd/control.h b/ldpd/control.h new file mode 100644 index 000000000..fd6e47071 --- /dev/null +++ b/ldpd/control.h @@ -0,0 +1,38 @@ +/* $OpenBSD$ */ + +/* + * 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. + */ + +#ifndef _CONTROL_H_ +#define _CONTROL_H_ + +#include +#include + +struct ctl_conn { + TAILQ_ENTRY(ctl_conn) entry; + struct imsgev iev; +}; +TAILQ_HEAD(ctl_conns, ctl_conn); + +extern struct ctl_conns ctl_conns; + +int control_init(void); +int control_listen(void); +void control_cleanup(void); +int control_imsg_relay(struct imsg *); + +#endif /* _CONTROL_H_ */ diff --git a/ldpd/hello.c b/ldpd/hello.c new file mode 100644 index 000000000..d0daed347 --- /dev/null +++ b/ldpd/hello.c @@ -0,0 +1,574 @@ +/* $OpenBSD$ */ + +/* + * Copyright (c) 2013, 2016 Renato Westphal + * Copyright (c) 2009 Michele Marchetto + * + * 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 "ldpd.h" +#include "ldpe.h" +#include "log.h" + +static int gen_hello_prms_tlv(struct ibuf *buf, uint16_t, uint16_t); +static int gen_opt4_hello_prms_tlv(struct ibuf *, uint16_t, uint32_t); +static int gen_opt16_hello_prms_tlv(struct ibuf *, uint16_t, uint8_t *); +static int gen_ds_hello_prms_tlv(struct ibuf *, uint32_t); +static int tlv_decode_hello_prms(char *, uint16_t, uint16_t *, uint16_t *); +static int tlv_decode_opt_hello_prms(char *, uint16_t, int *, int, + union ldpd_addr *, uint32_t *, uint16_t *); + +int +send_hello(enum hello_type type, struct iface_af *ia, struct tnbr *tnbr) +{ + int af; + union ldpd_addr dst; + uint16_t size, holdtime = 0, flags = 0; + int fd = 0; + struct ibuf *buf; + int err = 0; + + switch (type) { + case HELLO_LINK: + af = ia->af; + holdtime = ia->hello_holdtime; + flags = 0; + fd = (ldp_af_global_get(&global, af))->ldp_disc_socket; + + /* multicast destination address */ + switch (af) { + case AF_INET: + if (!(leconf->ipv4.flags & F_LDPD_AF_NO_GTSM)) + flags |= F_HELLO_GTSM; + dst.v4 = global.mcast_addr_v4; + break; + case AF_INET6: + dst.v6 = global.mcast_addr_v6; + break; + default: + fatalx("send_hello: unknown af"); + } + break; + case HELLO_TARGETED: + af = tnbr->af; + holdtime = tnbr->hello_holdtime; + flags = F_HELLO_TARGETED; + if ((tnbr->flags & F_TNBR_CONFIGURED) || tnbr->pw_count) + flags |= F_HELLO_REQ_TARG; + fd = (ldp_af_global_get(&global, af))->ldp_edisc_socket; + + /* unicast destination address */ + dst = tnbr->addr; + break; + default: + fatalx("send_hello: unknown hello type"); + } + + /* calculate message size */ + size = LDP_HDR_SIZE + LDP_MSG_SIZE + sizeof(struct hello_prms_tlv); + switch (af) { + case AF_INET: + size += sizeof(struct hello_prms_opt4_tlv); + break; + case AF_INET6: + size += sizeof(struct hello_prms_opt16_tlv); + break; + default: + fatalx("send_hello: unknown af"); + } + size += sizeof(struct hello_prms_opt4_tlv); + if (ldp_is_dual_stack(leconf)) + size += sizeof(struct hello_prms_opt4_tlv); + + /* generate message */ + if ((buf = ibuf_open(size)) == NULL) + fatal(__func__); + + err |= gen_ldp_hdr(buf, size); + size -= LDP_HDR_SIZE; + err |= gen_msg_hdr(buf, MSG_TYPE_HELLO, size); + err |= gen_hello_prms_tlv(buf, holdtime, flags); + + /* + * RFC 7552 - Section 6.1: + * "An LSR MUST include only the transport address whose address + * family is the same as that of the IP packet carrying the Hello + * message". + */ + switch (af) { + case AF_INET: + err |= gen_opt4_hello_prms_tlv(buf, TLV_TYPE_IPV4TRANSADDR, + leconf->ipv4.trans_addr.v4.s_addr); + break; + case AF_INET6: + err |= gen_opt16_hello_prms_tlv(buf, TLV_TYPE_IPV6TRANSADDR, + leconf->ipv6.trans_addr.v6.s6_addr); + break; + default: + fatalx("send_hello: unknown af"); + } + + err |= gen_opt4_hello_prms_tlv(buf, TLV_TYPE_CONFIG, + htonl(global.conf_seqnum)); + + /* + * RFC 7552 - Section 6.1.1: + * "A Dual-stack LSR (i.e., an LSR supporting Dual-stack LDP for a peer) + * MUST include the Dual-Stack capability TLV in all of its LDP Hellos". + */ + if (ldp_is_dual_stack(leconf)) + err |= gen_ds_hello_prms_tlv(buf, leconf->trans_pref); + + if (err) { + ibuf_free(buf); + return (-1); + } + + send_packet(fd, af, &dst, ia, buf->buf, buf->wpos); + ibuf_free(buf); + + return (0); +} + +void +recv_hello(struct in_addr lsr_id, struct ldp_msg *msg, int af, + union ldpd_addr *src, struct iface *iface, int multicast, char *buf, + uint16_t len) +{ + struct adj *adj = NULL; + struct nbr *nbr, *nbrt; + uint16_t holdtime, flags; + int tlvs_rcvd; + int ds_tlv; + union ldpd_addr trans_addr; + uint32_t scope_id = 0; + uint32_t conf_seqnum; + uint16_t trans_pref; + int r; + struct hello_source source; + struct iface_af *ia = NULL; + struct tnbr *tnbr = NULL; + + r = tlv_decode_hello_prms(buf, len, &holdtime, &flags); + if (r == -1) { + log_debug("%s: lsr-id %s: failed to decode params", __func__, + inet_ntoa(lsr_id)); + return; + } + /* safety checks */ + if (holdtime != 0 && holdtime < MIN_HOLDTIME) { + log_debug("%s: lsr-id %s: invalid hello holdtime (%u)", + __func__, inet_ntoa(lsr_id), holdtime); + return; + } + if (multicast && (flags & F_HELLO_TARGETED)) { + log_debug("%s: lsr-id %s: multicast targeted hello", __func__, + inet_ntoa(lsr_id)); + return; + } + if (!multicast && !((flags & F_HELLO_TARGETED))) { + log_debug("%s: lsr-id %s: unicast link hello", __func__, + inet_ntoa(lsr_id)); + return; + } + buf += r; + len -= r; + + r = tlv_decode_opt_hello_prms(buf, len, &tlvs_rcvd, af, &trans_addr, + &conf_seqnum, &trans_pref); + if (r == -1) { + log_debug("%s: lsr-id %s: failed to decode optional params", + __func__, inet_ntoa(lsr_id)); + return; + } + if (r != len) { + log_debug("%s: lsr-id %s: unexpected data in message", + __func__, inet_ntoa(lsr_id)); + return; + } + + /* implicit transport address */ + if (!(tlvs_rcvd & F_HELLO_TLV_RCVD_ADDR)) + trans_addr = *src; + if (bad_addr(af, &trans_addr)) { + log_debug("%s: lsr-id %s: invalid transport address %s", + __func__, inet_ntoa(lsr_id), log_addr(af, &trans_addr)); + return; + } + if (af == AF_INET6 && IN6_IS_SCOPE_EMBED(&trans_addr.v6)) { + /* + * RFC 7552 - Section 6.1: + * "An LSR MUST use a global unicast IPv6 address in an IPv6 + * Transport Address optional object of outgoing targeted + * Hellos and check for the same in incoming targeted Hellos + * (i.e., MUST discard the targeted Hello if it failed the + * check)". + */ + if (flags & F_HELLO_TARGETED) { + log_debug("%s: lsr-id %s: invalid targeted hello " + "transport address %s", __func__, inet_ntoa(lsr_id), + log_addr(af, &trans_addr)); + return; + } + scope_id = iface->ifindex; + } + + memset(&source, 0, sizeof(source)); + if (flags & F_HELLO_TARGETED) { + /* + * RFC 7552 - Section 5.2: + * "The link-local IPv6 addresses MUST NOT be used as the + * targeted LDP Hello packet's source or destination addresses". + */ + if (af == AF_INET6 && IN6_IS_SCOPE_EMBED(&src->v6)) { + log_debug("%s: lsr-id %s: targeted hello with " + "link-local source address", __func__, + inet_ntoa(lsr_id)); + return; + } + + tnbr = tnbr_find(leconf, af, src); + + /* remove the dynamic tnbr if the 'R' bit was cleared */ + if (tnbr && (tnbr->flags & F_TNBR_DYNAMIC) && + !((flags & F_HELLO_REQ_TARG))) { + tnbr->flags &= ~F_TNBR_DYNAMIC; + tnbr = tnbr_check(tnbr); + } + + if (!tnbr) { + if (!((flags & F_HELLO_REQ_TARG) && + ((ldp_af_conf_get(leconf, af))->flags & + F_LDPD_AF_THELLO_ACCEPT))) + return; + + tnbr = tnbr_new(leconf, af, src); + tnbr->flags |= F_TNBR_DYNAMIC; + tnbr_update(tnbr); + LIST_INSERT_HEAD(&leconf->tnbr_list, tnbr, entry); + } + + source.type = HELLO_TARGETED; + source.target = tnbr; + } else { + ia = iface_af_get(iface, af); + source.type = HELLO_LINK; + source.link.ia = ia; + source.link.src_addr = *src; + } + + adj = adj_find(&source); + nbr = nbr_find_ldpid(lsr_id.s_addr); + + /* check dual-stack tlv */ + ds_tlv = (tlvs_rcvd & F_HELLO_TLV_RCVD_DS) ? 1 : 0; + if (ds_tlv && trans_pref != leconf->trans_pref) { + /* + * RFC 7552 - Section 6.1.1: + * "If the Dual-Stack capability TLV is present and the remote + * preference does not match the local preference (or does not + * get recognized), then the LSR MUST discard the Hello message + * and log an error. + * If an LDP session was already in place, then the LSR MUST + * send a fatal Notification message with status code of + * 'Transport Connection Mismatch' and reset the session". + */ + log_debug("%s: lsr-id %s: remote transport preference does not " + "match the local preference", __func__, inet_ntoa(lsr_id)); + if (nbr) + session_shutdown(nbr, S_TRANS_MISMTCH, msg->id, + msg->type); + if (adj) + adj_del(adj, S_SHUTDOWN); + return; + } + + /* + * Check for noncompliant dual-stack neighbor according to + * RFC 7552 section 6.1.1. + */ + if (nbr && !ds_tlv) { + switch (af) { + case AF_INET: + if (nbr_adj_count(nbr, AF_INET6) > 0) { + session_shutdown(nbr, S_DS_NONCMPLNCE, + msg->id, msg->type); + return; + } + break; + case AF_INET6: + if (nbr_adj_count(nbr, AF_INET) > 0) { + session_shutdown(nbr, S_DS_NONCMPLNCE, + msg->id, msg->type); + return; + } + break; + default: + fatalx("recv_hello: unknown af"); + } + } + + /* + * Protections against misconfigured networks and buggy implementations. + */ + if (nbr && nbr->af == af && + (ldp_addrcmp(af, &nbr->raddr, &trans_addr) || + nbr->raddr_scope != scope_id)) { + log_warnx("%s: lsr-id %s: hello packet advertising a different " + "transport address", __func__, inet_ntoa(lsr_id)); + if (adj) + adj_del(adj, S_SHUTDOWN); + return; + } + if (nbr == NULL) { + nbrt = nbr_find_addr(af, &trans_addr); + if (nbrt) { + log_debug("%s: transport address %s is already being " + "used by lsr-id %s", __func__, log_addr(af, + &trans_addr), inet_ntoa(nbrt->id)); + if (adj) + adj_del(adj, S_SHUTDOWN); + return; + } + } + + if (adj == NULL) { + adj = adj_new(lsr_id, &source, &trans_addr); + if (nbr) { + adj->nbr = nbr; + LIST_INSERT_HEAD(&nbr->adj_list, adj, nbr_entry); + } + } + + /* + * If the hello adjacency's address-family doesn't match the local + * preference, then an adjacency is still created but we don't attempt + * to start an LDP session. + */ + if (nbr == NULL && (!ds_tlv || + ((trans_pref == DUAL_STACK_LDPOV4 && af == AF_INET) || + (trans_pref == DUAL_STACK_LDPOV6 && af == AF_INET6)))) + nbr = nbr_new(lsr_id, af, ds_tlv, &trans_addr, scope_id); + + /* dynamic LDPv4 GTSM negotiation as per RFC 6720 */ + if (nbr) { + if (flags & F_HELLO_GTSM) + nbr->flags |= F_NBR_GTSM_NEGOTIATED; + else + nbr->flags &= ~F_NBR_GTSM_NEGOTIATED; + } + + /* update neighbor's configuration sequence number */ + if (nbr && (tlvs_rcvd & F_HELLO_TLV_RCVD_CONF)) { + if (conf_seqnum > nbr->conf_seqnum && + nbr_pending_idtimer(nbr)) + nbr_stop_idtimer(nbr); + nbr->conf_seqnum = conf_seqnum; + } + + /* always update the holdtime to properly handle runtime changes */ + switch (source.type) { + case HELLO_LINK: + if (holdtime == 0) + holdtime = LINK_DFLT_HOLDTIME; + + adj->holdtime = min(ia->hello_holdtime, holdtime); + break; + case HELLO_TARGETED: + if (holdtime == 0) + holdtime = TARGETED_DFLT_HOLDTIME; + + adj->holdtime = min(tnbr->hello_holdtime, holdtime); + } + if (adj->holdtime != INFINITE_HOLDTIME) + adj_start_itimer(adj); + else + adj_stop_itimer(adj); + + if (nbr && nbr->state == NBR_STA_PRESENT && !nbr_pending_idtimer(nbr) && + nbr_session_active_role(nbr) && !nbr_pending_connect(nbr)) + nbr_establish_connection(nbr); +} + +static int +gen_hello_prms_tlv(struct ibuf *buf, uint16_t holdtime, uint16_t flags) +{ + struct hello_prms_tlv parms; + + memset(&parms, 0, sizeof(parms)); + parms.type = htons(TLV_TYPE_COMMONHELLO); + parms.length = htons(sizeof(parms.holdtime) + sizeof(parms.flags)); + parms.holdtime = htons(holdtime); + parms.flags = htons(flags); + + return (ibuf_add(buf, &parms, sizeof(parms))); +} + +static int +gen_opt4_hello_prms_tlv(struct ibuf *buf, uint16_t type, uint32_t value) +{ + struct hello_prms_opt4_tlv parms; + + memset(&parms, 0, sizeof(parms)); + parms.type = htons(type); + parms.length = htons(sizeof(parms.value)); + parms.value = value; + + return (ibuf_add(buf, &parms, sizeof(parms))); +} + +static int +gen_opt16_hello_prms_tlv(struct ibuf *buf, uint16_t type, uint8_t *value) +{ + struct hello_prms_opt16_tlv parms; + + memset(&parms, 0, sizeof(parms)); + parms.type = htons(type); + parms.length = htons(sizeof(parms.value)); + memcpy(&parms.value, value, sizeof(parms.value)); + + return (ibuf_add(buf, &parms, sizeof(parms))); +} + +static int +gen_ds_hello_prms_tlv(struct ibuf *buf, uint32_t value) +{ + if (leconf->flags & F_LDPD_DS_CISCO_INTEROP) + value = htonl(value); + else + value = htonl(value << 28); + + return (gen_opt4_hello_prms_tlv(buf, TLV_TYPE_DUALSTACK, value)); +} + +static int +tlv_decode_hello_prms(char *buf, uint16_t len, uint16_t *holdtime, + uint16_t *flags) +{ + struct hello_prms_tlv tlv; + + if (len < sizeof(tlv)) + return (-1); + memcpy(&tlv, buf, sizeof(tlv)); + + if (tlv.type != htons(TLV_TYPE_COMMONHELLO)) + return (-1); + if (ntohs(tlv.length) != sizeof(tlv) - TLV_HDR_SIZE) + return (-1); + + *holdtime = ntohs(tlv.holdtime); + *flags = ntohs(tlv.flags); + + return (sizeof(tlv)); +} + +static int +tlv_decode_opt_hello_prms(char *buf, uint16_t len, int *tlvs_rcvd, int af, + union ldpd_addr *addr, uint32_t *conf_number, uint16_t *trans_pref) +{ + struct tlv tlv; + uint16_t tlv_len; + int total = 0; + + *tlvs_rcvd = 0; + memset(addr, 0, sizeof(*addr)); + *conf_number = 0; + *trans_pref = 0; + + /* + * RFC 7552 - Section 6.1: + * "An LSR SHOULD accept the Hello message that contains both IPv4 and + * IPv6 Transport Address optional objects but MUST use only the + * transport address whose address family is the same as that of the + * IP packet carrying the Hello message. An LSR SHOULD accept only + * the first Transport Address optional object for a given address + * family in the received Hello message and ignore the rest if the + * LSR receives more than one Transport Address optional object for a + * given address family". + */ + while (len >= sizeof(tlv)) { + memcpy(&tlv, buf, TLV_HDR_SIZE); + tlv_len = ntohs(tlv.length); + if (tlv_len + TLV_HDR_SIZE > len) + return (-1); + buf += TLV_HDR_SIZE; + len -= TLV_HDR_SIZE; + total += TLV_HDR_SIZE; + + switch (ntohs(tlv.type)) { + case TLV_TYPE_IPV4TRANSADDR: + if (tlv_len != sizeof(addr->v4)) + return (-1); + if (af != AF_INET) + return (-1); + if (*tlvs_rcvd & F_HELLO_TLV_RCVD_ADDR) + break; + memcpy(&addr->v4, buf, sizeof(addr->v4)); + *tlvs_rcvd |= F_HELLO_TLV_RCVD_ADDR; + break; + case TLV_TYPE_IPV6TRANSADDR: + if (tlv_len != sizeof(addr->v6)) + return (-1); + if (af != AF_INET6) + return (-1); + if (*tlvs_rcvd & F_HELLO_TLV_RCVD_ADDR) + break; + memcpy(&addr->v6, buf, sizeof(addr->v6)); + *tlvs_rcvd |= F_HELLO_TLV_RCVD_ADDR; + break; + case TLV_TYPE_CONFIG: + if (tlv_len != sizeof(uint32_t)) + return (-1); + memcpy(conf_number, buf, sizeof(uint32_t)); + *tlvs_rcvd |= F_HELLO_TLV_RCVD_CONF; + break; + case TLV_TYPE_DUALSTACK: + if (tlv_len != sizeof(uint32_t)) + return (-1); + /* + * RFC 7552 - Section 6.1: + * "A Single-stack LSR does not need to use the + * Dual-Stack capability in Hello messages and SHOULD + * ignore this capability if received". + */ + if (!ldp_is_dual_stack(leconf)) + break; + /* Shame on you, Cisco! */ + if (leconf->flags & F_LDPD_DS_CISCO_INTEROP) { + memcpy(trans_pref, buf + sizeof(uint16_t), + sizeof(uint16_t)); + *trans_pref = ntohs(*trans_pref); + } else { + memcpy(trans_pref, buf , sizeof(uint16_t)); + *trans_pref = ntohs(*trans_pref) >> 12; + } + *tlvs_rcvd |= F_HELLO_TLV_RCVD_DS; + break; + default: + /* if unknown flag set, ignore TLV */ + if (!(ntohs(tlv.type) & UNKNOWN_FLAG)) + return (-1); + break; + } + buf += tlv_len; + len -= tlv_len; + total += tlv_len; + } + + return (total); +} diff --git a/ldpd/init.c b/ldpd/init.c new file mode 100644 index 000000000..eb22bf52a --- /dev/null +++ b/ldpd/init.c @@ -0,0 +1,166 @@ +/* $OpenBSD$ */ + +/* + * Copyright (c) 2009 Michele Marchetto + * + * 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 "ldpd.h" +#include "ldpe.h" +#include "log.h" + +static int gen_init_prms_tlv(struct ibuf *, struct nbr *); + +void +send_init(struct nbr *nbr) +{ + struct ibuf *buf; + uint16_t size; + int err = 0; + + log_debug("%s: lsr-id %s", __func__, inet_ntoa(nbr->id)); + + size = LDP_HDR_SIZE + LDP_MSG_SIZE + SESS_PRMS_SIZE; + if ((buf = ibuf_open(size)) == NULL) + fatal(__func__); + + err |= gen_ldp_hdr(buf, size); + size -= LDP_HDR_SIZE; + err |= gen_msg_hdr(buf, MSG_TYPE_INIT, size); + size -= LDP_MSG_SIZE; + err |= gen_init_prms_tlv(buf, nbr); + if (err) { + ibuf_free(buf); + return; + } + + evbuf_enqueue(&nbr->tcp->wbuf, buf); +} + +int +recv_init(struct nbr *nbr, char *buf, uint16_t len) +{ + struct ldp_msg msg; + struct sess_prms_tlv sess; + uint16_t max_pdu_len; + + log_debug("%s: lsr-id %s", __func__, inet_ntoa(nbr->id)); + + memcpy(&msg, buf, sizeof(msg)); + buf += LDP_MSG_SIZE; + len -= LDP_MSG_SIZE; + + if (len < SESS_PRMS_SIZE) { + session_shutdown(nbr, S_BAD_MSG_LEN, msg.id, msg.type); + return (-1); + } + memcpy(&sess, buf, sizeof(sess)); + if (ntohs(sess.length) != SESS_PRMS_LEN) { + session_shutdown(nbr, S_BAD_TLV_LEN, msg.id, msg.type); + return (-1); + } + if (ntohs(sess.proto_version) != LDP_VERSION) { + session_shutdown(nbr, S_BAD_PROTO_VER, msg.id, msg.type); + return (-1); + } + if (ntohs(sess.keepalive_time) < MIN_KEEPALIVE) { + session_shutdown(nbr, S_KEEPALIVE_BAD, msg.id, msg.type); + return (-1); + } + if (sess.lsr_id != leconf->rtr_id.s_addr || + ntohs(sess.lspace_id) != 0) { + session_shutdown(nbr, S_NO_HELLO, msg.id, msg.type); + return (-1); + } + + buf += SESS_PRMS_SIZE; + len -= SESS_PRMS_SIZE; + + /* Optional Parameters */ + while (len > 0) { + struct tlv tlv; + uint16_t tlv_len; + + if (len < sizeof(tlv)) { + session_shutdown(nbr, S_BAD_TLV_LEN, msg.id, msg.type); + return (-1); + } + + memcpy(&tlv, buf, TLV_HDR_SIZE); + tlv_len = ntohs(tlv.length); + if (tlv_len + TLV_HDR_SIZE > len) { + session_shutdown(nbr, S_BAD_TLV_LEN, msg.id, msg.type); + return (-1); + } + buf += TLV_HDR_SIZE; + len -= TLV_HDR_SIZE; + + switch (ntohs(tlv.type)) { + case TLV_TYPE_ATMSESSIONPAR: + session_shutdown(nbr, S_BAD_TLV_VAL, msg.id, msg.type); + return (-1); + case TLV_TYPE_FRSESSION: + session_shutdown(nbr, S_BAD_TLV_VAL, msg.id, msg.type); + return (-1); + default: + if (!(ntohs(tlv.type) & UNKNOWN_FLAG)) + send_notification_nbr(nbr, S_UNKNOWN_TLV, + msg.id, msg.type); + /* ignore unknown tlv */ + break; + } + buf += tlv_len; + len -= tlv_len; + } + + nbr->keepalive = min(nbr_get_keepalive(nbr->af, nbr->id), + ntohs(sess.keepalive_time)); + + max_pdu_len = ntohs(sess.max_pdu_len); + /* + * RFC 5036 - Section 3.5.3: + * "A value of 255 or less specifies the default maximum length of + * 4096 octets". + */ + if (max_pdu_len <= 255) + max_pdu_len = LDP_MAX_LEN; + nbr->max_pdu_len = min(max_pdu_len, LDP_MAX_LEN); + + nbr_fsm(nbr, NBR_EVT_INIT_RCVD); + + return (0); +} + +static int +gen_init_prms_tlv(struct ibuf *buf, struct nbr *nbr) +{ + struct sess_prms_tlv parms; + + memset(&parms, 0, sizeof(parms)); + parms.type = htons(TLV_TYPE_COMMONSESSION); + parms.length = htons(SESS_PRMS_LEN); + parms.proto_version = htons(LDP_VERSION); + parms.keepalive_time = htons(nbr_get_keepalive(nbr->af, nbr->id)); + parms.reserved = 0; + parms.pvlim = 0; + parms.max_pdu_len = 0; + parms.lsr_id = nbr->id.s_addr; + parms.lspace_id = 0; + + return (ibuf_add(buf, &parms, SESS_PRMS_SIZE)); +} diff --git a/ldpd/interface.c b/ldpd/interface.c new file mode 100644 index 000000000..ff4c8f169 --- /dev/null +++ b/ldpd/interface.c @@ -0,0 +1,531 @@ +/* $OpenBSD$ */ + +/* + * Copyright (c) 2013, 2016 Renato Westphal + * Copyright (c) 2005 Claudio Jeker + * Copyright (c) 2004, 2005, 2008 Esben Norby + * + * 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 "ldpd.h" +#include "ldpe.h" +#include "log.h" + +static struct if_addr *if_addr_new(struct kaddr *); +static struct if_addr *if_addr_lookup(struct if_addr_head *, struct kaddr *); +static int if_start(struct iface *, int); +static int if_reset(struct iface *, int); +static void if_update_af(struct iface_af *, int); +static void if_hello_timer(int, short, void *); +static void if_start_hello_timer(struct iface_af *); +static void if_stop_hello_timer(struct iface_af *); +static int if_join_ipv4_group(struct iface *, struct in_addr *); +static int if_leave_ipv4_group(struct iface *, struct in_addr *); +static int if_join_ipv6_group(struct iface *, struct in6_addr *); +static int if_leave_ipv6_group(struct iface *, struct in6_addr *); + +struct iface * +if_new(struct kif *kif) +{ + struct iface *iface; + + if ((iface = calloc(1, sizeof(*iface))) == NULL) + fatal("if_new: calloc"); + + strlcpy(iface->name, kif->ifname, sizeof(iface->name)); + + /* get type */ + if (kif->flags & IFF_POINTOPOINT) + iface->type = IF_TYPE_POINTOPOINT; + if (kif->flags & IFF_BROADCAST && + kif->flags & IFF_MULTICAST) + iface->type = IF_TYPE_BROADCAST; + + /* get index and flags */ + LIST_INIT(&iface->addr_list); + iface->ifindex = kif->ifindex; + iface->flags = kif->flags; + iface->linkstate = kif->link_state; + iface->if_type = kif->if_type; + + /* ipv4 */ + iface->ipv4.af = AF_INET; + iface->ipv4.iface = iface; + iface->ipv4.enabled = 0; + iface->ipv4.state = IF_STA_DOWN; + LIST_INIT(&iface->ipv4.adj_list); + + /* ipv6 */ + iface->ipv6.af = AF_INET6; + iface->ipv6.iface = iface; + iface->ipv6.enabled = 0; + iface->ipv6.state = IF_STA_DOWN; + LIST_INIT(&iface->ipv6.adj_list); + + return (iface); +} + +struct iface * +if_lookup(struct ldpd_conf *xconf, unsigned short ifindex) +{ + struct iface *iface; + + LIST_FOREACH(iface, &xconf->iface_list, entry) + if (iface->ifindex == ifindex) + return (iface); + + return (NULL); +} + +void +if_exit(struct iface *iface) +{ + struct if_addr *if_addr; + + log_debug("%s: interface %s", __func__, iface->name); + + if (iface->ipv4.state == IF_STA_ACTIVE) + if_reset(iface, AF_INET); + if (iface->ipv6.state == IF_STA_ACTIVE) + if_reset(iface, AF_INET6); + + while ((if_addr = LIST_FIRST(&iface->addr_list)) != NULL) { + LIST_REMOVE(if_addr, entry); + free(if_addr); + } +} + +struct iface_af * +iface_af_get(struct iface *iface, int af) +{ + switch (af) { + case AF_INET: + return (&iface->ipv4); + case AF_INET6: + return (&iface->ipv6); + default: + fatalx("iface_af_get: unknown af"); + } +} + +static struct if_addr * +if_addr_new(struct kaddr *ka) +{ + struct if_addr *if_addr; + + if ((if_addr = calloc(1, sizeof(*if_addr))) == NULL) + fatal(__func__); + + if_addr->af = ka->af; + if_addr->addr = ka->addr; + if_addr->prefixlen = ka->prefixlen; + if_addr->dstbrd = ka->dstbrd; + + return (if_addr); +} + +static struct if_addr * +if_addr_lookup(struct if_addr_head *addr_list, struct kaddr *ka) +{ + struct if_addr *if_addr; + int af = ka->af; + + LIST_FOREACH(if_addr, addr_list, entry) + if (!ldp_addrcmp(af, &if_addr->addr, &ka->addr) && + if_addr->prefixlen == ka->prefixlen && + !ldp_addrcmp(af, &if_addr->dstbrd, &ka->dstbrd)) + return (if_addr); + + return (NULL); +} + +void +if_addr_add(struct kaddr *ka) +{ + struct iface *iface; + struct if_addr *if_addr; + struct nbr *nbr; + + if (if_addr_lookup(&global.addr_list, ka) == NULL) { + if_addr = if_addr_new(ka); + + LIST_INSERT_HEAD(&global.addr_list, if_addr, entry); + RB_FOREACH(nbr, nbr_id_head, &nbrs_by_id) { + if (nbr->state != NBR_STA_OPER) + continue; + if (if_addr->af == AF_INET && !nbr->v4_enabled) + continue; + if (if_addr->af == AF_INET6 && !nbr->v6_enabled) + continue; + + send_address_single(nbr, if_addr, 0); + } + } + + iface = if_lookup(leconf, ka->ifindex); + if (iface) { + if (ka->af == AF_INET6 && IN6_IS_ADDR_LINKLOCAL(&ka->addr.v6)) + iface->linklocal = ka->addr.v6; + + if (if_addr_lookup(&iface->addr_list, ka) == NULL) { + if_addr = if_addr_new(ka); + LIST_INSERT_HEAD(&iface->addr_list, if_addr, entry); + if_update(iface, if_addr->af); + } + } +} + +void +if_addr_del(struct kaddr *ka) +{ + struct iface *iface; + struct if_addr *if_addr; + struct nbr *nbr; + + iface = if_lookup(leconf, ka->ifindex); + if (iface) { + if (ka->af == AF_INET6 && + IN6_ARE_ADDR_EQUAL(&iface->linklocal, &ka->addr.v6)) + memset(&iface->linklocal, 0, sizeof(iface->linklocal)); + + if_addr = if_addr_lookup(&iface->addr_list, ka); + if (if_addr) { + LIST_REMOVE(if_addr, entry); + if_update(iface, if_addr->af); + free(if_addr); + } + } + + if_addr = if_addr_lookup(&global.addr_list, ka); + if (if_addr) { + RB_FOREACH(nbr, nbr_id_head, &nbrs_by_id) { + if (nbr->state != NBR_STA_OPER) + continue; + if (if_addr->af == AF_INET && !nbr->v4_enabled) + continue; + if (if_addr->af == AF_INET6 && !nbr->v6_enabled) + continue; + send_address_single(nbr, if_addr, 1); + } + LIST_REMOVE(if_addr, entry); + free(if_addr); + } +} + +static int +if_start(struct iface *iface, int af) +{ + struct iface_af *ia; + struct timeval now; + + log_debug("%s: %s address-family %s", __func__, iface->name, + af_name(af)); + + ia = iface_af_get(iface, af); + + gettimeofday(&now, NULL); + ia->uptime = now.tv_sec; + + switch (af) { + case AF_INET: + if (if_join_ipv4_group(iface, &global.mcast_addr_v4)) + return (-1); + break; + case AF_INET6: + if (if_join_ipv6_group(iface, &global.mcast_addr_v6)) + return (-1); + break; + default: + fatalx("if_start: unknown af"); + } + + send_hello(HELLO_LINK, ia, NULL); + + evtimer_set(&ia->hello_timer, if_hello_timer, ia); + if_start_hello_timer(ia); + return (0); +} + +static int +if_reset(struct iface *iface, int af) +{ + struct iface_af *ia; + struct adj *adj; + + log_debug("%s: %s address-family %s", __func__, iface->name, + af_name(af)); + + ia = iface_af_get(iface, af); + if_stop_hello_timer(ia); + + while ((adj = LIST_FIRST(&ia->adj_list)) != NULL) + adj_del(adj, S_SHUTDOWN); + + /* try to cleanup */ + switch (af) { + case AF_INET: + if (global.ipv4.ldp_disc_socket != -1) + if_leave_ipv4_group(iface, &global.mcast_addr_v4); + break; + case AF_INET6: + if (global.ipv6.ldp_disc_socket != -1) + if_leave_ipv6_group(iface, &global.mcast_addr_v6); + break; + default: + fatalx("if_start: unknown af"); + } + + return (0); +} + +static void +if_update_af(struct iface_af *ia, int link_ok) +{ + int addr_ok = 0, socket_ok, rtr_id_ok; + struct if_addr *if_addr; + + switch (ia->af) { + case AF_INET: + /* + * NOTE: for LDPv4, each interface should have at least one + * valid IP address otherwise they can not be enabled. + */ + LIST_FOREACH(if_addr, &ia->iface->addr_list, entry) { + if (if_addr->af == AF_INET) { + addr_ok = 1; + break; + } + } + break; + case AF_INET6: + /* for IPv6 the link-local address is enough. */ + if (IN6_IS_ADDR_LINKLOCAL(&ia->iface->linklocal)) + addr_ok = 1; + break; + default: + fatalx("if_update_af: unknown af"); + } + + if ((ldp_af_global_get(&global, ia->af))->ldp_disc_socket != -1) + socket_ok = 1; + else + socket_ok = 0; + + if (leconf->rtr_id.s_addr != INADDR_ANY) + rtr_id_ok = 1; + else + rtr_id_ok = 0; + + if (ia->state == IF_STA_DOWN) { + if (!ia->enabled || !link_ok || !addr_ok || !socket_ok || + !rtr_id_ok) + return; + + ia->state = IF_STA_ACTIVE; + if_start(ia->iface, ia->af); + } else if (ia->state == IF_STA_ACTIVE) { + if (ia->enabled && link_ok && addr_ok && socket_ok && rtr_id_ok) + return; + + ia->state = IF_STA_DOWN; + if_reset(ia->iface, ia->af); + } +} + +void +if_update(struct iface *iface, int af) +{ + int link_ok; + + link_ok = (iface->flags & IFF_UP) && + LINK_STATE_IS_UP(iface->linkstate); + + if (af == AF_INET || af == AF_UNSPEC) + if_update_af(&iface->ipv4, link_ok); + if (af == AF_INET6 || af == AF_UNSPEC) + if_update_af(&iface->ipv6, link_ok); +} + +void +if_update_all(int af) +{ + struct iface *iface; + + LIST_FOREACH(iface, &leconf->iface_list, entry) + if_update(iface, af); +} + +/* timers */ +/* ARGSUSED */ +static void +if_hello_timer(int fd, short event, void *arg) +{ + struct iface_af *ia = arg; + + send_hello(HELLO_LINK, ia, NULL); + if_start_hello_timer(ia); +} + +static void +if_start_hello_timer(struct iface_af *ia) +{ + struct timeval tv; + + timerclear(&tv); + tv.tv_sec = ia->hello_interval; + if (evtimer_add(&ia->hello_timer, &tv) == -1) + fatal(__func__); +} + +static void +if_stop_hello_timer(struct iface_af *ia) +{ + if (evtimer_pending(&ia->hello_timer, NULL) && + evtimer_del(&ia->hello_timer) == -1) + fatal(__func__); +} + +struct ctl_iface * +if_to_ctl(struct iface_af *ia) +{ + static struct ctl_iface ictl; + struct timeval now; + struct adj *adj; + + ictl.af = ia->af; + memcpy(ictl.name, ia->iface->name, sizeof(ictl.name)); + ictl.ifindex = ia->iface->ifindex; + ictl.state = ia->state; + ictl.flags = ia->iface->flags; + ictl.linkstate = ia->iface->linkstate; + ictl.type = ia->iface->type; + ictl.if_type = ia->iface->if_type; + ictl.hello_holdtime = ia->hello_holdtime; + ictl.hello_interval = ia->hello_interval; + + gettimeofday(&now, NULL); + if (ia->state != IF_STA_DOWN && + ia->uptime != 0) { + ictl.uptime = now.tv_sec - ia->uptime; + } else + ictl.uptime = 0; + + ictl.adj_cnt = 0; + LIST_FOREACH(adj, &ia->adj_list, ia_entry) + ictl.adj_cnt++; + + return (&ictl); +} + +/* multicast membership sockopts */ +in_addr_t +if_get_ipv4_addr(struct iface *iface) +{ + struct if_addr *if_addr; + + LIST_FOREACH(if_addr, &iface->addr_list, entry) + if (if_addr->af == AF_INET) + return (if_addr->addr.v4.s_addr); + + return (INADDR_ANY); +} + +static int +if_join_ipv4_group(struct iface *iface, struct in_addr *addr) +{ + struct ip_mreq mreq; + + log_debug("%s: interface %s addr %s", __func__, iface->name, + inet_ntoa(*addr)); + + mreq.imr_multiaddr = *addr; + mreq.imr_interface.s_addr = if_get_ipv4_addr(iface); + + if (setsockopt(global.ipv4.ldp_disc_socket, IPPROTO_IP, + IP_ADD_MEMBERSHIP, (void *)&mreq, sizeof(mreq)) < 0) { + log_warn("%s: error IP_ADD_MEMBERSHIP, interface %s address %s", + __func__, iface->name, inet_ntoa(*addr)); + return (-1); + } + return (0); +} + +static int +if_leave_ipv4_group(struct iface *iface, struct in_addr *addr) +{ + struct ip_mreq mreq; + + log_debug("%s: interface %s addr %s", __func__, iface->name, + inet_ntoa(*addr)); + + mreq.imr_multiaddr = *addr; + mreq.imr_interface.s_addr = if_get_ipv4_addr(iface); + + if (setsockopt(global.ipv4.ldp_disc_socket, IPPROTO_IP, + IP_DROP_MEMBERSHIP, (void *)&mreq, sizeof(mreq)) < 0) { + log_warn("%s: error IP_DROP_MEMBERSHIP, interface %s " + "address %s", __func__, iface->name, inet_ntoa(*addr)); + return (-1); + } + + return (0); +} + +static int +if_join_ipv6_group(struct iface *iface, struct in6_addr *addr) +{ + struct ipv6_mreq mreq; + + log_debug("%s: interface %s addr %s", __func__, iface->name, + log_in6addr(addr)); + + mreq.ipv6mr_multiaddr = *addr; + mreq.ipv6mr_interface = iface->ifindex; + + if (setsockopt(global.ipv6.ldp_disc_socket, IPPROTO_IPV6, + IPV6_JOIN_GROUP, &mreq, sizeof(mreq)) < 0) { + log_warn("%s: error IPV6_JOIN_GROUP, interface %s address %s", + __func__, iface->name, log_in6addr(addr)); + return (-1); + } + + return (0); +} + +static int +if_leave_ipv6_group(struct iface *iface, struct in6_addr *addr) +{ + struct ipv6_mreq mreq; + + log_debug("%s: interface %s addr %s", __func__, iface->name, + log_in6addr(addr)); + + mreq.ipv6mr_multiaddr = *addr; + mreq.ipv6mr_interface = iface->ifindex; + + if (setsockopt(global.ipv6.ldp_disc_socket, IPPROTO_IPV6, + IPV6_LEAVE_GROUP, (void *)&mreq, sizeof(mreq)) < 0) { + log_warn("%s: error IPV6_LEAVE_GROUP, interface %s address %s", + __func__, iface->name, log_in6addr(addr)); + return (-1); + } + + return (0); +} diff --git a/ldpd/keepalive.c b/ldpd/keepalive.c new file mode 100644 index 000000000..4cd49d485 --- /dev/null +++ b/ldpd/keepalive.c @@ -0,0 +1,58 @@ +/* $OpenBSD$ */ + +/* + * Copyright (c) 2009 Michele Marchetto + * + * 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 "ldpd.h" +#include "ldpe.h" +#include "log.h" + +void +send_keepalive(struct nbr *nbr) +{ + struct ibuf *buf; + uint16_t size; + + size = LDP_HDR_SIZE + LDP_MSG_SIZE; + if ((buf = ibuf_open(size)) == NULL) + fatal(__func__); + + gen_ldp_hdr(buf, size); + size -= LDP_HDR_SIZE; + gen_msg_hdr(buf, MSG_TYPE_KEEPALIVE, size); + + evbuf_enqueue(&nbr->tcp->wbuf, buf); +} + +int +recv_keepalive(struct nbr *nbr, char *buf, uint16_t len) +{ + struct ldp_msg msg; + + memcpy(&msg, buf, sizeof(msg)); + if (len != LDP_MSG_SIZE) { + session_shutdown(nbr, S_BAD_MSG_LEN, msg.id, msg.type); + return (-1); + } + + if (nbr->state != NBR_STA_OPER) + nbr_fsm(nbr, NBR_EVT_KEEPALIVE_RCVD); + + return (0); +} diff --git a/ldpd/l2vpn.c b/ldpd/l2vpn.c new file mode 100644 index 000000000..22c98745e --- /dev/null +++ b/ldpd/l2vpn.c @@ -0,0 +1,510 @@ +/* $OpenBSD$ */ + +/* + * Copyright (c) 2015 Renato Westphal + * Copyright (c) 2009 Michele Marchetto + * Copyright (c) 2005 Claudio Jeker + * Copyright (c) 2004, 2005, 2008 Esben Norby + * + * 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 "ldpd.h" +#include "ldpe.h" +#include "lde.h" +#include "log.h" + +static void l2vpn_pw_fec(struct l2vpn_pw *, struct fec *); + +struct l2vpn * +l2vpn_new(const char *name) +{ + struct l2vpn *l2vpn; + + if ((l2vpn = calloc(1, sizeof(*l2vpn))) == NULL) + fatal("l2vpn_new: calloc"); + + strlcpy(l2vpn->name, name, sizeof(l2vpn->name)); + + /* set default values */ + l2vpn->mtu = DEFAULT_L2VPN_MTU; + l2vpn->pw_type = DEFAULT_PW_TYPE; + + LIST_INIT(&l2vpn->if_list); + LIST_INIT(&l2vpn->pw_list); + + return (l2vpn); +} + +struct l2vpn * +l2vpn_find(struct ldpd_conf *xconf, const char *name) +{ + struct l2vpn *l2vpn; + + LIST_FOREACH(l2vpn, &xconf->l2vpn_list, entry) + if (strcmp(l2vpn->name, name) == 0) + return (l2vpn); + + return (NULL); +} + +void +l2vpn_del(struct l2vpn *l2vpn) +{ + struct l2vpn_if *lif; + struct l2vpn_pw *pw; + + while ((lif = LIST_FIRST(&l2vpn->if_list)) != NULL) { + LIST_REMOVE(lif, entry); + free(lif); + } + while ((pw = LIST_FIRST(&l2vpn->pw_list)) != NULL) { + LIST_REMOVE(pw, entry); + free(pw); + } + + free(l2vpn); +} + +void +l2vpn_init(struct l2vpn *l2vpn) +{ + struct l2vpn_pw *pw; + + LIST_FOREACH(pw, &l2vpn->pw_list, entry) + l2vpn_pw_init(pw); +} + +void +l2vpn_exit(struct l2vpn *l2vpn) +{ + struct l2vpn_pw *pw; + + LIST_FOREACH(pw, &l2vpn->pw_list, entry) + l2vpn_pw_exit(pw); +} + +struct l2vpn_if * +l2vpn_if_new(struct l2vpn *l2vpn, struct kif *kif) +{ + struct l2vpn_if *lif; + + if ((lif = calloc(1, sizeof(*lif))) == NULL) + fatal("l2vpn_if_new: calloc"); + + lif->l2vpn = l2vpn; + strlcpy(lif->ifname, kif->ifname, sizeof(lif->ifname)); + lif->ifindex = kif->ifindex; + lif->flags = kif->flags; + lif->link_state = kif->link_state; + + return (lif); +} + +struct l2vpn_if * +l2vpn_if_find(struct l2vpn *l2vpn, unsigned int ifindex) +{ + struct l2vpn_if *lif; + + LIST_FOREACH(lif, &l2vpn->if_list, entry) + if (lif->ifindex == ifindex) + return (lif); + + return (NULL); +} + +struct l2vpn_pw * +l2vpn_pw_new(struct l2vpn *l2vpn, struct kif *kif) +{ + struct l2vpn_pw *pw; + + if ((pw = calloc(1, sizeof(*pw))) == NULL) + fatal("l2vpn_pw_new: calloc"); + + pw->l2vpn = l2vpn; + strlcpy(pw->ifname, kif->ifname, sizeof(pw->ifname)); + pw->ifindex = kif->ifindex; + + return (pw); +} + +struct l2vpn_pw * +l2vpn_pw_find(struct l2vpn *l2vpn, unsigned int ifindex) +{ + struct l2vpn_pw *pw; + + LIST_FOREACH(pw, &l2vpn->pw_list, entry) + if (pw->ifindex == ifindex) + return (pw); + + return (NULL); +} + +void +l2vpn_pw_init(struct l2vpn_pw *pw) +{ + struct fec fec; + + l2vpn_pw_reset(pw); + + l2vpn_pw_fec(pw, &fec); + lde_kernel_insert(&fec, AF_INET, (union ldpd_addr*)&pw->lsr_id, 0, + 0, (void *)pw); +} + +void +l2vpn_pw_exit(struct l2vpn_pw *pw) +{ + struct fec fec; + + l2vpn_pw_fec(pw, &fec); + lde_kernel_remove(&fec, AF_INET, (union ldpd_addr*)&pw->lsr_id, 0); +} + +static void +l2vpn_pw_fec(struct l2vpn_pw *pw, struct fec *fec) +{ + memset(fec, 0, sizeof(*fec)); + fec->type = FEC_TYPE_PWID; + fec->u.pwid.type = pw->l2vpn->pw_type; + fec->u.pwid.pwid = pw->pwid; + fec->u.pwid.lsr_id = pw->lsr_id; +} + +void +l2vpn_pw_reset(struct l2vpn_pw *pw) +{ + pw->remote_group = 0; + pw->remote_mtu = 0; + pw->remote_status = 0; + + if (pw->flags & F_PW_CWORD_CONF) + pw->flags |= F_PW_CWORD; + else + pw->flags &= ~F_PW_CWORD; + + if (pw->flags & F_PW_STATUSTLV_CONF) + pw->flags |= F_PW_STATUSTLV; + else + pw->flags &= ~F_PW_STATUSTLV; +} + +int +l2vpn_pw_ok(struct l2vpn_pw *pw, struct fec_nh *fnh) +{ + struct fec fec; + struct fec_node *fn; + + /* check for a remote label */ + if (fnh->remote_label == NO_LABEL) + return (0); + + /* MTUs must match */ + if (pw->l2vpn->mtu != pw->remote_mtu) + return (0); + + /* check pw status if applicable */ + if ((pw->flags & F_PW_STATUSTLV) && + pw->remote_status != PW_FORWARDING) + return (0); + + /* check for a working lsp to the nexthop */ + memset(&fec, 0, sizeof(fec)); + switch (pw->af) { + case AF_INET: + fec.type = FEC_TYPE_IPV4; + fec.u.ipv4.prefix = pw->addr.v4; + fec.u.ipv4.prefixlen = 32; + break; + case AF_INET6: + fec.type = FEC_TYPE_IPV6; + fec.u.ipv6.prefix = pw->addr.v6; + fec.u.ipv6.prefixlen = 128; + break; + default: + fatalx("l2vpn_pw_ok: unknown af"); + } + + fn = (struct fec_node *)fec_find(&ft, &fec); + if (fn == NULL || fn->local_label == NO_LABEL) + return (0); + /* + * Need to ensure that there's a label binding for all nexthops. + * Otherwise, ECMP for this route could render the pseudowire unusable. + */ + LIST_FOREACH(fnh, &fn->nexthops, entry) + if (fnh->remote_label == NO_LABEL) + return (0); + + return (1); +} + +int +l2vpn_pw_negotiate(struct lde_nbr *ln, struct fec_node *fn, struct map *map) +{ + struct l2vpn_pw *pw; + struct status_tlv st; + + /* NOTE: thanks martini & friends for all this mess */ + + pw = (struct l2vpn_pw *) fn->data; + if (pw == NULL) + /* + * pseudowire not configured, return and record + * the mapping later + */ + return (0); + + /* RFC4447 - Section 6.2: control word negotiation */ + if (fec_find(&ln->sent_map, &fn->fec)) { + if ((map->flags & F_MAP_PW_CWORD) && + !(pw->flags & F_PW_CWORD_CONF)) { + /* ignore the received label mapping */ + return (1); + } else if (!(map->flags & F_MAP_PW_CWORD) && + (pw->flags & F_PW_CWORD_CONF)) { + /* append a "Wrong C-bit" status code */ + st.status_code = S_WRONG_CBIT; + st.msg_id = map->msg_id; + st.msg_type = htons(MSG_TYPE_LABELMAPPING); + lde_send_labelwithdraw(ln, fn, NO_LABEL, &st); + + pw->flags &= ~F_PW_CWORD; + lde_send_labelmapping(ln, fn, 1); + } + } else if (map->flags & F_MAP_PW_CWORD) { + if (pw->flags & F_PW_CWORD_CONF) + pw->flags |= F_PW_CWORD; + else + /* act as if no label mapping had been received */ + return (1); + } else + pw->flags &= ~F_PW_CWORD; + + /* RFC4447 - Section 5.4.3: pseudowire status negotiation */ + if (fec_find(&ln->recv_map, &fn->fec) == NULL && + !(map->flags & F_MAP_PW_STATUS)) + pw->flags &= ~F_PW_STATUSTLV; + + return (0); +} + +void +l2vpn_send_pw_status(uint32_t peerid, uint32_t status, struct fec *fec) +{ + struct notify_msg nm; + + memset(&nm, 0, sizeof(nm)); + nm.status_code = S_PW_STATUS; + nm.pw_status = status; + nm.flags |= F_NOTIF_PW_STATUS; + lde_fec2map(fec, &nm.fec); + nm.flags |= F_NOTIF_FEC; + + lde_imsg_compose_ldpe(IMSG_NOTIFICATION_SEND, peerid, 0, + &nm, sizeof(nm)); +} + +void +l2vpn_recv_pw_status(struct lde_nbr *ln, struct notify_msg *nm) +{ + struct fec fec; + struct fec_node *fn; + struct fec_nh *fnh; + struct l2vpn_pw *pw; + + /* TODO group wildcard */ + if (!(nm->fec.flags & F_MAP_PW_ID)) + return; + + lde_map2fec(&nm->fec, ln->id, &fec); + fn = (struct fec_node *)fec_find(&ft, &fec); + if (fn == NULL) + /* unknown fec */ + return; + + pw = (struct l2vpn_pw *) fn->data; + if (pw == NULL) + return; + + fnh = fec_nh_find(fn, AF_INET, (union ldpd_addr *)&ln->id, 0); + if (fnh == NULL) + return; + + /* remote status didn't change */ + if (pw->remote_status == nm->pw_status) + return; + + pw->remote_status = nm->pw_status; + + if (l2vpn_pw_ok(pw, fnh)) + lde_send_change_klabel(fn, fnh); + else + lde_send_delete_klabel(fn, fnh); +} + +void +l2vpn_sync_pws(int af, union ldpd_addr *addr) +{ + struct l2vpn *l2vpn; + struct l2vpn_pw *pw; + struct fec fec; + struct fec_node *fn; + struct fec_nh *fnh; + + LIST_FOREACH(l2vpn, &ldeconf->l2vpn_list, entry) { + LIST_FOREACH(pw, &l2vpn->pw_list, entry) { + if (af != pw->af || ldp_addrcmp(af, &pw->addr, addr)) + continue; + + l2vpn_pw_fec(pw, &fec); + fn = (struct fec_node *)fec_find(&ft, &fec); + if (fn == NULL) + continue; + fnh = fec_nh_find(fn, AF_INET, (union ldpd_addr *) + &pw->lsr_id, 0); + if (fnh == NULL) + continue; + + if (l2vpn_pw_ok(pw, fnh)) + lde_send_change_klabel(fn, fnh); + else + lde_send_delete_klabel(fn, fnh); + } + } +} + +void +l2vpn_pw_ctl(pid_t pid) +{ + struct l2vpn *l2vpn; + struct l2vpn_pw *pw; + static struct ctl_pw pwctl; + + LIST_FOREACH(l2vpn, &ldeconf->l2vpn_list, entry) + LIST_FOREACH(pw, &l2vpn->pw_list, entry) { + memset(&pwctl, 0, sizeof(pwctl)); + strlcpy(pwctl.ifname, pw->ifname, + sizeof(pwctl.ifname)); + pwctl.pwid = pw->pwid; + pwctl.lsr_id = pw->lsr_id; + pwctl.status = pw->flags & F_PW_STATUS_UP; + + lde_imsg_compose_ldpe(IMSG_CTL_SHOW_L2VPN_PW, 0, + pid, &pwctl, sizeof(pwctl)); + } +} + +void +l2vpn_binding_ctl(pid_t pid) +{ + struct fec *f; + struct fec_node *fn; + struct lde_map *me; + struct l2vpn_pw *pw; + static struct ctl_pw pwctl; + + RB_FOREACH(f, fec_tree, &ft) { + if (f->type != FEC_TYPE_PWID) + continue; + + fn = (struct fec_node *)f; + if (fn->local_label == NO_LABEL && + LIST_EMPTY(&fn->downstream)) + continue; + + memset(&pwctl, 0, sizeof(pwctl)); + pwctl.type = f->u.pwid.type; + pwctl.pwid = f->u.pwid.pwid; + pwctl.lsr_id = f->u.pwid.lsr_id; + + pw = (struct l2vpn_pw *) fn->data; + if (pw) { + pwctl.local_label = fn->local_label; + pwctl.local_gid = 0; + pwctl.local_ifmtu = pw->l2vpn->mtu; + } else + pwctl.local_label = NO_LABEL; + + LIST_FOREACH(me, &fn->downstream, entry) + if (f->u.pwid.lsr_id.s_addr == me->nexthop->id.s_addr) + break; + + if (me) { + pwctl.remote_label = me->map.label; + pwctl.remote_gid = me->map.fec.pwid.group_id; + if (me->map.flags & F_MAP_PW_IFMTU) + pwctl.remote_ifmtu = me->map.fec.pwid.ifmtu; + + lde_imsg_compose_ldpe(IMSG_CTL_SHOW_L2VPN_BINDING, + 0, pid, &pwctl, sizeof(pwctl)); + } else if (pw) { + pwctl.remote_label = NO_LABEL; + + lde_imsg_compose_ldpe(IMSG_CTL_SHOW_L2VPN_BINDING, + 0, pid, &pwctl, sizeof(pwctl)); + } + } +} + +/* ldpe */ + +void +ldpe_l2vpn_init(struct l2vpn *l2vpn) +{ + struct l2vpn_pw *pw; + + LIST_FOREACH(pw, &l2vpn->pw_list, entry) + ldpe_l2vpn_pw_init(pw); +} + +void +ldpe_l2vpn_exit(struct l2vpn *l2vpn) +{ + struct l2vpn_pw *pw; + + LIST_FOREACH(pw, &l2vpn->pw_list, entry) + ldpe_l2vpn_pw_exit(pw); +} + +void +ldpe_l2vpn_pw_init(struct l2vpn_pw *pw) +{ + struct tnbr *tnbr; + + tnbr = tnbr_find(leconf, pw->af, &pw->addr); + if (tnbr == NULL) { + tnbr = tnbr_new(leconf, pw->af, &pw->addr); + tnbr_update(tnbr); + LIST_INSERT_HEAD(&leconf->tnbr_list, tnbr, entry); + } + + tnbr->pw_count++; +} + +void +ldpe_l2vpn_pw_exit(struct l2vpn_pw *pw) +{ + struct tnbr *tnbr; + + tnbr = tnbr_find(leconf, pw->af, &pw->addr); + if (tnbr) { + tnbr->pw_count--; + tnbr_check(tnbr); + } +} diff --git a/ldpd/labelmapping.c b/ldpd/labelmapping.c new file mode 100644 index 000000000..88e64071b --- /dev/null +++ b/ldpd/labelmapping.c @@ -0,0 +1,764 @@ +/* $OpenBSD$ */ + +/* + * Copyright (c) 2014, 2015 Renato Westphal + * Copyright (c) 2009 Michele Marchetto + * + * 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 "ldpd.h" +#include "ldpe.h" +#include "log.h" + +static void enqueue_pdu(struct nbr *, struct ibuf *, uint16_t); +static int gen_label_tlv(struct ibuf *, uint32_t); +static int tlv_decode_label(struct nbr *, struct ldp_msg *, char *, + uint16_t, uint32_t *); +static int gen_reqid_tlv(struct ibuf *, uint32_t); + +static void +enqueue_pdu(struct nbr *nbr, struct ibuf *buf, uint16_t size) +{ + struct ldp_hdr *ldp_hdr; + + ldp_hdr = ibuf_seek(buf, 0, sizeof(struct ldp_hdr)); + ldp_hdr->length = htons(size); + evbuf_enqueue(&nbr->tcp->wbuf, buf); +} + +/* Generic function that handles all Label Message types */ +void +send_labelmessage(struct nbr *nbr, uint16_t type, struct mapping_head *mh) +{ + struct ibuf *buf = NULL; + struct mapping_entry *me; + uint16_t msg_size, size = 0; + int first = 1; + int err = 0; + + /* nothing to send */ + if (TAILQ_EMPTY(mh)) + return; + + while ((me = TAILQ_FIRST(mh)) != NULL) { + /* generate pdu */ + if (first) { + if ((buf = ibuf_open(nbr->max_pdu_len + + LDP_HDR_DEAD_LEN)) == NULL) + fatal(__func__); + + /* real size will be set up later */ + err |= gen_ldp_hdr(buf, 0); + + size = LDP_HDR_PDU_LEN; + first = 0; + } + + /* calculate size */ + msg_size = LDP_MSG_SIZE + TLV_HDR_SIZE; + switch (me->map.type) { + case MAP_TYPE_WILDCARD: + msg_size += FEC_ELM_WCARD_LEN; + break; + case MAP_TYPE_PREFIX: + msg_size += FEC_ELM_PREFIX_MIN_LEN + + PREFIX_SIZE(me->map.fec.prefix.prefixlen); + break; + case MAP_TYPE_PWID: + msg_size += FEC_PWID_ELM_MIN_LEN; + if (me->map.flags & F_MAP_PW_ID) + msg_size += PW_STATUS_TLV_LEN; + if (me->map.flags & F_MAP_PW_IFMTU) + msg_size += FEC_SUBTLV_IFMTU_SIZE; + if (me->map.flags & F_MAP_PW_STATUS) + msg_size += PW_STATUS_TLV_SIZE; + break; + } + if (me->map.label != NO_LABEL) + msg_size += LABEL_TLV_SIZE; + if (me->map.flags & F_MAP_REQ_ID) + msg_size += REQID_TLV_SIZE; + if (me->map.flags & F_MAP_STATUS) + msg_size += STATUS_SIZE; + + /* maximum pdu length exceeded, we need a new ldp pdu */ + if (size + msg_size > nbr->max_pdu_len) { + enqueue_pdu(nbr, buf, size); + first = 1; + continue; + } + + size += msg_size; + + /* append message and tlvs */ + err |= gen_msg_hdr(buf, type, msg_size); + err |= gen_fec_tlv(buf, &me->map); + if (me->map.label != NO_LABEL) + err |= gen_label_tlv(buf, me->map.label); + if (me->map.flags & F_MAP_REQ_ID) + err |= gen_reqid_tlv(buf, me->map.requestid); + if (me->map.flags & F_MAP_PW_STATUS) + err |= gen_pw_status_tlv(buf, me->map.pw_status); + if (me->map.flags & F_MAP_STATUS) + err |= gen_status_tlv(buf, me->map.st.status_code, + me->map.st.msg_id, me->map.st.msg_type); + if (err) { + ibuf_free(buf); + mapping_list_clr(mh); + return; + } + + log_debug("msg-out: %s: lsr-id %s, fec %s, label %s", + msg_name(type), inet_ntoa(nbr->id), log_map(&me->map), + log_label(me->map.label)); + + TAILQ_REMOVE(mh, me, entry); + free(me); + } + + enqueue_pdu(nbr, buf, size); + + nbr_fsm(nbr, NBR_EVT_PDU_SENT); +} + +/* Generic function that handles all Label Message types */ +int +recv_labelmessage(struct nbr *nbr, char *buf, uint16_t len, uint16_t type) +{ + struct ldp_msg msg; + struct tlv ft; + uint32_t label = NO_LABEL, reqid = 0; + uint32_t pw_status = 0; + uint8_t flags = 0; + int feclen, lbllen, tlen; + struct mapping_entry *me; + struct mapping_head mh; + struct map map; + + memcpy(&msg, buf, sizeof(msg)); + buf += LDP_MSG_SIZE; + len -= LDP_MSG_SIZE; + + /* FEC TLV */ + if (len < sizeof(ft)) { + session_shutdown(nbr, S_BAD_TLV_LEN, msg.id, msg.type); + return (-1); + } + + memcpy(&ft, buf, sizeof(ft)); + if (ntohs(ft.type) != TLV_TYPE_FEC) { + send_notification_nbr(nbr, S_MISS_MSG, msg.id, msg.type); + return (-1); + } + feclen = ntohs(ft.length); + if (feclen > len - TLV_HDR_SIZE) { + session_shutdown(nbr, S_BAD_TLV_LEN, msg.id, msg.type); + return (-1); + } + + buf += TLV_HDR_SIZE; /* just advance to the end of the fec header */ + len -= TLV_HDR_SIZE; + + TAILQ_INIT(&mh); + do { + memset(&map, 0, sizeof(map)); + map.msg_id = msg.id; + + if ((tlen = tlv_decode_fec_elm(nbr, &msg, buf, feclen, + &map)) == -1) + goto err; + if (map.type == MAP_TYPE_PWID && + !(map.flags & F_MAP_PW_ID) && + type != MSG_TYPE_LABELWITHDRAW && + type != MSG_TYPE_LABELRELEASE) { + send_notification_nbr(nbr, S_MISS_MSG, msg.id, + msg.type); + return (-1); + } + + /* + * The Wildcard FEC Element can be used only in the + * Label Withdraw and Label Release messages. + */ + if (map.type == MAP_TYPE_WILDCARD) { + switch (type) { + case MSG_TYPE_LABELMAPPING: + case MSG_TYPE_LABELREQUEST: + case MSG_TYPE_LABELABORTREQ: + session_shutdown(nbr, S_UNKNOWN_FEC, msg.id, + msg.type); + goto err; + default: + break; + } + } + + /* + * LDP supports the use of multiple FEC Elements per + * FEC for the Label Mapping message only. + */ + if (type != MSG_TYPE_LABELMAPPING && + tlen != feclen) { + session_shutdown(nbr, S_BAD_TLV_VAL, msg.id, msg.type); + goto err; + } + + mapping_list_add(&mh, &map); + + buf += tlen; + len -= tlen; + feclen -= tlen; + } while (feclen > 0); + + /* Mandatory Label TLV */ + if (type == MSG_TYPE_LABELMAPPING) { + lbllen = tlv_decode_label(nbr, &msg, buf, len, &label); + if (lbllen == -1) + goto err; + + buf += lbllen; + len -= lbllen; + } + + /* Optional Parameters */ + while (len > 0) { + struct tlv tlv; + uint16_t tlv_len; + uint32_t reqbuf, labelbuf, statusbuf; + + if (len < sizeof(tlv)) { + session_shutdown(nbr, S_BAD_TLV_LEN, msg.id, msg.type); + goto err; + } + + memcpy(&tlv, buf, TLV_HDR_SIZE); + tlv_len = ntohs(tlv.length); + if (tlv_len + TLV_HDR_SIZE > len) { + session_shutdown(nbr, S_BAD_TLV_LEN, msg.id, msg.type); + goto err; + } + buf += TLV_HDR_SIZE; + len -= TLV_HDR_SIZE; + + switch (ntohs(tlv.type)) { + case TLV_TYPE_LABELREQUEST: + switch (type) { + case MSG_TYPE_LABELMAPPING: + case MSG_TYPE_LABELREQUEST: + if (tlv_len != REQID_TLV_LEN) { + session_shutdown(nbr, S_BAD_TLV_LEN, + msg.id, msg.type); + goto err; + } + + flags |= F_MAP_REQ_ID; + memcpy(&reqbuf, buf, sizeof(reqbuf)); + reqid = ntohl(reqbuf); + break; + default: + /* ignore */ + break; + } + break; + case TLV_TYPE_HOPCOUNT: + case TLV_TYPE_PATHVECTOR: + /* ignore */ + break; + case TLV_TYPE_GENERICLABEL: + switch (type) { + case MSG_TYPE_LABELWITHDRAW: + case MSG_TYPE_LABELRELEASE: + if (tlv_len != LABEL_TLV_LEN) { + session_shutdown(nbr, S_BAD_TLV_LEN, + msg.id, msg.type); + goto err; + } + + memcpy(&labelbuf, buf, sizeof(labelbuf)); + label = ntohl(labelbuf); + break; + default: + /* ignore */ + break; + } + break; + case TLV_TYPE_ATMLABEL: + case TLV_TYPE_FRLABEL: + switch (type) { + case MSG_TYPE_LABELWITHDRAW: + case MSG_TYPE_LABELRELEASE: + /* unsupported */ + session_shutdown(nbr, S_BAD_TLV_VAL, msg.id, + msg.type); + goto err; + break; + default: + /* ignore */ + break; + } + break; + case TLV_TYPE_STATUS: + if (tlv_len != STATUS_TLV_LEN) { + session_shutdown(nbr, S_BAD_TLV_LEN, msg.id, + msg.type); + goto err; + } + /* ignore */ + break; + case TLV_TYPE_PW_STATUS: + switch (type) { + case MSG_TYPE_LABELMAPPING: + if (tlv_len != PW_STATUS_TLV_LEN) { + session_shutdown(nbr, S_BAD_TLV_LEN, + msg.id, msg.type); + goto err; + } + + flags |= F_MAP_PW_STATUS; + memcpy(&statusbuf, buf, sizeof(statusbuf)); + pw_status = ntohl(statusbuf); + break; + default: + /* ignore */ + break; + } + break; + default: + if (!(ntohs(tlv.type) & UNKNOWN_FLAG)) + send_notification_nbr(nbr, S_UNKNOWN_TLV, + msg.id, msg.type); + /* ignore unknown tlv */ + break; + } + buf += tlv_len; + len -= tlv_len; + } + + /* notify lde about the received message. */ + while ((me = TAILQ_FIRST(&mh)) != NULL) { + int imsg_type = IMSG_NONE; + + me->map.flags |= flags; + switch (me->map.type) { + case MAP_TYPE_PREFIX: + switch (me->map.fec.prefix.af) { + case AF_INET: + if (label == MPLS_LABEL_IPV6NULL) { + session_shutdown(nbr, S_BAD_TLV_VAL, + msg.id, msg.type); + goto err; + } + if (!nbr->v4_enabled) + goto next; + break; + case AF_INET6: + if (label == MPLS_LABEL_IPV4NULL) { + session_shutdown(nbr, S_BAD_TLV_VAL, + msg.id, msg.type); + goto err; + } + if (!nbr->v6_enabled) + goto next; + break; + default: + fatalx("recv_labelmessage: unknown af"); + } + break; + case MAP_TYPE_PWID: + if (label <= MPLS_LABEL_RESERVED_MAX) { + session_shutdown(nbr, S_BAD_TLV_VAL, msg.id, + msg.type); + goto err; + } + if (me->map.flags & F_MAP_PW_STATUS) + me->map.pw_status = pw_status; + break; + default: + break; + } + me->map.label = label; + if (me->map.flags & F_MAP_REQ_ID) + me->map.requestid = reqid; + + log_debug("msg-in: label mapping: lsr-id %s, fec %s, label %s", + inet_ntoa(nbr->id), log_map(&me->map), + log_label(me->map.label)); + + switch (type) { + case MSG_TYPE_LABELMAPPING: + imsg_type = IMSG_LABEL_MAPPING; + break; + case MSG_TYPE_LABELREQUEST: + imsg_type = IMSG_LABEL_REQUEST; + break; + case MSG_TYPE_LABELWITHDRAW: + imsg_type = IMSG_LABEL_WITHDRAW; + break; + case MSG_TYPE_LABELRELEASE: + imsg_type = IMSG_LABEL_RELEASE; + break; + case MSG_TYPE_LABELABORTREQ: + imsg_type = IMSG_LABEL_ABORT; + break; + default: + break; + } + + ldpe_imsg_compose_lde(imsg_type, nbr->peerid, 0, &me->map, + sizeof(struct map)); + +next: + TAILQ_REMOVE(&mh, me, entry); + free(me); + } + + return (0); + +err: + mapping_list_clr(&mh); + + return (-1); +} + +/* Other TLV related functions */ +static int +gen_label_tlv(struct ibuf *buf, uint32_t label) +{ + struct label_tlv lt; + + lt.type = htons(TLV_TYPE_GENERICLABEL); + lt.length = htons(LABEL_TLV_LEN); + lt.label = htonl(label); + + return (ibuf_add(buf, <, sizeof(lt))); +} + +static int +tlv_decode_label(struct nbr *nbr, struct ldp_msg *msg, char *buf, + uint16_t len, uint32_t *label) +{ + struct label_tlv lt; + + if (len < sizeof(lt)) { + session_shutdown(nbr, S_BAD_TLV_LEN, msg->id, msg->type); + return (-1); + } + memcpy(<, buf, sizeof(lt)); + + if (!(ntohs(lt.type) & TLV_TYPE_GENERICLABEL)) { + send_notification_nbr(nbr, S_MISS_MSG, msg->id, msg->type); + return (-1); + } + + switch (htons(lt.type)) { + case TLV_TYPE_GENERICLABEL: + if (ntohs(lt.length) != sizeof(lt) - TLV_HDR_SIZE) { + session_shutdown(nbr, S_BAD_TLV_LEN, msg->id, + msg->type); + return (-1); + } + + *label = ntohl(lt.label); + if (*label > MPLS_LABEL_MAX || + (*label <= MPLS_LABEL_RESERVED_MAX && + *label != MPLS_LABEL_IPV4NULL && + *label != MPLS_LABEL_IPV6NULL && + *label != MPLS_LABEL_IMPLNULL)) { + session_shutdown(nbr, S_BAD_TLV_VAL, msg->id, + msg->type); + return (-1); + } + break; + case TLV_TYPE_ATMLABEL: + case TLV_TYPE_FRLABEL: + default: + /* unsupported */ + session_shutdown(nbr, S_BAD_TLV_VAL, msg->id, msg->type); + return (-1); + } + + return (sizeof(lt)); +} + +static int +gen_reqid_tlv(struct ibuf *buf, uint32_t reqid) +{ + struct reqid_tlv rt; + + rt.type = htons(TLV_TYPE_LABELREQUEST); + rt.length = htons(REQID_TLV_LEN); + rt.reqid = htonl(reqid); + + return (ibuf_add(buf, &rt, sizeof(rt))); +} + +int +gen_pw_status_tlv(struct ibuf *buf, uint32_t status) +{ + struct pw_status_tlv st; + + st.type = htons(TLV_TYPE_PW_STATUS); + st.length = htons(PW_STATUS_TLV_LEN); + st.value = htonl(status); + + return (ibuf_add(buf, &st, sizeof(st))); +} + +int +gen_fec_tlv(struct ibuf *buf, struct map *map) +{ + struct tlv ft; + uint16_t family, len, pw_type, ifmtu; + uint8_t pw_len = 0; + uint32_t group_id, pwid; + int err = 0; + + ft.type = htons(TLV_TYPE_FEC); + + switch (map->type) { + case MAP_TYPE_WILDCARD: + ft.length = htons(sizeof(uint8_t)); + err |= ibuf_add(buf, &ft, sizeof(ft)); + err |= ibuf_add(buf, &map->type, sizeof(map->type)); + break; + case MAP_TYPE_PREFIX: + len = PREFIX_SIZE(map->fec.prefix.prefixlen); + ft.length = htons(sizeof(map->type) + sizeof(family) + + sizeof(map->fec.prefix.prefixlen) + len); + err |= ibuf_add(buf, &ft, sizeof(ft)); + err |= ibuf_add(buf, &map->type, sizeof(map->type)); + switch (map->fec.prefix.af) { + case AF_INET: + family = htons(AF_IPV4); + break; + case AF_INET6: + family = htons(AF_IPV6); + break; + default: + fatalx("gen_fec_tlv: unknown af"); + break; + } + err |= ibuf_add(buf, &family, sizeof(family)); + err |= ibuf_add(buf, &map->fec.prefix.prefixlen, + sizeof(map->fec.prefix.prefixlen)); + if (len) + err |= ibuf_add(buf, &map->fec.prefix.prefix, len); + break; + case MAP_TYPE_PWID: + if (map->flags & F_MAP_PW_ID) + pw_len += PW_STATUS_TLV_LEN; + if (map->flags & F_MAP_PW_IFMTU) + pw_len += FEC_SUBTLV_IFMTU_SIZE; + + len = FEC_PWID_ELM_MIN_LEN + pw_len; + + ft.length = htons(len); + err |= ibuf_add(buf, &ft, sizeof(ft)); + + err |= ibuf_add(buf, &map->type, sizeof(uint8_t)); + pw_type = map->fec.pwid.type; + if (map->flags & F_MAP_PW_CWORD) + pw_type |= CONTROL_WORD_FLAG; + pw_type = htons(pw_type); + err |= ibuf_add(buf, &pw_type, sizeof(uint16_t)); + err |= ibuf_add(buf, &pw_len, sizeof(uint8_t)); + group_id = htonl(map->fec.pwid.group_id); + err |= ibuf_add(buf, &group_id, sizeof(uint32_t)); + if (map->flags & F_MAP_PW_ID) { + pwid = htonl(map->fec.pwid.pwid); + err |= ibuf_add(buf, &pwid, sizeof(uint32_t)); + } + if (map->flags & F_MAP_PW_IFMTU) { + struct subtlv stlv; + + stlv.type = SUBTLV_IFMTU; + stlv.length = FEC_SUBTLV_IFMTU_SIZE; + err |= ibuf_add(buf, &stlv, sizeof(uint16_t)); + + ifmtu = htons(map->fec.pwid.ifmtu); + err |= ibuf_add(buf, &ifmtu, sizeof(uint16_t)); + } + break; + default: + break; + } + + return (err); +} + +int +tlv_decode_fec_elm(struct nbr *nbr, struct ldp_msg *msg, char *buf, + uint16_t len, struct map *map) +{ + uint16_t off = 0; + uint8_t pw_len; + + map->type = *buf; + off += sizeof(uint8_t); + + switch (map->type) { + case MAP_TYPE_WILDCARD: + if (len == FEC_ELM_WCARD_LEN) + return (off); + else { + session_shutdown(nbr, S_BAD_TLV_VAL, msg->id, + msg->type); + return (-1); + } + break; + case MAP_TYPE_PREFIX: + if (len < FEC_ELM_PREFIX_MIN_LEN) { + session_shutdown(nbr, S_BAD_TLV_LEN, msg->id, + msg->type); + return (-1); + } + + /* Address Family */ + memcpy(&map->fec.prefix.af, buf + off, + sizeof(map->fec.prefix.af)); + off += sizeof(map->fec.prefix.af); + map->fec.prefix.af = ntohs(map->fec.prefix.af); + switch (map->fec.prefix.af) { + case AF_IPV4: + map->fec.prefix.af = AF_INET; + break; + case AF_IPV6: + map->fec.prefix.af = AF_INET6; + break; + default: + send_notification_nbr(nbr, S_UNSUP_ADDR, msg->id, + msg->type); + return (-1); + } + + /* Prefix Length */ + map->fec.prefix.prefixlen = buf[off]; + off += sizeof(uint8_t); + if (len < off + PREFIX_SIZE(map->fec.prefix.prefixlen)) { + session_shutdown(nbr, S_BAD_TLV_LEN, msg->id, + msg->type); + return (-1); + } + + /* Prefix */ + memset(&map->fec.prefix.prefix, 0, + sizeof(map->fec.prefix.prefix)); + memcpy(&map->fec.prefix.prefix, buf + off, + PREFIX_SIZE(map->fec.prefix.prefixlen)); + + /* Just in case... */ + ldp_applymask(map->fec.prefix.af, &map->fec.prefix.prefix, + &map->fec.prefix.prefix, map->fec.prefix.prefixlen); + + return (off + PREFIX_SIZE(map->fec.prefix.prefixlen)); + case MAP_TYPE_PWID: + if (len < FEC_PWID_ELM_MIN_LEN) { + session_shutdown(nbr, S_BAD_TLV_LEN, msg->id, + msg->type); + return (-1); + } + + /* PW type */ + memcpy(&map->fec.pwid.type, buf + off, sizeof(uint16_t)); + map->fec.pwid.type = ntohs(map->fec.pwid.type); + if (map->fec.pwid.type & CONTROL_WORD_FLAG) { + map->flags |= F_MAP_PW_CWORD; + map->fec.pwid.type &= ~CONTROL_WORD_FLAG; + } + off += sizeof(uint16_t); + + /* PW info Length */ + pw_len = buf[off]; + off += sizeof(uint8_t); + + if (len != FEC_PWID_ELM_MIN_LEN + pw_len) { + session_shutdown(nbr, S_BAD_TLV_LEN, msg->id, + msg->type); + return (-1); + } + + /* Group ID */ + memcpy(&map->fec.pwid.group_id, buf + off, sizeof(uint32_t)); + map->fec.pwid.group_id = ntohl(map->fec.pwid.group_id); + off += sizeof(uint32_t); + + /* PW ID */ + if (pw_len == 0) + return (off); + + if (pw_len < sizeof(uint32_t)) { + session_shutdown(nbr, S_BAD_TLV_LEN, msg->id, + msg->type); + return (-1); + } + + memcpy(&map->fec.pwid.pwid, buf + off, sizeof(uint32_t)); + map->fec.pwid.pwid = ntohl(map->fec.pwid.pwid); + map->flags |= F_MAP_PW_ID; + off += sizeof(uint32_t); + pw_len -= sizeof(uint32_t); + + /* Optional Interface Parameter Sub-TLVs */ + while (pw_len > 0) { + struct subtlv stlv; + + if (pw_len < sizeof(stlv)) { + session_shutdown(nbr, S_BAD_TLV_LEN, msg->id, + msg->type); + return (-1); + } + + memcpy(&stlv, buf + off, sizeof(stlv)); + if (stlv.length > pw_len) { + session_shutdown(nbr, S_BAD_TLV_LEN, msg->id, + msg->type); + return (-1); + } + + switch (stlv.type) { + case SUBTLV_IFMTU: + if (stlv.length != FEC_SUBTLV_IFMTU_SIZE) { + session_shutdown(nbr, S_BAD_TLV_LEN, + msg->id, msg->type); + return (-1); + } + memcpy(&map->fec.pwid.ifmtu, buf + off + + SUBTLV_HDR_SIZE, sizeof(uint16_t)); + map->fec.pwid.ifmtu = ntohs(map->fec.pwid.ifmtu); + map->flags |= F_MAP_PW_IFMTU; + break; + default: + /* ignore */ + break; + } + off += stlv.length; + pw_len -= stlv.length; + } + + return (off); + default: + send_notification_nbr(nbr, S_UNKNOWN_FEC, msg->id, msg->type); + break; + } + + return (-1); +} diff --git a/ldpd/lde.c b/ldpd/lde.c new file mode 100644 index 000000000..8859ed25b --- /dev/null +++ b/ldpd/lde.c @@ -0,0 +1,1335 @@ +/* $OpenBSD$ */ + +/* + * Copyright (c) 2013, 2016 Renato Westphal + * Copyright (c) 2004, 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 "ldp.h" +#include "ldpd.h" +#include "ldpe.h" +#include "log.h" +#include "lde.h" + +static void lde_sig_handler(int sig, short, void *); +static __dead void lde_shutdown(void); +static int lde_imsg_compose_parent(int, pid_t, void *, uint16_t); +static void lde_dispatch_imsg(int, short, void *); +static void lde_dispatch_parent(int, short, void *); +static __inline int lde_nbr_compare(struct lde_nbr *, + struct lde_nbr *); +static struct lde_nbr *lde_nbr_new(uint32_t, struct lde_nbr *); +static void lde_nbr_del(struct lde_nbr *); +static struct lde_nbr *lde_nbr_find(uint32_t); +static void lde_nbr_clear(void); +static void lde_nbr_addr_update(struct lde_nbr *, + struct lde_addr *, int); +static void lde_map_free(void *); +static int lde_address_add(struct lde_nbr *, struct lde_addr *); +static int lde_address_del(struct lde_nbr *, struct lde_addr *); +static void lde_address_list_free(struct lde_nbr *); + +RB_GENERATE(nbr_tree, lde_nbr, entry, lde_nbr_compare) + +struct ldpd_conf *ldeconf; +struct nbr_tree lde_nbrs = RB_INITIALIZER(&lde_nbrs); + +static struct imsgev *iev_ldpe; +static struct imsgev *iev_main; + +/* ARGSUSED */ +static void +lde_sig_handler(int sig, short event, void *arg) +{ + /* + * signal handler rules don't apply, libevent decouples for us + */ + + switch (sig) { + case SIGINT: + case SIGTERM: + lde_shutdown(); + /* NOTREACHED */ + default: + fatalx("unexpected signal"); + } +} + +/* label decision engine */ +void +lde(int debug, int verbose) +{ + struct event ev_sigint, ev_sigterm; + struct timeval now; + struct passwd *pw; + + ldeconf = config_new_empty(); + + log_init(debug); + log_verbose(verbose); + + setproctitle("label decision engine"); + ldpd_process = PROC_LDE_ENGINE; + + if ((pw = getpwnam(LDPD_USER)) == NULL) + fatal("getpwnam"); + + if (chroot(pw->pw_dir) == -1) + fatal("chroot"); + if (chdir("/") == -1) + fatal("chdir(\"/\")"); + + 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"); + + if (pledge("stdio recvfd", NULL) == -1) + fatal("pledge"); + + event_init(); + + /* setup signal handler */ + signal_set(&ev_sigint, SIGINT, lde_sig_handler, NULL); + signal_set(&ev_sigterm, SIGTERM, lde_sig_handler, NULL); + signal_add(&ev_sigint, NULL); + signal_add(&ev_sigterm, NULL); + signal(SIGPIPE, SIG_IGN); + signal(SIGHUP, SIG_IGN); + + /* setup pipe and event handler to the parent process */ + if ((iev_main = malloc(sizeof(struct imsgev))) == NULL) + fatal(NULL); + imsg_init(&iev_main->ibuf, 3); + iev_main->handler = lde_dispatch_parent; + 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); + + /* setup and start the LIB garbage collector */ + evtimer_set(&gc_timer, lde_gc_timer, NULL); + lde_gc_start_timer(); + + gettimeofday(&now, NULL); + global.uptime = now.tv_sec; + + event_dispatch(); + + lde_shutdown(); +} + +static __dead void +lde_shutdown(void) +{ + /* close pipes */ + msgbuf_clear(&iev_ldpe->ibuf.w); + close(iev_ldpe->ibuf.fd); + msgbuf_clear(&iev_main->ibuf.w); + close(iev_main->ibuf.fd); + + lde_gc_stop_timer(); + lde_nbr_clear(); + fec_tree_clear(); + + config_clear(ldeconf); + + free(iev_ldpe); + free(iev_main); + + log_info("label decision engine exiting"); + exit(0); +} + +/* imesg */ +static int +lde_imsg_compose_parent(int type, pid_t pid, void *data, uint16_t datalen) +{ + return (imsg_compose_event(iev_main, type, 0, pid, -1, data, datalen)); +} + +int +lde_imsg_compose_ldpe(int type, uint32_t peerid, pid_t pid, void *data, + uint16_t datalen) +{ + return (imsg_compose_event(iev_ldpe, type, peerid, pid, + -1, data, datalen)); +} + +/* ARGSUSED */ +static void +lde_dispatch_imsg(int fd, short event, void *bula) +{ + struct imsgev *iev = bula; + struct imsgbuf *ibuf = &iev->ibuf; + struct imsg imsg; + struct lde_nbr *ln; + struct map map; + struct lde_addr lde_addr; + struct notify_msg nm; + ssize_t n; + int shut = 0, verbose; + + if (event & EV_READ) { + if ((n = imsg_read(ibuf)) == -1 && errno != EAGAIN) + fatal("imsg_read error"); + if (n == 0) /* connection closed */ + shut = 1; + } + if (event & EV_WRITE) { + if ((n = msgbuf_write(&ibuf->w)) == -1 && errno != EAGAIN) + fatal("msgbuf_write"); + if (n == 0) /* connection closed */ + shut = 1; + } + + for (;;) { + if ((n = imsg_get(ibuf, &imsg)) == -1) + fatal("lde_dispatch_imsg: imsg_get error"); + if (n == 0) + break; + + switch (imsg.hdr.type) { + case IMSG_LABEL_MAPPING_FULL: + ln = lde_nbr_find(imsg.hdr.peerid); + if (ln == NULL) { + log_debug("%s: cannot find lde neighbor", + __func__); + break; + } + + fec_snap(ln); + break; + case IMSG_LABEL_MAPPING: + case IMSG_LABEL_REQUEST: + case IMSG_LABEL_RELEASE: + case IMSG_LABEL_WITHDRAW: + case IMSG_LABEL_ABORT: + if (imsg.hdr.len - IMSG_HEADER_SIZE != sizeof(map)) + fatalx("lde_dispatch_imsg: wrong imsg len"); + memcpy(&map, imsg.data, sizeof(map)); + + ln = lde_nbr_find(imsg.hdr.peerid); + if (ln == NULL) { + log_debug("%s: cannot find lde neighbor", + __func__); + break; + } + + switch (imsg.hdr.type) { + case IMSG_LABEL_MAPPING: + lde_check_mapping(&map, ln); + break; + case IMSG_LABEL_REQUEST: + lde_check_request(&map, ln); + break; + case IMSG_LABEL_RELEASE: + if (map.type == MAP_TYPE_WILDCARD) + lde_check_release_wcard(&map, ln); + else + lde_check_release(&map, ln); + break; + case IMSG_LABEL_WITHDRAW: + if (map.type == MAP_TYPE_WILDCARD) + lde_check_withdraw_wcard(&map, ln); + else + lde_check_withdraw(&map, ln); + break; + case IMSG_LABEL_ABORT: + /* not necessary */ + break; + } + break; + case IMSG_ADDRESS_ADD: + if (imsg.hdr.len - IMSG_HEADER_SIZE != sizeof(lde_addr)) + fatalx("lde_dispatch_imsg: wrong imsg len"); + memcpy(&lde_addr, imsg.data, sizeof(lde_addr)); + + ln = lde_nbr_find(imsg.hdr.peerid); + if (ln == NULL) { + log_debug("%s: cannot find lde neighbor", + __func__); + break; + } + if (lde_address_add(ln, &lde_addr) < 0) { + log_debug("%s: cannot add address %s, it " + "already exists", __func__, + log_addr(lde_addr.af, &lde_addr.addr)); + } + break; + case IMSG_ADDRESS_DEL: + if (imsg.hdr.len - IMSG_HEADER_SIZE != sizeof(lde_addr)) + fatalx("lde_dispatch_imsg: wrong imsg len"); + memcpy(&lde_addr, imsg.data, sizeof(lde_addr)); + + ln = lde_nbr_find(imsg.hdr.peerid); + if (ln == NULL) { + log_debug("%s: cannot find lde neighbor", + __func__); + break; + } + if (lde_address_del(ln, &lde_addr) < 0) { + log_debug("%s: cannot delete address %s, it " + "does not exist", __func__, + log_addr(lde_addr.af, &lde_addr.addr)); + } + break; + case IMSG_NOTIFICATION: + if (imsg.hdr.len - IMSG_HEADER_SIZE != sizeof(nm)) + fatalx("lde_dispatch_imsg: wrong imsg len"); + memcpy(&nm, imsg.data, sizeof(nm)); + + ln = lde_nbr_find(imsg.hdr.peerid); + if (ln == NULL) { + log_debug("%s: cannot find lde neighbor", + __func__); + break; + } + + switch (nm.status_code) { + case S_PW_STATUS: + l2vpn_recv_pw_status(ln, &nm); + break; + default: + break; + } + break; + case IMSG_NEIGHBOR_UP: + if (imsg.hdr.len - IMSG_HEADER_SIZE != + sizeof(struct lde_nbr)) + fatalx("lde_dispatch_imsg: wrong imsg len"); + + if (lde_nbr_find(imsg.hdr.peerid)) + fatalx("lde_dispatch_imsg: " + "neighbor already exists"); + lde_nbr_new(imsg.hdr.peerid, imsg.data); + break; + case IMSG_NEIGHBOR_DOWN: + lde_nbr_del(lde_nbr_find(imsg.hdr.peerid)); + break; + case IMSG_CTL_SHOW_LIB: + rt_dump(imsg.hdr.pid); + + lde_imsg_compose_ldpe(IMSG_CTL_END, 0, + imsg.hdr.pid, NULL, 0); + break; + case IMSG_CTL_SHOW_L2VPN_PW: + l2vpn_pw_ctl(imsg.hdr.pid); + + lde_imsg_compose_ldpe(IMSG_CTL_END, 0, + imsg.hdr.pid, NULL, 0); + break; + case IMSG_CTL_SHOW_L2VPN_BINDING: + l2vpn_binding_ctl(imsg.hdr.pid); + + lde_imsg_compose_ldpe(IMSG_CTL_END, 0, + imsg.hdr.pid, NULL, 0); + break; + case IMSG_CTL_LOG_VERBOSE: + /* already checked by ldpe */ + memcpy(&verbose, imsg.data, sizeof(verbose)); + log_verbose(verbose); + break; + default: + log_debug("%s: unexpected imsg %d", __func__, + 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 */ +static void +lde_dispatch_parent(int fd, short event, void *bula) +{ + static struct ldpd_conf *nconf; + struct iface *niface; + struct tnbr *ntnbr; + struct nbr_params *nnbrp; + static struct l2vpn *nl2vpn; + struct l2vpn_if *nlif; + struct l2vpn_pw *npw; + struct imsg imsg; + struct kroute kr; + struct imsgev *iev = bula; + struct imsgbuf *ibuf = &iev->ibuf; + ssize_t n; + int shut = 0; + struct fec fec; + + if (event & EV_READ) { + if ((n = imsg_read(ibuf)) == -1 && errno != EAGAIN) + fatal("imsg_read error"); + if (n == 0) /* connection closed */ + shut = 1; + } + if (event & EV_WRITE) { + if ((n = msgbuf_write(&ibuf->w)) == -1 && errno != EAGAIN) + fatal("msgbuf_write"); + if (n == 0) /* connection closed */ + shut = 1; + } + + for (;;) { + if ((n = imsg_get(ibuf, &imsg)) == -1) + fatal("lde_dispatch_parent: imsg_get error"); + if (n == 0) + break; + + switch (imsg.hdr.type) { + case IMSG_NETWORK_ADD: + case IMSG_NETWORK_DEL: + if (imsg.hdr.len != IMSG_HEADER_SIZE + sizeof(kr)) { + log_warnx("%s: wrong imsg len", __func__); + break; + } + memcpy(&kr, imsg.data, sizeof(kr)); + + switch (kr.af) { + case AF_INET: + fec.type = FEC_TYPE_IPV4; + fec.u.ipv4.prefix = kr.prefix.v4; + fec.u.ipv4.prefixlen = kr.prefixlen; + break; + case AF_INET6: + fec.type = FEC_TYPE_IPV6; + fec.u.ipv6.prefix = kr.prefix.v6; + fec.u.ipv6.prefixlen = kr.prefixlen; + break; + default: + fatalx("lde_dispatch_parent: unknown af"); + } + + switch (imsg.hdr.type) { + case IMSG_NETWORK_ADD: + lde_kernel_insert(&fec, kr.af, &kr.nexthop, + kr.priority, kr.flags & F_CONNECTED, NULL); + break; + case IMSG_NETWORK_DEL: + lde_kernel_remove(&fec, kr.af, &kr.nexthop, + kr.priority); + break; + } + break; + case IMSG_SOCKET_IPC: + if (iev_ldpe) { + log_warnx("%s: received unexpected imsg fd " + "to ldpe", __func__); + break; + } + if ((fd = imsg.fd) == -1) { + log_warnx("%s: expected to receive imsg fd to " + "ldpe but didn't receive any", __func__); + break; + } + + if ((iev_ldpe = malloc(sizeof(struct imsgev))) == NULL) + fatal(NULL); + imsg_init(&iev_ldpe->ibuf, fd); + iev_ldpe->handler = lde_dispatch_imsg; + iev_ldpe->events = EV_READ; + event_set(&iev_ldpe->ev, iev_ldpe->ibuf.fd, + iev_ldpe->events, iev_ldpe->handler, iev_ldpe); + event_add(&iev_ldpe->ev, NULL); + break; + case IMSG_RECONF_CONF: + if ((nconf = malloc(sizeof(struct ldpd_conf))) == + NULL) + fatal(NULL); + memcpy(nconf, imsg.data, sizeof(struct ldpd_conf)); + + LIST_INIT(&nconf->iface_list); + LIST_INIT(&nconf->tnbr_list); + LIST_INIT(&nconf->nbrp_list); + LIST_INIT(&nconf->l2vpn_list); + break; + case IMSG_RECONF_IFACE: + if ((niface = malloc(sizeof(struct iface))) == NULL) + fatal(NULL); + memcpy(niface, imsg.data, sizeof(struct iface)); + + LIST_INIT(&niface->addr_list); + LIST_INIT(&niface->ipv4.adj_list); + LIST_INIT(&niface->ipv6.adj_list); + niface->ipv4.iface = niface; + niface->ipv6.iface = niface; + + LIST_INSERT_HEAD(&nconf->iface_list, niface, entry); + break; + case IMSG_RECONF_TNBR: + if ((ntnbr = malloc(sizeof(struct tnbr))) == NULL) + fatal(NULL); + memcpy(ntnbr, imsg.data, sizeof(struct tnbr)); + + LIST_INSERT_HEAD(&nconf->tnbr_list, ntnbr, entry); + break; + case IMSG_RECONF_NBRP: + if ((nnbrp = malloc(sizeof(struct nbr_params))) == NULL) + fatal(NULL); + memcpy(nnbrp, imsg.data, sizeof(struct nbr_params)); + + LIST_INSERT_HEAD(&nconf->nbrp_list, nnbrp, entry); + break; + case IMSG_RECONF_L2VPN: + if ((nl2vpn = malloc(sizeof(struct l2vpn))) == NULL) + fatal(NULL); + memcpy(nl2vpn, imsg.data, sizeof(struct l2vpn)); + + LIST_INIT(&nl2vpn->if_list); + LIST_INIT(&nl2vpn->pw_list); + + LIST_INSERT_HEAD(&nconf->l2vpn_list, nl2vpn, entry); + break; + case IMSG_RECONF_L2VPN_IF: + if ((nlif = malloc(sizeof(struct l2vpn_if))) == NULL) + fatal(NULL); + memcpy(nlif, imsg.data, sizeof(struct l2vpn_if)); + + nlif->l2vpn = nl2vpn; + LIST_INSERT_HEAD(&nl2vpn->if_list, nlif, entry); + break; + case IMSG_RECONF_L2VPN_PW: + if ((npw = malloc(sizeof(struct l2vpn_pw))) == NULL) + fatal(NULL); + memcpy(npw, imsg.data, sizeof(struct l2vpn_pw)); + + npw->l2vpn = nl2vpn; + LIST_INSERT_HEAD(&nl2vpn->pw_list, npw, entry); + break; + case IMSG_RECONF_END: + merge_config(ldeconf, nconf); + nconf = NULL; + break; + default: + log_debug("%s: unexpected imsg %d", __func__, + 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); + } +} + +uint32_t +lde_assign_label(void) +{ + static uint32_t label = MPLS_LABEL_RESERVED_MAX; + + /* XXX some checks needed */ + label++; + return (label); +} + +void +lde_send_change_klabel(struct fec_node *fn, struct fec_nh *fnh) +{ + struct kroute kr; + struct kpw kpw; + struct l2vpn_pw *pw; + + switch (fn->fec.type) { + case FEC_TYPE_IPV4: + memset(&kr, 0, sizeof(kr)); + kr.af = AF_INET; + kr.prefix.v4 = fn->fec.u.ipv4.prefix; + kr.prefixlen = fn->fec.u.ipv4.prefixlen; + kr.nexthop.v4 = fnh->nexthop.v4; + kr.local_label = fn->local_label; + kr.remote_label = fnh->remote_label; + kr.priority = fnh->priority; + + lde_imsg_compose_parent(IMSG_KLABEL_CHANGE, 0, &kr, + sizeof(kr)); + + if (fn->fec.u.ipv4.prefixlen == 32) + l2vpn_sync_pws(AF_INET, (union ldpd_addr *) + &fn->fec.u.ipv4.prefix); + break; + case FEC_TYPE_IPV6: + memset(&kr, 0, sizeof(kr)); + kr.af = AF_INET6; + kr.prefix.v6 = fn->fec.u.ipv6.prefix; + kr.prefixlen = fn->fec.u.ipv6.prefixlen; + kr.nexthop.v6 = fnh->nexthop.v6; + kr.local_label = fn->local_label; + kr.remote_label = fnh->remote_label; + kr.priority = fnh->priority; + + lde_imsg_compose_parent(IMSG_KLABEL_CHANGE, 0, &kr, + sizeof(kr)); + + if (fn->fec.u.ipv6.prefixlen == 128) + l2vpn_sync_pws(AF_INET6, (union ldpd_addr *) + &fn->fec.u.ipv6.prefix); + break; + case FEC_TYPE_PWID: + if (fn->local_label == NO_LABEL || + fnh->remote_label == NO_LABEL) + return; + + pw = (struct l2vpn_pw *) fn->data; + pw->flags |= F_PW_STATUS_UP; + + memset(&kpw, 0, sizeof(kpw)); + kpw.ifindex = pw->ifindex; + kpw.pw_type = fn->fec.u.pwid.type; + kpw.af = pw->af; + kpw.nexthop = pw->addr; + kpw.local_label = fn->local_label; + kpw.remote_label = fnh->remote_label; + kpw.flags = pw->flags; + + lde_imsg_compose_parent(IMSG_KPWLABEL_CHANGE, 0, &kpw, + sizeof(kpw)); + break; + } +} + +void +lde_send_delete_klabel(struct fec_node *fn, struct fec_nh *fnh) +{ + struct kroute kr; + struct kpw kpw; + struct l2vpn_pw *pw; + + switch (fn->fec.type) { + case FEC_TYPE_IPV4: + memset(&kr, 0, sizeof(kr)); + kr.af = AF_INET; + kr.prefix.v4 = fn->fec.u.ipv4.prefix; + kr.prefixlen = fn->fec.u.ipv4.prefixlen; + kr.nexthop.v4 = fnh->nexthop.v4; + kr.local_label = fn->local_label; + kr.remote_label = fnh->remote_label; + kr.priority = fnh->priority; + + lde_imsg_compose_parent(IMSG_KLABEL_DELETE, 0, &kr, + sizeof(kr)); + + if (fn->fec.u.ipv4.prefixlen == 32) + l2vpn_sync_pws(AF_INET, (union ldpd_addr *) + &fn->fec.u.ipv4.prefix); + break; + case FEC_TYPE_IPV6: + memset(&kr, 0, sizeof(kr)); + kr.af = AF_INET6; + kr.prefix.v6 = fn->fec.u.ipv6.prefix; + kr.prefixlen = fn->fec.u.ipv6.prefixlen; + kr.nexthop.v6 = fnh->nexthop.v6; + kr.local_label = fn->local_label; + kr.remote_label = fnh->remote_label; + kr.priority = fnh->priority; + + lde_imsg_compose_parent(IMSG_KLABEL_DELETE, 0, &kr, + sizeof(kr)); + + if (fn->fec.u.ipv6.prefixlen == 128) + l2vpn_sync_pws(AF_INET6, (union ldpd_addr *) + &fn->fec.u.ipv6.prefix); + break; + case FEC_TYPE_PWID: + pw = (struct l2vpn_pw *) fn->data; + if (!(pw->flags & F_PW_STATUS_UP)) + return; + pw->flags &= ~F_PW_STATUS_UP; + + memset(&kpw, 0, sizeof(kpw)); + kpw.ifindex = pw->ifindex; + kpw.pw_type = fn->fec.u.pwid.type; + kpw.af = pw->af; + kpw.nexthop = pw->addr; + kpw.local_label = fn->local_label; + kpw.remote_label = fnh->remote_label; + kpw.flags = pw->flags; + + lde_imsg_compose_parent(IMSG_KPWLABEL_DELETE, 0, &kpw, + sizeof(kpw)); + break; + } +} + +void +lde_fec2map(struct fec *fec, struct map *map) +{ + memset(map, 0, sizeof(*map)); + + switch (fec->type) { + case FEC_TYPE_IPV4: + map->type = MAP_TYPE_PREFIX; + map->fec.prefix.af = AF_INET; + map->fec.prefix.prefix.v4 = fec->u.ipv4.prefix; + map->fec.prefix.prefixlen = fec->u.ipv4.prefixlen; + break; + case FEC_TYPE_IPV6: + map->type = MAP_TYPE_PREFIX; + map->fec.prefix.af = AF_INET6; + map->fec.prefix.prefix.v6 = fec->u.ipv6.prefix; + map->fec.prefix.prefixlen = fec->u.ipv6.prefixlen; + break; + case FEC_TYPE_PWID: + map->type = MAP_TYPE_PWID; + map->fec.pwid.type = fec->u.pwid.type; + map->fec.pwid.group_id = 0; + map->flags |= F_MAP_PW_ID; + map->fec.pwid.pwid = fec->u.pwid.pwid; + break; + } +} + +void +lde_map2fec(struct map *map, struct in_addr lsr_id, struct fec *fec) +{ + memset(fec, 0, sizeof(*fec)); + + switch (map->type) { + case MAP_TYPE_PREFIX: + switch (map->fec.prefix.af) { + case AF_INET: + fec->type = FEC_TYPE_IPV4; + fec->u.ipv4.prefix = map->fec.prefix.prefix.v4; + fec->u.ipv4.prefixlen = map->fec.prefix.prefixlen; + break; + case AF_INET6: + fec->type = FEC_TYPE_IPV6; + fec->u.ipv6.prefix = map->fec.prefix.prefix.v6; + fec->u.ipv6.prefixlen = map->fec.prefix.prefixlen; + break; + default: + fatalx("lde_map2fec: unknown af"); + break; + } + break; + case MAP_TYPE_PWID: + fec->type = FEC_TYPE_PWID; + fec->u.pwid.type = map->fec.pwid.type; + fec->u.pwid.pwid = map->fec.pwid.pwid; + fec->u.pwid.lsr_id = lsr_id; + break; + } +} + +void +lde_send_labelmapping(struct lde_nbr *ln, struct fec_node *fn, int single) +{ + struct lde_req *lre; + struct lde_map *me; + struct map map; + struct l2vpn_pw *pw; + + /* + * This function skips SL.1 - 3 and SL.9 - 14 because the label + * allocation is done way earlier (because of the merging nature of + * ldpd). + */ + + lde_fec2map(&fn->fec, &map); + switch (fn->fec.type) { + case FEC_TYPE_IPV4: + if (!ln->v4_enabled) + return; + break; + case FEC_TYPE_IPV6: + if (!ln->v6_enabled) + return; + break; + case FEC_TYPE_PWID: + pw = (struct l2vpn_pw *) fn->data; + if (pw == NULL || pw->lsr_id.s_addr != ln->id.s_addr) + /* not the remote end of the pseudowire */ + return; + + map.flags |= F_MAP_PW_IFMTU; + map.fec.pwid.ifmtu = pw->l2vpn->mtu; + if (pw->flags & F_PW_CWORD) + map.flags |= F_MAP_PW_CWORD; + if (pw->flags & F_PW_STATUSTLV) { + map.flags |= F_MAP_PW_STATUS; + /* VPLS are always up */ + map.pw_status = PW_FORWARDING; + } + break; + } + map.label = fn->local_label; + + /* SL.6: is there a pending request for this mapping? */ + lre = (struct lde_req *)fec_find(&ln->recv_req, &fn->fec); + if (lre) { + /* set label request msg id in the mapping response. */ + map.requestid = lre->msg_id; + map.flags = F_MAP_REQ_ID; + + /* SL.7: delete record of pending request */ + lde_req_del(ln, lre, 0); + } + + /* SL.4: send label mapping */ + lde_imsg_compose_ldpe(IMSG_MAPPING_ADD, ln->peerid, 0, + &map, sizeof(map)); + if (single) + lde_imsg_compose_ldpe(IMSG_MAPPING_ADD_END, ln->peerid, 0, + NULL, 0); + + /* SL.5: record sent label mapping */ + me = (struct lde_map *)fec_find(&ln->sent_map, &fn->fec); + if (me == NULL) + me = lde_map_add(ln, fn, 1); + me->map = map; +} + +void +lde_send_labelwithdraw(struct lde_nbr *ln, struct fec_node *fn, uint32_t label, + struct status_tlv *st) +{ + struct lde_wdraw *lw; + struct map map; + struct fec *f; + struct l2vpn_pw *pw; + + if (fn) { + lde_fec2map(&fn->fec, &map); + switch (fn->fec.type) { + case FEC_TYPE_IPV4: + if (!ln->v4_enabled) + return; + break; + case FEC_TYPE_IPV6: + if (!ln->v6_enabled) + return; + break; + case FEC_TYPE_PWID: + pw = (struct l2vpn_pw *) fn->data; + if (pw == NULL || pw->lsr_id.s_addr != ln->id.s_addr) + /* not the remote end of the pseudowire */ + return; + + if (pw->flags & F_PW_CWORD) + map.flags |= F_MAP_PW_CWORD; + break; + } + map.label = fn->local_label; + } else { + memset(&map, 0, sizeof(map)); + map.type = MAP_TYPE_WILDCARD; + map.label = label; + } + + if (st) { + map.st.status_code = st->status_code; + map.st.msg_id = st->msg_id; + map.st.msg_type = st->msg_type; + map.flags |= F_MAP_STATUS; + } + + /* SWd.1: send label withdraw. */ + lde_imsg_compose_ldpe(IMSG_WITHDRAW_ADD, ln->peerid, 0, + &map, sizeof(map)); + lde_imsg_compose_ldpe(IMSG_WITHDRAW_ADD_END, ln->peerid, 0, NULL, 0); + + /* SWd.2: record label withdraw. */ + if (fn) { + lw = (struct lde_wdraw *)fec_find(&ln->sent_wdraw, &fn->fec); + if (lw == NULL) + lw = lde_wdraw_add(ln, fn); + lw->label = map.label; + } else { + RB_FOREACH(f, fec_tree, &ft) { + fn = (struct fec_node *)f; + + lw = (struct lde_wdraw *)fec_find(&ln->sent_wdraw, + &fn->fec); + if (lw == NULL) + lw = lde_wdraw_add(ln, fn); + lw->label = map.label; + } + } +} + +void +lde_send_labelwithdraw_all(struct fec_node *fn, uint32_t label) +{ + struct lde_nbr *ln; + + RB_FOREACH(ln, nbr_tree, &lde_nbrs) + lde_send_labelwithdraw(ln, fn, label, NULL); +} + +void +lde_send_labelrelease(struct lde_nbr *ln, struct fec_node *fn, uint32_t label) +{ + struct map map; + struct l2vpn_pw *pw; + + if (fn) { + lde_fec2map(&fn->fec, &map); + switch (fn->fec.type) { + case FEC_TYPE_IPV4: + if (!ln->v4_enabled) + return; + break; + case FEC_TYPE_IPV6: + if (!ln->v6_enabled) + return; + break; + case FEC_TYPE_PWID: + pw = (struct l2vpn_pw *) fn->data; + if (pw == NULL || pw->lsr_id.s_addr != ln->id.s_addr) + /* not the remote end of the pseudowire */ + return; + + if (pw->flags & F_PW_CWORD) + map.flags |= F_MAP_PW_CWORD; + break; + } + } else { + memset(&map, 0, sizeof(map)); + map.type = MAP_TYPE_WILDCARD; + } + map.label = label; + + lde_imsg_compose_ldpe(IMSG_RELEASE_ADD, ln->peerid, 0, + &map, sizeof(map)); + lde_imsg_compose_ldpe(IMSG_RELEASE_ADD_END, ln->peerid, 0, NULL, 0); +} + +void +lde_send_notification(uint32_t peerid, uint32_t status_code, uint32_t msg_id, + uint16_t msg_type) +{ + struct notify_msg nm; + + memset(&nm, 0, sizeof(nm)); + nm.status_code = status_code; + /* 'msg_id' and 'msg_type' should be in network byte order */ + nm.msg_id = msg_id; + nm.msg_type = msg_type; + + lde_imsg_compose_ldpe(IMSG_NOTIFICATION_SEND, peerid, 0, + &nm, sizeof(nm)); +} + +static __inline int +lde_nbr_compare(struct lde_nbr *a, struct lde_nbr *b) +{ + return (a->peerid - b->peerid); +} + +static struct lde_nbr * +lde_nbr_new(uint32_t peerid, struct lde_nbr *new) +{ + struct lde_nbr *ln; + + if ((ln = calloc(1, sizeof(*ln))) == NULL) + fatal(__func__); + + ln->id = new->id; + ln->v4_enabled = new->v4_enabled; + ln->v6_enabled = new->v6_enabled; + ln->peerid = peerid; + fec_init(&ln->recv_map); + fec_init(&ln->sent_map); + fec_init(&ln->recv_req); + fec_init(&ln->sent_req); + fec_init(&ln->sent_wdraw); + + TAILQ_INIT(&ln->addr_list); + + if (RB_INSERT(nbr_tree, &lde_nbrs, ln) != NULL) + fatalx("lde_nbr_new: RB_INSERT failed"); + + return (ln); +} + +static void +lde_nbr_del(struct lde_nbr *ln) +{ + struct fec *f; + struct fec_node *fn; + struct fec_nh *fnh; + struct l2vpn_pw *pw; + + if (ln == NULL) + return; + + /* uninstall received mappings */ + RB_FOREACH(f, fec_tree, &ft) { + fn = (struct fec_node *)f; + + LIST_FOREACH(fnh, &fn->nexthops, entry) { + switch (f->type) { + case FEC_TYPE_IPV4: + case FEC_TYPE_IPV6: + if (!lde_address_find(ln, fnh->af, + &fnh->nexthop)) + continue; + break; + case FEC_TYPE_PWID: + if (f->u.pwid.lsr_id.s_addr != ln->id.s_addr) + continue; + pw = (struct l2vpn_pw *) fn->data; + if (pw) + l2vpn_pw_reset(pw); + break; + default: + break; + } + + lde_send_delete_klabel(fn, fnh); + fnh->remote_label = NO_LABEL; + } + } + + lde_address_list_free(ln); + + fec_clear(&ln->recv_map, lde_map_free); + fec_clear(&ln->sent_map, lde_map_free); + fec_clear(&ln->recv_req, free); + fec_clear(&ln->sent_req, free); + fec_clear(&ln->sent_wdraw, free); + + RB_REMOVE(nbr_tree, &lde_nbrs, ln); + + free(ln); +} + +static struct lde_nbr * +lde_nbr_find(uint32_t peerid) +{ + struct lde_nbr ln; + + ln.peerid = peerid; + + return (RB_FIND(nbr_tree, &lde_nbrs, &ln)); +} + +struct lde_nbr * +lde_nbr_find_by_lsrid(struct in_addr addr) +{ + struct lde_nbr *ln; + + RB_FOREACH(ln, nbr_tree, &lde_nbrs) + if (ln->id.s_addr == addr.s_addr) + return (ln); + + return (NULL); +} + +struct lde_nbr * +lde_nbr_find_by_addr(int af, union ldpd_addr *addr) +{ + struct lde_nbr *ln; + + RB_FOREACH(ln, nbr_tree, &lde_nbrs) + if (lde_address_find(ln, af, addr) != NULL) + return (ln); + + return (NULL); +} + +static void +lde_nbr_clear(void) +{ + struct lde_nbr *ln; + + while ((ln = RB_ROOT(&lde_nbrs)) != NULL) + lde_nbr_del(ln); +} + +static void +lde_nbr_addr_update(struct lde_nbr *ln, struct lde_addr *lde_addr, int removed) +{ + struct fec *fec; + struct fec_node *fn; + struct fec_nh *fnh; + struct lde_map *me; + + RB_FOREACH(fec, fec_tree, &ln->recv_map) { + fn = (struct fec_node *)fec_find(&ft, fec); + switch (fec->type) { + case FEC_TYPE_IPV4: + if (lde_addr->af != AF_INET) + continue; + break; + case FEC_TYPE_IPV6: + if (lde_addr->af != AF_INET6) + continue; + break; + default: + continue; + } + + LIST_FOREACH(fnh, &fn->nexthops, entry) { + if (ldp_addrcmp(fnh->af, &fnh->nexthop, + &lde_addr->addr)) + continue; + + if (removed) { + lde_send_delete_klabel(fn, fnh); + fnh->remote_label = NO_LABEL; + } else { + me = (struct lde_map *)fec; + fnh->remote_label = me->map.label; + lde_send_change_klabel(fn, fnh); + } + break; + } + } +} + +struct lde_map * +lde_map_add(struct lde_nbr *ln, struct fec_node *fn, int sent) +{ + struct lde_map *me; + + me = calloc(1, sizeof(*me)); + if (me == NULL) + fatal(__func__); + + me->fec = fn->fec; + me->nexthop = ln; + + if (sent) { + LIST_INSERT_HEAD(&fn->upstream, me, entry); + if (fec_insert(&ln->sent_map, &me->fec)) + log_warnx("failed to add %s to sent map", + log_fec(&me->fec)); + /* XXX on failure more cleanup is needed */ + } else { + LIST_INSERT_HEAD(&fn->downstream, me, entry); + if (fec_insert(&ln->recv_map, &me->fec)) + log_warnx("failed to add %s to recv map", + log_fec(&me->fec)); + } + + return (me); +} + +void +lde_map_del(struct lde_nbr *ln, struct lde_map *me, int sent) +{ + if (sent) + fec_remove(&ln->sent_map, &me->fec); + else + fec_remove(&ln->recv_map, &me->fec); + + lde_map_free(me); +} + +static void +lde_map_free(void *ptr) +{ + struct lde_map *map = ptr; + + LIST_REMOVE(map, entry); + free(map); +} + +struct lde_req * +lde_req_add(struct lde_nbr *ln, struct fec *fec, int sent) +{ + struct fec_tree *t; + struct lde_req *lre; + + t = sent ? &ln->sent_req : &ln->recv_req; + + lre = calloc(1, sizeof(*lre)); + if (lre != NULL) { + lre->fec = *fec; + + if (fec_insert(t, &lre->fec)) { + log_warnx("failed to add %s to %s req", + log_fec(&lre->fec), sent ? "sent" : "recv"); + free(lre); + return (NULL); + } + } + + return (lre); +} + +void +lde_req_del(struct lde_nbr *ln, struct lde_req *lre, int sent) +{ + if (sent) + fec_remove(&ln->sent_req, &lre->fec); + else + fec_remove(&ln->recv_req, &lre->fec); + + free(lre); +} + +struct lde_wdraw * +lde_wdraw_add(struct lde_nbr *ln, struct fec_node *fn) +{ + struct lde_wdraw *lw; + + lw = calloc(1, sizeof(*lw)); + if (lw == NULL) + fatal(__func__); + + lw->fec = fn->fec; + + if (fec_insert(&ln->sent_wdraw, &lw->fec)) + log_warnx("failed to add %s to sent wdraw", + log_fec(&lw->fec)); + + return (lw); +} + +void +lde_wdraw_del(struct lde_nbr *ln, struct lde_wdraw *lw) +{ + fec_remove(&ln->sent_wdraw, &lw->fec); + free(lw); +} + +void +lde_change_egress_label(int af, int was_implicit) +{ + struct lde_nbr *ln; + struct fec *f; + struct fec_node *fn; + + RB_FOREACH(ln, nbr_tree, &lde_nbrs) { + /* explicit withdraw */ + if (was_implicit) + lde_send_labelwithdraw(ln, NULL, MPLS_LABEL_IMPLNULL, + NULL); + else { + if (ln->v4_enabled) + lde_send_labelwithdraw(ln, NULL, + MPLS_LABEL_IPV4NULL, NULL); + if (ln->v6_enabled) + lde_send_labelwithdraw(ln, NULL, + MPLS_LABEL_IPV6NULL, NULL); + } + + /* advertise new label of connected prefixes */ + RB_FOREACH(f, fec_tree, &ft) { + fn = (struct fec_node *)f; + if (fn->local_label > MPLS_LABEL_RESERVED_MAX) + continue; + + switch (af) { + case AF_INET: + if (fn->fec.type != FEC_TYPE_IPV4) + continue; + break; + case AF_INET6: + if (fn->fec.type != FEC_TYPE_IPV6) + continue; + break; + default: + fatalx("lde_change_egress_label: unknown af"); + } + + fn->local_label = egress_label(fn->fec.type); + lde_send_labelmapping(ln, fn, 0); + } + + lde_imsg_compose_ldpe(IMSG_MAPPING_ADD_END, ln->peerid, 0, + NULL, 0); + } +} + +static int +lde_address_add(struct lde_nbr *ln, struct lde_addr *lde_addr) +{ + struct lde_addr *new; + + if (lde_address_find(ln, lde_addr->af, &lde_addr->addr) != NULL) + return (-1); + + if ((new = calloc(1, sizeof(*new))) == NULL) + fatal(__func__); + + new->af = lde_addr->af; + new->addr = lde_addr->addr; + TAILQ_INSERT_TAIL(&ln->addr_list, new, entry); + + /* reevaluate the previously received mappings from this neighbor */ + lde_nbr_addr_update(ln, lde_addr, 0); + + return (0); +} + +static int +lde_address_del(struct lde_nbr *ln, struct lde_addr *lde_addr) +{ + lde_addr = lde_address_find(ln, lde_addr->af, &lde_addr->addr); + if (lde_addr == NULL) + return (-1); + + /* reevaluate the previously received mappings from this neighbor */ + lde_nbr_addr_update(ln, lde_addr, 1); + + TAILQ_REMOVE(&ln->addr_list, lde_addr, entry); + free(lde_addr); + + return (0); +} + +struct lde_addr * +lde_address_find(struct lde_nbr *ln, int af, union ldpd_addr *addr) +{ + struct lde_addr *lde_addr; + + TAILQ_FOREACH(lde_addr, &ln->addr_list, entry) + if (lde_addr->af == af && + ldp_addrcmp(af, &lde_addr->addr, addr) == 0) + return (lde_addr); + + return (NULL); +} + +static void +lde_address_list_free(struct lde_nbr *ln) +{ + struct lde_addr *lde_addr; + + while ((lde_addr = TAILQ_FIRST(&ln->addr_list)) != NULL) { + TAILQ_REMOVE(&ln->addr_list, lde_addr, entry); + free(lde_addr); + } +} diff --git a/ldpd/lde.h b/ldpd/lde.h new file mode 100644 index 000000000..b0f7b2043 --- /dev/null +++ b/ldpd/lde.h @@ -0,0 +1,202 @@ +/* $OpenBSD$ */ + +/* + * Copyright (c) 2013, 2016 Renato Westphal + * Copyright (c) 2009 Michele Marchetto + * Copyright (c) 2004, 2005 Esben Norby + * + * 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. + */ + +#ifndef _LDE_H_ +#define _LDE_H_ + +#include +#include +#include + +enum fec_type { + FEC_TYPE_IPV4, + FEC_TYPE_IPV6, + FEC_TYPE_PWID +}; + +struct fec { + RB_ENTRY(fec) entry; + enum fec_type type; + union { + struct { + struct in_addr prefix; + uint8_t prefixlen; + } ipv4; + struct { + struct in6_addr prefix; + uint8_t prefixlen; + } ipv6; + struct { + uint16_t type; + uint32_t pwid; + struct in_addr lsr_id; + } pwid; + } u; +}; +RB_HEAD(fec_tree, fec); +RB_PROTOTYPE(fec_tree, fec, entry, fec_compare) + +/* request entries */ +struct lde_req { + struct fec fec; + uint32_t msg_id; +}; + +/* mapping entries */ +struct lde_map { + struct fec fec; + LIST_ENTRY(lde_map) entry; + struct lde_nbr *nexthop; + struct map map; +}; + +/* withdraw entries */ +struct lde_wdraw { + struct fec fec; + uint32_t label; +}; + +/* Addresses belonging to neighbor */ +struct lde_addr { + TAILQ_ENTRY(lde_addr) entry; + int af; + union ldpd_addr addr; +}; + +/* just the info LDE needs */ +struct lde_nbr { + RB_ENTRY(lde_nbr) entry; + uint32_t peerid; + struct in_addr id; + int v4_enabled; /* announce/process v4 msgs */ + int v6_enabled; /* announce/process v6 msgs */ + struct fec_tree recv_req; + struct fec_tree sent_req; + struct fec_tree recv_map; + struct fec_tree sent_map; + struct fec_tree sent_wdraw; + TAILQ_HEAD(, lde_addr) addr_list; +}; +RB_HEAD(nbr_tree, lde_nbr); +RB_PROTOTYPE(nbr_tree, lde_nbr, entry, lde_nbr_compare) + +struct fec_nh { + LIST_ENTRY(fec_nh) entry; + int af; + union ldpd_addr nexthop; + uint32_t remote_label; + uint8_t priority; +}; + +struct fec_node { + struct fec fec; + + LIST_HEAD(, fec_nh) nexthops; /* fib nexthops */ + LIST_HEAD(, lde_map) downstream; /* recv mappings */ + LIST_HEAD(, lde_map) upstream; /* sent mappings */ + + uint32_t local_label; + void *data; /* fec specific data */ +}; + +#define LDE_GC_INTERVAL 300 + +extern struct ldpd_conf *ldeconf; +extern struct fec_tree ft; +extern struct nbr_tree lde_nbrs; +extern struct event gc_timer; + +/* lde.c */ +void lde(int, int); +int lde_imsg_compose_ldpe(int, uint32_t, pid_t, void *, uint16_t); +uint32_t lde_assign_label(void); +void lde_send_change_klabel(struct fec_node *, struct fec_nh *); +void lde_send_delete_klabel(struct fec_node *, struct fec_nh *); +void lde_fec2map(struct fec *, struct map *); +void lde_map2fec(struct map *, struct in_addr, struct fec *); +void lde_send_labelmapping(struct lde_nbr *, struct fec_node *, + int); +void lde_send_labelwithdraw(struct lde_nbr *, struct fec_node *, + uint32_t, struct status_tlv *); +void lde_send_labelwithdraw_all(struct fec_node *, uint32_t); +void lde_send_labelrelease(struct lde_nbr *, struct fec_node *, + uint32_t); +void lde_send_notification(uint32_t, uint32_t, uint32_t, uint16_t); +struct lde_nbr *lde_nbr_find_by_lsrid(struct in_addr); +struct lde_nbr *lde_nbr_find_by_addr(int, union ldpd_addr *); +struct lde_map *lde_map_add(struct lde_nbr *, struct fec_node *, int); +void lde_map_del(struct lde_nbr *, struct lde_map *, int); +struct lde_req *lde_req_add(struct lde_nbr *, struct fec *, int); +void lde_req_del(struct lde_nbr *, struct lde_req *, int); +struct lde_wdraw *lde_wdraw_add(struct lde_nbr *, struct fec_node *); +void lde_wdraw_del(struct lde_nbr *, struct lde_wdraw *); +void lde_change_egress_label(int, int); +struct lde_addr *lde_address_find(struct lde_nbr *, int, + union ldpd_addr *); + +/* lde_lib.c */ +void fec_init(struct fec_tree *); +struct fec *fec_find(struct fec_tree *, struct fec *); +int fec_insert(struct fec_tree *, struct fec *); +int fec_remove(struct fec_tree *, struct fec *); +void fec_clear(struct fec_tree *, void (*)(void *)); +void rt_dump(pid_t); +void fec_snap(struct lde_nbr *); +void fec_tree_clear(void); +struct fec_nh *fec_nh_find(struct fec_node *, int, union ldpd_addr *, + uint8_t); +uint32_t egress_label(enum fec_type); +void lde_kernel_insert(struct fec *, int, union ldpd_addr *, + uint8_t, int, void *); +void lde_kernel_remove(struct fec *, int, union ldpd_addr *, + uint8_t); +void lde_check_mapping(struct map *, struct lde_nbr *); +void lde_check_request(struct map *, struct lde_nbr *); +void lde_check_release(struct map *, struct lde_nbr *); +void lde_check_release_wcard(struct map *, struct lde_nbr *); +void lde_check_withdraw(struct map *, struct lde_nbr *); +void lde_check_withdraw_wcard(struct map *, struct lde_nbr *); +void lde_gc_timer(int, short, void *); +void lde_gc_start_timer(void); +void lde_gc_stop_timer(void); + +/* l2vpn.c */ +struct l2vpn *l2vpn_new(const char *); +struct l2vpn *l2vpn_find(struct ldpd_conf *, const char *); +void l2vpn_del(struct l2vpn *); +void l2vpn_init(struct l2vpn *); +void l2vpn_exit(struct l2vpn *); +struct l2vpn_if *l2vpn_if_new(struct l2vpn *, struct kif *); +struct l2vpn_if *l2vpn_if_find(struct l2vpn *, unsigned int); +struct l2vpn_pw *l2vpn_pw_new(struct l2vpn *, struct kif *); +struct l2vpn_pw *l2vpn_pw_find(struct l2vpn *, unsigned int); +void l2vpn_pw_init(struct l2vpn_pw *); +void l2vpn_pw_exit(struct l2vpn_pw *); +void l2vpn_pw_reset(struct l2vpn_pw *); +int l2vpn_pw_ok(struct l2vpn_pw *, struct fec_nh *); +int l2vpn_pw_negotiate(struct lde_nbr *, struct fec_node *, + struct map *); +void l2vpn_send_pw_status(uint32_t, uint32_t, struct fec *); +void l2vpn_recv_pw_status(struct lde_nbr *, struct notify_msg *); +void l2vpn_sync_pws(int, union ldpd_addr *); +void l2vpn_pw_ctl(pid_t); +void l2vpn_binding_ctl(pid_t); + +#endif /* _LDE_H_ */ diff --git a/ldpd/lde_lib.c b/ldpd/lde_lib.c new file mode 100644 index 000000000..d9c1f544f --- /dev/null +++ b/ldpd/lde_lib.c @@ -0,0 +1,784 @@ +/* $OpenBSD$ */ + +/* + * Copyright (c) 2013, 2016 Renato Westphal + * Copyright (c) 2009 Michele Marchetto + * + * 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 "ldpd.h" +#include "lde.h" +#include "log.h" + +static __inline int fec_compare(struct fec *, struct fec *); +static int lde_nbr_is_nexthop(struct fec_node *, + struct lde_nbr *); +static void fec_free(void *); +static struct fec_node *fec_add(struct fec *fec); +static struct fec_nh *fec_nh_add(struct fec_node *, int, union ldpd_addr *, + uint8_t priority); +static void fec_nh_del(struct fec_nh *); + +RB_GENERATE(fec_tree, fec, entry, fec_compare) + +struct fec_tree ft = RB_INITIALIZER(&ft); +struct event gc_timer; + +/* FEC tree functions */ +void +fec_init(struct fec_tree *fh) +{ + RB_INIT(fh); +} + +static __inline int +fec_compare(struct fec *a, struct fec *b) +{ + if (a->type < b->type) + return (-1); + if (a->type > b->type) + return (1); + + switch (a->type) { + case FEC_TYPE_IPV4: + if (ntohl(a->u.ipv4.prefix.s_addr) < + ntohl(b->u.ipv4.prefix.s_addr)) + return (-1); + if (ntohl(a->u.ipv4.prefix.s_addr) > + ntohl(b->u.ipv4.prefix.s_addr)) + return (1); + if (a->u.ipv4.prefixlen < b->u.ipv4.prefixlen) + return (-1); + if (a->u.ipv4.prefixlen > b->u.ipv4.prefixlen) + return (1); + return (0); + case FEC_TYPE_IPV6: + if (memcmp(&a->u.ipv6.prefix, &b->u.ipv6.prefix, + sizeof(struct in6_addr)) < 0) + return (-1); + if (memcmp(&a->u.ipv6.prefix, &b->u.ipv6.prefix, + sizeof(struct in6_addr)) > 0) + return (1); + if (a->u.ipv6.prefixlen < b->u.ipv6.prefixlen) + return (-1); + if (a->u.ipv6.prefixlen > b->u.ipv6.prefixlen) + return (1); + return (0); + case FEC_TYPE_PWID: + if (a->u.pwid.type < b->u.pwid.type) + return (-1); + if (a->u.pwid.type > b->u.pwid.type) + return (1); + if (a->u.pwid.pwid < b->u.pwid.pwid) + return (-1); + if (a->u.pwid.pwid > b->u.pwid.pwid) + return (1); + if (ntohl(a->u.pwid.lsr_id.s_addr) < + ntohl(b->u.pwid.lsr_id.s_addr)) + return (-1); + if (ntohl(a->u.pwid.lsr_id.s_addr) > + ntohl(b->u.pwid.lsr_id.s_addr)) + return (1); + return (0); + } + + return (-1); +} + +struct fec * +fec_find(struct fec_tree *fh, struct fec *f) +{ + return (RB_FIND(fec_tree, fh, f)); +} + +int +fec_insert(struct fec_tree *fh, struct fec *f) +{ + if (RB_INSERT(fec_tree, fh, f) != NULL) + return (-1); + return (0); +} + +int +fec_remove(struct fec_tree *fh, struct fec *f) +{ + if (RB_REMOVE(fec_tree, fh, f) == NULL) { + log_warnx("%s failed for %s", __func__, log_fec(f)); + return (-1); + } + return (0); +} + +void +fec_clear(struct fec_tree *fh, void (*free_cb)(void *)) +{ + struct fec *f; + + while ((f = RB_ROOT(fh)) != NULL) { + fec_remove(fh, f); + free_cb(f); + } +} + +/* routing table functions */ +static int +lde_nbr_is_nexthop(struct fec_node *fn, struct lde_nbr *ln) +{ + struct fec_nh *fnh; + + LIST_FOREACH(fnh, &fn->nexthops, entry) + if (lde_address_find(ln, fnh->af, &fnh->nexthop)) + return (1); + + return (0); +} + +void +rt_dump(pid_t pid) +{ + struct fec *f; + struct fec_node *fn; + struct lde_map *me; + static struct ctl_rt rtctl; + + RB_FOREACH(f, fec_tree, &ft) { + fn = (struct fec_node *)f; + if (fn->local_label == NO_LABEL && + LIST_EMPTY(&fn->downstream)) + continue; + + switch (fn->fec.type) { + case FEC_TYPE_IPV4: + rtctl.af = AF_INET; + rtctl.prefix.v4 = fn->fec.u.ipv4.prefix; + rtctl.prefixlen = fn->fec.u.ipv4.prefixlen; + break; + case FEC_TYPE_IPV6: + rtctl.af = AF_INET6; + rtctl.prefix.v6 = fn->fec.u.ipv6.prefix; + rtctl.prefixlen = fn->fec.u.ipv6.prefixlen; + break; + default: + continue; + } + + rtctl.local_label = fn->local_label; + LIST_FOREACH(me, &fn->downstream, entry) { + rtctl.in_use = lde_nbr_is_nexthop(fn, me->nexthop); + rtctl.nexthop = me->nexthop->id; + rtctl.remote_label = me->map.label; + + lde_imsg_compose_ldpe(IMSG_CTL_SHOW_LIB, 0, pid, + &rtctl, sizeof(rtctl)); + } + if (LIST_EMPTY(&fn->downstream)) { + rtctl.in_use = 0; + rtctl.nexthop.s_addr = INADDR_ANY; + rtctl.remote_label = NO_LABEL; + + lde_imsg_compose_ldpe(IMSG_CTL_SHOW_LIB, 0, pid, + &rtctl, sizeof(rtctl)); + } + } +} + +void +fec_snap(struct lde_nbr *ln) +{ + struct fec *f; + struct fec_node *fn; + + RB_FOREACH(f, fec_tree, &ft) { + fn = (struct fec_node *)f; + if (fn->local_label == NO_LABEL) + continue; + + lde_send_labelmapping(ln, fn, 0); + } + + lde_imsg_compose_ldpe(IMSG_MAPPING_ADD_END, ln->peerid, 0, NULL, 0); +} + +static void +fec_free(void *arg) +{ + struct fec_node *fn = arg; + struct fec_nh *fnh; + + while ((fnh = LIST_FIRST(&fn->nexthops))) + fec_nh_del(fnh); + if (!LIST_EMPTY(&fn->downstream)) + log_warnx("%s: fec %s downstream list not empty", __func__, + log_fec(&fn->fec)); + if (!LIST_EMPTY(&fn->upstream)) + log_warnx("%s: fec %s upstream list not empty", __func__, + log_fec(&fn->fec)); + + free(fn); +} + +void +fec_tree_clear(void) +{ + fec_clear(&ft, fec_free); +} + +static struct fec_node * +fec_add(struct fec *fec) +{ + struct fec_node *fn; + + fn = calloc(1, sizeof(*fn)); + if (fn == NULL) + fatal(__func__); + + fn->fec = *fec; + fn->local_label = NO_LABEL; + LIST_INIT(&fn->upstream); + LIST_INIT(&fn->downstream); + LIST_INIT(&fn->nexthops); + + if (fec_insert(&ft, &fn->fec)) + log_warnx("failed to add %s to ft tree", + log_fec(&fn->fec)); + + return (fn); +} + +struct fec_nh * +fec_nh_find(struct fec_node *fn, int af, union ldpd_addr *nexthop, + uint8_t priority) +{ + struct fec_nh *fnh; + + LIST_FOREACH(fnh, &fn->nexthops, entry) + if (fnh->af == af && + ldp_addrcmp(af, &fnh->nexthop, nexthop) == 0 && + fnh->priority == priority) + return (fnh); + + return (NULL); +} + +static struct fec_nh * +fec_nh_add(struct fec_node *fn, int af, union ldpd_addr *nexthop, + uint8_t priority) +{ + struct fec_nh *fnh; + + fnh = calloc(1, sizeof(*fnh)); + if (fnh == NULL) + fatal(__func__); + + fnh->af = af; + fnh->nexthop = *nexthop; + fnh->remote_label = NO_LABEL; + fnh->priority = priority; + LIST_INSERT_HEAD(&fn->nexthops, fnh, entry); + + return (fnh); +} + +static void +fec_nh_del(struct fec_nh *fnh) +{ + LIST_REMOVE(fnh, entry); + free(fnh); +} + +uint32_t +egress_label(enum fec_type fec_type) +{ + switch (fec_type) { + case FEC_TYPE_IPV4: + if (ldeconf->ipv4.flags & F_LDPD_AF_EXPNULL) + return (MPLS_LABEL_IPV4NULL); + break; + case FEC_TYPE_IPV6: + if (ldeconf->ipv6.flags & F_LDPD_AF_EXPNULL) + return (MPLS_LABEL_IPV6NULL); + break; + default: + fatalx("egress_label: unexpected fec type"); + } + + return (MPLS_LABEL_IMPLNULL); +} + +void +lde_kernel_insert(struct fec *fec, int af, union ldpd_addr *nexthop, + uint8_t priority, int connected, void *data) +{ + struct fec_node *fn; + struct fec_nh *fnh; + struct lde_map *me; + struct lde_nbr *ln; + + fn = (struct fec_node *)fec_find(&ft, fec); + if (fn == NULL) + fn = fec_add(fec); + if (fec_nh_find(fn, af, nexthop, priority) != NULL) + return; + + log_debug("lde add fec %s nexthop %s", + log_fec(&fn->fec), log_addr(af, nexthop)); + + if (fn->fec.type == FEC_TYPE_PWID) + fn->data = data; + + if (fn->local_label == NO_LABEL) { + if (connected) + fn->local_label = egress_label(fn->fec.type); + else + fn->local_label = lde_assign_label(); + + /* FEC.1: perform lsr label distribution procedure */ + RB_FOREACH(ln, nbr_tree, &lde_nbrs) + lde_send_labelmapping(ln, fn, 1); + } + + fnh = fec_nh_add(fn, af, nexthop, priority); + lde_send_change_klabel(fn, fnh); + + switch (fn->fec.type) { + case FEC_TYPE_IPV4: + case FEC_TYPE_IPV6: + ln = lde_nbr_find_by_addr(af, &fnh->nexthop); + break; + case FEC_TYPE_PWID: + ln = lde_nbr_find_by_lsrid(fn->fec.u.pwid.lsr_id); + break; + default: + ln = NULL; + break; + } + + if (ln) { + /* FEC.2 */ + me = (struct lde_map *)fec_find(&ln->recv_map, &fn->fec); + if (me) + /* FEC.5 */ + lde_check_mapping(&me->map, ln); + } +} + +void +lde_kernel_remove(struct fec *fec, int af, union ldpd_addr *nexthop, + uint8_t priority) +{ + struct fec_node *fn; + struct fec_nh *fnh; + + fn = (struct fec_node *)fec_find(&ft, fec); + if (fn == NULL) + /* route lost */ + return; + fnh = fec_nh_find(fn, af, nexthop, priority); + if (fnh == NULL) + /* route lost */ + return; + + log_debug("lde remove fec %s nexthop %s", + log_fec(&fn->fec), log_addr(af, nexthop)); + + lde_send_delete_klabel(fn, fnh); + fec_nh_del(fnh); + if (LIST_EMPTY(&fn->nexthops)) { + lde_send_labelwithdraw_all(fn, NO_LABEL); + fn->local_label = NO_LABEL; + if (fn->fec.type == FEC_TYPE_PWID) + fn->data = NULL; + } +} + +void +lde_check_mapping(struct map *map, struct lde_nbr *ln) +{ + struct fec fec; + struct fec_node *fn; + struct fec_nh *fnh; + struct lde_req *lre; + struct lde_map *me; + struct l2vpn_pw *pw; + int msgsource = 0; + + lde_map2fec(map, ln->id, &fec); + fn = (struct fec_node *)fec_find(&ft, &fec); + if (fn == NULL) + fn = fec_add(&fec); + + /* LMp.1: first check if we have a pending request running */ + lre = (struct lde_req *)fec_find(&ln->sent_req, &fn->fec); + if (lre) + /* LMp.2: delete record of outstanding label request */ + lde_req_del(ln, lre, 1); + + /* RFC 4447 control word and status tlv negotiation */ + if (map->type == MAP_TYPE_PWID && l2vpn_pw_negotiate(ln, fn, map)) + return; + + /* + * LMp.3 - LMp.8: loop detection - unnecessary for frame-mode + * mpls networks. + */ + + /* LMp.9 */ + me = (struct lde_map *)fec_find(&ln->recv_map, &fn->fec); + if (me) { + /* LMp.10 */ + if (me->map.label != map->label && lre == NULL) { + /* LMp.10a */ + lde_send_labelrelease(ln, fn, me->map.label); + + /* + * Can not use lde_nbr_find_by_addr() because there's + * the possibility of multipath. + */ + LIST_FOREACH(fnh, &fn->nexthops, entry) { + if (lde_address_find(ln, fnh->af, + &fnh->nexthop) == NULL) + continue; + + lde_send_delete_klabel(fn, fnh); + fnh->remote_label = NO_LABEL; + } + } + } + + /* + * LMp.11 - 12: consider multiple nexthops in order to + * support multipath + */ + LIST_FOREACH(fnh, &fn->nexthops, entry) { + /* LMp.15: install FEC in FIB */ + switch (fec.type) { + case FEC_TYPE_IPV4: + case FEC_TYPE_IPV6: + if (!lde_address_find(ln, fnh->af, &fnh->nexthop)) + continue; + + fnh->remote_label = map->label; + lde_send_change_klabel(fn, fnh); + break; + case FEC_TYPE_PWID: + pw = (struct l2vpn_pw *) fn->data; + if (pw == NULL) + continue; + + pw->remote_group = map->fec.pwid.group_id; + if (map->flags & F_MAP_PW_IFMTU) + pw->remote_mtu = map->fec.pwid.ifmtu; + if (map->flags & F_MAP_PW_STATUS) + pw->remote_status = map->pw_status; + fnh->remote_label = map->label; + if (l2vpn_pw_ok(pw, fnh)) + lde_send_change_klabel(fn, fnh); + break; + default: + break; + } + + msgsource = 1; + } + /* LMp.13 & LMp.16: Record the mapping from this peer */ + if (me == NULL) + me = lde_map_add(ln, fn, 0); + me->map = *map; + + if (msgsource == 0) + /* LMp.13: just return since we use liberal lbl retention */ + return; + + /* + * LMp.17 - LMp.27 are unnecessary since we don't need to implement + * loop detection. LMp.28 - LMp.30 are unnecessary because we are + * merging capable. + */ +} + +void +lde_check_request(struct map *map, struct lde_nbr *ln) +{ + struct fec fec; + struct lde_req *lre; + struct fec_node *fn; + struct fec_nh *fnh; + + /* LRq.1: skip loop detection (not necessary) */ + + /* LRq.2: is there a next hop for fec? */ + lde_map2fec(map, ln->id, &fec); + fn = (struct fec_node *)fec_find(&ft, &fec); + if (fn == NULL || LIST_EMPTY(&fn->nexthops)) { + /* LRq.5: send No Route notification */ + lde_send_notification(ln->peerid, S_NO_ROUTE, map->msg_id, + htons(MSG_TYPE_LABELREQUEST)); + return; + } + + /* LRq.3: is MsgSource the next hop? */ + LIST_FOREACH(fnh, &fn->nexthops, entry) { + switch (fec.type) { + case FEC_TYPE_IPV4: + case FEC_TYPE_IPV6: + if (!lde_address_find(ln, fnh->af, &fnh->nexthop)) + continue; + + /* LRq.4: send Loop Detected notification */ + lde_send_notification(ln->peerid, S_LOOP_DETECTED, + map->msg_id, htons(MSG_TYPE_LABELREQUEST)); + return; + default: + break; + } + } + + /* LRq.6: first check if we have a pending request running */ + lre = (struct lde_req *)fec_find(&ln->recv_req, &fn->fec); + if (lre != NULL) + /* LRq.7: duplicate request */ + return; + + /* LRq.8: record label request */ + lre = lde_req_add(ln, &fn->fec, 0); + if (lre != NULL) + lre->msg_id = ntohl(map->msg_id); + + /* LRq.9: perform LSR label distribution */ + lde_send_labelmapping(ln, fn, 1); + + /* + * LRq.10: do nothing (Request Never) since we use liberal + * label retention. + * LRq.11 - 12 are unnecessary since we are merging capable. + */ +} + +void +lde_check_release(struct map *map, struct lde_nbr *ln) +{ + struct fec fec; + struct fec_node *fn; + struct lde_wdraw *lw; + struct lde_map *me; + + /* TODO group wildcard */ + if (map->type == MAP_TYPE_PWID && !(map->flags & F_MAP_PW_ID)) + return; + + lde_map2fec(map, ln->id, &fec); + fn = (struct fec_node *)fec_find(&ft, &fec); + /* LRl.1: does FEC match a known FEC? */ + if (fn == NULL) + return; + + /* LRl.3: first check if we have a pending withdraw running */ + lw = (struct lde_wdraw *)fec_find(&ln->sent_wdraw, &fn->fec); + if (lw && (map->label == NO_LABEL || + (lw->label != NO_LABEL && map->label == lw->label))) { + /* LRl.4: delete record of outstanding label withdraw */ + lde_wdraw_del(ln, lw); + } + + /* LRl.6: check sent map list and remove it if available */ + me = (struct lde_map *)fec_find(&ln->sent_map, &fn->fec); + if (me && (map->label == NO_LABEL || map->label == me->map.label)) + lde_map_del(ln, me, 1); + + /* + * LRl.11 - 13 are unnecessary since we remove the label from + * forwarding/switching as soon as the FEC is unreachable. + */ +} + +void +lde_check_release_wcard(struct map *map, struct lde_nbr *ln) +{ + struct fec *f; + struct fec_node *fn; + struct lde_wdraw *lw; + struct lde_map *me; + + RB_FOREACH(f, fec_tree, &ft) { + fn = (struct fec_node *)f; + + /* LRl.3: first check if we have a pending withdraw running */ + lw = (struct lde_wdraw *)fec_find(&ln->sent_wdraw, &fn->fec); + if (lw && (map->label == NO_LABEL || + (lw->label != NO_LABEL && map->label == lw->label))) { + /* LRl.4: delete record of outstanding lbl withdraw */ + lde_wdraw_del(ln, lw); + } + + /* LRl.6: check sent map list and remove it if available */ + me = (struct lde_map *)fec_find(&ln->sent_map, &fn->fec); + if (me && + (map->label == NO_LABEL || map->label == me->map.label)) + lde_map_del(ln, me, 1); + + /* + * LRl.11 - 13 are unnecessary since we remove the label from + * forwarding/switching as soon as the FEC is unreachable. + */ + } +} + +void +lde_check_withdraw(struct map *map, struct lde_nbr *ln) +{ + struct fec fec; + struct fec_node *fn; + struct fec_nh *fnh; + struct lde_map *me; + struct l2vpn_pw *pw; + + /* TODO group wildcard */ + if (map->type == MAP_TYPE_PWID && !(map->flags & F_MAP_PW_ID)) + return; + + lde_map2fec(map, ln->id, &fec); + fn = (struct fec_node *)fec_find(&ft, &fec); + if (fn == NULL) + fn = fec_add(&fec); + + /* LWd.1: remove label from forwarding/switching use */ + LIST_FOREACH(fnh, &fn->nexthops, entry) { + switch (fec.type) { + case FEC_TYPE_IPV4: + case FEC_TYPE_IPV6: + if (!lde_address_find(ln, fnh->af, &fnh->nexthop)) + continue; + break; + case FEC_TYPE_PWID: + pw = (struct l2vpn_pw *) fn->data; + if (pw == NULL) + continue; + break; + default: + break; + } + lde_send_delete_klabel(fn, fnh); + fnh->remote_label = NO_LABEL; + } + + /* LWd.2: send label release */ + lde_send_labelrelease(ln, fn, map->label); + + /* LWd.3: check previously received label mapping */ + me = (struct lde_map *)fec_find(&ln->recv_map, &fn->fec); + if (me && (map->label == NO_LABEL || map->label == me->map.label)) + /* LWd.4: remove record of previously received lbl mapping */ + lde_map_del(ln, me, 0); +} + +void +lde_check_withdraw_wcard(struct map *map, struct lde_nbr *ln) +{ + struct fec *f; + struct fec_node *fn; + struct fec_nh *fnh; + struct lde_map *me; + + /* LWd.2: send label release */ + lde_send_labelrelease(ln, NULL, map->label); + + RB_FOREACH(f, fec_tree, &ft) { + fn = (struct fec_node *)f; + + /* LWd.1: remove label from forwarding/switching use */ + LIST_FOREACH(fnh, &fn->nexthops, entry) { + switch (f->type) { + case FEC_TYPE_IPV4: + case FEC_TYPE_IPV6: + if (!lde_address_find(ln, fnh->af, + &fnh->nexthop)) + continue; + break; + case FEC_TYPE_PWID: + if (f->u.pwid.lsr_id.s_addr != ln->id.s_addr) + continue; + break; + default: + break; + } + lde_send_delete_klabel(fn, fnh); + fnh->remote_label = NO_LABEL; + } + + /* LWd.3: check previously received label mapping */ + me = (struct lde_map *)fec_find(&ln->recv_map, &fn->fec); + if (me && (map->label == NO_LABEL || + map->label == me->map.label)) + /* + * LWd.4: remove record of previously received + * label mapping + */ + lde_map_del(ln, me, 0); + } +} + +/* gabage collector timer: timer to remove dead entries from the LIB */ + +/* ARGSUSED */ +void +lde_gc_timer(int fd, short event, void *arg) +{ + struct fec *fec, *safe; + struct fec_node *fn; + int count = 0; + + RB_FOREACH_SAFE(fec, fec_tree, &ft, safe) { + fn = (struct fec_node *) fec; + + if (!LIST_EMPTY(&fn->nexthops) || + !LIST_EMPTY(&fn->downstream) || + !LIST_EMPTY(&fn->upstream)) + continue; + + fec_remove(&ft, &fn->fec); + free(fn); + count++; + } + + if (count > 0) + log_debug("%s: %u entries removed", __func__, count); + + lde_gc_start_timer(); +} + +void +lde_gc_start_timer(void) +{ + struct timeval tv; + + timerclear(&tv); + tv.tv_sec = LDE_GC_INTERVAL; + if (evtimer_add(&gc_timer, &tv) == -1) + fatal(__func__); +} + +void +lde_gc_stop_timer(void) +{ + if (evtimer_pending(&gc_timer, NULL) && + evtimer_del(&gc_timer) == -1) + fatal(__func__); +} diff --git a/ldpd/ldp.h b/ldpd/ldp.h new file mode 100644 index 000000000..77034b30a --- /dev/null +++ b/ldpd/ldp.h @@ -0,0 +1,304 @@ +/* $OpenBSD$ */ + +/* + * Copyright (c) 2013, 2016 Renato Westphal + * Copyright (c) 2009 Michele Marchetto + * Copyright (c) 2004, 2005, 2008 Esben Norby + * + * 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. + */ + +/* LDP protocol definitions */ + +#ifndef _LDP_H_ +#define _LDP_H_ + +#include + +/* misc */ +#define LDP_VERSION 1 +#define LDP_PORT 646 +#define LDP_MAX_LEN 4096 + +/* All Routers on this Subnet group multicast addresses */ +#define AllRouters_v4 "224.0.0.2" +#define AllRouters_v6 "ff02::2" + +#define LINK_DFLT_HOLDTIME 15 +#define TARGETED_DFLT_HOLDTIME 45 +#define MIN_HOLDTIME 3 +#define MAX_HOLDTIME 0xffff +#define INFINITE_HOLDTIME 0xffff + +#define DEFAULT_KEEPALIVE 180 +#define MIN_KEEPALIVE 3 +#define MAX_KEEPALIVE 0xffff +#define KEEPALIVE_PER_PERIOD 3 +#define INIT_FSM_TIMEOUT 15 + +#define DEFAULT_HELLO_INTERVAL 5 +#define MIN_HELLO_INTERVAL 1 +#define MAX_HELLO_INTERVAL 0xffff + +#define INIT_DELAY_TMR 15 +#define MAX_DELAY_TMR 120 + +#define MIN_PWID_ID 1 +#define MAX_PWID_ID 0xffffffff + +#define DEFAULT_L2VPN_MTU 1500 +#define MIN_L2VPN_MTU 512 +#define MAX_L2VPN_MTU 0xffff + +/* LDP message types */ +#define MSG_TYPE_NOTIFICATION 0x0001 +#define MSG_TYPE_HELLO 0x0100 +#define MSG_TYPE_INIT 0x0200 +#define MSG_TYPE_KEEPALIVE 0x0201 +#define MSG_TYPE_ADDR 0x0300 +#define MSG_TYPE_ADDRWITHDRAW 0x0301 +#define MSG_TYPE_LABELMAPPING 0x0400 +#define MSG_TYPE_LABELREQUEST 0x0401 +#define MSG_TYPE_LABELWITHDRAW 0x0402 +#define MSG_TYPE_LABELRELEASE 0x0403 +#define MSG_TYPE_LABELABORTREQ 0x0404 + +/* LDP TLV types */ +#define TLV_TYPE_FEC 0x0100 +#define TLV_TYPE_ADDRLIST 0x0101 +#define TLV_TYPE_HOPCOUNT 0x0103 +#define TLV_TYPE_PATHVECTOR 0x0104 +#define TLV_TYPE_GENERICLABEL 0x0200 +#define TLV_TYPE_ATMLABEL 0x0201 +#define TLV_TYPE_FRLABEL 0x0202 +#define TLV_TYPE_STATUS 0x0300 +#define TLV_TYPE_EXTSTATUS 0x0301 +#define TLV_TYPE_RETURNEDPDU 0x0302 +#define TLV_TYPE_RETURNEDMSG 0x0303 +#define TLV_TYPE_COMMONHELLO 0x0400 +#define TLV_TYPE_IPV4TRANSADDR 0x0401 +#define TLV_TYPE_CONFIG 0x0402 +#define TLV_TYPE_IPV6TRANSADDR 0x0403 +#define TLV_TYPE_COMMONSESSION 0x0500 +#define TLV_TYPE_ATMSESSIONPAR 0x0501 +#define TLV_TYPE_FRSESSION 0x0502 +#define TLV_TYPE_LABELREQUEST 0x0600 +/* RFC 4447 */ +#define TLV_TYPE_PW_STATUS 0x096A +#define TLV_TYPE_PW_IF_PARAM 0x096B +#define TLV_TYPE_PW_GROUP_ID 0x096C +/* RFC 7552 */ +#define TLV_TYPE_DUALSTACK 0x8701 + +/* LDP header */ +struct ldp_hdr { + uint16_t version; + uint16_t length; + uint32_t lsr_id; + uint16_t lspace_id; +} __packed; + +#define LDP_HDR_SIZE 10 /* actual size of the LDP header */ +#define LDP_HDR_PDU_LEN 6 /* minimum "PDU Length" */ +#define LDP_HDR_DEAD_LEN 4 + +/* TLV record */ +struct tlv { + uint16_t type; + uint16_t length; +}; +#define TLV_HDR_SIZE 4 + +struct ldp_msg { + uint16_t type; + uint16_t length; + uint32_t id; + /* Mandatory Parameters */ + /* Optional Parameters */ +} __packed; + +#define LDP_MSG_SIZE 8 /* minimum size of LDP message */ +#define LDP_MSG_LEN 4 /* minimum "Message Length" */ +#define LDP_MSG_DEAD_LEN 4 + +#define UNKNOWN_FLAG 0x8000 +#define FORWARD_FLAG 0xc000 + +struct hello_prms_tlv { + uint16_t type; + uint16_t length; + uint16_t holdtime; + uint16_t flags; +}; +#define F_HELLO_TARGETED 0x8000 +#define F_HELLO_REQ_TARG 0x4000 +#define F_HELLO_GTSM 0x2000 + +struct hello_prms_opt4_tlv { + uint16_t type; + uint16_t length; + uint32_t value; +}; + +struct hello_prms_opt16_tlv { + uint16_t type; + uint16_t length; + uint8_t value[16]; +}; + +#define DUAL_STACK_LDPOV4 4 +#define DUAL_STACK_LDPOV6 6 + +#define F_HELLO_TLV_RCVD_ADDR 0x01 +#define F_HELLO_TLV_RCVD_CONF 0x02 +#define F_HELLO_TLV_RCVD_DS 0x04 + +#define S_SUCCESS 0x00000000 +#define S_BAD_LDP_ID 0x80000001 +#define S_BAD_PROTO_VER 0x80000002 +#define S_BAD_PDU_LEN 0x80000003 +#define S_UNKNOWN_MSG 0x00000004 +#define S_BAD_MSG_LEN 0x80000005 +#define S_UNKNOWN_TLV 0x00000006 +#define S_BAD_TLV_LEN 0x80000007 +#define S_BAD_TLV_VAL 0x80000008 +#define S_HOLDTIME_EXP 0x80000009 +#define S_SHUTDOWN 0x8000000A +#define S_LOOP_DETECTED 0x0000000B +#define S_UNKNOWN_FEC 0x0000000C +#define S_NO_ROUTE 0x0000000D +#define S_NO_LABEL_RES 0x0000000E +#define S_AVAILABLE 0x0000000F +#define S_NO_HELLO 0x80000010 +#define S_PARM_ADV_MODE 0x80000011 +#define S_MAX_PDU_LEN 0x80000012 +#define S_PARM_L_RANGE 0x80000013 +#define S_KEEPALIVE_TMR 0x80000014 +#define S_LAB_REQ_ABRT 0x00000015 +#define S_MISS_MSG 0x00000016 +#define S_UNSUP_ADDR 0x00000017 +#define S_KEEPALIVE_BAD 0x80000018 +#define S_INTERN_ERR 0x80000019 +/* RFC 4447 */ +#define S_ILLEGAL_CBIT 0x00000024 +#define S_WRONG_CBIT 0x00000025 +#define S_INCPT_BITRATE 0x00000026 +#define S_CEP_MISCONF 0x00000027 +#define S_PW_STATUS 0x00000028 +#define S_UNASSIGN_TAI 0x00000029 +#define S_MISCONF_ERR 0x0000002A +#define S_WITHDRAW_MTHD 0x0000002B +/* RFC 7552 */ +#define S_TRANS_MISMTCH 0x80000032 +#define S_DS_NONCMPLNCE 0x80000033 + +struct sess_prms_tlv { + uint16_t type; + uint16_t length; + uint16_t proto_version; + uint16_t keepalive_time; + uint8_t reserved; + uint8_t pvlim; + uint16_t max_pdu_len; + uint32_t lsr_id; + uint16_t lspace_id; +} __packed; + +#define SESS_PRMS_SIZE 18 +#define SESS_PRMS_LEN 14 + +struct status_tlv { + uint16_t type; + uint16_t length; + uint32_t status_code; + uint32_t msg_id; + uint16_t msg_type; +} __packed; + +#define STATUS_SIZE 14 +#define STATUS_TLV_LEN 10 +#define STATUS_FATAL 0x80000000 + +#define AF_IPV4 0x1 +#define AF_IPV6 0x2 + +struct address_list_tlv { + uint16_t type; + uint16_t length; + uint16_t family; + /* address entries */ +} __packed; + +#define ADDR_LIST_SIZE 6 + +#define FEC_ELM_WCARD_LEN 1 +#define FEC_ELM_PREFIX_MIN_LEN 4 +#define FEC_PWID_ELM_MIN_LEN 8 + +#define MAP_TYPE_WILDCARD 0x01 +#define MAP_TYPE_PREFIX 0x02 +#define MAP_TYPE_PWID 0x80 +#define MAP_TYPE_GENPWID 0x81 + +#define CONTROL_WORD_FLAG 0x8000 +#define PW_TYPE_ETHERNET_TAGGED 0x0004 +#define PW_TYPE_ETHERNET 0x0005 +#define DEFAULT_PW_TYPE PW_TYPE_ETHERNET + +/* RFC 4447 Sub-TLV record */ +struct subtlv { + uint8_t type; + uint8_t length; +}; +#define SUBTLV_HDR_SIZE 2 + +#define SUBTLV_IFMTU 0x01 +#define SUBTLV_VLANID 0x06 + +#define FEC_SUBTLV_IFMTU_SIZE 4 +#define FEC_SUBTLV_VLANID_SIZE 4 + +struct label_tlv { + uint16_t type; + uint16_t length; + uint32_t label; +}; +#define LABEL_TLV_SIZE 8 +#define LABEL_TLV_LEN 4 + +struct reqid_tlv { + uint16_t type; + uint16_t length; + uint32_t reqid; +}; +#define REQID_TLV_SIZE 8 +#define REQID_TLV_LEN 4 + +struct pw_status_tlv { + uint16_t type; + uint16_t length; + uint32_t value; +}; +#define PW_STATUS_TLV_SIZE 8 +#define PW_STATUS_TLV_LEN 4 + +#define PW_FORWARDING 0 +#define PW_NOT_FORWARDING (1 << 0) +#define PW_LOCAL_RX_FAULT (1 << 1) +#define PW_LOCAL_TX_FAULT (1 << 2) +#define PW_PSN_RX_FAULT (1 << 3) +#define PW_PSN_TX_FAULT (1 << 4) + +#define NO_LABEL UINT32_MAX + +#endif /* !_LDP_H_ */ diff --git a/ldpd/ldpd.c b/ldpd/ldpd.c new file mode 100644 index 000000000..4e1449198 --- /dev/null +++ b/ldpd/ldpd.c @@ -0,0 +1,1227 @@ +/* $OpenBSD$ */ + +/* + * Copyright (c) 2013, 2016 Renato Westphal + * Copyright (c) 2005 Claudio Jeker + * Copyright (c) 2004, 2008 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 "ldpd.h" +#include "ldpe.h" +#include "lde.h" +#include "log.h" + +static void main_sig_handler(int, short, void *); +static __dead void usage(void); +static __dead void ldpd_shutdown(void); +static pid_t start_child(enum ldpd_process, char *, int, int, int); +static void main_dispatch_ldpe(int, short, void *); +static void main_dispatch_lde(int, short, void *); +static int main_imsg_compose_both(enum imsg_type, void *, + uint16_t); +static int main_imsg_send_ipc_sockets(struct imsgbuf *, + struct imsgbuf *); +static void main_imsg_send_net_sockets(int); +static void main_imsg_send_net_socket(int, enum socket_type); +static int main_imsg_send_config(struct ldpd_conf *); +static int ldp_reload(void); +static void merge_global(struct ldpd_conf *, struct ldpd_conf *); +static void merge_af(int, struct ldpd_af_conf *, + struct ldpd_af_conf *); +static void merge_ifaces(struct ldpd_conf *, struct ldpd_conf *); +static void merge_iface_af(struct iface_af *, struct iface_af *); +static void merge_tnbrs(struct ldpd_conf *, struct ldpd_conf *); +static void merge_nbrps(struct ldpd_conf *, struct ldpd_conf *); +static void merge_l2vpns(struct ldpd_conf *, struct ldpd_conf *); +static void merge_l2vpn(struct ldpd_conf *, struct l2vpn *, + struct l2vpn *); + +struct ldpd_global global; +struct ldpd_conf *ldpd_conf; + +static char *conffile; +static struct imsgev *iev_ldpe; +static struct imsgev *iev_lde; +static pid_t ldpe_pid; +static pid_t lde_pid; + +/* ARGSUSED */ +static void +main_sig_handler(int sig, short event, void *arg) +{ + /* signal handler rules don't apply, libevent decouples for us */ + switch (sig) { + case SIGTERM: + case SIGINT: + ldpd_shutdown(); + /* NOTREACHED */ + case SIGHUP: + if (ldp_reload() == -1) + log_warnx("configuration reload failed"); + else + log_debug("configuration reloaded"); + break; + default: + fatalx("unexpected signal"); + /* NOTREACHED */ + } +} + +static __dead void +usage(void) +{ + extern char *__progname; + + fprintf(stderr, "usage: %s [-dnv] [-D macro=value] [-f file]\n", + __progname); + exit(1); +} + +int +main(int argc, char *argv[]) +{ + struct event ev_sigint, ev_sigterm, ev_sighup; + char *saved_argv0; + int ch; + int debug = 0, lflag = 0, eflag = 0; + int pipe_parent2ldpe[2]; + int pipe_parent2lde[2]; + + conffile = CONF_FILE; + ldpd_process = PROC_MAIN; + + log_init(1); /* log to stderr until daemonized */ + log_verbose(1); + + saved_argv0 = argv[0]; + if (saved_argv0 == NULL) + saved_argv0 = "ldpd"; + + while ((ch = getopt(argc, argv, "dD:f:nvLE")) != -1) { + switch (ch) { + case 'd': + debug = 1; + break; + case 'D': + if (cmdline_symset(optarg) < 0) + log_warnx("could not parse macro definition %s", + optarg); + break; + case 'f': + conffile = optarg; + break; + case 'n': + global.cmd_opts |= LDPD_OPT_NOACTION; + break; + case 'v': + if (global.cmd_opts & LDPD_OPT_VERBOSE) + global.cmd_opts |= LDPD_OPT_VERBOSE2; + global.cmd_opts |= LDPD_OPT_VERBOSE; + break; + case 'L': + lflag = 1; + break; + case 'E': + eflag = 1; + break; + default: + usage(); + /* NOTREACHED */ + } + } + + argc -= optind; + argv += optind; + if (argc > 0 || (lflag && eflag)) + usage(); + + if (lflag) + lde(debug, global.cmd_opts & LDPD_OPT_VERBOSE); + else if (eflag) + ldpe(debug, global.cmd_opts & LDPD_OPT_VERBOSE); + + /* fetch interfaces early */ + kif_init(); + + /* parse config file */ + if ((ldpd_conf = parse_config(conffile)) == NULL ) { + kif_clear(); + exit(1); + } + + if (global.cmd_opts & LDPD_OPT_NOACTION) { + if (global.cmd_opts & LDPD_OPT_VERBOSE) + print_config(ldpd_conf); + else + fprintf(stderr, "configuration OK\n"); + kif_clear(); + exit(0); + } + + /* check for root privileges */ + if (geteuid()) + errx(1, "need root privileges"); + + /* check for ldpd user */ + if (getpwnam(LDPD_USER) == NULL) + errx(1, "unknown user %s", LDPD_USER); + + log_init(debug); + log_verbose(global.cmd_opts & (LDPD_OPT_VERBOSE | LDPD_OPT_VERBOSE2)); + + if (!debug) + daemon(1, 0); + + log_info("startup"); + + if (socketpair(AF_UNIX, SOCK_STREAM | SOCK_NONBLOCK | SOCK_CLOEXEC, + PF_UNSPEC, pipe_parent2ldpe) == -1) + fatal("socketpair"); + if (socketpair(AF_UNIX, SOCK_STREAM | SOCK_NONBLOCK | SOCK_CLOEXEC, + PF_UNSPEC, pipe_parent2lde) == -1) + fatal("socketpair"); + + /* start children */ + lde_pid = start_child(PROC_LDE_ENGINE, saved_argv0, + pipe_parent2lde[1], debug, global.cmd_opts & LDPD_OPT_VERBOSE); + ldpe_pid = start_child(PROC_LDP_ENGINE, saved_argv0, + pipe_parent2ldpe[1], debug, global.cmd_opts & LDPD_OPT_VERBOSE); + + event_init(); + + /* setup signal handler */ + signal_set(&ev_sigint, SIGINT, main_sig_handler, NULL); + signal_set(&ev_sigterm, SIGTERM, main_sig_handler, NULL); + signal_set(&ev_sighup, SIGHUP, main_sig_handler, NULL); + signal_add(&ev_sigint, NULL); + signal_add(&ev_sigterm, NULL); + signal_add(&ev_sighup, NULL); + signal(SIGPIPE, SIG_IGN); + + /* setup pipes to children */ + if ((iev_ldpe = malloc(sizeof(struct imsgev))) == NULL || + (iev_lde = malloc(sizeof(struct imsgev))) == NULL) + fatal(NULL); + imsg_init(&iev_ldpe->ibuf, pipe_parent2ldpe[0]); + iev_ldpe->handler = main_dispatch_ldpe; + imsg_init(&iev_lde->ibuf, pipe_parent2lde[0]); + iev_lde->handler = main_dispatch_lde; + + /* setup event handler */ + iev_ldpe->events = EV_READ; + event_set(&iev_ldpe->ev, iev_ldpe->ibuf.fd, iev_ldpe->events, + iev_ldpe->handler, iev_ldpe); + event_add(&iev_ldpe->ev, NULL); + + iev_lde->events = EV_READ; + event_set(&iev_lde->ev, iev_lde->ibuf.fd, iev_lde->events, + iev_lde->handler, iev_lde); + event_add(&iev_lde->ev, NULL); + + if (main_imsg_send_ipc_sockets(&iev_ldpe->ibuf, &iev_lde->ibuf)) + fatal("could not establish imsg links"); + main_imsg_send_config(ldpd_conf); + + /* notify ldpe about existing interfaces and addresses */ + kif_redistribute(NULL); + + if (kr_init(!(ldpd_conf->flags & F_LDPD_NO_FIB_UPDATE)) == -1) + fatalx("kr_init failed"); + + if (ldpd_conf->ipv4.flags & F_LDPD_AF_ENABLED) + main_imsg_send_net_sockets(AF_INET); + if (ldpd_conf->ipv6.flags & F_LDPD_AF_ENABLED) + main_imsg_send_net_sockets(AF_INET6); + + /* remove unneded stuff from config */ + /* ... */ + + event_dispatch(); + + ldpd_shutdown(); + /* NOTREACHED */ + return (0); +} + +static __dead void +ldpd_shutdown(void) +{ + pid_t pid; + int status; + + /* close pipes */ + msgbuf_clear(&iev_ldpe->ibuf.w); + close(iev_ldpe->ibuf.fd); + msgbuf_clear(&iev_lde->ibuf.w); + close(iev_lde->ibuf.fd); + + kr_shutdown(); + config_clear(ldpd_conf); + + log_debug("waiting for children to terminate"); + do { + pid = wait(&status); + if (pid == -1) { + if (errno != EINTR && errno != ECHILD) + fatal("wait"); + } else if (WIFSIGNALED(status)) + log_warnx("%s terminated; signal %d", + (pid == lde_pid) ? "label decision engine" : + "ldp engine", WTERMSIG(status)); + } while (pid != -1 || (pid == -1 && errno == EINTR)); + + free(iev_ldpe); + free(iev_lde); + + log_info("terminating"); + exit(0); +} + +static pid_t +start_child(enum ldpd_process p, char *argv0, int fd, int debug, int verbose) +{ + char *argv[5]; + int argc = 0; + pid_t pid; + + switch (pid = fork()) { + case -1: + fatal("cannot fork"); + case 0: + break; + default: + close(fd); + return (pid); + } + + if (dup2(fd, 3) == -1) + fatal("cannot setup imsg fd"); + + argv[argc++] = argv0; + switch (p) { + case PROC_MAIN: + fatalx("Can not start main process"); + case PROC_LDE_ENGINE: + argv[argc++] = "-L"; + break; + case PROC_LDP_ENGINE: + argv[argc++] = "-E"; + break; + } + if (debug) + argv[argc++] = "-d"; + if (verbose) + argv[argc++] = "-v"; + argv[argc++] = NULL; + + execvp(argv0, argv); + fatal("execvp"); +} + +/* imsg handling */ +/* ARGSUSED */ +static void +main_dispatch_ldpe(int fd, short event, void *bula) +{ + struct imsgev *iev = bula; + struct imsgbuf *ibuf = &iev->ibuf; + struct imsg imsg; + int af; + ssize_t n; + int shut = 0, verbose; + + if (event & EV_READ) { + if ((n = imsg_read(ibuf)) == -1 && errno != EAGAIN) + fatal("imsg_read error"); + if (n == 0) /* connection closed */ + shut = 1; + } + if (event & EV_WRITE) { + if ((n = msgbuf_write(&ibuf->w)) == -1 && errno != EAGAIN) + fatal("msgbuf_write"); + if (n == 0) + shut = 1; + } + + for (;;) { + if ((n = imsg_get(ibuf, &imsg)) == -1) + fatal("imsg_get"); + + if (n == 0) + break; + + switch (imsg.hdr.type) { + case IMSG_REQUEST_SOCKETS: + af = imsg.hdr.pid; + main_imsg_send_net_sockets(af); + break; + case IMSG_CTL_RELOAD: + if (ldp_reload() == -1) + log_warnx("configuration reload failed"); + else + log_debug("configuration reloaded"); + break; + case IMSG_CTL_FIB_COUPLE: + kr_fib_couple(); + break; + case IMSG_CTL_FIB_DECOUPLE: + kr_fib_decouple(); + break; + case IMSG_CTL_KROUTE: + case IMSG_CTL_KROUTE_ADDR: + kr_show_route(&imsg); + break; + case IMSG_CTL_IFINFO: + if (imsg.hdr.len == IMSG_HEADER_SIZE) + kr_ifinfo(NULL, imsg.hdr.pid); + else if (imsg.hdr.len == IMSG_HEADER_SIZE + IFNAMSIZ) + kr_ifinfo(imsg.data, imsg.hdr.pid); + else + log_warnx("IFINFO request with wrong len"); + break; + case IMSG_CTL_LOG_VERBOSE: + /* already checked by ldpe */ + memcpy(&verbose, imsg.data, sizeof(verbose)); + log_verbose(verbose); + break; + default: + log_debug("%s: error handling imsg %d", __func__, + 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 */ +static void +main_dispatch_lde(int fd, short event, void *bula) +{ + struct imsgev *iev = bula; + struct imsgbuf *ibuf = &iev->ibuf; + struct imsg imsg; + ssize_t n; + int shut = 0; + + if (event & EV_READ) { + if ((n = imsg_read(ibuf)) == -1 && errno != EAGAIN) + fatal("imsg_read error"); + if (n == 0) /* connection closed */ + shut = 1; + } + if (event & EV_WRITE) { + if ((n = msgbuf_write(&ibuf->w)) == -1 && errno != EAGAIN) + fatal("msgbuf_write"); + if (n == 0) + shut = 1; + } + + for (;;) { + if ((n = imsg_get(ibuf, &imsg)) == -1) + fatal("imsg_get"); + + if (n == 0) + break; + + switch (imsg.hdr.type) { + case IMSG_KLABEL_CHANGE: + if (imsg.hdr.len - IMSG_HEADER_SIZE != + sizeof(struct kroute)) + fatalx("invalid size of IMSG_KLABEL_CHANGE"); + if (kr_change(imsg.data)) + log_warnx("%s: error changing route", __func__); + break; + case IMSG_KLABEL_DELETE: + if (imsg.hdr.len - IMSG_HEADER_SIZE != + sizeof(struct kroute)) + fatalx("invalid size of IMSG_KLABEL_DELETE"); + if (kr_delete(imsg.data)) + log_warnx("%s: error deleting route", __func__); + break; + case IMSG_KPWLABEL_CHANGE: + if (imsg.hdr.len - IMSG_HEADER_SIZE != + sizeof(struct kpw)) + fatalx("invalid size of IMSG_KPWLABEL_CHANGE"); + if (kmpw_set(imsg.data)) + log_warnx("%s: error changing pseudowire", + __func__); + break; + case IMSG_KPWLABEL_DELETE: + if (imsg.hdr.len - IMSG_HEADER_SIZE != + sizeof(struct kpw)) + fatalx("invalid size of IMSG_KPWLABEL_DELETE"); + if (kmpw_unset(imsg.data)) + log_warnx("%s: error unsetting pseudowire", + __func__); + break; + default: + log_debug("%s: error handling imsg %d", __func__, + 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); + } +} + +void +main_imsg_compose_ldpe(int type, pid_t pid, void *data, uint16_t datalen) +{ + if (iev_ldpe == NULL) + return; + imsg_compose_event(iev_ldpe, type, 0, pid, -1, data, datalen); +} + +void +main_imsg_compose_lde(int type, pid_t pid, void *data, uint16_t datalen) +{ + imsg_compose_event(iev_lde, type, 0, pid, -1, data, datalen); +} + +static int +main_imsg_compose_both(enum imsg_type type, void *buf, uint16_t len) +{ + if (imsg_compose_event(iev_ldpe, type, 0, 0, -1, buf, len) == -1) + return (-1); + if (imsg_compose_event(iev_lde, type, 0, 0, -1, buf, len) == -1) + return (-1); + return (0); +} + +void +imsg_event_add(struct imsgev *iev) +{ + iev->events = EV_READ; + if (iev->ibuf.w.queued) + iev->events |= EV_WRITE; + + event_del(&iev->ev); + event_set(&iev->ev, iev->ibuf.fd, iev->events, iev->handler, iev); + event_add(&iev->ev, NULL); +} + +int +imsg_compose_event(struct imsgev *iev, uint16_t type, uint32_t peerid, + pid_t pid, int fd, void *data, uint16_t datalen) +{ + int ret; + + if ((ret = imsg_compose(&iev->ibuf, type, peerid, + pid, fd, data, datalen)) != -1) + imsg_event_add(iev); + return (ret); +} + +void +evbuf_enqueue(struct evbuf *eb, struct ibuf *buf) +{ + ibuf_close(&eb->wbuf, buf); + evbuf_event_add(eb); +} + +void +evbuf_event_add(struct evbuf *eb) +{ + if (eb->wbuf.queued) + event_add(&eb->ev, NULL); +} + +void +evbuf_init(struct evbuf *eb, int fd, void (*handler)(int, short, void *), + void *arg) +{ + msgbuf_init(&eb->wbuf); + eb->wbuf.fd = fd; + event_set(&eb->ev, eb->wbuf.fd, EV_WRITE, handler, arg); +} + +void +evbuf_clear(struct evbuf *eb) +{ + event_del(&eb->ev); + msgbuf_clear(&eb->wbuf); + eb->wbuf.fd = -1; +} + +static int +main_imsg_send_ipc_sockets(struct imsgbuf *ldpe_buf, struct imsgbuf *lde_buf) +{ + int pipe_ldpe2lde[2]; + + if (socketpair(AF_UNIX, SOCK_STREAM | SOCK_CLOEXEC | SOCK_NONBLOCK, + PF_UNSPEC, pipe_ldpe2lde) == -1) + return (-1); + + if (imsg_compose(ldpe_buf, IMSG_SOCKET_IPC, 0, 0, pipe_ldpe2lde[0], + NULL, 0) == -1) + return (-1); + if (imsg_compose(lde_buf, IMSG_SOCKET_IPC, 0, 0, pipe_ldpe2lde[1], + NULL, 0) == -1) + return (-1); + + return (0); +} + +static void +main_imsg_send_net_sockets(int af) +{ + main_imsg_send_net_socket(af, LDP_SOCKET_DISC); + main_imsg_send_net_socket(af, LDP_SOCKET_EDISC); + main_imsg_send_net_socket(af, LDP_SOCKET_SESSION); + imsg_compose_event(iev_ldpe, IMSG_SETUP_SOCKETS, af, 0, -1, NULL, 0); +} + +static void +main_imsg_send_net_socket(int af, enum socket_type type) +{ + int fd; + + fd = ldp_create_socket(af, type); + if (fd == -1) { + log_warnx("%s: failed to create %s socket for address-family " + "%s", __func__, socket_name(type), af_name(af)); + return; + } + + imsg_compose_event(iev_ldpe, IMSG_SOCKET_NET, af, 0, fd, &type, + sizeof(type)); +} + +struct ldpd_af_conf * +ldp_af_conf_get(struct ldpd_conf *xconf, int af) +{ + switch (af) { + case AF_INET: + return (&xconf->ipv4); + case AF_INET6: + return (&xconf->ipv6); + default: + fatalx("ldp_af_conf_get: unknown af"); + } +} + +struct ldpd_af_global * +ldp_af_global_get(struct ldpd_global *xglobal, int af) +{ + switch (af) { + case AF_INET: + return (&xglobal->ipv4); + case AF_INET6: + return (&xglobal->ipv6); + default: + fatalx("ldp_af_global_get: unknown af"); + } +} + +int +ldp_is_dual_stack(struct ldpd_conf *xconf) +{ + return ((xconf->ipv4.flags & F_LDPD_AF_ENABLED) && + (xconf->ipv6.flags & F_LDPD_AF_ENABLED)); +} + +static int +main_imsg_send_config(struct ldpd_conf *xconf) +{ + struct iface *iface; + struct tnbr *tnbr; + struct nbr_params *nbrp; + struct l2vpn *l2vpn; + struct l2vpn_if *lif; + struct l2vpn_pw *pw; + + if (main_imsg_compose_both(IMSG_RECONF_CONF, xconf, + sizeof(*xconf)) == -1) + return (-1); + + LIST_FOREACH(iface, &xconf->iface_list, entry) { + if (main_imsg_compose_both(IMSG_RECONF_IFACE, iface, + sizeof(*iface)) == -1) + return (-1); + } + + LIST_FOREACH(tnbr, &xconf->tnbr_list, entry) { + if (main_imsg_compose_both(IMSG_RECONF_TNBR, tnbr, + sizeof(*tnbr)) == -1) + return (-1); + } + + LIST_FOREACH(nbrp, &xconf->nbrp_list, entry) { + if (main_imsg_compose_both(IMSG_RECONF_NBRP, nbrp, + sizeof(*nbrp)) == -1) + return (-1); + } + + LIST_FOREACH(l2vpn, &xconf->l2vpn_list, entry) { + if (main_imsg_compose_both(IMSG_RECONF_L2VPN, l2vpn, + sizeof(*l2vpn)) == -1) + return (-1); + + LIST_FOREACH(lif, &l2vpn->if_list, entry) { + if (main_imsg_compose_both(IMSG_RECONF_L2VPN_IF, lif, + sizeof(*lif)) == -1) + return (-1); + } + LIST_FOREACH(pw, &l2vpn->pw_list, entry) { + if (main_imsg_compose_both(IMSG_RECONF_L2VPN_PW, pw, + sizeof(*pw)) == -1) + return (-1); + } + } + + if (main_imsg_compose_both(IMSG_RECONF_END, NULL, 0) == -1) + return (-1); + + return (0); +} + +static int +ldp_reload(void) +{ + struct ldpd_conf *xconf; + + if ((xconf = parse_config(conffile)) == NULL) + return (-1); + + if (main_imsg_send_config(xconf) == -1) + return (-1); + + merge_config(ldpd_conf, xconf); + + return (0); +} + +void +merge_config(struct ldpd_conf *conf, struct ldpd_conf *xconf) +{ + merge_global(conf, xconf); + merge_af(AF_INET, &conf->ipv4, &xconf->ipv4); + merge_af(AF_INET6, &conf->ipv6, &xconf->ipv6); + merge_ifaces(conf, xconf); + merge_tnbrs(conf, xconf); + merge_nbrps(conf, xconf); + merge_l2vpns(conf, xconf); + free(xconf); +} + +static void +merge_global(struct ldpd_conf *conf, struct ldpd_conf *xconf) +{ + /* change of router-id requires resetting all neighborships */ + if (conf->rtr_id.s_addr != xconf->rtr_id.s_addr) { + if (ldpd_process == PROC_LDP_ENGINE) { + ldpe_reset_nbrs(AF_INET); + ldpe_reset_nbrs(AF_INET6); + if (conf->rtr_id.s_addr == INADDR_ANY || + xconf->rtr_id.s_addr == INADDR_ANY) { + if_update_all(AF_UNSPEC); + tnbr_update_all(AF_UNSPEC); + } + } + conf->rtr_id = xconf->rtr_id; + } + + if (conf->trans_pref != xconf->trans_pref) { + if (ldpd_process == PROC_LDP_ENGINE) + ldpe_reset_ds_nbrs(); + conf->trans_pref = xconf->trans_pref; + } + + if ((conf->flags & F_LDPD_DS_CISCO_INTEROP) != + (xconf->flags & F_LDPD_DS_CISCO_INTEROP)) { + if (ldpd_process == PROC_LDP_ENGINE) + ldpe_reset_ds_nbrs(); + } + + conf->flags = xconf->flags; +} + +static void +merge_af(int af, struct ldpd_af_conf *af_conf, struct ldpd_af_conf *xa) +{ + int egress_label_changed = 0; + int update_sockets = 0; + + if (af_conf->keepalive != xa->keepalive) { + af_conf->keepalive = xa->keepalive; + if (ldpd_process == PROC_LDP_ENGINE) + ldpe_stop_init_backoff(af); + } + af_conf->thello_holdtime = xa->thello_holdtime; + af_conf->thello_interval = xa->thello_interval; + + /* update flags */ + if (ldpd_process == PROC_LDP_ENGINE && + (af_conf->flags & F_LDPD_AF_THELLO_ACCEPT) && + !(xa->flags & F_LDPD_AF_THELLO_ACCEPT)) + ldpe_remove_dynamic_tnbrs(af); + + if ((af_conf->flags & F_LDPD_AF_NO_GTSM) != + (xa->flags & F_LDPD_AF_NO_GTSM)) { + if (af == AF_INET6) + /* need to set/unset IPV6_MINHOPCOUNT */ + update_sockets = 1; + else if (ldpd_process == PROC_LDP_ENGINE) + /* for LDPv4 just resetting the neighbors is enough */ + ldpe_reset_nbrs(af); + } + + if ((af_conf->flags & F_LDPD_AF_EXPNULL) != + (xa->flags & F_LDPD_AF_EXPNULL)) + egress_label_changed = 1; + + af_conf->flags = xa->flags; + + if (egress_label_changed) { + switch (ldpd_process) { + case PROC_LDE_ENGINE: + lde_change_egress_label(af, af_conf->flags & + F_LDPD_AF_EXPNULL); + break; + case PROC_MAIN: + kr_change_egress_label(af, af_conf->flags & + F_LDPD_AF_EXPNULL); + break; + default: + break; + } + } + + if (ldp_addrcmp(af, &af_conf->trans_addr, &xa->trans_addr)) { + af_conf->trans_addr = xa->trans_addr; + update_sockets = 1; + } + + if (ldpd_process == PROC_MAIN && update_sockets) + imsg_compose_event(iev_ldpe, IMSG_CLOSE_SOCKETS, af, 0, -1, + NULL, 0); +} + +static void +merge_ifaces(struct ldpd_conf *conf, struct ldpd_conf *xconf) +{ + struct iface *iface, *itmp, *xi; + + LIST_FOREACH_SAFE(iface, &conf->iface_list, entry, itmp) { + /* find deleted interfaces */ + if ((xi = if_lookup(xconf, iface->ifindex)) == NULL) { + LIST_REMOVE(iface, entry); + if (ldpd_process == PROC_LDP_ENGINE) + if_exit(iface); + free(iface); + } + } + LIST_FOREACH_SAFE(xi, &xconf->iface_list, entry, itmp) { + /* find new interfaces */ + if ((iface = if_lookup(conf, xi->ifindex)) == NULL) { + LIST_REMOVE(xi, entry); + LIST_INSERT_HEAD(&conf->iface_list, xi, entry); + + /* resend addresses to activate new interfaces */ + if (ldpd_process == PROC_MAIN) + kif_redistribute(xi->name); + continue; + } + + /* update existing interfaces */ + merge_iface_af(&iface->ipv4, &xi->ipv4); + merge_iface_af(&iface->ipv6, &xi->ipv6); + LIST_REMOVE(xi, entry); + free(xi); + } +} + +static void +merge_iface_af(struct iface_af *ia, struct iface_af *xi) +{ + if (ia->enabled != xi->enabled) { + ia->enabled = xi->enabled; + if (ldpd_process == PROC_LDP_ENGINE) + if_update(ia->iface, ia->af); + } + ia->hello_holdtime = xi->hello_holdtime; + ia->hello_interval = xi->hello_interval; +} + +static void +merge_tnbrs(struct ldpd_conf *conf, struct ldpd_conf *xconf) +{ + struct tnbr *tnbr, *ttmp, *xt; + + LIST_FOREACH_SAFE(tnbr, &conf->tnbr_list, entry, ttmp) { + if (!(tnbr->flags & F_TNBR_CONFIGURED)) + continue; + + /* find deleted tnbrs */ + if ((xt = tnbr_find(xconf, tnbr->af, &tnbr->addr)) == NULL) { + if (ldpd_process == PROC_LDP_ENGINE) { + tnbr->flags &= ~F_TNBR_CONFIGURED; + tnbr_check(tnbr); + } else { + LIST_REMOVE(tnbr, entry); + free(tnbr); + } + } + } + LIST_FOREACH_SAFE(xt, &xconf->tnbr_list, entry, ttmp) { + /* find new tnbrs */ + if ((tnbr = tnbr_find(conf, xt->af, &xt->addr)) == NULL) { + LIST_REMOVE(xt, entry); + LIST_INSERT_HEAD(&conf->tnbr_list, xt, entry); + + if (ldpd_process == PROC_LDP_ENGINE) + tnbr_update(xt); + continue; + } + + /* update existing tnbrs */ + if (!(tnbr->flags & F_TNBR_CONFIGURED)) + tnbr->flags |= F_TNBR_CONFIGURED; + tnbr->hello_holdtime = xt->hello_holdtime; + tnbr->hello_interval = xt->hello_interval; + LIST_REMOVE(xt, entry); + free(xt); + } +} + +static void +merge_nbrps(struct ldpd_conf *conf, struct ldpd_conf *xconf) +{ + struct nbr_params *nbrp, *ntmp, *xn; + struct nbr *nbr; + int nbrp_changed; + + LIST_FOREACH_SAFE(nbrp, &conf->nbrp_list, entry, ntmp) { + /* find deleted nbrps */ + if ((xn = nbr_params_find(xconf, nbrp->lsr_id)) == NULL) { + if (ldpd_process == PROC_LDP_ENGINE) { + nbr = nbr_find_ldpid(nbrp->lsr_id.s_addr); + if (nbr) { + session_shutdown(nbr, S_SHUTDOWN, 0, 0); + pfkey_remove(nbr); + if (nbr_session_active_role(nbr)) + nbr_establish_connection(nbr); + } + } + LIST_REMOVE(nbrp, entry); + free(nbrp); + } + } + LIST_FOREACH_SAFE(xn, &xconf->nbrp_list, entry, ntmp) { + /* find new nbrps */ + if ((nbrp = nbr_params_find(conf, xn->lsr_id)) == NULL) { + LIST_REMOVE(xn, entry); + LIST_INSERT_HEAD(&conf->nbrp_list, xn, entry); + + if (ldpd_process == PROC_LDP_ENGINE) { + nbr = nbr_find_ldpid(xn->lsr_id.s_addr); + if (nbr) { + session_shutdown(nbr, S_SHUTDOWN, 0, 0); + if (pfkey_establish(nbr, xn) == -1) + fatalx("pfkey setup failed"); + if (nbr_session_active_role(nbr)) + nbr_establish_connection(nbr); + } + } + continue; + } + + /* update existing nbrps */ + if (nbrp->flags != xn->flags || + nbrp->keepalive != xn->keepalive || + nbrp->gtsm_enabled != xn->gtsm_enabled || + nbrp->gtsm_hops != xn->gtsm_hops || + nbrp->auth.method != xn->auth.method || + strcmp(nbrp->auth.md5key, xn->auth.md5key) != 0) + nbrp_changed = 1; + else + nbrp_changed = 0; + + nbrp->keepalive = xn->keepalive; + nbrp->gtsm_enabled = xn->gtsm_enabled; + nbrp->gtsm_hops = xn->gtsm_hops; + nbrp->auth.method = xn->auth.method; + strlcpy(nbrp->auth.md5key, xn->auth.md5key, + sizeof(nbrp->auth.md5key)); + nbrp->auth.md5key_len = xn->auth.md5key_len; + nbrp->flags = xn->flags; + + if (ldpd_process == PROC_LDP_ENGINE) { + nbr = nbr_find_ldpid(nbrp->lsr_id.s_addr); + if (nbr && nbrp_changed) { + session_shutdown(nbr, S_SHUTDOWN, 0, 0); + pfkey_remove(nbr); + if (pfkey_establish(nbr, nbrp) == -1) + fatalx("pfkey setup failed"); + if (nbr_session_active_role(nbr)) + nbr_establish_connection(nbr); + } + } + LIST_REMOVE(xn, entry); + free(xn); + } +} + +static void +merge_l2vpns(struct ldpd_conf *conf, struct ldpd_conf *xconf) +{ + struct l2vpn *l2vpn, *ltmp, *xl; + + LIST_FOREACH_SAFE(l2vpn, &conf->l2vpn_list, entry, ltmp) { + /* find deleted l2vpns */ + if ((xl = l2vpn_find(xconf, l2vpn->name)) == NULL) { + LIST_REMOVE(l2vpn, entry); + + switch (ldpd_process) { + case PROC_LDE_ENGINE: + l2vpn_exit(l2vpn); + break; + case PROC_LDP_ENGINE: + ldpe_l2vpn_exit(l2vpn); + break; + case PROC_MAIN: + break; + } + l2vpn_del(l2vpn); + } + } + LIST_FOREACH_SAFE(xl, &xconf->l2vpn_list, entry, ltmp) { + /* find new l2vpns */ + if ((l2vpn = l2vpn_find(conf, xl->name)) == NULL) { + LIST_REMOVE(xl, entry); + LIST_INSERT_HEAD(&conf->l2vpn_list, xl, entry); + + switch (ldpd_process) { + case PROC_LDE_ENGINE: + l2vpn_init(xl); + break; + case PROC_LDP_ENGINE: + ldpe_l2vpn_init(xl); + break; + case PROC_MAIN: + break; + } + continue; + } + + /* update existing l2vpns */ + merge_l2vpn(conf, l2vpn, xl); + LIST_REMOVE(xl, entry); + free(xl); + } +} + +static void +merge_l2vpn(struct ldpd_conf *xconf, struct l2vpn *l2vpn, struct l2vpn *xl) +{ + struct l2vpn_if *lif, *ftmp, *xf; + struct l2vpn_pw *pw, *ptmp, *xp; + struct nbr *nbr; + int reset_nbr, reinstall_pwfec, reinstall_tnbr; + int previous_pw_type, previous_mtu; + + previous_pw_type = l2vpn->pw_type; + previous_mtu = l2vpn->mtu; + + /* merge intefaces */ + LIST_FOREACH_SAFE(lif, &l2vpn->if_list, entry, ftmp) { + /* find deleted interfaces */ + if ((xf = l2vpn_if_find(xl, lif->ifindex)) == NULL) { + LIST_REMOVE(lif, entry); + free(lif); + } + } + LIST_FOREACH_SAFE(xf, &xl->if_list, entry, ftmp) { + /* find new interfaces */ + if ((lif = l2vpn_if_find(l2vpn, xf->ifindex)) == NULL) { + LIST_REMOVE(xf, entry); + LIST_INSERT_HEAD(&l2vpn->if_list, xf, entry); + xf->l2vpn = l2vpn; + continue; + } + + LIST_REMOVE(xf, entry); + free(xf); + } + + /* merge pseudowires */ + LIST_FOREACH_SAFE(pw, &l2vpn->pw_list, entry, ptmp) { + /* find deleted pseudowires */ + if ((xp = l2vpn_pw_find(xl, pw->ifindex)) == NULL) { + switch (ldpd_process) { + case PROC_LDE_ENGINE: + l2vpn_pw_exit(pw); + break; + case PROC_LDP_ENGINE: + ldpe_l2vpn_pw_exit(pw); + break; + case PROC_MAIN: + break; + } + + LIST_REMOVE(pw, entry); + free(pw); + } + } + LIST_FOREACH_SAFE(xp, &xl->pw_list, entry, ptmp) { + /* find new pseudowires */ + if ((pw = l2vpn_pw_find(l2vpn, xp->ifindex)) == NULL) { + LIST_REMOVE(xp, entry); + LIST_INSERT_HEAD(&l2vpn->pw_list, xp, entry); + xp->l2vpn = l2vpn; + + switch (ldpd_process) { + case PROC_LDE_ENGINE: + l2vpn_pw_init(xp); + break; + case PROC_LDP_ENGINE: + ldpe_l2vpn_pw_init(xp); + break; + case PROC_MAIN: + break; + } + continue; + } + + /* update existing pseudowire */ + if (pw->af != xp->af || + ldp_addrcmp(pw->af, &pw->addr, &xp->addr)) + reinstall_tnbr = 1; + else + reinstall_tnbr = 0; + + /* changes that require a session restart */ + if ((pw->flags & (F_PW_STATUSTLV_CONF|F_PW_CWORD_CONF)) != + (xp->flags & (F_PW_STATUSTLV_CONF|F_PW_CWORD_CONF))) + reset_nbr = 1; + else + reset_nbr = 0; + + if (l2vpn->pw_type != xl->pw_type || l2vpn->mtu != xl->mtu || + pw->pwid != xp->pwid || reinstall_tnbr || reset_nbr || + pw->lsr_id.s_addr != xp->lsr_id.s_addr) + reinstall_pwfec = 1; + else + reinstall_pwfec = 0; + + if (ldpd_process == PROC_LDP_ENGINE) { + if (reinstall_tnbr) + ldpe_l2vpn_pw_exit(pw); + if (reset_nbr) { + nbr = nbr_find_ldpid(pw->lsr_id.s_addr); + if (nbr && nbr->state == NBR_STA_OPER) + session_shutdown(nbr, S_SHUTDOWN, 0, 0); + } + } + if (ldpd_process == PROC_LDE_ENGINE && + !reset_nbr && reinstall_pwfec) + l2vpn_pw_exit(pw); + pw->lsr_id = xp->lsr_id; + pw->af = xp->af; + pw->addr = xp->addr; + pw->pwid = xp->pwid; + strlcpy(pw->ifname, xp->ifname, sizeof(pw->ifname)); + pw->ifindex = xp->ifindex; + if (xp->flags & F_PW_CWORD_CONF) + pw->flags |= F_PW_CWORD_CONF; + else + pw->flags &= ~F_PW_CWORD_CONF; + if (xp->flags & F_PW_STATUSTLV_CONF) + pw->flags |= F_PW_STATUSTLV_CONF; + else + pw->flags &= ~F_PW_STATUSTLV_CONF; + if (ldpd_process == PROC_LDP_ENGINE && reinstall_tnbr) + ldpe_l2vpn_pw_init(pw); + if (ldpd_process == PROC_LDE_ENGINE && + !reset_nbr && reinstall_pwfec) { + l2vpn->pw_type = xl->pw_type; + l2vpn->mtu = xl->mtu; + l2vpn_pw_init(pw); + l2vpn->pw_type = previous_pw_type; + l2vpn->mtu = previous_mtu; + } + + LIST_REMOVE(xp, entry); + free(xp); + } + + l2vpn->pw_type = xl->pw_type; + l2vpn->mtu = xl->mtu; + strlcpy(l2vpn->br_ifname, xl->br_ifname, sizeof(l2vpn->br_ifname)); + l2vpn->br_ifindex = xl->br_ifindex; +} + +struct ldpd_conf * +config_new_empty(void) +{ + struct ldpd_conf *xconf; + + xconf = calloc(1, sizeof(*xconf)); + if (xconf == NULL) + fatal(NULL); + + LIST_INIT(&xconf->iface_list); + LIST_INIT(&xconf->tnbr_list); + LIST_INIT(&xconf->nbrp_list); + LIST_INIT(&xconf->l2vpn_list); + + return (xconf); +} + +void +config_clear(struct ldpd_conf *conf) +{ + struct ldpd_conf *xconf; + + /* + * Merge current config with an empty config, this will deactivate + * and deallocate all the interfaces, pseudowires and so on. Before + * merging, copy the router-id and other variables to avoid some + * unnecessary operations, like trying to reset the neighborships. + */ + xconf = config_new_empty(); + xconf->ipv4 = conf->ipv4; + xconf->ipv6 = conf->ipv6; + xconf->rtr_id = conf->rtr_id; + xconf->trans_pref = conf->trans_pref; + xconf->flags = conf->flags; + merge_config(conf, xconf); + free(conf); +} diff --git a/ldpd/ldpd.h b/ldpd/ldpd.h new file mode 100644 index 000000000..d32d23c6c --- /dev/null +++ b/ldpd/ldpd.h @@ -0,0 +1,606 @@ +/* $OpenBSD$ */ + +/* + * Copyright (c) 2013, 2016 Renato Westphal + * Copyright (c) 2009 Michele Marchetto + * 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. + */ + +#ifndef _LDPD_H_ +#define _LDPD_H_ + +#include +#include +#include +#include +#include +#include +#include +#include + +#include "ldp.h" + +#define CONF_FILE "/etc/ldpd.conf" +#define LDPD_SOCKET "/var/run/ldpd.sock" +#define LDPD_USER "_ldpd" + +#define LDPD_OPT_VERBOSE 0x00000001 +#define LDPD_OPT_VERBOSE2 0x00000002 +#define LDPD_OPT_NOACTION 0x00000004 + +#define TCP_MD5_KEY_LEN 80 +#define L2VPN_NAME_LEN 32 + +#define RT_BUF_SIZE 16384 +#define MAX_RTSOCK_BUF 128 * 1024 +#define LDP_BACKLOG 128 + +#define F_LDPD_INSERTED 0x0001 +#define F_CONNECTED 0x0002 +#define F_STATIC 0x0004 +#define F_DYNAMIC 0x0008 +#define F_REJECT 0x0010 +#define F_BLACKHOLE 0x0020 +#define F_REDISTRIBUTED 0x0040 + +struct evbuf { + struct msgbuf wbuf; + struct event ev; +}; + +struct imsgev { + struct imsgbuf ibuf; + void (*handler)(int, short, void *); + struct event ev; + short events; +}; + +enum imsg_type { + IMSG_NONE, + IMSG_CTL_RELOAD, + IMSG_CTL_SHOW_INTERFACE, + IMSG_CTL_SHOW_DISCOVERY, + IMSG_CTL_SHOW_NBR, + IMSG_CTL_SHOW_LIB, + IMSG_CTL_SHOW_L2VPN_PW, + IMSG_CTL_SHOW_L2VPN_BINDING, + IMSG_CTL_CLEAR_NBR, + IMSG_CTL_FIB_COUPLE, + IMSG_CTL_FIB_DECOUPLE, + IMSG_CTL_KROUTE, + IMSG_CTL_KROUTE_ADDR, + IMSG_CTL_IFINFO, + IMSG_CTL_END, + IMSG_CTL_LOG_VERBOSE, + IMSG_KLABEL_CHANGE, + IMSG_KLABEL_DELETE, + IMSG_KPWLABEL_CHANGE, + IMSG_KPWLABEL_DELETE, + IMSG_IFSTATUS, + IMSG_NEWADDR, + IMSG_DELADDR, + IMSG_LABEL_MAPPING, + IMSG_LABEL_MAPPING_FULL, + IMSG_LABEL_REQUEST, + IMSG_LABEL_RELEASE, + IMSG_LABEL_WITHDRAW, + IMSG_LABEL_ABORT, + IMSG_REQUEST_ADD, + IMSG_REQUEST_ADD_END, + IMSG_MAPPING_ADD, + IMSG_MAPPING_ADD_END, + IMSG_RELEASE_ADD, + IMSG_RELEASE_ADD_END, + IMSG_WITHDRAW_ADD, + IMSG_WITHDRAW_ADD_END, + IMSG_ADDRESS_ADD, + IMSG_ADDRESS_DEL, + IMSG_NOTIFICATION, + IMSG_NOTIFICATION_SEND, + IMSG_NEIGHBOR_UP, + IMSG_NEIGHBOR_DOWN, + IMSG_NETWORK_ADD, + IMSG_NETWORK_DEL, + IMSG_SOCKET_IPC, + IMSG_SOCKET_NET, + IMSG_CLOSE_SOCKETS, + IMSG_REQUEST_SOCKETS, + IMSG_SETUP_SOCKETS, + IMSG_RECONF_CONF, + IMSG_RECONF_IFACE, + IMSG_RECONF_TNBR, + IMSG_RECONF_NBRP, + IMSG_RECONF_L2VPN, + IMSG_RECONF_L2VPN_IF, + IMSG_RECONF_L2VPN_PW, + IMSG_RECONF_END +}; + +union ldpd_addr { + struct in_addr v4; + struct in6_addr v6; +}; + +#define IN6_IS_SCOPE_EMBED(a) \ + ((IN6_IS_ADDR_LINKLOCAL(a)) || \ + (IN6_IS_ADDR_MC_LINKLOCAL(a)) || \ + (IN6_IS_ADDR_MC_INTFACELOCAL(a))) + +/* interface states */ +#define IF_STA_DOWN 0x01 +#define IF_STA_ACTIVE 0x02 + +/* targeted neighbor states */ +#define TNBR_STA_DOWN 0x01 +#define TNBR_STA_ACTIVE 0x02 + +/* interface types */ +enum iface_type { + IF_TYPE_POINTOPOINT, + IF_TYPE_BROADCAST +}; + +/* neighbor states */ +#define NBR_STA_PRESENT 0x0001 +#define NBR_STA_INITIAL 0x0002 +#define NBR_STA_OPENREC 0x0004 +#define NBR_STA_OPENSENT 0x0008 +#define NBR_STA_OPER 0x0010 +#define NBR_STA_SESSION (NBR_STA_INITIAL | NBR_STA_OPENREC | \ + NBR_STA_OPENSENT | NBR_STA_OPER) + +/* neighbor events */ +enum nbr_event { + NBR_EVT_NOTHING, + NBR_EVT_MATCH_ADJ, + NBR_EVT_CONNECT_UP, + NBR_EVT_CLOSE_SESSION, + NBR_EVT_INIT_RCVD, + NBR_EVT_KEEPALIVE_RCVD, + NBR_EVT_PDU_RCVD, + NBR_EVT_PDU_SENT, + NBR_EVT_INIT_SENT +}; + +/* neighbor actions */ +enum nbr_action { + NBR_ACT_NOTHING, + NBR_ACT_RST_KTIMEOUT, + NBR_ACT_SESSION_EST, + NBR_ACT_RST_KTIMER, + NBR_ACT_CONNECT_SETUP, + NBR_ACT_PASSIVE_INIT, + NBR_ACT_KEEPALIVE_SEND, + NBR_ACT_CLOSE_SESSION +}; + +TAILQ_HEAD(mapping_head, mapping_entry); + +struct map { + uint8_t type; + uint32_t msg_id; + union { + struct { + uint16_t af; + union ldpd_addr prefix; + uint8_t prefixlen; + } prefix; + struct { + uint16_t type; + uint32_t pwid; + uint32_t group_id; + uint16_t ifmtu; + } pwid; + } fec; + struct { + uint32_t status_code; + uint32_t msg_id; + uint16_t msg_type; + } st; + uint32_t label; + uint32_t requestid; + uint32_t pw_status; + uint8_t flags; +}; +#define F_MAP_REQ_ID 0x01 /* optional request message id present */ +#define F_MAP_STATUS 0x02 /* status */ +#define F_MAP_PW_CWORD 0x04 /* pseudowire control word */ +#define F_MAP_PW_ID 0x08 /* pseudowire connection id */ +#define F_MAP_PW_IFMTU 0x10 /* pseudowire interface parameter */ +#define F_MAP_PW_STATUS 0x20 /* pseudowire status */ + +struct notify_msg { + uint32_t status_code; + uint32_t msg_id; /* network byte order */ + uint16_t msg_type; /* network byte order */ + uint32_t pw_status; + struct map fec; + uint8_t flags; +}; +#define F_NOTIF_PW_STATUS 0x01 /* pseudowire status tlv present */ +#define F_NOTIF_FEC 0x02 /* fec tlv present */ + +struct if_addr { + LIST_ENTRY(if_addr) entry; + int af; + union ldpd_addr addr; + uint8_t prefixlen; + union ldpd_addr dstbrd; +}; +LIST_HEAD(if_addr_head, if_addr); + +struct iface_af { + struct iface *iface; + int af; + int enabled; + int state; + LIST_HEAD(, adj) adj_list; + time_t uptime; + struct event hello_timer; + uint16_t hello_holdtime; + uint16_t hello_interval; +}; + +struct iface { + LIST_ENTRY(iface) entry; + char name[IF_NAMESIZE]; + unsigned int ifindex; + struct if_addr_head addr_list; + struct in6_addr linklocal; + enum iface_type type; + uint8_t if_type; + uint16_t flags; + uint8_t linkstate; + struct iface_af ipv4; + struct iface_af ipv6; +}; + +/* source of targeted hellos */ +struct tnbr { + LIST_ENTRY(tnbr) entry; + struct event hello_timer; + struct adj *adj; + int af; + union ldpd_addr addr; + int state; + uint16_t hello_holdtime; + uint16_t hello_interval; + uint16_t pw_count; + uint8_t flags; +}; +#define F_TNBR_CONFIGURED 0x01 +#define F_TNBR_DYNAMIC 0x02 + +enum auth_method { + AUTH_NONE, + AUTH_MD5SIG +}; + +/* neighbor specific parameters */ +struct nbr_params { + LIST_ENTRY(nbr_params) entry; + struct in_addr lsr_id; + uint16_t keepalive; + int gtsm_enabled; + uint8_t gtsm_hops; + struct { + enum auth_method method; + char md5key[TCP_MD5_KEY_LEN]; + uint8_t md5key_len; + } auth; + uint8_t flags; +}; +#define F_NBRP_KEEPALIVE 0x01 +#define F_NBRP_GTSM 0x02 +#define F_NBRP_GTSM_HOPS 0x04 + +struct l2vpn_if { + LIST_ENTRY(l2vpn_if) entry; + struct l2vpn *l2vpn; + char ifname[IF_NAMESIZE]; + unsigned int ifindex; + uint16_t flags; + uint8_t link_state; +}; + +struct l2vpn_pw { + LIST_ENTRY(l2vpn_pw) entry; + struct l2vpn *l2vpn; + struct in_addr lsr_id; + int af; + union ldpd_addr addr; + uint32_t pwid; + char ifname[IF_NAMESIZE]; + unsigned int ifindex; + uint32_t remote_group; + uint16_t remote_mtu; + uint32_t remote_status; + uint8_t flags; +}; +#define F_PW_STATUSTLV_CONF 0x01 /* status tlv configured */ +#define F_PW_STATUSTLV 0x02 /* status tlv negotiated */ +#define F_PW_CWORD_CONF 0x04 /* control word configured */ +#define F_PW_CWORD 0x08 /* control word negotiated */ +#define F_PW_STATUS_UP 0x10 /* pseudowire is operational */ + +struct l2vpn { + LIST_ENTRY(l2vpn) entry; + char name[L2VPN_NAME_LEN]; + int type; + int pw_type; + int mtu; + char br_ifname[IF_NAMESIZE]; + unsigned int br_ifindex; + LIST_HEAD(, l2vpn_if) if_list; + LIST_HEAD(, l2vpn_pw) pw_list; +}; +#define L2VPN_TYPE_VPWS 1 +#define L2VPN_TYPE_VPLS 2 + +/* ldp_conf */ +enum ldpd_process { + PROC_MAIN, + PROC_LDP_ENGINE, + PROC_LDE_ENGINE +} ldpd_process; + +enum socket_type { + LDP_SOCKET_DISC, + LDP_SOCKET_EDISC, + LDP_SOCKET_SESSION +}; + +enum hello_type { + HELLO_LINK, + HELLO_TARGETED +}; + +struct ldpd_af_conf { + uint16_t keepalive; + uint16_t thello_holdtime; + uint16_t thello_interval; + union ldpd_addr trans_addr; + int flags; +}; +#define F_LDPD_AF_ENABLED 0x0001 +#define F_LDPD_AF_THELLO_ACCEPT 0x0002 +#define F_LDPD_AF_EXPNULL 0x0004 +#define F_LDPD_AF_NO_GTSM 0x0008 + +struct ldpd_conf { + struct in_addr rtr_id; + struct ldpd_af_conf ipv4; + struct ldpd_af_conf ipv6; + LIST_HEAD(, iface) iface_list; + LIST_HEAD(, tnbr) tnbr_list; + LIST_HEAD(, nbr_params) nbrp_list; + LIST_HEAD(, l2vpn) l2vpn_list; + uint16_t trans_pref; + int flags; +}; +#define F_LDPD_NO_FIB_UPDATE 0x0001 +#define F_LDPD_DS_CISCO_INTEROP 0x0002 + +struct ldpd_af_global { + struct event disc_ev; + struct event edisc_ev; + int ldp_disc_socket; + int ldp_edisc_socket; + int ldp_session_socket; +}; + +struct ldpd_global { + int cmd_opts; + time_t uptime; + struct ldpd_af_global ipv4; + struct ldpd_af_global ipv6; + uint32_t conf_seqnum; + int pfkeysock; + struct if_addr_head addr_list; + LIST_HEAD(, adj) adj_list; + struct in_addr mcast_addr_v4; + struct in6_addr mcast_addr_v6; + TAILQ_HEAD(, pending_conn) pending_conns; +}; + +/* kroute */ +struct kroute { + int af; + union ldpd_addr prefix; + uint8_t prefixlen; + union ldpd_addr nexthop; + uint32_t local_label; + uint32_t remote_label; + unsigned short ifindex; + uint8_t priority; + uint16_t flags; +}; + +struct kpw { + unsigned short ifindex; + int pw_type; + int af; + union ldpd_addr nexthop; + uint32_t local_label; + uint32_t remote_label; + uint8_t flags; +}; + +struct kaddr { + unsigned short ifindex; + int af; + union ldpd_addr addr; + uint8_t prefixlen; + union ldpd_addr dstbrd; +}; + +struct kif { + char ifname[IF_NAMESIZE]; + unsigned short ifindex; + int flags; + uint8_t link_state; + int mtu; + uint8_t if_type; + uint64_t baudrate; +}; + +/* control data structures */ +struct ctl_iface { + int af; + char name[IF_NAMESIZE]; + unsigned int ifindex; + int state; + uint16_t flags; + uint8_t linkstate; + enum iface_type type; + uint8_t if_type; + uint16_t hello_holdtime; + uint16_t hello_interval; + time_t uptime; + uint16_t adj_cnt; +}; + +struct ctl_adj { + int af; + struct in_addr id; + enum hello_type type; + char ifname[IF_NAMESIZE]; + union ldpd_addr src_addr; + uint16_t holdtime; + union ldpd_addr trans_addr; +}; + +struct ctl_nbr { + int af; + struct in_addr id; + union ldpd_addr laddr; + union ldpd_addr raddr; + time_t uptime; + int nbr_state; +}; + +struct ctl_rt { + int af; + union ldpd_addr prefix; + uint8_t prefixlen; + struct in_addr nexthop; /* lsr-id */ + uint32_t local_label; + uint32_t remote_label; + uint8_t flags; + uint8_t in_use; +}; + +struct ctl_pw { + uint16_t type; + char ifname[IF_NAMESIZE]; + uint32_t pwid; + struct in_addr lsr_id; + uint32_t local_label; + uint32_t local_gid; + uint16_t local_ifmtu; + uint32_t remote_label; + uint32_t remote_gid; + uint16_t remote_ifmtu; + uint32_t status; +}; + +extern struct ldpd_conf *ldpd_conf; +extern struct ldpd_global global; + +/* parse.y */ +struct ldpd_conf *parse_config(char *); +int cmdline_symset(char *); + +/* kroute.c */ +int kif_init(void); +int kr_init(int); +void kif_redistribute(const char *); +int kr_change(struct kroute *); +int kr_delete(struct kroute *); +void kr_shutdown(void); +void kr_fib_couple(void); +void kr_fib_decouple(void); +void kr_change_egress_label(int, int); +void kr_show_route(struct imsg *); +void kr_ifinfo(char *, pid_t); +struct kif *kif_findname(char *); +void kif_clear(void); +int kmpw_set(struct kpw *); +int kmpw_unset(struct kpw *); + +/* util.c */ +uint8_t mask2prefixlen(in_addr_t); +uint8_t mask2prefixlen6(struct sockaddr_in6 *); +in_addr_t prefixlen2mask(uint8_t); +struct in6_addr *prefixlen2mask6(uint8_t); +void ldp_applymask(int, union ldpd_addr *, + const union ldpd_addr *, int); +int ldp_addrcmp(int, const union ldpd_addr *, + const union ldpd_addr *); +int ldp_addrisset(int, const union ldpd_addr *); +int ldp_prefixcmp(int, const union ldpd_addr *, + const union ldpd_addr *, uint8_t); +int bad_addr_v4(struct in_addr); +int bad_addr_v6(struct in6_addr *); +int bad_addr(int, union ldpd_addr *); +void embedscope(struct sockaddr_in6 *); +void recoverscope(struct sockaddr_in6 *); +void addscope(struct sockaddr_in6 *, uint32_t); +void clearscope(struct in6_addr *); +struct sockaddr *addr2sa(int af, union ldpd_addr *, uint16_t); +void sa2addr(struct sockaddr *, int *, union ldpd_addr *); + +/* ldpd.c */ +void main_imsg_compose_ldpe(int, pid_t, void *, uint16_t); +void main_imsg_compose_lde(int, pid_t, void *, uint16_t); +void imsg_event_add(struct imsgev *); +int imsg_compose_event(struct imsgev *, uint16_t, uint32_t, pid_t, + int, void *, uint16_t); +void evbuf_enqueue(struct evbuf *, struct ibuf *); +void evbuf_event_add(struct evbuf *); +void evbuf_init(struct evbuf *, int, void (*)(int, short, void *), void *); +void evbuf_clear(struct evbuf *); +struct ldpd_af_conf *ldp_af_conf_get(struct ldpd_conf *, int); +struct ldpd_af_global *ldp_af_global_get(struct ldpd_global *, int); +int ldp_is_dual_stack(struct ldpd_conf *); +void merge_config(struct ldpd_conf *, struct ldpd_conf *); +struct ldpd_conf *config_new_empty(void); +void config_clear(struct ldpd_conf *); + +/* socket.c */ +int ldp_create_socket(int, enum socket_type); +void sock_set_recvbuf(int); +int sock_set_reuse(int, int); +int sock_set_bindany(int, int); +int sock_set_ipv4_tos(int, int); +int sock_set_ipv4_recvif(int, int); +int sock_set_ipv4_minttl(int, int); +int sock_set_ipv4_ucast_ttl(int fd, int); +int sock_set_ipv4_mcast_ttl(int, uint8_t); +int sock_set_ipv4_mcast(struct iface *); +int sock_set_ipv4_mcast_loop(int); +int sock_set_ipv6_dscp(int, int); +int sock_set_ipv6_pktinfo(int, int); +int sock_set_ipv6_minhopcount(int, int); +int sock_set_ipv6_ucast_hops(int, int); +int sock_set_ipv6_mcast_hops(int, int); +int sock_set_ipv6_mcast(struct iface *); +int sock_set_ipv6_mcast_loop(int); + +/* printconf.c */ +void print_config(struct ldpd_conf *); + +#endif /* _LDPD_H_ */ diff --git a/ldpd/ldpe.c b/ldpd/ldpe.c new file mode 100644 index 000000000..1b4d2a67e --- /dev/null +++ b/ldpd/ldpe.c @@ -0,0 +1,808 @@ +/* $OpenBSD$ */ + +/* + * Copyright (c) 2013, 2016 Renato Westphal + * Copyright (c) 2005 Claudio Jeker + * Copyright (c) 2004, 2008 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 "ldpd.h" +#include "ldpe.h" +#include "lde.h" +#include "control.h" +#include "log.h" + +static void ldpe_sig_handler(int, short, void *); +static __dead void ldpe_shutdown(void); +static void ldpe_dispatch_main(int, short, void *); +static void ldpe_dispatch_lde(int, short, void *); +static void ldpe_dispatch_pfkey(int, short, void *); +static void ldpe_setup_sockets(int, int, int, int); +static void ldpe_close_sockets(int); +static void ldpe_iface_af_ctl(struct ctl_conn *, int, unsigned int); + +struct ldpd_conf *leconf; +struct ldpd_sysdep sysdep; + +static struct imsgev *iev_main; +static struct imsgev *iev_lde; +static struct event pfkey_ev; + +/* ARGSUSED */ +static void +ldpe_sig_handler(int sig, short event, void *bula) +{ + switch (sig) { + case SIGINT: + case SIGTERM: + ldpe_shutdown(); + /* NOTREACHED */ + default: + fatalx("unexpected signal"); + } +} + +/* label distribution protocol engine */ +void +ldpe(int debug, int verbose) +{ + struct passwd *pw; + struct event ev_sigint, ev_sigterm; + + leconf = config_new_empty(); + + log_init(debug); + log_verbose(verbose); + + setproctitle("ldp engine"); + ldpd_process = PROC_LDP_ENGINE; + + /* create ldpd control socket outside chroot */ + if (control_init() == -1) + fatalx("control socket setup failed"); + + LIST_INIT(&global.addr_list); + LIST_INIT(&global.adj_list); + TAILQ_INIT(&global.pending_conns); + if (inet_pton(AF_INET, AllRouters_v4, &global.mcast_addr_v4) != 1) + fatal("inet_pton"); + if (inet_pton(AF_INET6, AllRouters_v6, &global.mcast_addr_v6) != 1) + fatal("inet_pton"); + global.pfkeysock = pfkey_init(); + + if ((pw = getpwnam(LDPD_USER)) == NULL) + fatal("getpwnam"); + + if (chroot(pw->pw_dir) == -1) + fatal("chroot"); + if (chdir("/") == -1) + fatal("chdir(\"/\")"); + + 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"); + + if (pledge("stdio cpath inet mcast recvfd", NULL) == -1) + fatal("pledge"); + + event_init(); + accept_init(); + + /* setup signal handler */ + signal_set(&ev_sigint, SIGINT, ldpe_sig_handler, NULL); + signal_set(&ev_sigterm, SIGTERM, ldpe_sig_handler, NULL); + signal_add(&ev_sigint, NULL); + signal_add(&ev_sigterm, NULL); + signal(SIGPIPE, SIG_IGN); + signal(SIGHUP, SIG_IGN); + + /* setup pipe and event handler to the parent process */ + if ((iev_main = malloc(sizeof(struct imsgev))) == NULL) + fatal(NULL); + imsg_init(&iev_main->ibuf, 3); + iev_main->handler = ldpe_dispatch_main; + 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); + + if (sysdep.no_pfkey == 0) { + event_set(&pfkey_ev, global.pfkeysock, EV_READ | EV_PERSIST, + ldpe_dispatch_pfkey, NULL); + event_add(&pfkey_ev, NULL); + } + + /* mark sockets as closed */ + global.ipv4.ldp_disc_socket = -1; + global.ipv4.ldp_edisc_socket = -1; + global.ipv4.ldp_session_socket = -1; + global.ipv6.ldp_disc_socket = -1; + global.ipv6.ldp_edisc_socket = -1; + global.ipv6.ldp_session_socket = -1; + + /* listen on ldpd control socket */ + TAILQ_INIT(&ctl_conns); + control_listen(); + + if ((pkt_ptr = calloc(1, IBUF_READ_SIZE)) == NULL) + fatal(__func__); + + event_dispatch(); + + ldpe_shutdown(); +} + +static __dead void +ldpe_shutdown(void) +{ + struct if_addr *if_addr; + struct adj *adj; + + /* close pipes */ + msgbuf_write(&iev_lde->ibuf.w); + msgbuf_clear(&iev_lde->ibuf.w); + close(iev_lde->ibuf.fd); + msgbuf_write(&iev_main->ibuf.w); + msgbuf_clear(&iev_main->ibuf.w); + close(iev_main->ibuf.fd); + + control_cleanup(); + config_clear(leconf); + + if (sysdep.no_pfkey == 0) { + event_del(&pfkey_ev); + close(global.pfkeysock); + } + ldpe_close_sockets(AF_INET); + ldpe_close_sockets(AF_INET6); + + /* remove addresses from global list */ + while ((if_addr = LIST_FIRST(&global.addr_list)) != NULL) { + LIST_REMOVE(if_addr, entry); + free(if_addr); + } + while ((adj = LIST_FIRST(&global.adj_list)) != NULL) + adj_del(adj, S_SHUTDOWN); + + /* clean up */ + free(iev_lde); + free(iev_main); + free(pkt_ptr); + + log_info("ldp engine exiting"); + exit(0); +} + +/* imesg */ +int +ldpe_imsg_compose_parent(int type, pid_t pid, void *data, uint16_t datalen) +{ + return (imsg_compose_event(iev_main, type, 0, pid, -1, data, datalen)); +} + +int +ldpe_imsg_compose_lde(int type, uint32_t peerid, pid_t pid, void *data, + uint16_t datalen) +{ + return (imsg_compose_event(iev_lde, type, peerid, pid, -1, + data, datalen)); +} + +/* ARGSUSED */ +static void +ldpe_dispatch_main(int fd, short event, void *bula) +{ + static struct ldpd_conf *nconf; + struct iface *niface; + struct tnbr *ntnbr; + struct nbr_params *nnbrp; + static struct l2vpn *nl2vpn; + struct l2vpn_if *nlif; + struct l2vpn_pw *npw; + struct imsg imsg; + struct imsgev *iev = bula; + struct imsgbuf *ibuf = &iev->ibuf; + struct iface *iface = NULL; + struct kif *kif; + int af; + enum socket_type *socket_type; + static int disc_socket = -1; + static int edisc_socket = -1; + static int session_socket = -1; + struct nbr *nbr; + struct nbr_params *nbrp; + int n, shut = 0; + + if (event & EV_READ) { + if ((n = imsg_read(ibuf)) == -1 && errno != EAGAIN) + fatal("imsg_read error"); + if (n == 0) /* connection closed */ + shut = 1; + } + if (event & EV_WRITE) { + if ((n = msgbuf_write(&ibuf->w)) == -1 && errno != EAGAIN) + fatal("ldpe_dispatch_main: msgbuf_write"); + if (n == 0) + shut = 1; + } + + for (;;) { + if ((n = imsg_get(ibuf, &imsg)) == -1) + fatal("ldpe_dispatch_main: imsg_get error"); + if (n == 0) + break; + + switch (imsg.hdr.type) { + case IMSG_IFSTATUS: + if (imsg.hdr.len != IMSG_HEADER_SIZE + + sizeof(struct kif)) + fatalx("IFSTATUS imsg with wrong len"); + kif = imsg.data; + + iface = if_lookup(leconf, kif->ifindex); + if (!iface) + break; + + iface->flags = kif->flags; + iface->linkstate = kif->link_state; + if_update(iface, AF_UNSPEC); + break; + case IMSG_NEWADDR: + if (imsg.hdr.len != IMSG_HEADER_SIZE + + sizeof(struct kaddr)) + fatalx("NEWADDR imsg with wrong len"); + + if_addr_add(imsg.data); + break; + case IMSG_DELADDR: + if (imsg.hdr.len != IMSG_HEADER_SIZE + + sizeof(struct kaddr)) + fatalx("DELADDR imsg with wrong len"); + + if_addr_del(imsg.data); + break; + case IMSG_SOCKET_IPC: + if (iev_lde) { + log_warnx("%s: received unexpected imsg fd " + "to lde", __func__); + break; + } + if ((fd = imsg.fd) == -1) { + log_warnx("%s: expected to receive imsg fd to " + "lde but didn't receive any", __func__); + break; + } + + if ((iev_lde = malloc(sizeof(struct imsgev))) == NULL) + fatal(NULL); + imsg_init(&iev_lde->ibuf, fd); + iev_lde->handler = ldpe_dispatch_lde; + iev_lde->events = EV_READ; + event_set(&iev_lde->ev, iev_lde->ibuf.fd, + iev_lde->events, iev_lde->handler, iev_lde); + event_add(&iev_lde->ev, NULL); + break; + case IMSG_CLOSE_SOCKETS: + af = imsg.hdr.peerid; + + RB_FOREACH(nbr, nbr_id_head, &nbrs_by_id) { + if (nbr->af != af) + continue; + session_shutdown(nbr, S_SHUTDOWN, 0, 0); + pfkey_remove(nbr); + } + ldpe_close_sockets(af); + if_update_all(af); + tnbr_update_all(af); + + disc_socket = -1; + edisc_socket = -1; + session_socket = -1; + if ((ldp_af_conf_get(leconf, af))->flags & + F_LDPD_AF_ENABLED) + ldpe_imsg_compose_parent(IMSG_REQUEST_SOCKETS, + af, NULL, 0); + break; + case IMSG_SOCKET_NET: + if (imsg.hdr.len != IMSG_HEADER_SIZE + + sizeof(enum socket_type)) + fatalx("SOCKET_NET imsg with wrong len"); + socket_type = imsg.data; + + switch (*socket_type) { + case LDP_SOCKET_DISC: + disc_socket = imsg.fd; + break; + case LDP_SOCKET_EDISC: + edisc_socket = imsg.fd; + break; + case LDP_SOCKET_SESSION: + session_socket = imsg.fd; + break; + } + break; + case IMSG_SETUP_SOCKETS: + af = imsg.hdr.peerid; + if (disc_socket == -1 || edisc_socket == -1 || + session_socket == -1) { + if (disc_socket != -1) + close(disc_socket); + if (edisc_socket != -1) + close(edisc_socket); + if (session_socket != -1) + close(session_socket); + break; + } + + ldpe_setup_sockets(af, disc_socket, edisc_socket, + session_socket); + if_update_all(af); + tnbr_update_all(af); + RB_FOREACH(nbr, nbr_id_head, &nbrs_by_id) { + if (nbr->af != af) + continue; + nbr->laddr = (ldp_af_conf_get(leconf, + af))->trans_addr; + nbrp = nbr_params_find(leconf, nbr->id); + if (nbrp && pfkey_establish(nbr, nbrp) == -1) + fatalx("pfkey setup failed"); + if (nbr_session_active_role(nbr)) + nbr_establish_connection(nbr); + } + break; + case IMSG_RECONF_CONF: + if ((nconf = malloc(sizeof(struct ldpd_conf))) == + NULL) + fatal(NULL); + memcpy(nconf, imsg.data, sizeof(struct ldpd_conf)); + + LIST_INIT(&nconf->iface_list); + LIST_INIT(&nconf->tnbr_list); + LIST_INIT(&nconf->nbrp_list); + LIST_INIT(&nconf->l2vpn_list); + break; + case IMSG_RECONF_IFACE: + if ((niface = malloc(sizeof(struct iface))) == NULL) + fatal(NULL); + memcpy(niface, imsg.data, sizeof(struct iface)); + + LIST_INIT(&niface->addr_list); + LIST_INIT(&niface->ipv4.adj_list); + LIST_INIT(&niface->ipv6.adj_list); + niface->ipv4.iface = niface; + niface->ipv6.iface = niface; + + LIST_INSERT_HEAD(&nconf->iface_list, niface, entry); + break; + case IMSG_RECONF_TNBR: + if ((ntnbr = malloc(sizeof(struct tnbr))) == NULL) + fatal(NULL); + memcpy(ntnbr, imsg.data, sizeof(struct tnbr)); + + LIST_INSERT_HEAD(&nconf->tnbr_list, ntnbr, entry); + break; + case IMSG_RECONF_NBRP: + if ((nnbrp = malloc(sizeof(struct nbr_params))) == NULL) + fatal(NULL); + memcpy(nnbrp, imsg.data, sizeof(struct nbr_params)); + + LIST_INSERT_HEAD(&nconf->nbrp_list, nnbrp, entry); + break; + case IMSG_RECONF_L2VPN: + if ((nl2vpn = malloc(sizeof(struct l2vpn))) == NULL) + fatal(NULL); + memcpy(nl2vpn, imsg.data, sizeof(struct l2vpn)); + + LIST_INIT(&nl2vpn->if_list); + LIST_INIT(&nl2vpn->pw_list); + + LIST_INSERT_HEAD(&nconf->l2vpn_list, nl2vpn, entry); + break; + case IMSG_RECONF_L2VPN_IF: + if ((nlif = malloc(sizeof(struct l2vpn_if))) == NULL) + fatal(NULL); + memcpy(nlif, imsg.data, sizeof(struct l2vpn_if)); + + nlif->l2vpn = nl2vpn; + LIST_INSERT_HEAD(&nl2vpn->if_list, nlif, entry); + break; + case IMSG_RECONF_L2VPN_PW: + if ((npw = malloc(sizeof(struct l2vpn_pw))) == NULL) + fatal(NULL); + memcpy(npw, imsg.data, sizeof(struct l2vpn_pw)); + + npw->l2vpn = nl2vpn; + LIST_INSERT_HEAD(&nl2vpn->pw_list, npw, entry); + break; + case IMSG_RECONF_END: + merge_config(leconf, nconf); + nconf = NULL; + global.conf_seqnum++; + 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("ldpe_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 */ +static void +ldpe_dispatch_lde(int fd, short event, void *bula) +{ + struct imsgev *iev = bula; + struct imsgbuf *ibuf = &iev->ibuf; + struct imsg imsg; + struct map map; + struct notify_msg nm; + int n, shut = 0; + struct nbr *nbr = NULL; + + if (event & EV_READ) { + if ((n = imsg_read(ibuf)) == -1 && errno != EAGAIN) + fatal("imsg_read error"); + if (n == 0) /* connection closed */ + shut = 1; + } + if (event & EV_WRITE) { + if ((n = msgbuf_write(&ibuf->w)) == -1 && errno != EAGAIN) + fatal("ldpe_dispatch_lde: msgbuf_write"); + if (n == 0) + shut = 1; + } + + for (;;) { + if ((n = imsg_get(ibuf, &imsg)) == -1) + fatal("ldpe_dispatch_lde: imsg_get error"); + if (n == 0) + break; + + switch (imsg.hdr.type) { + case IMSG_MAPPING_ADD: + case IMSG_RELEASE_ADD: + case IMSG_REQUEST_ADD: + case IMSG_WITHDRAW_ADD: + if (imsg.hdr.len - IMSG_HEADER_SIZE != sizeof(map)) + fatalx("invalid size of map request"); + memcpy(&map, imsg.data, sizeof(map)); + + nbr = nbr_find_peerid(imsg.hdr.peerid); + if (nbr == NULL) { + log_debug("ldpe_dispatch_lde: cannot find " + "neighbor"); + break; + } + if (nbr->state != NBR_STA_OPER) + break; + + switch (imsg.hdr.type) { + case IMSG_MAPPING_ADD: + mapping_list_add(&nbr->mapping_list, &map); + break; + case IMSG_RELEASE_ADD: + mapping_list_add(&nbr->release_list, &map); + break; + case IMSG_REQUEST_ADD: + mapping_list_add(&nbr->request_list, &map); + break; + case IMSG_WITHDRAW_ADD: + mapping_list_add(&nbr->withdraw_list, &map); + break; + } + break; + case IMSG_MAPPING_ADD_END: + case IMSG_RELEASE_ADD_END: + case IMSG_REQUEST_ADD_END: + case IMSG_WITHDRAW_ADD_END: + nbr = nbr_find_peerid(imsg.hdr.peerid); + if (nbr == NULL) { + log_debug("ldpe_dispatch_lde: cannot find " + "neighbor"); + break; + } + if (nbr->state != NBR_STA_OPER) + break; + + switch (imsg.hdr.type) { + case IMSG_MAPPING_ADD_END: + send_labelmessage(nbr, MSG_TYPE_LABELMAPPING, + &nbr->mapping_list); + break; + case IMSG_RELEASE_ADD_END: + send_labelmessage(nbr, MSG_TYPE_LABELRELEASE, + &nbr->release_list); + break; + case IMSG_REQUEST_ADD_END: + send_labelmessage(nbr, MSG_TYPE_LABELREQUEST, + &nbr->request_list); + break; + case IMSG_WITHDRAW_ADD_END: + send_labelmessage(nbr, MSG_TYPE_LABELWITHDRAW, + &nbr->withdraw_list); + break; + } + break; + case IMSG_NOTIFICATION_SEND: + if (imsg.hdr.len - IMSG_HEADER_SIZE != sizeof(nm)) + fatalx("invalid size of OE request"); + memcpy(&nm, imsg.data, sizeof(nm)); + + nbr = nbr_find_peerid(imsg.hdr.peerid); + if (nbr == NULL) { + log_debug("ldpe_dispatch_lde: cannot find " + "neighbor"); + break; + } + if (nbr->state != NBR_STA_OPER) + break; + + send_notification_full(nbr->tcp, &nm); + break; + case IMSG_CTL_END: + case IMSG_CTL_SHOW_LIB: + case IMSG_CTL_SHOW_L2VPN_PW: + case IMSG_CTL_SHOW_L2VPN_BINDING: + control_imsg_relay(&imsg); + break; + default: + log_debug("ldpe_dispatch_lde: 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 */ +static void +ldpe_dispatch_pfkey(int fd, short event, void *bula) +{ + if (event & EV_READ) { + if (pfkey_read(fd, NULL) == -1) { + fatal("pfkey_read failed, exiting..."); + } + } +} + +static void +ldpe_setup_sockets(int af, int disc_socket, int edisc_socket, + int session_socket) +{ + struct ldpd_af_global *af_global; + + af_global = ldp_af_global_get(&global, af); + + /* discovery socket */ + af_global->ldp_disc_socket = disc_socket; + event_set(&af_global->disc_ev, af_global->ldp_disc_socket, + EV_READ|EV_PERSIST, disc_recv_packet, NULL); + event_add(&af_global->disc_ev, NULL); + + /* extended discovery socket */ + af_global->ldp_edisc_socket = edisc_socket; + event_set(&af_global->edisc_ev, af_global->ldp_edisc_socket, + EV_READ|EV_PERSIST, disc_recv_packet, NULL); + event_add(&af_global->edisc_ev, NULL); + + /* session socket */ + af_global->ldp_session_socket = session_socket; + accept_add(af_global->ldp_session_socket, session_accept, NULL); +} + +static void +ldpe_close_sockets(int af) +{ + struct ldpd_af_global *af_global; + + af_global = ldp_af_global_get(&global, af); + + /* discovery socket */ + if (event_initialized(&af_global->disc_ev)) + event_del(&af_global->disc_ev); + if (af_global->ldp_disc_socket != -1) { + close(af_global->ldp_disc_socket); + af_global->ldp_disc_socket = -1; + } + + /* extended discovery socket */ + if (event_initialized(&af_global->edisc_ev)) + event_del(&af_global->edisc_ev); + if (af_global->ldp_edisc_socket != -1) { + close(af_global->ldp_edisc_socket); + af_global->ldp_edisc_socket = -1; + } + + /* session socket */ + if (af_global->ldp_session_socket != -1) { + accept_del(af_global->ldp_session_socket); + close(af_global->ldp_session_socket); + af_global->ldp_session_socket = -1; + } +} + +void +ldpe_reset_nbrs(int af) +{ + struct nbr *nbr; + + RB_FOREACH(nbr, nbr_id_head, &nbrs_by_id) { + if (nbr->af == af) + session_shutdown(nbr, S_SHUTDOWN, 0, 0); + } +} + +void +ldpe_reset_ds_nbrs(void) +{ + struct nbr *nbr; + + RB_FOREACH(nbr, nbr_id_head, &nbrs_by_id) { + if (nbr->ds_tlv) + session_shutdown(nbr, S_SHUTDOWN, 0, 0); + } +} + +void +ldpe_remove_dynamic_tnbrs(int af) +{ + struct tnbr *tnbr, *safe; + + LIST_FOREACH_SAFE(tnbr, &leconf->tnbr_list, entry, safe) { + if (tnbr->af != af) + continue; + + tnbr->flags &= ~F_TNBR_DYNAMIC; + tnbr_check(tnbr); + } +} + +void +ldpe_stop_init_backoff(int af) +{ + struct nbr *nbr; + + RB_FOREACH(nbr, nbr_id_head, &nbrs_by_id) { + if (nbr->af == af && nbr_pending_idtimer(nbr)) { + nbr_stop_idtimer(nbr); + nbr_establish_connection(nbr); + } + } +} + +static void +ldpe_iface_af_ctl(struct ctl_conn *c, int af, unsigned int idx) +{ + struct iface *iface; + struct iface_af *ia; + struct ctl_iface *ictl; + + LIST_FOREACH(iface, &leconf->iface_list, entry) { + if (idx == 0 || idx == iface->ifindex) { + ia = iface_af_get(iface, af); + if (!ia->enabled) + continue; + + ictl = if_to_ctl(ia); + imsg_compose_event(&c->iev, + IMSG_CTL_SHOW_INTERFACE, + 0, 0, -1, ictl, sizeof(struct ctl_iface)); + } + } +} + +void +ldpe_iface_ctl(struct ctl_conn *c, unsigned int idx) +{ + ldpe_iface_af_ctl(c, AF_INET, idx); + ldpe_iface_af_ctl(c, AF_INET6, idx); +} + +void +ldpe_adj_ctl(struct ctl_conn *c) +{ + struct nbr *nbr; + struct adj *adj; + struct ctl_adj *actl; + + RB_FOREACH(nbr, nbr_addr_head, &nbrs_by_addr) { + LIST_FOREACH(adj, &nbr->adj_list, nbr_entry) { + actl = adj_to_ctl(adj); + imsg_compose_event(&c->iev, IMSG_CTL_SHOW_DISCOVERY, + 0, 0, -1, actl, sizeof(struct ctl_adj)); + } + } + /* show adjacencies not associated with any neighbor */ + LIST_FOREACH(adj, &global.adj_list, global_entry) { + if (adj->nbr != NULL) + continue; + + actl = adj_to_ctl(adj); + imsg_compose_event(&c->iev, IMSG_CTL_SHOW_DISCOVERY, 0, 0, + -1, actl, sizeof(struct ctl_adj)); + } + + imsg_compose_event(&c->iev, IMSG_CTL_END, 0, 0, -1, NULL, 0); +} + +void +ldpe_nbr_ctl(struct ctl_conn *c) +{ + struct nbr *nbr; + struct ctl_nbr *nctl; + + RB_FOREACH(nbr, nbr_addr_head, &nbrs_by_addr) { + 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 +mapping_list_add(struct mapping_head *mh, struct map *map) +{ + struct mapping_entry *me; + + me = calloc(1, sizeof(*me)); + if (me == NULL) + fatal(__func__); + me->map = *map; + + TAILQ_INSERT_TAIL(mh, me, entry); +} + +void +mapping_list_clr(struct mapping_head *mh) +{ + struct mapping_entry *me; + + while ((me = TAILQ_FIRST(mh)) != NULL) { + TAILQ_REMOVE(mh, me, entry); + free(me); + } +} diff --git a/ldpd/ldpe.h b/ldpd/ldpe.h new file mode 100644 index 000000000..e640dd67d --- /dev/null +++ b/ldpd/ldpe.h @@ -0,0 +1,282 @@ +/* $OpenBSD$ */ + +/* + * Copyright (c) 2013, 2016 Renato Westphal + * Copyright (c) 2009 Michele Marchetto + * Copyright (c) 2004, 2005, 2008 Esben Norby + * + * 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. + */ + +#ifndef _LDPE_H_ +#define _LDPE_H_ + +#include +#include +#include +#include + +#include "ldpd.h" + +#define min(x,y) ((x) <= (y) ? (x) : (y)) +#define max(x,y) ((x) > (y) ? (x) : (y)) + +struct hello_source { + enum hello_type type; + struct { + struct iface_af *ia; + union ldpd_addr src_addr; + } link; + struct tnbr *target; +}; + +struct adj { + LIST_ENTRY(adj) global_entry; + LIST_ENTRY(adj) nbr_entry; + LIST_ENTRY(adj) ia_entry; + struct in_addr lsr_id; + struct nbr *nbr; + int ds_tlv; + struct hello_source source; + struct event inactivity_timer; + uint16_t holdtime; + union ldpd_addr trans_addr; +}; + +struct tcp_conn { + struct nbr *nbr; + int fd; + struct ibuf_read *rbuf; + struct evbuf wbuf; + struct event rev; +}; + +struct nbr { + RB_ENTRY(nbr) id_tree, addr_tree, pid_tree; + struct tcp_conn *tcp; + LIST_HEAD(, adj) adj_list; /* adjacencies */ + struct event ev_connect; + struct event keepalive_timer; + struct event keepalive_timeout; + struct event init_timeout; + struct event initdelay_timer; + + struct mapping_head mapping_list; + struct mapping_head withdraw_list; + struct mapping_head request_list; + struct mapping_head release_list; + struct mapping_head abortreq_list; + + uint32_t peerid; /* unique ID in DB */ + int af; + int ds_tlv; + int v4_enabled; /* announce/process v4 msgs */ + int v6_enabled; /* announce/process v6 msgs */ + struct in_addr id; /* lsr id */ + union ldpd_addr laddr; /* local address */ + union ldpd_addr raddr; /* remote address */ + uint32_t raddr_scope; /* remote address scope (v6) */ + time_t uptime; + int fd; + int state; + uint32_t conf_seqnum; + int idtimer_cnt; + uint16_t keepalive; + uint16_t max_pdu_len; + + struct { + uint8_t established; + uint32_t spi_in; + uint32_t spi_out; + enum auth_method method; + char md5key[TCP_MD5_KEY_LEN]; + } auth; + int flags; +}; +#define F_NBR_GTSM_NEGOTIATED 0x01 + +RB_HEAD(nbr_id_head, nbr); +RB_PROTOTYPE(nbr_id_head, nbr, id_tree, nbr_id_compare) +RB_HEAD(nbr_addr_head, nbr); +RB_PROTOTYPE(nbr_addr_head, nbr, addr_tree, nbr_addr_compare) +RB_HEAD(nbr_pid_head, nbr); +RB_PROTOTYPE(nbr_pid_head, nbr, pid_tree, nbr_pid_compare) + +struct pending_conn { + TAILQ_ENTRY(pending_conn) entry; + int fd; + int af; + union ldpd_addr addr; + struct event ev_timeout; +}; +#define PENDING_CONN_TIMEOUT 5 + +struct mapping_entry { + TAILQ_ENTRY(mapping_entry) entry; + struct map map; +}; + +struct ldpd_sysdep { + uint8_t no_pfkey; + uint8_t no_md5sig; +}; + +extern struct ldpd_conf *leconf; +extern struct ldpd_sysdep sysdep; +extern struct nbr_id_head nbrs_by_id; +extern struct nbr_addr_head nbrs_by_addr; +extern struct nbr_pid_head nbrs_by_pid; + +/* accept.c */ +void accept_init(void); +int accept_add(int, void (*)(int, short, void *), void *); +void accept_del(int); +void accept_pause(void); +void accept_unpause(void); + +/* hello.c */ +int send_hello(enum hello_type, struct iface_af *, struct tnbr *); +void recv_hello(struct in_addr, struct ldp_msg *, int, union ldpd_addr *, + struct iface *, int, char *, uint16_t); + +/* init.c */ +void send_init(struct nbr *); +int recv_init(struct nbr *, char *, uint16_t); + +/* keepalive.c */ +void send_keepalive(struct nbr *); +int recv_keepalive(struct nbr *, char *, uint16_t); + +/* notification.c */ +void send_notification_full(struct tcp_conn *, struct notify_msg *); +void send_notification(uint32_t, struct tcp_conn *, uint32_t, + uint16_t); +void send_notification_nbr(struct nbr *, uint32_t, uint32_t, uint16_t); +int recv_notification(struct nbr *, char *, uint16_t); +int gen_status_tlv(struct ibuf *, uint32_t, uint32_t, uint16_t); + +/* address.c */ +void send_address_single(struct nbr *, struct if_addr *, int); +void send_address_all(struct nbr *, int); +int recv_address(struct nbr *, char *, uint16_t); + +/* labelmapping.c */ +#define PREFIX_SIZE(x) (((x) + 7) / 8) +void send_labelmessage(struct nbr *, uint16_t, struct mapping_head *); +int recv_labelmessage(struct nbr *, char *, uint16_t, uint16_t); +int gen_pw_status_tlv(struct ibuf *, uint32_t); +int gen_fec_tlv(struct ibuf *, struct map *); +int tlv_decode_fec_elm(struct nbr *, struct ldp_msg *, char *, + uint16_t, struct map *); + +/* ldpe.c */ +void ldpe(int, int); +int ldpe_imsg_compose_parent(int, pid_t, void *, + uint16_t); +int ldpe_imsg_compose_lde(int, uint32_t, pid_t, void *, + uint16_t); +void ldpe_reset_nbrs(int); +void ldpe_reset_ds_nbrs(void); +void ldpe_remove_dynamic_tnbrs(int); +void ldpe_stop_init_backoff(int); +struct ctl_conn; +void ldpe_iface_ctl(struct ctl_conn *, unsigned int); +void ldpe_adj_ctl(struct ctl_conn *); +void ldpe_nbr_ctl(struct ctl_conn *); +void mapping_list_add(struct mapping_head *, struct map *); +void mapping_list_clr(struct mapping_head *); + +/* interface.c */ +struct iface *if_new(struct kif *); +void if_exit(struct iface *); +struct iface *if_lookup(struct ldpd_conf *, unsigned short); +struct iface_af *iface_af_get(struct iface *, int); +void if_addr_add(struct kaddr *); +void if_addr_del(struct kaddr *); +void if_update(struct iface *, int); +void if_update_all(int); +struct ctl_iface *if_to_ctl(struct iface_af *); +in_addr_t if_get_ipv4_addr(struct iface *); + +/* adjacency.c */ +struct adj *adj_new(struct in_addr, struct hello_source *, + union ldpd_addr *); +void adj_del(struct adj *, uint32_t); +struct adj *adj_find(struct hello_source *); +int adj_get_af(struct adj *adj); +void adj_start_itimer(struct adj *); +void adj_stop_itimer(struct adj *); +struct tnbr *tnbr_new(struct ldpd_conf *, int, union ldpd_addr *); +struct tnbr *tnbr_find(struct ldpd_conf *, int, union ldpd_addr *); +struct tnbr *tnbr_check(struct tnbr *); +void tnbr_update(struct tnbr *); +void tnbr_update_all(int); +struct ctl_adj *adj_to_ctl(struct adj *); + +/* neighbor.c */ +int nbr_fsm(struct nbr *, enum nbr_event); +struct nbr *nbr_new(struct in_addr, int, int, union ldpd_addr *, + uint32_t); +void nbr_del(struct nbr *); +struct nbr *nbr_find_ldpid(uint32_t); +struct nbr *nbr_find_addr(int, union ldpd_addr *); +struct nbr *nbr_find_peerid(uint32_t); +int nbr_adj_count(struct nbr *, int); +int nbr_session_active_role(struct nbr *); +void nbr_stop_ktimer(struct nbr *); +void nbr_stop_ktimeout(struct nbr *); +void nbr_stop_itimeout(struct nbr *); +void nbr_start_idtimer(struct nbr *); +void nbr_stop_idtimer(struct nbr *); +int nbr_pending_idtimer(struct nbr *); +int nbr_pending_connect(struct nbr *); +int nbr_establish_connection(struct nbr *); +int nbr_gtsm_enabled(struct nbr *, struct nbr_params *); +int nbr_gtsm_setup(int, int, struct nbr_params *); +int nbr_gtsm_check(int, struct nbr *, struct nbr_params *); +struct nbr_params *nbr_params_new(struct in_addr); +struct nbr_params *nbr_params_find(struct ldpd_conf *, struct in_addr); +uint16_t nbr_get_keepalive(int, struct in_addr); +struct ctl_nbr *nbr_to_ctl(struct nbr *); +void nbr_clear_ctl(struct ctl_nbr *); + +/* packet.c */ +int gen_ldp_hdr(struct ibuf *, uint16_t); +int gen_msg_hdr(struct ibuf *, uint16_t, uint16_t); +int send_packet(int, int, union ldpd_addr *, + struct iface_af *, void *, size_t); +void disc_recv_packet(int, short, void *); +void session_accept(int, short, void *); +void session_accept_nbr(struct nbr *, int); +void session_shutdown(struct nbr *, uint32_t, uint32_t, + uint32_t); +void session_close(struct nbr *); +struct tcp_conn *tcp_new(int, struct nbr *); +void pending_conn_del(struct pending_conn *); +struct pending_conn *pending_conn_find(int, union ldpd_addr *); + +char *pkt_ptr; /* packet buffer */ + +/* pfkey.c */ +int pfkey_read(int, struct sadb_msg *); +int pfkey_establish(struct nbr *, struct nbr_params *); +int pfkey_remove(struct nbr *); +int pfkey_init(void); + +/* l2vpn.c */ +void ldpe_l2vpn_init(struct l2vpn *); +void ldpe_l2vpn_exit(struct l2vpn *); +void ldpe_l2vpn_pw_init(struct l2vpn_pw *); +void ldpe_l2vpn_pw_exit(struct l2vpn_pw *); + +#endif /* _LDPE_H_ */ diff --git a/ldpd/log.c b/ldpd/log.c new file mode 100644 index 000000000..e14b6e51e --- /dev/null +++ b/ldpd/log.c @@ -0,0 +1,600 @@ +/* $OpenBSD$ */ + +/* + * 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 "ldpd.h" +#include "ldpe.h" +#include "lde.h" +#include "log.h" + +static const char * const procnames[] = { + "parent", + "ldpe", + "lde" +}; + +static void vlog(int, const char *, va_list); + +static int debug; +static int verbose; + +void +log_init(int n_debug) +{ + extern char *__progname; + + debug = n_debug; + + if (!debug) + openlog(__progname, LOG_PID | LOG_NDELAY, LOG_DAEMON); + + tzset(); +} + +void +log_verbose(int v) +{ + verbose = v; +} + +void +logit(int pri, const char *fmt, ...) +{ + va_list ap; + + va_start(ap, fmt); + vlog(pri, fmt, ap); + va_end(ap); +} + +static void +vlog(int pri, const char *fmt, va_list ap) +{ + char *nfmt; + + if (debug) { + /* best effort in out of mem situations */ + if (asprintf(&nfmt, "%s\n", fmt) == -1) { + vfprintf(stderr, fmt, ap); + fprintf(stderr, "\n"); + } else { + vfprintf(stderr, nfmt, ap); + free(nfmt); + } + fflush(stderr); + } else + vsyslog(pri, fmt, ap); +} + +void +log_warn(const char *emsg, ...) +{ + char *nfmt; + va_list ap; + + /* best effort to even work in out of memory situations */ + if (emsg == NULL) + logit(LOG_CRIT, "%s", strerror(errno)); + else { + va_start(ap, emsg); + + if (asprintf(&nfmt, "%s: %s", emsg, strerror(errno)) == -1) { + /* we tried it... */ + vlog(LOG_CRIT, emsg, ap); + logit(LOG_CRIT, "%s", strerror(errno)); + } else { + vlog(LOG_CRIT, nfmt, ap); + free(nfmt); + } + va_end(ap); + } +} + +void +log_warnx(const char *emsg, ...) +{ + va_list ap; + + va_start(ap, emsg); + vlog(LOG_CRIT, emsg, ap); + va_end(ap); +} + +void +log_info(const char *emsg, ...) +{ + va_list ap; + + va_start(ap, emsg); + vlog(LOG_INFO, emsg, ap); + va_end(ap); +} + +void +log_debug(const char *emsg, ...) +{ + va_list ap; + + if (verbose & LDPD_OPT_VERBOSE) { + va_start(ap, emsg); + vlog(LOG_DEBUG, emsg, ap); + va_end(ap); + } +} + +void +fatal(const char *emsg) +{ + if (emsg == NULL) + logit(LOG_CRIT, "fatal in %s: %s", procnames[ldpd_process], + strerror(errno)); + else + if (errno) + logit(LOG_CRIT, "fatal in %s: %s: %s", + procnames[ldpd_process], emsg, strerror(errno)); + else + logit(LOG_CRIT, "fatal in %s: %s", + procnames[ldpd_process], emsg); + + exit(1); +} + +void +fatalx(const char *emsg) +{ + errno = 0; + fatal(emsg); +} + +#define NUM_LOGS 4 +const char * +log_sockaddr(void *vp) +{ + static char buf[NUM_LOGS][NI_MAXHOST]; + static int round = 0; + struct sockaddr *sa = vp; + + round = (round + 1) % NUM_LOGS; + + if (getnameinfo(sa, sa->sa_len, buf[round], NI_MAXHOST, NULL, 0, + NI_NUMERICHOST)) + return ("(unknown)"); + else + return (buf[round]); +} + +const char * +log_in6addr(const struct in6_addr *addr) +{ + struct sockaddr_in6 sa_in6; + + memset(&sa_in6, 0, sizeof(sa_in6)); + sa_in6.sin6_len = sizeof(sa_in6); + sa_in6.sin6_family = AF_INET6; + sa_in6.sin6_addr = *addr; + + recoverscope(&sa_in6); + + return (log_sockaddr(&sa_in6)); +} + +const char * +log_in6addr_scope(const struct in6_addr *addr, unsigned int ifindex) +{ + struct sockaddr_in6 sa_in6; + + memset(&sa_in6, 0, sizeof(sa_in6)); + sa_in6.sin6_len = sizeof(sa_in6); + sa_in6.sin6_family = AF_INET6; + sa_in6.sin6_addr = *addr; + + addscope(&sa_in6, ifindex); + + return (log_sockaddr(&sa_in6)); +} + +const char * +log_addr(int af, const union ldpd_addr *addr) +{ + static char buf[NUM_LOGS][INET6_ADDRSTRLEN]; + static int round = 0; + + switch (af) { + case AF_INET: + round = (round + 1) % NUM_LOGS; + if (inet_ntop(AF_INET, &addr->v4, buf[round], + sizeof(buf[round])) == NULL) + return ("???"); + return (buf[round]); + case AF_INET6: + return (log_in6addr(&addr->v6)); + default: + break; + } + + return ("???"); +} + +#define TF_BUFS 4 +#define TF_LEN 32 + +char * +log_label(uint32_t label) +{ + char *buf; + static char tfbuf[TF_BUFS][TF_LEN]; /* ring buffer */ + static int idx = 0; + + buf = tfbuf[idx++]; + if (idx == TF_BUFS) + idx = 0; + + switch (label) { + case NO_LABEL: + snprintf(buf, TF_LEN, "-"); + break; + case MPLS_LABEL_IMPLNULL: + snprintf(buf, TF_LEN, "imp-null"); + break; + case MPLS_LABEL_IPV4NULL: + case MPLS_LABEL_IPV6NULL: + snprintf(buf, TF_LEN, "exp-null"); + break; + default: + snprintf(buf, TF_LEN, "%u", label); + break; + } + + return (buf); +} + +char * +log_hello_src(const struct hello_source *src) +{ + static char buf[64]; + + switch (src->type) { + case HELLO_LINK: + snprintf(buf, sizeof(buf), "iface %s", + src->link.ia->iface->name); + break; + case HELLO_TARGETED: + snprintf(buf, sizeof(buf), "source %s", + log_addr(src->target->af, &src->target->addr)); + break; + } + + return (buf); +} + +const char * +log_map(const struct map *map) +{ + static char buf[64]; + + switch (map->type) { + case MAP_TYPE_WILDCARD: + if (snprintf(buf, sizeof(buf), "wildcard") < 0) + return ("???"); + break; + case MAP_TYPE_PREFIX: + if (snprintf(buf, sizeof(buf), "%s/%u", + log_addr(map->fec.prefix.af, &map->fec.prefix.prefix), + map->fec.prefix.prefixlen) == -1) + return ("???"); + break; + case MAP_TYPE_PWID: + if (snprintf(buf, sizeof(buf), "pwid %u (%s)", + map->fec.pwid.pwid, + pw_type_name(map->fec.pwid.type)) == -1) + return ("???"); + break; + default: + return ("???"); + } + + return (buf); +} + +const char * +log_fec(const struct fec *fec) +{ + static char buf[64]; + union ldpd_addr addr; + + switch (fec->type) { + case FEC_TYPE_IPV4: + addr.v4 = fec->u.ipv4.prefix; + if (snprintf(buf, sizeof(buf), "ipv4 %s/%u", + log_addr(AF_INET, &addr), fec->u.ipv4.prefixlen) == -1) + return ("???"); + break; + case FEC_TYPE_IPV6: + addr.v6 = fec->u.ipv6.prefix; + if (snprintf(buf, sizeof(buf), "ipv6 %s/%u", + log_addr(AF_INET6, &addr), fec->u.ipv6.prefixlen) == -1) + return ("???"); + break; + case FEC_TYPE_PWID: + if (snprintf(buf, sizeof(buf), + "pwid %u (%s) - %s", + fec->u.pwid.pwid, pw_type_name(fec->u.pwid.type), + inet_ntoa(fec->u.pwid.lsr_id)) == -1) + return ("???"); + break; + default: + return ("???"); + } + + return (buf); +} + +/* names */ +const char * +af_name(int af) +{ + switch (af) { + case AF_INET: + return ("ipv4"); + case AF_INET6: + return ("ipv6"); + case AF_MPLS: + return ("mpls"); + default: + return ("UNKNOWN"); + } +} + +const char * +socket_name(int type) +{ + switch (type) { + case LDP_SOCKET_DISC: + return ("discovery"); + case LDP_SOCKET_EDISC: + return ("extended discovery"); + case LDP_SOCKET_SESSION: + return ("session"); + default: + return ("UNKNOWN"); + } +} + +const char * +nbr_state_name(int state) +{ + switch (state) { + case NBR_STA_PRESENT: + return ("PRESENT"); + case NBR_STA_INITIAL: + return ("INITIALIZED"); + case NBR_STA_OPENREC: + return ("OPENREC"); + case NBR_STA_OPENSENT: + return ("OPENSENT"); + case NBR_STA_OPER: + return ("OPERATIONAL"); + default: + return ("UNKNOWN"); + } +} + +const char * +if_state_name(int state) +{ + switch (state) { + case IF_STA_DOWN: + return ("DOWN"); + case IF_STA_ACTIVE: + return ("ACTIVE"); + default: + return ("UNKNOWN"); + } +} + +const char * +if_type_name(enum iface_type type) +{ + switch (type) { + case IF_TYPE_POINTOPOINT: + return ("POINTOPOINT"); + case IF_TYPE_BROADCAST: + return ("BROADCAST"); + } + /* NOTREACHED */ + return ("UNKNOWN"); +} + +const char * +msg_name(uint16_t msg) +{ + static char buf[16]; + + switch (msg) { + case MSG_TYPE_NOTIFICATION: + return ("notification"); + case MSG_TYPE_HELLO: + return ("hello"); + case MSG_TYPE_INIT: + return ("initialization"); + case MSG_TYPE_KEEPALIVE: + return ("keepalive"); + case MSG_TYPE_ADDR: + return ("address"); + case MSG_TYPE_ADDRWITHDRAW: + return ("address withdraw"); + case MSG_TYPE_LABELMAPPING: + return ("label mapping"); + case MSG_TYPE_LABELREQUEST: + return ("label request"); + case MSG_TYPE_LABELWITHDRAW: + return ("label withdraw"); + case MSG_TYPE_LABELRELEASE: + return ("label release"); + case MSG_TYPE_LABELABORTREQ: + default: + snprintf(buf, sizeof(buf), "[%08x]", msg); + return (buf); + } +} + +const char * +status_code_name(uint32_t status) +{ + static char buf[16]; + + switch (status) { + case S_SUCCESS: + return ("Success"); + case S_BAD_LDP_ID: + return ("Bad LDP Identifier"); + case S_BAD_PROTO_VER: + return ("Bad Protocol Version"); + case S_BAD_PDU_LEN: + return ("Bad PDU Length"); + case S_UNKNOWN_MSG: + return ("Unknown Message Type"); + case S_BAD_MSG_LEN: + return ("Bad Message Length"); + case S_UNKNOWN_TLV: + return ("Unknown TLV"); + case S_BAD_TLV_LEN: + return ("Bad TLV Length"); + case S_BAD_TLV_VAL: + return ("Malformed TLV Value"); + case S_HOLDTIME_EXP: + return ("Hold Timer Expired"); + case S_SHUTDOWN: + return ("Shutdown"); + case S_LOOP_DETECTED: + return ("Loop Detected"); + case S_UNKNOWN_FEC: + return ("Unknown FEC"); + case S_NO_ROUTE: + return ("No Route"); + case S_NO_LABEL_RES: + return ("No Label Resources"); + case S_AVAILABLE: + return ("Label Resources Available"); + case S_NO_HELLO: + return ("Session Rejected, No Hello"); + case S_PARM_ADV_MODE: + return ("Rejected Advertisement Mode Parameter"); + case S_MAX_PDU_LEN: + return ("Rejected Max PDU Length Parameter"); + case S_PARM_L_RANGE: + return ("Rejected Label Range Parameter"); + case S_KEEPALIVE_TMR: + return ("KeepAlive Timer Expired"); + case S_LAB_REQ_ABRT: + return ("Label Request Aborted"); + case S_MISS_MSG: + return ("Missing Message Parameters"); + case S_UNSUP_ADDR: + return ("Unsupported Address Family"); + case S_KEEPALIVE_BAD: + return ("Bad KeepAlive Time"); + case S_INTERN_ERR: + return ("Internal Error"); + case S_ILLEGAL_CBIT: + return ("Illegal C-Bit"); + case S_WRONG_CBIT: + return ("Wrong C-Bit"); + case S_INCPT_BITRATE: + return ("Incompatible bit-rate"); + case S_CEP_MISCONF: + return ("CEP-TDM mis-configuration"); + case S_PW_STATUS: + return ("PW Status"); + case S_UNASSIGN_TAI: + return ("Unassigned/Unrecognized TAI"); + case S_MISCONF_ERR: + return ("Generic Misconfiguration Error"); + case S_WITHDRAW_MTHD: + return ("Label Withdraw PW Status Method"); + case S_TRANS_MISMTCH: + return ("Transport Connection Mismatch"); + case S_DS_NONCMPLNCE: + return ("Dual-Stack Noncompliance"); + default: + snprintf(buf, sizeof(buf), "[%08x]", status); + return (buf); + } +} + +const char * +pw_type_name(uint16_t pw_type) +{ + static char buf[64]; + + switch (pw_type) { + case PW_TYPE_ETHERNET_TAGGED: + return ("Eth Tagged"); + case PW_TYPE_ETHERNET: + return ("Ethernet"); + default: + snprintf(buf, sizeof(buf), "[%0x]", pw_type); + return (buf); + } +} + +static char *msgtypes[] = { + "", + "RTM_ADD: Add Route", + "RTM_DELETE: Delete Route", + "RTM_CHANGE: Change Metrics or flags", + "RTM_GET: Report Metrics", + "RTM_LOSING: Kernel Suspects Partitioning", + "RTM_REDIRECT: Told to use different route", + "RTM_MISS: Lookup failed on this address", + "RTM_LOCK: fix specified metrics", + "RTM_OLDADD: caused by SIOCADDRT", + "RTM_OLDDEL: caused by SIOCDELRT", + "RTM_RESOLVE: Route created by cloning", + "RTM_NEWADDR: address being added to iface", + "RTM_DELADDR: address being removed from iface", + "RTM_IFINFO: iface status change", + "RTM_IFANNOUNCE: iface arrival/departure", + "RTM_DESYNC: route socket overflow", +}; + +void +log_rtmsg(unsigned char rtm_type) +{ + if (!(verbose & LDPD_OPT_VERBOSE2)) + return; + + if (rtm_type > 0 && + rtm_type < sizeof(msgtypes)/sizeof(msgtypes[0])) + log_debug("kernel message: %s", msgtypes[rtm_type]); + else + log_debug("kernel message: rtm_type %d out of range", + rtm_type); +} diff --git a/ldpd/log.h b/ldpd/log.h new file mode 100644 index 000000000..94c463041 --- /dev/null +++ b/ldpd/log.h @@ -0,0 +1,63 @@ +/* $OpenBSD$ */ + +/* + * 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. + */ + +#ifndef _LOG_H_ +#define _LOG_H_ + +#include + +struct in6_addr; +union ldpd_addr; +struct hello_source; +struct fec; + +void log_init(int); +void log_verbose(int); +void logit(int, const char *, ...) + __attribute__((__format__ (printf, 2, 3))); +void log_warn(const char *, ...) + __attribute__((__format__ (printf, 1, 2))); +void log_warnx(const char *, ...) + __attribute__((__format__ (printf, 1, 2))); +void log_info(const char *, ...) + __attribute__((__format__ (printf, 1, 2))); +void log_debug(const char *, ...) + __attribute__((__format__ (printf, 1, 2))); +void fatal(const char *) __dead + __attribute__((__format__ (printf, 1, 0))); +void fatalx(const char *) __dead + __attribute__((__format__ (printf, 1, 0))); +const char *log_sockaddr(void *); +const char *log_in6addr(const struct in6_addr *); +const char *log_in6addr_scope(const struct in6_addr *, unsigned int); +const char *log_addr(int, const union ldpd_addr *); +char *log_label(uint32_t); +char *log_hello_src(const struct hello_source *); +const char *log_map(const struct map *); +const char *log_fec(const struct fec *); +const char *af_name(int); +const char *socket_name(int); +const char *nbr_state_name(int); +const char *if_state_name(int); +const char *if_type_name(enum iface_type); +const char *msg_name(uint16_t); +const char *status_code_name(uint32_t); +const char *pw_type_name(uint16_t); +void log_rtmsg(unsigned char); + +#endif /* _LOG_H_ */ diff --git a/ldpd/neighbor.c b/ldpd/neighbor.c new file mode 100644 index 000000000..d3f83734f --- /dev/null +++ b/ldpd/neighbor.c @@ -0,0 +1,827 @@ +/* $OpenBSD$ */ + +/* + * Copyright (c) 2013, 2016 Renato Westphal + * Copyright (c) 2009 Michele Marchetto + * Copyright (c) 2005 Claudio Jeker + * Copyright (c) 2004, 2005, 2008 Esben Norby + * + * 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 "ldpd.h" +#include "ldpe.h" +#include "lde.h" +#include "log.h" + +static __inline int nbr_id_compare(struct nbr *, struct nbr *); +static __inline int nbr_addr_compare(struct nbr *, struct nbr *); +static __inline int nbr_pid_compare(struct nbr *, struct nbr *); +static void nbr_update_peerid(struct nbr *); +static void nbr_ktimer(int, short, void *); +static void nbr_start_ktimer(struct nbr *); +static void nbr_ktimeout(int, short, void *); +static void nbr_start_ktimeout(struct nbr *); +static void nbr_itimeout(int, short, void *); +static void nbr_start_itimeout(struct nbr *); +static void nbr_idtimer(int, short, void *); +static int nbr_act_session_operational(struct nbr *); +static void nbr_send_labelmappings(struct nbr *); + +RB_GENERATE(nbr_id_head, nbr, id_tree, nbr_id_compare) +RB_GENERATE(nbr_addr_head, nbr, addr_tree, nbr_addr_compare) +RB_GENERATE(nbr_pid_head, nbr, pid_tree, nbr_pid_compare) + +struct { + int state; + enum nbr_event event; + enum nbr_action action; + int new_state; +} nbr_fsm_tbl[] = { + /* current state event that happened action to take resulting state */ +/* Passive Role */ + {NBR_STA_PRESENT, NBR_EVT_MATCH_ADJ, NBR_ACT_NOTHING, NBR_STA_INITIAL}, + {NBR_STA_INITIAL, NBR_EVT_INIT_RCVD, NBR_ACT_PASSIVE_INIT, NBR_STA_OPENREC}, + {NBR_STA_OPENREC, NBR_EVT_KEEPALIVE_RCVD, NBR_ACT_SESSION_EST, NBR_STA_OPER}, +/* Active Role */ + {NBR_STA_PRESENT, NBR_EVT_CONNECT_UP, NBR_ACT_CONNECT_SETUP, NBR_STA_INITIAL}, + {NBR_STA_INITIAL, NBR_EVT_INIT_SENT, NBR_ACT_NOTHING, NBR_STA_OPENSENT}, + {NBR_STA_OPENSENT, NBR_EVT_INIT_RCVD, NBR_ACT_KEEPALIVE_SEND, NBR_STA_OPENREC}, +/* Session Maintenance */ + {NBR_STA_OPER, NBR_EVT_PDU_RCVD, NBR_ACT_RST_KTIMEOUT, 0}, + {NBR_STA_SESSION, NBR_EVT_PDU_RCVD, NBR_ACT_NOTHING, 0}, + {NBR_STA_OPER, NBR_EVT_PDU_SENT, NBR_ACT_RST_KTIMER, 0}, + {NBR_STA_SESSION, NBR_EVT_PDU_SENT, NBR_ACT_NOTHING, 0}, +/* Session Close */ + {NBR_STA_PRESENT, NBR_EVT_CLOSE_SESSION, NBR_ACT_NOTHING, 0}, + {NBR_STA_SESSION, NBR_EVT_CLOSE_SESSION, NBR_ACT_CLOSE_SESSION, NBR_STA_PRESENT}, + {-1, NBR_EVT_NOTHING, NBR_ACT_NOTHING, 0}, +}; + +const char * const nbr_event_names[] = { + "NOTHING", + "ADJACENCY MATCHED", + "CONNECTION UP", + "SESSION CLOSE", + "INIT RECEIVED", + "KEEPALIVE RECEIVED", + "PDU RECEIVED", + "PDU SENT", + "INIT SENT" +}; + +const char * const nbr_action_names[] = { + "NOTHING", + "RESET KEEPALIVE TIMEOUT", + "START NEIGHBOR SESSION", + "RESET KEEPALIVE TIMER", + "SETUP NEIGHBOR CONNECTION", + "SEND INIT AND KEEPALIVE", + "SEND KEEPALIVE", + "CLOSE SESSION" +}; + +struct nbr_id_head nbrs_by_id = RB_INITIALIZER(&nbrs_by_id); +struct nbr_addr_head nbrs_by_addr = RB_INITIALIZER(&nbrs_by_addr); +struct nbr_pid_head nbrs_by_pid = RB_INITIALIZER(&nbrs_by_pid); + +static __inline int +nbr_id_compare(struct nbr *a, struct nbr *b) +{ + return (ntohl(a->id.s_addr) - ntohl(b->id.s_addr)); +} + +static __inline int +nbr_addr_compare(struct nbr *a, struct nbr *b) +{ + if (a->af < b->af) + return (-1); + if (a->af > b->af) + return (1); + + return (ldp_addrcmp(a->af, &a->raddr, &b->raddr)); +} + +static __inline int +nbr_pid_compare(struct nbr *a, struct nbr *b) +{ + return (a->peerid - b->peerid); +} + +int +nbr_fsm(struct nbr *nbr, enum nbr_event event) +{ + struct timeval now; + int old_state; + int new_state = 0; + int i; + + old_state = nbr->state; + for (i = 0; nbr_fsm_tbl[i].state != -1; i++) + if ((nbr_fsm_tbl[i].state & old_state) && + (nbr_fsm_tbl[i].event == event)) { + new_state = nbr_fsm_tbl[i].new_state; + break; + } + + if (nbr_fsm_tbl[i].state == -1) { + /* event outside of the defined fsm, ignore it. */ + log_warnx("%s: lsr-id %s, event %s not expected in " + "state %s", __func__, inet_ntoa(nbr->id), + nbr_event_names[event], nbr_state_name(old_state)); + return (0); + } + + if (new_state != 0) + nbr->state = new_state; + + if (old_state != nbr->state) { + log_debug("%s: event %s resulted in action %s and " + "changing state for lsr-id %s from %s to %s", + __func__, nbr_event_names[event], + nbr_action_names[nbr_fsm_tbl[i].action], + inet_ntoa(nbr->id), nbr_state_name(old_state), + nbr_state_name(nbr->state)); + + if (nbr->state == NBR_STA_OPER) { + gettimeofday(&now, NULL); + nbr->uptime = now.tv_sec; + } + } + + if (nbr->state == NBR_STA_OPER || nbr->state == NBR_STA_PRESENT) + nbr_stop_itimeout(nbr); + else + nbr_start_itimeout(nbr); + + switch (nbr_fsm_tbl[i].action) { + case NBR_ACT_RST_KTIMEOUT: + nbr_start_ktimeout(nbr); + break; + case NBR_ACT_RST_KTIMER: + nbr_start_ktimer(nbr); + break; + case NBR_ACT_SESSION_EST: + nbr_act_session_operational(nbr); + nbr_start_ktimer(nbr); + nbr_start_ktimeout(nbr); + if (nbr->v4_enabled) + send_address_all(nbr, AF_INET); + if (nbr->v6_enabled) + send_address_all(nbr, AF_INET6); + nbr_send_labelmappings(nbr); + break; + case NBR_ACT_CONNECT_SETUP: + nbr->tcp = tcp_new(nbr->fd, nbr); + + /* trigger next state */ + send_init(nbr); + nbr_fsm(nbr, NBR_EVT_INIT_SENT); + break; + case NBR_ACT_PASSIVE_INIT: + send_init(nbr); + send_keepalive(nbr); + break; + case NBR_ACT_KEEPALIVE_SEND: + nbr_start_ktimeout(nbr); + send_keepalive(nbr); + break; + case NBR_ACT_CLOSE_SESSION: + ldpe_imsg_compose_lde(IMSG_NEIGHBOR_DOWN, nbr->peerid, 0, + NULL, 0); + session_close(nbr); + break; + case NBR_ACT_NOTHING: + /* do nothing */ + break; + } + + return (0); +} + +struct nbr * +nbr_new(struct in_addr id, int af, int ds_tlv, union ldpd_addr *addr, + uint32_t scope_id) +{ + struct nbr *nbr; + struct nbr_params *nbrp; + struct adj *adj; + struct pending_conn *pconn; + + log_debug("%s: lsr-id %s transport-address %s", __func__, + inet_ntoa(id), log_addr(af, addr)); + + if ((nbr = calloc(1, sizeof(*nbr))) == NULL) + fatal(__func__); + + LIST_INIT(&nbr->adj_list); + nbr->state = NBR_STA_PRESENT; + nbr->peerid = 0; + nbr->af = af; + nbr->ds_tlv = ds_tlv; + if (af == AF_INET || ds_tlv) + nbr->v4_enabled = 1; + if (af == AF_INET6 || ds_tlv) + nbr->v6_enabled = 1; + nbr->id = id; + nbr->laddr = (ldp_af_conf_get(leconf, af))->trans_addr; + nbr->raddr = *addr; + nbr->raddr_scope = scope_id; + nbr->conf_seqnum = 0; + + LIST_FOREACH(adj, &global.adj_list, global_entry) { + if (adj->lsr_id.s_addr == nbr->id.s_addr) { + adj->nbr = nbr; + LIST_INSERT_HEAD(&nbr->adj_list, adj, nbr_entry); + } + } + + if (RB_INSERT(nbr_id_head, &nbrs_by_id, nbr) != NULL) + fatalx("nbr_new: RB_INSERT(nbrs_by_id) failed"); + if (RB_INSERT(nbr_addr_head, &nbrs_by_addr, nbr) != NULL) + fatalx("nbr_new: RB_INSERT(nbrs_by_addr) failed"); + + TAILQ_INIT(&nbr->mapping_list); + TAILQ_INIT(&nbr->withdraw_list); + TAILQ_INIT(&nbr->request_list); + TAILQ_INIT(&nbr->release_list); + TAILQ_INIT(&nbr->abortreq_list); + + /* set event structures */ + evtimer_set(&nbr->keepalive_timeout, nbr_ktimeout, nbr); + evtimer_set(&nbr->keepalive_timer, nbr_ktimer, nbr); + evtimer_set(&nbr->init_timeout, nbr_itimeout, nbr); + evtimer_set(&nbr->initdelay_timer, nbr_idtimer, nbr); + + nbrp = nbr_params_find(leconf, nbr->id); + if (nbrp && pfkey_establish(nbr, nbrp) == -1) + fatalx("pfkey setup failed"); + + pconn = pending_conn_find(nbr->af, &nbr->raddr); + if (pconn) { + session_accept_nbr(nbr, pconn->fd); + pending_conn_del(pconn); + } + + return (nbr); +} + +void +nbr_del(struct nbr *nbr) +{ + log_debug("%s: lsr-id %s", __func__, inet_ntoa(nbr->id)); + + nbr_fsm(nbr, NBR_EVT_CLOSE_SESSION); + pfkey_remove(nbr); + + if (nbr_pending_connect(nbr)) + event_del(&nbr->ev_connect); + nbr_stop_ktimer(nbr); + nbr_stop_ktimeout(nbr); + nbr_stop_itimeout(nbr); + nbr_stop_idtimer(nbr); + + mapping_list_clr(&nbr->mapping_list); + mapping_list_clr(&nbr->withdraw_list); + mapping_list_clr(&nbr->request_list); + mapping_list_clr(&nbr->release_list); + mapping_list_clr(&nbr->abortreq_list); + + if (nbr->peerid) + RB_REMOVE(nbr_pid_head, &nbrs_by_pid, nbr); + RB_REMOVE(nbr_id_head, &nbrs_by_id, nbr); + RB_REMOVE(nbr_addr_head, &nbrs_by_addr, nbr); + + free(nbr); +} + +static void +nbr_update_peerid(struct nbr *nbr) +{ + static uint32_t peercnt = 1; + + if (nbr->peerid) + RB_REMOVE(nbr_pid_head, &nbrs_by_pid, nbr); + + /* get next unused peerid */ + while (nbr_find_peerid(++peercnt)) + ; + nbr->peerid = peercnt; + + if (RB_INSERT(nbr_pid_head, &nbrs_by_pid, nbr) != NULL) + fatalx("nbr_update_peerid: RB_INSERT(nbrs_by_pid) failed"); +} + +struct nbr * +nbr_find_ldpid(uint32_t lsr_id) +{ + struct nbr n; + n.id.s_addr = lsr_id; + return (RB_FIND(nbr_id_head, &nbrs_by_id, &n)); +} + +struct nbr * +nbr_find_addr(int af, union ldpd_addr *addr) +{ + struct nbr n; + n.af = af; + n.raddr = *addr; + return (RB_FIND(nbr_addr_head, &nbrs_by_addr, &n)); +} + +struct nbr * +nbr_find_peerid(uint32_t peerid) +{ + struct nbr n; + n.peerid = peerid; + return (RB_FIND(nbr_pid_head, &nbrs_by_pid, &n)); +} + +int +nbr_adj_count(struct nbr *nbr, int af) +{ + struct adj *adj; + int total = 0; + + LIST_FOREACH(adj, &nbr->adj_list, nbr_entry) + if (adj_get_af(adj) == af) + total++; + + return (total); +} + +int +nbr_session_active_role(struct nbr *nbr) +{ + if (ldp_addrcmp(nbr->af, &nbr->laddr, &nbr->raddr) > 0) + return (1); + + return (0); +} + +/* timers */ + +/* Keepalive timer: timer to send keepalive message to neighbors */ + +static void +nbr_ktimer(int fd, short event, void *arg) +{ + struct nbr *nbr = arg; + + send_keepalive(nbr); + nbr_start_ktimer(nbr); +} + +static void +nbr_start_ktimer(struct nbr *nbr) +{ + struct timeval tv; + + /* send three keepalives per period */ + timerclear(&tv); + tv.tv_sec = (time_t)(nbr->keepalive / KEEPALIVE_PER_PERIOD); + if (evtimer_add(&nbr->keepalive_timer, &tv) == -1) + fatal(__func__); +} + +void +nbr_stop_ktimer(struct nbr *nbr) +{ + if (evtimer_pending(&nbr->keepalive_timer, NULL) && + evtimer_del(&nbr->keepalive_timer) == -1) + fatal(__func__); +} + +/* Keepalive timeout: if the nbr hasn't sent keepalive */ + +static void +nbr_ktimeout(int fd, short event, void *arg) +{ + struct nbr *nbr = arg; + + log_debug("%s: lsr-id %s", __func__, inet_ntoa(nbr->id)); + + session_shutdown(nbr, S_KEEPALIVE_TMR, 0, 0); +} + +static void +nbr_start_ktimeout(struct nbr *nbr) +{ + struct timeval tv; + + timerclear(&tv); + tv.tv_sec = nbr->keepalive; + + if (evtimer_add(&nbr->keepalive_timeout, &tv) == -1) + fatal(__func__); +} + +void +nbr_stop_ktimeout(struct nbr *nbr) +{ + if (evtimer_pending(&nbr->keepalive_timeout, NULL) && + evtimer_del(&nbr->keepalive_timeout) == -1) + fatal(__func__); +} + +/* Session initialization timeout: if nbr got stuck in the initialization FSM */ + +static void +nbr_itimeout(int fd, short event, void *arg) +{ + struct nbr *nbr = arg; + + log_debug("%s: lsr-id %s", __func__, inet_ntoa(nbr->id)); + + nbr_fsm(nbr, NBR_EVT_CLOSE_SESSION); +} + +static void +nbr_start_itimeout(struct nbr *nbr) +{ + struct timeval tv; + + timerclear(&tv); + tv.tv_sec = INIT_FSM_TIMEOUT; + if (evtimer_add(&nbr->init_timeout, &tv) == -1) + fatal(__func__); +} + +void +nbr_stop_itimeout(struct nbr *nbr) +{ + if (evtimer_pending(&nbr->init_timeout, NULL) && + evtimer_del(&nbr->init_timeout) == -1) + fatal(__func__); +} + +/* Init delay timer: timer to retry to iniziatize session */ + +static void +nbr_idtimer(int fd, short event, void *arg) +{ + struct nbr *nbr = arg; + + log_debug("%s: lsr-id %s", __func__, inet_ntoa(nbr->id)); + + nbr_establish_connection(nbr); +} + +void +nbr_start_idtimer(struct nbr *nbr) +{ + struct timeval tv; + + timerclear(&tv); + + tv.tv_sec = INIT_DELAY_TMR; + switch(nbr->idtimer_cnt) { + default: + /* do not further increase the counter */ + tv.tv_sec = MAX_DELAY_TMR; + break; + case 2: + tv.tv_sec *= 2; + /* FALLTHROUGH */ + case 1: + tv.tv_sec *= 2; + /* FALLTHROUGH */ + case 0: + nbr->idtimer_cnt++; + break; + } + + if (evtimer_add(&nbr->initdelay_timer, &tv) == -1) + fatal(__func__); +} + +void +nbr_stop_idtimer(struct nbr *nbr) +{ + if (evtimer_pending(&nbr->initdelay_timer, NULL) && + evtimer_del(&nbr->initdelay_timer) == -1) + fatal(__func__); +} + +int +nbr_pending_idtimer(struct nbr *nbr) +{ + if (evtimer_pending(&nbr->initdelay_timer, NULL)) + return (1); + + return (0); +} + +int +nbr_pending_connect(struct nbr *nbr) +{ + if (event_initialized(&nbr->ev_connect) && + event_pending(&nbr->ev_connect, EV_WRITE, NULL)) + return (1); + + return (0); +} + +static void +nbr_connect_cb(int fd, short event, void *arg) +{ + struct nbr *nbr = arg; + int error; + socklen_t len; + + len = sizeof(error); + if (getsockopt(nbr->fd, SOL_SOCKET, SO_ERROR, &error, &len) < 0) { + log_warn("%s: getsockopt SOL_SOCKET SO_ERROR", __func__); + return; + } + + if (error) { + close(nbr->fd); + errno = error; + log_debug("%s: error while connecting to %s: %s", __func__, + log_addr(nbr->af, &nbr->raddr), strerror(errno)); + return; + } + + nbr_fsm(nbr, NBR_EVT_CONNECT_UP); +} + +int +nbr_establish_connection(struct nbr *nbr) +{ + struct sockaddr_storage local_sa; + struct sockaddr_storage remote_sa; + struct adj *adj; + struct nbr_params *nbrp; + int opt = 1; + + nbr->fd = socket(nbr->af, + SOCK_STREAM | SOCK_NONBLOCK | SOCK_CLOEXEC, 0); + if (nbr->fd == -1) { + log_warn("%s: error while creating socket", __func__); + return (-1); + } + + nbrp = nbr_params_find(leconf, nbr->id); + if (nbrp && nbrp->auth.method == AUTH_MD5SIG) { + if (sysdep.no_pfkey || sysdep.no_md5sig) { + log_warnx("md5sig configured but not available"); + close(nbr->fd); + return (-1); + } + if (setsockopt(nbr->fd, IPPROTO_TCP, TCP_MD5SIG, + &opt, sizeof(opt)) == -1) { + log_warn("setsockopt md5sig"); + close(nbr->fd); + return (-1); + } + } + + memcpy(&local_sa, addr2sa(nbr->af, &nbr->laddr, 0), sizeof(local_sa)); + memcpy(&remote_sa, addr2sa(nbr->af, &nbr->raddr, LDP_PORT), + sizeof(local_sa)); + if (nbr->af == AF_INET6 && nbr->raddr_scope) + addscope((struct sockaddr_in6 *)&remote_sa, nbr->raddr_scope); + + if (bind(nbr->fd, (struct sockaddr *)&local_sa, + local_sa.ss_len) == -1) { + log_warn("%s: error while binding socket to %s", __func__, + log_sockaddr((struct sockaddr *)&local_sa)); + close(nbr->fd); + return (-1); + } + + if (nbr_gtsm_check(nbr->fd, nbr, nbrp)) { + close(nbr->fd); + return (-1); + } + + /* + * Send an extra hello to guarantee that the remote peer has formed + * an adjacency as well. + */ + LIST_FOREACH(adj, &nbr->adj_list, nbr_entry) + send_hello(adj->source.type, adj->source.link.ia, + adj->source.target); + + if (connect(nbr->fd, (struct sockaddr *)&remote_sa, + remote_sa.ss_len) == -1) { + if (errno == EINPROGRESS) { + event_set(&nbr->ev_connect, nbr->fd, EV_WRITE, + nbr_connect_cb, nbr); + event_add(&nbr->ev_connect, NULL); + return (0); + } + log_warn("%s: error while connecting to %s", __func__, + log_sockaddr((struct sockaddr *)&remote_sa)); + close(nbr->fd); + return (-1); + } + + /* connection completed immediately */ + nbr_fsm(nbr, NBR_EVT_CONNECT_UP); + + return (0); +} + +int +nbr_gtsm_enabled(struct nbr *nbr, struct nbr_params *nbrp) +{ + /* + * RFC 6720 - Section 3: + * "This document allows for the implementation to provide an option to + * statically (e.g., via configuration) and/or dynamically override the + * default behavior and enable/disable GTSM on a per-peer basis". + */ + if (nbrp && (nbrp->flags & F_NBRP_GTSM)) + return (nbrp->gtsm_enabled); + + if ((ldp_af_conf_get(leconf, nbr->af))->flags & F_LDPD_AF_NO_GTSM) + return (0); + + /* By default, GTSM support has to be negotiated for LDPv4 */ + if (nbr->af == AF_INET && !(nbr->flags & F_NBR_GTSM_NEGOTIATED)) + return (0); + + return (1); +} + +int +nbr_gtsm_setup(int fd, int af, struct nbr_params *nbrp) +{ + int ttl = 255; + + if (nbrp && (nbrp->flags & F_NBRP_GTSM_HOPS)) + ttl = 256 - nbrp->gtsm_hops; + + switch (af) { + case AF_INET: + if (sock_set_ipv4_minttl(fd, ttl) == -1) + return (-1); + ttl = 255; + if (sock_set_ipv4_ucast_ttl(fd, ttl) == -1) + return (-1); + break; + case AF_INET6: + if (sock_set_ipv6_minhopcount(fd, ttl) == -1) + return (-1); + ttl = 255; + if (sock_set_ipv6_ucast_hops(fd, ttl) == -1) + return (-1); + break; + default: + fatalx("nbr_gtsm_setup: unknown af"); + } + + return (0); +} + +int +nbr_gtsm_check(int fd, struct nbr *nbr, struct nbr_params *nbrp) +{ + if (!nbr_gtsm_enabled(nbr, nbrp)) { + switch (nbr->af) { + case AF_INET: + sock_set_ipv4_ucast_ttl(fd, -1); + break; + case AF_INET6: + /* + * Send packets with a Hop Limit of 255 even when GSTM + * is disabled to guarantee interoperability. + */ + sock_set_ipv6_ucast_hops(fd, 255); + break; + default: + fatalx("nbr_gtsm_check: unknown af"); + break; + } + return (0); + } + + if (nbr_gtsm_setup(fd, nbr->af, nbrp) == -1) { + log_warnx("%s: error enabling GTSM for lsr-id %s", __func__, + inet_ntoa(nbr->id)); + return (-1); + } + + return (0); +} + +static int +nbr_act_session_operational(struct nbr *nbr) +{ + struct lde_nbr lde_nbr; + + nbr->idtimer_cnt = 0; + + /* this is necessary to avoid ipc synchronization issues */ + nbr_update_peerid(nbr); + + memset(&lde_nbr, 0, sizeof(lde_nbr)); + lde_nbr.id = nbr->id; + lde_nbr.v4_enabled = nbr->v4_enabled; + lde_nbr.v6_enabled = nbr->v6_enabled; + return (ldpe_imsg_compose_lde(IMSG_NEIGHBOR_UP, nbr->peerid, 0, + &lde_nbr, sizeof(lde_nbr))); +} + +static void +nbr_send_labelmappings(struct nbr *nbr) +{ + ldpe_imsg_compose_lde(IMSG_LABEL_MAPPING_FULL, nbr->peerid, 0, + NULL, 0); +} + +struct nbr_params * +nbr_params_new(struct in_addr lsr_id) +{ + struct nbr_params *nbrp; + + if ((nbrp = calloc(1, sizeof(*nbrp))) == NULL) + fatal(__func__); + + nbrp->lsr_id = lsr_id; + nbrp->auth.method = AUTH_NONE; + + return (nbrp); +} + +struct nbr_params * +nbr_params_find(struct ldpd_conf *xconf, struct in_addr lsr_id) +{ + struct nbr_params *nbrp; + + LIST_FOREACH(nbrp, &xconf->nbrp_list, entry) + if (nbrp->lsr_id.s_addr == lsr_id.s_addr) + return (nbrp); + + return (NULL); +} + +uint16_t +nbr_get_keepalive(int af, struct in_addr lsr_id) +{ + struct nbr_params *nbrp; + + nbrp = nbr_params_find(leconf, lsr_id); + if (nbrp && (nbrp->flags & F_NBRP_KEEPALIVE)) + return (nbrp->keepalive); + + return ((ldp_af_conf_get(leconf, af))->keepalive); +} + +struct ctl_nbr * +nbr_to_ctl(struct nbr *nbr) +{ + static struct ctl_nbr nctl; + struct timeval now; + + nctl.af = nbr->af; + nctl.id = nbr->id; + nctl.laddr = nbr->laddr; + nctl.raddr = nbr->raddr; + nctl.nbr_state = nbr->state; + + gettimeofday(&now, NULL); + if (nbr->state == NBR_STA_OPER) { + nctl.uptime = now.tv_sec - nbr->uptime; + } else + nctl.uptime = 0; + + return (&nctl); +} + +void +nbr_clear_ctl(struct ctl_nbr *nctl) +{ + struct nbr *nbr; + + RB_FOREACH(nbr, nbr_addr_head, &nbrs_by_addr) { + if (ldp_addrisset(nctl->af, &nctl->raddr) && + ldp_addrcmp(nctl->af, &nctl->raddr, &nbr->raddr)) + continue; + + log_debug("%s: neighbor %s manually cleared", __func__, + log_addr(nbr->af, &nbr->raddr)); + session_shutdown(nbr, S_SHUTDOWN, 0, 0); + } +} diff --git a/ldpd/notification.c b/ldpd/notification.c new file mode 100644 index 000000000..f30646bb8 --- /dev/null +++ b/ldpd/notification.c @@ -0,0 +1,239 @@ +/* $OpenBSD$ */ + +/* + * Copyright (c) 2009 Michele Marchetto + * + * 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 "ldpd.h" +#include "ldp.h" +#include "log.h" +#include "ldpe.h" + +void +send_notification_full(struct tcp_conn *tcp, struct notify_msg *nm) +{ + struct ibuf *buf; + uint16_t size; + int err = 0; + + /* calculate size */ + size = LDP_HDR_SIZE + LDP_MSG_SIZE + STATUS_SIZE; + if (nm->flags & F_NOTIF_PW_STATUS) + size += PW_STATUS_TLV_SIZE; + if (nm->flags & F_NOTIF_FEC) { + size += TLV_HDR_SIZE; + switch (nm->fec.type) { + case MAP_TYPE_PWID: + size += FEC_PWID_ELM_MIN_LEN; + if (nm->fec.flags & F_MAP_PW_ID) + size += sizeof(uint32_t); + break; + } + } + + if ((buf = ibuf_open(size)) == NULL) + fatal(__func__); + + err |= gen_ldp_hdr(buf, size); + size -= LDP_HDR_SIZE; + err |= gen_msg_hdr(buf, MSG_TYPE_NOTIFICATION, size); + err |= gen_status_tlv(buf, nm->status_code, nm->msg_id, nm->msg_type); + /* optional tlvs */ + if (nm->flags & F_NOTIF_PW_STATUS) + err |= gen_pw_status_tlv(buf, nm->pw_status); + if (nm->flags & F_NOTIF_FEC) + err |= gen_fec_tlv(buf, &nm->fec); + if (err) { + ibuf_free(buf); + return; + } + + if (tcp->nbr) + log_debug("msg-out: notification: lsr-id %s, status %s%s", + inet_ntoa(tcp->nbr->id), status_code_name(nm->status_code), + (nm->status_code & STATUS_FATAL) ? " (fatal)" : ""); + + evbuf_enqueue(&tcp->wbuf, buf); +} + +/* send a notification without optional tlvs */ +void +send_notification(uint32_t status_code, struct tcp_conn *tcp, uint32_t msg_id, + uint16_t msg_type) +{ + struct notify_msg nm; + + memset(&nm, 0, sizeof(nm)); + nm.status_code = status_code; + nm.msg_id = msg_id; + nm.msg_type = msg_type; + + send_notification_full(tcp, &nm); +} + +void +send_notification_nbr(struct nbr *nbr, uint32_t status_code, uint32_t msg_id, + uint16_t msg_type) +{ + send_notification(status_code, nbr->tcp, msg_id, msg_type); + nbr_fsm(nbr, NBR_EVT_PDU_SENT); +} + +int +recv_notification(struct nbr *nbr, char *buf, uint16_t len) +{ + struct ldp_msg msg; + struct status_tlv st; + struct notify_msg nm; + int tlen; + + memcpy(&msg, buf, sizeof(msg)); + buf += LDP_MSG_SIZE; + len -= LDP_MSG_SIZE; + + if (len < STATUS_SIZE) { + session_shutdown(nbr, S_BAD_MSG_LEN, msg.id, msg.type); + return (-1); + } + memcpy(&st, buf, sizeof(st)); + + if (ntohs(st.length) > STATUS_SIZE - TLV_HDR_SIZE || + ntohs(st.length) > len - TLV_HDR_SIZE) { + session_shutdown(nbr, S_BAD_TLV_LEN, msg.id, msg.type); + return (-1); + } + buf += STATUS_SIZE; + len -= STATUS_SIZE; + + memset(&nm, 0, sizeof(nm)); + nm.status_code = ntohl(st.status_code); + + /* Optional Parameters */ + while (len > 0) { + struct tlv tlv; + uint16_t tlv_len; + + if (len < sizeof(tlv)) { + session_shutdown(nbr, S_BAD_TLV_LEN, msg.id, msg.type); + return (-1); + } + + memcpy(&tlv, buf, TLV_HDR_SIZE); + tlv_len = ntohs(tlv.length); + if (tlv_len + TLV_HDR_SIZE > len) { + session_shutdown(nbr, S_BAD_TLV_LEN, msg.id, msg.type); + return (-1); + } + buf += TLV_HDR_SIZE; + len -= TLV_HDR_SIZE; + + switch (ntohs(tlv.type)) { + case TLV_TYPE_EXTSTATUS: + case TLV_TYPE_RETURNEDPDU: + case TLV_TYPE_RETURNEDMSG: + /* TODO is there any use for this? */ + break; + case TLV_TYPE_PW_STATUS: + if (tlv_len != 4) { + session_shutdown(nbr, S_BAD_TLV_LEN, + msg.id, msg.type); + return (-1); + } + + nm.pw_status = ntohl(*(uint32_t *)buf); + nm.flags |= F_NOTIF_PW_STATUS; + break; + case TLV_TYPE_FEC: + if ((tlen = tlv_decode_fec_elm(nbr, &msg, buf, + tlv_len, &nm.fec)) == -1) + return (-1); + /* allow only one fec element */ + if (tlen != tlv_len) { + session_shutdown(nbr, S_BAD_TLV_VAL, + msg.id, msg.type); + return (-1); + } + nm.flags |= F_NOTIF_FEC; + break; + default: + if (!(ntohs(tlv.type) & UNKNOWN_FLAG)) + send_notification_nbr(nbr, S_UNKNOWN_TLV, + msg.id, msg.type); + /* ignore unknown tlv */ + break; + } + buf += tlv_len; + len -= tlv_len; + } + + if (nm.status_code == S_PW_STATUS) { + if (!(nm.flags & (F_NOTIF_PW_STATUS|F_NOTIF_FEC))) { + send_notification_nbr(nbr, S_MISS_MSG, + msg.id, msg.type); + return (-1); + } + + switch (nm.fec.type) { + case MAP_TYPE_PWID: + break; + default: + send_notification_nbr(nbr, S_BAD_TLV_VAL, + msg.id, msg.type); + return (-1); + } + } + + log_warnx("msg-in: notification: lsr-id %s, status %s%s", + inet_ntoa(nbr->id), status_code_name(ntohl(st.status_code)), + (st.status_code & htonl(STATUS_FATAL)) ? " (fatal)" : ""); + + if (st.status_code & htonl(STATUS_FATAL)) { + if (nbr->state == NBR_STA_OPENSENT) + nbr_start_idtimer(nbr); + + nbr_fsm(nbr, NBR_EVT_CLOSE_SESSION); + return (-1); + } + + if (nm.status_code == S_PW_STATUS) + ldpe_imsg_compose_lde(IMSG_NOTIFICATION, nbr->peerid, 0, + &nm, sizeof(nm)); + + return (0); +} + +int +gen_status_tlv(struct ibuf *buf, uint32_t status_code, uint32_t msg_id, + uint16_t msg_type) +{ + struct status_tlv st; + + memset(&st, 0, sizeof(st)); + st.type = htons(TLV_TYPE_STATUS); + st.length = htons(STATUS_TLV_LEN); + st.status_code = htonl(status_code); + /* + * For convenience, msg_id and msg_type are already in network + * byte order. + */ + st.msg_id = msg_id; + st.msg_type = msg_type; + + return (ibuf_add(buf, &st, STATUS_SIZE)); +} diff --git a/ldpd/packet.c b/ldpd/packet.c new file mode 100644 index 000000000..7cc375c31 --- /dev/null +++ b/ldpd/packet.c @@ -0,0 +1,788 @@ +/* $OpenBSD$ */ + +/* + * Copyright (c) 2013, 2016 Renato Westphal + * Copyright (c) 2009 Michele Marchetto + * Copyright (c) 2004, 2005, 2008 Esben Norby + * + * 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 "ldpd.h" +#include "ldpe.h" +#include "log.h" + +static struct iface *disc_find_iface(unsigned int, int, + union ldpd_addr *, int); +static void session_read(int, short, void *); +static void session_write(int, short, void *); +static ssize_t session_get_pdu(struct ibuf_read *, char **); +static void tcp_close(struct tcp_conn *); +static struct pending_conn *pending_conn_new(int, int, union ldpd_addr *); +static void pending_conn_timeout(int, short, void *); + +int +gen_ldp_hdr(struct ibuf *buf, uint16_t size) +{ + struct ldp_hdr ldp_hdr; + + memset(&ldp_hdr, 0, sizeof(ldp_hdr)); + ldp_hdr.version = htons(LDP_VERSION); + /* exclude the 'Version' and 'PDU Length' fields from the total */ + ldp_hdr.length = htons(size - LDP_HDR_DEAD_LEN); + ldp_hdr.lsr_id = leconf->rtr_id.s_addr; + ldp_hdr.lspace_id = 0; + + return (ibuf_add(buf, &ldp_hdr, LDP_HDR_SIZE)); +} + +int +gen_msg_hdr(struct ibuf *buf, uint16_t type, uint16_t size) +{ + static int msgcnt = 0; + struct ldp_msg msg; + + memset(&msg, 0, sizeof(msg)); + msg.type = htons(type); + /* exclude the 'Type' and 'Length' fields from the total */ + msg.length = htons(size - LDP_MSG_DEAD_LEN); + msg.id = htonl(++msgcnt); + + return (ibuf_add(buf, &msg, sizeof(msg))); +} + +/* send packets */ +int +send_packet(int fd, int af, union ldpd_addr *dst, struct iface_af *ia, + void *pkt, size_t len) +{ + struct sockaddr *sa; + + switch (af) { + case AF_INET: + if (ia && IN_MULTICAST(ntohl(dst->v4.s_addr))) { + /* set outgoing interface for multicast traffic */ + if (sock_set_ipv4_mcast(ia->iface) == -1) { + log_debug("%s: error setting multicast " + "interface, %s", __func__, ia->iface->name); + return (-1); + } + } + break; + case AF_INET6: + if (ia && IN6_IS_ADDR_MULTICAST(&dst->v6)) { + /* set outgoing interface for multicast traffic */ + if (sock_set_ipv6_mcast(ia->iface) == -1) { + log_debug("%s: error setting multicast " + "interface, %s", __func__, ia->iface->name); + return (-1); + } + } + break; + default: + fatalx("send_packet: unknown af"); + } + + sa = addr2sa(af, dst, LDP_PORT); + if (sendto(fd, pkt, len, 0, sa, sa->sa_len) == -1) { + log_warn("%s: error sending packet to %s", __func__, + log_sockaddr(sa)); + return (-1); + } + + return (0); +} + +/* Discovery functions */ +#define CMSG_MAXLEN max(sizeof(struct sockaddr_dl), sizeof(struct in6_pktinfo)) +void +disc_recv_packet(int fd, short event, void *bula) +{ + union { + struct cmsghdr hdr; + char buf[CMSG_SPACE(CMSG_MAXLEN)]; + } cmsgbuf; + struct msghdr m; + struct sockaddr_storage from; + struct iovec iov; + char *buf; + struct cmsghdr *cmsg; + ssize_t r; + int multicast; + int af; + union ldpd_addr src; + unsigned int ifindex = 0; + struct iface *iface; + uint16_t len; + struct ldp_hdr ldp_hdr; + uint16_t pdu_len; + struct ldp_msg msg; + uint16_t msg_len; + struct in_addr lsr_id; + + if (event != EV_READ) + return; + + /* setup buffer */ + memset(&m, 0, sizeof(m)); + iov.iov_base = buf = pkt_ptr; + iov.iov_len = IBUF_READ_SIZE; + m.msg_name = &from; + m.msg_namelen = sizeof(from); + m.msg_iov = &iov; + m.msg_iovlen = 1; + m.msg_control = &cmsgbuf.buf; + m.msg_controllen = sizeof(cmsgbuf.buf); + + if ((r = recvmsg(fd, &m, 0)) == -1) { + if (errno != EAGAIN && errno != EINTR) + log_debug("%s: read error: %s", __func__, + strerror(errno)); + return; + } + + multicast = (m.msg_flags & MSG_MCAST) ? 1 : 0; + sa2addr((struct sockaddr *)&from, &af, &src); + if (bad_addr(af, &src)) { + log_debug("%s: invalid source address: %s", __func__, + log_addr(af, &src)); + return; + } + + for (cmsg = CMSG_FIRSTHDR(&m); cmsg != NULL; + cmsg = CMSG_NXTHDR(&m, cmsg)) { + if (af == AF_INET && cmsg->cmsg_level == IPPROTO_IP && + cmsg->cmsg_type == IP_RECVIF) { + ifindex = ((struct sockaddr_dl *) + CMSG_DATA(cmsg))->sdl_index; + break; + } + if (af == AF_INET6 && cmsg->cmsg_level == IPPROTO_IPV6 && + cmsg->cmsg_type == IPV6_PKTINFO) { + ifindex = ((struct in6_pktinfo *) + CMSG_DATA(cmsg))->ipi6_ifindex; + break; + } + } + + /* find a matching interface */ + iface = disc_find_iface(ifindex, af, &src, multicast); + if (iface == NULL) + return; + + /* check packet size */ + len = (uint16_t)r; + if (len < (LDP_HDR_SIZE + LDP_MSG_SIZE) || len > LDP_MAX_LEN) { + log_debug("%s: bad packet size, source %s", __func__, + log_addr(af, &src)); + return; + } + + /* LDP header sanity checks */ + memcpy(&ldp_hdr, buf, sizeof(ldp_hdr)); + if (ntohs(ldp_hdr.version) != LDP_VERSION) { + log_debug("%s: invalid LDP version %d, source %s", __func__, + ntohs(ldp_hdr.version), log_addr(af, &src)); + return; + } + if (ntohs(ldp_hdr.lspace_id) != 0) { + log_debug("%s: invalid label space %u, source %s", __func__, + ntohs(ldp_hdr.lspace_id), log_addr(af, &src)); + return; + } + /* check "PDU Length" field */ + pdu_len = ntohs(ldp_hdr.length); + if ((pdu_len < (LDP_HDR_PDU_LEN + LDP_MSG_SIZE)) || + (pdu_len > (len - LDP_HDR_DEAD_LEN))) { + log_debug("%s: invalid LDP packet length %u, source %s", + __func__, ntohs(ldp_hdr.length), log_addr(af, &src)); + return; + } + buf += LDP_HDR_SIZE; + len -= LDP_HDR_SIZE; + + lsr_id.s_addr = ldp_hdr.lsr_id; + + /* + * For UDP, we process only the first message of each packet. This does + * not impose any restrictions since LDP uses UDP only for sending Hello + * packets. + */ + memcpy(&msg, buf, sizeof(msg)); + + /* check "Message Length" field */ + msg_len = ntohs(msg.length); + if (msg_len < LDP_MSG_LEN || ((msg_len + LDP_MSG_DEAD_LEN) > pdu_len)) { + log_debug("%s: invalid LDP message length %u, source %s", + __func__, ntohs(msg.length), log_addr(af, &src)); + return; + } + buf += LDP_MSG_SIZE; + len -= LDP_MSG_SIZE; + + /* switch LDP packet type */ + switch (ntohs(msg.type)) { + case MSG_TYPE_HELLO: + recv_hello(lsr_id, &msg, af, &src, iface, multicast, buf, len); + break; + default: + log_debug("%s: unknown LDP packet type, source %s", __func__, + log_addr(af, &src)); + } +} + +static struct iface * +disc_find_iface(unsigned int ifindex, int af, union ldpd_addr *src, + int multicast) +{ + struct iface *iface; + struct iface_af *ia; + struct if_addr *if_addr; + in_addr_t mask; + + iface = if_lookup(leconf, ifindex); + if (iface == NULL) + return (NULL); + + /* + * For unicast packets, we just need to make sure that the interface + * is enabled for the given address-family. + */ + if (!multicast) { + ia = iface_af_get(iface, af); + if (ia->enabled) + return (iface); + return (NULL); + } + + switch (af) { + case AF_INET: + LIST_FOREACH(if_addr, &iface->addr_list, entry) { + if (if_addr->af != AF_INET) + continue; + + switch (iface->type) { + case IF_TYPE_POINTOPOINT: + if (if_addr->dstbrd.v4.s_addr == src->v4.s_addr) + return (iface); + break; + default: + mask = prefixlen2mask(if_addr->prefixlen); + if ((if_addr->addr.v4.s_addr & mask) == + (src->v4.s_addr & mask)) + return (iface); + break; + } + } + break; + case AF_INET6: + if (IN6_IS_ADDR_LINKLOCAL(&src->v6)) + return (iface); + break; + default: + fatalx("disc_find_iface: unknown af"); + } + + return (NULL); +} + +void +session_accept(int fd, short event, void *bula) +{ + struct sockaddr_storage src; + socklen_t len = sizeof(src); + int newfd; + int af; + union ldpd_addr addr; + struct nbr *nbr; + struct pending_conn *pconn; + + if (!(event & EV_READ)) + return; + + newfd = accept4(fd, (struct sockaddr *)&src, &len, + SOCK_NONBLOCK | SOCK_CLOEXEC); + if (newfd == -1) { + /* + * Pause accept if we are out of file descriptors, or + * libevent will haunt us here too. + */ + if (errno == ENFILE || errno == EMFILE) { + accept_pause(); + } else if (errno != EWOULDBLOCK && errno != EINTR && + errno != ECONNABORTED) + log_debug("%s: accept error: %s", __func__, + strerror(errno)); + return; + } + + sa2addr((struct sockaddr *)&src, &af, &addr); + + /* + * Since we don't support label spaces, we can identify this neighbor + * just by its source address. This way we don't need to wait for its + * Initialization message to know who we are talking to. + */ + nbr = nbr_find_addr(af, &addr); + if (nbr == NULL) { + /* + * According to RFC 5036, we would need to send a No Hello + * Error Notification message and close this TCP connection + * right now. But doing so would trigger the backoff exponential + * timer in the remote peer, which would considerably slow down + * the session establishment process. The trick here is to wait + * five seconds before sending the Notification Message. There's + * a good chance that the remote peer will send us a Hello + * message within this interval, so it's worth waiting before + * taking a more drastic measure. + */ + pconn = pending_conn_find(af, &addr); + if (pconn) + close(newfd); + else + pending_conn_new(newfd, af, &addr); + return; + } + /* protection against buggy implementations */ + if (nbr_session_active_role(nbr)) { + close(newfd); + return; + } + if (nbr->state != NBR_STA_PRESENT) { + log_debug("%s: lsr-id %s: rejecting additional transport " + "connection", __func__, inet_ntoa(nbr->id)); + close(newfd); + return; + } + + session_accept_nbr(nbr, newfd); +} + +void +session_accept_nbr(struct nbr *nbr, int fd) +{ + struct nbr_params *nbrp; + int opt; + socklen_t len; + + nbrp = nbr_params_find(leconf, nbr->id); + if (nbr_gtsm_check(fd, nbr, nbrp)) { + close(fd); + return; + } + + if (nbrp && nbrp->auth.method == AUTH_MD5SIG) { + if (sysdep.no_pfkey || sysdep.no_md5sig) { + log_warnx("md5sig configured but not available"); + close(fd); + return; + } + + len = sizeof(opt); + if (getsockopt(fd, IPPROTO_TCP, TCP_MD5SIG, &opt, &len) == -1) + fatal("getsockopt TCP_MD5SIG"); + if (!opt) { /* non-md5'd connection! */ + log_warnx("connection attempt without md5 signature"); + close(fd); + return; + } + } + + nbr->tcp = tcp_new(fd, nbr); + nbr_fsm(nbr, NBR_EVT_MATCH_ADJ); +} + +static void +session_read(int fd, short event, void *arg) +{ + struct nbr *nbr = arg; + struct tcp_conn *tcp = nbr->tcp; + struct ldp_hdr *ldp_hdr; + struct ldp_msg *msg; + char *buf, *pdu; + ssize_t n, len; + uint16_t pdu_len, msg_len, msg_size, max_pdu_len; + int ret; + + if (event != EV_READ) + return; + + if ((n = read(fd, tcp->rbuf->buf + tcp->rbuf->wpos, + sizeof(tcp->rbuf->buf) - tcp->rbuf->wpos)) == -1) { + if (errno != EINTR && errno != EAGAIN) { + log_warn("%s: read error", __func__); + nbr_fsm(nbr, NBR_EVT_CLOSE_SESSION); + return; + } + /* retry read */ + return; + } + if (n == 0) { + /* connection closed */ + log_debug("%s: connection closed by remote end", __func__); + nbr_fsm(nbr, NBR_EVT_CLOSE_SESSION); + return; + } + tcp->rbuf->wpos += n; + + while ((len = session_get_pdu(tcp->rbuf, &buf)) > 0) { + pdu = buf; + ldp_hdr = (struct ldp_hdr *)pdu; + if (ntohs(ldp_hdr->version) != LDP_VERSION) { + session_shutdown(nbr, S_BAD_PROTO_VER, 0, 0); + free(buf); + return; + } + + pdu_len = ntohs(ldp_hdr->length); + /* + * RFC 5036 - Section 3.5.3: + * "Prior to completion of the negotiation, the maximum + * allowable length is 4096 bytes". + */ + if (nbr->state == NBR_STA_OPER) + max_pdu_len = nbr->max_pdu_len; + else + max_pdu_len = LDP_MAX_LEN; + if (pdu_len < (LDP_HDR_PDU_LEN + LDP_MSG_SIZE) || + pdu_len > max_pdu_len) { + session_shutdown(nbr, S_BAD_PDU_LEN, 0, 0); + free(buf); + return; + } + pdu_len -= LDP_HDR_PDU_LEN; + if (ldp_hdr->lsr_id != nbr->id.s_addr || + ldp_hdr->lspace_id != 0) { + session_shutdown(nbr, S_BAD_LDP_ID, 0, 0); + free(buf); + return; + } + pdu += LDP_HDR_SIZE; + len -= LDP_HDR_SIZE; + + nbr_fsm(nbr, NBR_EVT_PDU_RCVD); + + while (len >= LDP_MSG_SIZE) { + uint16_t type; + + msg = (struct ldp_msg *)pdu; + type = ntohs(msg->type); + msg_len = ntohs(msg->length); + if (msg_len < LDP_MSG_LEN || + (msg_len + LDP_MSG_DEAD_LEN) > pdu_len) { + session_shutdown(nbr, S_BAD_TLV_LEN, msg->id, + msg->type); + free(buf); + return; + } + msg_size = msg_len + LDP_MSG_DEAD_LEN; + pdu_len -= msg_size; + + /* check for error conditions earlier */ + switch (type) { + case MSG_TYPE_INIT: + if ((nbr->state != NBR_STA_INITIAL) && + (nbr->state != NBR_STA_OPENSENT)) { + session_shutdown(nbr, S_SHUTDOWN, + msg->id, msg->type); + free(buf); + return; + } + break; + case MSG_TYPE_KEEPALIVE: + if ((nbr->state == NBR_STA_INITIAL) || + (nbr->state == NBR_STA_OPENSENT)) { + session_shutdown(nbr, S_SHUTDOWN, + msg->id, msg->type); + free(buf); + return; + } + break; + case MSG_TYPE_ADDR: + case MSG_TYPE_ADDRWITHDRAW: + case MSG_TYPE_LABELMAPPING: + case MSG_TYPE_LABELREQUEST: + case MSG_TYPE_LABELWITHDRAW: + case MSG_TYPE_LABELRELEASE: + case MSG_TYPE_LABELABORTREQ: + if (nbr->state != NBR_STA_OPER) { + session_shutdown(nbr, S_SHUTDOWN, + msg->id, msg->type); + free(buf); + return; + } + break; + default: + break; + } + + /* switch LDP packet type */ + switch (type) { + case MSG_TYPE_NOTIFICATION: + ret = recv_notification(nbr, pdu, msg_size); + break; + case MSG_TYPE_INIT: + ret = recv_init(nbr, pdu, msg_size); + break; + case MSG_TYPE_KEEPALIVE: + ret = recv_keepalive(nbr, pdu, msg_size); + break; + case MSG_TYPE_ADDR: + case MSG_TYPE_ADDRWITHDRAW: + ret = recv_address(nbr, pdu, msg_size); + break; + case MSG_TYPE_LABELMAPPING: + case MSG_TYPE_LABELREQUEST: + case MSG_TYPE_LABELWITHDRAW: + case MSG_TYPE_LABELRELEASE: + case MSG_TYPE_LABELABORTREQ: + ret = recv_labelmessage(nbr, pdu, msg_size, + type); + break; + default: + log_debug("%s: unknown LDP message from nbr %s", + __func__, inet_ntoa(nbr->id)); + if (!(ntohs(msg->type) & UNKNOWN_FLAG)) + send_notification_nbr(nbr, + S_UNKNOWN_MSG, msg->id, msg->type); + /* ignore the message */ + ret = 0; + break; + } + + if (ret == -1) { + /* parser failed, giving up */ + free(buf); + return; + } + + /* Analyse the next message */ + pdu += msg_size; + len -= msg_size; + } + free(buf); + if (len != 0) { + session_shutdown(nbr, S_BAD_PDU_LEN, 0, 0); + return; + } + } +} + +static void +session_write(int fd, short event, void *arg) +{ + struct tcp_conn *tcp = arg; + struct nbr *nbr = tcp->nbr; + + if (!(event & EV_WRITE)) + return; + + if (msgbuf_write(&tcp->wbuf.wbuf) <= 0) + if (errno != EAGAIN && nbr) + nbr_fsm(nbr, NBR_EVT_CLOSE_SESSION); + + if (nbr == NULL && !tcp->wbuf.wbuf.queued) { + /* + * We are done sending the notification message, now we can + * close the socket. + */ + tcp_close(tcp); + return; + } + + evbuf_event_add(&tcp->wbuf); +} + +void +session_shutdown(struct nbr *nbr, uint32_t status, uint32_t msg_id, + uint32_t msg_type) +{ + switch (nbr->state) { + case NBR_STA_PRESENT: + if (nbr_pending_connect(nbr)) + event_del(&nbr->ev_connect); + break; + case NBR_STA_INITIAL: + case NBR_STA_OPENREC: + case NBR_STA_OPENSENT: + case NBR_STA_OPER: + log_debug("%s: lsr-id %s", __func__, inet_ntoa(nbr->id)); + + send_notification_nbr(nbr, status, msg_id, msg_type); + + nbr_fsm(nbr, NBR_EVT_CLOSE_SESSION); + break; + default: + fatalx("session_shutdown: unknown neighbor state"); + } +} + +void +session_close(struct nbr *nbr) +{ + log_debug("%s: closing session with lsr-id %s", __func__, + inet_ntoa(nbr->id)); + + tcp_close(nbr->tcp); + nbr_stop_ktimer(nbr); + nbr_stop_ktimeout(nbr); + nbr_stop_itimeout(nbr); +} + +static ssize_t +session_get_pdu(struct ibuf_read *r, char **b) +{ + struct ldp_hdr l; + size_t av, dlen, left; + + av = r->wpos; + if (av < sizeof(l)) + return (0); + + memcpy(&l, r->buf, sizeof(l)); + dlen = ntohs(l.length) + LDP_HDR_DEAD_LEN; + if (dlen > av) + return (0); + + if ((*b = malloc(dlen)) == NULL) + return (-1); + + memcpy(*b, r->buf, dlen); + if (dlen < av) { + left = av - dlen; + memmove(r->buf, r->buf + dlen, left); + r->wpos = left; + } else + r->wpos = 0; + + return (dlen); +} + +struct tcp_conn * +tcp_new(int fd, struct nbr *nbr) +{ + struct tcp_conn *tcp; + + if ((tcp = calloc(1, sizeof(*tcp))) == NULL) + fatal(__func__); + + tcp->fd = fd; + evbuf_init(&tcp->wbuf, tcp->fd, session_write, tcp); + + if (nbr) { + if ((tcp->rbuf = calloc(1, sizeof(struct ibuf_read))) == NULL) + fatal(__func__); + + event_set(&tcp->rev, tcp->fd, EV_READ | EV_PERSIST, + session_read, nbr); + event_add(&tcp->rev, NULL); + tcp->nbr = nbr; + } + + return (tcp); +} + +static void +tcp_close(struct tcp_conn *tcp) +{ + /* try to flush write buffer */ + msgbuf_write(&tcp->wbuf.wbuf); + evbuf_clear(&tcp->wbuf); + + if (tcp->nbr) { + event_del(&tcp->rev); + free(tcp->rbuf); + tcp->nbr->tcp = NULL; + } + + close(tcp->fd); + accept_unpause(); + free(tcp); +} + +static struct pending_conn * +pending_conn_new(int fd, int af, union ldpd_addr *addr) +{ + struct pending_conn *pconn; + struct timeval tv; + + if ((pconn = calloc(1, sizeof(*pconn))) == NULL) + fatal(__func__); + + pconn->fd = fd; + pconn->af = af; + pconn->addr = *addr; + evtimer_set(&pconn->ev_timeout, pending_conn_timeout, pconn); + TAILQ_INSERT_TAIL(&global.pending_conns, pconn, entry); + + timerclear(&tv); + tv.tv_sec = PENDING_CONN_TIMEOUT; + if (evtimer_add(&pconn->ev_timeout, &tv) == -1) + fatal(__func__); + + return (pconn); +} + +void +pending_conn_del(struct pending_conn *pconn) +{ + if (evtimer_pending(&pconn->ev_timeout, NULL) && + evtimer_del(&pconn->ev_timeout) == -1) + fatal(__func__); + + TAILQ_REMOVE(&global.pending_conns, pconn, entry); + free(pconn); +} + +struct pending_conn * +pending_conn_find(int af, union ldpd_addr *addr) +{ + struct pending_conn *pconn; + + TAILQ_FOREACH(pconn, &global.pending_conns, entry) + if (af == pconn->af && + ldp_addrcmp(af, addr, &pconn->addr) == 0) + return (pconn); + + return (NULL); +} + +static void +pending_conn_timeout(int fd, short event, void *arg) +{ + struct pending_conn *pconn = arg; + struct tcp_conn *tcp; + + log_debug("%s: no adjacency with remote end: %s", __func__, + log_addr(pconn->af, &pconn->addr)); + + /* + * Create a write buffer detached from any neighbor to send a + * notification message reliably. + */ + tcp = tcp_new(pconn->fd, NULL); + send_notification(S_NO_HELLO, tcp, 0, 0); + msgbuf_write(&tcp->wbuf.wbuf); + + pending_conn_del(pconn); +} diff --git a/ldpd/pfkey.c b/ldpd/pfkey.c new file mode 100644 index 000000000..f0f16c867 --- /dev/null +++ b/ldpd/pfkey.c @@ -0,0 +1,466 @@ +/* $OpenBSD$ */ + +/* + * Copyright (c) 2003, 2004 Henning Brauer + * Copyright (c) 2003, 2004 Markus Friedl + * + * 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 "ldpd.h" +#include "ldpe.h" +#include "log.h" + +static int pfkey_send(int, uint8_t, uint8_t, uint8_t, + int, union ldpd_addr *, union ldpd_addr *, + uint32_t, uint8_t, int, char *, uint8_t, int, char *, + uint16_t, uint16_t); +static int pfkey_reply(int, uint32_t *); +static int pfkey_sa_add(int, union ldpd_addr *, union ldpd_addr *, + uint8_t, char *, uint32_t *); +static int pfkey_sa_remove(int, union ldpd_addr *, union ldpd_addr *, + uint32_t *); +static int pfkey_md5sig_establish(struct nbr *, struct nbr_params *nbrp); +static int pfkey_md5sig_remove(struct nbr *); + +#define PFKEY2_CHUNK sizeof(uint64_t) +#define ROUNDUP(x) (((x) + (PFKEY2_CHUNK - 1)) & ~(PFKEY2_CHUNK - 1)) +#define IOV_CNT 20 + +static uint32_t sadb_msg_seq; +static uint32_t pid; /* should pid_t but pfkey needs uint32_t */ +static int fd; + +static int +pfkey_send(int sd, uint8_t satype, uint8_t mtype, uint8_t dir, + int af, union ldpd_addr *src, union ldpd_addr *dst, uint32_t spi, + uint8_t aalg, int alen, char *akey, uint8_t ealg, int elen, char *ekey, + uint16_t sport, uint16_t dport) +{ + struct sadb_msg smsg; + struct sadb_sa sa; + struct sadb_address sa_src, sa_dst; + struct sadb_key sa_akey, sa_ekey; + struct sadb_spirange sa_spirange; + struct iovec iov[IOV_CNT]; + ssize_t n; + int len = 0; + int iov_cnt; + struct sockaddr_storage ssrc, sdst, smask, dmask; + struct sockaddr *saptr; + + if (!pid) + pid = getpid(); + + /* we need clean sockaddr... no ports set */ + memset(&ssrc, 0, sizeof(ssrc)); + memset(&smask, 0, sizeof(smask)); + if ((saptr = addr2sa(af, src, 0))) + memcpy(&ssrc, saptr, sizeof(ssrc)); + switch (af) { + case AF_INET: + memset(&((struct sockaddr_in *)&smask)->sin_addr, 0xff, 32/8); + break; + case AF_INET6: + memset(&((struct sockaddr_in6 *)&smask)->sin6_addr, 0xff, + 128/8); + break; + default: + return (-1); + } + smask.ss_family = ssrc.ss_family; + smask.ss_len = ssrc.ss_len; + + memset(&sdst, 0, sizeof(sdst)); + memset(&dmask, 0, sizeof(dmask)); + if ((saptr = addr2sa(af, dst, 0))) + memcpy(&sdst, saptr, sizeof(sdst)); + switch (af) { + case AF_INET: + memset(&((struct sockaddr_in *)&dmask)->sin_addr, 0xff, 32/8); + break; + case AF_INET6: + memset(&((struct sockaddr_in6 *)&dmask)->sin6_addr, 0xff, + 128/8); + break; + default: + return (-1); + } + dmask.ss_family = sdst.ss_family; + dmask.ss_len = sdst.ss_len; + + memset(&smsg, 0, sizeof(smsg)); + smsg.sadb_msg_version = PF_KEY_V2; + smsg.sadb_msg_seq = ++sadb_msg_seq; + smsg.sadb_msg_pid = pid; + smsg.sadb_msg_len = sizeof(smsg) / 8; + smsg.sadb_msg_type = mtype; + smsg.sadb_msg_satype = satype; + + switch (mtype) { + case SADB_GETSPI: + memset(&sa_spirange, 0, sizeof(sa_spirange)); + sa_spirange.sadb_spirange_exttype = SADB_EXT_SPIRANGE; + sa_spirange.sadb_spirange_len = sizeof(sa_spirange) / 8; + sa_spirange.sadb_spirange_min = 0x100; + sa_spirange.sadb_spirange_max = 0xffffffff; + sa_spirange.sadb_spirange_reserved = 0; + break; + case SADB_ADD: + case SADB_UPDATE: + case SADB_DELETE: + memset(&sa, 0, sizeof(sa)); + sa.sadb_sa_exttype = SADB_EXT_SA; + sa.sadb_sa_len = sizeof(sa) / 8; + sa.sadb_sa_replay = 0; + sa.sadb_sa_spi = spi; + sa.sadb_sa_state = SADB_SASTATE_MATURE; + break; + } + + memset(&sa_src, 0, sizeof(sa_src)); + sa_src.sadb_address_exttype = SADB_EXT_ADDRESS_SRC; + sa_src.sadb_address_len = (sizeof(sa_src) + ROUNDUP(ssrc.ss_len)) / 8; + + memset(&sa_dst, 0, sizeof(sa_dst)); + sa_dst.sadb_address_exttype = SADB_EXT_ADDRESS_DST; + sa_dst.sadb_address_len = (sizeof(sa_dst) + ROUNDUP(sdst.ss_len)) / 8; + + sa.sadb_sa_auth = aalg; + sa.sadb_sa_encrypt = SADB_X_EALG_AES; /* XXX */ + + switch (mtype) { + case SADB_ADD: + case SADB_UPDATE: + memset(&sa_akey, 0, sizeof(sa_akey)); + sa_akey.sadb_key_exttype = SADB_EXT_KEY_AUTH; + sa_akey.sadb_key_len = (sizeof(sa_akey) + + ((alen + 7) / 8) * 8) / 8; + sa_akey.sadb_key_bits = 8 * alen; + + memset(&sa_ekey, 0, sizeof(sa_ekey)); + sa_ekey.sadb_key_exttype = SADB_EXT_KEY_ENCRYPT; + sa_ekey.sadb_key_len = (sizeof(sa_ekey) + + ((elen + 7) / 8) * 8) / 8; + sa_ekey.sadb_key_bits = 8 * elen; + + break; + } + + iov_cnt = 0; + + /* msghdr */ + iov[iov_cnt].iov_base = &smsg; + iov[iov_cnt].iov_len = sizeof(smsg); + iov_cnt++; + + switch (mtype) { + case SADB_ADD: + case SADB_UPDATE: + case SADB_DELETE: + /* SA hdr */ + iov[iov_cnt].iov_base = &sa; + iov[iov_cnt].iov_len = sizeof(sa); + smsg.sadb_msg_len += sa.sadb_sa_len; + iov_cnt++; + break; + case SADB_GETSPI: + /* SPI range */ + iov[iov_cnt].iov_base = &sa_spirange; + iov[iov_cnt].iov_len = sizeof(sa_spirange); + smsg.sadb_msg_len += sa_spirange.sadb_spirange_len; + iov_cnt++; + break; + } + + /* dest addr */ + iov[iov_cnt].iov_base = &sa_dst; + iov[iov_cnt].iov_len = sizeof(sa_dst); + iov_cnt++; + iov[iov_cnt].iov_base = &sdst; + iov[iov_cnt].iov_len = ROUNDUP(sdst.ss_len); + smsg.sadb_msg_len += sa_dst.sadb_address_len; + iov_cnt++; + + /* src addr */ + iov[iov_cnt].iov_base = &sa_src; + iov[iov_cnt].iov_len = sizeof(sa_src); + iov_cnt++; + iov[iov_cnt].iov_base = &ssrc; + iov[iov_cnt].iov_len = ROUNDUP(ssrc.ss_len); + smsg.sadb_msg_len += sa_src.sadb_address_len; + iov_cnt++; + + switch (mtype) { + case SADB_ADD: + case SADB_UPDATE: + if (alen) { + /* auth key */ + iov[iov_cnt].iov_base = &sa_akey; + iov[iov_cnt].iov_len = sizeof(sa_akey); + iov_cnt++; + iov[iov_cnt].iov_base = akey; + iov[iov_cnt].iov_len = ((alen + 7) / 8) * 8; + smsg.sadb_msg_len += sa_akey.sadb_key_len; + iov_cnt++; + } + if (elen) { + /* encryption key */ + iov[iov_cnt].iov_base = &sa_ekey; + iov[iov_cnt].iov_len = sizeof(sa_ekey); + iov_cnt++; + iov[iov_cnt].iov_base = ekey; + iov[iov_cnt].iov_len = ((elen + 7) / 8) * 8; + smsg.sadb_msg_len += sa_ekey.sadb_key_len; + iov_cnt++; + } + break; + } + + len = smsg.sadb_msg_len * 8; + do { + n = writev(sd, iov, iov_cnt); + } while (n == -1 && (errno == EAGAIN || errno == EINTR)); + + if (n == -1) { + log_warn("writev (%d/%d)", iov_cnt, len); + return (-1); + } + + return (0); +} + +int +pfkey_read(int sd, struct sadb_msg *h) +{ + struct sadb_msg hdr; + + if (recv(sd, &hdr, sizeof(hdr), MSG_PEEK) != sizeof(hdr)) { + if (errno == EAGAIN || errno == EINTR) + return (0); + log_warn("pfkey peek"); + return (-1); + } + + /* XXX: Only one message can be outstanding. */ + if (hdr.sadb_msg_seq == sadb_msg_seq && + hdr.sadb_msg_pid == pid) { + if (h) + *h = hdr; + return (0); + } + + /* not ours, discard */ + if (read(sd, &hdr, sizeof(hdr)) == -1) { + if (errno == EAGAIN || errno == EINTR) + return (0); + log_warn("pfkey read"); + return (-1); + } + + return (1); +} + +static int +pfkey_reply(int sd, uint32_t *spip) +{ + struct sadb_msg hdr, *msg; + struct sadb_ext *ext; + struct sadb_sa *sa; + uint8_t *data; + ssize_t len; + int rv; + + do { + rv = pfkey_read(sd, &hdr); + if (rv == -1) + return (-1); + } while (rv); + + if (hdr.sadb_msg_errno != 0) { + errno = hdr.sadb_msg_errno; + if (errno == ESRCH) + return (0); + else { + log_warn("pfkey"); + return (-1); + } + } + if ((data = reallocarray(NULL, hdr.sadb_msg_len, PFKEY2_CHUNK)) == NULL) { + log_warn("pfkey malloc"); + return (-1); + } + len = hdr.sadb_msg_len * PFKEY2_CHUNK; + if (read(sd, data, len) != len) { + log_warn("pfkey read"); + explicit_bzero(data, len); + free(data); + return (-1); + } + + if (hdr.sadb_msg_type == SADB_GETSPI) { + if (spip == NULL) { + explicit_bzero(data, len); + free(data); + return (0); + } + + msg = (struct sadb_msg *)data; + for (ext = (struct sadb_ext *)(msg + 1); + (size_t)((uint8_t *)ext - (uint8_t *)msg) < + msg->sadb_msg_len * PFKEY2_CHUNK; + ext = (struct sadb_ext *)((uint8_t *)ext + + ext->sadb_ext_len * PFKEY2_CHUNK)) { + if (ext->sadb_ext_type == SADB_EXT_SA) { + sa = (struct sadb_sa *) ext; + *spip = sa->sadb_sa_spi; + break; + } + } + } + explicit_bzero(data, len); + free(data); + return (0); +} + +static int +pfkey_sa_add(int af, union ldpd_addr *src, union ldpd_addr *dst, uint8_t keylen, + char *key, uint32_t *spi) +{ + if (pfkey_send(fd, SADB_X_SATYPE_TCPSIGNATURE, SADB_GETSPI, 0, + af, src, dst, 0, 0, 0, NULL, 0, 0, NULL, 0, 0) < 0) + return (-1); + if (pfkey_reply(fd, spi) < 0) + return (-1); + if (pfkey_send(fd, SADB_X_SATYPE_TCPSIGNATURE, SADB_UPDATE, 0, + af, src, dst, *spi, 0, keylen, key, 0, 0, NULL, 0, 0) < 0) + return (-1); + if (pfkey_reply(fd, NULL) < 0) + return (-1); + return (0); +} + +static int +pfkey_sa_remove(int af, union ldpd_addr *src, union ldpd_addr *dst, + uint32_t *spi) +{ + if (pfkey_send(fd, SADB_X_SATYPE_TCPSIGNATURE, SADB_DELETE, 0, + af, src, dst, *spi, 0, 0, NULL, 0, 0, NULL, 0, 0) < 0) + return (-1); + if (pfkey_reply(fd, NULL) < 0) + return (-1); + *spi = 0; + return (0); +} + +static int +pfkey_md5sig_establish(struct nbr *nbr, struct nbr_params *nbrp) +{ + sleep(1); + + if (!nbr->auth.spi_out) + if (pfkey_sa_add(nbr->af, &nbr->laddr, &nbr->raddr, + nbrp->auth.md5key_len, nbrp->auth.md5key, + &nbr->auth.spi_out) == -1) + return (-1); + if (!nbr->auth.spi_in) + if (pfkey_sa_add(nbr->af, &nbr->raddr, &nbr->laddr, + nbrp->auth.md5key_len, nbrp->auth.md5key, + &nbr->auth.spi_in) == -1) + return (-1); + + nbr->auth.established = 1; + return (0); +} + +static int +pfkey_md5sig_remove(struct nbr *nbr) +{ + if (nbr->auth.spi_out) + if (pfkey_sa_remove(nbr->af, &nbr->laddr, &nbr->raddr, + &nbr->auth.spi_out) == -1) + return (-1); + if (nbr->auth.spi_in) + if (pfkey_sa_remove(nbr->af, &nbr->raddr, &nbr->laddr, + &nbr->auth.spi_in) == -1) + return (-1); + + nbr->auth.established = 0; + nbr->auth.spi_in = 0; + nbr->auth.spi_out = 0; + nbr->auth.method = AUTH_NONE; + memset(nbr->auth.md5key, 0, sizeof(nbr->auth.md5key)); + + return (0); +} + +int +pfkey_establish(struct nbr *nbr, struct nbr_params *nbrp) +{ + if (nbrp->auth.method == AUTH_NONE) + return (0); + + /* + * make sure we keep copies of everything we need to + * remove SAs and flows later again. + */ + nbr->auth.method = nbrp->auth.method; + + switch (nbr->auth.method) { + case AUTH_MD5SIG: + strlcpy(nbr->auth.md5key, nbrp->auth.md5key, + sizeof(nbr->auth.md5key)); + return (pfkey_md5sig_establish(nbr, nbrp)); + default: + break; + } + + return (0); +} + +int +pfkey_remove(struct nbr *nbr) +{ + if (nbr->auth.method == AUTH_NONE || !nbr->auth.established) + return (0); + + switch (nbr->auth.method) { + case AUTH_MD5SIG: + return (pfkey_md5sig_remove(nbr)); + default: + break; + } + + return (0); +} + +int +pfkey_init(void) +{ + if ((fd = socket(PF_KEY, SOCK_RAW | SOCK_CLOEXEC | SOCK_NONBLOCK, + PF_KEY_V2)) == -1) { + if (errno == EPROTONOSUPPORT) { + log_warnx("PF_KEY not available"); + sysdep.no_pfkey = 1; + return (-1); + } else + fatal("pfkey setup failed"); + } + return (fd); +} diff --git a/ldpd/socket.c b/ldpd/socket.c new file mode 100644 index 000000000..8f26771df --- /dev/null +++ b/ldpd/socket.c @@ -0,0 +1,391 @@ +/* $OpenBSD$ */ + +/* + * Copyright (c) 2016 Renato Westphal + * Copyright (c) 2009 Michele Marchetto + * Copyright (c) 2005 Claudio Jeker + * 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 "ldpd.h" +#include "ldpe.h" +#include "log.h" + +int +ldp_create_socket(int af, enum socket_type type) +{ + int fd, domain, proto; + union ldpd_addr addr; + struct sockaddr_storage local_sa; + int opt; + + /* create socket */ + switch (type) { + case LDP_SOCKET_DISC: + case LDP_SOCKET_EDISC: + domain = SOCK_DGRAM; + proto = IPPROTO_UDP; + break; + case LDP_SOCKET_SESSION: + domain = SOCK_STREAM; + proto = IPPROTO_TCP; + break; + default: + fatalx("ldp_create_socket: unknown socket type"); + } + fd = socket(af, domain | SOCK_NONBLOCK | SOCK_CLOEXEC, proto); + if (fd == -1) { + log_warn("%s: error creating socket", __func__); + return (-1); + } + + /* bind to a local address/port */ + switch (type) { + case LDP_SOCKET_DISC: + /* listen on all addresses */ + memset(&addr, 0, sizeof(addr)); + memcpy(&local_sa, addr2sa(af, &addr, LDP_PORT), + sizeof(local_sa)); + break; + case LDP_SOCKET_EDISC: + case LDP_SOCKET_SESSION: + addr = (ldp_af_conf_get(ldpd_conf, af))->trans_addr; + memcpy(&local_sa, addr2sa(af, &addr, LDP_PORT), + sizeof(local_sa)); + if (sock_set_bindany(fd, 1) == -1) { + close(fd); + return (-1); + } + break; + } + if (sock_set_reuse(fd, 1) == -1) { + close(fd); + return (-1); + } + if (bind(fd, (struct sockaddr *)&local_sa, local_sa.ss_len) == -1) { + log_warn("%s: error binding socket", __func__); + close(fd); + return (-1); + } + + /* set options */ + switch (af) { + case AF_INET: + if (sock_set_ipv4_tos(fd, IPTOS_PREC_INTERNETCONTROL) == -1) { + close(fd); + return (-1); + } + if (type == LDP_SOCKET_DISC) { + if (sock_set_ipv4_mcast_ttl(fd, + IP_DEFAULT_MULTICAST_TTL) == -1) { + close(fd); + return (-1); + } + if (sock_set_ipv4_mcast_loop(fd) == -1) { + close(fd); + return (-1); + } + } + if (type == LDP_SOCKET_DISC || type == LDP_SOCKET_EDISC) { + if (sock_set_ipv4_recvif(fd, 1) == -1) { + close(fd); + return (-1); + } + } + if (type == LDP_SOCKET_SESSION) { + if (sock_set_ipv4_ucast_ttl(fd, 255) == -1) { + close(fd); + return (-1); + } + } + break; + case AF_INET6: + if (sock_set_ipv6_dscp(fd, IPTOS_PREC_INTERNETCONTROL) == -1) { + close(fd); + return (-1); + } + if (type == LDP_SOCKET_DISC) { + if (sock_set_ipv6_mcast_loop(fd) == -1) { + close(fd); + return (-1); + } + if (sock_set_ipv6_mcast_hops(fd, 255) == -1) { + close(fd); + return (-1); + } + if (!(ldpd_conf->ipv6.flags & F_LDPD_AF_NO_GTSM)) { + if (sock_set_ipv6_minhopcount(fd, 255) == -1) { + close(fd); + return (-1); + } + } + } + if (type == LDP_SOCKET_DISC || type == LDP_SOCKET_EDISC) { + if (sock_set_ipv6_pktinfo(fd, 1) == -1) { + close(fd); + return (-1); + } + } + if (type == LDP_SOCKET_SESSION) { + if (sock_set_ipv6_ucast_hops(fd, 255) == -1) { + close(fd); + return (-1); + } + } + break; + } + switch (type) { + case LDP_SOCKET_DISC: + case LDP_SOCKET_EDISC: + sock_set_recvbuf(fd); + break; + case LDP_SOCKET_SESSION: + if (listen(fd, LDP_BACKLOG) == -1) + log_warn("%s: error listening on socket", __func__); + + opt = 1; + if (setsockopt(fd, IPPROTO_TCP, TCP_MD5SIG, &opt, + sizeof(opt)) == -1) { + if (errno == ENOPROTOOPT) { /* system w/o md5sig */ + log_warnx("md5sig not available, disabling"); + sysdep.no_md5sig = 1; + } else { + close(fd); + return (-1); + } + } + break; + } + + return (fd); +} + +void +sock_set_recvbuf(int fd) +{ + int bsize; + + bsize = 65535; + while (setsockopt(fd, SOL_SOCKET, SO_RCVBUF, &bsize, + sizeof(bsize)) == -1) + bsize /= 2; +} + +int +sock_set_reuse(int fd, int enable) +{ + if (setsockopt(fd, SOL_SOCKET, SO_REUSEADDR, &enable, + sizeof(int)) < 0) { + log_warn("%s: error setting SO_REUSEADDR", __func__); + return (-1); + } + + return (0); +} + +int +sock_set_bindany(int fd, int enable) +{ + if (setsockopt(fd, SOL_SOCKET, SO_BINDANY, &enable, + sizeof(int)) < 0) { + log_warn("%s: error setting SO_BINDANY", __func__); + return (-1); + } + + return (0); +} + +int +sock_set_ipv4_tos(int fd, int tos) +{ + if (setsockopt(fd, IPPROTO_IP, IP_TOS, (int *)&tos, sizeof(tos)) < 0) { + log_warn("%s: error setting IP_TOS to 0x%x", __func__, tos); + return (-1); + } + + return (0); +} + +int +sock_set_ipv4_recvif(int fd, int enable) +{ + if (setsockopt(fd, IPPROTO_IP, IP_RECVIF, &enable, + sizeof(enable)) < 0) { + log_warn("%s: error setting IP_RECVIF", __func__); + return (-1); + } + return (0); +} + +int +sock_set_ipv4_minttl(int fd, int ttl) +{ + if (setsockopt(fd, IPPROTO_IP, IP_MINTTL, &ttl, sizeof(ttl)) < 0) { + log_warn("%s: error setting IP_MINTTL", __func__); + return (-1); + } + + return (0); +} + +int +sock_set_ipv4_ucast_ttl(int fd, int ttl) +{ + if (setsockopt(fd, IPPROTO_IP, IP_TTL, &ttl, sizeof(ttl)) < 0) { + log_warn("%s: error setting IP_TTL", __func__); + return (-1); + } + + return (0); +} + +int +sock_set_ipv4_mcast_ttl(int fd, uint8_t ttl) +{ + if (setsockopt(fd, IPPROTO_IP, IP_MULTICAST_TTL, + (char *)&ttl, sizeof(ttl)) < 0) { + log_warn("%s: error setting IP_MULTICAST_TTL to %d", + __func__, ttl); + return (-1); + } + + return (0); +} + +int +sock_set_ipv4_mcast(struct iface *iface) +{ + in_addr_t addr; + + addr = if_get_ipv4_addr(iface); + + if (setsockopt(global.ipv4.ldp_disc_socket, IPPROTO_IP, IP_MULTICAST_IF, + &addr, sizeof(addr)) < 0) { + log_warn("%s: error setting IP_MULTICAST_IF, interface %s", + __func__, iface->name); + return (-1); + } + + return (0); +} + +int +sock_set_ipv4_mcast_loop(int fd) +{ + uint8_t loop = 0; + + if (setsockopt(fd, IPPROTO_IP, IP_MULTICAST_LOOP, + (char *)&loop, sizeof(loop)) < 0) { + log_warn("%s: error setting IP_MULTICAST_LOOP", __func__); + return (-1); + } + + return (0); +} + +int +sock_set_ipv6_dscp(int fd, int dscp) +{ + if (setsockopt(fd, IPPROTO_IPV6, IPV6_TCLASS, &dscp, + sizeof(dscp)) < 0) { + log_warn("%s: error setting IPV6_TCLASS", __func__); + return (-1); + } + + return (0); +} + +int +sock_set_ipv6_pktinfo(int fd, int enable) +{ + if (setsockopt(fd, IPPROTO_IPV6, IPV6_RECVPKTINFO, &enable, + sizeof(enable)) < 0) { + log_warn("%s: error setting IPV6_RECVPKTINFO", __func__); + return (-1); + } + + return (0); +} + +int +sock_set_ipv6_minhopcount(int fd, int hoplimit) +{ + if (setsockopt(fd, IPPROTO_IPV6, IPV6_MINHOPCOUNT, + &hoplimit, sizeof(hoplimit)) < 0) { + log_warn("%s: error setting IPV6_MINHOPCOUNT", __func__); + return (-1); + } + + return (0); +} + +int +sock_set_ipv6_ucast_hops(int fd, int hoplimit) +{ + if (setsockopt(fd, IPPROTO_IPV6, IPV6_UNICAST_HOPS, + &hoplimit, sizeof(hoplimit)) < 0) { + log_warn("%s: error setting IPV6_UNICAST_HOPS", __func__); + return (-1); + } + + return (0); +} + +int +sock_set_ipv6_mcast_hops(int fd, int hoplimit) +{ + if (setsockopt(fd, IPPROTO_IPV6, IPV6_MULTICAST_HOPS, + &hoplimit, sizeof(hoplimit)) < 0) { + log_warn("%s: error setting IPV6_MULTICAST_HOPS", __func__); + return (-1); + } + + return (0); +} + +int +sock_set_ipv6_mcast(struct iface *iface) +{ + if (setsockopt(global.ipv6.ldp_disc_socket, IPPROTO_IPV6, + IPV6_MULTICAST_IF, &iface->ifindex, sizeof(iface->ifindex)) < 0) { + log_warn("%s: error setting IPV6_MULTICAST_IF, interface %s", + __func__, iface->name); + return (-1); + } + + return (0); +} + +int +sock_set_ipv6_mcast_loop(int fd) +{ + unsigned int loop = 0; + + if (setsockopt(fd, IPPROTO_IPV6, IPV6_MULTICAST_LOOP, + &loop, sizeof(loop)) < 0) { + log_warn("%s: error setting IPV6_MULTICAST_LOOP", __func__); + return (-1); + } + + return (0); +} diff --git a/ldpd/util.c b/ldpd/util.c new file mode 100644 index 000000000..981a5c8c2 --- /dev/null +++ b/ldpd/util.c @@ -0,0 +1,356 @@ +/* $OpenBSD$ */ + +/* + * Copyright (c) 2015 Renato Westphal + * Copyright (c) 2012 Alexander Bluhm + * 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 "ldpd.h" +#include "log.h" + +uint8_t +mask2prefixlen(in_addr_t ina) +{ + if (ina == 0) + return (0); + else + return (33 - ffs(ntohl(ina))); +} + +uint8_t +mask2prefixlen6(struct sockaddr_in6 *sa_in6) +{ + uint8_t l = 0, *ap, *ep; + + /* + * sin6_len is the size of the sockaddr so substract the offset of + * the possibly truncated sin6_addr struct. + */ + ap = (uint8_t *)&sa_in6->sin6_addr; + ep = (uint8_t *)sa_in6 + sa_in6->sin6_len; + for (; ap < ep; ap++) { + /* this "beauty" is adopted from sbin/route/show.c ... */ + switch (*ap) { + case 0xff: + l += 8; + break; + case 0xfe: + l += 7; + return (l); + case 0xfc: + l += 6; + return (l); + case 0xf8: + l += 5; + return (l); + case 0xf0: + l += 4; + return (l); + case 0xe0: + l += 3; + return (l); + case 0xc0: + l += 2; + return (l); + case 0x80: + l += 1; + return (l); + case 0x00: + return (l); + default: + fatalx("non contiguous inet6 netmask"); + } + } + + return (l); +} + +in_addr_t +prefixlen2mask(uint8_t prefixlen) +{ + if (prefixlen == 0) + return (0); + + return (htonl(0xffffffff << (32 - prefixlen))); +} + +struct in6_addr * +prefixlen2mask6(uint8_t prefixlen) +{ + static struct in6_addr mask; + int i; + + memset(&mask, 0, sizeof(mask)); + for (i = 0; i < prefixlen / 8; i++) + mask.s6_addr[i] = 0xff; + i = prefixlen % 8; + if (i) + mask.s6_addr[prefixlen / 8] = 0xff00 >> i; + + return (&mask); +} + +void +ldp_applymask(int af, union ldpd_addr *dest, const union ldpd_addr *src, + int prefixlen) +{ + struct in6_addr mask; + int i; + + switch (af) { + case AF_INET: + dest->v4.s_addr = src->v4.s_addr & prefixlen2mask(prefixlen); + break; + case AF_INET6: + memset(&mask, 0, sizeof(mask)); + for (i = 0; i < prefixlen / 8; i++) + mask.s6_addr[i] = 0xff; + i = prefixlen % 8; + if (i) + mask.s6_addr[prefixlen / 8] = 0xff00 >> i; + + for (i = 0; i < 16; i++) + dest->v6.s6_addr[i] = src->v6.s6_addr[i] & + mask.s6_addr[i]; + break; + default: + fatalx("ldp_applymask: unknown af"); + } +} + +int +ldp_addrcmp(int af, const union ldpd_addr *a, const union ldpd_addr *b) +{ + switch (af) { + case AF_INET: + if (a->v4.s_addr == b->v4.s_addr) + return (0); + return ((ntohl(a->v4.s_addr) > ntohl(b->v4.s_addr)) ? 1 : -1); + case AF_INET6: + return (memcmp(&a->v6, &b->v6, sizeof(struct in6_addr))); + default: + fatalx("ldp_addrcmp: unknown af"); + } +} + +int +ldp_addrisset(int af, const union ldpd_addr *addr) +{ + switch (af) { + case AF_UNSPEC: + return (0); + case AF_INET: + if (addr->v4.s_addr != INADDR_ANY) + return (1); + break; + case AF_INET6: + if (!IN6_IS_ADDR_UNSPECIFIED(&addr->v6)) + return (1); + break; + default: + fatalx("ldp_addrisset: unknown af"); + } + + return (0); +} + +int +ldp_prefixcmp(int af, const union ldpd_addr *a, const union ldpd_addr *b, + uint8_t prefixlen) +{ + in_addr_t mask, aa, ba; + int i; + uint8_t m; + + switch (af) { + case AF_INET: + if (prefixlen == 0) + return (0); + if (prefixlen > 32) + fatalx("ldp_prefixcmp: bad IPv4 prefixlen"); + mask = htonl(prefixlen2mask(prefixlen)); + aa = htonl(a->v4.s_addr) & mask; + ba = htonl(b->v4.s_addr) & mask; + return (aa - ba); + case AF_INET6: + if (prefixlen == 0) + return (0); + if (prefixlen > 128) + fatalx("ldp_prefixcmp: bad IPv6 prefixlen"); + for (i = 0; i < prefixlen / 8; i++) + if (a->v6.s6_addr[i] != b->v6.s6_addr[i]) + return (a->v6.s6_addr[i] - b->v6.s6_addr[i]); + i = prefixlen % 8; + if (i) { + m = 0xff00 >> i; + if ((a->v6.s6_addr[prefixlen / 8] & m) != + (b->v6.s6_addr[prefixlen / 8] & m)) + return ((a->v6.s6_addr[prefixlen / 8] & m) - + (b->v6.s6_addr[prefixlen / 8] & m)); + } + return (0); + default: + fatalx("ldp_prefixcmp: unknown af"); + } + return (-1); +} + +int +bad_addr_v4(struct in_addr addr) +{ + uint32_t a = ntohl(addr.s_addr); + + if (((a >> IN_CLASSA_NSHIFT) == 0) || + ((a >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) || + IN_MULTICAST(a) || IN_BADCLASS(a)) + return (1); + + return (0); +} + +int +bad_addr_v6(struct in6_addr *addr) +{ + if (IN6_IS_ADDR_UNSPECIFIED(addr) || + IN6_IS_ADDR_LOOPBACK(addr) || + IN6_IS_ADDR_MULTICAST(addr) || + IN6_IS_ADDR_SITELOCAL(addr) || + IN6_IS_ADDR_V4MAPPED(addr) || + IN6_IS_ADDR_V4COMPAT(addr)) + return (1); + + return (0); +} + +int +bad_addr(int af, union ldpd_addr *addr) +{ + switch (af) { + case AF_INET: + return (bad_addr_v4(addr->v4)); + case AF_INET6: + return (bad_addr_v6(&addr->v6)); + default: + fatalx("bad_addr: unknown af"); + } +} + +void +embedscope(struct sockaddr_in6 *sin6) +{ + uint16_t tmp16; + + if (IN6_IS_SCOPE_EMBED(&sin6->sin6_addr)) { + memcpy(&tmp16, &sin6->sin6_addr.s6_addr[2], sizeof(tmp16)); + if (tmp16 != 0) { + log_warnx("%s: address %s already has embeded scope %u", + __func__, log_sockaddr(sin6), ntohs(tmp16)); + } + tmp16 = htons(sin6->sin6_scope_id); + memcpy(&sin6->sin6_addr.s6_addr[2], &tmp16, sizeof(tmp16)); + sin6->sin6_scope_id = 0; + } +} + +void +recoverscope(struct sockaddr_in6 *sin6) +{ + uint16_t tmp16; + + if (sin6->sin6_scope_id != 0) + log_warnx("%s: address %s already has scope id %u", + __func__, log_sockaddr(sin6), sin6->sin6_scope_id); + + if (IN6_IS_SCOPE_EMBED(&sin6->sin6_addr)) { + memcpy(&tmp16, &sin6->sin6_addr.s6_addr[2], sizeof(tmp16)); + sin6->sin6_scope_id = ntohs(tmp16); + sin6->sin6_addr.s6_addr[2] = 0; + sin6->sin6_addr.s6_addr[3] = 0; + } +} + +void +addscope(struct sockaddr_in6 *sin6, uint32_t id) +{ + if (sin6->sin6_scope_id != 0) + log_warnx("%s: address %s already has scope id %u", __func__, + log_sockaddr(sin6), sin6->sin6_scope_id); + + if (IN6_IS_SCOPE_EMBED(&sin6->sin6_addr)) + sin6->sin6_scope_id = id; +} + +void +clearscope(struct in6_addr *in6) +{ + if (IN6_IS_SCOPE_EMBED(in6)) { + in6->s6_addr[2] = 0; + in6->s6_addr[3] = 0; + } +} + +struct sockaddr * +addr2sa(int af, union ldpd_addr *addr, uint16_t port) +{ + static struct sockaddr_storage ss; + struct sockaddr_in *sa_in = (struct sockaddr_in *)&ss; + struct sockaddr_in6 *sa_in6 = (struct sockaddr_in6 *)&ss; + + memset(&ss, 0, sizeof(ss)); + switch (af) { + case AF_INET: + sa_in->sin_family = AF_INET; + sa_in->sin_len = sizeof(struct sockaddr_in); + sa_in->sin_addr = addr->v4; + sa_in->sin_port = htons(port); + break; + case AF_INET6: + sa_in6->sin6_family = AF_INET6; + sa_in6->sin6_len = sizeof(struct sockaddr_in6); + sa_in6->sin6_addr = addr->v6; + sa_in6->sin6_port = htons(port); + break; + default: + fatalx("addr2sa: unknown af"); + } + + return ((struct sockaddr *)&ss); +} + +void +sa2addr(struct sockaddr *sa, int *af, union ldpd_addr *addr) +{ + struct sockaddr_in *sa_in = (struct sockaddr_in *)sa; + struct sockaddr_in6 *sa_in6 = (struct sockaddr_in6 *)sa; + + memset(addr, 0, sizeof(*addr)); + switch (sa->sa_family) { + case AF_INET: + *af = AF_INET; + addr->v4 = sa_in->sin_addr; + break; + case AF_INET6: + *af = AF_INET6; + addr->v6 = sa_in6->sin6_addr; + break; + default: + fatalx("sa2addr: unknown af"); + } +} diff --git a/lib/imsg-buffer.c b/lib/imsg-buffer.c new file mode 100644 index 000000000..61b2c095a --- /dev/null +++ b/lib/imsg-buffer.c @@ -0,0 +1,309 @@ +/* $OpenBSD$ */ + +/* + * 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 "imsg.h" + +int ibuf_realloc(struct ibuf *, size_t); +void ibuf_enqueue(struct msgbuf *, struct ibuf *); +void ibuf_dequeue(struct msgbuf *, struct ibuf *); + +struct ibuf * +ibuf_open(size_t len) +{ + struct ibuf *buf; + + if ((buf = calloc(1, sizeof(struct ibuf))) == NULL) + return (NULL); + if ((buf->buf = malloc(len)) == NULL) { + free(buf); + return (NULL); + } + buf->size = buf->max = len; + buf->fd = -1; + + return (buf); +} + +struct ibuf * +ibuf_dynamic(size_t len, size_t max) +{ + struct ibuf *buf; + + if (max < len) + return (NULL); + + if ((buf = ibuf_open(len)) == NULL) + return (NULL); + + if (max > 0) + buf->max = max; + + return (buf); +} + +int +ibuf_realloc(struct ibuf *buf, size_t len) +{ + u_char *b; + + /* on static buffers max is eq size and so the following fails */ + if (buf->wpos + len > buf->max) { + errno = ERANGE; + return (-1); + } + + b = realloc(buf->buf, buf->wpos + len); + if (b == NULL) + return (-1); + buf->buf = b; + buf->size = buf->wpos + len; + + return (0); +} + +int +ibuf_add(struct ibuf *buf, const void *data, size_t len) +{ + if (buf->wpos + len > buf->size) + if (ibuf_realloc(buf, len) == -1) + return (-1); + + memcpy(buf->buf + buf->wpos, data, len); + buf->wpos += len; + return (0); +} + +void * +ibuf_reserve(struct ibuf *buf, size_t len) +{ + void *b; + + if (buf->wpos + len > buf->size) + if (ibuf_realloc(buf, len) == -1) + return (NULL); + + b = buf->buf + buf->wpos; + buf->wpos += len; + return (b); +} + +void * +ibuf_seek(struct ibuf *buf, size_t pos, size_t len) +{ + /* only allowed to seek in already written parts */ + if (pos + len > buf->wpos) + return (NULL); + + return (buf->buf + pos); +} + +size_t +ibuf_size(struct ibuf *buf) +{ + return (buf->wpos); +} + +size_t +ibuf_left(struct ibuf *buf) +{ + return (buf->max - buf->wpos); +} + +void +ibuf_close(struct msgbuf *msgbuf, struct ibuf *buf) +{ + ibuf_enqueue(msgbuf, buf); +} + +int +ibuf_write(struct msgbuf *msgbuf) +{ + struct iovec iov[IOV_MAX]; + struct ibuf *buf; + unsigned int i = 0; + ssize_t n; + + memset(&iov, 0, sizeof(iov)); + TAILQ_FOREACH(buf, &msgbuf->bufs, entry) { + if (i >= IOV_MAX) + break; + iov[i].iov_base = buf->buf + buf->rpos; + iov[i].iov_len = buf->wpos - buf->rpos; + i++; + } + +again: + if ((n = writev(msgbuf->fd, iov, i)) == -1) { + if (errno == EINTR) + goto again; + if (errno == ENOBUFS) + errno = EAGAIN; + return (-1); + } + + if (n == 0) { /* connection closed */ + errno = 0; + return (0); + } + + msgbuf_drain(msgbuf, n); + + return (1); +} + +void +ibuf_free(struct ibuf *buf) +{ + if (buf == NULL) + return; + free(buf->buf); + free(buf); +} + +void +msgbuf_init(struct msgbuf *msgbuf) +{ + msgbuf->queued = 0; + msgbuf->fd = -1; + TAILQ_INIT(&msgbuf->bufs); +} + +void +msgbuf_drain(struct msgbuf *msgbuf, size_t n) +{ + struct ibuf *buf, *next; + + for (buf = TAILQ_FIRST(&msgbuf->bufs); buf != NULL && n > 0; + buf = next) { + next = TAILQ_NEXT(buf, entry); + if (buf->rpos + n >= buf->wpos) { + n -= buf->wpos - buf->rpos; + ibuf_dequeue(msgbuf, buf); + } else { + buf->rpos += n; + n = 0; + } + } +} + +void +msgbuf_clear(struct msgbuf *msgbuf) +{ + struct ibuf *buf; + + while ((buf = TAILQ_FIRST(&msgbuf->bufs)) != NULL) + ibuf_dequeue(msgbuf, buf); +} + +int +msgbuf_write(struct msgbuf *msgbuf) +{ + struct iovec iov[IOV_MAX]; + struct ibuf *buf; + unsigned int i = 0; + ssize_t n; + struct msghdr msg; + struct cmsghdr *cmsg; + union { + struct cmsghdr hdr; + char buf[CMSG_SPACE(sizeof(int))]; + } cmsgbuf; + + memset(&iov, 0, sizeof(iov)); + memset(&msg, 0, sizeof(msg)); + memset(&cmsgbuf, 0, sizeof(cmsgbuf)); + TAILQ_FOREACH(buf, &msgbuf->bufs, entry) { + if (i >= IOV_MAX) + break; + iov[i].iov_base = buf->buf + buf->rpos; + iov[i].iov_len = buf->wpos - buf->rpos; + i++; + if (buf->fd != -1) + break; + } + + msg.msg_iov = iov; + msg.msg_iovlen = i; + + if (buf != NULL && buf->fd != -1) { + msg.msg_control = (caddr_t)&cmsgbuf.buf; + msg.msg_controllen = sizeof(cmsgbuf.buf); + cmsg = CMSG_FIRSTHDR(&msg); + cmsg->cmsg_len = CMSG_LEN(sizeof(int)); + cmsg->cmsg_level = SOL_SOCKET; + cmsg->cmsg_type = SCM_RIGHTS; + *(int *)CMSG_DATA(cmsg) = buf->fd; + } + +again: + if ((n = sendmsg(msgbuf->fd, &msg, 0)) == -1) { + if (errno == EINTR) + goto again; + if (errno == ENOBUFS) + errno = EAGAIN; + return (-1); + } + + if (n == 0) { /* connection closed */ + errno = 0; + return (0); + } + + /* + * assumption: fd got sent if sendmsg sent anything + * this works because fds are passed one at a time + */ + if (buf != NULL && buf->fd != -1) { + close(buf->fd); + buf->fd = -1; + } + + msgbuf_drain(msgbuf, n); + + return (1); +} + +void +ibuf_enqueue(struct msgbuf *msgbuf, struct ibuf *buf) +{ + TAILQ_INSERT_TAIL(&msgbuf->bufs, buf, entry); + msgbuf->queued++; +} + +void +ibuf_dequeue(struct msgbuf *msgbuf, struct ibuf *buf) +{ + TAILQ_REMOVE(&msgbuf->bufs, buf, entry); + + if (buf->fd != -1) + close(buf->fd); + + msgbuf->queued--; + ibuf_free(buf); +} diff --git a/lib/imsg.c b/lib/imsg.c new file mode 100644 index 000000000..0c1cb8220 --- /dev/null +++ b/lib/imsg.c @@ -0,0 +1,302 @@ +/* $OpenBSD$ */ + +/* + * 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 "imsg.h" + +int imsg_fd_overhead = 0; + +int imsg_get_fd(struct imsgbuf *); + +void +imsg_init(struct imsgbuf *ibuf, int fd) +{ + msgbuf_init(&ibuf->w); + memset(&ibuf->r, 0, sizeof(ibuf->r)); + ibuf->fd = fd; + ibuf->w.fd = fd; + ibuf->pid = getpid(); + TAILQ_INIT(&ibuf->fds); +} + +ssize_t +imsg_read(struct imsgbuf *ibuf) +{ + struct msghdr msg; + struct cmsghdr *cmsg; + union { + struct cmsghdr hdr; + char buf[CMSG_SPACE(sizeof(int) * 1)]; + } cmsgbuf; + struct iovec iov; + ssize_t n = -1; + int fd; + struct imsg_fd *ifd; + + memset(&msg, 0, sizeof(msg)); + memset(&cmsgbuf, 0, sizeof(cmsgbuf)); + + iov.iov_base = ibuf->r.buf + ibuf->r.wpos; + iov.iov_len = sizeof(ibuf->r.buf) - ibuf->r.wpos; + msg.msg_iov = &iov; + msg.msg_iovlen = 1; + msg.msg_control = &cmsgbuf.buf; + msg.msg_controllen = sizeof(cmsgbuf.buf); + + if ((ifd = calloc(1, sizeof(struct imsg_fd))) == NULL) + return (-1); + +again: + if (getdtablecount() + imsg_fd_overhead + + (int)((CMSG_SPACE(sizeof(int))-CMSG_SPACE(0))/sizeof(int)) + >= getdtablesize()) { + errno = EAGAIN; + free(ifd); + return (-1); + } + + if ((n = recvmsg(ibuf->fd, &msg, 0)) == -1) { + if (errno == EINTR) + goto again; + goto fail; + } + + ibuf->r.wpos += n; + + for (cmsg = CMSG_FIRSTHDR(&msg); cmsg != NULL; + cmsg = CMSG_NXTHDR(&msg, cmsg)) { + if (cmsg->cmsg_level == SOL_SOCKET && + cmsg->cmsg_type == SCM_RIGHTS) { + int i; + int j; + + /* + * We only accept one file descriptor. Due to C + * padding rules, our control buffer might contain + * more than one fd, and we must close them. + */ + j = ((char *)cmsg + cmsg->cmsg_len - + (char *)CMSG_DATA(cmsg)) / sizeof(int); + for (i = 0; i < j; i++) { + fd = ((int *)CMSG_DATA(cmsg))[i]; + if (ifd != NULL) { + ifd->fd = fd; + TAILQ_INSERT_TAIL(&ibuf->fds, ifd, + entry); + ifd = NULL; + } else + close(fd); + } + } + /* we do not handle other ctl data level */ + } + +fail: + free(ifd); + return (n); +} + +ssize_t +imsg_get(struct imsgbuf *ibuf, struct imsg *imsg) +{ + size_t av, left, datalen; + + av = ibuf->r.wpos; + + if (IMSG_HEADER_SIZE > av) + return (0); + + memcpy(&imsg->hdr, ibuf->r.buf, sizeof(imsg->hdr)); + if (imsg->hdr.len < IMSG_HEADER_SIZE || + imsg->hdr.len > MAX_IMSGSIZE) { + errno = ERANGE; + return (-1); + } + if (imsg->hdr.len > av) + return (0); + datalen = imsg->hdr.len - IMSG_HEADER_SIZE; + ibuf->r.rptr = ibuf->r.buf + IMSG_HEADER_SIZE; + if (datalen == 0) + imsg->data = NULL; + else if ((imsg->data = malloc(datalen)) == NULL) + return (-1); + + if (imsg->hdr.flags & IMSGF_HASFD) + imsg->fd = imsg_get_fd(ibuf); + else + imsg->fd = -1; + + memcpy(imsg->data, ibuf->r.rptr, datalen); + + if (imsg->hdr.len < av) { + left = av - imsg->hdr.len; + memmove(&ibuf->r.buf, ibuf->r.buf + imsg->hdr.len, left); + ibuf->r.wpos = left; + } else + ibuf->r.wpos = 0; + + return (datalen + IMSG_HEADER_SIZE); +} + +int +imsg_compose(struct imsgbuf *ibuf, u_int32_t type, u_int32_t peerid, + pid_t pid, int fd, const void *data, u_int16_t datalen) +{ + struct ibuf *wbuf; + + if ((wbuf = imsg_create(ibuf, type, peerid, pid, datalen)) == NULL) + return (-1); + + if (imsg_add(wbuf, data, datalen) == -1) + return (-1); + + wbuf->fd = fd; + + imsg_close(ibuf, wbuf); + + return (1); +} + +int +imsg_composev(struct imsgbuf *ibuf, u_int32_t type, u_int32_t peerid, + pid_t pid, int fd, const struct iovec *iov, int iovcnt) +{ + struct ibuf *wbuf; + int i, datalen = 0; + + for (i = 0; i < iovcnt; i++) + datalen += iov[i].iov_len; + + if ((wbuf = imsg_create(ibuf, type, peerid, pid, datalen)) == NULL) + return (-1); + + for (i = 0; i < iovcnt; i++) + if (imsg_add(wbuf, iov[i].iov_base, iov[i].iov_len) == -1) + return (-1); + + wbuf->fd = fd; + + imsg_close(ibuf, wbuf); + + return (1); +} + +/* ARGSUSED */ +struct ibuf * +imsg_create(struct imsgbuf *ibuf, u_int32_t type, u_int32_t peerid, + pid_t pid, u_int16_t datalen) +{ + struct ibuf *wbuf; + struct imsg_hdr hdr; + + datalen += IMSG_HEADER_SIZE; + if (datalen > MAX_IMSGSIZE) { + errno = ERANGE; + return (NULL); + } + + hdr.type = type; + hdr.flags = 0; + hdr.peerid = peerid; + if ((hdr.pid = pid) == 0) + hdr.pid = ibuf->pid; + if ((wbuf = ibuf_dynamic(datalen, MAX_IMSGSIZE)) == NULL) { + return (NULL); + } + if (imsg_add(wbuf, &hdr, sizeof(hdr)) == -1) + return (NULL); + + return (wbuf); +} + +int +imsg_add(struct ibuf *msg, const void *data, u_int16_t datalen) +{ + if (datalen) + if (ibuf_add(msg, data, datalen) == -1) { + ibuf_free(msg); + return (-1); + } + return (datalen); +} + +void +imsg_close(struct imsgbuf *ibuf, struct ibuf *msg) +{ + struct imsg_hdr *hdr; + + hdr = (struct imsg_hdr *)msg->buf; + + hdr->flags &= ~IMSGF_HASFD; + if (msg->fd != -1) + hdr->flags |= IMSGF_HASFD; + + hdr->len = (u_int16_t)msg->wpos; + + ibuf_close(&ibuf->w, msg); +} + +void +imsg_free(struct imsg *imsg) +{ + free(imsg->data); +} + +int +imsg_get_fd(struct imsgbuf *ibuf) +{ + int fd; + struct imsg_fd *ifd; + + if ((ifd = TAILQ_FIRST(&ibuf->fds)) == NULL) + return (-1); + + fd = ifd->fd; + TAILQ_REMOVE(&ibuf->fds, ifd, entry); + free(ifd); + + return (fd); +} + +int +imsg_flush(struct imsgbuf *ibuf) +{ + while (ibuf->w.queued) + if (msgbuf_write(&ibuf->w) <= 0) + return (-1); + return (0); +} + +void +imsg_clear(struct imsgbuf *ibuf) +{ + int fd; + + msgbuf_clear(&ibuf->w); + while ((fd = imsg_get_fd(ibuf)) != -1) + close(fd); +} diff --git a/lib/imsg.h b/lib/imsg.h new file mode 100644 index 000000000..d053d0195 --- /dev/null +++ b/lib/imsg.h @@ -0,0 +1,112 @@ +/* $OpenBSD$ */ + +/* + * Copyright (c) 2006, 2007 Pierre-Yves Ritschard + * Copyright (c) 2006, 2007, 2008 Reyk Floeter + * 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. + */ + +#ifndef _IMSG_H_ +#define _IMSG_H_ + +#define IBUF_READ_SIZE 65535 +#define IMSG_HEADER_SIZE sizeof(struct imsg_hdr) +#define MAX_IMSGSIZE 16384 + +struct ibuf { + TAILQ_ENTRY(ibuf) entry; + u_char *buf; + size_t size; + size_t max; + size_t wpos; + size_t rpos; + int fd; +}; + +struct msgbuf { + TAILQ_HEAD(, ibuf) bufs; + u_int32_t queued; + int fd; +}; + +struct ibuf_read { + u_char buf[IBUF_READ_SIZE]; + u_char *rptr; + size_t wpos; +}; + +struct imsg_fd { + TAILQ_ENTRY(imsg_fd) entry; + int fd; +}; + +struct imsgbuf { + TAILQ_HEAD(, imsg_fd) fds; + struct ibuf_read r; + struct msgbuf w; + int fd; + pid_t pid; +}; + +#define IMSGF_HASFD 1 + +struct imsg_hdr { + u_int32_t type; + u_int16_t len; + u_int16_t flags; + u_int32_t peerid; + u_int32_t pid; +}; + +struct imsg { + struct imsg_hdr hdr; + int fd; + void *data; +}; + + +/* buffer.c */ +struct ibuf *ibuf_open(size_t); +struct ibuf *ibuf_dynamic(size_t, size_t); +int ibuf_add(struct ibuf *, const void *, size_t); +void *ibuf_reserve(struct ibuf *, size_t); +void *ibuf_seek(struct ibuf *, size_t, size_t); +size_t ibuf_size(struct ibuf *); +size_t ibuf_left(struct ibuf *); +void ibuf_close(struct msgbuf *, struct ibuf *); +int ibuf_write(struct msgbuf *); +void ibuf_free(struct ibuf *); +void msgbuf_init(struct msgbuf *); +void msgbuf_clear(struct msgbuf *); +int msgbuf_write(struct msgbuf *); +void msgbuf_drain(struct msgbuf *, size_t); + +/* imsg.c */ +void imsg_init(struct imsgbuf *, int); +ssize_t imsg_read(struct imsgbuf *); +ssize_t imsg_get(struct imsgbuf *, struct imsg *); +int imsg_compose(struct imsgbuf *, u_int32_t, u_int32_t, pid_t, + int, const void *, u_int16_t); +int imsg_composev(struct imsgbuf *, u_int32_t, u_int32_t, pid_t, + int, const struct iovec *, int); +struct ibuf *imsg_create(struct imsgbuf *, u_int32_t, u_int32_t, pid_t, + u_int16_t); +int imsg_add(struct ibuf *, const void *, u_int16_t); +void imsg_close(struct imsgbuf *, struct ibuf *); +void imsg_free(struct imsg *); +int imsg_flush(struct imsgbuf *); +void imsg_clear(struct imsgbuf *); + +#endif diff --git a/lib/openbsd-queue.h b/lib/openbsd-queue.h new file mode 100644 index 000000000..5e81fdd13 --- /dev/null +++ b/lib/openbsd-queue.h @@ -0,0 +1,533 @@ +/* $OpenBSD: queue.h,v 1.43 2015/12/28 19:38:40 millert Exp $ */ +/* $NetBSD: queue.h,v 1.11 1996/05/16 05:17:14 mycroft Exp $ */ + +/* + * Copyright (c) 1991, 1993 + * 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. + * 3. 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. + * + * @(#)queue.h 8.5 (Berkeley) 8/20/94 + */ + +#ifndef _SYS_QUEUE_H_ +#define _SYS_QUEUE_H_ + +/* + * This file defines five types of data structures: singly-linked lists, + * lists, simple queues, tail queues and XOR simple queues. + * + * + * A singly-linked list is headed by a single forward pointer. The elements + * are singly linked for minimum space and pointer manipulation overhead at + * the expense of O(n) removal for arbitrary elements. New elements can be + * added to the list after an existing element or at the head of the list. + * Elements being removed from the head of the list should use the explicit + * macro for this purpose for optimum efficiency. A singly-linked list may + * only be traversed in the forward direction. Singly-linked lists are ideal + * for applications with large datasets and few or no removals or for + * implementing a LIFO queue. + * + * A list is headed by a single forward pointer (or an array of forward + * pointers for a hash table header). The elements are doubly linked + * so that an arbitrary element can be removed without a need to + * traverse the list. New elements can be added to the list before + * or after an existing element or at the head of the list. A list + * may only be traversed in the forward direction. + * + * A simple queue is headed by a pair of pointers, one to the head of the + * list and the other to the tail of the list. The elements are singly + * linked to save space, so elements can only be removed from the + * head of the list. New elements can be added to the list before or after + * an existing element, at the head of the list, or at the end of the + * list. A simple queue may only be traversed in the forward direction. + * + * A tail queue is headed by a pair of pointers, one to the head of the + * list and the other to the tail of the list. The elements are doubly + * linked so that an arbitrary element can be removed without a need to + * traverse the list. New elements can be added to the list before or + * after an existing element, at the head of the list, or at the end of + * the list. A tail queue may be traversed in either direction. + * + * An XOR simple queue is used in the same way as a regular simple queue. + * The difference is that the head structure also includes a "cookie" that + * is XOR'd with the queue pointer (first, last or next) to generate the + * real pointer value. + * + * For details on the use of these macros, see the queue(3) manual page. + */ + +#if defined(QUEUE_MACRO_DEBUG) || (defined(_KERNEL) && defined(DIAGNOSTIC)) +#define _Q_INVALIDATE(a) (a) = ((void *)-1) +#else +#define _Q_INVALIDATE(a) +#endif + +/* + * Singly-linked List definitions. + */ +#define SLIST_HEAD(name, type) \ +struct name { \ + struct type *slh_first; /* first element */ \ +} + +#define SLIST_HEAD_INITIALIZER(head) \ + { NULL } + +#define SLIST_ENTRY(type) \ +struct { \ + struct type *sle_next; /* next element */ \ +} + +/* + * Singly-linked List access methods. + */ +#define SLIST_FIRST(head) ((head)->slh_first) +#define SLIST_END(head) NULL +#define SLIST_EMPTY(head) (SLIST_FIRST(head) == SLIST_END(head)) +#define SLIST_NEXT(elm, field) ((elm)->field.sle_next) + +#define SLIST_FOREACH(var, head, field) \ + for((var) = SLIST_FIRST(head); \ + (var) != SLIST_END(head); \ + (var) = SLIST_NEXT(var, field)) + +#define SLIST_FOREACH_SAFE(var, head, field, tvar) \ + for ((var) = SLIST_FIRST(head); \ + (var) && ((tvar) = SLIST_NEXT(var, field), 1); \ + (var) = (tvar)) + +/* + * Singly-linked List functions. + */ +#define SLIST_INIT(head) { \ + SLIST_FIRST(head) = SLIST_END(head); \ +} + +#define SLIST_INSERT_AFTER(slistelm, elm, field) do { \ + (elm)->field.sle_next = (slistelm)->field.sle_next; \ + (slistelm)->field.sle_next = (elm); \ +} while (0) + +#define SLIST_INSERT_HEAD(head, elm, field) do { \ + (elm)->field.sle_next = (head)->slh_first; \ + (head)->slh_first = (elm); \ +} while (0) + +#define SLIST_REMOVE_AFTER(elm, field) do { \ + (elm)->field.sle_next = (elm)->field.sle_next->field.sle_next; \ +} while (0) + +#define SLIST_REMOVE_HEAD(head, field) do { \ + (head)->slh_first = (head)->slh_first->field.sle_next; \ +} while (0) + +#define SLIST_REMOVE(head, elm, type, field) do { \ + if ((head)->slh_first == (elm)) { \ + SLIST_REMOVE_HEAD((head), field); \ + } else { \ + struct type *curelm = (head)->slh_first; \ + \ + while (curelm->field.sle_next != (elm)) \ + curelm = curelm->field.sle_next; \ + curelm->field.sle_next = \ + curelm->field.sle_next->field.sle_next; \ + } \ + _Q_INVALIDATE((elm)->field.sle_next); \ +} while (0) + +/* + * List definitions. + */ +#define LIST_HEAD(name, type) \ +struct name { \ + struct type *lh_first; /* first element */ \ +} + +#define LIST_HEAD_INITIALIZER(head) \ + { NULL } + +#define LIST_ENTRY(type) \ +struct { \ + struct type *le_next; /* next element */ \ + struct type **le_prev; /* address of previous next element */ \ +} + +/* + * List access methods. + */ +#define LIST_FIRST(head) ((head)->lh_first) +#define LIST_END(head) NULL +#define LIST_EMPTY(head) (LIST_FIRST(head) == LIST_END(head)) +#define LIST_NEXT(elm, field) ((elm)->field.le_next) + +#define LIST_FOREACH(var, head, field) \ + for((var) = LIST_FIRST(head); \ + (var)!= LIST_END(head); \ + (var) = LIST_NEXT(var, field)) + +#define LIST_FOREACH_SAFE(var, head, field, tvar) \ + for ((var) = LIST_FIRST(head); \ + (var) && ((tvar) = LIST_NEXT(var, field), 1); \ + (var) = (tvar)) + +/* + * List functions. + */ +#define LIST_INIT(head) do { \ + LIST_FIRST(head) = LIST_END(head); \ +} while (0) + +#define LIST_INSERT_AFTER(listelm, elm, field) do { \ + if (((elm)->field.le_next = (listelm)->field.le_next) != NULL) \ + (listelm)->field.le_next->field.le_prev = \ + &(elm)->field.le_next; \ + (listelm)->field.le_next = (elm); \ + (elm)->field.le_prev = &(listelm)->field.le_next; \ +} while (0) + +#define LIST_INSERT_BEFORE(listelm, elm, field) do { \ + (elm)->field.le_prev = (listelm)->field.le_prev; \ + (elm)->field.le_next = (listelm); \ + *(listelm)->field.le_prev = (elm); \ + (listelm)->field.le_prev = &(elm)->field.le_next; \ +} while (0) + +#define LIST_INSERT_HEAD(head, elm, field) do { \ + if (((elm)->field.le_next = (head)->lh_first) != NULL) \ + (head)->lh_first->field.le_prev = &(elm)->field.le_next;\ + (head)->lh_first = (elm); \ + (elm)->field.le_prev = &(head)->lh_first; \ +} while (0) + +#define LIST_REMOVE(elm, field) do { \ + if ((elm)->field.le_next != NULL) \ + (elm)->field.le_next->field.le_prev = \ + (elm)->field.le_prev; \ + *(elm)->field.le_prev = (elm)->field.le_next; \ + _Q_INVALIDATE((elm)->field.le_prev); \ + _Q_INVALIDATE((elm)->field.le_next); \ +} while (0) + +#define LIST_REPLACE(elm, elm2, field) do { \ + if (((elm2)->field.le_next = (elm)->field.le_next) != NULL) \ + (elm2)->field.le_next->field.le_prev = \ + &(elm2)->field.le_next; \ + (elm2)->field.le_prev = (elm)->field.le_prev; \ + *(elm2)->field.le_prev = (elm2); \ + _Q_INVALIDATE((elm)->field.le_prev); \ + _Q_INVALIDATE((elm)->field.le_next); \ +} while (0) + +/* + * Simple queue definitions. + */ +#define SIMPLEQ_HEAD(name, type) \ +struct name { \ + struct type *sqh_first; /* first element */ \ + struct type **sqh_last; /* addr of last next element */ \ +} + +#define SIMPLEQ_HEAD_INITIALIZER(head) \ + { NULL, &(head).sqh_first } + +#define SIMPLEQ_ENTRY(type) \ +struct { \ + struct type *sqe_next; /* next element */ \ +} + +/* + * Simple queue access methods. + */ +#define SIMPLEQ_FIRST(head) ((head)->sqh_first) +#define SIMPLEQ_END(head) NULL +#define SIMPLEQ_EMPTY(head) (SIMPLEQ_FIRST(head) == SIMPLEQ_END(head)) +#define SIMPLEQ_NEXT(elm, field) ((elm)->field.sqe_next) + +#define SIMPLEQ_FOREACH(var, head, field) \ + for((var) = SIMPLEQ_FIRST(head); \ + (var) != SIMPLEQ_END(head); \ + (var) = SIMPLEQ_NEXT(var, field)) + +#define SIMPLEQ_FOREACH_SAFE(var, head, field, tvar) \ + for ((var) = SIMPLEQ_FIRST(head); \ + (var) && ((tvar) = SIMPLEQ_NEXT(var, field), 1); \ + (var) = (tvar)) + +/* + * Simple queue functions. + */ +#define SIMPLEQ_INIT(head) do { \ + (head)->sqh_first = NULL; \ + (head)->sqh_last = &(head)->sqh_first; \ +} while (0) + +#define SIMPLEQ_INSERT_HEAD(head, elm, field) do { \ + if (((elm)->field.sqe_next = (head)->sqh_first) == NULL) \ + (head)->sqh_last = &(elm)->field.sqe_next; \ + (head)->sqh_first = (elm); \ +} while (0) + +#define SIMPLEQ_INSERT_TAIL(head, elm, field) do { \ + (elm)->field.sqe_next = NULL; \ + *(head)->sqh_last = (elm); \ + (head)->sqh_last = &(elm)->field.sqe_next; \ +} while (0) + +#define SIMPLEQ_INSERT_AFTER(head, listelm, elm, field) do { \ + if (((elm)->field.sqe_next = (listelm)->field.sqe_next) == NULL)\ + (head)->sqh_last = &(elm)->field.sqe_next; \ + (listelm)->field.sqe_next = (elm); \ +} while (0) + +#define SIMPLEQ_REMOVE_HEAD(head, field) do { \ + if (((head)->sqh_first = (head)->sqh_first->field.sqe_next) == NULL) \ + (head)->sqh_last = &(head)->sqh_first; \ +} while (0) + +#define SIMPLEQ_REMOVE_AFTER(head, elm, field) do { \ + if (((elm)->field.sqe_next = (elm)->field.sqe_next->field.sqe_next) \ + == NULL) \ + (head)->sqh_last = &(elm)->field.sqe_next; \ +} while (0) + +#define SIMPLEQ_CONCAT(head1, head2) do { \ + if (!SIMPLEQ_EMPTY((head2))) { \ + *(head1)->sqh_last = (head2)->sqh_first; \ + (head1)->sqh_last = (head2)->sqh_last; \ + SIMPLEQ_INIT((head2)); \ + } \ +} while (0) + +/* + * XOR Simple queue definitions. + */ +#define XSIMPLEQ_HEAD(name, type) \ +struct name { \ + struct type *sqx_first; /* first element */ \ + struct type **sqx_last; /* addr of last next element */ \ + unsigned long sqx_cookie; \ +} + +#define XSIMPLEQ_ENTRY(type) \ +struct { \ + struct type *sqx_next; /* next element */ \ +} + +/* + * XOR Simple queue access methods. + */ +#define XSIMPLEQ_XOR(head, ptr) ((__typeof(ptr))((head)->sqx_cookie ^ \ + (unsigned long)(ptr))) +#define XSIMPLEQ_FIRST(head) XSIMPLEQ_XOR(head, ((head)->sqx_first)) +#define XSIMPLEQ_END(head) NULL +#define XSIMPLEQ_EMPTY(head) (XSIMPLEQ_FIRST(head) == XSIMPLEQ_END(head)) +#define XSIMPLEQ_NEXT(head, elm, field) XSIMPLEQ_XOR(head, ((elm)->field.sqx_next)) + + +#define XSIMPLEQ_FOREACH(var, head, field) \ + for ((var) = XSIMPLEQ_FIRST(head); \ + (var) != XSIMPLEQ_END(head); \ + (var) = XSIMPLEQ_NEXT(head, var, field)) + +#define XSIMPLEQ_FOREACH_SAFE(var, head, field, tvar) \ + for ((var) = XSIMPLEQ_FIRST(head); \ + (var) && ((tvar) = XSIMPLEQ_NEXT(head, var, field), 1); \ + (var) = (tvar)) + +/* + * XOR Simple queue functions. + */ +#define XSIMPLEQ_INIT(head) do { \ + arc4random_buf(&(head)->sqx_cookie, sizeof((head)->sqx_cookie)); \ + (head)->sqx_first = XSIMPLEQ_XOR(head, NULL); \ + (head)->sqx_last = XSIMPLEQ_XOR(head, &(head)->sqx_first); \ +} while (0) + +#define XSIMPLEQ_INSERT_HEAD(head, elm, field) do { \ + if (((elm)->field.sqx_next = (head)->sqx_first) == \ + XSIMPLEQ_XOR(head, NULL)) \ + (head)->sqx_last = XSIMPLEQ_XOR(head, &(elm)->field.sqx_next); \ + (head)->sqx_first = XSIMPLEQ_XOR(head, (elm)); \ +} while (0) + +#define XSIMPLEQ_INSERT_TAIL(head, elm, field) do { \ + (elm)->field.sqx_next = XSIMPLEQ_XOR(head, NULL); \ + *(XSIMPLEQ_XOR(head, (head)->sqx_last)) = XSIMPLEQ_XOR(head, (elm)); \ + (head)->sqx_last = XSIMPLEQ_XOR(head, &(elm)->field.sqx_next); \ +} while (0) + +#define XSIMPLEQ_INSERT_AFTER(head, listelm, elm, field) do { \ + if (((elm)->field.sqx_next = (listelm)->field.sqx_next) == \ + XSIMPLEQ_XOR(head, NULL)) \ + (head)->sqx_last = XSIMPLEQ_XOR(head, &(elm)->field.sqx_next); \ + (listelm)->field.sqx_next = XSIMPLEQ_XOR(head, (elm)); \ +} while (0) + +#define XSIMPLEQ_REMOVE_HEAD(head, field) do { \ + if (((head)->sqx_first = XSIMPLEQ_XOR(head, \ + (head)->sqx_first)->field.sqx_next) == XSIMPLEQ_XOR(head, NULL)) \ + (head)->sqx_last = XSIMPLEQ_XOR(head, &(head)->sqx_first); \ +} while (0) + +#define XSIMPLEQ_REMOVE_AFTER(head, elm, field) do { \ + if (((elm)->field.sqx_next = XSIMPLEQ_XOR(head, \ + (elm)->field.sqx_next)->field.sqx_next) \ + == XSIMPLEQ_XOR(head, NULL)) \ + (head)->sqx_last = \ + XSIMPLEQ_XOR(head, &(elm)->field.sqx_next); \ +} while (0) + + +/* + * Tail queue definitions. + */ +#define TAILQ_HEAD(name, type) \ +struct name { \ + struct type *tqh_first; /* first element */ \ + struct type **tqh_last; /* addr of last next element */ \ +} + +#define TAILQ_HEAD_INITIALIZER(head) \ + { NULL, &(head).tqh_first } + +#define TAILQ_ENTRY(type) \ +struct { \ + struct type *tqe_next; /* next element */ \ + struct type **tqe_prev; /* address of previous next element */ \ +} + +/* + * Tail queue access methods. + */ +#define TAILQ_FIRST(head) ((head)->tqh_first) +#define TAILQ_END(head) NULL +#define TAILQ_NEXT(elm, field) ((elm)->field.tqe_next) +#define TAILQ_LAST(head, headname) \ + (*(((struct headname *)((head)->tqh_last))->tqh_last)) +/* XXX */ +#define TAILQ_PREV(elm, headname, field) \ + (*(((struct headname *)((elm)->field.tqe_prev))->tqh_last)) +#define TAILQ_EMPTY(head) \ + (TAILQ_FIRST(head) == TAILQ_END(head)) + +#define TAILQ_FOREACH(var, head, field) \ + for((var) = TAILQ_FIRST(head); \ + (var) != TAILQ_END(head); \ + (var) = TAILQ_NEXT(var, field)) + +#define TAILQ_FOREACH_SAFE(var, head, field, tvar) \ + for ((var) = TAILQ_FIRST(head); \ + (var) != TAILQ_END(head) && \ + ((tvar) = TAILQ_NEXT(var, field), 1); \ + (var) = (tvar)) + + +#define TAILQ_FOREACH_REVERSE(var, head, headname, field) \ + for((var) = TAILQ_LAST(head, headname); \ + (var) != TAILQ_END(head); \ + (var) = TAILQ_PREV(var, headname, field)) + +#define TAILQ_FOREACH_REVERSE_SAFE(var, head, headname, field, tvar) \ + for ((var) = TAILQ_LAST(head, headname); \ + (var) != TAILQ_END(head) && \ + ((tvar) = TAILQ_PREV(var, headname, field), 1); \ + (var) = (tvar)) + +/* + * Tail queue functions. + */ +#define TAILQ_INIT(head) do { \ + (head)->tqh_first = NULL; \ + (head)->tqh_last = &(head)->tqh_first; \ +} while (0) + +#define TAILQ_INSERT_HEAD(head, elm, field) do { \ + if (((elm)->field.tqe_next = (head)->tqh_first) != NULL) \ + (head)->tqh_first->field.tqe_prev = \ + &(elm)->field.tqe_next; \ + else \ + (head)->tqh_last = &(elm)->field.tqe_next; \ + (head)->tqh_first = (elm); \ + (elm)->field.tqe_prev = &(head)->tqh_first; \ +} while (0) + +#define TAILQ_INSERT_TAIL(head, elm, field) do { \ + (elm)->field.tqe_next = NULL; \ + (elm)->field.tqe_prev = (head)->tqh_last; \ + *(head)->tqh_last = (elm); \ + (head)->tqh_last = &(elm)->field.tqe_next; \ +} while (0) + +#define TAILQ_INSERT_AFTER(head, listelm, elm, field) do { \ + if (((elm)->field.tqe_next = (listelm)->field.tqe_next) != NULL)\ + (elm)->field.tqe_next->field.tqe_prev = \ + &(elm)->field.tqe_next; \ + else \ + (head)->tqh_last = &(elm)->field.tqe_next; \ + (listelm)->field.tqe_next = (elm); \ + (elm)->field.tqe_prev = &(listelm)->field.tqe_next; \ +} while (0) + +#define TAILQ_INSERT_BEFORE(listelm, elm, field) do { \ + (elm)->field.tqe_prev = (listelm)->field.tqe_prev; \ + (elm)->field.tqe_next = (listelm); \ + *(listelm)->field.tqe_prev = (elm); \ + (listelm)->field.tqe_prev = &(elm)->field.tqe_next; \ +} while (0) + +#define TAILQ_REMOVE(head, elm, field) do { \ + if (((elm)->field.tqe_next) != NULL) \ + (elm)->field.tqe_next->field.tqe_prev = \ + (elm)->field.tqe_prev; \ + else \ + (head)->tqh_last = (elm)->field.tqe_prev; \ + *(elm)->field.tqe_prev = (elm)->field.tqe_next; \ + _Q_INVALIDATE((elm)->field.tqe_prev); \ + _Q_INVALIDATE((elm)->field.tqe_next); \ +} while (0) + +#define TAILQ_REPLACE(head, elm, elm2, field) do { \ + if (((elm2)->field.tqe_next = (elm)->field.tqe_next) != NULL) \ + (elm2)->field.tqe_next->field.tqe_prev = \ + &(elm2)->field.tqe_next; \ + else \ + (head)->tqh_last = &(elm2)->field.tqe_next; \ + (elm2)->field.tqe_prev = (elm)->field.tqe_prev; \ + *(elm2)->field.tqe_prev = (elm2); \ + _Q_INVALIDATE((elm)->field.tqe_prev); \ + _Q_INVALIDATE((elm)->field.tqe_next); \ +} while (0) + +#define TAILQ_CONCAT(head1, head2, field) do { \ + if (!TAILQ_EMPTY(head2)) { \ + *(head1)->tqh_last = (head2)->tqh_first; \ + (head2)->tqh_first->field.tqe_prev = (head1)->tqh_last; \ + (head1)->tqh_last = (head2)->tqh_last; \ + TAILQ_INIT((head2)); \ + } \ +} while (0) + +#endif /* !_SYS_QUEUE_H_ */ diff --git a/lib/openbsd-tree.h b/lib/openbsd-tree.h new file mode 100644 index 000000000..e6502b1e7 --- /dev/null +++ b/lib/openbsd-tree.h @@ -0,0 +1,748 @@ +/* $OpenBSD: tree.h,v 1.14 2015/05/25 03:07:49 deraadt Exp $ */ +/* + * Copyright 2002 Niels Provos + * 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 THE AUTHOR ``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 AUTHOR 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. + */ + +#ifndef _SYS_TREE_H_ +#define _SYS_TREE_H_ + +/* + * This file defines data structures for different types of trees: + * splay trees and red-black trees. + * + * A splay tree is a self-organizing data structure. Every operation + * on the tree causes a splay to happen. The splay moves the requested + * node to the root of the tree and partly rebalances it. + * + * This has the benefit that request locality causes faster lookups as + * the requested nodes move to the top of the tree. On the other hand, + * every lookup causes memory writes. + * + * The Balance Theorem bounds the total access time for m operations + * and n inserts on an initially empty tree as O((m + n)lg n). The + * amortized cost for a sequence of m accesses to a splay tree is O(lg n); + * + * A red-black tree is a binary search tree with the node color as an + * extra attribute. It fulfills a set of conditions: + * - every search path from the root to a leaf consists of the + * same number of black nodes, + * - each red node (except for the root) has a black parent, + * - each leaf node is black. + * + * Every operation on a red-black tree is bounded as O(lg n). + * The maximum height of a red-black tree is 2lg (n+1). + */ + +#define SPLAY_HEAD(name, type) \ +struct name { \ + struct type *sph_root; /* root of the tree */ \ +} + +#define SPLAY_INITIALIZER(root) \ + { NULL } + +#define SPLAY_INIT(root) do { \ + (root)->sph_root = NULL; \ +} while (0) + +#define SPLAY_ENTRY(type) \ +struct { \ + struct type *spe_left; /* left element */ \ + struct type *spe_right; /* right element */ \ +} + +#define SPLAY_LEFT(elm, field) (elm)->field.spe_left +#define SPLAY_RIGHT(elm, field) (elm)->field.spe_right +#define SPLAY_ROOT(head) (head)->sph_root +#define SPLAY_EMPTY(head) (SPLAY_ROOT(head) == NULL) + +/* SPLAY_ROTATE_{LEFT,RIGHT} expect that tmp hold SPLAY_{RIGHT,LEFT} */ +#define SPLAY_ROTATE_RIGHT(head, tmp, field) do { \ + SPLAY_LEFT((head)->sph_root, field) = SPLAY_RIGHT(tmp, field); \ + SPLAY_RIGHT(tmp, field) = (head)->sph_root; \ + (head)->sph_root = tmp; \ +} while (0) + +#define SPLAY_ROTATE_LEFT(head, tmp, field) do { \ + SPLAY_RIGHT((head)->sph_root, field) = SPLAY_LEFT(tmp, field); \ + SPLAY_LEFT(tmp, field) = (head)->sph_root; \ + (head)->sph_root = tmp; \ +} while (0) + +#define SPLAY_LINKLEFT(head, tmp, field) do { \ + SPLAY_LEFT(tmp, field) = (head)->sph_root; \ + tmp = (head)->sph_root; \ + (head)->sph_root = SPLAY_LEFT((head)->sph_root, field); \ +} while (0) + +#define SPLAY_LINKRIGHT(head, tmp, field) do { \ + SPLAY_RIGHT(tmp, field) = (head)->sph_root; \ + tmp = (head)->sph_root; \ + (head)->sph_root = SPLAY_RIGHT((head)->sph_root, field); \ +} while (0) + +#define SPLAY_ASSEMBLE(head, node, left, right, field) do { \ + SPLAY_RIGHT(left, field) = SPLAY_LEFT((head)->sph_root, field); \ + SPLAY_LEFT(right, field) = SPLAY_RIGHT((head)->sph_root, field);\ + SPLAY_LEFT((head)->sph_root, field) = SPLAY_RIGHT(node, field); \ + SPLAY_RIGHT((head)->sph_root, field) = SPLAY_LEFT(node, field); \ +} while (0) + +/* Generates prototypes and inline functions */ + +#define SPLAY_PROTOTYPE(name, type, field, cmp) \ +void name##_SPLAY(struct name *, struct type *); \ +void name##_SPLAY_MINMAX(struct name *, int); \ +struct type *name##_SPLAY_INSERT(struct name *, struct type *); \ +struct type *name##_SPLAY_REMOVE(struct name *, struct type *); \ + \ +/* Finds the node with the same key as elm */ \ +static __inline struct type * \ +name##_SPLAY_FIND(struct name *head, struct type *elm) \ +{ \ + if (SPLAY_EMPTY(head)) \ + return(NULL); \ + name##_SPLAY(head, elm); \ + if ((cmp)(elm, (head)->sph_root) == 0) \ + return (head->sph_root); \ + return (NULL); \ +} \ + \ +static __inline struct type * \ +name##_SPLAY_NEXT(struct name *head, struct type *elm) \ +{ \ + name##_SPLAY(head, elm); \ + if (SPLAY_RIGHT(elm, field) != NULL) { \ + elm = SPLAY_RIGHT(elm, field); \ + while (SPLAY_LEFT(elm, field) != NULL) { \ + elm = SPLAY_LEFT(elm, field); \ + } \ + } else \ + elm = NULL; \ + return (elm); \ +} \ + \ +static __inline struct type * \ +name##_SPLAY_MIN_MAX(struct name *head, int val) \ +{ \ + name##_SPLAY_MINMAX(head, val); \ + return (SPLAY_ROOT(head)); \ +} + +/* Main splay operation. + * Moves node close to the key of elm to top + */ +#define SPLAY_GENERATE(name, type, field, cmp) \ +struct type * \ +name##_SPLAY_INSERT(struct name *head, struct type *elm) \ +{ \ + if (SPLAY_EMPTY(head)) { \ + SPLAY_LEFT(elm, field) = SPLAY_RIGHT(elm, field) = NULL; \ + } else { \ + int __comp; \ + name##_SPLAY(head, elm); \ + __comp = (cmp)(elm, (head)->sph_root); \ + if(__comp < 0) { \ + SPLAY_LEFT(elm, field) = SPLAY_LEFT((head)->sph_root, field);\ + SPLAY_RIGHT(elm, field) = (head)->sph_root; \ + SPLAY_LEFT((head)->sph_root, field) = NULL; \ + } else if (__comp > 0) { \ + SPLAY_RIGHT(elm, field) = SPLAY_RIGHT((head)->sph_root, field);\ + SPLAY_LEFT(elm, field) = (head)->sph_root; \ + SPLAY_RIGHT((head)->sph_root, field) = NULL; \ + } else \ + return ((head)->sph_root); \ + } \ + (head)->sph_root = (elm); \ + return (NULL); \ +} \ + \ +struct type * \ +name##_SPLAY_REMOVE(struct name *head, struct type *elm) \ +{ \ + struct type *__tmp; \ + if (SPLAY_EMPTY(head)) \ + return (NULL); \ + name##_SPLAY(head, elm); \ + if ((cmp)(elm, (head)->sph_root) == 0) { \ + if (SPLAY_LEFT((head)->sph_root, field) == NULL) { \ + (head)->sph_root = SPLAY_RIGHT((head)->sph_root, field);\ + } else { \ + __tmp = SPLAY_RIGHT((head)->sph_root, field); \ + (head)->sph_root = SPLAY_LEFT((head)->sph_root, field);\ + name##_SPLAY(head, elm); \ + SPLAY_RIGHT((head)->sph_root, field) = __tmp; \ + } \ + return (elm); \ + } \ + return (NULL); \ +} \ + \ +void \ +name##_SPLAY(struct name *head, struct type *elm) \ +{ \ + struct type __node, *__left, *__right, *__tmp; \ + int __comp; \ +\ + SPLAY_LEFT(&__node, field) = SPLAY_RIGHT(&__node, field) = NULL;\ + __left = __right = &__node; \ +\ + while ((__comp = (cmp)(elm, (head)->sph_root))) { \ + if (__comp < 0) { \ + __tmp = SPLAY_LEFT((head)->sph_root, field); \ + if (__tmp == NULL) \ + break; \ + if ((cmp)(elm, __tmp) < 0){ \ + SPLAY_ROTATE_RIGHT(head, __tmp, field); \ + if (SPLAY_LEFT((head)->sph_root, field) == NULL)\ + break; \ + } \ + SPLAY_LINKLEFT(head, __right, field); \ + } else if (__comp > 0) { \ + __tmp = SPLAY_RIGHT((head)->sph_root, field); \ + if (__tmp == NULL) \ + break; \ + if ((cmp)(elm, __tmp) > 0){ \ + SPLAY_ROTATE_LEFT(head, __tmp, field); \ + if (SPLAY_RIGHT((head)->sph_root, field) == NULL)\ + break; \ + } \ + SPLAY_LINKRIGHT(head, __left, field); \ + } \ + } \ + SPLAY_ASSEMBLE(head, &__node, __left, __right, field); \ +} \ + \ +/* Splay with either the minimum or the maximum element \ + * Used to find minimum or maximum element in tree. \ + */ \ +void name##_SPLAY_MINMAX(struct name *head, int __comp) \ +{ \ + struct type __node, *__left, *__right, *__tmp; \ +\ + SPLAY_LEFT(&__node, field) = SPLAY_RIGHT(&__node, field) = NULL;\ + __left = __right = &__node; \ +\ + while (1) { \ + if (__comp < 0) { \ + __tmp = SPLAY_LEFT((head)->sph_root, field); \ + if (__tmp == NULL) \ + break; \ + if (__comp < 0){ \ + SPLAY_ROTATE_RIGHT(head, __tmp, field); \ + if (SPLAY_LEFT((head)->sph_root, field) == NULL)\ + break; \ + } \ + SPLAY_LINKLEFT(head, __right, field); \ + } else if (__comp > 0) { \ + __tmp = SPLAY_RIGHT((head)->sph_root, field); \ + if (__tmp == NULL) \ + break; \ + if (__comp > 0) { \ + SPLAY_ROTATE_LEFT(head, __tmp, field); \ + if (SPLAY_RIGHT((head)->sph_root, field) == NULL)\ + break; \ + } \ + SPLAY_LINKRIGHT(head, __left, field); \ + } \ + } \ + SPLAY_ASSEMBLE(head, &__node, __left, __right, field); \ +} + +#define SPLAY_NEGINF -1 +#define SPLAY_INF 1 + +#define SPLAY_INSERT(name, x, y) name##_SPLAY_INSERT(x, y) +#define SPLAY_REMOVE(name, x, y) name##_SPLAY_REMOVE(x, y) +#define SPLAY_FIND(name, x, y) name##_SPLAY_FIND(x, y) +#define SPLAY_NEXT(name, x, y) name##_SPLAY_NEXT(x, y) +#define SPLAY_MIN(name, x) (SPLAY_EMPTY(x) ? NULL \ + : name##_SPLAY_MIN_MAX(x, SPLAY_NEGINF)) +#define SPLAY_MAX(name, x) (SPLAY_EMPTY(x) ? NULL \ + : name##_SPLAY_MIN_MAX(x, SPLAY_INF)) + +#define SPLAY_FOREACH(x, name, head) \ + for ((x) = SPLAY_MIN(name, head); \ + (x) != NULL; \ + (x) = SPLAY_NEXT(name, head, x)) + +/* Macros that define a red-black tree */ +#define RB_HEAD(name, type) \ +struct name { \ + struct type *rbh_root; /* root of the tree */ \ +} + +#define RB_INITIALIZER(root) \ + { NULL } + +#define RB_INIT(root) do { \ + (root)->rbh_root = NULL; \ +} while (0) + +#define RB_BLACK 0 +#define RB_RED 1 +#define RB_ENTRY(type) \ +struct { \ + struct type *rbe_left; /* left element */ \ + struct type *rbe_right; /* right element */ \ + struct type *rbe_parent; /* parent element */ \ + int rbe_color; /* node color */ \ +} + +#define RB_LEFT(elm, field) (elm)->field.rbe_left +#define RB_RIGHT(elm, field) (elm)->field.rbe_right +#define RB_PARENT(elm, field) (elm)->field.rbe_parent +#define RB_COLOR(elm, field) (elm)->field.rbe_color +#define RB_ROOT(head) (head)->rbh_root +#define RB_EMPTY(head) (RB_ROOT(head) == NULL) + +#define RB_SET(elm, parent, field) do { \ + RB_PARENT(elm, field) = parent; \ + RB_LEFT(elm, field) = RB_RIGHT(elm, field) = NULL; \ + RB_COLOR(elm, field) = RB_RED; \ +} while (0) + +#define RB_SET_BLACKRED(black, red, field) do { \ + RB_COLOR(black, field) = RB_BLACK; \ + RB_COLOR(red, field) = RB_RED; \ +} while (0) + +#ifndef RB_AUGMENT +#define RB_AUGMENT(x) do {} while (0) +#endif + +#define RB_ROTATE_LEFT(head, elm, tmp, field) do { \ + (tmp) = RB_RIGHT(elm, field); \ + if ((RB_RIGHT(elm, field) = RB_LEFT(tmp, field))) { \ + RB_PARENT(RB_LEFT(tmp, field), field) = (elm); \ + } \ + RB_AUGMENT(elm); \ + if ((RB_PARENT(tmp, field) = RB_PARENT(elm, field))) { \ + if ((elm) == RB_LEFT(RB_PARENT(elm, field), field)) \ + RB_LEFT(RB_PARENT(elm, field), field) = (tmp); \ + else \ + RB_RIGHT(RB_PARENT(elm, field), field) = (tmp); \ + } else \ + (head)->rbh_root = (tmp); \ + RB_LEFT(tmp, field) = (elm); \ + RB_PARENT(elm, field) = (tmp); \ + RB_AUGMENT(tmp); \ + if ((RB_PARENT(tmp, field))) \ + RB_AUGMENT(RB_PARENT(tmp, field)); \ +} while (0) + +#define RB_ROTATE_RIGHT(head, elm, tmp, field) do { \ + (tmp) = RB_LEFT(elm, field); \ + if ((RB_LEFT(elm, field) = RB_RIGHT(tmp, field))) { \ + RB_PARENT(RB_RIGHT(tmp, field), field) = (elm); \ + } \ + RB_AUGMENT(elm); \ + if ((RB_PARENT(tmp, field) = RB_PARENT(elm, field))) { \ + if ((elm) == RB_LEFT(RB_PARENT(elm, field), field)) \ + RB_LEFT(RB_PARENT(elm, field), field) = (tmp); \ + else \ + RB_RIGHT(RB_PARENT(elm, field), field) = (tmp); \ + } else \ + (head)->rbh_root = (tmp); \ + RB_RIGHT(tmp, field) = (elm); \ + RB_PARENT(elm, field) = (tmp); \ + RB_AUGMENT(tmp); \ + if ((RB_PARENT(tmp, field))) \ + RB_AUGMENT(RB_PARENT(tmp, field)); \ +} while (0) + +/* Generates prototypes and inline functions */ +#define RB_PROTOTYPE(name, type, field, cmp) \ + RB_PROTOTYPE_INTERNAL(name, type, field, cmp,) +#define RB_PROTOTYPE_STATIC(name, type, field, cmp) \ + RB_PROTOTYPE_INTERNAL(name, type, field, cmp, __attribute__((__unused__)) static) +#define RB_PROTOTYPE_INTERNAL(name, type, field, cmp, attr) \ +attr void name##_RB_INSERT_COLOR(struct name *, struct type *); \ +attr void name##_RB_REMOVE_COLOR(struct name *, struct type *, struct type *);\ +attr struct type *name##_RB_REMOVE(struct name *, struct type *); \ +attr struct type *name##_RB_INSERT(struct name *, struct type *); \ +attr struct type *name##_RB_FIND(struct name *, struct type *); \ +attr struct type *name##_RB_NFIND(struct name *, struct type *); \ +attr struct type *name##_RB_NEXT(struct type *); \ +attr struct type *name##_RB_PREV(struct type *); \ +attr struct type *name##_RB_MINMAX(struct name *, int); \ + \ + +/* Main rb operation. + * Moves node close to the key of elm to top + */ +#define RB_GENERATE(name, type, field, cmp) \ + RB_GENERATE_INTERNAL(name, type, field, cmp,) +#define RB_GENERATE_STATIC(name, type, field, cmp) \ + RB_GENERATE_INTERNAL(name, type, field, cmp, __attribute__((__unused__)) static) +#define RB_GENERATE_INTERNAL(name, type, field, cmp, attr) \ +attr void \ +name##_RB_INSERT_COLOR(struct name *head, struct type *elm) \ +{ \ + struct type *parent, *gparent, *tmp; \ + while ((parent = RB_PARENT(elm, field)) && \ + RB_COLOR(parent, field) == RB_RED) { \ + gparent = RB_PARENT(parent, field); \ + if (parent == RB_LEFT(gparent, field)) { \ + tmp = RB_RIGHT(gparent, field); \ + if (tmp && RB_COLOR(tmp, field) == RB_RED) { \ + RB_COLOR(tmp, field) = RB_BLACK; \ + RB_SET_BLACKRED(parent, gparent, field);\ + elm = gparent; \ + continue; \ + } \ + if (RB_RIGHT(parent, field) == elm) { \ + RB_ROTATE_LEFT(head, parent, tmp, field);\ + tmp = parent; \ + parent = elm; \ + elm = tmp; \ + } \ + RB_SET_BLACKRED(parent, gparent, field); \ + RB_ROTATE_RIGHT(head, gparent, tmp, field); \ + } else { \ + tmp = RB_LEFT(gparent, field); \ + if (tmp && RB_COLOR(tmp, field) == RB_RED) { \ + RB_COLOR(tmp, field) = RB_BLACK; \ + RB_SET_BLACKRED(parent, gparent, field);\ + elm = gparent; \ + continue; \ + } \ + if (RB_LEFT(parent, field) == elm) { \ + RB_ROTATE_RIGHT(head, parent, tmp, field);\ + tmp = parent; \ + parent = elm; \ + elm = tmp; \ + } \ + RB_SET_BLACKRED(parent, gparent, field); \ + RB_ROTATE_LEFT(head, gparent, tmp, field); \ + } \ + } \ + RB_COLOR(head->rbh_root, field) = RB_BLACK; \ +} \ + \ +attr void \ +name##_RB_REMOVE_COLOR(struct name *head, struct type *parent, struct type *elm) \ +{ \ + struct type *tmp; \ + while ((elm == NULL || RB_COLOR(elm, field) == RB_BLACK) && \ + elm != RB_ROOT(head)) { \ + if (RB_LEFT(parent, field) == elm) { \ + tmp = RB_RIGHT(parent, field); \ + if (RB_COLOR(tmp, field) == RB_RED) { \ + RB_SET_BLACKRED(tmp, parent, field); \ + RB_ROTATE_LEFT(head, parent, tmp, field);\ + tmp = RB_RIGHT(parent, field); \ + } \ + if ((RB_LEFT(tmp, field) == NULL || \ + RB_COLOR(RB_LEFT(tmp, field), field) == RB_BLACK) &&\ + (RB_RIGHT(tmp, field) == NULL || \ + RB_COLOR(RB_RIGHT(tmp, field), field) == RB_BLACK)) {\ + RB_COLOR(tmp, field) = RB_RED; \ + elm = parent; \ + parent = RB_PARENT(elm, field); \ + } else { \ + if (RB_RIGHT(tmp, field) == NULL || \ + RB_COLOR(RB_RIGHT(tmp, field), field) == RB_BLACK) {\ + struct type *oleft; \ + if ((oleft = RB_LEFT(tmp, field)))\ + RB_COLOR(oleft, field) = RB_BLACK;\ + RB_COLOR(tmp, field) = RB_RED; \ + RB_ROTATE_RIGHT(head, tmp, oleft, field);\ + tmp = RB_RIGHT(parent, field); \ + } \ + RB_COLOR(tmp, field) = RB_COLOR(parent, field);\ + RB_COLOR(parent, field) = RB_BLACK; \ + if (RB_RIGHT(tmp, field)) \ + RB_COLOR(RB_RIGHT(tmp, field), field) = RB_BLACK;\ + RB_ROTATE_LEFT(head, parent, tmp, field);\ + elm = RB_ROOT(head); \ + break; \ + } \ + } else { \ + tmp = RB_LEFT(parent, field); \ + if (RB_COLOR(tmp, field) == RB_RED) { \ + RB_SET_BLACKRED(tmp, parent, field); \ + RB_ROTATE_RIGHT(head, parent, tmp, field);\ + tmp = RB_LEFT(parent, field); \ + } \ + if ((RB_LEFT(tmp, field) == NULL || \ + RB_COLOR(RB_LEFT(tmp, field), field) == RB_BLACK) &&\ + (RB_RIGHT(tmp, field) == NULL || \ + RB_COLOR(RB_RIGHT(tmp, field), field) == RB_BLACK)) {\ + RB_COLOR(tmp, field) = RB_RED; \ + elm = parent; \ + parent = RB_PARENT(elm, field); \ + } else { \ + if (RB_LEFT(tmp, field) == NULL || \ + RB_COLOR(RB_LEFT(tmp, field), field) == RB_BLACK) {\ + struct type *oright; \ + if ((oright = RB_RIGHT(tmp, field)))\ + RB_COLOR(oright, field) = RB_BLACK;\ + RB_COLOR(tmp, field) = RB_RED; \ + RB_ROTATE_LEFT(head, tmp, oright, field);\ + tmp = RB_LEFT(parent, field); \ + } \ + RB_COLOR(tmp, field) = RB_COLOR(parent, field);\ + RB_COLOR(parent, field) = RB_BLACK; \ + if (RB_LEFT(tmp, field)) \ + RB_COLOR(RB_LEFT(tmp, field), field) = RB_BLACK;\ + RB_ROTATE_RIGHT(head, parent, tmp, field);\ + elm = RB_ROOT(head); \ + break; \ + } \ + } \ + } \ + if (elm) \ + RB_COLOR(elm, field) = RB_BLACK; \ +} \ + \ +attr struct type * \ +name##_RB_REMOVE(struct name *head, struct type *elm) \ +{ \ + struct type *child, *parent, *old = elm; \ + int color; \ + if (RB_LEFT(elm, field) == NULL) \ + child = RB_RIGHT(elm, field); \ + else if (RB_RIGHT(elm, field) == NULL) \ + child = RB_LEFT(elm, field); \ + else { \ + struct type *left; \ + elm = RB_RIGHT(elm, field); \ + while ((left = RB_LEFT(elm, field))) \ + elm = left; \ + child = RB_RIGHT(elm, field); \ + parent = RB_PARENT(elm, field); \ + color = RB_COLOR(elm, field); \ + if (child) \ + RB_PARENT(child, field) = parent; \ + if (parent) { \ + if (RB_LEFT(parent, field) == elm) \ + RB_LEFT(parent, field) = child; \ + else \ + RB_RIGHT(parent, field) = child; \ + RB_AUGMENT(parent); \ + } else \ + RB_ROOT(head) = child; \ + if (RB_PARENT(elm, field) == old) \ + parent = elm; \ + (elm)->field = (old)->field; \ + if (RB_PARENT(old, field)) { \ + if (RB_LEFT(RB_PARENT(old, field), field) == old)\ + RB_LEFT(RB_PARENT(old, field), field) = elm;\ + else \ + RB_RIGHT(RB_PARENT(old, field), field) = elm;\ + RB_AUGMENT(RB_PARENT(old, field)); \ + } else \ + RB_ROOT(head) = elm; \ + RB_PARENT(RB_LEFT(old, field), field) = elm; \ + if (RB_RIGHT(old, field)) \ + RB_PARENT(RB_RIGHT(old, field), field) = elm; \ + if (parent) { \ + left = parent; \ + do { \ + RB_AUGMENT(left); \ + } while ((left = RB_PARENT(left, field))); \ + } \ + goto color; \ + } \ + parent = RB_PARENT(elm, field); \ + color = RB_COLOR(elm, field); \ + if (child) \ + RB_PARENT(child, field) = parent; \ + if (parent) { \ + if (RB_LEFT(parent, field) == elm) \ + RB_LEFT(parent, field) = child; \ + else \ + RB_RIGHT(parent, field) = child; \ + RB_AUGMENT(parent); \ + } else \ + RB_ROOT(head) = child; \ +color: \ + if (color == RB_BLACK) \ + name##_RB_REMOVE_COLOR(head, parent, child); \ + return (old); \ +} \ + \ +/* Inserts a node into the RB tree */ \ +attr struct type * \ +name##_RB_INSERT(struct name *head, struct type *elm) \ +{ \ + struct type *tmp; \ + struct type *parent = NULL; \ + int comp = 0; \ + tmp = RB_ROOT(head); \ + while (tmp) { \ + parent = tmp; \ + comp = (cmp)(elm, parent); \ + if (comp < 0) \ + tmp = RB_LEFT(tmp, field); \ + else if (comp > 0) \ + tmp = RB_RIGHT(tmp, field); \ + else \ + return (tmp); \ + } \ + RB_SET(elm, parent, field); \ + if (parent != NULL) { \ + if (comp < 0) \ + RB_LEFT(parent, field) = elm; \ + else \ + RB_RIGHT(parent, field) = elm; \ + RB_AUGMENT(parent); \ + } else \ + RB_ROOT(head) = elm; \ + name##_RB_INSERT_COLOR(head, elm); \ + return (NULL); \ +} \ + \ +/* Finds the node with the same key as elm */ \ +attr struct type * \ +name##_RB_FIND(struct name *head, struct type *elm) \ +{ \ + struct type *tmp = RB_ROOT(head); \ + int comp; \ + while (tmp) { \ + comp = cmp(elm, tmp); \ + if (comp < 0) \ + tmp = RB_LEFT(tmp, field); \ + else if (comp > 0) \ + tmp = RB_RIGHT(tmp, field); \ + else \ + return (tmp); \ + } \ + return (NULL); \ +} \ + \ +/* Finds the first node greater than or equal to the search key */ \ +attr struct type * \ +name##_RB_NFIND(struct name *head, struct type *elm) \ +{ \ + struct type *tmp = RB_ROOT(head); \ + struct type *res = NULL; \ + int comp; \ + while (tmp) { \ + comp = cmp(elm, tmp); \ + if (comp < 0) { \ + res = tmp; \ + tmp = RB_LEFT(tmp, field); \ + } \ + else if (comp > 0) \ + tmp = RB_RIGHT(tmp, field); \ + else \ + return (tmp); \ + } \ + return (res); \ +} \ + \ +/* ARGSUSED */ \ +attr struct type * \ +name##_RB_NEXT(struct type *elm) \ +{ \ + if (RB_RIGHT(elm, field)) { \ + elm = RB_RIGHT(elm, field); \ + while (RB_LEFT(elm, field)) \ + elm = RB_LEFT(elm, field); \ + } else { \ + if (RB_PARENT(elm, field) && \ + (elm == RB_LEFT(RB_PARENT(elm, field), field))) \ + elm = RB_PARENT(elm, field); \ + else { \ + while (RB_PARENT(elm, field) && \ + (elm == RB_RIGHT(RB_PARENT(elm, field), field)))\ + elm = RB_PARENT(elm, field); \ + elm = RB_PARENT(elm, field); \ + } \ + } \ + return (elm); \ +} \ + \ +/* ARGSUSED */ \ +attr struct type * \ +name##_RB_PREV(struct type *elm) \ +{ \ + if (RB_LEFT(elm, field)) { \ + elm = RB_LEFT(elm, field); \ + while (RB_RIGHT(elm, field)) \ + elm = RB_RIGHT(elm, field); \ + } else { \ + if (RB_PARENT(elm, field) && \ + (elm == RB_RIGHT(RB_PARENT(elm, field), field))) \ + elm = RB_PARENT(elm, field); \ + else { \ + while (RB_PARENT(elm, field) && \ + (elm == RB_LEFT(RB_PARENT(elm, field), field)))\ + elm = RB_PARENT(elm, field); \ + elm = RB_PARENT(elm, field); \ + } \ + } \ + return (elm); \ +} \ + \ +attr struct type * \ +name##_RB_MINMAX(struct name *head, int val) \ +{ \ + struct type *tmp = RB_ROOT(head); \ + struct type *parent = NULL; \ + while (tmp) { \ + parent = tmp; \ + if (val < 0) \ + tmp = RB_LEFT(tmp, field); \ + else \ + tmp = RB_RIGHT(tmp, field); \ + } \ + return (parent); \ +} + +#define RB_NEGINF -1 +#define RB_INF 1 + +#define RB_INSERT(name, x, y) name##_RB_INSERT(x, y) +#define RB_REMOVE(name, x, y) name##_RB_REMOVE(x, y) +#define RB_FIND(name, x, y) name##_RB_FIND(x, y) +#define RB_NFIND(name, x, y) name##_RB_NFIND(x, y) +#define RB_NEXT(name, x, y) name##_RB_NEXT(y) +#define RB_PREV(name, x, y) name##_RB_PREV(y) +#define RB_MIN(name, x) name##_RB_MINMAX(x, RB_NEGINF) +#define RB_MAX(name, x) name##_RB_MINMAX(x, RB_INF) + +#define RB_FOREACH(x, name, head) \ + for ((x) = RB_MIN(name, head); \ + (x) != NULL; \ + (x) = name##_RB_NEXT(x)) + +#define RB_FOREACH_SAFE(x, name, head, y) \ + for ((x) = RB_MIN(name, head); \ + ((x) != NULL) && ((y) = name##_RB_NEXT(x), 1); \ + (x) = (y)) + +#define RB_FOREACH_REVERSE(x, name, head) \ + for ((x) = RB_MAX(name, head); \ + (x) != NULL; \ + (x) = name##_RB_PREV(x)) + +#define RB_FOREACH_REVERSE_SAFE(x, name, head, y) \ + for ((x) = RB_MAX(name, head); \ + ((x) != NULL) && ((y) = name##_RB_PREV(x), 1); \ + (x) = (y)) + +#endif /* _SYS_TREE_H_ */ -- 2.39.5