bird/proto/bgp/bgp.c

2110 lines
56 KiB
C
Raw Normal View History

2000-03-20 06:09:07 +08:00
/*
* BIRD -- The Border Gateway Protocol
*
* (c) 2000 Martin Mares <mj@ucw.cz>
* (c) 2008--2016 Ondrej Zajicek <santiago@crfreenet.org>
* (c) 2008--2016 CZ.NIC z.s.p.o.
2000-03-20 06:09:07 +08:00
*
* Can be freely distributed and used under the terms of the GNU GPL.
*/
2000-06-05 01:06:18 +08:00
/**
* DOC: Border Gateway Protocol
*
* The BGP protocol is implemented in three parts: |bgp.c| which takes care of
* the connection and most of the interface with BIRD core, |packets.c| handling
2000-06-05 01:06:18 +08:00
* both incoming and outgoing BGP packets and |attrs.c| containing functions for
* manipulation with BGP attribute lists.
*
* As opposed to the other existing routing daemons, BIRD has a sophisticated
* core architecture which is able to keep all the information needed by BGP in
* the primary routing table, therefore no complex data structures like a
* central BGP table are needed. This increases memory footprint of a BGP router
* with many connections, but not too much and, which is more important, it
* makes BGP much easier to implement.
2000-06-05 01:06:18 +08:00
*
* Each instance of BGP (corresponding to a single BGP peer) is described by a
* &bgp_proto structure to which are attached individual connections represented
* by &bgp_connection (usually, there exists only one connection, but during BGP
* session setup, there can be more of them). The connections are handled
* according to the BGP state machine defined in the RFC with all the timers and
* all the parameters configurable.
2000-06-05 01:06:18 +08:00
*
* In incoming direction, we listen on the connection's socket and each time we
* receive some input, we pass it to bgp_rx(). It decodes packet headers and the
* markers and passes complete packets to bgp_rx_packet() which distributes the
* packet according to its type.
2000-06-05 01:06:18 +08:00
*
* In outgoing direction, we gather all the routing updates and sort them to
* buckets (&bgp_bucket) according to their attributes (we keep a hash table for
* fast comparison of &rta's and a &fib which helps us to find if we already
* have another route for the same destination queued for sending, so that we
* can replace it with the new one immediately instead of sending both
* updates). There also exists a special bucket holding all the route
* withdrawals which cannot be queued anywhere else as they don't have any
* attributes. If we have any packet to send (due to either new routes or the
* connection tracking code wanting to send a Open, Keepalive or Notification
* message), we call bgp_schedule_packet() which sets the corresponding bit in a
* @packet_to_send bit field in &bgp_conn and as soon as the transmit socket
* buffer becomes empty, we call bgp_fire_tx(). It inspects state of all the
* packet type bits and calls the corresponding bgp_create_xx() functions,
* eventually rescheduling the same packet type if we have more data of the same
* type to send.
2000-06-05 01:06:18 +08:00
*
* The processing of attributes consists of two functions: bgp_decode_attrs()
* for checking of the attribute blocks and translating them to the language of
* BIRD's extended attributes and bgp_encode_attrs() which does the
* converse. Both functions are built around a @bgp_attr_table array describing
* all important characteristics of all known attributes. Unknown transitive
* attributes are attached to the route as %EAF_TYPE_OPAQUE byte streams.
*
* BGP protocol implements graceful restart in both restarting (local restart)
* and receiving (neighbor restart) roles. The first is handled mostly by the
* graceful restart code in the nest, BGP protocol just handles capabilities,
* sets @gr_wait and locks graceful restart until end-of-RIB mark is received.
* The second is implemented by internal restart of the BGP state to %BS_IDLE
* and protocol state to %PS_START, but keeping the protocol up from the core
* point of view and therefore maintaining received routes. Routing table
* refresh cycle (rt_refresh_begin(), rt_refresh_end()) is used for removing
* stale routes after reestablishment of BGP session during graceful restart.
*
* Supported standards:
* <itemize>
* <item> <rfc id="4271"> - Border Gateway Protocol 4 (BGP)
* <item> <rfc id="1997"> - BGP Communities Attribute
* <item> <rfc id="2385"> - Protection of BGP Sessions via TCP MD5 Signature
* <item> <rfc id="2545"> - Use of BGP Multiprotocol Extensions for IPv6
* <item> <rfc id="2918"> - Route Refresh Capability
* <item> <rfc id="3107"> - Carrying Label Information in BGP
* <item> <rfc id="4360"> - BGP Extended Communities Attribute
* <item> <rfc id="4364"> - BGP/MPLS IPv4 Virtual Private Networks
* <item> <rfc id="4456"> - BGP Route Reflection
* <item> <rfc id="4486"> - Subcodes for BGP Cease Notification Message
* <item> <rfc id="4659"> - BGP/MPLS IPv6 Virtual Private Networks
* <item> <rfc id="4724"> - Graceful Restart Mechanism for BGP
* <item> <rfc id="4760"> - Multiprotocol extensions for BGP
* <item> <rfc id="4798"> - Connecting IPv6 Islands over IPv4 MPLS
* <item> <rfc id="5065"> - AS confederations for BGP
* <item> <rfc id="5082"> - Generalized TTL Security Mechanism
* <item> <rfc id="5492"> - Capabilities Advertisement with BGP
* <item> <rfc id="5549"> - Advertising IPv4 NLRI with an IPv6 Next Hop
* <item> <rfc id="5575"> - Dissemination of Flow Specification Rules
* <item> <rfc id="5668"> - 4-Octet AS Specific BGP Extended Community
* <item> <rfc id="6286"> - AS-Wide Unique BGP Identifier
* <item> <rfc id="6608"> - Subcodes for BGP Finite State Machine Error
* <item> <rfc id="6793"> - BGP Support for 4-Octet AS Numbers
* <item> <rfc id="7313"> - Enhanced Route Refresh Capability for BGP
* <item> <rfc id="7606"> - Revised Error Handling for BGP UPDATE Messages
* <item> <rfc id="7911"> - Advertisement of Multiple Paths in BGP
* <item> <rfc id="7947"> - Internet Exchange BGP Route Server
* <item> <rfc id="8092"> - BGP Large Communities Attribute
* </itemize>
*/
2000-06-05 01:06:18 +08:00
#undef LOCAL_DEBUG
2000-03-20 06:09:07 +08:00
#include "nest/bird.h"
#include "nest/iface.h"
#include "nest/protocol.h"
#include "nest/route.h"
#include "nest/cli.h"
#include "nest/locks.h"
2000-03-20 06:09:07 +08:00
#include "conf/conf.h"
#include "lib/socket.h"
2000-03-31 02:44:23 +08:00
#include "lib/resource.h"
2002-11-13 16:47:06 +08:00
#include "lib/string.h"
2000-03-20 06:09:07 +08:00
#include "bgp.h"
2000-03-31 02:44:23 +08:00
struct linpool *bgp_linpool; /* Global temporary pool */
struct linpool *bgp_linpool2; /* Global temporary pool for bgp_rt_notify() */
static list bgp_sockets; /* Global list of listening sockets */
static void bgp_connect(struct bgp_proto *p);
static void bgp_active(struct bgp_proto *p);
static void bgp_update_bfd(struct bgp_proto *p, int use_bfd);
2000-03-20 06:09:07 +08:00
static int bgp_incoming_connection(sock *sk, uint dummy UNUSED);
static void bgp_listen_sock_err(sock *sk UNUSED, int err);
/**
* bgp_open - open a BGP instance
* @p: BGP instance
*
* This function allocates and configures shared BGP resources, mainly listening
* sockets. Should be called as the last step during initialization (when lock
* is acquired and neighbor is ready). When error, caller should change state to
* PS_DOWN and return immediately.
*/
static int
bgp_open(struct bgp_proto *p)
{
struct bgp_socket *bs = NULL;
struct iface *ifa = p->cf->strict_bind ? p->cf->iface : NULL;
ip_addr addr = p->cf->strict_bind ? p->cf->local_ip :
(ipa_is_ip4(p->cf->remote_ip) ? IPA_NONE4 : IPA_NONE6);
uint port = p->cf->local_port;
/* FIXME: Add some global init? */
if (!bgp_linpool)
init_list(&bgp_sockets);
/* We assume that cf->iface is defined iff cf->local_ip is link-local */
WALK_LIST(bs, bgp_sockets)
if (ipa_equal(bs->sk->saddr, addr) && (bs->sk->iface == ifa) && (bs->sk->sport == port))
{
bs->uc++;
p->sock = bs;
return 0;
}
sock *sk = sk_new(proto_pool);
sk->type = SK_TCP_PASSIVE;
sk->ttl = 255;
sk->saddr = addr;
sk->sport = port;
sk->flags = 0;
sk->tos = IP_PREC_INTERNET_CONTROL;
sk->rbsize = BGP_RX_BUFFER_SIZE;
sk->tbsize = BGP_TX_BUFFER_SIZE;
sk->rx_hook = bgp_incoming_connection;
sk->err_hook = bgp_listen_sock_err;
if (sk_open(sk) < 0)
goto err;
bs = mb_allocz(proto_pool, sizeof(struct bgp_socket));
bs->sk = sk;
bs->uc = 1;
p->sock = bs;
2014-05-18 17:42:26 +08:00
add_tail(&bgp_sockets, &bs->n);
if (!bgp_linpool)
{
2017-05-16 20:31:16 +08:00
bgp_linpool = lp_new_default(proto_pool);
bgp_linpool2 = lp_new_default(proto_pool);
}
return 0;
err:
sk_log_error(sk, p->p.name);
log(L_ERR "%s: Cannot open listening socket", p->p.name);
rfree(sk);
return -1;
}
/**
* bgp_close - close a BGP instance
* @p: BGP instance
*
* This function frees and deconfigures shared BGP resources.
*/
static void
bgp_close(struct bgp_proto *p)
{
struct bgp_socket *bs = p->sock;
ASSERT(bs && bs->uc);
if (--bs->uc)
return;
rfree(bs->sk);
rem_node(&bs->n);
mb_free(bs);
if (!EMPTY_LIST(bgp_sockets))
return;
rfree(bgp_linpool);
bgp_linpool = NULL;
rfree(bgp_linpool2);
bgp_linpool2 = NULL;
}
static inline int
bgp_setup_auth(struct bgp_proto *p, int enable)
{
if (p->cf->password)
{
int rv = sk_set_md5_auth(p->sock->sk,
p->cf->local_ip, p->cf->remote_ip, p->cf->iface,
enable ? p->cf->password : NULL, p->cf->setkey);
if (rv < 0)
sk_log_error(p->sock->sk, p->p.name);
return rv;
}
else
return 0;
}
static inline struct bgp_channel *
bgp_find_channel(struct bgp_proto *p, u32 afi)
{
struct bgp_channel *c;
WALK_LIST(c, p->p.channels)
if (c->afi == afi)
return c;
return NULL;
}
static void
bgp_startup(struct bgp_proto *p)
{
BGP_TRACE(D_EVENTS, "Started");
p->start_state = BSS_CONNECT;
2009-09-25 04:12:11 +08:00
if (!p->cf->passive)
bgp_active(p);
}
static void
bgp_startup_timeout(timer *t)
{
bgp_startup(t->data);
}
static void
bgp_initiate(struct bgp_proto *p)
{
int err_val;
if (bgp_open(p) < 0)
{ err_val = BEM_NO_SOCKET; goto err1; }
if (bgp_setup_auth(p, 1) < 0)
{ err_val = BEM_INVALID_MD5; goto err2; }
if (p->cf->bfd)
bgp_update_bfd(p, p->cf->bfd);
if (p->startup_delay)
{
p->start_state = BSS_DELAY;
BGP_TRACE(D_EVENTS, "Startup delayed by %d seconds due to errors", p->startup_delay);
bgp_start_timer(p->startup_timer, p->startup_delay);
}
else
bgp_startup(p);
return;
err2:
bgp_close(p);
err1:
p->p.disabled = 1;
bgp_store_error(p, NULL, BE_MISC, err_val);
proto_notify_state(&p->p, PS_DOWN);
return;
}
2000-06-05 01:06:18 +08:00
/**
* bgp_start_timer - start a BGP timer
* @t: timer
* @value: time to fire (0 to disable the timer)
*
* This functions calls tm_start() on @t with time @value and the amount of
* randomization suggested by the BGP standard. Please use it for all BGP
* timers.
2000-06-05 01:06:18 +08:00
*/
void
bgp_start_timer(timer *t, int value)
{
if (value)
{
/* The randomization procedure is specified in RFC 1771: 9.2.3.3 */
t->randomize = value / 4;
tm_start(t, value - t->randomize);
}
2000-03-31 01:39:48 +08:00
else
tm_stop(t);
}
2000-06-05 01:06:18 +08:00
/**
* bgp_close_conn - close a BGP connection
* @conn: connection to close
*
* This function takes a connection described by the &bgp_conn structure, closes
* its socket and frees all resources associated with it.
2000-06-05 01:06:18 +08:00
*/
2000-03-31 01:39:48 +08:00
void
bgp_close_conn(struct bgp_conn *conn)
{
// struct bgp_proto *p = conn->bgp;
2000-03-31 01:39:48 +08:00
DBG("BGP: Closing connection\n");
conn->packets_to_send = 0;
conn->channels_to_send = 0;
rfree(conn->connect_timer);
conn->connect_timer = NULL;
2000-03-31 01:39:48 +08:00
rfree(conn->keepalive_timer);
conn->keepalive_timer = NULL;
rfree(conn->hold_timer);
conn->hold_timer = NULL;
rfree(conn->tx_ev);
conn->tx_ev = NULL;
rfree(conn->sk);
conn->sk = NULL;
mb_free(conn->local_caps);
conn->local_caps = NULL;
mb_free(conn->remote_caps);
conn->remote_caps = NULL;
}
/**
* bgp_update_startup_delay - update a startup delay
* @p: BGP instance
*
* This function updates a startup delay that is used to postpone next BGP
* connect. It also handles disable_after_error and might stop BGP instance
* when error happened and disable_after_error is on.
*
* It should be called when BGP protocol error happened.
*/
void
bgp_update_startup_delay(struct bgp_proto *p)
{
struct bgp_config *cf = p->cf;
DBG("BGP: Updating startup delay\n");
if (p->last_proto_error && ((now - p->last_proto_error) >= (int) cf->error_amnesia_time))
p->startup_delay = 0;
p->last_proto_error = now;
if (cf->disable_after_error)
{
p->startup_delay = 0;
p->p.disabled = 1;
return;
}
if (!p->startup_delay)
p->startup_delay = cf->error_delay_time_min;
else
p->startup_delay = MIN(2 * p->startup_delay, cf->error_delay_time_max);
}
static void
bgp_graceful_close_conn(struct bgp_conn *conn, uint subcode)
{
switch (conn->state)
{
case BS_IDLE:
case BS_CLOSE:
return;
case BS_CONNECT:
case BS_ACTIVE:
bgp_conn_enter_idle_state(conn);
return;
case BS_OPENSENT:
case BS_OPENCONFIRM:
case BS_ESTABLISHED:
bgp_error(conn, 6, subcode, NULL, 0);
return;
default:
bug("bgp_graceful_close_conn: Unknown state %d", conn->state);
}
}
static void
bgp_down(struct bgp_proto *p)
{
if (p->start_state > BSS_PREPARE)
{
bgp_setup_auth(p, 0);
bgp_close(p);
}
BGP_TRACE(D_EVENTS, "Down");
proto_notify_state(&p->p, PS_DOWN);
}
static void
bgp_decision(void *vp)
{
struct bgp_proto *p = vp;
DBG("BGP: Decision start\n");
if ((p->p.proto_state == PS_START) &&
(p->outgoing_conn.state == BS_IDLE) &&
(p->incoming_conn.state != BS_OPENCONFIRM) &&
!p->cf->passive)
bgp_active(p);
if ((p->p.proto_state == PS_STOP) &&
(p->outgoing_conn.state == BS_IDLE) &&
(p->incoming_conn.state == BS_IDLE))
bgp_down(p);
}
void
bgp_stop(struct bgp_proto *p, uint subcode)
{
proto_notify_state(&p->p, PS_STOP);
bgp_graceful_close_conn(&p->outgoing_conn, subcode);
bgp_graceful_close_conn(&p->incoming_conn, subcode);
ev_schedule(p->event);
}
2010-01-03 19:17:52 +08:00
static inline void
bgp_conn_set_state(struct bgp_conn *conn, uint new_state)
2010-01-03 19:17:52 +08:00
{
if (conn->bgp->p.mrtdump & MD_STATES)
mrt_dump_bgp_state_change(conn, conn->state, new_state);
conn->state = new_state;
}
void
bgp_conn_enter_openconfirm_state(struct bgp_conn *conn)
{
/* Really, most of the work is done in bgp_rx_open(). */
bgp_conn_set_state(conn, BS_OPENCONFIRM);
}
static const struct bgp_af_caps dummy_af_caps = { };
void
bgp_conn_enter_established_state(struct bgp_conn *conn)
{
struct bgp_proto *p = conn->bgp;
struct bgp_caps *local = conn->local_caps;
struct bgp_caps *peer = conn->remote_caps;
struct bgp_channel *c;
BGP_TRACE(D_EVENTS, "BGP session established");
/* For multi-hop BGP sessions */
if (ipa_zero(p->source_addr))
p->source_addr = conn->sk->saddr;
conn->sk->fast_rx = 0;
p->conn = conn;
p->last_error_class = 0;
p->last_error_code = 0;
p->as4_session = conn->as4_session;
p->route_refresh = peer->route_refresh;
p->enhanced_refresh = local->enhanced_refresh && peer->enhanced_refresh;
/* Whether we may handle possible GR of peer (it has some AF GR-able) */
p->gr_ready = 0; /* Updated later */
/* Whether peer is ready to handle our GR recovery */
int peer_gr_ready = peer->gr_aware && !(peer->gr_flags & BGP_GRF_RESTART);
if (p->gr_active_num)
tm_stop(p->gr_timer);
/* Number of active channels */
int num = 0;
WALK_LIST(c, p->p.channels)
{
const struct bgp_af_caps *loc = bgp_find_af_caps(local, c->afi);
const struct bgp_af_caps *rem = bgp_find_af_caps(peer, c->afi);
/* Ignore AFIs that were not announced in multiprotocol capability */
if (!loc || !loc->ready)
loc = &dummy_af_caps;
if (!rem || !rem->ready)
rem = &dummy_af_caps;
int active = loc->ready && rem->ready;
c->c.disabled = !active;
c->c.reloadable = p->route_refresh;
c->index = active ? num++ : 0;
c->feed_state = BFS_NONE;
c->load_state = BFS_NONE;
/* Channels where peer may do GR */
c->gr_ready = active && local->gr_aware && rem->gr_able;
p->gr_ready = p->gr_ready || c->gr_ready;
/* Channels not able to recover gracefully */
if (p->p.gr_recovery && (!active || !peer_gr_ready))
channel_graceful_restart_unlock(&c->c);
/* Channels waiting for local convergence */
if (p->p.gr_recovery && loc->gr_able && peer_gr_ready)
c->c.gr_wait = 1;
/* Channels where peer is not able to recover gracefully */
if (c->gr_active && ! (c->gr_ready && (rem->gr_af_flags & BGP_GRF_FORWARDING)))
bgp_graceful_restart_done(c);
/* GR capability implies that neighbor will send End-of-RIB */
if (peer->gr_aware)
c->load_state = BFS_LOADING;
c->ext_next_hop = c->cf->ext_next_hop && (bgp_channel_is_ipv6(c) || rem->ext_next_hop);
c->add_path_rx = (loc->add_path & BGP_ADD_PATH_RX) && (rem->add_path & BGP_ADD_PATH_TX);
c->add_path_tx = (loc->add_path & BGP_ADD_PATH_TX) && (rem->add_path & BGP_ADD_PATH_RX);
2016-12-24 06:03:26 +08:00
/* Update RA mode */
if (c->add_path_tx)
c->c.ra_mode = RA_ANY;
2016-12-24 06:03:26 +08:00
else if (c->cf->secondary)
c->c.ra_mode = RA_ACCEPTED;
else
c->c.ra_mode = RA_OPTIMAL;
}
p->afi_map = mb_alloc(p->p.pool, num * sizeof(u32));
p->channel_map = mb_alloc(p->p.pool, num * sizeof(void *));
p->channel_count = num;
WALK_LIST(c, p->p.channels)
{
if (c->c.disabled)
continue;
p->afi_map[c->index] = c->afi;
p->channel_map[c->index] = c;
}
/* proto_notify_state() will likely call bgp_feed_begin(), setting c->feed_state */
2010-01-03 19:17:52 +08:00
bgp_conn_set_state(conn, BS_ESTABLISHED);
proto_notify_state(&p->p, PS_UP);
}
static void
bgp_conn_leave_established_state(struct bgp_proto *p)
{
BGP_TRACE(D_EVENTS, "BGP session closed");
p->conn = NULL;
2017-02-08 21:34:48 +08:00
// XXXX free these tables to avoid memory leak during graceful restart
// bgp_free_prefix_table(p);
// bgp_free_bucket_table(p);
if (p->p.proto_state == PS_UP)
bgp_stop(p, 0);
}
void
bgp_conn_enter_close_state(struct bgp_conn *conn)
{
struct bgp_proto *p = conn->bgp;
int os = conn->state;
2010-01-03 19:17:52 +08:00
bgp_conn_set_state(conn, BS_CLOSE);
tm_stop(conn->keepalive_timer);
conn->sk->rx_hook = NULL;
/* Timeout for CLOSE state, if we cannot send notification soon then we just hangup */
bgp_start_timer(conn->hold_timer, 10);
if (os == BS_ESTABLISHED)
bgp_conn_leave_established_state(p);
}
void
bgp_conn_enter_idle_state(struct bgp_conn *conn)
{
struct bgp_proto *p = conn->bgp;
int os = conn->state;
bgp_close_conn(conn);
2010-01-03 19:17:52 +08:00
bgp_conn_set_state(conn, BS_IDLE);
ev_schedule(p->event);
if (os == BS_ESTABLISHED)
bgp_conn_leave_established_state(p);
}
/**
* bgp_handle_graceful_restart - handle detected BGP graceful restart
* @p: BGP instance
*
* This function is called when a BGP graceful restart of the neighbor is
* detected (when the TCP connection fails or when a new TCP connection
* appears). The function activates processing of the restart - starts routing
* table refresh cycle and activates BGP restart timer. The protocol state goes
* back to %PS_START, but changing BGP state back to %BS_IDLE is left for the
* caller.
*/
void
bgp_handle_graceful_restart(struct bgp_proto *p)
{
ASSERT(p->conn && (p->conn->state == BS_ESTABLISHED) && p->gr_ready);
BGP_TRACE(D_EVENTS, "Neighbor graceful restart detected%s",
p->gr_active_num ? " - already pending" : "");
p->gr_active_num = 0;
struct bgp_channel *c;
WALK_LIST(c, p->p.channels)
{
if (c->gr_ready)
{
if (c->gr_active)
rt_refresh_end(c->c.table, &c->c);
c->gr_active = 1;
p->gr_active_num++;
rt_refresh_begin(c->c.table, &c->c);
}
else
{
/* Just flush the routes */
rt_refresh_begin(c->c.table, &c->c);
rt_refresh_end(c->c.table, &c->c);
}
}
proto_notify_state(&p->p, PS_START);
bgp_start_timer(p->gr_timer, p->conn->local_caps->gr_time);
}
/**
* bgp_graceful_restart_done - finish active BGP graceful restart
* @c: BGP channel
*
* This function is called when the active BGP graceful restart of the neighbor
* should be finished for channel @c - either successfully (the neighbor sends
* all paths and reports end-of-RIB for given AFI/SAFI on the new session) or
* unsuccessfully (the neighbor does not support BGP graceful restart on the new
* session). The function ends the routing table refresh cycle.
*/
void
bgp_graceful_restart_done(struct bgp_channel *c)
{
struct bgp_proto *p = (void *) c->c.proto;
ASSERT(c->gr_active);
c->gr_active = 0;
p->gr_active_num--;
if (!p->gr_active_num)
BGP_TRACE(D_EVENTS, "Neighbor graceful restart done");
rt_refresh_end(c->c.table, &c->c);
}
/**
* bgp_graceful_restart_timeout - timeout of graceful restart 'restart timer'
* @t: timer
*
* This function is a timeout hook for @gr_timer, implementing BGP restart time
* limit for reestablisment of the BGP session after the graceful restart. When
* fired, we just proceed with the usual protocol restart.
*/
static void
bgp_graceful_restart_timeout(timer *t)
{
struct bgp_proto *p = t->data;
BGP_TRACE(D_EVENTS, "Neighbor graceful restart timeout");
bgp_stop(p, 0);
}
/**
* bgp_refresh_begin - start incoming enhanced route refresh sequence
* @c: BGP channel
*
* This function is called when an incoming enhanced route refresh sequence is
* started by the neighbor, demarcated by the BoRR packet. The function updates
* the load state and starts the routing table refresh cycle. Note that graceful
* restart also uses routing table refresh cycle, but RFC 7313 and load states
* ensure that these two sequences do not overlap.
*/
void
bgp_refresh_begin(struct bgp_channel *c)
{
struct bgp_proto *p = (void *) c->c.proto;
if (c->load_state == BFS_LOADING)
{ log(L_WARN "%s: BEGIN-OF-RR received before END-OF-RIB, ignoring", p->p.name); return; }
c->load_state = BFS_REFRESHING;
rt_refresh_begin(c->c.table, &c->c);
}
/**
* bgp_refresh_end - finish incoming enhanced route refresh sequence
* @c: BGP channel
*
* This function is called when an incoming enhanced route refresh sequence is
* finished by the neighbor, demarcated by the EoRR packet. The function updates
* the load state and ends the routing table refresh cycle. Routes not received
* during the sequence are removed by the nest.
*/
void
bgp_refresh_end(struct bgp_channel *c)
{
struct bgp_proto *p = (void *) c->c.proto;
if (c->load_state != BFS_REFRESHING)
{ log(L_WARN "%s: END-OF-RR received without prior BEGIN-OF-RR, ignoring", p->p.name); return; }
c->load_state = BFS_NONE;
rt_refresh_end(c->c.table, &c->c);
}
static void
bgp_send_open(struct bgp_conn *conn)
{
DBG("BGP: Sending open\n");
conn->sk->rx_hook = bgp_rx;
2000-03-31 01:39:48 +08:00
conn->sk->tx_hook = bgp_tx;
tm_stop(conn->connect_timer);
bgp_schedule_packet(conn, NULL, PKT_OPEN);
2010-01-03 19:17:52 +08:00
bgp_conn_set_state(conn, BS_OPENSENT);
bgp_start_timer(conn->hold_timer, conn->bgp->cf->initial_hold_time);
}
static void
bgp_connected(sock *sk)
{
struct bgp_conn *conn = sk->data;
2000-05-03 00:07:41 +08:00
struct bgp_proto *p = conn->bgp;
2000-05-03 00:07:41 +08:00
BGP_TRACE(D_EVENTS, "Connected");
bgp_send_open(conn);
}
static void
bgp_connect_timeout(timer *t)
{
struct bgp_conn *conn = t->data;
2000-05-03 00:07:41 +08:00
struct bgp_proto *p = conn->bgp;
2000-05-03 00:07:41 +08:00
DBG("BGP: connect_timeout\n");
if (p->p.proto_state == PS_START)
{
bgp_close_conn(conn);
bgp_connect(p);
}
else
bgp_conn_enter_idle_state(conn);
}
static void
bgp_sock_err(sock *sk, int err)
{
struct bgp_conn *conn = sk->data;
2000-05-03 00:07:41 +08:00
struct bgp_proto *p = conn->bgp;
/*
* This error hook may be called either asynchronously from main
* loop, or synchronously from sk_send(). But sk_send() is called
* only from bgp_tx() and bgp_kick_tx(), which are both called
* asynchronously from main loop. Moreover, they end if err hook is
* called. Therefore, we could suppose that it is always called
* asynchronously.
*/
bgp_store_error(p, conn, BE_SOCKET, err);
if (err)
BGP_TRACE(D_EVENTS, "Connection lost (%M)", err);
else
BGP_TRACE(D_EVENTS, "Connection closed");
if ((conn->state == BS_ESTABLISHED) && p->gr_ready)
bgp_handle_graceful_restart(p);
bgp_conn_enter_idle_state(conn);
}
static void
bgp_hold_timeout(timer *t)
{
struct bgp_conn *conn = t->data;
struct bgp_proto *p = conn->bgp;
DBG("BGP: Hold timeout\n");
/* We are already closing the connection - just do hangup */
if (conn->state == BS_CLOSE)
{
BGP_TRACE(D_EVENTS, "Connection stalled");
bgp_conn_enter_idle_state(conn);
return;
}
/* If there is something in input queue, we are probably congested
and perhaps just not processed BGP packets in time. */
if (sk_rx_ready(conn->sk) > 0)
bgp_start_timer(conn->hold_timer, 10);
else
bgp_error(conn, 4, 0, NULL, 0);
}
static void
bgp_keepalive_timeout(timer *t)
{
struct bgp_conn *conn = t->data;
DBG("BGP: Keepalive timer\n");
bgp_schedule_packet(conn, NULL, PKT_KEEPALIVE);
/* Kick TX a bit faster */
if (ev_active(conn->tx_ev))
ev_run(conn->tx_ev);
}
static void
bgp_setup_conn(struct bgp_proto *p, struct bgp_conn *conn)
{
conn->sk = NULL;
conn->bgp = p;
2000-03-21 23:53:50 +08:00
conn->packets_to_send = 0;
conn->channels_to_send = 0;
conn->last_channel = 0;
conn->last_channel_count = 0;
conn->connect_timer = tm_new_set(p->p.pool, bgp_connect_timeout, conn, 0, 0);
conn->hold_timer = tm_new_set(p->p.pool, bgp_hold_timeout, conn, 0, 0);
conn->keepalive_timer = tm_new_set(p->p.pool, bgp_keepalive_timeout, conn, 0, 0);
conn->tx_ev = ev_new(p->p.pool);
conn->tx_ev->hook = bgp_kick_tx;
conn->tx_ev->data = conn;
}
static void
bgp_setup_sk(struct bgp_conn *conn, sock *s)
{
s->data = conn;
s->err_hook = bgp_sock_err;
s->fast_rx = 1;
conn->sk = s;
}
static void
bgp_active(struct bgp_proto *p)
{
int delay = MAX(1, p->cf->connect_delay_time);
struct bgp_conn *conn = &p->outgoing_conn;
BGP_TRACE(D_EVENTS, "Connect delayed by %d seconds", delay);
bgp_setup_conn(p, conn);
2010-01-03 19:17:52 +08:00
bgp_conn_set_state(conn, BS_ACTIVE);
bgp_start_timer(conn->connect_timer, delay);
}
2000-06-05 01:06:18 +08:00
/**
* bgp_connect - initiate an outgoing connection
* @p: BGP instance
*
* The bgp_connect() function creates a new &bgp_conn and initiates
* a TCP connection to the peer. The rest of connection setup is governed
* by the BGP state machine as described in the standard.
*/
static void
bgp_connect(struct bgp_proto *p) /* Enter Connect state and start establishing connection */
{
2000-03-31 01:39:48 +08:00
struct bgp_conn *conn = &p->outgoing_conn;
int hops = p->cf->multihop ? : 1;
DBG("BGP: Connecting\n");
sock *s = sk_new(p->p.pool);
s->type = SK_TCP_ACTIVE;
s->saddr = p->source_addr;
s->daddr = p->cf->remote_ip;
s->dport = p->cf->remote_port;
s->iface = p->neigh ? p->neigh->iface : NULL;
s->ttl = p->cf->ttl_security ? 255 : hops;
s->rbsize = p->cf->enable_extended_messages ? BGP_RX_BUFFER_EXT_SIZE : BGP_RX_BUFFER_SIZE;
s->tbsize = p->cf->enable_extended_messages ? BGP_TX_BUFFER_EXT_SIZE : BGP_TX_BUFFER_SIZE;
s->tos = IP_PREC_INTERNET_CONTROL;
s->password = p->cf->password;
s->tx_hook = bgp_connected;
BGP_TRACE(D_EVENTS, "Connecting to %I%J from local address %I%J", s->daddr, p->cf->iface,
2014-10-24 17:11:43 +08:00
s->saddr, ipa_is_link_local(s->saddr) ? s->iface : NULL);
bgp_setup_conn(p, conn);
bgp_setup_sk(conn, s);
2010-01-03 19:17:52 +08:00
bgp_conn_set_state(conn, BS_CONNECT);
if (sk_open(s) < 0)
2014-05-18 17:42:26 +08:00
goto err;
/* Set minimal receive TTL if needed */
if (p->cf->ttl_security)
if (sk_set_min_ttl(s, 256 - hops) < 0)
2014-05-18 17:42:26 +08:00
goto err;
DBG("BGP: Waiting for connect success\n");
bgp_start_timer(conn->connect_timer, p->cf->connect_retry_time);
2014-05-18 17:42:26 +08:00
return;
err:
2014-05-18 17:42:26 +08:00
sk_log_error(s, p->p.name);
bgp_sock_err(s, 0);
return;
}
/**
* bgp_find_proto - find existing proto for incoming connection
* @sk: TCP socket
*
*/
static struct bgp_proto *
bgp_find_proto(sock *sk)
{
struct bgp_proto *p;
WALK_LIST(p, proto_list)
if ((p->p.proto == &proto_bgp) &&
ipa_equal(p->cf->remote_ip, sk->daddr) &&
2017-04-28 17:19:12 +08:00
(!p->cf->iface || (p->cf->iface == sk->iface)) &&
(ipa_zero(p->cf->local_ip) || ipa_equal(p->cf->local_ip, sk->saddr)) &&
(p->cf->local_port == sk->sport))
return p;
return NULL;
}
2000-06-05 01:06:18 +08:00
/**
* bgp_incoming_connection - handle an incoming connection
* @sk: TCP socket
* @dummy: unused
*
* This function serves as a socket hook for accepting of new BGP
* connections. It searches a BGP instance corresponding to the peer
* which has connected and if such an instance exists, it creates a
* &bgp_conn structure, attaches it to the instance and either sends
* an Open message or (if there already is an active connection) it
* closes the new connection by sending a Notification message.
*/
static int
bgp_incoming_connection(sock *sk, uint dummy UNUSED)
{
struct bgp_proto *p;
int acc, hops;
DBG("BGP: Incoming connection from %I port %d\n", sk->daddr, sk->dport);
p = bgp_find_proto(sk);
if (!p)
{
log(L_WARN "BGP: Unexpected connect from unknown address %I%J (port %d)",
sk->daddr, ipa_is_link_local(sk->daddr) ? sk->iface : NULL, sk->dport);
rfree(sk);
return 0;
}
/*
* BIRD should keep multiple incoming connections in OpenSent state (for
* details RFC 4271 8.2.1 par 3), but it keeps just one. Duplicate incoming
* connections are rejected istead. The exception is the case where an
* incoming connection triggers a graceful restart.
*/
acc = (p->p.proto_state == PS_START || p->p.proto_state == PS_UP) &&
(p->start_state >= BSS_CONNECT) && (!p->incoming_conn.sk);
if (p->conn && (p->conn->state == BS_ESTABLISHED) && p->gr_ready)
{
bgp_store_error(p, NULL, BE_MISC, BEM_GRACEFUL_RESTART);
bgp_handle_graceful_restart(p);
bgp_conn_enter_idle_state(p->conn);
acc = 1;
/* There might be separate incoming connection in OpenSent state */
if (p->incoming_conn.state > BS_ACTIVE)
bgp_close_conn(&p->incoming_conn);
}
BGP_TRACE(D_EVENTS, "Incoming connection from %I%J (port %d) %s",
sk->daddr, ipa_is_link_local(sk->daddr) ? sk->iface : NULL,
sk->dport, acc ? "accepted" : "rejected");
if (!acc)
{
rfree(sk);
return 0;
}
hops = p->cf->multihop ? : 1;
if (sk_set_ttl(sk, p->cf->ttl_security ? 255 : hops) < 0)
goto err;
if (p->cf->ttl_security)
if (sk_set_min_ttl(sk, 256 - hops) < 0)
goto err;
if (p->cf->enable_extended_messages)
{
sk->rbsize = BGP_RX_BUFFER_EXT_SIZE;
sk->tbsize = BGP_TX_BUFFER_EXT_SIZE;
sk_reallocate(sk);
}
bgp_setup_conn(p, &p->incoming_conn);
bgp_setup_sk(&p->incoming_conn, sk);
bgp_send_open(&p->incoming_conn);
return 0;
err:
sk_log_error(sk, p->p.name);
log(L_ERR "%s: Incoming connection aborted", p->p.name);
rfree(sk);
return 0;
}
static void
bgp_listen_sock_err(sock *sk UNUSED, int err)
{
if (err == ECONNABORTED)
log(L_WARN "BGP: Incoming connection aborted");
else
log(L_ERR "BGP: Error on listening socket: %M", err);
}
static void
bgp_start_neighbor(struct bgp_proto *p)
{
/* Called only for single-hop BGP sessions */
if (ipa_zero(p->source_addr))
p->source_addr = p->neigh->ifa->ip;
if (ipa_is_link_local(p->source_addr))
p->link_addr = p->source_addr;
else
{
/* Find some link-local address for given iface */
struct ifa *a;
WALK_LIST(a, p->neigh->iface->addrs)
if (a->scope == SCOPE_LINK)
{
p->link_addr = a->ip;
break;
}
DBG("%s: Selected link-local address %I\n", p->p.name, p->link_addr);
}
bgp_initiate(p);
}
static void
bgp_neigh_notify(neighbor *n)
{
struct bgp_proto *p = (struct bgp_proto *) n->proto;
int ps = p->p.proto_state;
if (n != p->neigh)
return;
if ((ps == PS_DOWN) || (ps == PS_STOP))
return;
int prepare = (ps == PS_START) && (p->start_state == BSS_PREPARE);
if (n->scope <= 0)
{
if (!prepare)
{
BGP_TRACE(D_EVENTS, "Neighbor lost");
bgp_store_error(p, NULL, BE_MISC, BEM_NEIGHBOR_LOST);
/* Perhaps also run bgp_update_startup_delay(p)? */
bgp_stop(p, 0);
}
}
else if (p->cf->check_link && !(n->iface->flags & IF_LINK_UP))
{
if (!prepare)
{
BGP_TRACE(D_EVENTS, "Link down");
bgp_store_error(p, NULL, BE_MISC, BEM_LINK_DOWN);
if (ps == PS_UP)
bgp_update_startup_delay(p);
bgp_stop(p, 0);
}
}
else
{
if (prepare)
{
BGP_TRACE(D_EVENTS, "Neighbor ready");
bgp_start_neighbor(p);
}
}
}
static void
bgp_bfd_notify(struct bfd_request *req)
{
struct bgp_proto *p = req->data;
int ps = p->p.proto_state;
if (req->down && ((ps == PS_START) || (ps == PS_UP)))
{
BGP_TRACE(D_EVENTS, "BFD session down");
bgp_store_error(p, NULL, BE_MISC, BEM_BFD_DOWN);
if (ps == PS_UP)
bgp_update_startup_delay(p);
bgp_stop(p, 0);
}
}
static void
bgp_update_bfd(struct bgp_proto *p, int use_bfd)
{
if (use_bfd && !p->bfd_req)
p->bfd_req = bfd_request_session(p->p.pool, p->cf->remote_ip, p->source_addr,
p->cf->multihop ? NULL : p->neigh->iface,
bgp_bfd_notify, p);
if (!use_bfd && p->bfd_req)
{
rfree(p->bfd_req);
p->bfd_req = NULL;
}
}
static void
bgp_reload_routes(struct channel *C)
2009-11-27 03:47:59 +08:00
{
struct bgp_proto *p = (void *) C->proto;
struct bgp_channel *c = (void *) C;
2009-11-27 03:47:59 +08:00
ASSERT(p->conn && p->route_refresh);
bgp_schedule_packet(p->conn, c, PKT_ROUTE_REFRESH);
2009-11-27 03:47:59 +08:00
}
static void
bgp_feed_begin(struct channel *C, int initial)
{
struct bgp_proto *p = (void *) C->proto;
struct bgp_channel *c = (void *) C;
/* This should not happen */
if (!p->conn)
return;
if (initial && p->cf->gr_mode)
c->feed_state = BFS_LOADING;
/* It is refeed and both sides support enhanced route refresh */
if (!initial && p->enhanced_refresh)
{
/* BoRR must not be sent before End-of-RIB */
if (c->feed_state == BFS_LOADING || c->feed_state == BFS_LOADED)
return;
c->feed_state = BFS_REFRESHING;
bgp_schedule_packet(p->conn, c, PKT_BEGIN_REFRESH);
}
}
static void
bgp_feed_end(struct channel *C)
{
struct bgp_proto *p = (void *) C->proto;
struct bgp_channel *c = (void *) C;
/* This should not happen */
if (!p->conn)
return;
/* Non-demarcated feed ended, nothing to do */
if (c->feed_state == BFS_NONE)
return;
/* Schedule End-of-RIB packet */
if (c->feed_state == BFS_LOADING)
c->feed_state = BFS_LOADED;
/* Schedule EoRR packet */
if (c->feed_state == BFS_REFRESHING)
c->feed_state = BFS_REFRESHED;
/* Kick TX hook */
bgp_schedule_packet(p->conn, c, PKT_UPDATE);
}
static void
bgp_start_locked(struct object_lock *lock)
{
struct bgp_proto *p = lock->data;
struct bgp_config *cf = p->cf;
if (p->p.proto_state != PS_START)
{
DBG("BGP: Got lock in different state %d\n", p->p.proto_state);
return;
}
DBG("BGP: Got lock\n");
if (cf->multihop)
{
/* Multi-hop sessions do not use neighbor entries */
bgp_initiate(p);
return;
}
neighbor *n = neigh_find2(&p->p, &cf->remote_ip, cf->iface, NEF_STICKY);
if (!n)
{
log(L_ERR "%s: Invalid remote address %I%J", p->p.name, cf->remote_ip, cf->iface);
/* As we do not start yet, we can just disable protocol */
p->p.disabled = 1;
bgp_store_error(p, NULL, BE_MISC, BEM_INVALID_NEXT_HOP);
proto_notify_state(&p->p, PS_DOWN);
return;
}
p->neigh = n;
if (n->scope <= 0)
BGP_TRACE(D_EVENTS, "Waiting for %I%J to become my neighbor", cf->remote_ip, cf->iface);
else if (p->cf->check_link && !(n->iface->flags & IF_LINK_UP))
BGP_TRACE(D_EVENTS, "Waiting for link on %s", n->iface->name);
else
bgp_start_neighbor(p);
}
2000-03-20 06:09:07 +08:00
static int
bgp_start(struct proto *P)
{
struct bgp_proto *p = (struct bgp_proto *) P;
struct object_lock *lock;
2000-03-31 01:39:48 +08:00
DBG("BGP: Startup.\n");
p->start_state = BSS_PREPARE;
2000-03-31 01:39:48 +08:00
p->outgoing_conn.state = BS_IDLE;
p->incoming_conn.state = BS_IDLE;
p->neigh = NULL;
p->bfd_req = NULL;
p->gr_ready = 0;
p->gr_active_num = 0;
p->event = ev_new(p->p.pool);
p->event->hook = bgp_decision;
p->event->data = p;
p->startup_timer = tm_new(p->p.pool);
p->startup_timer->hook = bgp_startup_timeout;
p->startup_timer->data = p;
p->gr_timer = tm_new(p->p.pool);
p->gr_timer->hook = bgp_graceful_restart_timeout;
p->gr_timer->data = p;
p->local_id = proto_get_router_id(P->cf);
if (p->rr_client)
p->rr_cluster_id = p->cf->rr_cluster_id ? p->cf->rr_cluster_id : p->local_id;
p->remote_id = 0;
p->source_addr = p->cf->local_ip;
p->link_addr = IPA_NONE;
/* XXXX */
if (p->p.gr_recovery && p->cf->gr_mode)
{
struct bgp_channel *c;
WALK_LIST(c, p->p.channels)
channel_graceful_restart_lock(&c->c);
}
/*
* Before attempting to create the connection, we need to lock the port,
* so that we are the only instance attempting to talk with that neighbor.
*/
lock = p->lock = olock_new(P->pool);
lock->addr = p->cf->remote_ip;
lock->port = p->cf->remote_port;
lock->iface = p->cf->iface;
lock->type = OBJLOCK_TCP;
lock->hook = bgp_start_locked;
lock->data = p;
olock_acquire(lock);
return PS_START;
2000-03-20 06:09:07 +08:00
}
extern int proto_restart;
2000-03-20 06:09:07 +08:00
static int
bgp_shutdown(struct proto *P)
{
struct bgp_proto *p = (struct bgp_proto *) P;
uint subcode = 0;
2000-05-03 00:07:41 +08:00
BGP_TRACE(D_EVENTS, "Shutdown requested");
switch (P->down_code)
{
case PDC_CF_REMOVE:
case PDC_CF_DISABLE:
subcode = 3; // Errcode 6, 3 - peer de-configured
break;
case PDC_CF_RESTART:
subcode = 6; // Errcode 6, 6 - other configuration change
break;
case PDC_CMD_DISABLE:
case PDC_CMD_SHUTDOWN:
subcode = 2; // Errcode 6, 2 - administrative shutdown
break;
case PDC_CMD_RESTART:
subcode = 4; // Errcode 6, 4 - administrative reset
break;
case PDC_RX_LIMIT_HIT:
case PDC_IN_LIMIT_HIT:
subcode = 1; // Errcode 6, 1 - max number of prefixes reached
/* log message for compatibility */
log(L_WARN "%s: Route limit exceeded, shutting down", p->p.name);
goto limit;
case PDC_OUT_LIMIT_HIT:
subcode = proto_restart ? 4 : 2; // Administrative reset or shutdown
limit:
bgp_store_error(p, NULL, BE_AUTO_DOWN, BEA_ROUTE_LIMIT_EXCEEDED);
if (proto_restart)
bgp_update_startup_delay(p);
else
p->startup_delay = 0;
goto done;
}
bgp_store_error(p, NULL, BE_MAN_DOWN, 0);
p->startup_delay = 0;
done:
bgp_stop(p, subcode);
return p->p.proto_state;
2000-03-20 06:09:07 +08:00
}
static struct proto *
bgp_init(struct proto_config *CF)
{
struct proto *P = proto_new(CF);
struct bgp_proto *p = (struct bgp_proto *) P;
struct bgp_config *cf = (struct bgp_config *) CF;
P->rt_notify = bgp_rt_notify;
P->import_control = bgp_import_control;
P->neigh_notify = bgp_neigh_notify;
2009-11-27 03:47:59 +08:00
P->reload_routes = bgp_reload_routes;
P->feed_begin = bgp_feed_begin;
P->feed_end = bgp_feed_end;
P->rte_better = bgp_rte_better;
P->rte_mergable = bgp_rte_mergable;
P->rte_recalculate = cf->deterministic_med ? bgp_rte_recalculate : NULL;
p->cf = cf;
p->local_as = cf->local_as;
p->remote_as = cf->remote_as;
p->public_as = cf->local_as;
p->is_internal = (cf->local_as == cf->remote_as);
p->is_interior = p->is_internal || cf->confederation_member;
p->rs_client = cf->rs_client;
p->rr_client = cf->rr_client;
/* Confederation ID is used for truly external peers */
if (cf->confederation && !p->is_interior)
p->public_as = cf->confederation;
/* Add all channels */
struct bgp_channel_config *cc;
WALK_LIST(cc, CF->channels)
proto_add_channel(P, &cc->c);
return P;
}
static void
bgp_channel_init(struct channel *C, struct channel_config *CF)
{
struct bgp_channel *c = (void *) C;
struct bgp_channel_config *cf = (void *) CF;
c->cf = cf;
c->afi = cf->afi;
c->desc = cf->desc;
if (cf->igp_table_ip4)
c->igp_table_ip4 = cf->igp_table_ip4->table;
if (cf->igp_table_ip6)
c->igp_table_ip6 = cf->igp_table_ip6->table;
}
static int
bgp_channel_start(struct channel *C)
{
struct bgp_proto *p = (void *) C->proto;
struct bgp_channel *c = (void *) C;
ip_addr src = p->source_addr;
if (c->igp_table_ip4)
rt_lock_table(c->igp_table_ip4);
if (c->igp_table_ip6)
rt_lock_table(c->igp_table_ip6);
c->pool = p->p.pool; // XXXX
bgp_init_bucket_table(c);
bgp_init_prefix_table(c);
c->next_hop_addr = c->cf->next_hop_addr;
c->link_addr = IPA_NONE;
c->packets_to_send = 0;
/* Try to use source address as next hop address */
if (ipa_zero(c->next_hop_addr))
{
if (bgp_channel_is_ipv4(c) && (ipa_is_ip4(src) || c->ext_next_hop))
c->next_hop_addr = src;
if (bgp_channel_is_ipv6(c) && (ipa_is_ip6(src) || c->ext_next_hop))
c->next_hop_addr = src;
}
/* Exit if no feasible next hop address is found */
if (ipa_zero(c->next_hop_addr))
{
log(L_WARN "%s: Missing next hop address", p->p.name);
return 0;
}
/* Set link-local address for IPv6 single-hop BGP */
if (ipa_is_ip6(c->next_hop_addr) && p->neigh)
{
c->link_addr = p->link_addr;
if (ipa_zero(c->link_addr))
log(L_WARN "%s: Missing link-local address", p->p.name);
}
/* Link local address is already in c->link_addr */
if (ipa_is_link_local(c->next_hop_addr))
c->next_hop_addr = IPA_NONE;
return 0; /* XXXX: Currently undefined */
}
static void
bgp_channel_shutdown(struct channel *C)
{
struct bgp_channel *c = (void *) C;
/* XXXX: cleanup bucket and prefix tables */
c->next_hop_addr = IPA_NONE;
c->link_addr = IPA_NONE;
}
static void
bgp_channel_cleanup(struct channel *C)
{
struct bgp_channel *c = (void *) C;
if (c->igp_table_ip4)
rt_unlock_table(c->igp_table_ip4);
if (c->igp_table_ip6)
rt_unlock_table(c->igp_table_ip6);
}
static inline struct bgp_channel_config *
bgp_find_channel_config(struct bgp_config *cf, u32 afi)
{
struct bgp_channel_config *cc;
WALK_LIST(cc, cf->c.channels)
if (cc->afi == afi)
return cc;
return NULL;
}
struct rtable_config *
bgp_default_igp_table(struct bgp_config *cf, struct bgp_channel_config *cc, u32 type)
{
struct bgp_channel_config *cc2;
struct rtable_config *tab;
/* First, try table connected by the channel */
if (cc->c.table->addr_type == type)
return cc->c.table;
/* Find paired channel with the same SAFI but the other AFI */
u32 afi2 = cc->afi ^ 0x30000;
cc2 = bgp_find_channel_config(cf, afi2);
/* Second, try IGP table configured in the paired channel */
if (cc2 && (tab = (type == NET_IP4) ? cc2->igp_table_ip4 : cc2->igp_table_ip6))
return tab;
/* Third, try table connected by the paired channel */
if (cc2 && (cc2->c.table->addr_type == type))
return cc2->c.table;
/* Last, try default table of given type */
if (tab = cf->c.global->def_tables[type])
return tab;
cf_error("Undefined IGP table");
}
void
bgp_postconfig(struct proto_config *CF)
{
struct bgp_config *cf = (void *) CF;
int internal = (cf->local_as == cf->remote_as);
/* Do not check templates at all */
if (cf->c.class == SYM_TEMPLATE)
return;
/* EBGP direct by default, IBGP multihop by default */
if (cf->multihop < 0)
cf->multihop = internal ? 64 : 0;
if (!cf->local_as)
cf_error("Local AS number must be set");
if (ipa_zero(cf->remote_ip))
cf_error("Neighbor must be configured");
if (!cf->remote_as)
cf_error("Remote AS number must be set");
2017-04-28 17:19:12 +08:00
if (ipa_is_link_local(cf->remote_ip) && !cf->iface)
cf_error("Link-local neighbor address requires specified interface");
if (!(cf->capabilities && cf->enable_as4) && (cf->remote_as > 0xFFFF))
cf_error("Neighbor AS number out of range (AS4 not available)");
if (!internal && cf->rr_client)
cf_error("Only internal neighbor can be RR client");
if (internal && cf->rs_client)
cf_error("Only external neighbor can be RS client");
if (!cf->confederation && cf->confederation_member)
cf_error("Confederation ID must be set for member sessions");
if (cf->multihop && (ipa_is_link_local(cf->local_ip) ||
ipa_is_link_local(cf->remote_ip)))
cf_error("Multihop BGP cannot be used with link-local addresses");
2017-04-28 17:19:12 +08:00
if (cf->multihop && cf->iface)
cf_error("Multihop BGP cannot be bound to interface");
if (cf->multihop && cf->check_link)
cf_error("Multihop BGP cannot depend on link state");
if (cf->multihop && cf->bfd && ipa_zero(cf->local_ip))
cf_error("Multihop BGP with BFD requires specified local address");
struct bgp_channel_config *cc;
WALK_LIST(cc, CF->channels)
{
/* Disable after error incompatible with restart limit action */
if ((cc->c.in_limit.action == PLA_RESTART) && cf->disable_after_error)
cc->c.in_limit.action = PLA_DISABLE;
/* Different default based on rs_client */
if (!cc->missing_lladdr)
cc->missing_lladdr = cf->rs_client ? MLL_IGNORE : MLL_SELF;
/* Different default for gw_mode */
if (!cc->gw_mode)
cc->gw_mode = cf->multihop ? GW_RECURSIVE : GW_DIRECT;
/* Default based on proto config */
if (cc->gr_able == 0xff)
cc->gr_able = (cf->gr_mode == BGP_GR_ABLE);
2012-07-05 03:31:03 +08:00
2017-03-29 21:31:04 +08:00
/* Default values of IGP tables */
if ((cc->gw_mode == GW_RECURSIVE) && !cc->desc->no_igp)
{
if (!cc->igp_table_ip4 && (bgp_cc_is_ipv4(cc) || cc->ext_next_hop))
cc->igp_table_ip4 = bgp_default_igp_table(cf, cc, NET_IP4);
if (!cc->igp_table_ip6 && (bgp_cc_is_ipv6(cc) || cc->ext_next_hop))
cc->igp_table_ip6 = bgp_default_igp_table(cf, cc, NET_IP6);
2017-03-29 21:31:04 +08:00
if (cc->igp_table_ip4 && bgp_cc_is_ipv6(cc) && !cc->ext_next_hop)
cf_error("Mismatched IGP table type");
if (cc->igp_table_ip6 && bgp_cc_is_ipv4(cc) && !cc->ext_next_hop)
cf_error("Mismatched IGP table type");
}
if (cf->multihop && (cc->gw_mode == GW_DIRECT))
cf_error("Multihop BGP cannot use direct gateway mode");
2012-07-05 03:31:03 +08:00
if ((cc->gw_mode == GW_RECURSIVE) && cc->c.table->sorted)
cf_error("BGP in recursive mode prohibits sorted table");
if (cf->deterministic_med && cc->c.table->sorted)
cf_error("BGP with deterministic MED prohibits sorted table");
if (cc->secondary && !cc->c.table->sorted)
cf_error("BGP with secondary option requires sorted table");
}
}
static int
bgp_reconfigure(struct proto *P, struct proto_config *CF)
{
struct bgp_proto *p = (void *) P;
struct bgp_config *new = (void *) CF;
struct bgp_config *old = p->cf;
if (proto_get_router_id(CF) != p->local_id)
return 0;
int same = !memcmp(((byte *) old) + sizeof(struct proto_config),
((byte *) new) + sizeof(struct proto_config),
// password item is last and must be checked separately
OFFSETOF(struct bgp_config, password) - sizeof(struct proto_config))
&& ((!old->password && !new->password)
|| (old->password && new->password && !strcmp(old->password, new->password)));
/* FIXME: Move channel reconfiguration to generic protocol code ? */
struct channel *C, *C2;
struct bgp_channel_config *cc;
WALK_LIST(C, p->p.channels)
C->stale = 1;
WALK_LIST(cc, new->c.channels)
{
C = (struct channel *) bgp_find_channel(p, cc->afi);
same = proto_configure_channel(P, &C, &cc->c) && same;
C->stale = 0;
}
WALK_LIST_DELSAFE(C, C2, p->p.channels)
if (C->stale)
same = proto_configure_channel(P, &C, NULL) && same;
if (same && (p->start_state > BSS_PREPARE))
bgp_update_bfd(p, new->bfd);
/* We should update our copy of configuration ptr as old configuration will be freed */
if (same)
p->cf = new;
return same;
}
#define IGP_TABLE(cf, sym) ((cf)->igp_table_##sym ? (cf)->igp_table_##sym ->table : NULL )
static int
bgp_channel_reconfigure(struct channel *C, struct channel_config *CC)
{
struct bgp_channel *c = (void *) C;
struct bgp_channel_config *new = (void *) CC;
struct bgp_channel_config *old = c->cf;
if (memcmp(((byte *) old) + sizeof(struct channel_config),
((byte *) new) + sizeof(struct channel_config),
/* Remaining items must be checked separately */
OFFSETOF(struct bgp_channel_config, rest) - sizeof(struct channel_config)))
return 0;
/* Check change in IGP tables */
if ((IGP_TABLE(old, ip4) != IGP_TABLE(new, ip4)) ||
(IGP_TABLE(old, ip6) != IGP_TABLE(new, ip6)))
return 0;
c->cf = new;
return 1;
}
static void
bgp_copy_config(struct proto_config *dest UNUSED, struct proto_config *src UNUSED)
{
/* Just a shallow copy */
}
2000-06-05 01:06:18 +08:00
/**
* bgp_error - report a protocol error
* @c: connection
* @code: error code (according to the RFC)
2000-06-07 20:29:08 +08:00
* @subcode: error sub-code
2000-06-05 01:06:18 +08:00
* @data: data to be passed in the Notification message
* @len: length of the data
*
* bgp_error() sends a notification packet to tell the other side that a protocol
2000-06-07 20:29:08 +08:00
* error has occurred (including the data considered erroneous if possible) and
2000-06-05 01:06:18 +08:00
* closes the connection.
*/
void
bgp_error(struct bgp_conn *c, uint code, uint subcode, byte *data, int len)
{
struct bgp_proto *p = c->bgp;
if (c->state == BS_CLOSE)
return;
bgp_log_error(p, BE_BGP_TX, "Error", code, subcode, data, ABS(len));
bgp_store_error(p, c, BE_BGP_TX, (code << 16) | subcode);
bgp_conn_enter_close_state(c);
c->notify_code = code;
c->notify_subcode = subcode;
c->notify_data = data;
c->notify_size = (len > 0) ? len : 0;
bgp_schedule_packet(c, NULL, PKT_NOTIFICATION);
if (code != 6)
{
bgp_update_startup_delay(p);
bgp_stop(p, 0);
}
}
/**
* bgp_store_error - store last error for status report
* @p: BGP instance
* @c: connection
* @class: error class (BE_xxx constants)
* @code: error code (class specific)
*
* bgp_store_error() decides whether given error is interesting enough
* and store that error to last_error variables of @p
*/
void
bgp_store_error(struct bgp_proto *p, struct bgp_conn *c, u8 class, u32 code)
{
/* During PS_UP, we ignore errors on secondary connection */
if ((p->p.proto_state == PS_UP) && c && (c != p->conn))
return;
/* During PS_STOP, we ignore any errors, as we want to report
* the error that caused transition to PS_STOP
*/
if (p->p.proto_state == PS_STOP)
return;
p->last_error_class = class;
p->last_error_code = code;
}
static char *bgp_state_names[] = { "Idle", "Connect", "Active", "OpenSent", "OpenConfirm", "Established", "Close" };
2009-06-04 19:31:09 +08:00
static char *bgp_err_classes[] = { "", "Error: ", "Socket: ", "Received: ", "BGP Error: ", "Automatic shutdown: ", ""};
static char *bgp_misc_errors[] = { "", "Neighbor lost", "Invalid next hop", "Kernel MD5 auth failed", "No listening socket", "Link down", "BFD session down", "Graceful restart"};
2009-06-04 19:31:09 +08:00
static char *bgp_auto_errors[] = { "", "Route limit exceeded"};
static const char *
bgp_last_errmsg(struct bgp_proto *p)
2000-03-31 02:44:23 +08:00
{
switch (p->last_error_class)
{
case BE_MISC:
return bgp_misc_errors[p->last_error_code];
case BE_SOCKET:
return (p->last_error_code == 0) ? "Connection closed" : strerror(p->last_error_code);
case BE_BGP_RX:
case BE_BGP_TX:
return bgp_error_dsc(p->last_error_code >> 16, p->last_error_code & 0xFF);
case BE_AUTO_DOWN:
return bgp_auto_errors[p->last_error_code];
default:
return "";
}
}
static const char *
bgp_state_dsc(struct bgp_proto *p)
{
if (p->p.proto_state == PS_DOWN)
return "Down";
int state = MAX(p->incoming_conn.state, p->outgoing_conn.state);
if ((state == BS_IDLE) && (p->start_state >= BSS_CONNECT) && p->cf->passive)
return "Passive";
return bgp_state_names[state];
}
static void
bgp_get_status(struct proto *P, byte *buf)
{
struct bgp_proto *p = (struct bgp_proto *) P;
const char *err1 = bgp_err_classes[p->last_error_class];
const char *err2 = bgp_last_errmsg(p);
if (P->proto_state == PS_DOWN)
bsprintf(buf, "%s%s", err1, err2);
else
bsprintf(buf, "%-14s%s%s", bgp_state_dsc(p), err1, err2);
}
static void
bgp_show_afis(int code, char *s, u32 *afis, uint count)
{
buffer b;
LOG_BUFFER_INIT(b);
buffer_puts(&b, s);
for (u32 *af = afis; af < (afis + count); af++)
{
const struct bgp_af_desc *desc = bgp_get_af_desc(*af);
if (desc)
buffer_print(&b, " %s", desc->name);
else
buffer_print(&b, " <%u/%u>", BGP_AFI(*af), BGP_SAFI(*af));
}
if (b.pos == b.end)
strcpy(b.end - 32, " ... <too long>");
cli_msg(code, b.start);
}
static void
bgp_show_capabilities(struct bgp_proto *p UNUSED, struct bgp_caps *caps)
{
struct bgp_af_caps *ac;
uint any_mp_bgp = 0;
uint any_gr_able = 0;
uint any_add_path = 0;
uint any_ext_next_hop = 0;
u32 *afl1 = alloca(caps->af_count * sizeof(u32));
u32 *afl2 = alloca(caps->af_count * sizeof(u32));
uint afn1, afn2;
WALK_AF_CAPS(caps, ac)
{
any_mp_bgp |= ac->ready;
any_gr_able |= ac->gr_able;
any_add_path |= ac->add_path;
any_ext_next_hop |= ac->ext_next_hop;
}
if (any_mp_bgp)
{
cli_msg(-1006, " Multiprotocol");
afn1 = 0;
WALK_AF_CAPS(caps, ac)
if (ac->ready)
afl1[afn1++] = ac->afi;
bgp_show_afis(-1006, " AF announced:", afl1, afn1);
}
if (caps->route_refresh)
cli_msg(-1006, " Route refresh");
if (any_ext_next_hop)
{
cli_msg(-1006, " Extended next hop");
afn1 = 0;
WALK_AF_CAPS(caps, ac)
if (ac->ext_next_hop)
afl1[afn1++] = ac->afi;
bgp_show_afis(-1006, " IPv6 nexthop:", afl1, afn1);
}
if (caps->ext_messages)
cli_msg(-1006, " Extended message");
if (caps->gr_aware)
cli_msg(-1006, " Graceful restart");
if (any_gr_able)
{
/* Continues from gr_aware */
cli_msg(-1006, " Restart time: %u", caps->gr_time);
if (caps->gr_flags & BGP_GRF_RESTART)
cli_msg(-1006, " Restart recovery");
afn1 = afn2 = 0;
WALK_AF_CAPS(caps, ac)
{
if (ac->gr_able)
afl1[afn1++] = ac->afi;
if (ac->gr_af_flags & BGP_GRF_FORWARDING)
afl2[afn2++] = ac->afi;
}
bgp_show_afis(-1006, " AF supported:", afl1, afn1);
bgp_show_afis(-1006, " AF preserved:", afl2, afn2);
}
if (caps->as4_support)
cli_msg(-1006, " 4-octet AS numbers");
if (any_add_path)
{
cli_msg(-1006, " ADD-PATH");
afn1 = afn2 = 0;
WALK_AF_CAPS(caps, ac)
{
if (ac->add_path & BGP_ADD_PATH_RX)
afl1[afn1++] = ac->afi;
if (ac->add_path & BGP_ADD_PATH_TX)
afl2[afn2++] = ac->afi;
}
bgp_show_afis(-1006, " RX:", afl1, afn1);
bgp_show_afis(-1006, " TX:", afl2, afn2);
}
if (caps->enhanced_refresh)
cli_msg(-1006, " Enhanced refresh");
}
static void
bgp_show_proto_info(struct proto *P)
{
struct bgp_proto *p = (struct bgp_proto *) P;
cli_msg(-1006, " BGP state: %s", bgp_state_dsc(p));
cli_msg(-1006, " Neighbor address: %I%J", p->cf->remote_ip, p->cf->iface);
cli_msg(-1006, " Neighbor AS: %u", p->remote_as);
if (p->gr_active_num)
cli_msg(-1006, " Neighbor graceful restart active");
if (P->proto_state == PS_START)
{
struct bgp_conn *oc = &p->outgoing_conn;
if ((p->start_state < BSS_CONNECT) &&
(p->startup_timer->expires))
cli_msg(-1006, " Error wait: %d/%d",
p->startup_timer->expires - now, p->startup_delay);
if ((oc->state == BS_ACTIVE) &&
(oc->connect_timer->expires))
cli_msg(-1006, " Connect delay: %d/%d",
oc->connect_timer->expires - now, p->cf->connect_delay_time);
if (p->gr_active_num && p->gr_timer->expires)
cli_msg(-1006, " Restart timer: %d/-", p->gr_timer->expires - now);
}
else if (P->proto_state == PS_UP)
{
cli_msg(-1006, " Neighbor ID: %R", p->remote_id);
cli_msg(-1006, " Local capabilities");
bgp_show_capabilities(p, p->conn->local_caps);
cli_msg(-1006, " Neighbor capabilities");
bgp_show_capabilities(p, p->conn->remote_caps);
/* XXXX
cli_msg(-1006, " Session: %s%s%s%s%s%s%s%s",
p->is_internal ? "internal" : "external",
p->cf->multihop ? " multihop" : "",
p->rr_client ? " route-reflector" : "",
p->rs_client ? " route-server" : "",
p->as4_session ? " AS4" : "",
p->add_path_rx ? " add-path-rx" : "",
p->add_path_tx ? " add-path-tx" : "",
p->ext_messages ? " ext-messages" : "");
*/
cli_msg(-1006, " Source address: %I", p->source_addr);
cli_msg(-1006, " Hold timer: %d/%d",
tm_remains(p->conn->hold_timer), p->conn->hold_time);
cli_msg(-1006, " Keepalive timer: %d/%d",
tm_remains(p->conn->keepalive_timer), p->conn->keepalive_time);
}
if ((p->last_error_class != BE_NONE) &&
(p->last_error_class != BE_MAN_DOWN))
{
const char *err1 = bgp_err_classes[p->last_error_class];
const char *err2 = bgp_last_errmsg(p);
cli_msg(-1006, " Last error: %s%s", err1, err2);
}
{
/* XXXX ?? */
struct bgp_channel *c;
WALK_LIST(c, p->p.channels)
{
channel_show_info(&c->c);
if (c->igp_table_ip4)
cli_msg(-1006, " IGP IPv4 table: %s", c->igp_table_ip4->name);
if (c->igp_table_ip6)
cli_msg(-1006, " IGP IPv6 table: %s", c->igp_table_ip6->name);
}
}
2000-03-31 02:44:23 +08:00
}
struct channel_class channel_bgp = {
.channel_size = sizeof(struct bgp_channel),
.config_size = sizeof(struct bgp_channel_config),
.init = bgp_channel_init,
.start = bgp_channel_start,
.shutdown = bgp_channel_shutdown,
.cleanup = bgp_channel_cleanup,
.reconfigure = bgp_channel_reconfigure,
};
2000-03-20 06:09:07 +08:00
struct protocol proto_bgp = {
.name = "BGP",
.template = "bgp%d",
.attr_class = EAP_BGP,
.preference = DEF_PREF_BGP,
.channel_mask = NB_IP | NB_VPN | NB_FLOW,
.proto_size = sizeof(struct bgp_proto),
.config_size = sizeof(struct bgp_config),
.postconfig = bgp_postconfig,
.init = bgp_init,
.start = bgp_start,
.shutdown = bgp_shutdown,
.reconfigure = bgp_reconfigure,
.copy_config = bgp_copy_config,
.get_status = bgp_get_status,
.get_attr = bgp_get_attr,
.get_route_info = bgp_get_route_info,
.show_proto_info = bgp_show_proto_info
2000-03-20 06:09:07 +08:00
};