From 1ec522538fb81a56b068c087d0a842faf7aa7869 Mon Sep 17 00:00:00 2001
From: Ondrej Zajicek <santiago@crfreenet.org>
Date: Tue, 19 Nov 2013 22:33:48 +0100
Subject: BFD protocol, ready for release.

Supports OSPF and BGP and also statically configured sessions.
---
 aclocal.m4            |  12 +
 configure.in          |  35 ++-
 doc/bird.sgml         | 195 +++++++++++++-
 lib/birdlib.h         |   2 +
 lib/printf.c          |   2 +-
 nest/proto.c          |   7 +-
 proto/bfd/bfd.c       | 685 ++++++++++++++++++++++++++++++++++++++++----------
 proto/bfd/bfd.h       |  42 ++--
 proto/bfd/config.Y    |  82 +++---
 proto/bfd/io.c        |  36 ++-
 proto/bfd/io.h        |  11 +-
 proto/bfd/packets.c   |  37 +--
 proto/bgp/bgp.c       |  48 +++-
 proto/bgp/bgp.h       |   5 +
 proto/bgp/config.Y    |   3 +-
 proto/ospf/config.Y   |   1 +
 proto/ospf/hello.c    |   3 +
 proto/ospf/iface.c    |  14 ++
 proto/ospf/neighbor.c |  30 +++
 proto/ospf/neighbor.h |   1 +
 proto/ospf/ospf.h     |   4 +
 proto/radv/radv.c     |   2 +-
 sysdep/autoconf.h.in  |   4 +
 sysdep/unix/io.c      |   2 +
 sysdep/unix/log.c     |   9 +-
 25 files changed, 1044 insertions(+), 228 deletions(-)

diff --git a/aclocal.m4 b/aclocal.m4
index 3ceb6eb6..02c0f76b 100644
--- a/aclocal.m4
+++ b/aclocal.m4
@@ -133,6 +133,18 @@ if test "$bird_cv_struct_ip_mreqn" = yes ; then
 fi
 ])
 
+AC_DEFUN(BIRD_CHECK_PTHREADS,
+[
+  bird_tmp_cflags="$CFLAGS"
+
+  CFLAGS="$CFLAGS -pthread"
+  AC_CACHE_CHECK([whether POSIX threads are available], bird_cv_lib_pthreads,
+    [AC_LINK_IFELSE([AC_LANG_PROGRAM([[#include <pthread.h>]], [[pthread_t pt; pthread_create(&pt, NULL, NULL, NULL); pthread_spinlock_t lock; pthread_spin_lock(&lock); ]])],
+		    [bird_cv_lib_pthreads=yes], [bird_cv_lib_pthreads=no])])
+
+  CFLAGS="$bird_tmp_cflags"
+])
+
 AC_DEFUN(BIRD_CHECK_GCC_OPTION,
 [
   bird_tmp_cflags="$CFLAGS"
diff --git a/configure.in b/configure.in
index fc18657d..9b5dc3e2 100644
--- a/configure.in
+++ b/configure.in
@@ -10,6 +10,7 @@ AC_ARG_ENABLE(debug,	[  --enable-debug          enable internal debugging routin
 AC_ARG_ENABLE(memcheck,	[  --enable-memcheck       check memory allocations when debugging (default: enabled)],,enable_memcheck=yes)
 AC_ARG_ENABLE(client,	[  --enable-client         enable building of BIRD client (default: enabled)],,enable_client=yes)
 AC_ARG_ENABLE(ipv6,	[  --enable-ipv6           enable building of IPv6 version (default: disabled)],,enable_ipv6=no)
+AC_ARG_ENABLE(pthreads,	[  --enable-pthreads       enable POSIX threads support (default: detect)],,enable_pthreads=try)
 AC_ARG_WITH(suffix,	[  --with-suffix=STRING    use specified suffix for BIRD files (default: 6 for IPv6 version)],[given_suffix="yes"])
 AC_ARG_WITH(sysconfig,	[  --with-sysconfig=FILE   use specified BIRD system configuration file])
 AC_ARG_WITH(protocols,	[  --with-protocols=LIST   include specified routing protocols (default: all)],,[with_protocols="all"])
@@ -47,11 +48,10 @@ AC_SUBST(runtimedir)
 if test "$enable_ipv6" = yes ; then
 	ip=ipv6
 	SUFFIX=6
-	all_protocols=bfd,bgp,ospf,pipe,radv,rip,static
+	proto_radv=radv
 else
 	ip=ipv4
 	SUFFIX=""
-	all_protocols=bfd,bgp,ospf,pipe,rip,static
 fi
 
 if test "$given_suffix" = yes ; then
@@ -59,10 +59,6 @@ if test "$given_suffix" = yes ; then
 fi
 AC_SUBST(SUFFIX)
 
-if test "$with_protocols" = all ; then
-	with_protocols="$all_protocols"
-fi
-
 if test "$enable_debug" = yes ; then
 	CONFIG_FILE="bird$SUFFIX.conf"
 	CONTROL_SOCKET="bird$SUFFIX.ctl"
@@ -87,12 +83,29 @@ if test -z "$GCC" ; then
 	AC_MSG_ERROR([This program requires the GNU C Compiler.])
 fi
 
+if test "$enable_pthreads" != no ; then
+	BIRD_CHECK_PTHREADS
+
+	if test "$bird_cv_lib_pthreads" = yes ; then
+		AC_DEFINE(USE_PTHREADS)
+		CFLAGS="$CFLAGS -pthread"
+		LDFLAGS="$LDFLAGS -pthread"
+		proto_bfd=bfd
+	elif test "$enable_pthreads" = yes ; then
+		AC_MSG_ERROR([POSIX threads not available.])
+	fi
+
+	if test "$enable_pthreads" = try ; then
+		enable_pthreads="$bird_cv_lib_pthreads"
+	fi
+fi
+
 if test "$bird_cflags_default" = yes ; then
 	BIRD_CHECK_GCC_OPTION(bird_cv_c_option_wno_pointer_sign, -Wno-pointer-sign, -Wall)
 	BIRD_CHECK_GCC_OPTION(bird_cv_c_option_fno_strict_aliasing, -fno-strict-aliasing)
 	BIRD_CHECK_GCC_OPTION(bird_cv_c_option_fno_strict_overflow, -fno-strict-overflow)
 
-	CFLAGS="$CFLAGS -pthread -Wall -Wstrict-prototypes -Wno-parentheses"
+	CFLAGS="$CFLAGS -Wall -Wstrict-prototypes -Wno-parentheses"
 	BIRD_ADD_GCC_OPTION(bird_cv_c_option_wno_pointer_sign, -Wno-pointer-sign)
 	BIRD_ADD_GCC_OPTION(bird_cv_c_option_fno_strict_aliasing, -fno-strict-aliasing)
 	BIRD_ADD_GCC_OPTION(bird_cv_c_option_fno_strict_overflow, -fno-strict-overflow)
@@ -183,6 +196,13 @@ fi
 
 AC_SUBST(iproutedir)
 
+all_protocols="$proto_bfd bgp ospf pipe $proto_radv rip static"
+all_protocols=`echo $all_protocols | sed 's/ /,/g'`
+
+if test "$with_protocols" = all ; then
+	with_protocols="$all_protocols"
+fi
+
 AC_MSG_CHECKING([protocols])
 protocols=`echo "$with_protocols" | sed 's/,/ /g'`
 if test "$protocols" = no ; then protocols= ; fi
@@ -272,6 +292,7 @@ BIRD was configured with the following options:
 	Iproute2 directory:	$iproutedir
 	System configuration:	$sysdesc
 	Debugging:		$enable_debug
+	POSIX threads:		$enable_pthreads
 	Routing protocols:	$protocols
 	Client:			$enable_client
 EOF
diff --git a/doc/bird.sgml b/doc/bird.sgml
index 3cd80c32..3bc0e453 100644
--- a/doc/bird.sgml
+++ b/doc/bird.sgml
@@ -1244,6 +1244,178 @@ undefined value is regarded as empty clist for most purposes.
 
 <chapt>Protocols
 
+<sect><label id="sect-bfd">BFD
+
+<sect1>Introduction
+
+<p>Bidirectional Forwarding Detection (BFD) is not a routing protocol itself, it
+is an independent tool providing liveness and failure detection. Routing
+protocols like OSPF and BGP use integrated periodic "hello" messages to monitor
+liveness of neighbors, but detection times of these mechanisms are high (e.g. 40
+seconds by default in OSPF, could be set down to several seconds). BFD offers
+universal, fast and low-overhead mechanism for failure detection, which could be
+attached to any routing protocol in an advisory role.
+
+<p>BFD consists of mostly independent BFD sessions. Each session monitors an
+unicast bidirectional path between two BFD-enabled routers. This is done by
+periodically sending control packets in both directions. BFD does not handle
+neighbor discovery, BFD sessions are created on demand by request of other
+protocols (like OSPF or BGP), which supply appropriate information like IP
+addresses and associated interfaces. When a session changes its state, these
+protocols are notified and act accordingly (e.g. break an OSPF adjacency when
+the BFD session went down).
+
+<p>BIRD implements basic BFD behavior as defined in
+RFC 5880<htmlurl url="ftp://ftp.rfc-editor.org/in-notes/rfc5880.txt">
+(some advanced features like the echo mode or authentication are not implemented),
+IP transport for BFD as defined in
+RFC 5881<htmlurl url="ftp://ftp.rfc-editor.org/in-notes/rfc5881.txt"> and
+RFC 5883<htmlurl url="ftp://ftp.rfc-editor.org/in-notes/rfc5883.txt">
+and interaction with client protocols as defined in
+RFC 5882<htmlurl url="ftp://ftp.rfc-editor.org/in-notes/rfc5882.txt">.
+
+<p>Note that BFD implementation in BIRD is currently a new feature in
+development, expect some rough edges and possible UI and configuration changes
+in the future. Also note that we currently support at most one protocol instance.
+
+<sect1>Configuration
+
+<p>BFD configuration consists mainly of multiple definitions of interfaces.
+Most BFD config options are session specific. When a new session is requested
+and dynamically created, it is configured from one of these definitions. For
+sessions to directly connected neighbors, <cf/interface/ definitions are chosen
+based on the interface associated with the session, while <cf/multihop/
+definition is used for multihop sessions. If no definition is relevant, the
+session is just created with the default configuration. Therefore, an empty BFD
+configuration is often sufficient.
+
+<p>Note that to use BFD for other protocols like OSPF or BGP, these protocols
+also have to be configured to request BFD sessions, usually by <cf/bfd/ option.
+
+<p>Some of BFD session options require <m/time/ value, which has to be specified
+with the appropriate unit: <m/num/ <cf/s/|<cf/ms/|<cf/us/. Although microseconds
+are allowed as units, practical minimum values are usually in order of tens of
+milliseconds.
+
+<code>
+protocol bfd [&lt;name&gt;] {
+	interface &lt;interface pattern&gt; {
+		interval &lt;time&gt;;
+		min rx interval &lt;time&gt;;
+		min tx interval &lt;time&gt;;
+		idle tx interval &lt;time&gt;;
+		multiplier &lt;num&gt;;
+		passive &lt;switch&gt;;
+	};
+	multihop {
+		interval &lt;time&gt;;
+		min rx interval &lt;time&gt;;
+		min tx interval &lt;time&gt;;
+		idle tx interval &lt;time&gt;;
+		multiplier &lt;num&gt;;
+		passive &lt;switch&gt;;
+	};
+	neighbor &lt;ip&gt; [dev "&lt;interface&gt;"] [local &lt;ip&gt;] [multihop &lt;switch&gt;];
+}
+</code>
+
+<descrip>
+	<tag>interface <m/pattern [, ...]/  { <m/options/ }</tag>
+	Interface definitions allow to specify options for sessions associated
+	with such interfaces and also may contain interface specific options.
+	See <ref id="dsc-iface" name="interface"> common option for a detailed
+	description of interface patterns. Note that contrary to the behavior of
+	<cf/interface/ definitions of other protocols, BFD protocol would accept
+	sessions (in default configuration) even on interfaces not covered by
+	such definitions.
+
+	<tag>multihop { <m/options/ }</tag>
+	Multihop definitions allow to specify options for multihop BFD sessions,
+	in the same manner as <cf/interface/ definitions are used for directly
+	connected sessions. Currently only one such definition (for all multihop
+	sessions) could be used.
+
+	<tag>neighbor <m/ip/ [dev "<m/interface/"] [local <m/ip/] [multihop <m/switch/]</tag>
+	BFD sessions are usually created on demand as requested by other
+	protocols (like OSPF or BGP). This option allows to explicitly add
+	a BFD session to the specified neighbor regardless of such requests.
+	
+	The session is identified by the IP address of the neighbor, with
+	optional specification of used interface and local IP.  By default
+	the neighbor must be directly connected, unless the the session is
+	configured as multihop. Note that local IP must be specified for
+	multihop sessions.
+</descrip>
+
+<p>Session specific options (part of <cf/interface/ and <cf/multihop/ definitions):
+
+<descrip>
+	<tag>interval <m/time/</tag>
+	BFD ensures availability of the forwarding path associated with the
+	session by periodically sending BFD control packets in both
+	directions. The rate of such packets is controlled by two options,
+	<cf/min rx interval/ and <cf/min tx interval/ (see below). This option
+	is just a shorthand to set both of these options together.
+
+	<tag>min rx interval <m/time/</tag>
+	This option specifies the minimum RX interval, which is announced to the
+	neighbor and used there to limit the neighbor's rate of generated BFD
+	control packets. Default: 10 ms.
+
+	<tag>min tx interval <m/time/</tag>
+	This option specifies the desired TX interval, which controls the rate
+	of generated BFD control packets (together with <cf/min rx interval/
+	announced by the neighbor). Note that this value is used only if the BFD
+	session is up, otherwise the value of <cf/idle tx interval/ is used
+	instead. Default: 100 ms.
+
+	<tag>idle tx interval <m/time/</tag>
+	In order to limit unnecessary traffic in cases where a neighbor is not
+	available or not running BFD, the rate of generated BFD control packets
+	is lower when the BFD session is not up. This option specifies the
+	desired TX interval in such cases instead of <cf/min tx interval/.
+	Default: 1 s.
+
+	<tag>multiplier <m/num/</tag>
+	Failure detection time for BFD sessions is based on established rate of
+	BFD control packets (<cf>min rx/tx interval</cf>) multiplied by this
+	multiplier, which is essentially (ignoring jitter) a number of missed
+	packets after which the session is declared down. Note that rates and
+	multipliers could be different in each direction of a BFD session.
+	Default: 5.
+
+	<tag>passive <m/switch/</tag>
+	Generally, both BFD session endpoinds try to establish the session by
+	sending control packets to the other side. This option allows to enable
+	passive mode, which means that the router does not send BFD packets
+	until it has received one from the other side. Default: disabled.
+</descrip>
+
+<sect1>Example
+
+<p><code>
+protocol bfd {
+	interface "eth*" {
+		min rx interval 20 ms;
+		min tx interval 50 ms;
+		idle tx interval 300 ms;
+	};
+	interface "gre*" {
+		interval 200 ms;
+		multiplier 10;
+		passive;
+	};
+	multihop {
+		interval 200 ms;
+		multiplier 10;
+	};
+
+	neighbor 192.168.1.10;
+	neighbor 192.168.2.2 dev "eth2";
+	neighbor 192.168.10.1 local 192.168.1.1 multihop;
+}
+</code>
+
 <sect>BGP
 
 <p>The Border Gateway Protocol is the routing protocol used for backbone
@@ -1258,8 +1430,8 @@ AS). Each AS is a part of the network with common management and
 common routing policy. It is identified by a unique 16-bit number
 (ASN).  Routers within each AS usually exchange AS-internal routing
 information with each other using an interior gateway protocol (IGP,
-such as OSPF or RIP). Boundary routers at the border of
-the AS communicate global (inter-AS) network reachability information with
+such as OSPF or RIP). Boundary routers at the border of the AS
+communicate global (inter-AS) network reachability information with
 their neighbors in the neighboring AS'es via exterior BGP (eBGP) and
 redistribute received information to other routers in the AS via
 interior BGP (iBGP).
@@ -1412,7 +1584,15 @@ for each neighbor using the following configuration parameters:
 	<tag>igp table <m/name/</tag> Specifies a table that is used
 	as an IGP routing table. Default: the same as the table BGP is
 	connected to.
-	
+
+	<tag>bfd <M>switch</M></tag>
+	BGP could use BFD protocol as an advisory mechanism for neighbor
+	liveness and failure detection. If enabled, BIRD setups a BFD session
+	for the BGP neighbor and tracks its liveness by it. This has an
+	advantage of an order of magnitude lower detection times in case of
+	failure. Note that BFD protocol also has to be configured, see
+	<ref id="sect-bfd" name="BFD"> section for details. Default: disabled.
+
 	<tag>ttl security <m/switch/</tag> Use GTSM (RFC 5082 - the
 	generalized TTL security mechanism). GTSM protects against
 	spoofed packets by ignoring received packets with a smaller
@@ -1986,6 +2166,7 @@ protocol ospf &lt;name&gt; {
 			real broadcast &lt;switch&gt;;
 			ptp netmask &lt;switch&gt;;
 			check link &lt;switch&gt;;
+			bfd &lt;switch&gt;;
 			ecmp weight &lt;num&gt;;
 			ttl security [&lt;switch&gt;; | tx only]
 			tx class|dscp &lt;num&gt;;
@@ -2260,6 +2441,14 @@ protocol ospf &lt;name&gt; {
 	 prefix) is propagated. It is possible that some hardware
 	 drivers or platforms do not implement this feature. Default value is no.
 
+	<tag>bfd <M>switch</M></tag>
+	OSPF could use BFD protocol as an advisory mechanism for neighbor
+	liveness and failure detection. If enabled, BIRD setups a BFD session
+	for each OSPF neighbor and tracks its liveness by it. This has an
+	advantage of an order of magnitude lower detection times in case of
+	failure. Note that BFD protocol also has to be configured, see
+	<ref id="sect-bfd" name="BFD"> section for details. Default value is no.
+
 	<tag>ttl security [<m/switch/ | tx only]</tag>
 	 TTL security is a feature that protects routing protocols
 	 from remote spoofed packets by using TTL 255 instead of TTL 1
diff --git a/lib/birdlib.h b/lib/birdlib.h
index 2d6849e1..a5424958 100644
--- a/lib/birdlib.h
+++ b/lib/birdlib.h
@@ -24,6 +24,8 @@
 #define _MAX(a,b) (((a)>(b))?(a):(b))
 
 #ifndef PARSER
+#undef MIN
+#undef MAX
 #define MIN(a,b) _MIN(a,b)
 #define MAX(a,b) _MAX(a,b)
 #endif
diff --git a/lib/printf.c b/lib/printf.c
index d8600b61..41e1cc0d 100644
--- a/lib/printf.c
+++ b/lib/printf.c
@@ -276,7 +276,7 @@ int bvsnprintf(char *buf, int size, const char *fmt, va_list args)
 				ip_ntox(va_arg(args, ip_addr), ipbuf);
 			else {
 				ip_ntop(va_arg(args, ip_addr), ipbuf);
-				if (field_width > 0)
+				if (field_width == 1)
 					field_width = STD_ADDRESS_P_LENGTH;
 			}
 			s = ipbuf;
diff --git a/nest/proto.c b/nest/proto.c
index 0c85c2d9..c15247be 100644
--- a/nest/proto.c
+++ b/nest/proto.c
@@ -681,6 +681,9 @@ proto_build(struct protocol *p)
     }
 }
 
+/* FIXME: convert this call to some protocol hook */
+extern void bfd_init_all(void);
+
 /**
  * protos_build - build a protocol list
  *
@@ -718,8 +721,10 @@ protos_build(void)
 #ifdef CONFIG_BGP
   proto_build(&proto_bgp);
 #endif
-  // XXX
+#ifdef CONFIG_BFD
   proto_build(&proto_bfd);
+  bfd_init_all();
+#endif
 
   proto_pool = rp_new(&root_pool, "Protocols");
   proto_flush_event = ev_new(proto_pool);
diff --git a/proto/bfd/bfd.c b/proto/bfd/bfd.c
index 3e2af9d5..5ebfadc1 100644
--- a/proto/bfd/bfd.c
+++ b/proto/bfd/bfd.c
@@ -4,6 +4,103 @@
  *	Can be freely distributed and used under the terms of the GNU GPL.
  */
 
+/**
+ * DOC: Bidirectional Forwarding Detection
+ *
+ * The BFD protocol is implemented in three files: |bfd.c| containing the
+ * protocol logic and the protocol glue with BIRD core, |packets.c| handling BFD
+ * packet processing, RX, TX and protocol sockets. |io.c| then contains generic
+ * code for the event loop, threads and event sources (sockets, microsecond
+ * timers). This generic code will be merged to the main BIRD I/O code in the
+ * future.
+ *
+ * The BFD implementation uses a separate thread with an internal event loop for
+ * handling the protocol logic, which requires high-res and low-latency timing,
+ * so it is not affected by the rest of BIRD, which has several low-granularity
+ * hooks in the main loop, uses second-based timers and cannot offer good
+ * latency. The core of BFD protocol (the code related to BFD sessions,
+ * interfaces and packets) runs in the BFD thread, while the rest (the code
+ * related to BFD requests, BFD neighbors and the protocol glue) runs in the
+ * main thread.
+ *
+ * BFD sessions are represented by structure &bfd_session that contains a state
+ * related to the session and two timers (TX timer for periodic packets and hold
+ * timer for session timeout). These sessions are allocated from @session_slab
+ * and are accessible by two hash tables, @session_hash_id (by session ID) and
+ * @session_hash_ip (by IP addresses of neighbors). Slab and both hashes are in
+ * the main protocol structure &bfd_proto. The protocol logic related to BFD
+ * sessions is implemented in internal functions bfd_session_*(), which are
+ * expected to be called from the context of BFD thread, and external functions
+ * bfd_add_session(), bfd_remove_session() and bfd_reconfigure_session(), which
+ * form an interface to the BFD core for the rest and are expected to be called
+ * from the context of main thread.
+ *
+ * Each BFD session has an associated BFD interface, represented by structure
+ * &bfd_iface. A BFD interface contains a socket used for TX (the one for RX is
+ * shared in &bfd_proto), an interface configuration and reference counter.
+ * Compared to interface structures of other protocols, these structures are not
+ * created and removed based on interface notification events, but according to
+ * the needs of BFD sessions. When a new session is created, it requests a
+ * proper BFD interface by function bfd_get_iface(), which either finds an
+ * existing one in &iface_list (from &bfd_proto) or allocates a new one. When a
+ * session is removed, an associated iface is dicharged by bfd_free_iface().
+ *
+ * BFD requests are the external API for the other protocols. When a protocol
+ * wants a BFD session, it calls bfd_request_session(), which creates a
+ * structure &bfd_request containing approprite information and an notify hook.
+ * This structure is a resource associated with the caller's resource pool. When
+ * a BFD protocol is available, a BFD request is submitted to the protocol, an
+ * appropriate BFD session is found or created and the request is attached to
+ * the session. When a session changes state, all attached requests (and related
+ * protocols) are notified. Note that BFD requests do not depend on BFD protocol
+ * running. When the BFD protocol is stopped or removed (or not available from
+ * beginning), related BFD requests are stored in @bfd_wait_list, where waits
+ * for a new protocol.
+ *
+ * BFD neighbors are just a way to statically configure BFD sessions without
+ * requests from other protocol. Structures &bfd_neighbor are part of BFD
+ * configuration (like static routes in the static protocol). BFD neighbors are
+ * handled by BFD protocol like it is a BFD client -- when a BFD neighbor is
+ * ready, the protocol just creates a BFD request like any other protocol.
+ * 
+ * The protocol uses a new generic event loop (structure &birdloop) from |io.c|,
+ * which supports sockets, timers and events like the main loop. Timers
+ * (structure &timer2) are new microsecond based timers, while sockets and
+ * events are the same. A birdloop is associated with a thread (field @thread)
+ * in which event hooks are executed. Most functions for setting event sources
+ * (like sk_start() or tm2_start()) must be called from the context of that
+ * thread. Birdloop allows to temporarily acquire the context of that thread for
+ * the main thread by calling birdloop_enter() and then birdloop_leave(), which
+ * also ensures mutual exclusion with all event hooks. Note that resources
+ * associated with a birdloop (like timers) should be attached to the
+ * independent resource pool, detached from the main resource tree.
+ *
+ * There are two kinds of interaction between the BFD core (running in the BFD
+ * thread) and the rest of BFD (running in the main thread). The first kind are
+ * configuration calls from main thread to the BFD thread (like bfd_add_session()).
+ * These calls are synchronous and use birdloop_enter() mechanism for mutual
+ * exclusion. The second kind is a notification about session changes from the
+ * BFD thread to the main thread. This is done in an asynchronous way, sesions
+ * with pending notifications are linked (in the BFD thread) to @notify_list in
+ * &bfd_proto, and then bfd_notify_hook() in the main thread is activated using
+ * bfd_notify_kick() and a pipe. The hook then processes scheduled sessions and
+ * calls hooks from associated BFD requests. This @notify_list (and state fields
+ * in structure &bfd_session) is protected by a spinlock in &bfd_proto and
+ * functions bfd_lock_sessions() / bfd_unlock_sessions().
+ *
+ * There are few data races (accessing @p->p.debug from TRACE() from the BFD
+ * thread and accessing some some private fields of %bfd_session from
+ * bfd_show_sessions() from the main thread, but these are harmless (i hope).
+ *
+ * TODO: document functions and access restrictions for fields in BFD structures.
+ *
+ * Supported standards:
+ * - RFC 5880 - main BFD standard
+ * - RFC 5881 - BFD for IP links
+ * - RFC 5882 - generic application of BFD
+ * - RFC 5883 - BFD for multihop paths
+ */
+
 #include "bfd.h"
 
 
@@ -17,23 +114,34 @@
 #define HASH_IP_EQ(a,b)		ipa_equal(a,b)
 #define HASH_IP_FN(k)		ipa_hash(k)
 
+static list bfd_proto_list;
+static list bfd_wait_list;
 
 const char *bfd_state_names[] = { "AdminDown", "Down", "Init", "Up" };
 
+static void bfd_session_set_min_tx(struct bfd_session *s, u32 val);
+static struct bfd_iface *bfd_get_iface(struct bfd_proto *p, ip_addr local, struct iface *iface);
+static void bfd_free_iface(struct bfd_iface *ifa);
 static inline void bfd_notify_kick(struct bfd_proto *p);
 
+
+/*
+ *	BFD sessions
+ */
+
 static void 
 bfd_session_update_state(struct bfd_session *s, uint state, uint diag)
 {
-  struct bfd_proto *p = s->bfd;
+  struct bfd_proto *p = s->ifa->bfd;
+  uint old_state = s->loc_state; 
   int notify;
 
-  if (s->loc_state == state)
+  if (state == old_state)
     return;
 
-  TRACE(D_EVENTS, "Session changed %I %d %d", s->addr, state, diag);
-  debug("STATE %I %d %d %d\n", s->addr, s->loc_state, state, diag);
-    
+  TRACE(D_EVENTS, "Session to %I changed state from %s to %s",
+	s->addr, bfd_state_names[old_state], bfd_state_names[state]);
+
   bfd_lock_sessions(p);
   s->loc_state = state;
   s->loc_diag = diag;
@@ -43,21 +151,14 @@ bfd_session_update_state(struct bfd_session *s, uint state, uint diag)
     add_tail(&p->notify_list, &s->n);
   bfd_unlock_sessions(p);
 
-  if (notify)
-    bfd_notify_kick(p);
-}
+  if (state == BFD_STATE_UP)
+    bfd_session_set_min_tx(s, s->ifa->cf->min_tx_int);
 
-static void 
-bfd_session_timeout(struct bfd_session *s)
-{
-  s->rem_state = BFD_STATE_DOWN;
-  s->rem_id = 0;
-  s->rem_min_tx_int = 0;
-  s->rem_min_rx_int = 1;
-  s->rem_demand_mode = 0;
-  s->rem_detect_mult = 0;
+  if (old_state == BFD_STATE_UP)
+    bfd_session_set_min_tx(s, s->ifa->cf->idle_tx_int);
 
-  bfd_session_update_state(s, BFD_STATE_DOWN, BFD_DIAG_TIMEOUT);
+  if (notify)
+    bfd_notify_kick(p);
 }
 
 static void
@@ -93,10 +194,9 @@ bfd_session_update_detection_time(struct bfd_session *s, int kick)
 }
 
 static void
-bfd_session_control_tx_timer(struct bfd_session *s)
+bfd_session_control_tx_timer(struct bfd_session *s, int reset)
 {
-  if (!s->opened)
-    goto stop;
+  // if (!s->opened) goto stop;
 
   if (s->passive && (s->rem_id == 0))
     goto stop;
@@ -111,10 +211,12 @@ bfd_session_control_tx_timer(struct bfd_session *s)
     goto stop;
 
   /* So TX timer should run */
-  if (tm2_active(s->tx_timer))
-    return;
+  if (reset || !tm2_active(s->tx_timer))
+  {
+    s->last_tx = 0;
+    tm2_start(s->tx_timer, 0);
+  }
 
-  tm2_start(s->tx_timer, 0);
   return;
 
  stop:
@@ -125,6 +227,10 @@ bfd_session_control_tx_timer(struct bfd_session *s)
 static void
 bfd_session_request_poll(struct bfd_session *s, u8 request)
 {
+  /* Not sure about this, but doing poll in this case does not make sense */
+  if (s->rem_id == 0)
+    return;
+
   s->poll_scheduled |= request;
 
   if (s->poll_active)
@@ -132,7 +238,8 @@ bfd_session_request_poll(struct bfd_session *s, u8 request)
 
   s->poll_active = s->poll_scheduled;
   s->poll_scheduled = 0;
-  bfd_send_ctl(s->bfd, s, 0);
+
+  bfd_session_control_tx_timer(s, 1);
 }
 
 static void
@@ -146,11 +253,10 @@ bfd_session_terminate_poll(struct bfd_session *s)
   if (poll_done & BFD_POLL_RX)
     s->req_min_rx_int = s->req_min_rx_new;
 
-  s->poll_active = 0;
+  s->poll_active = s->poll_scheduled;
+  s->poll_scheduled = 0;
 
   /* Timers are updated by caller - bfd_session_process_ctl() */
-
-  // xxx_restart_poll();
 }
 
 void
@@ -191,10 +297,32 @@ bfd_session_process_ctl(struct bfd_session *s, u8 flags, u32 old_tx_int, u32 old
   if (next_state)
     bfd_session_update_state(s, next_state, diag);
 
-  bfd_session_control_tx_timer(s);
+  bfd_session_control_tx_timer(s, 0);
 
   if (flags & BFD_FLAG_POLL)
-    bfd_send_ctl(s->bfd, s, 1);
+    bfd_send_ctl(s->ifa->bfd, s, 1);
+}
+
+static void 
+bfd_session_timeout(struct bfd_session *s)
+{
+  struct bfd_proto *p = s->ifa->bfd;
+
+  TRACE(D_EVENTS, "Session to %I expired", s->addr);
+
+  s->rem_state = BFD_STATE_DOWN;
+  s->rem_id = 0;
+  s->rem_min_tx_int = 0;
+  s->rem_min_rx_int = 1;
+  s->rem_demand_mode = 0;
+  s->rem_detect_mult = 0;
+
+  s->poll_active = 0;
+  s->poll_scheduled = 0;
+
+  bfd_session_update_state(s, BFD_STATE_DOWN, BFD_DIAG_TIMEOUT);
+
+  bfd_session_control_tx_timer(s, 1);
 }
 
 static void
@@ -255,8 +383,7 @@ bfd_tx_timer_hook(timer2 *t)
   struct bfd_session *s = t->data;
 
   s->last_tx = current_time();
-  //  debug("TX %d\n", (s32)  (s->last_tx TO_MS));
-  bfd_send_ctl(s->bfd, s, 0);
+  bfd_send_ctl(s->ifa->bfd, s, 0);
 }
 
 static void
@@ -277,48 +404,53 @@ bfd_get_free_id(struct bfd_proto *p)
 }
 
 static struct bfd_session *
-bfd_add_session(struct bfd_proto *p, ip_addr addr, struct bfd_session_config *opts)
+bfd_add_session(struct bfd_proto *p, ip_addr addr, ip_addr local, struct iface *iface)
 {
   birdloop_enter(p->loop);
 
+  struct bfd_iface *ifa = bfd_get_iface(p, local, iface);
+
   struct bfd_session *s = sl_alloc(p->session_slab);
   bzero(s, sizeof(struct bfd_session));
 
   s->addr = addr;
+  s->ifa = ifa;
   s->loc_id = bfd_get_free_id(p);
-  debug("XXX INS1 %d %d %u %I\n", p->session_hash_id.count,  p->session_hash_ip.count, s->loc_id, s->addr);
+
   HASH_INSERT(p->session_hash_id, HASH_ID, s);
-  debug("XXX INS2 %d %d\n", p->session_hash_id.count,  p->session_hash_ip.count);
   HASH_INSERT(p->session_hash_ip, HASH_IP, s);
-  debug("XXX INS3 %d %d\n", p->session_hash_id.count,  p->session_hash_ip.count);
-  s->bfd = p;
+
 
   /* Initialization of state variables - see RFC 5880 6.8.1 */
   s->loc_state = BFD_STATE_DOWN;
   s->rem_state = BFD_STATE_DOWN;
-  s->des_min_tx_int = s->des_min_tx_new = opts->min_tx_int; // XXX  opts->idle_tx_int;
-  s->req_min_rx_int = s->req_min_rx_new = opts->min_rx_int;
+  s->des_min_tx_int = s->des_min_tx_new = ifa->cf->idle_tx_int;
+  s->req_min_rx_int = s->req_min_rx_new = ifa->cf->min_rx_int;
   s->rem_min_rx_int = 1;
-  s->detect_mult = opts->multiplier;
-  s->passive = opts->passive;
+  s->detect_mult = ifa->cf->multiplier;
+  s->passive = ifa->cf->passive;
 
   s->tx_timer = tm2_new_init(p->tpool, bfd_tx_timer_hook, s, 0, 0);
   s->hold_timer = tm2_new_init(p->tpool, bfd_hold_timer_hook, s, 0, 0);
   bfd_session_update_tx_interval(s);
+  bfd_session_control_tx_timer(s, 1);
+
+  init_list(&s->request_list);
+  s->last_state_change = now;
+
+  TRACE(D_EVENTS, "Session to %I added", s->addr);
 
   birdloop_leave(p->loop);
 
   return s;
 }
 
+/*
 static void
 bfd_open_session(struct bfd_proto *p, struct bfd_session *s, ip_addr local, struct iface *ifa)
 {
   birdloop_enter(p->loop);
 
-  s->bsock = bfd_get_socket(p, local, ifa);
-  // s->local = local;
-  // s->iface = ifa;
   s->opened = 1;
 
   bfd_session_control_tx_timer(s);
@@ -331,10 +463,6 @@ bfd_close_session(struct bfd_proto *p, struct bfd_session *s)
 {
   birdloop_enter(p->loop);
 
-  bfd_free_socket(s->bsock);
-  s->bsock = NULL;
-  // s->local = IPA_NONE;
-  // s->iface = NULL;
   s->opened = 0;
 
   bfd_session_update_state(s, BFD_STATE_DOWN, BFD_DIAG_PATH_DOWN);
@@ -342,54 +470,295 @@ bfd_close_session(struct bfd_proto *p, struct bfd_session *s)
 
   birdloop_leave(p->loop);
 }
+*/
 
 static void
 bfd_remove_session(struct bfd_proto *p, struct bfd_session *s)
 {
+  ip_addr ip = s->addr;
+
   birdloop_enter(p->loop);
 
-  bfd_free_socket(s->bsock);
+  bfd_free_iface(s->ifa);
 
   rfree(s->tx_timer);
   rfree(s->hold_timer);
 
-  debug("XXX REM1 %d %d %u %I\n", p->session_hash_id.count,  p->session_hash_ip.count, s->loc_id, s->addr);
   HASH_REMOVE(p->session_hash_id, HASH_ID, s);
-  debug("XXX REM2 %d %d\n", p->session_hash_id.count,  p->session_hash_ip.count);
   HASH_REMOVE(p->session_hash_ip, HASH_IP, s);
-  debug("XXX REM3 %d %d\n", p->session_hash_id.count,  p->session_hash_ip.count);
 
   sl_free(p->session_slab, s);
 
+  TRACE(D_EVENTS, "Session to %I removed", ip);
+
   birdloop_leave(p->loop);
 }
 
 static void
-bfd_configure_session(struct bfd_proto *p, struct bfd_session *s,
-		      struct bfd_session_config *opts)
+bfd_reconfigure_session(struct bfd_proto *p, struct bfd_session *s)
 {
   birdloop_enter(p->loop);
 
- // XXX  opts->idle_tx_int;
+  struct bfd_iface_config *cf = s->ifa->cf;
 
-  bfd_session_set_min_tx(s, opts->min_tx_int);
-  bfd_session_set_min_rx(s, opts->min_rx_int);
-  s->detect_mult = opts->multiplier;
-  s->passive = opts->passive;
+  u32 tx = (s->loc_state == BFD_STATE_UP) ? cf->min_tx_int : cf->idle_tx_int;
+  bfd_session_set_min_tx(s, tx);
+  bfd_session_set_min_rx(s, cf->min_rx_int);
+  s->detect_mult = cf->multiplier;
+  s->passive = cf->passive;
 
-  bfd_session_control_tx_timer(s);
+  bfd_session_control_tx_timer(s, 0);
 
   birdloop_leave(p->loop);
+
+  TRACE(D_EVENTS, "Session to %I reconfigured", s->addr);
+}
+
+
+/*
+ *	BFD interfaces
+ */
+
+static struct bfd_iface_config bfd_default_iface = {
+  .min_rx_int = BFD_DEFAULT_MIN_RX_INT,
+  .min_tx_int = BFD_DEFAULT_MIN_TX_INT,
+  .idle_tx_int = BFD_DEFAULT_IDLE_TX_INT,
+  .multiplier = BFD_DEFAULT_MULTIPLIER
+};
+
+static inline struct bfd_iface_config *
+bfd_find_iface_config(struct bfd_config *cf, struct iface *iface)
+{
+  struct bfd_iface_config *ic;
+
+  ic = iface ? (void *) iface_patt_find(&cf->patt_list, iface, NULL) : cf->multihop;
+
+  return ic ? ic : &bfd_default_iface;
+}
+
+static struct bfd_iface *
+bfd_get_iface(struct bfd_proto *p, ip_addr local, struct iface *iface)
+{
+  struct bfd_iface *ifa;
+
+  WALK_LIST(ifa, p->iface_list)
+    if (ipa_equal(ifa->local, local) && (ifa->iface == iface))
+      return ifa->uc++, ifa;
+
+  struct bfd_config *cf = (struct bfd_config *) (p->p.cf);
+  struct bfd_iface_config *ic = bfd_find_iface_config(cf, iface);
+
+  ifa = mb_allocz(p->tpool, sizeof(struct bfd_iface));
+  ifa->local = local;
+  ifa->iface = iface;
+  ifa->cf = ic;
+  ifa->bfd = p;
+
+  ifa->sk = bfd_open_tx_sk(p, local, iface);
+  ifa->uc = 1;
+
+  add_tail(&p->iface_list, &ifa->n);
+
+  return ifa;
+}
+
+static void
+bfd_free_iface(struct bfd_iface *ifa)
+{
+  if (!ifa || --ifa->uc)
+    return;
+
+  rem_node(&ifa->n);
+  sk_stop(ifa->sk);
+  rfree(ifa->sk);
+  mb_free(ifa);
+}
+
+static void
+bfd_reconfigure_iface(struct bfd_proto *p, struct bfd_iface *ifa, struct bfd_config *nc)
+{
+  struct bfd_iface_config *nic = bfd_find_iface_config(nc, ifa->iface);
+  ifa->changed = !!memcmp(nic, ifa->cf, sizeof(struct bfd_iface_config));
+
+  /* This should be probably changed to not access ifa->cf from the BFD thread */
+  birdloop_enter(p->loop);
+  ifa->cf = nic;
+  birdloop_leave(p->loop);
+}
+
+
+/*
+ *	BFD requests
+ */
+
+static void
+bfd_request_notify(struct bfd_request *req, u8 state, u8 diag)
+{
+  u8 old_state = req->state;
+
+  if (state == old_state)
+    return;
+
+  req->state = state;
+  req->diag = diag;
+  req->old_state = old_state;
+  req->down = (old_state == BFD_STATE_UP) && (state == BFD_STATE_DOWN);
+
+  if (req->hook)
+    req->hook(req);
+}
+
+static int
+bfd_add_request(struct bfd_proto *p, struct bfd_request *req)
+{
+  struct bfd_session *s = bfd_find_session_by_addr(p, req->addr);
+  u8 state, diag;
+
+  if (!s)
+    s = bfd_add_session(p, req->addr, req->local, req->iface);
+
+  rem_node(&req->n);
+  add_tail(&s->request_list, &req->n);
+  req->session = s;
+
+  bfd_lock_sessions(p);
+  state = s->loc_state;
+  diag = s->loc_diag;
+  bfd_unlock_sessions(p);
+
+  bfd_request_notify(req, state, diag);
+
+  return 1;
+}
+
+static void
+bfd_submit_request(struct bfd_request *req)
+{
+  node *n;
+
+  WALK_LIST(n, bfd_proto_list)
+    if (bfd_add_request(SKIP_BACK(struct bfd_proto, bfd_node, n), req))
+      return;
+
+  rem_node(&req->n);
+  add_tail(&bfd_wait_list, &req->n);
+  req->session = NULL;
+  bfd_request_notify(req, BFD_STATE_ADMIN_DOWN, 0);
+}
+
+static void
+bfd_take_requests(struct bfd_proto *p)
+{
+  node *n, *nn;
+
+  WALK_LIST_DELSAFE(n, nn, bfd_wait_list)
+    bfd_add_request(p, SKIP_BACK(struct bfd_request, n, n));
+}
+
+static void
+bfd_drop_requests(struct bfd_proto *p)
+{
+  node *n;
+
+  HASH_WALK(p->session_hash_id, next_id, s)
+  {
+    /* We assume that p is not in bfd_proto_list */
+    WALK_LIST_FIRST(n, s->request_list)
+      bfd_submit_request(SKIP_BACK(struct bfd_request, n, n));
+  }
+  HASH_WALK_END;
+}
+
+static struct resclass bfd_request_class;
+
+struct bfd_request *
+bfd_request_session(pool *p, ip_addr addr, ip_addr local, struct iface *iface,
+		    void (*hook)(struct bfd_request *), void *data)
+{
+  struct bfd_request *req = ralloc(p, &bfd_request_class);
+
+  /* Hack: self-link req->n, we will call rem_node() on it */
+  req->n.prev = req->n.next = &req->n;
+
+  req->addr = addr;
+  req->local = local;
+  req->iface = iface;
+
+  bfd_submit_request(req);
+
+  req->hook = hook;
+  req->data = data;
+
+  return req;
+}
+
+static void
+bfd_request_free(resource *r)
+{
+  struct bfd_request *req = (struct bfd_request *) r;
+  struct bfd_session *s = req->session;
+
+  rem_node(&req->n);
+
+  /* Remove the session if there is no request for it. Skip that if
+     inside notify hooks, will be handled by bfd_notify_hook() itself */
+
+  if (s && EMPTY_LIST(s->request_list) && !s->notify_running)
+    bfd_remove_session(s->ifa->bfd, s);
+}
+
+static void
+bfd_request_dump(resource *r)
+{
+  struct bfd_request *req = (struct bfd_request *) r;
+
+  debug("(code %p, data %p)\n", req->hook, req->data);
+}
+
+static struct resclass bfd_request_class = {
+  "BFD request",
+  sizeof(struct bfd_request),
+  bfd_request_free,
+  bfd_request_dump,
+  NULL,
+  NULL
+};
+
+
+/*
+ *	BFD neighbors
+ */
+
+static void
+bfd_neigh_notify(struct neighbor *nb)
+{
+  struct bfd_proto *p = (struct bfd_proto *) nb->proto;
+  struct bfd_neighbor *n = nb->data;
+
+  if (!n)
+    return;
+
+  if ((nb->scope > 0) && !n->req)
+  {
+    ip_addr local = ipa_nonzero(n->local) ? n->local : nb->iface->addr->ip;
+    n->req = bfd_request_session(p->p.pool, n->addr, local, nb->iface, NULL, NULL);
+  }
+
+  if ((nb->scope <= 0) && n->req)
+  {
+    rfree(n->req);
+    n->req = NULL;
+  }
 }
 
 static void
 bfd_start_neighbor(struct bfd_proto *p, struct bfd_neighbor *n)
 {
-  n->session = bfd_add_session(p, n->addr, n->opts);
+  n->active = 1;
 
-  if (n->opts->multihop)
+  if (n->multihop)
   {
-    bfd_open_session(p, n->session, n->local, NULL);
+    n->req = bfd_request_session(p->p.pool, n->addr, n->local, NULL, NULL, NULL);
     return;
   }
 
@@ -402,14 +771,15 @@ bfd_start_neighbor(struct bfd_proto *p, struct bfd_neighbor *n)
 
   if (nb->data)
   {
-    log(L_ERR "%s: Duplicate remote address %I", p->p.name, n->addr);
+    log(L_ERR "%s: Duplicate neighbor %I", p->p.name, n->addr);
     return;
   }
 
-  nb->data = n->session;
+  n->neigh = nb;
+  nb->data = n;
 
   if (nb->scope > 0)
-    bfd_open_session(p, n->session, nb->iface->addr->ip, nb->iface);
+    bfd_neigh_notify(nb);
   else
     TRACE(D_EVENTS, "Waiting for %I%J to become my neighbor", n->addr, n->iface);
 }
@@ -417,33 +787,54 @@ bfd_start_neighbor(struct bfd_proto *p, struct bfd_neighbor *n)
 static void
 bfd_stop_neighbor(struct bfd_proto *p, struct bfd_neighbor *n)
 {
-  if (!n->opts->multihop)
-  {
-    struct neighbor *nb = neigh_find2(&p->p, &n->addr, n->iface, 0);
-    if (nb)
-      nb->data = NULL;
-  }
+  if (n->neigh)
+    n->neigh->data = NULL;
+  n->neigh = NULL;
 
-  bfd_remove_session(p, n->session);
+  rfree(n->req);
+  n->req = NULL;
 }
 
-static void
-bfd_neigh_notify(struct neighbor *nb)
+static inline int
+bfd_same_neighbor(struct bfd_neighbor *x, struct bfd_neighbor *y)
 {
-  struct bfd_proto *p = (struct bfd_proto *) nb->proto;
-  struct bfd_session *s = nb->data;
+  return ipa_equal(x->addr, y->addr) && ipa_equal(x->local, y->local) &&
+    (x->iface == y->iface) && (x->multihop == y->multihop);
+}
 
-  if (!s)
-    return;
+static void
+bfd_reconfigure_neighbors(struct bfd_proto *p, struct bfd_config *new)
+{
+  struct bfd_config *old = (struct bfd_config *) (p->p.cf);
+  struct bfd_neighbor *on, *nn;
 
-  if ((nb->scope > 0) && !s->opened)
-    bfd_open_session(p, s, nb->iface->addr->ip, nb->iface);
+  WALK_LIST(on, old->neigh_list)
+  {
+    WALK_LIST(nn, new->neigh_list)
+      if (bfd_same_neighbor(nn, on))
+      {
+	nn->neigh = on->neigh;
+	if (nn->neigh)
+	  nn->neigh->data = nn;
+
+	nn->req = on->req;
+	nn->active = 1;
+	return;
+      }
+
+    bfd_stop_neighbor(p, on);
+  }
 
-  if ((nb->scope <= 0) && s->opened)
-    bfd_close_session(p, s);
+  WALK_LIST(nn, new->neigh_list)
+    if (!nn->active)
+      bfd_start_neighbor(p, nn);
 }
 
 
+/*
+ *	BFD notify socket
+ */
+
 /* This core notify code should be replaced after main loop transition to birdloop */
 
 int pipe(int pipefd[2]);
@@ -456,6 +847,8 @@ bfd_notify_hook(sock *sk, int len)
   struct bfd_proto *p = sk->data;
   struct bfd_session *s;
   list tmp_list;
+  u8 state, diag;
+  node *n, *nn;
 
   pipe_drain(sk->fd);
 
@@ -469,10 +862,21 @@ bfd_notify_hook(sock *sk, int len)
   {
     bfd_lock_sessions(p);
     rem2_node(&s->n);
+    state = s->loc_state;
+    diag = s->loc_diag;
     bfd_unlock_sessions(p);
 
-    // XXX do something
-    TRACE(D_EVENTS, "Notify: session changed %I %d %d", s->addr, s->loc_state, s->loc_diag);
+    /* FIXME: convert to btime and move to bfd_session_update_state() */ 
+    s->last_state_change = now;
+
+    s->notify_running = 1;
+    WALK_LIST_DELSAFE(n, nn, s->request_list)
+      bfd_request_notify(SKIP_BACK(struct bfd_request, n, n), state, diag);
+    s->notify_running = 0;
+
+    /* Remove the session if all requests were removed in notify hooks */
+    if (EMPTY_LIST(s->request_list))
+      bfd_remove_session(p, s);
   }
 
   return 0;
@@ -523,6 +927,17 @@ bfd_notify_init(struct bfd_proto *p)
 }
 
 
+/*
+ *	BFD protocol glue
+ */
+
+void
+bfd_init_all(void)
+{
+  init_list(&bfd_proto_list);
+  init_list(&bfd_wait_list);
+}
+
 static struct proto *
 bfd_init(struct proto_config *c)
 {
@@ -539,25 +954,28 @@ bfd_start(struct proto *P)
   struct bfd_proto *p = (struct bfd_proto *) P;
   struct bfd_config *cf = (struct bfd_config *) (P->cf);
 
-  p->loop = birdloop_new(P->pool);
+  p->loop = birdloop_new();
   p->tpool = rp_new(NULL, "BFD thread root");
   pthread_spin_init(&p->lock, PTHREAD_PROCESS_PRIVATE);
 
   p->session_slab = sl_new(P->pool, sizeof(struct bfd_session));
-  HASH_INIT(p->session_hash_id, P->pool, 4);
-  HASH_INIT(p->session_hash_ip, P->pool, 4);
-
-  init_list(&p->sock_list);
+  HASH_INIT(p->session_hash_id, P->pool, 8);
+  HASH_INIT(p->session_hash_ip, P->pool, 8);
 
+  init_list(&p->iface_list);
 
   init_list(&p->notify_list);
   bfd_notify_init(p);
 
+  add_tail(&bfd_proto_list, &p->bfd_node);
+
   birdloop_enter(p->loop);
   p->rx_1 = bfd_open_rx_sk(p, 0);
   p->rx_m = bfd_open_rx_sk(p, 1);
   birdloop_leave(p->loop);
 
+  bfd_take_requests(p);
+
   struct bfd_neighbor *n;
   WALK_LIST(n, cf->neigh_list)
     bfd_start_neighbor(p, n);
@@ -572,76 +990,79 @@ static int
 bfd_shutdown(struct proto *P)
 {
   struct bfd_proto *p = (struct bfd_proto *) P;
+  struct bfd_config *cf = (struct bfd_config *) (P->cf);
+
+  rem_node(&p->bfd_node);
 
   birdloop_stop(p->loop);
 
+  struct bfd_neighbor *n;
+  WALK_LIST(n, cf->neigh_list)
+    bfd_stop_neighbor(p, n);
+
+  bfd_drop_requests(p);
+
   /* FIXME: This is hack */
   birdloop_enter(p->loop);
   rfree(p->tpool);
   birdloop_leave(p->loop);
 
-  return PS_DOWN;
-}
-
-static inline int
-bfd_same_neighbor(struct bfd_neighbor *x, struct bfd_neighbor *y)
-{
-  return ipa_equal(x->addr, y->addr) && ipa_equal(x->local, y->local) &&
-    (x->iface == y->iface) && (x->opts->multihop == y->opts->multihop);
-}
-
-static void
-bfd_match_neighbor(struct bfd_proto *p, struct bfd_neighbor *on, struct bfd_config *new)
-{
-  struct bfd_neighbor *nn;
-
-  WALK_LIST(nn, new->neigh_list)
-    if (bfd_same_neighbor(nn, on))
-    {
-      nn->session = on->session;
-      bfd_configure_session(p, nn->session, nn->opts);
-      return;
-    }
+  birdloop_free(p->loop);
 
-  bfd_stop_neighbor(p, on);
+  return PS_DOWN;
 }
 
 static int
 bfd_reconfigure(struct proto *P, struct proto_config *c)
 {
   struct bfd_proto *p = (struct bfd_proto *) P;
-  struct bfd_config *old = (struct bfd_config *) (P->cf);
+  // struct bfd_config *old = (struct bfd_config *) (P->cf);
   struct bfd_config *new = (struct bfd_config *) c;
-  struct bfd_neighbor *n;
+  struct bfd_iface *ifa;
 
   birdloop_mask_wakeups(p->loop);
 
-  WALK_LIST(n, old->neigh_list)
-    bfd_match_neighbor(p, n, new);
+  WALK_LIST(ifa, p->iface_list)
+    bfd_reconfigure_iface(p, ifa, new);
+
+  HASH_WALK(p->session_hash_id, next_id, s)
+  {
+    if (s->ifa->changed)
+      bfd_reconfigure_session(p, s);
+  }
+  HASH_WALK_END;
 
-  WALK_LIST(n, new->neigh_list)
-    if (!n->session)
-      bfd_start_neighbor(p, n);
+  bfd_reconfigure_neighbors(p, new);
 
   birdloop_unmask_wakeups(p->loop);
 
   return 1;
 }
 
+/* Ensure one instance */
+struct bfd_config *bfd_cf;
+
+static void
+bfd_preconfig(struct protocol *P UNUSED, struct config *c UNUSED)
+{
+  bfd_cf = NULL;
+}
+
 static void
 bfd_copy_config(struct proto_config *dest, struct proto_config *src)
 {
   struct bfd_config *d = (struct bfd_config *) dest;
   // struct bfd_config *s = (struct bfd_config *) src;
 
-  /* We clean up neigh_list, ifaces are non-sharable */
+  /* We clean up patt_list and neigh_list, neighbors and ifaces are non-sharable */
+  init_list(&d->patt_list);  
   init_list(&d->neigh_list);
-
 }
 
 void
 bfd_show_sessions(struct proto *P)
 {
+  byte tbuf[TM_DATETIME_BUFFER_SIZE];
   struct bfd_proto *p = (struct bfd_proto *) P;
   uint state, diag;
   u32 tx_int, timeout;
@@ -655,22 +1076,25 @@ bfd_show_sessions(struct proto *P)
   }
 
   cli_msg(-1013, "%s:", p->p.name);
-  cli_msg(-1013, "%-12s\t%s\t%s\t%s\t%s", "Router IP", "Iface",
-	  "State", "TX Int", "Timeout");
+  cli_msg(-1013, "%-25s %-10s %-10s %-10s  %8s %8s",
+	  "IP address", "Interface", "State", "Since", "Interval", "Timeout");
 
-  debug("XXX WALK %d %d\n", p->session_hash_id.count,  p->session_hash_ip.count);
 
   HASH_WALK(p->session_hash_id, next_id, s)
   {
-    // FIXME this is unsafe
+    /* FIXME: this is thread-unsafe, but perhaps harmless */
     state = s->loc_state;
     diag = s->loc_diag;
-    ifname = (s->bsock && s->bsock->sk->iface) ? s->bsock->sk->iface->name : "---";
-    tx_int = (MAX(s->des_min_tx_int, s->rem_min_rx_int) TO_MS);
+    ifname = (s->ifa && s->ifa->sk->iface) ? s->ifa->sk->iface->name : "---";
+    tx_int = s->last_tx ? (MAX(s->des_min_tx_int, s->rem_min_rx_int) TO_MS) : 0;
     timeout = (MAX(s->req_min_rx_int, s->rem_min_tx_int) TO_MS) * s->rem_detect_mult;
 
-    cli_msg(-1013, "%I\t%s\t%s %d\t%u\t%u",
-	    s->addr, ifname, bfd_state_names[state], diag, tx_int, timeout);
+    state = (state < 4) ? state : 0;
+    tm_format_datetime(tbuf, &config->tf_proto, s->last_state_change);
+
+    cli_msg(-1013, "%-25I %-10s %-10s %-10s  %3u.%03u  %3u.%03u",
+	    s->addr, ifname, bfd_state_names[state], tbuf,
+	    tx_int / 1000, tx_int % 1000, timeout / 1000, timeout % 1000);
   }
   HASH_WALK_END;
 
@@ -685,5 +1109,6 @@ struct protocol proto_bfd = {
   .start =		bfd_start,
   .shutdown =		bfd_shutdown,
   .reconfigure =	bfd_reconfigure,
+  .preconfig = 		bfd_preconfig,
   .copy_config =	bfd_copy_config,
 };
diff --git a/proto/bfd/bfd.h b/proto/bfd/bfd.h
index 66c6ed17..f4ab3fcc 100644
--- a/proto/bfd/bfd.h
+++ b/proto/bfd/bfd.h
@@ -20,6 +20,7 @@
 #include "lib/socket.h"
 #include "lib/string.h"
 
+#include "nest/bfd.h"
 #include "io.h"
 
 
@@ -33,19 +34,23 @@
 #define BFD_DEFAULT_MULTIPLIER	5
 
 
+struct bfd_iface_config;
+
 struct bfd_config
 {
   struct proto_config c;
-  list neigh_list;		/* List of struct bfd_neighbor */
+  list patt_list;		/* List of iface configs (struct bfd_iface_config) */
+  list neigh_list;		/* List of configured neighbors (struct bfd_neighbor) */
+  struct bfd_iface_config *multihop; /* Multihop pseudoiface config */
 };
 
-struct bfd_session_config
+struct bfd_iface_config
 {
+  struct iface_patt i;
   u32 min_rx_int;
   u32 min_tx_int;
   u32 idle_tx_int;
   u8 multiplier;
-  u8 multihop;
   u8 passive;
 };
 
@@ -55,9 +60,12 @@ struct bfd_neighbor
   ip_addr addr;
   ip_addr local;
   struct iface *iface;
-  struct bfd_session_config *opts;
 
-  struct bfd_session *session;
+  struct neighbor *neigh;
+  struct bfd_request *req;
+
+  u8 multihop;
+  u8 active;
 };
 
 struct bfd_proto
@@ -66,6 +74,7 @@ struct bfd_proto
   struct birdloop *loop;
   pool *tpool;
   pthread_spinlock_t lock;
+  node bfd_node;
 
   slab *session_slab;
   HASH(struct bfd_session) session_hash_id;
@@ -77,25 +86,31 @@ struct bfd_proto
 
   sock *rx_1;
   sock *rx_m;
-  list sock_list;
+  list iface_list;
 };
 
-struct bfd_socket
+struct bfd_iface
 {
   node n;
+  ip_addr local;
+  struct iface *iface;
+  struct bfd_iface_config *cf;
+  struct bfd_proto *bfd;
+
   sock *sk;
   u32 uc;
+  u8 changed;
 };
 
 struct bfd_session
 {
   node n;
   ip_addr addr;				/* Address of session */
+  struct bfd_iface *ifa;		/* Iface associated with session */
   struct bfd_session *next_id;		/* Next in bfd.session_hash_id */
   struct bfd_session *next_ip;		/* Next in bfd.session_hash_ip */
-  struct bfd_proto *bfd;
 
-  u8 opened;
+  u8 opened_unused;
   u8 passive;
   u8 poll_active;
   u8 poll_scheduled;
@@ -123,7 +138,9 @@ struct bfd_session
   timer2 *tx_timer;			/* Periodic control packet timer */
   timer2 *hold_timer;			/* Timer for session down detection time */
 
-  struct bfd_socket *bsock;		/* Socket associated with session */
+  list request_list;			/* List of client requests (struct bfd_request) */
+  bird_clock_t last_state_change;	/* Time of last state change */
+  u8 notify_running;			/* 1 if notify hooks are running */
 };
 
 
@@ -168,10 +185,7 @@ void bfd_show_sessions(struct proto *P);
 /* packets.c */
 void bfd_send_ctl(struct bfd_proto *p, struct bfd_session *s, int final);
 sock * bfd_open_rx_sk(struct bfd_proto *p, int multihop);
-struct bfd_socket * bfd_get_socket(struct bfd_proto *p, ip_addr local, struct iface *ifa);
-void bfd_free_socket(struct bfd_socket *sk);
-
-
+sock * bfd_open_tx_sk(struct bfd_proto *p, ip_addr local, struct iface *ifa);
 
 
 #endif /* _BIRD_BFD_H_ */
diff --git a/proto/bfd/config.Y b/proto/bfd/config.Y
index f1193d70..1bf8764f 100644
--- a/proto/bfd/config.Y
+++ b/proto/bfd/config.Y
@@ -12,20 +12,21 @@ CF_HDR
 CF_DEFINES
 
 #define BFD_CFG ((struct bfd_config *) this_proto)
-#define BFD_SESSION this_bfd_session
+#define BFD_IFACE ((struct bfd_iface_config *) this_ipatt)
 #define BFD_NEIGHBOR this_bfd_neighbor
 
-static struct bfd_session_config *this_bfd_session;
 static struct bfd_neighbor *this_bfd_neighbor;
 
+extern struct bfd_config *bfd_cf;
 
 CF_DECLS
 
-CF_KEYWORDS(BFD, MIN, IDLE, RX, TX, INTERVAL, MULTIPLIER, MULTIHOP, PASSIVE,
-	NEIGHBOR, DEV)
+CF_KEYWORDS(BFD, MIN, IDLE, RX, TX, INTERVAL, MULTIPLIER, PASSIVE,
+	INTERFACE, MULTIHOP, NEIGHBOR, DEV, LOCAL)
 
 %type <iface> bfd_neigh_iface
 %type <a> bfd_neigh_local
+%type <i> bfd_neigh_multihop
 
 CF_GRAMMAR
 
@@ -34,12 +35,19 @@ CF_ADDTO(proto, bfd_proto)
 bfd_proto_start: proto_start BFD
 {
   this_proto = proto_config_new(&proto_bfd, sizeof(struct bfd_config), $1);
+  init_list(&BFD_CFG->patt_list);
   init_list(&BFD_CFG->neigh_list);
+
+  if (bfd_cf)
+    cf_error("Only one BFD instance allowed");
+  bfd_cf = BFD_CFG;
 };
 
 bfd_proto_item:
    proto_item
- | bfd_neighbor
+ | INTERFACE bfd_iface
+ | MULTIHOP bfd_multihop
+ | NEIGHBOR bfd_neighbor
  ;
 
 bfd_proto_opts:
@@ -51,38 +59,41 @@ bfd_proto:
    bfd_proto_start proto_name '{' bfd_proto_opts '}';
 
 
-bfd_session_start:
+bfd_iface_start:
 {
-  this_bfd_session = cfg_allocz(sizeof(struct bfd_session_config));
+  this_ipatt = cfg_allocz(sizeof(struct bfd_iface_config));
+  init_list(&this_ipatt->ipn_list);
 
-  BFD_SESSION->min_rx_int = BFD_DEFAULT_MIN_RX_INT;
-  BFD_SESSION->min_tx_int = BFD_DEFAULT_MIN_TX_INT;
-  BFD_SESSION->idle_tx_int = BFD_DEFAULT_IDLE_TX_INT;
-  BFD_SESSION->multiplier = BFD_DEFAULT_MULTIPLIER;
+  BFD_IFACE->min_rx_int = BFD_DEFAULT_MIN_RX_INT;
+  BFD_IFACE->min_tx_int = BFD_DEFAULT_MIN_TX_INT;
+  BFD_IFACE->idle_tx_int = BFD_DEFAULT_IDLE_TX_INT;
+  BFD_IFACE->multiplier = BFD_DEFAULT_MULTIPLIER;
 };
 
-bfd_session_item:
-   INTERVAL expr_us { BFD_SESSION->min_rx_int = BFD_SESSION->min_tx_int = $2; }
- | MIN RX INTERVAL expr_us { BFD_SESSION->min_rx_int = $4; }
- | MIN TX INTERVAL expr_us { BFD_SESSION->min_tx_int = $4; }
- | IDLE TX INTERVAL expr_us { BFD_SESSION->idle_tx_int = $4; }
- | MULTIPLIER expr { BFD_SESSION->multiplier = $2; }
- | MULTIHOP bool { BFD_SESSION->multihop = $2; }
- | PASSIVE bool { BFD_SESSION->passive = $2; }
+bfd_iface_item:
+   INTERVAL expr_us { BFD_IFACE->min_rx_int = BFD_IFACE->min_tx_int = $2; }
+ | MIN RX INTERVAL expr_us { BFD_IFACE->min_rx_int = $4; }
+ | MIN TX INTERVAL expr_us { BFD_IFACE->min_tx_int = $4; }
+ | IDLE TX INTERVAL expr_us { BFD_IFACE->idle_tx_int = $4; }
+ | MULTIPLIER expr { BFD_IFACE->multiplier = $2; }
+ | PASSIVE bool { BFD_IFACE->passive = $2; }
  ;
 
-bfd_session_opts:
+bfd_iface_opts:
    /* empty */
- | bfd_session_opts bfd_session_item ';'
+ | bfd_iface_opts bfd_iface_item ';'
  ;
 
-bfd_session_opt_list:
+bfd_iface_opt_list:
    /* empty */
- | '{' bfd_session_opts '}'
+ | '{' bfd_iface_opts '}'
  ;
 
-bfd_session: 
-   bfd_session_start bfd_session_opt_list;
+bfd_iface: bfd_iface_start iface_patt_list bfd_iface_opt_list
+{ add_tail(&BFD_CFG->patt_list, NODE this_ipatt); };
+
+bfd_multihop: bfd_iface_start bfd_iface_opt_list
+{ BFD_CFG->multihop = BFD_IFACE; };
 
 
 bfd_neigh_iface:
@@ -96,15 +107,26 @@ bfd_neigh_local:
  | LOCAL ipa { $$ = $2; }
  ;
 
-bfd_neighbor: NEIGHBOR ipa bfd_neigh_iface bfd_neigh_local bfd_session
+bfd_neigh_multihop:
+   /* empty */ { $$ = 0; }
+ | MULTIHOP bool { $$ = $2; }
+ ;
+
+bfd_neighbor: ipa bfd_neigh_iface bfd_neigh_local bfd_neigh_multihop
 {
   this_bfd_neighbor = cfg_allocz(sizeof(struct bfd_neighbor));
   add_tail(&BFD_CFG->neigh_list, NODE this_bfd_neighbor);
 
-  BFD_NEIGHBOR->addr = $2;
-  BFD_NEIGHBOR->local = $4;
-  BFD_NEIGHBOR->iface = $3;
-  BFD_NEIGHBOR->opts = BFD_SESSION;
+  BFD_NEIGHBOR->addr = $1;
+  BFD_NEIGHBOR->local = $3;
+  BFD_NEIGHBOR->iface = $2;
+  BFD_NEIGHBOR->multihop = $4;
+
+  if ($4 && $2)
+    cf_error("Neighbor cannot set both interface and multihop");
+
+  if ($4 && ipa_zero($3))
+    cf_error("Multihop neighbor requires specified local address");
 };
 
 
diff --git a/proto/bfd/io.c b/proto/bfd/io.c
index 2c1f7b03..fb150040 100644
--- a/proto/bfd/io.c
+++ b/proto/bfd/io.c
@@ -52,6 +52,9 @@ struct birdloop
 };
 
 
+/*
+ *	Current thread context
+ */
 
 static pthread_key_t current_loop_key;
 
@@ -74,6 +77,9 @@ birdloop_init_current(void)
 }
 
 
+/*
+ *	Time clock
+ */
 
 static void times_update_alt(struct birdloop *loop);
 
@@ -163,6 +169,9 @@ current_time(void)
 }
 
 
+/*
+ *	Wakeup code for birdloop
+ */
 
 static void
 pipe_new(int *pfds)
@@ -244,6 +253,9 @@ wakeup_kick(struct birdloop *loop)
 }
 
 
+/*
+ *	Events
+ */
 
 static inline uint
 events_waiting(struct birdloop *loop)
@@ -279,12 +291,14 @@ ev2_schedule(event *e)
 }
 
 
+/*
+ *	Timers
+ */
 
 #define TIMER_LESS(a,b)		((a)->expires < (b)->expires)
 #define TIMER_SWAP(heap,a,b,t)	(t = heap[a], heap[a] = heap[b], heap[b] = t, \
 				   heap[a]->index = (a), heap[b]->index = (b))
 
-
 static inline uint timers_count(struct birdloop *loop)
 { return loop->timers.used - 1; }
 
@@ -425,6 +439,9 @@ timers_fire(struct birdloop *loop)
 }
 
 
+/*
+ *	Sockets
+ */
 
 static void
 sockets_init(struct birdloop *loop)
@@ -494,12 +511,16 @@ sk_stop(sock *s)
 static inline uint sk_want_events(sock *s)
 { return (s->rx_hook ? POLLIN : 0) | ((s->ttx != s->tpos) ? POLLOUT : 0); }
 
+/*
+FIXME: this should be called from sock code
+
 static void
 sockets_update(struct birdloop *loop, sock *s)
 {
   if (s->index >= 0)
     loop->poll_fd.data[s->index].events = sk_want_events(s);
 }
+*/
 
 static void
 sockets_prepare(struct birdloop *loop)
@@ -594,17 +615,21 @@ sockets_fire(struct birdloop *loop)
 }
 
 
+/*
+ *	Birdloop
+ */
 
 static void * birdloop_main(void *arg);
 
 struct birdloop *
-birdloop_new(pool *p)
+birdloop_new(void)
 {
   /* FIXME: this init should be elsewhere and thread-safe */
   static int init = 0;
   if (!init)
     { birdloop_init_current(); init = 1; }
 
+  pool *p = rp_new(NULL, "Birdloop root");
   struct birdloop *loop = mb_allocz(p, sizeof(struct birdloop));
   loop->pool = p;
   pthread_mutex_init(&loop->mutex, NULL);
@@ -640,6 +665,12 @@ birdloop_stop(struct birdloop *loop)
     die("pthread_join(): %M", rv);
 }
 
+void
+birdloop_free(struct birdloop *loop)
+{
+  rfree(loop->pool);
+}
+
 
 void
 birdloop_enter(struct birdloop *loop)
@@ -735,4 +766,3 @@ birdloop_main(void *arg)
 }
 
 
-
diff --git a/proto/bfd/io.h b/proto/bfd/io.h
index 4f7c678d..3f166a47 100644
--- a/proto/bfd/io.h
+++ b/proto/bfd/io.h
@@ -63,6 +63,13 @@ tm2_new_init(pool *p, void (*hook)(struct timer2 *), void *data, uint rec, uint
   return t;
 }
 
+static inline void
+tm2_set_max(timer2 *t, btime when)
+{
+  if (when > t->expires)
+    tm2_set(t, when);
+}
+
 /*
 static inline void
 tm2_start_max(timer2 *t, btime after)
@@ -78,9 +85,11 @@ void sk_stop(sock *s);
 
 
 
-struct birdloop *birdloop_new(pool *p);
+struct birdloop *birdloop_new(void);
 void birdloop_start(struct birdloop *loop);
 void birdloop_stop(struct birdloop *loop);
+void birdloop_free(struct birdloop *loop);
+
 void birdloop_enter(struct birdloop *loop);
 void birdloop_leave(struct birdloop *loop);
 void birdloop_mask_wakeups(struct birdloop *loop);
diff --git a/proto/bfd/packets.c b/proto/bfd/packets.c
index 0e24114b..fc2616ca 100644
--- a/proto/bfd/packets.c
+++ b/proto/bfd/packets.c
@@ -62,7 +62,7 @@ bfd_format_flags(u8 flags, char *buf)
 void
 bfd_send_ctl(struct bfd_proto *p, struct bfd_session *s, int final)
 {
-  sock *sk = s->bsock->sk;
+  sock *sk = s->ifa->sk;
   struct bfd_ctl_packet *pkt = (struct bfd_ctl_packet *) sk->tbuf;
   char fb[8];
 
@@ -143,7 +143,7 @@ bfd_rx_hook(sock *sk, int len)
     s = bfd_find_session_by_addr(p, sk->faddr);
 
     /* FIXME: better session matching and message */
-    if (!s || !s->opened)
+    if (!s)
       return 1;
   }
 
@@ -155,7 +155,7 @@ bfd_rx_hook(sock *sk, int len)
   u32 old_tx_int = s->des_min_tx_int;
   u32 old_rx_int = s->rem_min_rx_int;
 
-  s->rem_id = ntohl(pkt->snd_id);
+  s->rem_id= ntohl(pkt->snd_id);
   s->rem_state = bfd_pkt_get_state(pkt);
   s->rem_diag = bfd_pkt_get_diag(pkt);
   s->rem_demand_mode = pkt->flags & BFD_FLAG_DEMAND;
@@ -213,7 +213,7 @@ bfd_open_rx_sk(struct bfd_proto *p, int multihop)
   return NULL;
 }
 
-static inline sock *
+sock *
 bfd_open_tx_sk(struct bfd_proto *p, ip_addr local, struct iface *ifa)
 {
   sock *sk = sk_new(p->tpool);
@@ -246,32 +246,3 @@ bfd_open_tx_sk(struct bfd_proto *p, ip_addr local, struct iface *ifa)
   rfree(sk);
   return NULL;
 }
-
-struct bfd_socket *
-bfd_get_socket(struct bfd_proto *p, ip_addr local, struct iface *ifa)
-{
-  struct bfd_socket *sk;
-
-  WALK_LIST(sk, p->sock_list)
-    if (ipa_equal(sk->sk->saddr, local) && (sk->sk->iface == ifa))
-      return sk->uc++, sk;
-
-  sk = mb_allocz(p->tpool, sizeof(struct bfd_socket));
-  sk->sk = bfd_open_tx_sk(p, local, ifa);
-  sk->uc = 1;
-  add_tail(&p->sock_list, &sk->n);
-
-  return sk;
-}
-
-void
-bfd_free_socket(struct bfd_socket *sk)
-{
-  if (!sk || --sk->uc)
-    return;
-
-  rem_node(&sk->n);
-  sk_stop(sk->sk);
-  rfree(sk->sk);
-  mb_free(sk);
-}
diff --git a/proto/bgp/bgp.c b/proto/bgp/bgp.c
index 7cad75df..07ad31f3 100644
--- a/proto/bgp/bgp.c
+++ b/proto/bgp/bgp.c
@@ -59,8 +59,9 @@
 #include "nest/iface.h"
 #include "nest/protocol.h"
 #include "nest/route.h"
-#include "nest/locks.h"
+#include "nest/bfd.h"
 #include "nest/cli.h"
+#include "nest/locks.h"
 #include "conf/conf.h"
 #include "lib/socket.h"
 #include "lib/resource.h"
@@ -76,6 +77,7 @@ static void bgp_close(struct bgp_proto *p, int apply_md5);
 static void bgp_connect(struct bgp_proto *p);
 static void bgp_active(struct bgp_proto *p);
 static sock *bgp_setup_listen_sk(ip_addr addr, unsigned port, u32 flags);
+static void bgp_update_bfd(struct bgp_proto *p, int use_bfd);
 
 
 /**
@@ -153,8 +155,12 @@ bgp_initiate(struct bgp_proto *p)
   if (rv < 0)
     return;
 
+  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", p->startup_delay);
       bgp_start_timer(p->startup_timer, p->startup_delay);
     }
@@ -765,6 +771,37 @@ bgp_neigh_notify(neighbor *n)
     }
 }
 
+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 int
 bgp_reload_routes(struct proto *P)
 {
@@ -825,6 +862,7 @@ bgp_start(struct proto *P)
   p->outgoing_conn.state = BS_IDLE;
   p->incoming_conn.state = BS_IDLE;
   p->neigh = NULL;
+  p->bfd_req = NULL;
 
   rt_lock_table(p->igp_table);
 
@@ -992,6 +1030,9 @@ bgp_check_config(struct bgp_config *c)
 		      ipa_has_link_scope(c->source_addr)))
     cf_error("Multihop BGP cannot be used with link-local addresses");
 
+  if (c->multihop && c->bfd && ipa_zero(c->source_addr))
+    cf_error("Multihop BGP with BFD requires specified source address");
+
 
   /* Different default based on rs_client */
   if (!c->missing_lladdr)
@@ -1034,6 +1075,9 @@ bgp_reconfigure(struct proto *P, struct proto_config *C)
 	|| (old->password && new->password && !strcmp(old->password, new->password)))
     && (get_igp_table(old) == get_igp_table(new));
 
+  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;
@@ -1115,7 +1159,7 @@ bgp_store_error(struct bgp_proto *p, struct bgp_conn *c, u8 class, u32 code)
 
 static char *bgp_state_names[] = { "Idle", "Connect", "Active", "OpenSent", "OpenConfirm", "Established", "Close" };
 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" };
+static char *bgp_misc_errors[] = { "", "Neighbor lost", "Invalid next hop", "Kernel MD5 auth failed", "No listening socket", "BFD session down" };
 static char *bgp_auto_errors[] = { "", "Route limit exceeded"};
 
 static const char *
diff --git a/proto/bgp/bgp.h b/proto/bgp/bgp.h
index 77a36715..6da38949 100644
--- a/proto/bgp/bgp.h
+++ b/proto/bgp/bgp.h
@@ -14,6 +14,7 @@
 
 struct linpool;
 struct eattr;
+struct bfd_request;
 
 struct bgp_config {
   struct proto_config c;
@@ -52,8 +53,10 @@ struct bgp_config {
   unsigned error_delay_time_min;	/* Time to wait after an error is detected */
   unsigned error_delay_time_max;
   unsigned disable_after_error;		/* Disable the protocol when error is detected */
+
   char *password;			/* Password used for MD5 authentication */
   struct rtable_config *igp_table;	/* Table used for recursive next hop lookups */
+  int bfd;				/* Use BFD for liveness detection */
 };
 
 #define MLL_SELF 1
@@ -99,6 +102,7 @@ struct bgp_proto {
   struct bgp_conn incoming_conn;	/* Incoming connection we have neither accepted nor rejected yet */
   struct object_lock *lock;		/* Lock for neighbor connection */
   struct neighbor *neigh;		/* Neighbor entry corresponding to remote ip, NULL if multihop */
+  struct bfd_request *bfd_req;		/* BFD request, if BFD is used */
   ip_addr source_addr;			/* Local address used as an advertised next hop */
   rtable *igp_table;			/* Table used for recursive next hop lookups */
   struct event *event;			/* Event for respawning and shutting process */
@@ -287,6 +291,7 @@ void bgp_log_error(struct bgp_proto *p, u8 class, char *msg, unsigned code, unsi
 #define BEM_INVALID_NEXT_HOP	2
 #define BEM_INVALID_MD5		3	/* MD5 authentication kernel request failed (possibly not supported) */
 #define BEM_NO_SOCKET		4
+#define BEM_BFD_DOWN		5
 
 /* Automatic shutdown error codes */
 
diff --git a/proto/bgp/config.Y b/proto/bgp/config.Y
index d5e5aaca..0292c234 100644
--- a/proto/bgp/config.Y
+++ b/proto/bgp/config.Y
@@ -26,7 +26,7 @@ CF_KEYWORDS(BGP, LOCAL, NEIGHBOR, AS, HOLD, TIME, CONNECT, RETRY,
 	PREFER, OLDER, MISSING, LLADDR, DROP, IGNORE, ROUTE, REFRESH,
 	INTERPRET, COMMUNITIES, BGP_ORIGINATOR_ID, BGP_CLUSTER_LIST, IGP,
 	TABLE, GATEWAY, DIRECT, RECURSIVE, MED, TTL, SECURITY, DETERMINISTIC,
-	SECONDARY)
+	SECONDARY, BFD)
 
 CF_GRAMMAR
 
@@ -110,6 +110,7 @@ bgp_proto:
  | bgp_proto SECONDARY bool ';' { BGP_CFG->secondary = $3; }
  | bgp_proto IGP TABLE rtable ';' { BGP_CFG->igp_table = $4; }
  | bgp_proto TTL SECURITY bool ';' { BGP_CFG->ttl_security = $4; }
+ | bgp_proto BFD bool ';' { BGP_CFG->bfd = $3; cf_check_bfd($3); }
  ;
 
 CF_ADDTO(dynamic_attr, BGP_ORIGIN
diff --git a/proto/ospf/config.Y b/proto/ospf/config.Y
index f042e1aa..f06dd311 100644
--- a/proto/ospf/config.Y
+++ b/proto/ospf/config.Y
@@ -309,6 +309,7 @@ ospf_iface_item:
  | TX PRIORITY expr { OSPF_PATT->tx_priority = $3; }
  | TTL SECURITY bool { OSPF_PATT->ttl_security = $3; }
  | TTL SECURITY TX ONLY { OSPF_PATT->ttl_security = 2; }
+ | BFD bool { OSPF_PATT->bfd = $2; cf_check_bfd($2); }
  | password_list
  ;
 
diff --git a/proto/ospf/hello.c b/proto/ospf/hello.c
index d5aa1b95..bac2a589 100644
--- a/proto/ospf/hello.c
+++ b/proto/ospf/hello.c
@@ -140,6 +140,9 @@ ospf_hello_receive(struct ospf_packet *ps_i, struct ospf_iface *ifa,
 #ifdef OSPFv3
     n->iface_id = ntohl(ps->iface_id);
 #endif
+
+    if (n->ifa->cf->bfd)
+      ospf_neigh_update_bfd(n, n->ifa->bfd);
   }
   ospf_neigh_sm(n, INM_HELLOREC);
 
diff --git a/proto/ospf/iface.c b/proto/ospf/iface.c
index 63c26466..f1409840 100644
--- a/proto/ospf/iface.c
+++ b/proto/ospf/iface.c
@@ -536,6 +536,7 @@ ospf_iface_new(struct ospf_area *oa, struct ifa *addr, struct ospf_iface_patt *i
   ifa->check_link = ip->check_link;
   ifa->ecmp_weight = ip->ecmp_weight;
   ifa->check_ttl = (ip->ttl_security == 1);
+  ifa->bfd = ip->bfd;
 
 #ifdef OSPFv2
   ifa->autype = ip->autype;
@@ -840,6 +841,19 @@ ospf_iface_reconfigure(struct ospf_iface *ifa, struct ospf_iface_patt *new)
     ifa->ecmp_weight = new->ecmp_weight;
   }
 
+  /* BFD */
+  if (ifa->bfd != new->bfd)
+  {
+    OSPF_TRACE(D_EVENTS, "%s BFD on interface %s",
+	       new->bfd ? "Enabling" : "Disabling", ifname);
+    ifa->bfd = new->bfd;
+
+    struct ospf_neighbor *n;
+    WALK_LIST(n, ifa->neigh_list)
+      ospf_neigh_update_bfd(n, ifa->bfd);
+  }
+
+
   /* instance_id is not updated - it is part of key */
 
   return 1;
diff --git a/proto/ospf/neighbor.c b/proto/ospf/neighbor.c
index 26d81dce..61224ec2 100644
--- a/proto/ospf/neighbor.c
+++ b/proto/ospf/neighbor.c
@@ -582,6 +582,36 @@ ospf_neigh_remove(struct ospf_neighbor *n)
   OSPF_TRACE(D_EVENTS, "Deleting neigbor.");
 }
 
+static void
+ospf_neigh_bfd_hook(struct bfd_request *req)
+{
+  struct ospf_neighbor *n = req->data;
+  struct proto *p = &n->ifa->oa->po->proto;
+
+  if (req->down)
+  {
+    OSPF_TRACE(D_EVENTS, "BFD session down for %I on %s",
+	       n->ip, n->ifa->iface->name);
+
+    ospf_neigh_remove(n);
+  }
+}
+
+void
+ospf_neigh_update_bfd(struct ospf_neighbor *n, int use_bfd)
+{
+  if (use_bfd && !n->bfd_req)
+    n->bfd_req = bfd_request_session(n->pool, n->ip, n->ifa->addr->ip, n->ifa->iface,
+				     ospf_neigh_bfd_hook, n);
+
+  if (!use_bfd && n->bfd_req)
+  {
+    rfree(n->bfd_req);
+    n->bfd_req = NULL;
+  }
+}
+
+
 void
 ospf_sh_neigh_info(struct ospf_neighbor *n)
 {
diff --git a/proto/ospf/neighbor.h b/proto/ospf/neighbor.h
index f593faed..e674927d 100644
--- a/proto/ospf/neighbor.h
+++ b/proto/ospf/neighbor.h
@@ -16,6 +16,7 @@ void bdr_election(struct ospf_iface *ifa);
 struct ospf_neighbor *find_neigh(struct ospf_iface *ifa, u32 rid);
 struct ospf_neighbor *find_neigh_by_ip(struct ospf_iface *ifa, ip_addr ip);
 void ospf_neigh_remove(struct ospf_neighbor *n);
+void ospf_neigh_update_bfd(struct ospf_neighbor *n, int use_bfd);
 void ospf_sh_neigh_info(struct ospf_neighbor *n);
 
 #endif /* _BIRD_OSPF_NEIGHBOR_H_ */
diff --git a/proto/ospf/ospf.h b/proto/ospf/ospf.h
index f1409af3..46a1c3c1 100644
--- a/proto/ospf/ospf.h
+++ b/proto/ospf/ospf.h
@@ -46,6 +46,7 @@ do { if ((p->debug & D_PACKETS) || OSPF_FORCE_DEBUG) \
 #include "nest/route.h"
 #include "nest/cli.h"
 #include "nest/locks.h"
+#include "nest/bfd.h"
 #include "conf/conf.h"
 #include "lib/string.h"
 
@@ -276,6 +277,7 @@ struct ospf_iface
   u8 ecmp_weight;		/* Weight used for ECMP */
   u8 ptp_netmask;		/* Send real netmask for P2P */
   u8 check_ttl;			/* Check incoming packets for TTL 255 */
+  u8 bfd;			/* Use BFD on iface */
 };
 
 struct ospf_md5
@@ -708,6 +710,7 @@ struct ospf_neighbor
 #define ACKL_DIRECT 0
 #define ACKL_DELAY 1
   timer *ackd_timer;		/* Delayed ack timer */
+  struct bfd_request *bfd_req;	/* BFD request, if BFD is used */
   u32 csn;                      /* Last received crypt seq number (for MD5) */
 };
 
@@ -818,6 +821,7 @@ struct ospf_iface_patt
   u8 real_bcast;		/* Not really used in OSPFv3 */
   u8 ptp_netmask;		/* bool + 2 for unspecified */
   u8 ttl_security;		/* bool + 2 for TX only */
+  u8 bfd;
 
 #ifdef OSPFv2
   list *passwords;
diff --git a/proto/radv/radv.c b/proto/radv/radv.c
index a6b9b16c..90408536 100644
--- a/proto/radv/radv.c
+++ b/proto/radv/radv.c
@@ -15,7 +15,7 @@
  * The RAdv protocol is implemented in two files: |radv.c| containing
  * the interface with BIRD core and the protocol logic and |packets.c|
  * handling low level protocol stuff (RX, TX and packet formats).
- * The protocol does not import or export any routes.
+ * The protocol does not export any routes.
  *
  * The RAdv is structured in the usual way - for each handled interface
  * there is a structure &radv_iface that contains a state related to
diff --git a/sysdep/autoconf.h.in b/sysdep/autoconf.h.in
index ac6f7a87..a9e46e27 100644
--- a/sysdep/autoconf.h.in
+++ b/sysdep/autoconf.h.in
@@ -39,10 +39,14 @@
 #undef CONFIG_STATIC
 #undef CONFIG_RIP
 #undef CONFIG_RADV
+#undef CONFIG_BFD
 #undef CONFIG_BGP
 #undef CONFIG_OSPF
 #undef CONFIG_PIPE
 
+/* We use multithreading */
+#undef USE_PTHREADS
+
 /* We have <syslog.h> and syslog() */
 #undef HAVE_SYSLOG
 
diff --git a/sysdep/unix/io.c b/sysdep/unix/io.c
index f0ec6dae..da8343f9 100644
--- a/sysdep/unix/io.c
+++ b/sysdep/unix/io.c
@@ -538,6 +538,8 @@ sk_free(resource *r)
   if (s->fd >= 0)
     {
       close(s->fd);
+
+      /* FIXME: we should call sk_stop() for SKF_THREAD sockets */
       if (s->flags & SKF_THREAD)
 	return;
 
diff --git a/sysdep/unix/log.c b/sysdep/unix/log.c
index 67b70773..9dd4d66f 100644
--- a/sysdep/unix/log.c
+++ b/sysdep/unix/log.c
@@ -36,13 +36,20 @@ static const bird_clock_t rate_limit_time = 5;
 static const int rate_limit_count = 5;
 
 
-// XXX add ifdef for threads
+#ifdef USE_PTHREADS
 
 #include <pthread.h>
 static pthread_mutex_t log_mutex;
 static inline void log_lock(void) { pthread_mutex_lock(&log_mutex); }
 static inline void log_unlock(void) { pthread_mutex_unlock(&log_mutex); }
 
+#else
+
+static inline void log_lock(void) {  }
+static inline void log_unlock(void) {  }
+
+#endif
+
 
 #ifdef HAVE_SYSLOG
 #include <sys/syslog.h>
-- 
cgit v1.2.3