Merge branch 'int-new' into bash-completion

.gitignore

@@ -11,3 +11,4 @@
 /config.status
 /configure
 /sysdep/autoconf.h.in
+/sysdep/autoconf.h.in~

.gitlab-ci.yml

@@ -0,0 +1,315 @@
+variables:
+  DEBIAN_FRONTEND: noninteractive
+  LC_ALL: C
+  GIT_STRATEGY: fetch
+  DOCKER_CMD: docker --config="$HOME/.docker/$CI_JOB_ID/"
+  IMG_BASE: registry.labs.nic.cz/labs/bird
+
+stages:
+  - image
+  - build
+
+.docker: &docker_build
+  stage: image
+  allow_failure: true
+  script:
+  - $DOCKER_CMD login -u gitlab-ci-token -p $CI_BUILD_TOKEN registry.labs.nic.cz
+  # Make sure we refresh the base image if it updates (eg. security updates, etc)
+  # If we do just the build, cache is always reused and the freshness of the
+  # base image is never checked. However, pull always asks and updates the
+  # image only if it changed ‒ therefore, the cache is used unless there's a
+  # change.
+  - $DOCKER_CMD pull `sed -ne 's/^FROM //p' "misc/docker/$IMG_NAME/Dockerfile"`
+  - $DOCKER_CMD build -t "bird:$IMG_NAME" "misc/docker/$IMG_NAME"
+  - $DOCKER_CMD tag "bird:$IMG_NAME" "$IMG_BASE:$IMG_NAME"
+  - $DOCKER_CMD push "$IMG_BASE:$IMG_NAME"
+  after_script:
+  - rm -f "$HOME/.docker/$CI_JOB_ID/" # cleanup the credentials
+  tags:
+  # That's Docker in Docker
+  - dind
+
+docker_debian-7-amd64:
+  variables:
+    IMG_NAME: "debian-7-amd64"
+  <<: *docker_build
+
+docker_debian-8-amd64:
+  variables:
+    IMG_NAME: "debian-8-amd64"
+  <<: *docker_build
+
+docker_debian-9-amd64:
+  variables:
+    IMG_NAME: "debian-9-amd64"
+  <<: *docker_build
+
+docker_debian-testing-amd64:
+  variables:
+    IMG_NAME: "debian-testing-amd64"
+  <<: *docker_build
+
+docker_debian-7-i386:
+  variables:
+    IMG_NAME: "debian-7-i386"
+  <<: *docker_build
+
+docker_debian-8-i386:
+  variables:
+    IMG_NAME: "debian-8-i386"
+  <<: *docker_build
+
+docker_debian-9-i386:
+  variables:
+    IMG_NAME: "debian-9-i386"
+  <<: *docker_build
+
+docker_debian-testing-i386:
+  variables:
+    IMG_NAME: "debian-testing-i386"
+  <<: *docker_build
+
+docker_fedora-25-amd64:
+  variables:
+    IMG_NAME: "fedora-25-amd64"
+  <<: *docker_build
+
+docker_fedora-26-amd64:
+  variables:
+    IMG_NAME: "fedora-26-amd64"
+  <<: *docker_build
+
+docker_centos-6-amd64:
+  variables:
+    IMG_NAME: "centos-6-amd64"
+  <<: *docker_build
+
+docker_centos-7-amd64:
+  variables:
+    IMG_NAME: "centos-7-amd64"
+  <<: *docker_build
+
+docker_opensuse-42_3-amd64:
+  variables:
+    IMG_NAME: "opensuse-42.3-amd64"
+  <<: *docker_build
+
+docker_ubuntu-14_04-amd64:
+  variables:
+    IMG_NAME: "ubuntu-14.04-amd64"
+  <<: *docker_build
+
+docker_ubuntu-16_04-amd64:
+  variables:
+    IMG_NAME: "ubuntu-16.04-amd64"
+  <<: *docker_build
+
+.debian-7-i386: &debian-7-i386_env
+  image: registry.labs.nic.cz/labs/bird:debian-7-i386
+  tags:
+  - docker
+  - linux
+  - amd64
+
+.debian-8-i386: &debian-8-i386_env
+  image: registry.labs.nic.cz/labs/bird:debian-8-i386
+  tags:
+  - docker
+  - linux
+  - amd64
+
+.debian-9-i386: &debian-9-i386_env
+  image: registry.labs.nic.cz/labs/bird:debian-9-i386
+  tags:
+  - docker
+  - linux
+  - amd64
+
+.debian-testing-i386: &debian-testing-i386_env
+  image: registry.labs.nic.cz/labs/bird:debian-testing-i386
+  tags:
+  - docker
+  - linux
+  - amd64
+
+.debian-7-amd64: &debian-7-amd64_env
+  image: registry.labs.nic.cz/labs/bird:debian-7-amd64
+  tags:
+  - docker
+  - linux
+  - amd64
+
+.debian-8-amd64: &debian-8-amd64_env
+  image: registry.labs.nic.cz/labs/bird:debian-8-amd64
+  tags:
+  - docker
+  - linux
+  - amd64
+
+.debian-9-amd64: &debian-9-amd64_env
+  image: registry.labs.nic.cz/labs/bird:debian-9-amd64
+  tags:
+  - docker
+  - linux
+  - amd64
+
+.debian-testing-amd64: &debian-testing-amd64_env
+  image: registry.labs.nic.cz/labs/bird:debian-testing-amd64
+  tags:
+  - docker
+  - linux
+  - amd64
+
+.fedora-25-amd64: &fedora-25-amd64_env
+  image: registry.labs.nic.cz/labs/bird:fedora-25-amd64
+  tags:
+  - docker
+  - linux
+  - amd64
+
+.fedora-26-amd64: &fedora-26-amd64_env
+  image: registry.labs.nic.cz/labs/bird:fedora-26-amd64
+  tags:
+  - docker
+  - linux
+  - amd64
+
+.centos-6-amd64: &centos-6-amd64_env
+  image: registry.labs.nic.cz/labs/bird:centos-6-amd64
+  tags:
+  - docker
+  - linux
+  - amd64
+
+.centos-7-amd64: &centos-7-amd64_env
+  image: registry.labs.nic.cz/labs/bird:centos-7-amd64
+  tags:
+  - docker
+  - linux
+  - amd64
+
+.opensuse-42_3-amd64: &opensuse-42_3-amd64_env
+  image: registry.labs.nic.cz/labs/bird:opensuse-42.3-amd64
+  tags:
+  - docker
+  - linux
+  - amd64
+
+.ubuntu-14_04-amd64: &ubuntu-14_04-amd64_env
+  image: registry.labs.nic.cz/labs/bird:ubuntu-14.04-amd64
+  tags:
+  - docker
+  - linux
+  - amd64
+
+.ubuntu-16_04-amd64: &ubuntu-16_04-amd64_env
+  image: registry.labs.nic.cz/labs/bird:ubuntu-16.04-amd64
+  tags:
+  - docker
+  - linux
+  - amd64
+
+# TODO We want to copy these BSDs to our own virtual machines, to make sure someone doesn't update them by accident.
+.freebsd-11-i386: &freebsd-11-i386_env
+  variables:
+    CPPFLAGS: "-I/usr/local/include"
+    LDFLAGS: "-L/usr/local/lib"
+  tags:
+  - freebsd
+  - i386
+  #only:
+  #- master
+  #- triggers
+  #- tags
+
+.freebsd-11-amd64: &freebsd-11-amd64_env
+  variables:
+    CPPFLAGS: "-I/usr/local/include"
+    LDFLAGS: "-L/usr/local/lib"
+  tags:
+  - freebsd
+  - amd64
+  #only:
+  #- master
+  #- triggers
+  #- tags
+
+.build: &build_job
+  stage: build
+  script:
+  - autoreconf
+  - ./configure CPPFLAGS="$CPPFLAGS" LDFLAGS="$LDFLAGS"
+  # Detect which make is available
+  - MAKE=make
+  - which gmake 2>/dev/null >/dev/null && MAKE=gmake
+  - $MAKE
+  # Run tests if they are available
+  - $MAKE check
+
+build-debian-7-amd64:
+  <<: *debian-7-amd64_env
+  <<: *build_job
+
+build-debian-8-amd64:
+  <<: *debian-8-amd64_env
+  <<: *build_job
+
+build-debian-9-amd64:
+  <<: *debian-9-amd64_env
+  <<: *build_job
+
+build-debian-testing-amd64:
+  <<: *debian-testing-amd64_env
+  <<: *build_job
+
+build-fedora-25-amd64:
+  <<: *fedora-25-amd64_env
+  <<: *build_job
+
+build-fedora-26-amd64:
+  <<: *fedora-26-amd64_env
+  <<: *build_job
+
+build-centos-6-amd64:
+  <<: *centos-6-amd64_env
+  <<: *build_job
+
+build-centos-7-amd64:
+  <<: *centos-7-amd64_env
+  <<: *build_job
+
+build-opensuse-42_3-amd64:
+  <<: *opensuse-42_3-amd64_env
+  <<: *build_job
+
+build-ubuntu-14_04-amd64:
+  <<: *ubuntu-14_04-amd64_env
+  <<: *build_job
+
+build-ubuntu-16_04-amd64:
+  <<: *ubuntu-16_04-amd64_env
+  <<: *build_job
+
+build-debian-7-i386:
+  <<: *debian-7-i386_env
+  <<: *build_job
+
+build-debian-8-i386:
+  <<: *debian-8-i386_env
+  <<: *build_job
+
+build-debian-9-i386:
+  <<: *debian-9-i386_env
+  <<: *build_job
+
+build-debian-testing-i386:
+  <<: *debian-testing-i386_env
+  <<: *build_job
+
+build-freebsd-11-amd64:
+  <<: *freebsd-11-amd64_env
+  <<: *build_job
+
+build-freebsd-11-i386:
+  <<: *freebsd-11-i386_env
+  <<: *build_job

NEWS

@@ -1,45 +1,46 @@
-Version 2.0.0-pre1 (2017-04-29)
+Version 2.0.0 (2017-12-11)
+  o Integrated IPv4 + IPv6 design
   o Support for MPLS next hops
-  o VPNv4 and VPNv6 network types
+  o Support for VPNv4 and VPNv6 networks
+  o Microsecond timers infrastructure
+  o Basic VRF support
+  o Babel: Support for dual-stack IPv4/IPv6
+  o Babel: Many improvements and bugfixes
+  o Major BGP protocol redesign
+  o Full support for Multiprotocol BGP
+  o BGP multicast support (SAFI 2)
+  o BGP flowspec support (RFC 5575)
   o BGP with MPLS labels (RFC 3107)
   o BGP MPLS/VPN support (RFC 4364)
   o BGP 6PE - IPv6 NLRI over IPv4 MPLS (RFC 4798)
   o BGP IPv4 NLRI with an IPv6 Next Hop (RFC 5549)
   o BGP Confederations (RFC 5065)
-  o BGP: Simplify igp table options
+  o BGP Shutdown communication (RFC 8203)
   o BGP: Allow exchanging LOCAL_PREF with eBGP peers
   o BGP: Allow to specify interface for regular sessions
-  o Babel support restored
+  o OSPF: Support of address families in OSPFv3
-  o Static: Minor overhaul
+  o OSPF: Enable ECMP and Link detection by default
-  o Netlink: Default kernel metric changed to 32
+  o RAdv: Support for more specific routes (RFC 4191)
-  o Flowspec: Limit tcp mask length to 12 bits
+  o RAdv: Proper handling of prefix retraction
-  o Update of show route command
+  o RIP: Enable ECMP and Link detection by default
-
+  o Redesign of RPKI handling
-  Notes:
-
-  Definitions of OSPFv2, OSPFv3 and RIP NG protocols now use keywords
-  'ospf v2', 'ospf v3' and 'rip ng' instead of 'ospf2', 'ospf3' and 'ripng'.
-
-  Flows and ROAs no longer use phony next hops, so there is no need to use
-  'drop' or 'unreachable' in their static route definitions.
-
-  See doc/bird.conf.example2 for configuration examples.
-
-
-Version 2.0.0-pre0 (2016-12-07)
-  o Integrated IPv4 + IPv6 design
-  o Major BGP protocol redesign
-  o BGP multicast support (SAFI 2)
-  o BGP flowspec support (RFC 5575)
   o New RPKI-Router protocol
+  o Static: Minor overhaul
+  o Static: Support for all new route types
+  o Kenrel: Default Linux kernel metric changed to 32
+  o Kernel: Fix IPv6 ECMP handling with Linux 4.11+
+  o Update of show route command
+  o BIRD client persistent history
   o New build system
   o Unit tests
+  o ...
 
   Notes:
 
-  Protocols and tables are now connected using explicit channels, most related
+  Tables are now defined with appropriate net type keyword. Protocols and tables
-  protocol options (table, import, export, ...) are now channel options. See
+  are now connected by explicit channels, most related protocol options (table,
-  doc/bird.conf.example2 for configuration examples.
+  import, export, ...) are now channel options. See doc/bird.conf.example2 for
+  configuration examples. Some options were removed/replaced.
 
 
 Version 1.6.3 (2016-12-21)

README

@@ -6,7 +6,7 @@
                 (c) 1998--2008  Martin Mares <mj@ucw.cz>
                 (c) 1998--2000  Pavel Machek <pavel@ucw.cz>
                 (c) 1998--2008  Ondrej Filip <feela@network.cz>
-                (c) 2009--2016  CZ.NIC z.s.p.o.
+                (c) 2009--2017  CZ.NIC z.s.p.o.
 
 ================================================================================
 
@@ -19,7 +19,7 @@ Public License.
 What do we support
 ==================
 
-	o  Both IPv4 and IPv6 (use --enable-ipv6 when configuring)
+	o  Both IPv4 and IPv6
 	o  Multiple routing tables
 	o  Border Gateway Protocol (BGPv4)
 	o  Routing Information Protocol (RIPv2, RIPng)

TODO

@@ -1,45 +0,0 @@
-Core
-~~~~
-- socket open failure should not be fatal
-- &&,||: priorities
-- static: allow specifying a per-route filter program for setting route attributes?
-
-Globals
-~~~~~~~
-- right usage of DBG vs. debug
-- logging and tracing; use appropriate log levels
-- check incoming packets and log errors!!
-- check log calls for trailing newlines and log levels followed by comma
-- check if all protocols set proper packet priorities and TTL's.
-- try compiling with -Wunused
-- does everybody test return value of sk_open?
-- protocols: implement CLI hooks and per-procotol CLI commands
-- protocols: implement reconfigure hook
-- protocols: use locking
-- check use of system includes and sprintf()
-
-Various ideas
-~~~~~~~~~~~~~
-- client: Ctrl-R eats one more enter
-- bgp: timing of updates?
-- netlink: import Linux route attributes to our rta's, so that they can be filtered?
-- config: executable config files
-- filters: user defined attributes?
-- io: use poll if available
-- route recalculation timing and flap dampening [see RFC2439 for algorithms]
-- aggregate engine: standard route aggregation and summarization [RFC2519]
-- aggregate engine: injection of manually configured pseudo-static routes
-- generate default route if any working BGP connection exists (aggregate engine again?)
-- generate default route to IGP's (aggregate engine yet another time?)
-- look at RFC 2386 (QoS-based routing)
-- cli: show tables?
-
-OSPF
-~~~~
-	- check incoming packets using neighbor cache
-	- RFC2328 appendix E: Use a better algorithm
-	- automatic generation of external route tags (RFC1403)
-	- RFC2370 opaque LSA's
-	- Limit export rate of external LSAs (like Gated does)
-	- Bugfix in link state retransmission list (aging)
-	- Graceful OSPF restart - RFC3623

aclocal.m4

@@ -31,6 +31,41 @@ AC_DEFUN([BIRD_CHECK_PTHREADS],
   CFLAGS="$bird_tmp_cflags"
 ])
 
+AC_DEFUN([BIRD_CHECK_MPLS_KERNEL],
+[
+  AC_CACHE_CHECK(
+    [for Linux MPLS headers],
+    [bird_cv_mpls_kernel],
+    [
+      AC_COMPILE_IFELSE(
+	[
+	  AC_LANG_PROGRAM(
+	    [
+	      #include <linux/lwtunnel.h>
+	      #include <linux/netlink.h>
+	      #include <linux/rtnetlink.h>
+	      #include <sys/socket.h>
+	      void t(int arg);
+	    ],
+	    [
+	      t(AF_MPLS);
+	      t(RTA_VIA);
+	      t(RTA_NEWDST);
+	      t(RTA_ENCAP_TYPE);
+	      t(RTA_ENCAP);
+	      struct rtvia rtvia;
+	      t(LWTUNNEL_ENCAP_MPLS);
+	    ]
+	  )
+	],
+	[bird_cv_mpls_kernel=yes],
+	[bird_cv_mpls_kernel=no]
+      )
+    ]
+  )
+])
+
+
 AC_DEFUN([BIRD_CHECK_GCC_OPTION],
 [
   bird_tmp_cflags="$CFLAGS"

bird.conf

@@ -52,6 +52,6 @@ protocol rip {
 	ipv4;
 }
 
-protocol ripng {
+protocol rip ng {
 	ipv6;
 }

client/client.c

@@ -25,9 +25,10 @@
 #include <fcntl.h>
 #include <unistd.h>
 #include <errno.h>
+#include <sys/select.h>
 #include <sys/socket.h>
-#include <sys/un.h>
 #include <sys/types.h>
+#include <sys/un.h>
 
 #include "nest/bird.h"
 #include "lib/resource.h"

conf/conf.c

@@ -52,7 +52,7 @@
 #include "lib/resource.h"
 #include "lib/string.h"
 #include "lib/event.h"
-#include "sysdep/unix/timer.h"
+#include "lib/timer.h"
 #include "conf/conf.h"
 #include "filter/filter.h"
 
@@ -102,9 +102,9 @@ config_alloc(const char *name)
   c->pool = p;
   c->mem = l;
   c->file_name = ndup;
-  c->load_time = now;
+  c->load_time = current_time();
-  c->tf_route = c->tf_proto = (struct timeformat){"%T", "%F", 20*3600};
+  c->tf_route = c->tf_proto = TM_ISO_SHORT_MS;
-  c->tf_base = c->tf_log = (struct timeformat){"%F %T", NULL, 0};
+  c->tf_base = c->tf_log = TM_ISO_LONG_MS;
   c->gr_wait = DEFAULT_GR_WAIT;
 
   return c;
@@ -219,11 +219,6 @@ global_commit(struct config *new, struct config *old)
   if (!old)
     return 0;
 
-  if (!ipa_equal(old->listen_bgp_addr, new->listen_bgp_addr) ||
-      (old->listen_bgp_port != new->listen_bgp_port) ||
-      (old->listen_bgp_flags != new->listen_bgp_flags))
-    log(L_WARN "Reconfiguration of BGP listening socket not implemented, please restart BIRD.");
-
   if (!new->router_id)
     {
       new->router_id = old->router_id;
@@ -307,7 +302,7 @@ config_done(void *unused UNUSED)
  * config_commit - commit a configuration
  * @c: new configuration
  * @type: type of reconfiguration (RECONFIG_SOFT or RECONFIG_HARD)
- * @timeout: timeout for undo (or 0 for no timeout)
+ * @timeout: timeout for undo (in seconds; or 0 for no timeout)
  *
  * When a configuration is parsed and prepared for use, the
  * config_commit() function starts the process of reconfiguration.
@@ -331,7 +326,7 @@ config_done(void *unused UNUSED)
  * are accepted.
  */
 int
-config_commit(struct config *c, int type, int timeout)
+config_commit(struct config *c, int type, uint timeout)
 {
   if (shutting_down)
     {
@@ -340,8 +335,8 @@ config_commit(struct config *c, int type, int timeout)
     }
 
   undo_available = 1;
-  if (timeout > 0)
+  if (timeout)
-    tm_start(config_timer, timeout);
+    tm_start(config_timer, timeout S);
   else
     tm_stop(config_timer);
 
@@ -453,7 +448,7 @@ extern void cmd_reconfig_undo_notify(void);
 extern void cmd_reconfig_msg(int r);
 
 static void
-config_timeout(struct timer *t UNUSED)
+config_timeout(timer *t UNUSED)
 {
   log(L_INFO "Config timeout expired, starting undo");
   cmd_reconfig_undo_notify();

conf/conf.h

@@ -13,7 +13,7 @@
 #include "lib/ip.h"
 #include "lib/hash.h"
 #include "lib/resource.h"
-#include "sysdep/unix/timer.h"
+#include "lib/timer.h"
 
 
 /* Configuration structure */
@@ -32,16 +32,13 @@ struct config {
   struct iface_patt *router_id_from;	/* Configured list of router ID iface patterns */
 
   u32 router_id;			/* Our Router ID */
-  ip_addr listen_bgp_addr;		/* Listening BGP socket should use this address */
-  unsigned listen_bgp_port;		/* Listening BGP socket should use this port (0 is default) */
-  u32 listen_bgp_flags;			/* Listening BGP socket should use these flags */
   unsigned proto_default_debug;		/* Default protocol debug mask */
   unsigned proto_default_mrtdump;	/* Default protocol mrtdump mask */
   struct timeformat tf_route;		/* Time format for 'show route' */
   struct timeformat tf_proto;		/* Time format for 'show protocol' */
   struct timeformat tf_log;		/* Time format for the logfile */
   struct timeformat tf_base;		/* Time format for other purposes */
-  u32 gr_wait;				/* Graceful restart wait timeout */
+  u32 gr_wait;				/* Graceful restart wait timeout (sec) */
 
   int cli_debug;			/* Tracing of CLI connections and commands */
   int latency_debug;			/* I/O loop tracks duration of each event */
@@ -57,7 +54,7 @@ struct config {
   struct config *fallback;		/* Link to regular config for CLI parsing */
   int obstacle_count;			/* Number of items blocking freeing of this config */
   int shutdown;				/* This is a pseudo-config for daemon shutdown */
-  bird_clock_t load_time;		/* When we've got this configuration */
+  btime load_time;			/* When we've got this configuration */
 };
 
 /* Please don't use these variables in protocols. Use proto_config->global instead. */
@@ -68,7 +65,7 @@ struct config *config_alloc(const char *name);
 int config_parse(struct config *);
 int cli_parse(struct config *);
 void config_free(struct config *);
-int config_commit(struct config *, int type, int timeout);
+int config_commit(struct config *, int type, uint timeout);
 int config_confirm(void);
 int config_undo(void);
 void config_init(void);

conf/confbase.Y

@@ -14,7 +14,7 @@ CF_HDR
 #include "conf/conf.h"
 #include "lib/resource.h"
 #include "lib/socket.h"
-#include "sysdep/unix/timer.h"
+#include "lib/timer.h"
 #include "lib/string.h"
 #include "nest/protocol.h"
 #include "nest/iface.h"
@@ -60,7 +60,7 @@ CF_DECLS
   struct lsadb_show_data *ld;
   struct iface *iface;
   void *g;
-  bird_clock_t time;
+  btime time;
   struct f_prefix px;
   struct proto_spec ps;
   struct channel_limit cl;
@@ -80,11 +80,10 @@ CF_DECLS
 %type <iface> ipa_scope
 
 %type <i> expr bool pxlen4
-%type <i32> expr_us
+%type <time> expr_us time
-%type <time> datetime
 %type <a> ipa
 %type <net> net_ip4_ net_ip6_ net_ip6 net_ip_ net_ip net_or_ipa
-%type <net_ptr> net_ net_any net_vpn4_ net_vpn6_ net_vpn_ net_roa4_ net_roa6_ net_roa_
+%type <net_ptr> net_ net_any net_vpn4_ net_vpn6_ net_vpn_ net_roa4_ net_roa6_ net_roa_ net_mpls_
 %type <mls> label_stack_start label_stack
 
 %type <t> text opttext
@@ -97,7 +96,7 @@ CF_DECLS
 %left '!'
 %nonassoc '.'
 
-CF_KEYWORDS(DEFINE, ON, OFF, YES, NO, S, MS, US, PORT, VPN)
+CF_KEYWORDS(DEFINE, ON, OFF, YES, NO, S, MS, US, PORT, VPN, MPLS)
 
 CF_GRAMMAR
 
@@ -137,9 +136,9 @@ expr:
 
 
 expr_us:
-   expr S  { $$ = (u32) $1 * 1000000; }
+   expr S  { $$ = $1 S_; }
- | expr MS { $$ = (u32) $1 * 1000; }
+ | expr MS { $$ = $1 MS_; }
- | expr US { $$ = (u32) $1 * 1; }
+ | expr US { $$ = $1 US_; }
  ;
 
 /* Switches */
@@ -235,6 +234,12 @@ net_roa6_: net_ip6_ MAX NUM AS NUM
     cf_error("Invalid max prefix length %u", $3);
 };
 
+net_mpls_: MPLS NUM
+{
+  $$ = cfg_alloc(sizeof(net_addr_roa6));
+  net_fill_mpls($$, $2);
+}
+
 net_ip_: net_ip4_ | net_ip6_ ;
 net_vpn_: net_vpn4_ | net_vpn6_ ;
 net_roa_: net_roa4_ | net_roa6_ ;
@@ -244,6 +249,7 @@ net_:
  | net_vpn_
  | net_roa_
  | net_flow_
+ | net_mpls_
  ;
 
 
@@ -308,11 +314,11 @@ label_stack:
   }
 ;
 
-datetime:
+time:
    TEXT {
-     $$ = tm_parse_datetime($1);
+     $$ = tm_parse_time($1);
      if (!$$)
-       cf_error("Invalid date and time");
+       cf_error("Invalid date/time");
    }
  ;
 

conf/flowspec.Y

@@ -11,16 +11,6 @@ CF_HDR
 #define PARSER 1
 
 #include "nest/bird.h"
-#include "conf/conf.h"
-#include "lib/resource.h"
-#include "lib/socket.h"
-#include "sysdep/unix/timer.h"
-#include "lib/string.h"
-#include "nest/protocol.h"
-#include "nest/iface.h"
-#include "nest/route.h"
-#include "nest/cli.h"
-#include "filter/filter.h"
 #include "lib/flowspec.h"
 
 

configure.ac

@@ -36,6 +36,12 @@ AC_ARG_ENABLE([libssh],
   [enable_libssh=try]
 )
 
+AC_ARG_ENABLE([mpls-kernel],
+  [AS_HELP_STRING([--enable-mpls-kernel], [enable MPLS support in kernel protocol @<:@try@:>@])],
+  [],
+  [enable_mpls_kernel=try]
+)
+
 AC_ARG_WITH([protocols],
   [AS_HELP_STRING([--with-protocols=LIST], [include specified routing protocols @<:@all@:>@])],
   [],
@@ -239,6 +245,20 @@ if test "$enable_libssh" != no ; then
   fi
 fi
 
+if test "$enable_mpls_kernel" != no ; then
+  BIRD_CHECK_MPLS_KERNEL
+
+  if test "$bird_cv_mpls_kernel" = yes ; then
+    AC_DEFINE([HAVE_MPLS_KERNEL], [1], [Define to 1 if kernel is MPLS capable])
+  elif test "$enable_mpls_kernel" = yes ; then
+    AC_MSG_ERROR([Kernel MPLS support not found.])
+  fi
+
+  if test "$enable_mpls_kernel" = try ; then
+    enable_mpls_kernel="$bird_cv_mpls_kernel"
+  fi
+fi
+
 all_protocols="$proto_bfd babel bgp ospf pipe radv rip $proto_rpki static"
 
 all_protocols=`echo $all_protocols | sed 's/ /,/g'`
@@ -287,18 +307,6 @@ esac
 AC_CHECK_HEADERS_ONCE([alloca.h syslog.h])
 AC_CHECK_MEMBERS([struct sockaddr.sa_len], [], [], [#include <sys/socket.h>])
 
-AC_CHECK_HEADERS([linux/lwtunnel.h],
-  [AC_DEFINE([HAVE_LWTUNNEL], [1], [Define to 1 if you have the <linux/lwtunnel.h> header file.])],
-  [],
-  [] dnl Force new AC_CHECK_HEADER semantics
-)
-
-AC_CHECK_MEMBERS([struct rtvia.rtvia_family],
-  [AC_DEFINE([HAVE_STRUCT_RTVIA], [1], [Define to 1 if you have rtvia structure.])],
-  [],
-  [#include <linux/rtnetlink.h>]
-)
-
 AC_C_BIGENDIAN(
   [AC_DEFINE([CPU_BIG_ENDIAN], [1], [Define to 1 if cpu is big endian])],
   [AC_DEFINE([CPU_LITTLE_ENDIAN], [1], [Define to 1 if cpu is little endian])],
@@ -314,7 +322,7 @@ if test "$enable_debug" = yes ; then
     [
       AC_DEFINE([HAVE_EXECINFO_H], [1], [Define to 1 if you have the <execinfo.h> header file.])
       AC_SEARCH_LIBS([backtrace], [execinfo],
-	[]
+	[],
 	[AC_MSG_ERROR([Function backtrace not available.])]
       )
     ]
@@ -396,6 +404,7 @@ AC_MSG_RESULT([        System configuration:	$sysdesc])
 AC_MSG_RESULT([        Debugging:		$enable_debug])
 AC_MSG_RESULT([        POSIX threads:		$enable_pthreads])
 AC_MSG_RESULT([        Routing protocols:	$protocols])
+AC_MSG_RESULT([        Kernel MPLS support:	$enable_mpls_kernel])
 AC_MSG_RESULT([        Client:			$enable_client])
 
 rm -f $objdir/.*-stamp

filter/config.Y

@@ -112,7 +112,7 @@ f_new_ec_item(u32 kind, u32 ipv4_used, u32 key, u32 vf, u32 vt)
 {
   u64 fm, to;
 
-  if (ipv4_used || (key >= 0x10000)) {
+  if ((kind != EC_GENERIC) && (ipv4_used || (key >= 0x10000))) {
     check_u16(vf);
     if (vt == EC_ALL)
       vt = 0xFFFF;
@@ -421,7 +421,7 @@ CF_KEYWORDS(FUNCTION, PRINT, PRINTN, UNSET, RETURN,
 %type <v> set_atom switch_atom fipa
 %type <px> fprefix
 %type <s> decls declsn one_decl function_params
-%type <h> bgp_path bgp_path_tail1 bgp_path_tail2
+%type <h> bgp_path bgp_path_tail
 %type <t> get_cf_position
 
 CF_GRAMMAR
@@ -763,25 +763,18 @@ bgp_path_expr:
  ;
 
 bgp_path:
-   PO  bgp_path_tail1 PC  { $$ = $2; }
+   PO  bgp_path_tail PC  { $$ = $2; }
- | '/' bgp_path_tail2 '/' { $$ = $2; }
  ;
 
-bgp_path_tail1:
+bgp_path_tail:
-   NUM bgp_path_tail1		{ $$ = cfg_allocz(sizeof(struct f_path_mask)); $$->next = $2; $$->kind = PM_ASN; $$->val = $1; }
+   NUM bgp_path_tail		{ $$ = cfg_allocz(sizeof(struct f_path_mask)); $$->next = $2; $$->kind = PM_ASN; $$->val = $1; }
- | NUM DDOT NUM bgp_path_tail1	{ $$ = cfg_allocz(sizeof(struct f_path_mask)); $$->next = $4; $$->kind = PM_ASN_RANGE; $$->val = $1; $$->val2 = $3; }
+ | NUM DDOT NUM bgp_path_tail	{ $$ = cfg_allocz(sizeof(struct f_path_mask)); $$->next = $4; $$->kind = PM_ASN_RANGE; $$->val = $1; $$->val2 = $3; }
- | '*' bgp_path_tail1		{ $$ = cfg_allocz(sizeof(struct f_path_mask)); $$->next = $2; $$->kind = PM_ASTERISK; }
+ | '*' bgp_path_tail		{ $$ = cfg_allocz(sizeof(struct f_path_mask)); $$->next = $2; $$->kind = PM_ASTERISK; }
- | '?' bgp_path_tail1		{ $$ = cfg_allocz(sizeof(struct f_path_mask)); $$->next = $2; $$->kind = PM_QUESTION; }
+ | '?' bgp_path_tail		{ $$ = cfg_allocz(sizeof(struct f_path_mask)); $$->next = $2; $$->kind = PM_QUESTION; }
- | bgp_path_expr bgp_path_tail1	{ $$ = cfg_allocz(sizeof(struct f_path_mask)); $$->next = $2; $$->kind = PM_ASN_EXPR; $$->val = (uintptr_t) $1; }
+ | bgp_path_expr bgp_path_tail	{ $$ = cfg_allocz(sizeof(struct f_path_mask)); $$->next = $2; $$->kind = PM_ASN_EXPR; $$->val = (uintptr_t) $1; }
  | 				{ $$ = NULL; }
  ;
 
-bgp_path_tail2:
-   NUM bgp_path_tail2 { $$ = cfg_allocz(sizeof(struct f_path_mask)); $$->next = $2; $$->kind = PM_ASN; $$->val = $1; }
- | '?' bgp_path_tail2 { $$ = cfg_allocz(sizeof(struct f_path_mask)); $$->next = $2; $$->kind = PM_ASTERISK; }
- | 		      { $$ = NULL; }
- ;
-
 constant:
    NUM    { $$ = f_new_inst(); $$->code = 'c'; $$->aux = T_INT;  $$->a2.i = $1; }
  | TRUE   { $$ = f_new_inst(); $$->code = 'c'; $$->aux = T_BOOL; $$->a2.i = 1;  }

filter/filter.c

@@ -1089,7 +1089,8 @@ interpret(struct f_inst *what)
 
       switch (what->aux & EAF_TYPE_MASK) {
       case EAF_TYPE_INT:
-	if (v1.type != T_INT)
+	// Enums are also ints, so allow them in.
+	if (v1.type != T_INT && (v1.type < T_ENUM_LO || v1.type > T_ENUM_HI))
 	  runtime( "Setting int attribute to non-int value" );
 	l->attrs[0].u.data = v1.val.i;
 	break;

filter/filter.h

@@ -147,6 +147,7 @@ void val_format(struct f_val v, buffer *buf);
 #define T_ENUM_RTD 0x34
 #define T_ENUM_ROA 0x35
 #define T_ENUM_NETTYPE 0x36
+#define T_ENUM_RA_PREFERENCE 0x37
 
 /* new enums go here */
 #define T_ENUM_EMPTY 0x3f	/* Special hack for atomic_aggr */

filter/test.conf

@@ -589,14 +589,11 @@ function mkpath(int a; int b)
 
 function t_path()
 bgpmask pm1;
-bgpmask pm2;
 bgppath p2;
 {
-	pm1 =  / 4 3 2 1 /;
+	pm1 = [= 4 3 2 1 =];
-	pm2 = [= 4 3 2 1 =];
 
-	bt_assert(pm1 = pm2);
+	bt_assert(format(pm1) = "[= 4 3 2 1 =]");
-	bt_assert(format(pm2) = "[= 4 3 2 1 =]");
 
 	bt_assert(+empty+ = +empty+);
 	bt_assert(10 !~ +empty+);
@@ -609,17 +606,14 @@ bgppath p2;
 	bt_assert(format(p2) = "(path 4 3 2 1)");
 	bt_assert(p2.len = 4);
 	bt_assert(p2 ~ pm1);
-	bt_assert(p2 ~ pm2);
 	bt_assert(3 ~ p2);
 	bt_assert(p2 ~ [2, 10..20]);
 	bt_assert(p2 ~ [4, 10..20]);
 
 	p2 = prepend(p2, 5);
 	bt_assert(p2 !~ pm1);
-	bt_assert(p2 !~ pm2);
 	bt_assert(10 !~ p2);
 	bt_assert(p2 !~ [8, ten..(2*ten)]);
-	bt_assert(p2 ~  / ? 4 3 2 1 /);
 	bt_assert(p2 ~ [= * 4 3 * 1 =]);
 	bt_assert(p2 ~ [= (3+2) (2*2) 3 2 1 =]);
 	bt_assert(p2 ~ mkpath(5, 4));

lib/Makefile

@@ -1,4 +1,4 @@
-src := bitops.c checksum.c event.c flowspec.c idm.c ip.c lists.c mac.c md5.c mempool.c net.c patmatch.c printf.c resource.c sha1.c sha256.c sha512.c slab.c slists.c tbf.c xmalloc.c
+src := bitops.c checksum.c event.c flowspec.c idm.c ip.c lists.c mac.c md5.c mempool.c net.c patmatch.c printf.c resource.c sha1.c sha256.c sha512.c slab.c slists.c tbf.c timer.c xmalloc.c
 obj := $(src-o-files)
 $(all-daemon)
 

lib/birdlib.h

@@ -9,7 +9,6 @@
 #ifndef _BIRD_BIRDLIB_H_
 #define _BIRD_BIRDLIB_H_
 
-#include "sysdep/unix/timer.h"
 #include "lib/alloca.h"
 
 /* Ugly structure offset handling macros */
@@ -70,10 +69,11 @@ static inline int u64_cmp(u64 i1, u64 i2)
 /* Microsecond time */
 
 typedef s64 btime;
+//typedef s64 bird_clock_t;
 
-#define S_	*1000000
+#define S_	* (btime) 1000000
-#define MS_	*1000
+#define MS_	* (btime) 1000
-#define US_	*1
+#define US_	* (btime) 1
 #define TO_S	/1000000
 #define TO_MS	/1000
 #define TO_US	/1
@@ -82,39 +82,26 @@ typedef s64 btime;
 #define S	S_
 #define MS	MS_
 #define US	US_
+#define NS	/1000
 #endif
 
+#define TIME_INFINITY ((s64) 0x7fffffffffffffff)
+
 
 /* Rate limiting */
 
 struct tbf {
-  bird_clock_t timestamp;		/* Last update */
+  btime timestamp;			/* Last update */
-  u16 count;				/* Available tokens */
+  u64 count;				/* Available micro-tokens */
   u16 burst;				/* Max number of tokens */
-  u16 rate;				/* Rate of replenishment */
+  u16 rate;				/* Rate of replenishment (tokens / sec) */
-  u16 mark;				/* Whether last op was limited */
+  u32 drop;				/* Number of failed request since last successful */
 };
 
 /* Default TBF values for rate limiting log messages */
 #define TBF_DEFAULT_LOG_LIMITS { .rate = 1, .burst = 5 }
 
-void tbf_update(struct tbf *f);
+int tbf_limit(struct tbf *f);
-
-static inline int
-tbf_limit(struct tbf *f)
-{
-  tbf_update(f);
-
-  if (!f->count)
-  {
-    f->mark = 1;
-    return 1;
-  }
-
-  f->count--;
-  f->mark = 0;
-  return 0;
-}
 
 
 /* Logging and dying */

lib/buffer.h

@@ -13,10 +13,14 @@
 #include "lib/resource.h"
 #include "sysdep/config.h"
 
-#define BUFFER(type)		struct { type *data; uint used, size; }
+#define BUFFER_(type)		struct { type *data; uint used, size; }
 #define BUFFER_TYPE(v)		typeof(* (v).data)
 #define BUFFER_SIZE(v)		((v).size * sizeof(* (v).data))
 
+#ifndef PARSER
+#define BUFFER(type) BUFFER_(type)
+#endif
+
 #define BUFFER_INIT(v,pool,isize)					\
   ({									\
     (v).used = 0;							\

lib/event_test.c

@@ -55,6 +55,7 @@ t_ev_run_list(void)
 
   resource_init();
   olock_init();
+  timer_init();
   io_init();
   rt_init();
   if_init();

lib/flowspec.h

@@ -91,7 +91,7 @@ const byte *flow6_next_part(const byte *pos, const byte *end);
 
 /* A data structure for keep a state of flow builder */
 struct flow_builder {
-  BUFFER(byte) data;
+  BUFFER_(byte) data;
   enum flow_type this_type;
   enum flow_type last_type;
   u16 last_op_offset;			/* Position of last operator in data.data */

lib/flowspec_test.c

@@ -70,8 +70,8 @@ t_first_part(void)
   net_addr_flow4 *f;
   NET_ADDR_FLOW4_(f, ip4_build(10,0,0,1), 24, ((byte[]) { 0x00, 0x00, 0xab }));
 
-  const byte const *under240 = &f->data[1];
+  const byte *under240 = &f->data[1];
-  const byte const *above240 = &f->data[2];
+  const byte *above240 = &f->data[2];
 
   /* Case 0x00 0x00 */
   bt_assert(flow4_first_part(f) == NULL);

lib/mempool.c

@@ -11,7 +11,7 @@
  *
  * Linear memory pools are collections of memory blocks which
  * support very fast allocation of new blocks, but are able to free only
- * the whole collection at once.
+ * the whole collection at once (or in stack order).
  *
  * Example: Each configuration is described by a complex system of structures,
  * linked lists and function trees which are all allocated from a single linear
@@ -37,7 +37,7 @@ const int lp_chunk_size = sizeof(struct lp_chunk);
 struct linpool {
   resource r;
   byte *ptr, *end;
-  struct lp_chunk *first, *current, **plast;	/* Normal (reusable) chunks */
+  struct lp_chunk *first, *current;		/* Normal (reusable) chunks */
   struct lp_chunk *first_large;			/* Large chunks */
   uint chunk_size, threshold, total, total_large;
 };
@@ -69,7 +69,6 @@ linpool
 *lp_new(pool *p, uint blk)
 {
   linpool *m = ralloc(p, &lp_class);
-  m->plast = &m->first;
   m->chunk_size = blk;
   m->threshold = 3*blk/4;
   return m;
@@ -114,22 +113,25 @@ lp_alloc(linpool *m, uint size)
 	}
       else
 	{
-	  if (m->current)
+	  if (m->current && m->current->next)
 	    {
 	      /* Still have free chunks from previous incarnation (before lp_flush()) */
-	      c = m->current;
+	      c = m->current->next;
-	      m->current = c->next;
 	    }
 	  else
 	    {
 	      /* Need to allocate a new chunk */
 	      c = xmalloc(sizeof(struct lp_chunk) + m->chunk_size);
 	      m->total += m->chunk_size;
-	      *m->plast = c;
-	      m->plast = &c->next;
 	      c->next = NULL;
 	      c->size = m->chunk_size;
+
+	      if (m->current)
+		m->current->next = c;
+	      else
+		m->first = c;
 	    }
+	  m->current = c;
 	  m->ptr = c->data + size;
 	  m->end = c->data + m->chunk_size;
 	}
@@ -190,9 +192,11 @@ lp_flush(linpool *m)
 {
   struct lp_chunk *c;
 
-  /* Relink all normal chunks to free list and free all large chunks */
+  /* Move ptr to the first chunk and free all large chunks */
-  m->ptr = m->end = NULL;
+  m->current = c = m->first;
-  m->current = m->first;
+  m->ptr = c ? c->data : NULL;
+  m->end = c ? c->data + m->chunk_size : NULL;
+
   while (c = m->first_large)
     {
       m->first_large = c->next;
@@ -201,6 +205,49 @@ lp_flush(linpool *m)
   m->total_large = 0;
 }
 
+/**
+ * lp_save - save the state of a linear memory pool
+ * @m: linear memory pool
+ * @p: state buffer
+ *
+ * This function saves the state of a linear memory pool. Saved state can be
+ * used later to restore the pool (to free memory allocated since).
+ */
+void
+lp_save(linpool *m, lp_state *p)
+{
+  p->current = m->current;
+  p->large = m->first_large;
+  p->ptr = m->ptr;
+}
+
+/**
+ * lp_restore - restore the state of a linear memory pool
+ * @m: linear memory pool
+ * @p: saved state
+ *
+ * This function restores the state of a linear memory pool, freeing all memory
+ * allocated since the state was saved. Note that the function cannot un-free
+ * the memory, therefore the function also invalidates other states that were
+ * saved between (on the same pool).
+ */
+void
+lp_restore(linpool *m, lp_state *p)
+{
+  struct lp_chunk *c;
+
+  /* Move ptr to the saved pos and free all newer large chunks */
+  m->current = c = p->current;
+  m->ptr = p->ptr;
+  m->end = c ? c->data + m->chunk_size : NULL;
+
+  while ((c = m->first_large) && (c != p->large))
+    {
+      m->first_large = c->next;
+      xfree(c);
+    }
+}
+
 static void
 lp_free(resource *r)
 {

lib/net.c

@@ -288,3 +288,21 @@ net_in_netX(const net_addr *a, const net_addr *n)
 
   return (net_pxlen(n) <= net_pxlen(a)) && ipa_in_netX(net_prefix(a), n);
 }
+
+#define CHECK_NET(T,S) \
+  ({ if (sizeof(T) != S) die("sizeof %s is %d/%d", #T, (int) sizeof(T), S); })
+
+void
+net_init(void)
+{
+  CHECK_NET(net_addr,		24);
+  CHECK_NET(net_addr_ip4,	 8);
+  CHECK_NET(net_addr_ip6,	20);
+  CHECK_NET(net_addr_vpn4,	16);
+  CHECK_NET(net_addr_vpn6,	32);
+  CHECK_NET(net_addr_roa4,	16);
+  CHECK_NET(net_addr_roa6,	28);
+  CHECK_NET(net_addr_flow4,	 8);
+  CHECK_NET(net_addr_flow6,	20);
+  CHECK_NET(net_addr_mpls,	 8);
+}

lib/net.h

@@ -45,7 +45,7 @@ typedef struct net_addr {
   u8 type;
   u8 pxlen;
   u16 length;
-  u8 data[16];
+  u8 data[20];
   u64 align[0];
 } net_addr;
 
@@ -76,6 +76,7 @@ typedef struct net_addr_vpn6 {
   u8 pxlen;
   u16 length;
   ip6_addr prefix;
+  u32 padding;
   u64 rd;
 } net_addr_vpn6;
 
@@ -152,7 +153,7 @@ extern const u16 net_max_text_length[];
   ((net_addr_vpn4) { NET_VPN4, pxlen, sizeof(net_addr_vpn4), prefix, rd })
 
 #define NET_ADDR_VPN6(prefix,pxlen,rd) \
-  ((net_addr_vpn6) { NET_VPN6, pxlen, sizeof(net_addr_vpn6), prefix, rd })
+  ((net_addr_vpn6) { NET_VPN6, pxlen, sizeof(net_addr_vpn6), prefix, 0, rd })
 
 #define NET_ADDR_ROA4(prefix,pxlen,max_pxlen,asn) \
   ((net_addr_roa4) { NET_ROA4, pxlen, sizeof(net_addr_roa4), prefix, max_pxlen, asn })
@@ -230,11 +231,14 @@ static inline int net_type_match(const net_addr *a, u32 mask)
 static inline int net_is_ip(const net_addr *a)
 { return (a->type == NET_IP4) || (a->type == NET_IP6); }
 
+static inline int net_is_vpn(const net_addr *a)
+{ return (a->type == NET_VPN4) || (a->type == NET_VPN6); }
+
 static inline int net_is_roa(const net_addr *a)
 { return (a->type == NET_ROA4) || (a->type == NET_ROA6); }
 
-static inline int net_is_vpn(const net_addr *a)
+static inline int net_is_flow(const net_addr *a)
-{ return (a->type == NET_VPN4) || (a->type == NET_VPN6); }
+{ return (a->type == NET_FLOW4) || (a->type == NET_FLOW6); }
 
 
 static inline ip4_addr net4_prefix(const net_addr *a)
@@ -354,10 +358,10 @@ static inline int net_zero_roa6(const net_addr_roa6 *a)
 { return !a->pxlen && ip6_zero(a->prefix) && !a->max_pxlen && !a->asn; }
 
 static inline int net_zero_flow4(const net_addr_flow4 *a)
-{ return !a->pxlen && ip4_zero(a->prefix) && !a->data; }
+{ return !a->pxlen && ip4_zero(a->prefix) && (a->length == sizeof(net_addr_flow4)); }
 
 static inline int net_zero_flow6(const net_addr_flow6 *a)
-{ return !a->pxlen && ip6_zero(a->prefix) && !a->data; }
+{ return !a->pxlen && ip6_zero(a->prefix) && (a->length == sizeof(net_addr_flow6)); }
 
 static inline int net_zero_mpls(const net_addr_mpls *a)
 { return !a->label; }
@@ -526,8 +530,21 @@ int net_classify(const net_addr *N);
 int net_format(const net_addr *N, char *buf, int buflen);
 int rd_format(const u64 rd, char *buf, int buflen);
 
+static inline int ipa_in_net_ip4(ip4_addr a, const net_addr_ip4 *n)
+{ return ip4_zero(ip4_and(ip4_xor(a, n->prefix), ip4_mkmask(n->pxlen))); }
+
+static inline int net_in_net_ip4(const net_addr_ip4 *a, const net_addr_ip4 *b)
+{ return (a->pxlen >= b->pxlen) && ipa_in_net_ip4(a->prefix, b); }
+
+static inline int ipa_in_net_ip6(ip6_addr a, const net_addr_ip6 *n)
+{ return ip6_zero(ip6_and(ip6_xor(a, n->prefix), ip6_mkmask(n->pxlen))); }
+
+static inline int net_in_net_ip6(const net_addr_ip6 *a, const net_addr_ip6 *b)
+{ return (a->pxlen >= b->pxlen) && ipa_in_net_ip6(a->prefix, b); }
+
 int ipa_in_netX(const ip_addr A, const net_addr *N);
 int net_in_netX(const net_addr *A, const net_addr *N);
 
+void net_init(void);
 
 #endif

lib/printf.c

@@ -125,9 +125,10 @@ static char * number(char * str, long num, int base, int size, int precision,
  * or |%I6| can be used for explicit ip4_addr / ip6_addr arguments, |%N| for
  * generic network addresses (net_addr *), |%R| for Router / Network ID (u32
  * value printed as IPv4 address), |%lR| for 64bit Router / Network ID (u64
- * value printed as eight :-separated octets) and |%m| resp. |%M| for error
+ * value printed as eight :-separated octets), |%t| for time values (btime) with
- * messages (uses strerror() to translate @errno code to message text). On the
+ * specified subsecond precision, and |%m| resp. |%M| for error messages (uses
- * other hand, it doesn't support floating point numbers.
+ * strerror() to translate @errno code to message text). On the other hand, it
+ * doesn't support floating point numbers.
  *
  * Result: number of characters of the output string or -1 if
  * the buffer space was insufficient.
@@ -139,6 +140,8 @@ int bvsnprintf(char *buf, int size, const char *fmt, va_list args)
 	int i, base;
 	u32 x;
 	u64 X;
+	btime t;
+	s64 t1, t2;
 	char *str, *start;
 	const char *s;
 	char ipbuf[NET_MAX_TEXT_LENGTH+1];
@@ -279,7 +282,6 @@ int bvsnprintf(char *buf, int size, const char *fmt, va_list args)
 				return -1;
 			continue;
 
-
 		case 'n':
 			if (qualifier == 'l') {
 				long * ip = va_arg(args, long *);
@@ -360,6 +362,50 @@ int bvsnprintf(char *buf, int size, const char *fmt, va_list args)
 			s = ipbuf;
 			goto str;
 
+		case 't':
+			t = va_arg(args, btime);
+			t1 = t TO_S;
+			t2 = t - t1 S;
+
+			if (precision < 0)
+			  precision = 3;
+
+			if (precision > 6)
+			  precision = 6;
+
+			/* Compute field_width for second part */
+			if ((precision > 0) && (field_width > 0))
+			  field_width -= (1 + precision);
+
+			if (field_width < 0)
+			  field_width = 0;
+
+			/* Print seconds */
+			flags |= SIGN;
+			str = number(str, t1, 10, field_width, 0, flags, size);
+			if (!str)
+			  return -1;
+
+			if (precision > 0)
+			{
+			  size -= (str-start);
+			  start = str;
+
+			  if ((1 + precision) > size)
+			    return -1;
+
+			  /* Convert microseconds to requested precision */
+			  for (i = precision; i < 6; i++)
+			    t2 /= 10;
+
+			  /* Print sub-seconds */
+			  *str++ = '.';
+			  str = number(str, t2, 10, precision, 0, ZEROPAD, size - 1);
+			  if (!str)
+			    return -1;
+			}
+			goto done;
+
 		/* integer number formats - set up the flags and "break" */
 		case 'o':
 			base = 8;
@@ -401,6 +447,7 @@ int bvsnprintf(char *buf, int size, const char *fmt, va_list args)
 		str = number(str, num, base, field_width, precision, flags, size);
 		if (!str)
 			return -1;
+	done:	;
 	}
 	if (!size)
 		return -1;

lib/printf_test.c

@@ -56,6 +56,15 @@ t_simple(void)
   BSPRINTF(2, "-1", buf, "%d", -1);
   BSPRINTF(11, "-2147483648", buf, "%d", -2147483648);
 
+  BSPRINTF(7, "123.456", buf, "%t", (btime) 123456789);
+  BSPRINTF(7, "123.456", buf, "%2t", (btime) 123456789);
+  BSPRINTF(8, " 123.456", buf, "%8t", (btime) 123456789);
+  BSPRINTF(4, " 123", buf, "%4.0t", (btime) 123456789);
+  BSPRINTF(8, "123.4567", buf, "%8.4t", (btime) 123456789);
+  BSPRINTF(9, "0123.4567", buf, "%09.4t", (btime) 123456789);
+  BSPRINTF(12, "  123.456789", buf, "%12.10t", (btime) 123456789);
+  BSPRINTF(8, " 123.004", buf, "%8t", (btime) 123004 MS);
+
   return 1;
 }
 

lib/resource.h

@@ -59,11 +59,18 @@ void mb_free(void *);
 
 typedef struct linpool linpool;
 
+typedef struct lp_state {
+  void *current, *large;
+  byte *ptr;
+} lp_state;
+
 linpool *lp_new(pool *, unsigned blk);
 void *lp_alloc(linpool *, unsigned size);	/* Aligned */
 void *lp_allocu(linpool *, unsigned size);	/* Unaligned */
 void *lp_allocz(linpool *, unsigned size);	/* With clear */
 void lp_flush(linpool *);			/* Free everything, but leave linpool */
+void lp_save(linpool *m, lp_state *p);		/* Save state */
+void lp_restore(linpool *m, lp_state *p);	/* Restore state */
 
 extern const int lp_chunk_size;
 #define LP_GAS		    1024

lib/socket.h

@@ -50,6 +50,7 @@ typedef struct birdsock {
   int ttl;				/* Time To Live, -1 = default */
   u32 flags;
   struct iface *iface;			/* Interface; specify this for broad/multicast sockets */
+  struct iface *vrf;			/* Related VRF instance, NULL if global */
 
   byte *rbuf, *rpos;			/* NULL=allocate automatically */
   uint fast_rx;				/* RX has higher priority in event loop */

lib/tbf.c

@@ -8,22 +8,30 @@
  */
 
 #include "nest/bird.h"
+#include "lib/timer.h"
 
-void
+int
-tbf_update(struct tbf *f)
+tbf_limit(struct tbf *f)
 {
-  bird_clock_t delta = now - f->timestamp;
+  btime delta = current_time() - f->timestamp;
 
-  if (delta == 0)
+  if (delta > 0)
-    return;
-
-  f->timestamp = now;
-
-  if ((0 < delta) && (delta < f->burst))
   {
-    u32 next = f->count + delta * f->rate;
+    u64 next = f->count + delta * f->rate;
-    f->count = MIN(next, f->burst);
+    u64 burst = (u64) f->burst << 20;
+    f->count = MIN(next, burst);
+    f->timestamp += delta;
+  }
+
+  if (f->count < 1000000)
+  {
+    f->drop++;
+    return 1;
   }
   else
-    f->count = f->burst;
+  {
+    f->count -= 1000000;
+    f->drop = 0;
+    return 0;
+  }
 }

lib/timer.c

@@ -0,0 +1,377 @@
+/*
+ *	BIRD -- Timers
+ *
+ *	(c) 2013--2017 Ondrej Zajicek <santiago@crfreenet.org>
+ *	(c) 2013--2017 CZ.NIC z.s.p.o.
+ *
+ *	Can be freely distributed and used under the terms of the GNU GPL.
+ */
+
+/**
+ * DOC: Timers
+ *
+ * Timers are resources which represent a wish of a module to call a function at
+ * the specified time. The timer code does not guarantee exact timing, only that
+ * a timer function will not be called before the requested time.
+ *
+ * In BIRD, time is represented by values of the &btime type which is signed
+ * 64-bit integer interpreted as a relative number of microseconds since some
+ * fixed time point in past. The current time can be obtained by current_time()
+ * function with reasonable accuracy and is monotonic. There is also a current
+ * 'wall-clock' real time obtainable by current_real_time() reported by OS.
+ *
+ * Each timer is described by a &timer structure containing a pointer to the
+ * handler function (@hook), data private to this function (@data), time the
+ * function should be called at (@expires, 0 for inactive timers), for the other
+ * fields see |timer.h|.
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+
+#include "nest/bird.h"
+
+#include "lib/heap.h"
+#include "lib/resource.h"
+#include "lib/timer.h"
+
+
+struct timeloop main_timeloop;
+
+
+#ifdef USE_PTHREADS
+
+#include <pthread.h>
+
+/* Data accessed and modified from proto/bfd/io.c */
+pthread_key_t current_time_key;
+
+static inline struct timeloop *
+timeloop_current(void)
+{
+  return pthread_getspecific(current_time_key);
+}
+
+static inline void
+timeloop_init_current(void)
+{
+  pthread_key_create(&current_time_key, NULL);
+  pthread_setspecific(current_time_key, &main_timeloop);
+}
+
+void wakeup_kick_current(void);
+
+#else
+
+/* Just use main timelooop */
+static inline struct timeloop * timeloop_current(void) { return &main_timeloop; }
+static inline void timeloop_init_current(void) { }
+
+#endif
+
+btime
+current_time(void)
+{
+  return timeloop_current()->last_time;
+}
+
+btime
+current_real_time(void)
+{
+  struct timeloop *loop = timeloop_current();
+
+  if (!loop->real_time)
+    times_update_real_time(loop);
+
+  return loop->real_time;
+}
+
+
+#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 void
+tm_free(resource *r)
+{
+  timer *t = (void *) r;
+
+  tm_stop(t);
+}
+
+static void
+tm_dump(resource *r)
+{
+  timer *t = (void *) r;
+
+  debug("(code %p, data %p, ", t->hook, t->data);
+  if (t->randomize)
+    debug("rand %d, ", t->randomize);
+  if (t->recurrent)
+    debug("recur %d, ", t->recurrent);
+  if (t->expires)
+    debug("expires in %d ms)\n", (t->expires - current_time()) TO_MS);
+  else
+    debug("inactive)\n");
+}
+
+
+static struct resclass tm_class = {
+  "Timer",
+  sizeof(timer),
+  tm_free,
+  tm_dump,
+  NULL,
+  NULL
+};
+
+timer *
+tm_new(pool *p)
+{
+  timer *t = ralloc(p, &tm_class);
+  t->index = -1;
+  return t;
+}
+
+void
+tm_set(timer *t, btime when)
+{
+  struct timeloop *loop = timeloop_current();
+  uint tc = timers_count(loop);
+
+  if (!t->expires)
+  {
+    t->index = ++tc;
+    t->expires = when;
+    BUFFER_PUSH(loop->timers) = t;
+    HEAP_INSERT(loop->timers.data, tc, timer *, TIMER_LESS, TIMER_SWAP);
+  }
+  else if (t->expires < when)
+  {
+    t->expires = when;
+    HEAP_INCREASE(loop->timers.data, tc, timer *, TIMER_LESS, TIMER_SWAP, t->index);
+  }
+  else if (t->expires > when)
+  {
+    t->expires = when;
+    HEAP_DECREASE(loop->timers.data, tc, timer *, TIMER_LESS, TIMER_SWAP, t->index);
+  }
+
+#ifdef CONFIG_BFD
+  /* Hack to notify BFD loops */
+  if ((loop != &main_timeloop) && (t->index == 1))
+    wakeup_kick_current();
+#endif
+}
+
+void
+tm_start(timer *t, btime after)
+{
+  tm_set(t, current_time() + MAX(after, 0));
+}
+
+void
+tm_stop(timer *t)
+{
+  if (!t->expires)
+    return;
+
+  struct timeloop *loop = timeloop_current();
+  uint tc = timers_count(loop);
+
+  HEAP_DELETE(loop->timers.data, tc, timer *, TIMER_LESS, TIMER_SWAP, t->index);
+  BUFFER_POP(loop->timers);
+
+  t->index = -1;
+  t->expires = 0;
+}
+
+void
+timers_init(struct timeloop *loop, pool *p)
+{
+  times_init(loop);
+
+  BUFFER_INIT(loop->timers, p, 4);
+  BUFFER_PUSH(loop->timers) = NULL;
+}
+
+void io_log_event(void *hook, void *data);
+
+void
+timers_fire(struct timeloop *loop)
+{
+  btime base_time;
+  timer *t;
+
+  times_update(loop);
+  base_time = loop->last_time;
+
+  while (t = timers_first(loop))
+  {
+    if (t->expires > base_time)
+      return;
+
+    if (t->recurrent)
+    {
+      btime when = t->expires + t->recurrent;
+
+      if (when <= loop->last_time)
+	when = loop->last_time + t->recurrent;
+
+      if (t->randomize)
+	when += random() % (t->randomize + 1);
+
+      tm_set(t, when);
+    }
+    else
+      tm_stop(t);
+
+    /* This is ugly hack, we want to log just timers executed from the main I/O loop */
+    if (loop == &main_timeloop)
+      io_log_event(t->hook, t->data);
+
+    t->hook(t);
+  }
+}
+
+void
+timer_init(void)
+{
+  timers_init(&main_timeloop, &root_pool);
+  timeloop_init_current();
+}
+
+
+/**
+ * tm_parse_time - parse a date and time
+ * @x: time string
+ *
+ * tm_parse_time() takes a textual representation of a date and time
+ * (yyyy-mm-dd[ hh:mm:ss[.sss]]) and converts it to the corresponding value of
+ * type &btime.
+ */
+btime
+tm_parse_time(char *x)
+{
+  struct tm tm;
+  int usec, n1, n2, n3, r;
+
+  r = sscanf(x, "%d-%d-%d%n %d:%d:%d%n.%d%n",
+	     &tm.tm_year, &tm.tm_mon, &tm.tm_mday, &n1,
+	     &tm.tm_hour, &tm.tm_min, &tm.tm_sec, &n2,
+	     &usec, &n3);
+
+  if ((r == 3) && !x[n1])
+    tm.tm_hour = tm.tm_min = tm.tm_sec = usec = 0;
+  else if ((r == 6) && !x[n2])
+    usec = 0;
+  else if ((r == 7) && !x[n3])
+  {
+    /* Convert subsecond digits to proper precision */
+    int digits = n3 - n2 - 1;
+    if ((usec < 0) || (usec > 999999) || (digits < 1) || (digits > 6))
+      return 0;
+
+    while (digits++ < 6)
+      usec *= 10;
+  }
+  else
+    return 0;
+
+  tm.tm_mon--;
+  tm.tm_year -= 1900;
+  s64 ts = mktime(&tm);
+  if ((ts == (s64) (time_t) -1) || (ts < 0) || (ts > ((s64) 1 << 40)))
+    return 0;
+
+  return ts S + usec;
+}
+
+/**
+ * tm_format_time - convert date and time to textual representation
+ * @x: destination buffer of size %TM_DATETIME_BUFFER_SIZE
+ * @fmt: specification of resulting textual representation of the time
+ * @t: time
+ *
+ * This function formats the given relative time value @t to a textual
+ * date/time representation (dd-mm-yyyy hh:mm:ss) in real time.
+ */
+void
+tm_format_time(char *x, struct timeformat *fmt, btime t)
+{
+  btime dt = current_time() - t;
+  btime rt = current_real_time() - dt;
+  int v1 = !fmt->limit || (dt < fmt->limit);
+
+  tm_format_real_time(x, v1 ? fmt->fmt1 : fmt->fmt2, rt);
+}
+
+/* Replace %f in format string with usec scaled to requested precision */
+static int
+strfusec(char *buf, int size, const char *fmt, uint usec)
+{
+  char *str = buf;
+  int parity = 0;
+
+  while (*fmt)
+  {
+    if (!size)
+      return 0;
+
+    if ((fmt[0] == '%') && (!parity) &&
+	((fmt[1] == 'f') || (fmt[1] >= '1') && (fmt[1] <= '6') && (fmt[2] == 'f')))
+    {
+      int digits = (fmt[1] == 'f') ? 6 : (fmt[1] - '0');
+      uint d = digits, u = usec;
+
+      /* Convert microseconds to requested precision */
+      while (d++ < 6)
+	u /= 10;
+
+      int num = bsnprintf(str, size, "%0*u", digits, u);
+      if (num < 0)
+	return 0;
+
+      fmt += (fmt[1] == 'f') ? 2 : 3;
+      ADVANCE(str, size, num);
+    }
+    else
+    {
+      /* Handle '%%' expression */
+      parity = (*fmt == '%') ? !parity : 0;
+      *str++ = *fmt++;
+      size--;
+    }
+  }
+
+  if (!size)
+    return 0;
+
+  *str = 0;
+  return str - buf;
+}
+
+void
+tm_format_real_time(char *x, const char *fmt, btime t)
+{
+  s64 t1 = t TO_S;
+  s64 t2 = t - t1 S;
+
+  time_t ts = t1;
+  struct tm tm;
+  if (!localtime_r(&ts, &tm))
+    goto err;
+
+  byte tbuf[TM_DATETIME_BUFFER_SIZE];
+  if (!strfusec(tbuf, TM_DATETIME_BUFFER_SIZE, fmt, t2))
+    goto err;
+
+  if (!strftime(x, TM_DATETIME_BUFFER_SIZE, tbuf, &tm))
+    goto err;
+
+  return;
+
+err:
+  strcpy(x, "<error>");
+}

lib/timer.h

@@ -0,0 +1,127 @@
+/*
+ *	BIRD -- Timers
+ *
+ *	(c) 2013--2017 Ondrej Zajicek <santiago@crfreenet.org>
+ *	(c) 2013--2017 CZ.NIC z.s.p.o.
+ *
+ *	Can be freely distributed and used under the terms of the GNU GPL.
+ */
+
+#ifndef _BIRD_TIMER_H_
+#define _BIRD_TIMER_H_
+
+#include "nest/bird.h"
+#include "lib/buffer.h"
+#include "lib/resource.h"
+
+
+typedef struct timer
+{
+  resource r;
+  void (*hook)(struct timer *);
+  void *data;
+
+  btime expires;			/* 0=inactive */
+  uint randomize;			/* Amount of randomization */
+  uint recurrent;			/* Timer recurrence */
+
+  int index;
+} timer;
+
+struct timeloop
+{
+  BUFFER_(timer *) timers;
+  btime last_time;
+  btime real_time;
+};
+
+static inline uint timers_count(struct timeloop *loop)
+{ return loop->timers.used - 1; }
+
+static inline timer *timers_first(struct timeloop *loop)
+{ return (loop->timers.used > 1) ? loop->timers.data[1] : NULL; }
+
+extern struct timeloop main_timeloop;
+
+btime current_time(void);
+btime current_real_time(void);
+
+//#define now (current_time() TO_S)
+//#define now_real (current_real_time() TO_S)
+extern btime boot_time;
+
+timer *tm_new(pool *p);
+void tm_set(timer *t, btime when);
+void tm_start(timer *t, btime after);
+void tm_stop(timer *t);
+
+static inline int
+tm_active(timer *t)
+{
+  return t->expires != 0;
+}
+
+static inline btime
+tm_remains(timer *t)
+{
+  btime now_ = current_time();
+  return (t->expires > now_) ? (t->expires - now_) : 0;
+}
+
+static inline timer *
+tm_new_init(pool *p, void (*hook)(struct timer *), void *data, uint rec, uint rand)
+{
+  timer *t = tm_new(p);
+  t->hook = hook;
+  t->data = data;
+  t->recurrent = rec;
+  t->randomize = rand;
+  return t;
+}
+
+static inline void
+tm_set_max(timer *t, btime when)
+{
+  if (when > t->expires)
+    tm_set(t, when);
+}
+
+static inline void
+tm_start_max(timer *t, btime after)
+{
+  btime rem = tm_remains(t);
+  tm_start(t, MAX_(rem, after));
+}
+
+/* In sysdep code */
+void times_init(struct timeloop *loop);
+void times_update(struct timeloop *loop);
+void times_update_real_time(struct timeloop *loop);
+
+/* For I/O loop */
+void timers_init(struct timeloop *loop, pool *p);
+void timers_fire(struct timeloop *loop);
+
+void timer_init(void);
+
+
+struct timeformat {
+  char *fmt1, *fmt2;
+  btime limit;
+};
+
+#define TM_ISO_SHORT_S	(struct timeformat){"%T",     "%F", (s64) (20*3600) S_}
+#define TM_ISO_SHORT_MS	(struct timeformat){"%T.%3f", "%F", (s64) (20*3600) S_}
+#define TM_ISO_SHORT_US	(struct timeformat){"%T.%6f", "%F", (s64) (20*3600) S_}
+
+#define TM_ISO_LONG_S	(struct timeformat){"%F %T",     NULL, 0}
+#define TM_ISO_LONG_MS	(struct timeformat){"%F %T.%3f", NULL, 0}
+#define TM_ISO_LONG_US	(struct timeformat){"%F %T.%6f", NULL, 0}
+
+#define TM_DATETIME_BUFFER_SIZE 32	/* Buffer size required by tm_format_time() */
+
+btime tm_parse_time(char *x);
+void tm_format_time(char *x, struct timeformat *fmt, btime t);
+void tm_format_real_time(char *x, const char *fmt, btime t);
+
+#endif

misc/docker/centos-6-amd64/Dockerfile

@@ -0,0 +1,11 @@
+FROM centos:6
+RUN yum -y upgrade
+RUN yum -y install \
+	autoconf \
+	flex \
+	bison \
+	pkgconfig \
+	'readline-devel' \
+	'pkgconfig(ncurses)' \
+	gcc \
+	make

misc/docker/centos-7-amd64/Dockerfile

@@ -0,0 +1,11 @@
+FROM centos:7
+RUN yum -y upgrade
+RUN yum -y install \
+	autoconf \
+	flex \
+	bison \
+	pkgconfig \
+	'readline-devel' \
+	'pkgconfig(ncurses)' \
+	gcc \
+	make

misc/docker/debian-7-amd64/Dockerfile

@@ -0,0 +1,12 @@
+FROM debian:wheezy-slim
+ENV DEBIAN_FRONTEND noninteractive
+RUN sed -i 's/deb.debian.org/ftp.cz.debian.org/' /etc/apt/sources.list
+RUN apt-get -y update
+RUN apt-get -y upgrade
+RUN apt-get -y install \
+	autoconf \
+	build-essential \
+	flex \
+	bison \
+	ncurses-dev \
+	libreadline-dev

misc/docker/debian-7-i386/Dockerfile

@@ -0,0 +1,12 @@
+FROM i386/debian:wheezy-slim
+ENV DEBIAN_FRONTEND noninteractive
+RUN sed -i 's/deb.debian.org/ftp.cz.debian.org/' /etc/apt/sources.list
+RUN apt-get -y update
+RUN apt-get -y upgrade
+RUN apt-get -y install \
+	autoconf \
+	build-essential \
+	flex \
+	bison \
+	ncurses-dev \
+	libreadline-dev

misc/docker/debian-8-amd64/Dockerfile

@@ -0,0 +1,12 @@
+FROM debian:jessie-slim
+ENV DEBIAN_FRONTEND noninteractive
+RUN sed -i 's/deb.debian.org/ftp.cz.debian.org/' /etc/apt/sources.list
+RUN apt-get -y update
+RUN apt-get -y upgrade
+RUN apt-get -y install \
+	autoconf \
+	build-essential \
+	flex \
+	bison \
+	ncurses-dev \
+	libreadline-dev

misc/docker/debian-8-i386/Dockerfile

@@ -0,0 +1,12 @@
+FROM i386/debian:jessie-slim
+ENV DEBIAN_FRONTEND noninteractive
+RUN sed -i 's/deb.debian.org/ftp.cz.debian.org/' /etc/apt/sources.list
+RUN apt-get -y update
+RUN apt-get -y upgrade
+RUN apt-get -y install \
+	autoconf \
+	build-essential \
+	flex \
+	bison \
+	ncurses-dev \
+	libreadline-dev

misc/docker/debian-9-amd64/Dockerfile

@@ -0,0 +1,12 @@
+FROM debian:stretch-slim
+ENV DEBIAN_FRONTEND noninteractive
+RUN sed -i 's/deb.debian.org/ftp.cz.debian.org/' /etc/apt/sources.list
+RUN apt-get -y update
+RUN apt-get -y upgrade
+RUN apt-get -y install \
+	autoconf \
+	build-essential \
+	flex \
+	bison \
+	ncurses-dev \
+	libreadline-dev

misc/docker/debian-9-i386/Dockerfile

@@ -0,0 +1,12 @@
+FROM i386/debian:stretch-slim
+ENV DEBIAN_FRONTEND noninteractive
+RUN sed -i 's/deb.debian.org/ftp.cz.debian.org/' /etc/apt/sources.list
+RUN apt-get -y update
+RUN apt-get -y upgrade
+RUN apt-get -y install \
+	autoconf \
+	build-essential \
+	flex \
+	bison \
+	ncurses-dev \
+	libreadline-dev

misc/docker/debian-testing-amd64/Dockerfile

@@ -0,0 +1,12 @@
+FROM debian:testing-slim
+ENV DEBIAN_FRONTEND noninteractive
+RUN sed -i 's/deb.debian.org/ftp.cz.debian.org/' /etc/apt/sources.list
+RUN apt-get -y update
+RUN apt-get -y upgrade
+RUN apt-get -y install \
+	autoconf \
+	build-essential \
+	flex \
+	bison \
+	ncurses-dev \
+	libreadline-dev

misc/docker/debian-testing-i386/Dockerfile

@@ -0,0 +1,12 @@
+FROM i386/debian:testing-slim
+ENV DEBIAN_FRONTEND noninteractive
+RUN sed -i 's/deb.debian.org/ftp.cz.debian.org/' /etc/apt/sources.list
+RUN apt-get -y update
+RUN apt-get -y upgrade
+RUN apt-get -y install \
+	autoconf \
+	build-essential \
+	flex \
+	bison \
+	ncurses-dev \
+	libreadline-dev

misc/docker/fedora-25-amd64/Dockerfile

@@ -0,0 +1,10 @@
+FROM fedora:25
+RUN dnf -y upgrade
+RUN dnf -y install \
+	autoconf \
+	flex \
+	bison \
+	pkgconfig \
+	'readline-devel' \
+	'pkgconfig(ncurses)' \
+	gcc

misc/docker/fedora-26-amd64/Dockerfile

@@ -0,0 +1,10 @@
+FROM fedora:26
+RUN dnf -y upgrade
+RUN dnf -y install \
+	autoconf \
+	flex \
+	bison \
+	pkgconfig \
+	'readline-devel' \
+	'pkgconfig(ncurses)' \
+	gcc

misc/docker/opensuse-42.3-amd64/Dockerfile

@@ -0,0 +1,11 @@
+FROM opensuse:42.3
+RUN zypper -n up
+RUN zypper -n install \
+	autoconf \
+	flex \
+	bison \
+	pkgconfig \
+	readline-devel \
+	ncurses-devel \
+	gcc \
+	gmake

misc/docker/ubuntu-14.04-amd64/Dockerfile

@@ -0,0 +1,12 @@
+FROM ubuntu:14.04
+ENV DEBIAN_FRONTEND noninteractive
+RUN sed -i 's/deb.debian.org/ftp.cz.debian.org/' /etc/apt/sources.list
+RUN apt-get -y update
+RUN apt-get -y upgrade
+RUN apt-get -y install \
+	autoconf \
+	build-essential \
+	flex \
+	bison \
+	ncurses-dev \
+	libreadline-dev

misc/docker/ubuntu-16.04-amd64/Dockerfile

@@ -0,0 +1,12 @@
+FROM ubuntu:16.04
+ENV DEBIAN_FRONTEND noninteractive
+RUN sed -i 's/deb.debian.org/ftp.cz.debian.org/' /etc/apt/sources.list
+RUN apt-get -y update
+RUN apt-get -y upgrade
+RUN apt-get -y install \
+	autoconf \
+	build-essential \
+	flex \
+	bison \
+	ncurses-dev \
+	libreadline-dev

nest/cmds.c

@@ -27,12 +27,12 @@ cmd_show_status(void)
   byte tim[TM_DATETIME_BUFFER_SIZE];
 
   cli_msg(-1000, "BIRD " BIRD_VERSION);
-  tm_format_datetime(tim, &config->tf_base, now);
+  tm_format_time(tim, &config->tf_base, current_time());
   cli_msg(-1011, "Router ID is %R", config->router_id);
   cli_msg(-1011, "Current server time is %s", tim);
-  tm_format_datetime(tim, &config->tf_base, boot_time);
+  tm_format_time(tim, &config->tf_base, boot_time);
   cli_msg(-1011, "Last reboot on %s", tim);
-  tm_format_datetime(tim, &config->tf_base, config->load_time);
+  tm_format_time(tim, &config->tf_base, config->load_time);
   cli_msg(-1011, "Last reconfiguration on %s", tim);
 
   graceful_restart_show_status();

nest/config.Y

@@ -65,14 +65,15 @@ proto_postconfig(void)
 CF_DECLS
 
 CF_KEYWORDS(ROUTER, ID, PROTOCOL, TEMPLATE, PREFERENCE, DISABLED, DEBUG, ALL, OFF, DIRECT)
-CF_KEYWORDS(INTERFACE, IMPORT, EXPORT, FILTER, NONE, TABLE, STATES, ROUTES, FILTERS)
+CF_KEYWORDS(INTERFACE, IMPORT, EXPORT, FILTER, NONE, VRF, TABLE, STATES, ROUTES, FILTERS)
 CF_KEYWORDS(IPV4, IPV6, VPN4, VPN6, ROA4, ROA6)
 CF_KEYWORDS(RECEIVE, LIMIT, ACTION, WARN, BLOCK, RESTART, DISABLE, KEEP, FILTERED)
 CF_KEYWORDS(PASSWORD, FROM, PASSIVE, TO, ID, EVENTS, PACKETS, PROTOCOLS, INTERFACES)
 CF_KEYWORDS(ALGORITHM, KEYED, HMAC, MD5, SHA1, SHA256, SHA384, SHA512)
 CF_KEYWORDS(PRIMARY, STATS, COUNT, BY, FOR, COMMANDS, PREEXPORT, NOEXPORT, GENERATE)
-CF_KEYWORDS(LISTEN, BGP, V6ONLY, DUAL, ADDRESS, PORT, PASSWORDS, DESCRIPTION, SORTED)
+CF_KEYWORDS(BGP, PASSWORDS, DESCRIPTION, SORTED)
 CF_KEYWORDS(RELOAD, IN, OUT, MRTDUMP, MESSAGES, RESTRICT, MEMORY, IGP_METRIC, CLASS, DSCP)
+CF_KEYWORDS(TIMEFORMAT, ISO, SHORT, LONG, ROUTE, PROTOCOL, BASE, LOG, S, MS, US)
 CF_KEYWORDS(GRACEFUL, RESTART, WAIT, MAX, FLUSH, AS)
 CF_KEYWORDS(REFRESH)
 
@@ -124,24 +125,6 @@ idval:
    }
  ;
 
-
-CF_ADDTO(conf, listen)
-
-listen: LISTEN BGP listen_opts ';' ;
-
-listen_opts:
-   /* Nothing */
- | listen_opts listen_opt
- ;
-
-listen_opt:
-   ADDRESS ipa { new_config->listen_bgp_addr = $2; }
- | PORT expr { new_config->listen_bgp_port = $2; }
- | V6ONLY { new_config->listen_bgp_flags = 0; }
- | DUAL { new_config->listen_bgp_flags = 1; }
- ;
-
-
 CF_ADDTO(conf, gr_opts)
 
 gr_opts: GRACEFUL RESTART WAIT expr ';' { new_config->gr_wait = $4; } ;
@@ -158,6 +141,7 @@ net_type:
  | ROA6 { $$ = NET_ROA6; }
  | FLOW4{ $$ = NET_FLOW4; }
  | FLOW6{ $$ = NET_FLOW6; }
+ | MPLS { $$ = NET_MPLS; }
  ;
 
 CF_ENUM(T_ENUM_NETTYPE, NET_, IP4, IP6, VPN4, VPN6, ROA4, ROA6, FLOW4, FLOW6)
@@ -225,6 +209,7 @@ proto_item:
  | MRTDUMP mrtdump_mask { this_proto->mrtdump = $2; }
  | ROUTER ID idval { this_proto->router_id = $3; }
  | DESCRIPTION text { this_proto->dsc = $2; }
+ | VRF text { this_proto->vrf = if_get_by_name($2); }
  ;
 
 
@@ -296,6 +281,7 @@ limit_spec:
  | OFF { $$ = (struct channel_limit){}; }
  ;
 
+
 CF_ADDTO(conf, debug_default)
 
 debug_default:
@@ -305,6 +291,31 @@ debug_default:
 
 /* MRTDUMP PROTOCOLS is in systep/unix/config.Y */
 
+CF_ADDTO(conf, timeformat_base)
+
+timeformat_which:
+   ROUTE { $$ = &new_config->tf_route; }
+ | PROTOCOL { $$ = &new_config->tf_proto; }
+ | BASE { $$ = &new_config->tf_base; }
+ | LOG { $$ = &new_config->tf_log; }
+ ;
+
+timeformat_spec:
+   timeformat_which TEXT { *$1 = (struct timeformat){$2, NULL, 0}; }
+ | timeformat_which TEXT expr TEXT { *$1 = (struct timeformat){$2, $4, (s64) $3 S_}; }
+ | timeformat_which ISO SHORT    { *$1 = TM_ISO_SHORT_S; }
+ | timeformat_which ISO SHORT MS { *$1 = TM_ISO_SHORT_MS; }
+ | timeformat_which ISO SHORT US { *$1 = TM_ISO_SHORT_US; }
+ | timeformat_which ISO LONG    { *$1 = TM_ISO_LONG_S; }
+ | timeformat_which ISO LONG MS { *$1 = TM_ISO_LONG_MS; }
+ | timeformat_which ISO LONG US { *$1 = TM_ISO_LONG_US; }
+ ;
+
+timeformat_base:
+   TIMEFORMAT timeformat_spec ';'
+ ;
+
+
 /* Interface patterns */
 
 iface_patt_node_init:
@@ -463,12 +474,12 @@ password_item_begin:
 
 password_item_params:
    /* empty */ { }
- | GENERATE FROM datetime ';' password_item_params { this_p_item->genfrom = $3; }
+ | GENERATE FROM time ';' password_item_params { this_p_item->genfrom = $3; }
- | GENERATE TO datetime ';' password_item_params { this_p_item->gento = $3; }
+ | GENERATE TO time ';' password_item_params { this_p_item->gento = $3; }
- | ACCEPT FROM datetime ';' password_item_params { this_p_item->accfrom = $3; }
+ | ACCEPT FROM time ';' password_item_params { this_p_item->accfrom = $3; }
- | ACCEPT TO datetime ';' password_item_params { this_p_item->accto = $3; }
+ | ACCEPT TO time ';' password_item_params { this_p_item->accto = $3; }
- | FROM datetime ';' password_item_params { this_p_item->genfrom = this_p_item->accfrom = $2; }
+ | FROM time ';' password_item_params { this_p_item->genfrom = this_p_item->accfrom = $2; }
- | TO datetime ';' password_item_params { this_p_item->gento = this_p_item->accto = $2; }
+ | TO time ';' password_item_params { this_p_item->gento = this_p_item->accto = $2; }
  | ID expr ';' password_item_params { this_p_item->id = $2; if ($2 <= 0) cf_error("Password ID has to be greated than zero."); }
  | ALGORITHM password_algorithm ';' password_item_params { this_p_item->alg = $2; }
  ;
@@ -747,12 +758,12 @@ echo_size:
    }
  ;
 
-CF_CLI(DISABLE, proto_patt, <protocol> | \"<pattern>\" | all, [[Disable protocol]], CLI_SF_PROTOCOL | CLI_SF_KW_ALL)
+CF_CLI(DISABLE, proto_patt opttext, (<protocol> | \"<pattern>\" | all) [message], [[Disable protocol]], CLI_SF_PROTOCOL | CLI_SF_KW_ALL)
-{ proto_apply_cmd($2, proto_cmd_disable, 1, 0); } ;
+{ proto_apply_cmd($2, proto_cmd_disable, 1, (uintptr_t) $3); } ;
-CF_CLI(ENABLE, proto_patt, <protocol> | \"<pattern>\" | all, [[Enable protocol]], CLI_SF_PROTOCOL | CLI_SF_KW_ALL)
+CF_CLI(ENABLE, proto_patt opttext, (<protocol> | \"<pattern>\" | all) [message], [[Enable protocol]], CLI_SF_PROTOCOL | CLI_SF_KW_ALL)
-{ proto_apply_cmd($2, proto_cmd_enable, 1, 0); } ;
+{ proto_apply_cmd($2, proto_cmd_enable, 1, (uintptr_t) $3); } ;
-CF_CLI(RESTART, proto_patt, <protocol> | \"<pattern>\" | all, [[Restart protocol]], CLI_SF_PROTOCOL | CLI_SF_KW_ALL)
+CF_CLI(RESTART, proto_patt opttext, (<protocol> | \"<pattern>\" | all) [message], [[Restart protocol]], CLI_SF_PROTOCOL | CLI_SF_KW_ALL)
-{ proto_apply_cmd($2, proto_cmd_restart, 1, 0); } ;
+{ proto_apply_cmd($2, proto_cmd_restart, 1, (uintptr_t) $3); } ;
 CF_CLI(RELOAD, proto_patt, <protocol> | \"<pattern>\" | all, [[Reload protocol]], CLI_SF_PROTOCOL | CLI_SF_KW_ALL)
 { proto_apply_cmd($2, proto_cmd_reload, 1, CMD_RELOAD); } ;
 CF_CLI(RELOAD IN, proto_patt, <protocol> | \"<pattern>\" | all, [[Reload protocol (just imported routes)]], CLI_SF_PROTOCOL | CLI_SF_KW_ALL)

nest/iface.c

@@ -32,11 +32,14 @@
 #include "lib/resource.h"
 #include "lib/string.h"
 #include "conf/conf.h"
+#include "sysdep/unix/krt.h"
 
 static pool *if_pool;
 
 list iface_list;
 
+static void if_recalc_preferred(struct iface *i);
+
 /**
  * ifa_dump - dump interface address
  * @a: interface address descriptor
@@ -46,10 +49,11 @@ list iface_list;
 void
 ifa_dump(struct ifa *a)
 {
-  debug("\t%I, net %N bc %I -> %I%s%s%s\n", a->ip, &a->prefix, a->brd, a->opposite,
+  debug("\t%I, net %N bc %I -> %I%s%s%s%s\n", a->ip, &a->prefix, a->brd, a->opposite,
-	(a->flags & IF_UP) ? "" : " DOWN",
+	(a->flags & IA_PRIMARY) ? " PRIMARY" : "",
-	(a->flags & IA_PRIMARY) ? "" : " SEC",
+	(a->flags & IA_SECONDARY) ? " SEC" : "",
-	(a->flags & IA_PEER) ? "PEER" : "");
+	(a->flags & IA_HOST) ? " HOST" : "",
+	(a->flags & IA_PEER) ? " PEER" : "");
 }
 
 /**
@@ -116,7 +120,7 @@ if_what_changed(struct iface *i, struct iface *j)
   unsigned c;
 
   if (((i->flags ^ j->flags) & ~(IF_UP | IF_SHUTDOWN | IF_UPDATED | IF_ADMIN_UP | IF_LINK_UP | IF_TMP_DOWN | IF_JUST_CREATED))
-      || i->index != j->index)
+      || (i->index != j->index) || (i->master != j->master))
     return IF_CHANGE_TOO_MUCH;
   c = 0;
   if ((i->flags ^ j->flags) & IF_UP)
@@ -133,17 +137,21 @@ if_copy(struct iface *to, struct iface *from)
 {
   to->flags = from->flags | (to->flags & IF_TMP_DOWN);
   to->mtu = from->mtu;
+  to->master_index = from->master_index;
+  to->master = from->master;
 }
 
 static inline void
 ifa_send_notify(struct proto *p, unsigned c, struct ifa *a)
 {
-  if (p->ifa_notify && (p->proto_state != PS_DOWN))
+  if (p->ifa_notify &&
+      (p->proto_state != PS_DOWN) &&
+      (!p->vrf || p->vrf == a->iface->master))
     {
       if (p->debug & D_IFACES)
-	log(L_TRACE "%s < %s address %N on interface %s %s",
+	log(L_TRACE "%s < address %N on interface %s %s",
-	    p->name, (a->flags & IA_PRIMARY) ? "primary" : "secondary",
+	    p->name, &a->prefix, a->iface->name,
-	    &a->prefix, a->iface->name, (c & IF_CHANGE_UP) ? "added" : "removed");
+	    (c & IF_CHANGE_UP) ? "added" : "removed");
       p->ifa_notify(p, c, a);
     }
 }
@@ -174,7 +182,9 @@ ifa_notify_change(unsigned c, struct ifa *a)
 static inline void
 if_send_notify(struct proto *p, unsigned c, struct iface *i)
 {
-  if (p->if_notify && (p->proto_state != PS_DOWN))
+  if (p->if_notify &&
+      (p->proto_state != PS_DOWN) &&
+      (!p->vrf || p->vrf == i->master))
     {
       if (p->debug & D_IFACES)
 	log(L_TRACE "%s < interface %s %s", p->name, i->name,
@@ -182,6 +192,7 @@ if_send_notify(struct proto *p, unsigned c, struct iface *i)
 	    (c & IF_CHANGE_DOWN) ? "goes down" :
 	    (c & IF_CHANGE_MTU) ? "changes MTU" :
 	    (c & IF_CHANGE_LINK) ? "changes link" :
+	    (c & IF_CHANGE_PREFERRED) ? "changes preferred address" :
 	    (c & IF_CHANGE_CREATE) ? "created" :
 	    "sends unknown event");
       p->if_notify(p, c, i);
@@ -210,10 +221,7 @@ if_notify_change(unsigned c, struct iface *i)
     neigh_if_down(i);
 
     WALK_LIST(a, i->addrs)
-      {
+      ifa_notify_change_(IF_CHANGE_DOWN, a);
-	a->flags = (i->flags & ~IA_FLAGS) | (a->flags & IA_FLAGS);
-	ifa_notify_change_(IF_CHANGE_DOWN, a);
-      }
 
     cli_notify_all_clients();
   }
@@ -224,10 +232,7 @@ if_notify_change(unsigned c, struct iface *i)
   if (c & IF_CHANGE_UP)
   {
     WALK_LIST(a, i->addrs)
-      {
+      ifa_notify_change_(IF_CHANGE_UP, a);
-	a->flags = (i->flags & ~IA_FLAGS) | (a->flags & IA_FLAGS);
-	ifa_notify_change_(IF_CHANGE_UP, a);
-      }
 
     neigh_if_up(i);
     cli_notify_all_clients();
@@ -237,24 +242,25 @@ if_notify_change(unsigned c, struct iface *i)
     neigh_if_link(i);
 }
 
-static unsigned
+static uint
-if_recalc_flags(struct iface *i, unsigned flags)
+if_recalc_flags(struct iface *i UNUSED, uint flags)
 {
-  if ((flags & (IF_SHUTDOWN | IF_TMP_DOWN)) ||
+  if ((flags & IF_ADMIN_UP) &&
-      !(flags & IF_ADMIN_UP) ||
+      !(flags & (IF_SHUTDOWN | IF_TMP_DOWN)) &&
-      !i->addr)
+      !(i->master_index && !i->master))
-    flags &= ~IF_UP;
-  else
     flags |= IF_UP;
+  else
+    flags &= ~IF_UP;
+
   return flags;
 }
 
 static void
-if_change_flags(struct iface *i, unsigned flags)
+if_change_flags(struct iface *i, uint flags)
 {
-  unsigned of = i->flags;
+  uint of = i->flags;
-
   i->flags = if_recalc_flags(i, flags);
+
   if ((i->flags ^ of) & IF_UP)
     if_notify_change((i->flags & IF_UP) ? IF_CHANGE_UP : IF_CHANGE_DOWN, i);
 }
@@ -302,7 +308,6 @@ if_update(struct iface *new)
   WALK_LIST(i, iface_list)
     if (!strcmp(new->name, i->name))
       {
-	new->addr = i->addr;
 	new->flags = if_recalc_flags(new, new->flags);
 	c = if_what_changed(i, new);
 	if (c & IF_CHANGE_TOO_MUCH)	/* Changed a lot, convert it to down/up */
@@ -310,10 +315,13 @@ if_update(struct iface *new)
 	    DBG("Interface %s changed too much -- forcing down/up transition\n", i->name);
 	    if_change_flags(i, i->flags | IF_TMP_DOWN);
 	    rem_node(&i->n);
-	    new->addr = i->addr;
+	    new->addr4 = i->addr4;
+	    new->addr6 = i->addr6;
+	    new->llv6 = i->llv6;
+	    new->sysdep = i->sysdep;
 	    memcpy(&new->addrs, &i->addrs, sizeof(i->addrs));
 	    memcpy(i, new, sizeof(*i));
-	    i->flags &= ~IF_UP; /* IF_TMP_DOWN will be added later */
+	    i->flags &= ~IF_UP; 	/* IF_TMP_DOWN will be added later */
 	    goto newif;
 	  }
 
@@ -344,13 +352,16 @@ if_start_update(void)
     {
       i->flags &= ~IF_UPDATED;
       WALK_LIST(a, i->addrs)
-	a->flags &= ~IF_UPDATED;
+	a->flags &= ~IA_UPDATED;
     }
 }
 
 void
 if_end_partial_update(struct iface *i)
 {
+  if (i->flags & IF_NEEDS_RECALC)
+    if_recalc_preferred(i);
+
   if (i->flags & IF_TMP_DOWN)
     if_change_flags(i, i->flags & ~IF_TMP_DOWN);
 }
@@ -368,7 +379,7 @@ if_end_update(void)
       else
 	{
 	  WALK_LIST_DELSAFE(a, b, i->addrs)
-	    if (!(a->flags & IF_UPDATED))
+	    if (!(a->flags & IA_UPDATED))
 	      ifa_delete(a);
 	  if_end_partial_update(i);
 	}
@@ -465,40 +476,99 @@ if_get_by_name(char *name)
   return i;
 }
 
-struct ifa *kif_choose_primary(struct iface *i);
+static inline void
-
+if_set_preferred(struct ifa **pos, struct ifa *new)
-static int
-ifa_recalc_primary(struct iface *i)
 {
-  struct ifa *a = kif_choose_primary(i);
+  if (*pos)
+    (*pos)->flags &= ~IA_PRIMARY;
+  if (new)
+    new->flags |= IA_PRIMARY;
 
-  if (a == i->addr)
+  *pos = new;
-    return 0;
+}
 
-  if (i->addr)
+static void
-    i->addr->flags &= ~IA_PRIMARY;
+if_recalc_preferred(struct iface *i)
+{
+  /*
+   * Preferred address selection priority:
+   * 1) Address configured in Device protocol
+   * 2) Sysdep IPv4 address (BSD)
+   * 3) Old preferred address
+   * 4) First address in list
+   */
 
-  if (a)
+  struct kif_iface_config *ic = kif_get_iface_config(i);
+  struct ifa *a4 = i->addr4, *a6 = i->addr6, *ll = i->llv6;
+  ip_addr pref_v4 = ic->pref_v4;
+  uint change = 0;
+
+  if (kif_update_sysdep_addr(i))
+    change |= IF_CHANGE_SYSDEP;
+
+  /* BSD sysdep address */
+  if (ipa_zero(pref_v4) && ip4_nonzero(i->sysdep))
+    pref_v4 = ipa_from_ip4(i->sysdep);
+
+  struct ifa *a;
+  WALK_LIST(a, i->addrs)
     {
-      a->flags |= IA_PRIMARY;
+      /* Secondary address is never selected */
-      rem_node(&a->n);
+      if (a->flags & IA_SECONDARY)
-      add_head(&i->addrs, &a->n);
+	continue;
+
+      if (ipa_is_ip4(a->ip)) {
+	if (!a4 || ipa_equal(a->ip, pref_v4))
+	  a4 = a;
+      } else if (!ipa_is_link_local(a->ip)) {
+	if (!a6 || ipa_equal(a->ip, ic->pref_v6))
+	  a6 = a;
+      } else {
+	if (!ll || ipa_equal(a->ip, ic->pref_ll))
+	  ll = a;
+      }
     }
 
-  i->addr = a;
+  if (a4 != i->addr4)
-  return 1;
+  {
+    if_set_preferred(&i->addr4, a4);
+    change |= IF_CHANGE_ADDR4;
+  }
+
+  if (a6 != i->addr6)
+  {
+    if_set_preferred(&i->addr6, a6);
+    change |= IF_CHANGE_ADDR6;
+  }
+
+  if (ll != i->llv6)
+  {
+    if_set_preferred(&i->llv6, ll);
+    change |= IF_CHANGE_LLV6;
+  }
+
+  i->flags &= ~IF_NEEDS_RECALC;
+
+  /*
+   * FIXME: There should be proper notification instead of iface restart:
+   * if_notify_change(change, i)
+   */
+  if (change)
+    if_change_flags(i, i->flags | IF_TMP_DOWN);
 }
 
 void
-ifa_recalc_all_primary_addresses(void)
+if_recalc_all_preferred_addresses(void)
 {
   struct iface *i;
 
   WALK_LIST(i, iface_list)
-    {
+  {
-      if (ifa_recalc_primary(i))
+    if_recalc_preferred(i);
-	if_change_flags(i, i->flags | IF_TMP_DOWN);
+
-    }
+    if (i->flags & IF_TMP_DOWN)
+      if_change_flags(i, i->flags & ~IF_TMP_DOWN);
+  }
 }
 
 static inline int
@@ -530,7 +600,7 @@ ifa_update(struct ifa *a)
 	    b->scope == a->scope &&
 	    !((b->flags ^ a->flags) & IA_PEER))
 	  {
-	    b->flags |= IF_UPDATED;
+	    b->flags |= IA_UPDATED;
 	    return b;
 	  }
 	ifa_delete(b);
@@ -538,15 +608,15 @@ ifa_update(struct ifa *a)
       }
 
   if ((a->prefix.type == NET_IP4) && (i->flags & IF_BROADCAST) && ipa_zero(a->brd))
-    log(L_ERR "Missing broadcast address for interface %s", i->name);
+    log(L_WARN "Missing broadcast address for interface %s", i->name);
 
   b = mb_alloc(if_pool, sizeof(struct ifa));
   memcpy(b, a, sizeof(struct ifa));
   add_tail(&i->addrs, &b->n);
-  b->flags = (i->flags & ~IA_FLAGS) | (a->flags & IA_FLAGS);
+  b->flags |= IA_UPDATED;
-  if (ifa_recalc_primary(i))
+
-    if_change_flags(i, i->flags | IF_TMP_DOWN);
+  i->flags |= IF_NEEDS_RECALC;
-  if (b->flags & IF_UP)
+  if (i->flags & IF_UP)
     ifa_notify_change(IF_CHANGE_CREATE | IF_CHANGE_UP, b);
   return b;
 }
@@ -569,16 +639,24 @@ ifa_delete(struct ifa *a)
     if (ifa_same(b, a))
       {
 	rem_node(&b->n);
-	if (b->flags & IF_UP)
+
-	  {
-	    b->flags &= ~IF_UP;
-	    ifa_notify_change(IF_CHANGE_DOWN, b);
-	  }
 	if (b->flags & IA_PRIMARY)
 	  {
-	    if_change_flags(i, i->flags | IF_TMP_DOWN);
+	    /*
-	    ifa_recalc_primary(i);
+	     * We unlink deleted preferred address and mark for recalculation.
+	     * FIXME: This could break if we make iface scan non-atomic, as
+	     * protocols still could use the freed address until they get
+	     * if_notify from preferred route recalculation.
+	     */
+	    if (b == i->addr4) i->addr4 = NULL;
+	    if (b == i->addr6) i->addr6 = NULL;
+	    if (b == i->llv6) i->llv6 = NULL;
+	    i->flags |= IF_NEEDS_RECALC;
 	  }
+
+	if (i->flags & IF_UP)
+	  ifa_notify_change(IF_CHANGE_DOWN, b);
+
 	mb_free(b);
 	return;
       }
@@ -745,16 +823,17 @@ iface_patts_equal(list *a, list *b, int (*comp)(struct iface_patt *, struct ifac
 static void
 if_show_addr(struct ifa *a)
 {
-  byte opp[IPA_MAX_TEXT_LENGTH + 16];
+  byte *flg, opp[IPA_MAX_TEXT_LENGTH + 16];
+
+  flg = (a->flags & IA_PRIMARY) ? "Preferred, " : (a->flags & IA_SECONDARY) ? "Secondary, " : "";
 
   if (ipa_nonzero(a->opposite))
-    bsprintf(opp, ", opposite %I", a->opposite);
+    bsprintf(opp, "opposite %I, ", a->opposite);
   else
     opp[0] = 0;
-  cli_msg(-1003, "\t%I/%d (%s%s, scope %s)",
+
-	  a->ip, a->prefix.pxlen,
+  cli_msg(-1003, "\t%I/%d (%s%sscope %s)",
-	  (a->flags & IA_PRIMARY) ? "Primary" : (a->flags & IA_SECONDARY) ? "Secondary" : "Unselected",
+	  a->ip, a->prefix.pxlen, flg, opp, ip_scope_text(a->scope));
-	  opp, ip_scope_text(a->scope));
 }
 
 void
@@ -769,7 +848,13 @@ if_show(void)
       if (i->flags & IF_SHUTDOWN)
 	continue;
 
-      cli_msg(-1001, "%s %s (index=%d)", i->name, (i->flags & IF_UP) ? "up" : "DOWN", i->index);
+      char mbuf[16 + sizeof(i->name)] = {};
+      if (i->master)
+	bsprintf(mbuf, " master=%s", i->master->name);
+      else if (i->master_index)
+	bsprintf(mbuf, " master=#%u", i->master_index);
+
+      cli_msg(-1001, "%s %s (index=%d%s)", i->name, (i->flags & IF_UP) ? "up" : "down", i->index, mbuf);
       if (!(i->flags & IF_MULTIACCESS))
 	type = "PtP";
       else
@@ -783,10 +868,13 @@ if_show(void)
 	      (i->flags & IF_LOOPBACK) ? " Loopback" : "",
 	      (i->flags & IF_IGNORE) ? " Ignored" : "",
 	      i->mtu);
-      if (i->addr)
+
-	if_show_addr(i->addr);
       WALK_LIST(a, i->addrs)
-	if (a != i->addr)
+	if (a->prefix.type == NET_IP4)
+	  if_show_addr(a);
+
+      WALK_LIST(a, i->addrs)
+	if (a->prefix.type == NET_IP6)
 	  if_show_addr(a);
     }
   cli_msg(0, "");
@@ -796,16 +884,25 @@ void
 if_show_summary(void)
 {
   struct iface *i;
-  byte addr[IPA_MAX_TEXT_LENGTH + 16];
 
-  cli_msg(-2005, "interface state address");
+  cli_msg(-2005, "%-10s %-6s %-18s %s", "Interface", "State", "IPv4 address", "IPv6 address");
   WALK_LIST(i, iface_list)
     {
-      if (i->addr)
+      byte a4[IPA_MAX_TEXT_LENGTH + 17];
-	bsprintf(addr, "%I/%d", i->addr->ip, i->addr->prefix.pxlen);
+      byte a6[IPA_MAX_TEXT_LENGTH + 17];
+
+      if (i->addr4)
+	bsprintf(a4, "%I/%d", i->addr4->ip, i->addr4->prefix.pxlen);
       else
-	addr[0] = 0;
+	a4[0] = 0;
-      cli_msg(-1005, "%-9s %-5s %s", i->name, (i->flags & IF_UP) ? "up" : "DOWN", addr);
+
+      if (i->addr6)
+	bsprintf(a6, "%I/%d", i->addr6->ip, i->addr6->prefix.pxlen);
+      else
+	a6[0] = 0;
+
+      cli_msg(-1005, "%-10s %-6s %-18s %s",
+	      i->name, (i->flags & IF_UP) ? "up" : "down", a4, a6);
     }
   cli_msg(0, "");
 }

nest/iface.h

@@ -34,12 +34,17 @@ struct iface {
   unsigned flags;
   unsigned mtu;
   unsigned index;			/* OS-dependent interface index */
+  unsigned master_index;		/* Interface index of master iface */
+  struct iface *master;			/* Master iface (e.g. for VRF) */
   list addrs;				/* Addresses assigned to this interface */
-  struct ifa *addr;			/* Primary address */
+  struct ifa *addr4;			/* Primary address for IPv4 */
+  struct ifa *addr6;			/* Primary address for IPv6 */
+  struct ifa *llv6;			/* Primary link-local address for IPv6 */
+  ip4_addr sysdep;			/* Arbitrary IPv4 address for internal sysdep use */
   list neighbors;			/* All neighbors on this interface */
 };
 
-#define IF_UP 1				/* IF_ADMIN_UP and IP address known */
+#define IF_UP 1				/* Currently just IF_ADMIN_UP */
 #define IF_MULTIACCESS 2
 #define IF_BROADCAST 4
 #define IF_MULTICAST 8
@@ -70,7 +75,10 @@ struct iface {
 
 #define IF_JUST_CREATED 0x10000000	/* Send creation event as soon as possible */
 #define IF_TMP_DOWN 0x20000000		/* Temporary shutdown due to interface reconfiguration */
-#define IF_UPDATED 0x40000000		/* Touched in last scan */
+#define IF_UPDATED 0x40000000		/* Iface touched in last scan */
+#define IF_NEEDS_RECALC	0x80000000	/* Preferred address recalculation is needed */
+
+#define IA_UPDATED IF_UPDATED		/* Address touched in last scan */
 
 /* Interface change events */
 
@@ -79,8 +87,14 @@ struct iface {
 #define IF_CHANGE_MTU 4
 #define IF_CHANGE_CREATE 8		/* Seen this interface for the first time */
 #define IF_CHANGE_LINK 0x10
+#define IF_CHANGE_ADDR4	0x100		/* Change of iface->addr4 */
+#define IF_CHANGE_ADDR6	0x200		/* ... */
+#define IF_CHANGE_LLV6 0x400
+#define IF_CHANGE_SYSDEP 0x800
 #define IF_CHANGE_TOO_MUCH 0x40000000	/* Used internally */
 
+#define IF_CHANGE_PREFERRED (IF_CHANGE_ADDR4 | IF_CHANGE_ADDR6 | IF_CHANGE_LLV6)
+
 void if_init(void);
 void if_dump(struct iface *);
 void if_dump_all(void);
@@ -99,7 +113,7 @@ void if_feed_baby(struct proto *);
 struct iface *if_find_by_index(unsigned);
 struct iface *if_find_by_name(char *);
 struct iface *if_get_by_name(char *);
-void ifa_recalc_all_primary_addresses(void);
+void if_recalc_all_preferred_addresses(void);
 
 
 /* The Neighbor Cache */

nest/locks.c

@@ -45,6 +45,7 @@ olock_same(struct object_lock *x, struct object_lock *y)
   return
     x->type == y->type &&
     x->iface == y->iface &&
+    x->vrf == y->vrf &&
     x->port == y->port &&
     x->inst == y->inst &&
     ipa_equal(x->addr, y->addr);

nest/locks.h

@@ -30,6 +30,7 @@ struct object_lock {
   uint port;		/* ... port number */
   uint inst;		/* ... instance ID */
   struct iface *iface;	/* ... interface */
+  struct iface *vrf;	/* ... or VRF (if iface is unknown) */
   void (*hook)(struct object_lock *);	/* Called when the lock succeeds */
   void *data;		/* User data */
   /* ... internal to lock manager, don't touch ... */

nest/neighbor.c

@@ -30,7 +30,8 @@
  * when the protocol has explicitly requested it via the %NEF_STICKY
  * flag because it wishes to be notified when the node will again become
  * a neighbor. Such entries are enqueued in a special list which is walked
- * whenever an interface changes its state to up.
+ * whenever an interface changes its state to up. Neighbor entry VRF
+ * association is implied by respective protocol.
  *
  * When a neighbor event occurs (a neighbor gets disconnected or a sticky
  * inactive neighbor becomes connected), the protocol hook neigh_notify()
@@ -153,8 +154,9 @@ neigh_find2(struct proto *p, ip_addr *a, struct iface *ifa, unsigned flags)
     }
   else
     WALK_LIST(i, iface_list)
-      if ((scope = if_connected(a, i, &addr)) >= 0)
+      if ((!p->vrf || p->vrf == i->master) &&
-	{
+	  ((scope = if_connected(a, i, &addr)) >= 0))
+        {
 	  ifa = i;
 	  break;
 	}

nest/password.c

@@ -10,6 +10,7 @@
 #include "nest/bird.h"
 #include "nest/password.h"
 #include "lib/string.h"
+#include "lib/timer.h"
 #include "lib/mac.h"
 
 struct password_item *last_password_item = NULL;
@@ -19,12 +20,13 @@ password_find(list *l, int first_fit)
 {
   struct password_item *pi;
   struct password_item *pf = NULL;
+  btime now_ = current_real_time();
 
   if (l)
   {
     WALK_LIST(pi, *l)
     {
-      if ((pi->genfrom < now_real) && (pi->gento > now_real))
+      if ((pi->genfrom < now_) && (pi->gento > now_))
       {
 	if (first_fit)
 	  return pi;
@@ -41,12 +43,13 @@ struct password_item *
 password_find_by_id(list *l, uint id)
 {
   struct password_item *pi;
+  btime now_ = current_real_time();
 
   if (!l)
     return NULL;
 
   WALK_LIST(pi, *l)
-    if ((pi->id == id) && (pi->accfrom <= now_real) && (now_real < pi->accto))
+    if ((pi->id == id) && (pi->accfrom <= now_) && (now_ < pi->accto))
       return pi;
 
   return NULL;
@@ -56,12 +59,13 @@ struct password_item *
 password_find_by_value(list *l, char *pass, uint size)
 {
   struct password_item *pi;
+  btime now_ = current_real_time();
 
   if (!l)
     return NULL;
 
   WALK_LIST(pi, *l)
-    if (password_verify(pi, pass, size) && (pi->accfrom <= now_real) && (now_real < pi->accto))
+    if (password_verify(pi, pass, size) && (pi->accfrom <= now_) && (now_ < pi->accto))
       return pi;
 
   return NULL;

nest/password.h

@@ -10,15 +10,13 @@
 #ifndef PASSWORD_H
 #define PASSWORD_H
 
-#include "sysdep/unix/timer.h"
-
 struct password_item {
   node n;
   char *password;			/* Key data, null terminated */
   uint length;				/* Key length, without null */
   uint id;				/* Key ID */
   uint alg;				/* MAC algorithm */
-  bird_clock_t accfrom, accto, genfrom, gento;
+  btime accfrom, accto, genfrom, gento;
 };
 
 extern struct password_item *last_password_item;

nest/proto.c

@@ -13,6 +13,7 @@
 #include "lib/resource.h"
 #include "lib/lists.h"
 #include "lib/event.h"
+#include "lib/timer.h"
 #include "lib/string.h"
 #include "conf/conf.h"
 #include "nest/route.h"
@@ -43,7 +44,7 @@ static char *c_states[] = { "DOWN", "START", "UP", "FLUSHING" };
 
 extern struct protocol proto_unix_iface;
 
-static void proto_shutdown_loop(struct timer *);
+static void proto_shutdown_loop(timer *);
 static void proto_rethink_goal(struct proto *p);
 static char *proto_state_name(struct proto *p);
 static void channel_verify_limits(struct channel *c);
@@ -162,7 +163,7 @@ proto_add_channel(struct proto *p, struct channel_config *cf)
 
   c->channel_state = CS_DOWN;
   c->export_state = ES_DOWN;
-  c->last_state_change = now;
+  c->last_state_change = current_time();
   c->reloadable = 1;
 
   CALL(c->channel->init, c, cf);
@@ -341,7 +342,7 @@ channel_set_state(struct channel *c, uint state)
     return;
 
   c->channel_state = state;
-  c->last_state_change = now;
+  c->last_state_change = current_time();
 
   switch (state)
   {
@@ -672,7 +673,8 @@ proto_init(struct proto_config *c, node *n)
   struct proto *p = pr->init(c);
 
   p->proto_state = PS_DOWN;
-  p->last_state_change = now;
+  p->last_state_change = current_time();
+  p->vrf = c->vrf;
   insert_node(&p->n, n);
 
   p->event = ev_new(proto_pool);
@@ -819,7 +821,8 @@ proto_reconfigure(struct proto *p, struct proto_config *oc, struct proto_config
   /* If there is a too big change in core attributes, ... */
   if ((nc->protocol != oc->protocol) ||
       (nc->net_type != oc->net_type) ||
-      (nc->disabled != p->disabled))
+      (nc->disabled != p->disabled) ||
+      (nc->vrf != oc->vrf))
     return 0;
 
   p->name = nc->name;
@@ -977,6 +980,7 @@ proto_rethink_goal(struct proto *p)
     proto_remove_channels(p);
     rem_node(&p->n);
     rfree(p->event);
+    mb_free(p->message);
     mb_free(p);
     if (!nc)
       return;
@@ -1046,7 +1050,7 @@ proto_rethink_goal(struct proto *p)
  *
  */
 
-static void graceful_restart_done(struct timer *t);
+static void graceful_restart_done(timer *t);
 
 /**
  * graceful_restart_recovery - request initial graceful restart recovery
@@ -1083,9 +1087,8 @@ graceful_restart_init(void)
   }
 
   graceful_restart_state = GRS_ACTIVE;
-  gr_wait_timer = tm_new(proto_pool);
+  gr_wait_timer = tm_new_init(proto_pool, graceful_restart_done, NULL, 0, 0);
-  gr_wait_timer->hook = graceful_restart_done;
+  tm_start(gr_wait_timer, config->gr_wait S);
-  tm_start(gr_wait_timer, config->gr_wait);
 }
 
 /**
@@ -1099,7 +1102,7 @@ graceful_restart_init(void)
  * restart wait timer fires (but there are still some locks).
  */
 static void
-graceful_restart_done(struct timer *t UNUSED)
+graceful_restart_done(timer *t UNUSED)
 {
   log(L_INFO "Graceful restart done");
   graceful_restart_state = GRS_DONE;
@@ -1136,7 +1139,7 @@ graceful_restart_show_status(void)
 
   cli_msg(-24, "Graceful restart recovery in progress");
   cli_msg(-24, "  Waiting for %d channels to recover", graceful_restart_locks);
-  cli_msg(-24, "  Wait timer is %d/%d", tm_remains(gr_wait_timer), config->gr_wait);
+  cli_msg(-24, "  Wait timer is %t/%u", tm_remains(gr_wait_timer), config->gr_wait);
 }
 
 /**
@@ -1298,7 +1301,7 @@ protos_build(void)
 int proto_restart;
 
 static void
-proto_shutdown_loop(struct timer *t UNUSED)
+proto_shutdown_loop(timer *t UNUSED)
 {
   struct proto *p, *p_next;
 
@@ -1329,7 +1332,40 @@ proto_schedule_down(struct proto *p, byte restart, byte code)
 
   p->down_sched = restart ? PDS_RESTART : PDS_DISABLE;
   p->down_code = code;
-  tm_start_max(proto_shutdown_timer, restart ? 2 : 0);
+  tm_start_max(proto_shutdown_timer, restart ? 250 MS : 0);
+}
+
+/**
+ * proto_set_message - set administrative message to protocol
+ * @p: protocol
+ * @msg: message
+ * @len: message length (-1 for NULL-terminated string)
+ *
+ * The function sets administrative message (string) related to protocol state
+ * change. It is called by the nest code for manual enable/disable/restart
+ * commands all routes to the protocol, and by protocol-specific code when the
+ * protocol state change is initiated by the protocol. Using NULL message clears
+ * the last message. The message string may be either NULL-terminated or with an
+ * explicit length.
+ */
+void
+proto_set_message(struct proto *p, char *msg, int len)
+{
+  mb_free(p->message);
+  p->message = NULL;
+
+  if (!msg || !len)
+    return;
+
+  if (len < 0)
+    len = strlen(msg);
+
+  if (!len)
+    return;
+
+  p->message = mb_alloc(proto_pool, len + 1);
+  memcpy(p->message, msg, len);
+  p->message[len] = 0;
 }
 
 
@@ -1500,7 +1536,7 @@ proto_notify_state(struct proto *p, uint state)
     return;
 
   p->proto_state = state;
-  p->last_state_change = now;
+  p->last_state_change = current_time();
 
   switch (state)
   {
@@ -1620,19 +1656,20 @@ channel_show_info(struct channel *c)
 }
 
 void
-proto_cmd_show(struct proto *p, uint verbose, int cnt)
+proto_cmd_show(struct proto *p, uintptr_t verbose, int cnt)
 {
   byte buf[256], tbuf[TM_DATETIME_BUFFER_SIZE];
 
   /* First protocol - show header */
   if (!cnt)
-    cli_msg(-2002, "name     proto    table    state  since       info");
+    cli_msg(-2002, "%-10s %-10s %-10s %-6s %-12s  %s",
+	    "Name", "Proto", "Table", "State", "Since", "Info");
 
   buf[0] = 0;
   if (p->proto->get_status)
     p->proto->get_status(p, buf);
-  tm_format_datetime(tbuf, &config->tf_proto, p->last_state_change);
+  tm_format_time(tbuf, &config->tf_proto, p->last_state_change);
-  cli_msg(-1002, "%-8s %-8s %-8s %-5s  %-10s  %s",
+  cli_msg(-1002, "%-10s %-10s %-10s %-6s %-12s  %s",
 	  p->name,
 	  p->proto->name,
 	  p->main_channel ? p->main_channel->table->name : "---",
@@ -1644,8 +1681,12 @@ proto_cmd_show(struct proto *p, uint verbose, int cnt)
   {
     if (p->cf->dsc)
       cli_msg(-1006, "  Description:    %s", p->cf->dsc);
+    if (p->message)
+      cli_msg(-1006, "  Message:        %s", p->message);
     if (p->cf->router_id)
       cli_msg(-1006, "  Router ID:      %R", p->cf->router_id);
+    if (p->vrf)
+      cli_msg(-1006, "  VRF:            %s", p->vrf->name);
 
     if (p->proto->show_proto_info)
       p->proto->show_proto_info(p);
@@ -1661,7 +1702,7 @@ proto_cmd_show(struct proto *p, uint verbose, int cnt)
 }
 
 void
-proto_cmd_disable(struct proto *p, uint arg UNUSED, int cnt UNUSED)
+proto_cmd_disable(struct proto *p, uintptr_t arg, int cnt UNUSED)
 {
   if (p->disabled)
   {
@@ -1672,12 +1713,13 @@ proto_cmd_disable(struct proto *p, uint arg UNUSED, int cnt UNUSED)
   log(L_INFO "Disabling protocol %s", p->name);
   p->disabled = 1;
   p->down_code = PDC_CMD_DISABLE;
+  proto_set_message(p, (char *) arg, -1);
   proto_rethink_goal(p);
   cli_msg(-9, "%s: disabled", p->name);
 }
 
 void
-proto_cmd_enable(struct proto *p, uint arg UNUSED, int cnt UNUSED)
+proto_cmd_enable(struct proto *p, uintptr_t arg, int cnt UNUSED)
 {
   if (!p->disabled)
   {
@@ -1687,12 +1729,13 @@ proto_cmd_enable(struct proto *p, uint arg UNUSED, int cnt UNUSED)
 
   log(L_INFO "Enabling protocol %s", p->name);
   p->disabled = 0;
+  proto_set_message(p, (char *) arg, -1);
   proto_rethink_goal(p);
   cli_msg(-11, "%s: enabled", p->name);
 }
 
 void
-proto_cmd_restart(struct proto *p, uint arg UNUSED, int cnt UNUSED)
+proto_cmd_restart(struct proto *p, uintptr_t arg, int cnt UNUSED)
 {
   if (p->disabled)
   {
@@ -1703,6 +1746,7 @@ proto_cmd_restart(struct proto *p, uint arg UNUSED, int cnt UNUSED)
   log(L_INFO "Restarting protocol %s", p->name);
   p->disabled = 1;
   p->down_code = PDC_CMD_RESTART;
+  proto_set_message(p, (char *) arg, -1);
   proto_rethink_goal(p);
   p->disabled = 0;
   proto_rethink_goal(p);
@@ -1710,7 +1754,7 @@ proto_cmd_restart(struct proto *p, uint arg UNUSED, int cnt UNUSED)
 }
 
 void
-proto_cmd_reload(struct proto *p, uint dir, int cnt UNUSED)
+proto_cmd_reload(struct proto *p, uintptr_t dir, int cnt UNUSED)
 {
   struct channel *c;
 
@@ -1749,19 +1793,19 @@ proto_cmd_reload(struct proto *p, uint dir, int cnt UNUSED)
 }
 
 void
-proto_cmd_debug(struct proto *p, uint mask, int cnt UNUSED)
+proto_cmd_debug(struct proto *p, uintptr_t mask, int cnt UNUSED)
 {
   p->debug = mask;
 }
 
 void
-proto_cmd_mrtdump(struct proto *p, uint mask, int cnt UNUSED)
+proto_cmd_mrtdump(struct proto *p, uintptr_t mask, int cnt UNUSED)
 {
   p->mrtdump = mask;
 }
 
 static void
-proto_apply_cmd_symbol(struct symbol *s, void (* cmd)(struct proto *, uint, int), uint arg)
+proto_apply_cmd_symbol(struct symbol *s, void (* cmd)(struct proto *, uintptr_t, int), uintptr_t arg)
 {
   if (s->class != SYM_PROTO)
   {
@@ -1774,7 +1818,7 @@ proto_apply_cmd_symbol(struct symbol *s, void (* cmd)(struct proto *, uint, int)
 }
 
 static void
-proto_apply_cmd_patt(char *patt, void (* cmd)(struct proto *, uint, int), uint arg)
+proto_apply_cmd_patt(char *patt, void (* cmd)(struct proto *, uintptr_t, int), uintptr_t arg)
 {
   struct proto *p;
   int cnt = 0;
@@ -1790,8 +1834,8 @@ proto_apply_cmd_patt(char *patt, void (* cmd)(struct proto *, uint, int), uint a
 }
 
 void
-proto_apply_cmd(struct proto_spec ps, void (* cmd)(struct proto *, uint, int),
+proto_apply_cmd(struct proto_spec ps, void (* cmd)(struct proto *, uintptr_t, int),
-		int restricted, uint arg)
+		int restricted, uintptr_t arg)
 {
   if (restricted && cli_access_restricted())
     return;

nest/protocol.h

@@ -12,7 +12,6 @@
 #include "lib/lists.h"
 #include "lib/resource.h"
 #include "lib/event.h"
-#include "sysdep/unix/timer.h"
 #include "nest/route.h"
 #include "conf/conf.h"
 
@@ -101,6 +100,7 @@ struct proto_config {
   u32 router_id;			/* Protocol specific router ID */
 
   list channels;			/* List of channel configs (struct channel_config) */
+  struct iface *vrf;			/* Related VRF instance, NULL if global */
 
   /* Check proto_reconfigure() and proto_copy_config() after changing struct proto_config */
 
@@ -143,6 +143,7 @@ struct proto {
   list channels;			/* List of channels to rtables (struct channel) */
   struct channel *main_channel;		/* Primary channel */
   struct rte_src *main_source;		/* Primary route source */
+  struct iface *vrf;			/* Related VRF instance, NULL if global */
 
   char *name;				/* Name of this instance (== cf->name) */
   u32 debug;				/* Debugging flags */
@@ -159,8 +160,9 @@ struct proto {
   byte down_sched;			/* Shutdown is scheduled for later (PDS_*) */
   byte down_code;			/* Reason for shutdown (PDC_* codes) */
   u32 hash_key;				/* Random key used for hashing of neighbors */
-  bird_clock_t last_state_change;	/* Time of last state transition */
+  btime last_state_change;		/* Time of last state transition */
   char *last_state_name_announced;	/* Last state name we've announced to the user */
+  char *message;			/* State-change message, allocated from proto_pool */
 
   /*
    *	General protocol hooks:
@@ -237,6 +239,7 @@ struct proto_spec {
 void *proto_new(struct proto_config *);
 void *proto_config_new(struct protocol *, int class);
 void proto_copy_config(struct proto_config *dest, struct proto_config *src);
+void proto_set_message(struct proto *p, char *msg, int len);
 
 void graceful_restart_recovery(void);
 void graceful_restart_init(void);
@@ -249,15 +252,15 @@ void channel_graceful_restart_unlock(struct channel *c);
 void channel_show_limit(struct channel_limit *l, const char *dsc);
 void channel_show_info(struct channel *c);
 
-void proto_cmd_show(struct proto *, uint, int);
+void proto_cmd_show(struct proto *, uintptr_t, int);
-void proto_cmd_disable(struct proto *, uint, int);
+void proto_cmd_disable(struct proto *, uintptr_t, int);
-void proto_cmd_enable(struct proto *, uint, int);
+void proto_cmd_enable(struct proto *, uintptr_t, int);
-void proto_cmd_restart(struct proto *, uint, int);
+void proto_cmd_restart(struct proto *, uintptr_t, int);
-void proto_cmd_reload(struct proto *, uint, int);
+void proto_cmd_reload(struct proto *, uintptr_t, int);
-void proto_cmd_debug(struct proto *, uint, int);
+void proto_cmd_debug(struct proto *, uintptr_t, int);
-void proto_cmd_mrtdump(struct proto *, uint, int);
+void proto_cmd_mrtdump(struct proto *, uintptr_t, int);
 
-void proto_apply_cmd(struct proto_spec ps, void (* cmd)(struct proto *, uint, int), int restricted, uint arg);
+void proto_apply_cmd(struct proto_spec ps, void (* cmd)(struct proto *, uintptr_t, int), int restricted, uintptr_t arg);
 struct proto *proto_get_named(struct symbol *, struct protocol *);
 
 #define CMD_RELOAD	0
@@ -508,7 +511,7 @@ struct channel {
   u8 gr_lock;				/* Graceful restart mechanism should wait for this channel */
   u8 gr_wait;				/* Route export to channel is postponed until graceful restart */
 
-  bird_clock_t last_state_change;	/* Time of last state transition */
+  btime last_state_change;		/* Time of last state transition */
 };
 
 

nest/route.h

@@ -11,7 +11,6 @@
 
 #include "lib/lists.h"
 #include "lib/resource.h"
-#include "sysdep/unix/timer.h"
 #include "lib/net.h"
 
 struct ea_list;
@@ -159,8 +158,8 @@ typedef struct rtable {
 					 * obstacle from this routing table.
 					 */
   struct event *rt_event;		/* Routing table event */
+  btime gc_time;			/* Time of last GC */
   int gc_counter;			/* Number of operations since last GC */
-  bird_clock_t gc_time;			/* Time of last GC */
   byte prune_state;			/* Table prune state, 1 -> scheduled, 2-> running */
   byte hcu_scheduled;			/* Hostcache update is scheduled */
   byte nhu_state;			/* Next Hop Update state */
@@ -213,7 +212,7 @@ typedef struct rte {
   byte flags;				/* Flags (REF_...) */
   byte pflags;				/* Protocol-specific flags */
   word pref;				/* Route preference */
-  bird_clock_t lastmod;			/* Last modified */
+  btime lastmod;			/* Last modified */
   union {				/* Protocol-dependent data (metrics etc.) */
 #ifdef CONFIG_RIP
     struct {
@@ -236,6 +235,7 @@ typedef struct rte {
 #endif
 #ifdef CONFIG_BABEL
     struct {
+      u16 seqno;			/* Babel seqno */
       u16 metric;			/* Babel metric */
       u64 router_id;			/* Babel router id */
     } babel;
@@ -309,6 +309,8 @@ int rt_feed_channel(struct channel *c);
 void rt_feed_channel_abort(struct channel *c);
 struct rtable_config *rt_new_table(struct symbol *s, uint addr_type);
 
+/* Default limit for ECMP next hops, defined in sysdep code */
+extern const int rt_default_ecmp;
 
 struct rt_show_data_rtable {
   node n;
@@ -470,7 +472,8 @@ typedef struct eattr {
 #define EAP_OSPF 3			/* OSPF */
 #define EAP_KRT 4			/* Kernel route attributes */
 #define EAP_BABEL 5			/* Babel attributes */
-#define EAP_MAX 6
+#define EAP_RADV 6			/* Router advertisment attributes */
+#define EAP_MAX 7
 
 #define EA_CODE(proto,id) (((proto) << 8) | (id))
 #define EA_PROTO(ea) ((ea) >> 8)

nest/rt-dev.c

@@ -33,7 +33,7 @@ dev_ifa_notify(struct proto *P, uint flags, struct ifa *ad)
   struct channel *c;
 
   if (!EMPTY_LIST(cf->iface_list) &&
-      !iface_patt_find(&cf->iface_list, ad->iface, ad->iface->addr))
+      !iface_patt_find(&cf->iface_list, ad->iface, ad))
     /* Empty list is automatically treated as "*" */
     return;
 

nest/rt-fib.c

@@ -32,6 +32,24 @@
  * Basic FIB operations are performed by functions defined by this module,
  * enumerating of FIB contents is accomplished by using the FIB_WALK() macro
  * or FIB_ITERATE_START() if you want to do it asynchronously.
+ *
+ * For simple iteration just place the body of the loop between FIB_WALK() and
+ * FIB_WALK_END(). You can't modify the FIB during the iteration (you can modify
+ * data in the node, but not add or remove nodes).
+ *
+ * If you need more freedom, you can use the FIB_ITERATE_*() group of macros.
+ * First, you initialize an iterator with FIB_ITERATE_INIT(). Then you can put
+ * the loop body in between FIB_ITERATE_START() and FIB_ITERATE_END(). In
+ * addition, the iteration can be suspended by calling FIB_ITERATE_PUT().
+ * This'll link the iterator inside the FIB. While suspended, you may modify the
+ * FIB, exit the current function, etc. To resume the iteration, enter the loop
+ * again. You can use FIB_ITERATE_UNLINK() to unlink the iterator (while
+ * iteration is suspended) in cases like premature end of FIB iteration.
+ *
+ * Note that the iterator must not be destroyed when the iteration is suspended,
+ * the FIB would then contain a pointer to invalid memory. Therefore, after each
+ * FIB_ITERATE_INIT() or FIB_ITERATE_PUT() there must be either
+ * FIB_ITERATE_START() or FIB_ITERATE_UNLINK() before the iterator is destroyed.
  */
 
 #undef LOCAL_DEBUG
@@ -195,6 +213,7 @@ fib_hash(struct fib *f, const net_addr *a)
   case NET_ROA6: return FIB_HASH(f, a, roa6);
   case NET_FLOW4: return FIB_HASH(f, a, flow4);
   case NET_FLOW6: return FIB_HASH(f, a, flow6);
+  case NET_MPLS: return FIB_HASH(f, a, mpls);
   default: bug("invalid type");
   }
 }
@@ -231,6 +250,7 @@ fib_find(struct fib *f, const net_addr *a)
   case NET_ROA6: return FIB_FIND(f, a, roa6);
   case NET_FLOW4: return FIB_FIND(f, a, flow4);
   case NET_FLOW6: return FIB_FIND(f, a, flow6);
+  case NET_MPLS: return FIB_FIND(f, a, mpls);
   default: bug("invalid type");
   }
 }
@@ -250,6 +270,7 @@ fib_insert(struct fib *f, const net_addr *a, struct fib_node *e)
   case NET_ROA6: FIB_INSERT(f, a, e, roa6); return;
   case NET_FLOW4: FIB_INSERT(f, a, e, flow4); return;
   case NET_FLOW6: FIB_INSERT(f, a, e, flow6); return;
+  case NET_MPLS: FIB_INSERT(f, a, e, mpls); return;
   default: bug("invalid type");
   }
 }

nest/rt-show.c

@@ -39,7 +39,7 @@ rt_show_rte(struct cli *c, byte *ia, rte *e, struct rt_show_data *d, ea_list *tm
   void (*get_route_info)(struct rte *, byte *buf, struct ea_list *attrs);
   struct nexthop *nh;
 
-  tm_format_datetime(tm, &config->tf_route, e->lastmod);
+  tm_format_time(tm, &config->tf_route, e->lastmod);
   if (ipa_nonzero(a->from) && !ipa_equal(a->from, a->nh.gw))
     bsprintf(from, " from %I", a->from);
   else
@@ -63,7 +63,7 @@ rt_show_rte(struct cli *c, byte *ia, rte *e, struct rt_show_data *d, ea_list *tm
   if (d->last_table != d->tab)
     rt_show_table(c, d);
 
-  cli_printf(c, -1007, "%-18s %s [%s %s%s]%s%s", ia, rta_dest_name(a->dest),
+  cli_printf(c, -1007, "%-20s %s [%s %s%s]%s%s", ia, rta_dest_name(a->dest),
 	     a->src->proto->name, tm, from, primary ? (sync_error ? " !" : " *") : "", info);
 
   if (a->dest == RTD_UNICAST)
@@ -71,6 +71,7 @@ rt_show_rte(struct cli *c, byte *ia, rte *e, struct rt_show_data *d, ea_list *tm
     {
       char mpls[MPLS_MAX_LABEL_STACK*12 + 5], *lsp = mpls;
       char *onlink = (nh->flags & RNF_ONLINK) ? " onlink" : "";
+      char weight[16] = "";
 
       if (nh->labels)
         {
@@ -81,11 +82,14 @@ rt_show_rte(struct cli *c, byte *ia, rte *e, struct rt_show_data *d, ea_list *tm
       *lsp = '\0';
 
       if (a->nh.next)
-	cli_printf(c, -1007, "\tvia %I%s on %s%s weight %d",
+	bsprintf(weight, " weight %d", nh->weight + 1);
-		   nh->gw, mpls, nh->iface->name, onlink, nh->weight + 1);
+
+      if (ipa_nonzero(nh->gw))
+	cli_printf(c, -1007, "\tvia %I on %s%s%s%s",
+		   nh->gw, nh->iface->name, mpls, onlink, weight);
       else
-	cli_printf(c, -1007, "\tvia %I%s on %s%s",
+	cli_printf(c, -1007, "\tdev %s%s%s",
-		   nh->gw, mpls, nh->iface->name, onlink);
+		   nh->iface->name, mpls,  onlink, weight);
     }
 
   if (d->verbose)

nest/rt-table.c

@@ -897,7 +897,9 @@ rte_validate(rte *e)
     return 0;
   }
 
-  c = net_classify(n->n.addr);
+  /* FIXME: better handling different nettypes */
+  c = !net_is_flow(n->n.addr) ?
+    net_classify(n->n.addr): (IADDR_HOST | SCOPE_UNIVERSE);
   if ((c < 0) || !(c & IADDR_HOST) || ((c & IADDR_SCOPE_MASK) <= SCOPE_LINK))
   {
     log(L_WARN "Ignoring bogus route %N received via %s",
@@ -1173,7 +1175,7 @@ rte_recalculate(struct channel *c, net *net, rte *new, struct rte_src *src)
     }
 
   if (new)
-    new->lastmod = now;
+    new->lastmod = current_time();
 
   /* Log the route change */
   if (p->debug & D_ROUTES)
@@ -1201,7 +1203,7 @@ rte_recalculate(struct channel *c, net *net, rte *new, struct rte_src *src)
 
   if (!net->routes &&
       (table->gc_counter++ >= table->config->gc_max_ops) &&
-      (table->gc_time + table->config->gc_min_time <= now))
+      (table->gc_time + table->config->gc_min_time <= current_time()))
     rt_schedule_prune(table);
 
   if (old_ok && p->rte_remove)
@@ -1497,7 +1499,7 @@ rte_dump(rte *e)
 {
   net *n = e->net;
   debug("%-1N ", n->n.addr);
-  debug("KF=%02x PF=%02x pref=%d lm=%d ", n->n.flags, e->pflags, e->pref, now-e->lastmod);
+  debug("KF=%02x PF=%02x pref=%d ", n->n.flags, e->pflags, e->pref);
   rta_dump(e->attrs);
   if (e->attrs->src->proto->proto->dump_attrs)
     e->attrs->src->proto->proto->dump_attrs(e);
@@ -1609,7 +1611,7 @@ rt_setup(pool *p, rtable *t, char *name, struct rtable_config *cf)
       t->rt_event = ev_new(p);
       t->rt_event->hook = rt_event;
       t->rt_event->data = t;
-      t->gc_time = now;
+      t->gc_time = current_time();
     }
 }
 
@@ -1708,7 +1710,7 @@ again:
 #endif
 
   tab->gc_counter = 0;
-  tab->gc_time = now;
+  tab->gc_time = current_time();
 
   /* state change 2->0, 3->1 */
   tab->prune_state &= 1;

proto/babel/babel.h

@@ -2,6 +2,8 @@
  *	BIRD -- The Babel protocol
  *
  *	Copyright (c) 2015--2016 Toke Hoiland-Jorgensen
+ * 	(c) 2016--2017 Ondrej Zajicek <santiago@crfreenet.org>
+ *	(c) 2016--2017 CZ.NIC z.s.p.o.
  *
  *	Can be freely distributed and used under the terms of the GNU GPL.
  *
@@ -21,7 +23,7 @@
 #include "lib/lists.h"
 #include "lib/socket.h"
 #include "lib/string.h"
-#include "sysdep/unix/timer.h"
+#include "lib/timer.h"
 
 #define EA_BABEL_METRIC		EA_CODE(EAP_BABEL, 0)
 #define EA_BABEL_ROUTER_ID	EA_CODE(EAP_BABEL, 1)
@@ -32,25 +34,28 @@
 #define BABEL_INFINITY		0xFFFF
 
 
-#define BABEL_HELLO_INTERVAL_WIRED	4	/* Default hello intervals in seconds */
+#define BABEL_HELLO_INTERVAL_WIRED	(4 S_)	/* Default hello intervals in seconds */
-#define BABEL_HELLO_INTERVAL_WIRELESS	4
+#define BABEL_HELLO_INTERVAL_WIRELESS	(4 S_)
+#define BABEL_HELLO_LIMIT		12
 #define BABEL_UPDATE_INTERVAL_FACTOR	4
 #define BABEL_IHU_INTERVAL_FACTOR	3
-#define BABEL_IHU_EXPIRY_FACTOR(X)	((X)*3/2)	/* 1.5 */
+#define BABEL_HOLD_TIME_FACTOR		4	/* How long we keep unreachable route relative to update interval */
-#define BABEL_HELLO_EXPIRY_FACTOR(X)	((X)*3/2)	/* 1.5 */
+#define BABEL_IHU_EXPIRY_FACTOR(X)	((btime)(X)*7/2)	/* 3.5 */
-#define BABEL_ROUTE_EXPIRY_FACTOR(X)	((X)*7/2)	/* 3.5 */
+#define BABEL_HELLO_EXPIRY_FACTOR(X)	((btime)(X)*3/2)	/* 1.5 */
-#define BABEL_ROUTE_REFRESH_INTERVAL	2	/* Seconds before route expiry to send route request */
+#define BABEL_ROUTE_EXPIRY_FACTOR(X)	((btime)(X)*7/2)	/* 3.5 */
-#define BABEL_HOLD_TIME			10	/* Expiry time for our own routes */
+#define BABEL_ROUTE_REFRESH_FACTOR(X)	((btime)(X)*5/2)	/* 2.5 */
+#define BABEL_SEQNO_REQUEST_RETRY	4
+#define BABEL_SEQNO_REQUEST_EXPIRY	(2 S_)
+#define BABEL_GARBAGE_INTERVAL		(300 S_)
 #define BABEL_RXCOST_WIRED		96
 #define BABEL_RXCOST_WIRELESS		256
 #define BABEL_INITIAL_HOP_COUNT		255
-#define BABEL_MAX_SEND_INTERVAL		5
+#define BABEL_MAX_SEND_INTERVAL		5	/* Unused ? */
-#define BABEL_TIME_UNITS		100	/* On-wire times are counted in centiseconds */
-#define BABEL_SEQNO_REQUEST_EXPIRY	60
-#define BABEL_GARBAGE_INTERVAL		300
 
 /* Max interval that will not overflow when carried as 16-bit centiseconds */
-#define BABEL_MAX_INTERVAL		(0xFFFF/BABEL_TIME_UNITS)
+#define BABEL_TIME_UNITS		10000	/* On-wire times are counted in centiseconds */
+#define BABEL_MIN_INTERVAL		(0x0001 * BABEL_TIME_UNITS)
+#define BABEL_MAX_INTERVAL		(0xFFFF * BABEL_TIME_UNITS)
 
 #define BABEL_OVERHEAD		(IP6_HEADER_LENGTH+UDP_HEADER_LENGTH)
 #define BABEL_MIN_MTU		(512 + BABEL_OVERHEAD)
@@ -102,8 +107,8 @@ enum babel_ae_type {
 
 struct babel_config {
   struct proto_config c;
-
+  list iface_list;			/* List of iface configs (struct babel_iface_config) */
-  list iface_list;              /* Patterns configured -- keep it first; see babel_reconfigure why */
+  uint hold_time;			/* Time to hold stale entries and unreachable routes */
 };
 
 struct babel_iface_config {
@@ -111,11 +116,12 @@ struct babel_iface_config {
 
   u16 rxcost;
   u8 type;
+  u8 limit;				/* Minimum number of Hellos to keep link up */
   u8 check_link;
   uint port;
-  u16 hello_interval;
+  uint hello_interval;			/* Hello interval, in us */
-  u16 ihu_interval;
+  uint ihu_interval;			/* IHU interval, in us */
-  u16 update_interval;
+  uint update_interval;			/* Update interval, in us */
 
   u16 rx_buffer;			/* RX buffer size, 0 for MTU */
   u16 tx_length;			/* TX packet length limit (including headers), 0 for MTU */
@@ -138,14 +144,13 @@ struct babel_proto {
   list interfaces;			/* Interfaces we really know about (struct babel_iface) */
   u64 router_id;
   u16 update_seqno;			/* To be increased on request */
+  u8 update_seqno_inc;			/* Request for update_seqno increase */
   u8 triggered;				/* For triggering global updates */
 
   slab *route_slab;
   slab *source_slab;
   slab *msg_slab;
-
   slab *seqno_slab;
-  list seqno_cache;			/* Seqno requests in the cache (struct babel_seqno_request) */
 
   struct tbf log_pkt_tbf;		/* TBF for packet messages */
 };
@@ -172,10 +177,10 @@ struct babel_iface {
 
   u16 hello_seqno;			/* To be increased on each hello */
 
-  bird_clock_t next_hello;
+  btime next_hello;
-  bird_clock_t next_regular;
+  btime next_regular;
-  bird_clock_t next_triggered;
+  btime next_triggered;
-  bird_clock_t want_triggered;
+  btime want_triggered;
 
   timer *timer;
   event *send_event;
@@ -186,13 +191,18 @@ struct babel_neighbor {
   struct babel_iface *ifa;
 
   ip_addr addr;
-  u16 txcost;
+  uint uc;				/* Reference counter for seqno requests */
+  u16 rxcost;				/* Sent in last IHU */
+  u16 txcost;				/* Received in last IHU */
+  u16 cost;				/* Computed neighbor cost */
+  s8 ihu_cnt;				/* IHU countdown, 0 to send it */
   u8 hello_cnt;
   u16 hello_map;
   u16 next_hello_seqno;
+  uint last_hello_int;
   /* expiry timers */
-  bird_clock_t hello_expiry;
+  btime hello_expiry;
-  bird_clock_t ihu_expiry;
+  btime ihu_expiry;
 
   list routes;				/* Routes this neighbour has sent us (struct babel_route) */
 };
@@ -203,7 +213,7 @@ struct babel_source {
   u64 router_id;
   u16 seqno;
   u16 metric;
-  bird_clock_t expires;
+  btime expires;
 };
 
 struct babel_route {
@@ -212,37 +222,46 @@ struct babel_route {
   struct babel_entry    *e;
   struct babel_neighbor *neigh;
 
+  u8 feasible;
   u16 seqno;
-  u16 advert_metric;
   u16 metric;
+  u16 advert_metric;
   u64 router_id;
   ip_addr next_hop;
-  bird_clock_t refresh_time;
+  btime refresh_time;
-  bird_clock_t expires;
+  btime expires;
-  u16 expiry_interval;
+};
+
+struct babel_seqno_request {
+  node n;
+  u64 router_id;
+  u16 seqno;
+  u8 hop_count;
+  u8 count;
+  btime expires;
+  struct babel_neighbor *nbr;
 };
 
 struct babel_entry {
-  struct babel_proto *proto;
+  struct babel_route *selected;
-  struct babel_route *selected_in;
-  struct babel_route *selected_out;
 
-  bird_clock_t updated;
-
-  list sources;				/* Source entries for this prefix (struct babel_source). */
   list routes;				/* Routes for this prefix (struct babel_route) */
+  list sources;				/* Source entries for this prefix (struct babel_source). */
+  list requests;
+
+  u8 valid;				/* Entry validity state (BABEL_ENTRY_*) */
+  u8 unreachable;			/* Unreachable route is announced */
+  u16 seqno;				/* Outgoing seqno */
+  u16 metric;				/* Outgoing metric */
+  u64 router_id;			/* Outgoing router ID */
+  btime updated;			/* Last change of outgoing rte, for triggered updates */
 
   struct fib_node n;
 };
 
-/* Stores forwarded seqno requests for duplicate suppression. */
+#define BABEL_ENTRY_DUMMY	0	/* No outgoing route */
-struct babel_seqno_request {
+#define BABEL_ENTRY_VALID	1	/* Valid outgoing route */
-  node n;
+#define BABEL_ENTRY_STALE	2	/* Stale outgoing route, waiting for GC */
-  net_addr net;
-  u64 router_id;
-  u16 seqno;
-  bird_clock_t updated;
-};
 
 
 /*
@@ -252,7 +271,7 @@ struct babel_seqno_request {
 struct babel_msg_ack_req {
   u8 type;
   u16 nonce;
-  u16 interval;
+  uint interval;
   ip_addr sender;
 };
 
@@ -264,7 +283,7 @@ struct babel_msg_ack {
 struct babel_msg_hello {
   u8 type;
   u16 seqno;
-  u16 interval;
+  uint interval;
   ip_addr sender;
 };
 
@@ -272,7 +291,7 @@ struct babel_msg_ihu {
   u8 type;
   u8 ae;
   u16 rxcost;
-  u16 interval;
+  uint interval;
   ip_addr addr;
   ip_addr sender;
 };
@@ -280,7 +299,7 @@ struct babel_msg_ihu {
 struct babel_msg_update {
   u8 type;
   u8 wildcard;
-  u16 interval;
+  uint interval;
   u16 seqno;
   u16 metric;
   u64 router_id;
@@ -334,6 +353,7 @@ void babel_handle_seqno_request(union babel_msg *msg, struct babel_iface *ifa);
 void babel_show_interfaces(struct proto *P, char *iff);
 void babel_show_neighbors(struct proto *P, char *iff);
 void babel_show_entries(struct proto *P);
+void babel_show_routes(struct proto *P);
 
 /* packets.c */
 void babel_enqueue(union babel_msg *msg, struct babel_iface *ifa);

proto/babel/config.Y

@@ -2,6 +2,8 @@
  *	BIRD -- Babel Configuration
  *
  *	Copyright (c) 2015-2016 Toke Hoiland-Jorgensen
+ * 	(c) 2016--2017 Ondrej Zajicek <santiago@crfreenet.org>
+ *	(c) 2016--2017 CZ.NIC z.s.p.o.
  *
  *	Can be freely distributed and used under the terms of the GNU GPL.
  */
@@ -32,6 +34,7 @@ babel_proto_start: proto_start BABEL
 {
   this_proto = proto_config_new(&proto_babel, $1);
   init_list(&BABEL_CFG->iface_list);
+  BABEL_CFG->hold_time = 1 S_;
 };
 
 babel_proto_item:
@@ -56,6 +59,7 @@ babel_iface_start:
   init_list(&this_ipatt->ipn_list);
   BABEL_IFACE->port = BABEL_PORT;
   BABEL_IFACE->type = BABEL_IFACE_TYPE_WIRED;
+  BABEL_IFACE->limit = BABEL_HELLO_LIMIT;
   BABEL_IFACE->tx_tos = IP_PREC_INTERNET_CONTROL;
   BABEL_IFACE->tx_priority = sk_priority_control;
   BABEL_IFACE->check_link = 1;
@@ -83,16 +87,19 @@ babel_iface_finish:
   if (!BABEL_IFACE->update_interval)
     BABEL_IFACE->update_interval = MIN_(BABEL_IFACE->hello_interval*BABEL_UPDATE_INTERVAL_FACTOR, BABEL_MAX_INTERVAL);
   BABEL_IFACE->ihu_interval = MIN_(BABEL_IFACE->hello_interval*BABEL_IHU_INTERVAL_FACTOR, BABEL_MAX_INTERVAL);
+
+  BABEL_CFG->hold_time = MAX_(BABEL_CFG->hold_time, BABEL_IFACE->update_interval*BABEL_HOLD_TIME_FACTOR);
 };
 
 
 babel_iface_item:
  | PORT expr { BABEL_IFACE->port = $2; if (($2<1) || ($2>65535)) cf_error("Invalid port number"); }
  | RXCOST expr { BABEL_IFACE->rxcost = $2; if (($2<1) || ($2>65535)) cf_error("Invalid rxcost"); }
- | HELLO INTERVAL expr { BABEL_IFACE->hello_interval = $3; if (($3<1) || ($3>BABEL_MAX_INTERVAL)) cf_error("Invalid hello interval"); }
+ | LIMIT expr { BABEL_IFACE->limit = $2; if (($2<1) || ($2>16)) cf_error("Limit must be in range 1-16"); }
- | UPDATE INTERVAL expr { BABEL_IFACE->update_interval = $3; if (($3<1) || ($3>BABEL_MAX_INTERVAL)) cf_error("Invalid update interval"); }
  | TYPE WIRED { BABEL_IFACE->type = BABEL_IFACE_TYPE_WIRED; }
  | TYPE WIRELESS { BABEL_IFACE->type = BABEL_IFACE_TYPE_WIRELESS; }
+ | HELLO INTERVAL expr_us { BABEL_IFACE->hello_interval = $3; if (($3<BABEL_MIN_INTERVAL) || ($3>BABEL_MAX_INTERVAL)) cf_error("Hello interval must be in range 10 ms - 655 s"); }
+ | UPDATE INTERVAL expr_us { BABEL_IFACE->update_interval = $3; if (($3<BABEL_MIN_INTERVAL) || ($3>BABEL_MAX_INTERVAL)) cf_error("Update interval must be in range 10 ms - 655 s"); }
  | RX BUFFER expr { BABEL_IFACE->rx_buffer = $3; if (($3<256) || ($3>65535)) cf_error("RX buffer must be in range 256-65535"); }
  | TX LENGTH expr { BABEL_IFACE->tx_length = $3; if (($3<256) || ($3>65535)) cf_error("TX length must be in range 256-65535"); }
  | TX tos { BABEL_IFACE->tx_tos = $2; }
@@ -129,6 +136,9 @@ CF_CLI(SHOW BABEL NEIGHBORS, optsym opttext, [<name>] [\"<interface>\"], [[Show
 CF_CLI(SHOW BABEL ENTRIES, optsym opttext, [<name>], [[Show information about Babel prefix entries]])
 { babel_show_entries(proto_get_named($4, &proto_babel)); };
 
+CF_CLI(SHOW BABEL ROUTES, optsym opttext, [<name>], [[Show information about Babel route entries]])
+{ babel_show_routes(proto_get_named($4, &proto_babel)); };
+
 CF_CODE
 
 CF_END

proto/babel/packets.c

@@ -2,6 +2,8 @@
  *	BIRD -- The Babel protocol
  *
  *	Copyright (c) 2015--2016 Toke Hoiland-Jorgensen
+ * 	(c) 2016--2017 Ondrej Zajicek <santiago@crfreenet.org>
+ *	(c) 2016--2017 CZ.NIC z.s.p.o.
  *
  *	Can be freely distributed and used under the terms of the GNU GPL.
  *
@@ -40,7 +42,7 @@ struct babel_tlv_ack {
 struct babel_tlv_hello {
   u8 type;
   u8 length;
-  u16 reserved;
+  u16 flags;
   u16 seqno;
   u16 interval;
 } PACKED;
@@ -104,8 +106,12 @@ struct babel_tlv_seqno_request {
 } PACKED;
 
 
-#define BABEL_FLAG_DEF_PREFIX	0x80
+/* Hello flags */
-#define BABEL_FLAG_ROUTER_ID	0x40
+#define BABEL_HF_UNICAST	0x8000
+
+/* Update flags */
+#define BABEL_UF_DEF_PREFIX	0x80
+#define BABEL_UF_ROUTER_ID	0x40
 
 
 struct babel_parse_state {
@@ -162,17 +168,17 @@ bytes_equal(u8 *b1, u8 *b2, uint maxlen)
   return i;
 }
 
-static inline u16
+static inline uint
 get_time16(const void *p)
 {
-  u16 v = get_u16(p) / BABEL_TIME_UNITS;
+  uint v = get_u16(p) * BABEL_TIME_UNITS;
-  return MAX(1, v);
+  return MAX(BABEL_MIN_INTERVAL, v);
 }
 
 static inline void
-put_time16(void *p, u16 v)
+put_time16(void *p, uint v)
 {
-  put_u16(p, v * BABEL_TIME_UNITS);
+  put_u16(p, v / BABEL_TIME_UNITS);
 }
 
 static inline void
@@ -341,6 +347,11 @@ babel_read_hello(struct babel_tlv *hdr, union babel_msg *m,
   struct babel_tlv_hello *tlv = (void *) hdr;
   struct babel_msg_hello *msg = &m->hello;
 
+  /* We currently don't support unicast Hello */
+  u16 flags = get_u16(&tlv->flags);
+  if (flags & BABEL_HF_UNICAST)
+    return PARSE_IGNORE;
+
   msg->type = BABEL_TLV_HELLO;
   msg->seqno = get_u16(&tlv->seqno);
   msg->interval = get_time16(&tlv->interval);
@@ -593,8 +604,8 @@ babel_read_update(struct babel_tlv *hdr, union babel_msg *m,
     if (tlv->omitted && !state->def_ip4_prefix_seen)
       return PARSE_ERROR;
 
-    /* Need next hop for v4 routes */
+    /* Update must have next hop, unless it is retraction */
-    if (ipa_zero(state->next_hop_ip4))
+    if (ipa_zero(state->next_hop_ip4) && (msg->metric != BABEL_INFINITY))
       return PARSE_ERROR;
 
     /* Merge saved prefix and received prefix parts */
@@ -604,7 +615,7 @@ babel_read_update(struct babel_tlv *hdr, union babel_msg *m,
     ip4_addr prefix4 = get_ip4(buf);
     net_fill_ip4(&msg->net, prefix4, tlv->plen);
 
-    if (tlv->flags & BABEL_FLAG_DEF_PREFIX)
+    if (tlv->flags & BABEL_UF_DEF_PREFIX)
     {
       put_ip4(state->def_ip4_prefix, prefix4);
       state->def_ip4_prefix_seen = 1;
@@ -629,13 +640,13 @@ babel_read_update(struct babel_tlv *hdr, union babel_msg *m,
     ip6_addr prefix6 = get_ip6(buf);
     net_fill_ip6(&msg->net, prefix6, tlv->plen);
 
-    if (tlv->flags & BABEL_FLAG_DEF_PREFIX)
+    if (tlv->flags & BABEL_UF_DEF_PREFIX)
     {
       put_ip6(state->def_ip6_prefix, prefix6);
       state->def_ip6_prefix_seen = 1;
     }
 
-    if (tlv->flags & BABEL_FLAG_ROUTER_ID)
+    if (tlv->flags & BABEL_UF_ROUTER_ID)
     {
       state->router_id = ((u64) _I2(prefix6)) << 32 | _I3(prefix6);
       state->router_id_seen = 1;
@@ -748,7 +759,7 @@ babel_write_update(struct babel_tlv *hdr, union babel_msg *m,
     else
     {
       put_ip6_px(tlv->addr, &msg->net);
-      tlv->flags |= BABEL_FLAG_DEF_PREFIX;
+      tlv->flags |= BABEL_UF_DEF_PREFIX;
 
       put_ip6(state->def_ip6_prefix, net6_prefix(&msg->net));
       state->def_ip6_pxlen = tlv->plen;
@@ -1294,7 +1305,7 @@ babel_rx_hook(sock *sk, uint len)
       sk->iface->name, sk->faddr, sk->laddr);
 
   /* Silently ignore my own packets */
-  if (ipa_equal(ifa->iface->addr->ip, sk->faddr))
+  if (ipa_equal(sk->faddr, sk->saddr))
     return 1;
 
   if (!ipa_is_link_local(sk->faddr))
@@ -1329,6 +1340,8 @@ babel_open_socket(struct babel_iface *ifa)
   sk->sport = ifa->cf->port;
   sk->dport = ifa->cf->port;
   sk->iface = ifa->iface;
+  sk->saddr = ifa->addr;
+  sk->vrf = p->p.vrf;
 
   sk->rx_hook = babel_rx_hook;
   sk->tx_hook = babel_tx_hook;

proto/bfd/bfd.c

@@ -64,16 +64,15 @@
  * 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
+ * which supports sockets, timers and events like the main loop. A birdloop is
- * (structure &timer2) are new microsecond based timers, while sockets and
+ * associated with a thread (field @thread) in which event hooks are executed.
- * events are the same. A birdloop is associated with a thread (field @thread)
+ * Most functions for setting event sources (like sk_start() or tm_start()) must
- * in which event hooks are executed. Most functions for setting event sources
+ * be called from the context of that thread. Birdloop allows to temporarily
- * (like sk_start() or tm2_start()) must be called from the context of that
+ * acquire the context of that thread for the main thread by calling
- * thread. Birdloop allows to temporarily acquire the context of that thread for
+ * birdloop_enter() and then birdloop_leave(), which also ensures mutual
- * the main thread by calling birdloop_enter() and then birdloop_leave(), which
+ * exclusion with all event hooks. Note that resources associated with a
- * also ensures mutual exclusion with all event hooks. Note that resources
+ * birdloop (like timers) should be attached to the independent resource pool,
- * associated with a birdloop (like timers) should be attached to the
+ * detached from the main resource tree.
- * 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
@@ -145,6 +144,7 @@ bfd_session_update_state(struct bfd_session *s, uint state, uint diag)
   bfd_lock_sessions(p);
   s->loc_state = state;
   s->loc_diag = diag;
+  s->last_state_change = current_time();
 
   notify = !NODE_VALID(&s->n);
   if (notify)
@@ -176,7 +176,7 @@ bfd_session_update_tx_interval(struct bfd_session *s)
     return;
 
   /* Set timer relative to last tx_timer event */
-  tm2_set(s->tx_timer, s->last_tx + tx_int_l);
+  tm_set(s->tx_timer, s->last_tx + tx_int_l);
 }
 
 static void
@@ -190,7 +190,7 @@ bfd_session_update_detection_time(struct bfd_session *s, int kick)
   if (!s->last_rx)
     return;
 
-  tm2_set(s->hold_timer, s->last_rx + timeout);
+  tm_set(s->hold_timer, s->last_rx + timeout);
 }
 
 static void
@@ -211,16 +211,16 @@ bfd_session_control_tx_timer(struct bfd_session *s, int reset)
     goto stop;
 
   /* So TX timer should run */
-  if (reset || !tm2_active(s->tx_timer))
+  if (reset || !tm_active(s->tx_timer))
   {
     s->last_tx = 0;
-    tm2_start(s->tx_timer, 0);
+    tm_start(s->tx_timer, 0);
   }
 
   return;
 
  stop:
-  tm2_stop(s->tx_timer);
+  tm_stop(s->tx_timer);
   s->last_tx = 0;
 }
 
@@ -379,7 +379,7 @@ bfd_find_session_by_addr(struct bfd_proto *p, ip_addr addr)
 }
 
 static void
-bfd_tx_timer_hook(timer2 *t)
+bfd_tx_timer_hook(timer *t)
 {
   struct bfd_session *s = t->data;
 
@@ -388,7 +388,7 @@ bfd_tx_timer_hook(timer2 *t)
 }
 
 static void
-bfd_hold_timer_hook(timer2 *t)
+bfd_hold_timer_hook(timer *t)
 {
   bfd_session_timeout(t->data);
 }
@@ -432,13 +432,13 @@ bfd_add_session(struct bfd_proto *p, ip_addr addr, ip_addr local, struct iface *
   s->passive = ifa->cf->passive;
   s->tx_csn = random_u32();
 
-  s->tx_timer = tm2_new_init(p->tpool, bfd_tx_timer_hook, s, 0, 0);
+  s->tx_timer = tm_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);
+  s->hold_timer = tm_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;
+  s->last_state_change = current_time();
 
   TRACE(D_EVENTS, "Session to %I added", s->addr);
 
@@ -879,9 +879,6 @@ bfd_notify_hook(sock *sk, uint len UNUSED)
     diag = s->loc_diag;
     bfd_unlock_sessions(p);
 
-    /* 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);
@@ -1080,7 +1077,7 @@ bfd_show_sessions(struct proto *P)
   byte tbuf[TM_DATETIME_BUFFER_SIZE];
   struct bfd_proto *p = (struct bfd_proto *) P;
   uint state, diag UNUSED;
-  u32 tx_int, timeout;
+  btime tx_int, timeout;
   const char *ifname;
 
   if (p->p.proto_state != PS_UP)
@@ -1101,15 +1098,14 @@ bfd_show_sessions(struct proto *P)
     state = s->loc_state;
     diag = s->loc_diag;
     ifname = (s->ifa && s->ifa->iface) ? s->ifa->iface->name : "---";
-    tx_int = s->last_tx ? (MAX(s->des_min_tx_int, s->rem_min_rx_int) TO_MS) : 0;
+    tx_int = s->last_tx ? MAX(s->des_min_tx_int, s->rem_min_rx_int) : 0;
-    timeout = (MAX(s->req_min_rx_int, s->rem_min_tx_int) TO_MS) * s->rem_detect_mult;
+    timeout = (btime) MAX(s->req_min_rx_int, s->rem_min_tx_int) * s->rem_detect_mult;
 
     state = (state < 4) ? state : 0;
-    tm_format_datetime(tbuf, &config->tf_proto, s->last_state_change);
+    tm_format_time(tbuf, &config->tf_proto, s->last_state_change);
 
-    cli_msg(-1020, "%-25I %-10s %-10s %-10s  %3u.%03u  %3u.%03u",
+    cli_msg(-1020, "%-25I %-10s %-10s %-10s  %7t  %7t",
-	    s->addr, ifname, bfd_state_names[state], tbuf,
+	    s->addr, ifname, bfd_state_names[state], tbuf, tx_int, timeout);
-	    tx_int / 1000, tx_int % 1000, timeout / 1000, timeout % 1000);
   }
   HASH_WALK_END;
 

proto/bfd/bfd.h

@@ -140,11 +140,11 @@ struct bfd_session
   btime last_tx;			/* Time of last sent periodic control packet */
   btime last_rx;			/* Time of last received valid control packet */
 
-  timer2 *tx_timer;			/* Periodic control packet timer */
+  timer *tx_timer;			/* Periodic control packet timer */
-  timer2 *hold_timer;			/* Timer for session down detection time */
+  timer *hold_timer;			/* Timer for session down detection time */
 
   list request_list;			/* List of client requests (struct bfd_request) */
-  bird_clock_t last_state_change;	/* Time of last state change */
+  btime last_state_change;		/* Time of last state change */
   u8 notify_running;			/* 1 if notify hooks are running */
 
   u8 rx_csn_known;			/* Received crypto sequence number is known */

proto/bfd/io.c

@@ -18,10 +18,10 @@
 #include "proto/bfd/io.h"
 
 #include "lib/buffer.h"
-#include "lib/heap.h"
 #include "lib/lists.h"
 #include "lib/resource.h"
 #include "lib/event.h"
+#include "lib/timer.h"
 #include "lib/socket.h"
 
 
@@ -31,16 +31,12 @@ struct birdloop
   pthread_t thread;
   pthread_mutex_t mutex;
 
-  btime last_time;
-  btime real_time;
-  u8 use_monotonic_clock;
-
   u8 stop_called;
   u8 poll_active;
   u8 wakeup_masked;
   int wakeup_fds[2];
 
-  BUFFER(timer2 *) timers;
+  struct timeloop time;
   list event_list;
   list sock_list;
   uint sock_num;
@@ -57,6 +53,7 @@ struct birdloop
  */
 
 static pthread_key_t current_loop_key;
+extern pthread_key_t current_time_key;
 
 static inline struct birdloop *
 birdloop_current(void)
@@ -68,6 +65,7 @@ static inline void
 birdloop_set_current(struct birdloop *loop)
 {
   pthread_setspecific(current_loop_key, loop);
+  pthread_setspecific(current_time_key, loop ? &loop->time : &main_timeloop);
 }
 
 static inline void
@@ -77,98 +75,6 @@ birdloop_init_current(void)
 }
 
 
-/*
- *	Time clock
- */
-
-static void times_update_alt(struct birdloop *loop);
-
-static void
-times_init(struct birdloop *loop)
-{
-  struct timespec ts;
-  int rv;
-
-  rv = clock_gettime(CLOCK_MONOTONIC, &ts);
-  if (rv < 0)
-  {
-    log(L_WARN "Monotonic clock is missing");
-
-    loop->use_monotonic_clock = 0;
-    loop->last_time = 0;
-    loop->real_time = 0;
-    times_update_alt(loop);
-    return;
-  }
-
-  if ((ts.tv_sec < 0) || (((s64) ts.tv_sec) > ((s64) 1 << 40)))
-    log(L_WARN "Monotonic clock is crazy");
-
-  loop->use_monotonic_clock = 1;
-  loop->last_time = ((s64) ts.tv_sec S) + (ts.tv_nsec / 1000);
-  loop->real_time = 0;
-}
-
-static void
-times_update_pri(struct birdloop *loop)
-{
-  struct timespec ts;
-  int rv;
-
-  rv = clock_gettime(CLOCK_MONOTONIC, &ts);
-  if (rv < 0)
-    die("clock_gettime: %m");
-
-  btime new_time = ((s64) ts.tv_sec S) + (ts.tv_nsec / 1000);
-
-  if (new_time < loop->last_time)
-    log(L_ERR "Monotonic clock is broken");
-
-  loop->last_time = new_time;
-  loop->real_time = 0;
-}
-
-static void
-times_update_alt(struct birdloop *loop)
-{
-  struct timeval tv;
-  int rv;
-
-  rv = gettimeofday(&tv, NULL);
-  if (rv < 0)
-    die("gettimeofday: %m");
-
-  btime new_time = ((s64) tv.tv_sec S) + tv.tv_usec;
-  btime delta = new_time - loop->real_time;
-
-  if ((delta < 0) || (delta > (60 S)))
-  {
-    if (loop->real_time)
-      log(L_WARN "Time jump, delta %d us", (int) delta);
-
-    delta = 100 MS;
-  }
-
-  loop->last_time += delta;
-  loop->real_time = new_time;
-}
-
-static void
-times_update(struct birdloop *loop)
-{
-  if (loop->use_monotonic_clock)
-    times_update_pri(loop);
-  else
-    times_update_alt(loop);
-}
-
-btime
-current_time(void)
-{
-  return birdloop_current()->last_time;
-}
-
-
 /*
  *	Wakeup code for birdloop
  */
@@ -238,7 +144,7 @@ wakeup_drain(struct birdloop *loop)
 }
 
 static inline void
-wakeup_do_kick(struct birdloop *loop) 
+wakeup_do_kick(struct birdloop *loop)
 {
   pipe_kick(loop->wakeup_fds[1]);
 }
@@ -252,6 +158,16 @@ wakeup_kick(struct birdloop *loop)
     loop->wakeup_masked = 2;
 }
 
+/* For notifications from outside */
+void
+wakeup_kick_current(void)
+{
+  struct birdloop *loop = birdloop_current();
+
+  if (loop && loop->poll_active)
+    wakeup_kick(loop);
+}
+
 
 /*
  *	Events
@@ -272,7 +188,7 @@ events_init(struct birdloop *loop)
 static void
 events_fire(struct birdloop *loop)
 {
-  times_update(loop);
+  times_update(&loop->time);
   ev_run_list(&loop->event_list);
 }
 
@@ -291,154 +207,6 @@ 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; }
-
-static inline timer2 *timers_first(struct birdloop *loop)
-{ return (loop->timers.used > 1) ? loop->timers.data[1] : NULL; }
-
-
-static void
-tm2_free(resource *r)
-{
-  timer2 *t = (timer2 *) r;
-
-  tm2_stop(t);
-}
-
-static void
-tm2_dump(resource *r)
-{
-  timer2 *t = (timer2 *) r;
-
-  debug("(code %p, data %p, ", t->hook, t->data);
-  if (t->randomize)
-    debug("rand %d, ", t->randomize);
-  if (t->recurrent)
-    debug("recur %d, ", t->recurrent);
-  if (t->expires)
-    debug("expires in %d ms)\n", (t->expires - current_time()) TO_MS);
-  else
-    debug("inactive)\n");
-}
-
-
-static struct resclass tm2_class = {
-  "Timer",
-  sizeof(timer2),
-  tm2_free,
-  tm2_dump,
-  NULL,
-  NULL
-};
-
-timer2 *
-tm2_new(pool *p)
-{
-  timer2 *t = ralloc(p, &tm2_class);
-  t->index = -1;
-  return t;
-}
-
-void
-tm2_set(timer2 *t, btime when)
-{
-  struct birdloop *loop = birdloop_current();
-  uint tc = timers_count(loop);
-
-  if (!t->expires)
-  {
-    t->index = ++tc;
-    t->expires = when;
-    BUFFER_PUSH(loop->timers) = t;
-    HEAP_INSERT(loop->timers.data, tc, timer2 *, TIMER_LESS, TIMER_SWAP);
-  }
-  else if (t->expires < when)
-  {
-    t->expires = when;
-    HEAP_INCREASE(loop->timers.data, tc, timer2 *, TIMER_LESS, TIMER_SWAP, t->index);
-  }
-  else if (t->expires > when)
-  {
-    t->expires = when;
-    HEAP_DECREASE(loop->timers.data, tc, timer2 *, TIMER_LESS, TIMER_SWAP, t->index);
-  }
-
-  if (loop->poll_active && (t->index == 1))
-    wakeup_kick(loop);
-}
-
-void
-tm2_start(timer2 *t, btime after)
-{
-  tm2_set(t, current_time() + MAX(after, 0));
-}
-
-void
-tm2_stop(timer2 *t)
-{
-  if (!t->expires)
-    return;
-
-  struct birdloop *loop = birdloop_current();
-  uint tc = timers_count(loop);
-
-  HEAP_DELETE(loop->timers.data, tc, timer2 *, TIMER_LESS, TIMER_SWAP, t->index);
-  BUFFER_POP(loop->timers);
-
-  t->index = -1;
-  t->expires = 0;
-}
-
-static void
-timers_init(struct birdloop *loop)
-{
-  BUFFER_INIT(loop->timers, loop->pool, 4);
-  BUFFER_PUSH(loop->timers) = NULL;
-}
-
-static void
-timers_fire(struct birdloop *loop)
-{
-  btime base_time;
-  timer2 *t;
-
-  times_update(loop);
-  base_time = loop->last_time;
-
-  while (t = timers_first(loop))
-  {
-    if (t->expires > base_time)
-      return;
-
-    if (t->recurrent)
-    {
-      btime when = t->expires + t->recurrent;
-      
-      if (when <= loop->last_time)
-	when = loop->last_time + t->recurrent;
-
-      if (t->randomize)
-	when += random() % (t->randomize + 1);
-
-      tm2_set(t, when);
-    }
-    else
-      tm2_stop(t);
-
-    t->hook(t);
-  }
-}
-
-
 /*
  *	Sockets
  */
@@ -586,7 +354,7 @@ sockets_fire(struct birdloop *loop)
   sock **psk = loop->poll_sk.data;
   int poll_num = loop->poll_fd.used - 1;
 
-  times_update(loop);
+  times_update(&loop->time);
 
   /* Last fd is internal wakeup fd */
   if (pfd[poll_num].revents & POLLIN)
@@ -634,11 +402,10 @@ birdloop_new(void)
   loop->pool = p;
   pthread_mutex_init(&loop->mutex, NULL);
 
-  times_init(loop);
   wakeup_init(loop);
 
   events_init(loop);
-  timers_init(loop);
+  timers_init(&loop->time, p);
   sockets_init(loop);
 
   return loop;
@@ -710,7 +477,7 @@ static void *
 birdloop_main(void *arg)
 {
   struct birdloop *loop = arg;
-  timer2 *t;
+  timer *t;
   int rv, timeout;
 
   birdloop_set_current(loop);
@@ -719,13 +486,13 @@ birdloop_main(void *arg)
   while (1)
   {
     events_fire(loop);
-    timers_fire(loop);
+    timers_fire(&loop->time);
 
-    times_update(loop);
+    times_update(&loop->time);
     if (events_waiting(loop))
       timeout = 0;
-    else if (t = timers_first(loop))
+    else if (t = timers_first(&loop->time))
-      timeout = (tm2_remains(t) TO_MS) + 1;
+      timeout = (tm_remains(t) TO_MS) + 1;
     else
       timeout = -1;
 
@@ -756,7 +523,7 @@ birdloop_main(void *arg)
     if (rv)
       sockets_fire(loop);
 
-    timers_fire(loop);
+    timers_fire(&loop->time);
   }
 
   loop->stop_called = 0;

proto/bfd/io.h

@@ -11,80 +11,15 @@
 #include "lib/lists.h"
 #include "lib/resource.h"
 #include "lib/event.h"
+#include "lib/timer.h"
 #include "lib/socket.h"
-// #include "sysdep/unix/timer.h"
 
 
-typedef struct timer2
-{
-  resource r;
-  void (*hook)(struct timer2 *);
-  void *data;
-
-  btime expires;			/* 0=inactive */
-  uint randomize;			/* Amount of randomization */
-  uint recurrent;			/* Timer recurrence */
-
-  int index;
-} timer2;
-
-
-btime current_time(void);
-
 void ev2_schedule(event *e);
 
-
-timer2 *tm2_new(pool *p);
-void tm2_set(timer2 *t, btime when);
-void tm2_start(timer2 *t, btime after);
-void tm2_stop(timer2 *t);
-
-static inline int
-tm2_active(timer2 *t)
-{
-  return t->expires != 0;
-}
-
-static inline btime 
-tm2_remains(timer2 *t)
-{
-  btime now = current_time();
-  return (t->expires > now) ? (t->expires - now) : 0;
-}
-
-static inline timer2 *
-tm2_new_init(pool *p, void (*hook)(struct timer2 *), void *data, uint rec, uint rand)
-{
-  timer2 *t = tm2_new(p);
-  t->hook = hook;
-  t->data = data;
-  t->recurrent = rec;
-  t->randomize = rand;
-  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)
-{
-  btime rem = tm2_remains(t);
-  tm2_start(t, MAX_(rem, after));
-}
-*/
-
-
 void sk_start(sock *s);
 void sk_stop(sock *s);
 
-
-
 struct birdloop *birdloop_new(void);
 void birdloop_start(struct birdloop *loop);
 void birdloop_stop(struct birdloop *loop);

proto/bgp/attrs.c

@@ -64,8 +64,6 @@
  * format - Optional hook that converts eattr to textual representation.
  */
 
-// XXXX review pool usage : c->c.proto->pool
-
 
 struct bgp_attr_desc {
   const char *name;
@@ -1175,6 +1173,22 @@ bgp_init_bucket_table(struct bgp_channel *c)
   c->withdraw_bucket = NULL;
 }
 
+void
+bgp_free_bucket_table(struct bgp_channel *c)
+{
+  HASH_FREE(c->bucket_hash);
+
+  struct bgp_bucket *b;
+  WALK_LIST_FIRST(b, c->bucket_queue)
+  {
+    rem_node(&b->send_node);
+    mb_free(b);
+  }
+
+  mb_free(c->withdraw_bucket);
+  c->withdraw_bucket = NULL;
+}
+
 static struct bgp_bucket *
 bgp_get_bucket(struct bgp_channel *c, ea_list *new)
 {

proto/bgp/bgp.c

@@ -98,11 +98,14 @@
  * <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
+ * <item> <rfc id="8203"> - BGP Administrative Shutdown Communication
  * </itemize>
 */
 
 #undef LOCAL_DEBUG
 
+#include <stdlib.h>
+
 #include "nest/bird.h"
 #include "nest/iface.h"
 #include "nest/protocol.h"
@@ -312,20 +315,21 @@ err1:
 /**
  * bgp_start_timer - start a BGP timer
  * @t: timer
- * @value: time to fire (0 to disable the timer)
+ * @value: time (in seconds) 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.
  */
 void
-bgp_start_timer(timer *t, int value)
+bgp_start_timer(timer *t, uint value)
 {
   if (value)
   {
-    /* The randomization procedure is specified in RFC 1771: 9.2.3.3 */
+    /* The randomization procedure is specified in RFC 4271 section 10 */
-    t->randomize = value / 4;
+    btime time = value S;
-    tm_start(t, value - t->randomize);
+    btime randomize = random() % ((time / 4) + 1);
+    tm_start(t, time - randomize);
   }
   else
     tm_stop(t);
@@ -381,10 +385,10 @@ bgp_update_startup_delay(struct bgp_proto *p)
 
   DBG("BGP: Updating startup delay\n");
 
-  if (p->last_proto_error && ((now - p->last_proto_error) >= (int) cf->error_amnesia_time))
+  if (p->last_proto_error && ((current_time() - p->last_proto_error) >= cf->error_amnesia_time S))
     p->startup_delay = 0;
 
-  p->last_proto_error = now;
+  p->last_proto_error = current_time();
 
   if (cf->disable_after_error)
   {
@@ -400,7 +404,7 @@ bgp_update_startup_delay(struct bgp_proto *p)
 }
 
 static void
-bgp_graceful_close_conn(struct bgp_conn *conn, uint subcode)
+bgp_graceful_close_conn(struct bgp_conn *conn, uint subcode, byte *data, uint len)
 {
   switch (conn->state)
   {
@@ -416,7 +420,7 @@ bgp_graceful_close_conn(struct bgp_conn *conn, uint subcode)
   case BS_OPENSENT:
   case BS_OPENCONFIRM:
   case BS_ESTABLISHED:
-    bgp_error(conn, 6, subcode, NULL, 0);
+    bgp_error(conn, 6, subcode, data, len);
     return;
 
   default:
@@ -456,11 +460,11 @@ bgp_decision(void *vp)
 }
 
 void
-bgp_stop(struct bgp_proto *p, uint subcode)
+bgp_stop(struct bgp_proto *p, uint subcode, byte *data, uint len)
 {
   proto_notify_state(&p->p, PS_STOP);
-  bgp_graceful_close_conn(&p->outgoing_conn, subcode);
+  bgp_graceful_close_conn(&p->outgoing_conn, subcode, data, len);
-  bgp_graceful_close_conn(&p->incoming_conn, subcode);
+  bgp_graceful_close_conn(&p->incoming_conn, subcode, data, len);
   ev_schedule(p->event);
 }
 
@@ -598,12 +602,8 @@ bgp_conn_leave_established_state(struct bgp_proto *p)
   BGP_TRACE(D_EVENTS, "BGP session closed");
   p->conn = NULL;
 
-  // 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);
+    bgp_stop(p, 0, NULL, 0);
 }
 
 void
@@ -661,6 +661,10 @@ bgp_handle_graceful_restart(struct bgp_proto *p)
   struct bgp_channel *c;
   WALK_LIST(c, p->p.channels)
   {
+    /* FIXME: perhaps check for channel state instead of disabled flag? */
+    if (c->c.disabled)
+      continue;
+
     if (c->gr_ready)
     {
       if (c->gr_active)
@@ -676,6 +680,13 @@ bgp_handle_graceful_restart(struct bgp_proto *p)
       rt_refresh_begin(c->c.table, &c->c);
       rt_refresh_end(c->c.table, &c->c);
     }
+
+    /* Reset bucket and prefix tables */
+    bgp_free_bucket_table(c);
+    bgp_free_prefix_table(c);
+    bgp_init_bucket_table(c);
+    bgp_init_prefix_table(c);
+    c->packets_to_send = 0;
   }
 
   proto_notify_state(&p->p, PS_START);
@@ -722,7 +733,7 @@ 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_stop(p, 0, NULL, 0);
 }
 
 
@@ -885,9 +896,9 @@ bgp_setup_conn(struct bgp_proto *p, struct bgp_conn *conn)
   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->connect_timer	= tm_new_init(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->hold_timer 	= tm_new_init(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->keepalive_timer	= tm_new_init(p->p.pool, bgp_keepalive_timeout, conn, 0, 0);
 
   conn->tx_ev = ev_new(p->p.pool);
   conn->tx_ev->hook = bgp_kick_tx;
@@ -936,6 +947,7 @@ bgp_connect(struct bgp_proto *p)	/* Enter Connect state and start establishing c
   s->daddr = p->cf->remote_ip;
   s->dport = p->cf->remote_port;
   s->iface = p->neigh ? p->neigh->iface : NULL;
+  s->vrf = p->p.vrf;
   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;
@@ -1094,19 +1106,8 @@ bgp_start_neighbor(struct bgp_proto *p)
 
   if (ipa_is_link_local(p->source_addr))
     p->link_addr = p->source_addr;
-  else
+  else if (p->neigh->iface->llv6)
-  {
+    p->link_addr = p->neigh->iface->llv6->ip;
-    /* 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);
 }
@@ -1132,7 +1133,7 @@ bgp_neigh_notify(neighbor *n)
       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);
+      bgp_stop(p, 0, NULL, 0);
     }
   }
   else if (p->cf->check_link && !(n->iface->flags & IF_LINK_UP))
@@ -1143,7 +1144,7 @@ bgp_neigh_notify(neighbor *n)
       bgp_store_error(p, NULL, BE_MISC, BEM_LINK_DOWN);
       if (ps == PS_UP)
 	bgp_update_startup_delay(p);
-      bgp_stop(p, 0);
+      bgp_stop(p, 0, NULL, 0);
     }
   }
   else
@@ -1168,7 +1169,7 @@ bgp_bfd_notify(struct bfd_request *req)
     bgp_store_error(p, NULL, BE_MISC, BEM_BFD_DOWN);
     if (ps == PS_UP)
       bgp_update_startup_delay(p);
-    bgp_stop(p, 0);
+    bgp_stop(p, 0, NULL, 0);
   }
 }
 
@@ -1311,13 +1312,8 @@ bgp_start(struct proto *P)
   p->event->hook = bgp_decision;
   p->event->data = p;
 
-  p->startup_timer = tm_new(p->p.pool);
+  p->startup_timer = tm_new_init(p->p.pool, bgp_startup_timeout, p, 0, 0);
-  p->startup_timer->hook = bgp_startup_timeout;
+  p->gr_timer = tm_new_init(p->p.pool, bgp_graceful_restart_timeout, p, 0, 0);
-  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)
@@ -1327,7 +1323,7 @@ bgp_start(struct proto *P)
   p->source_addr = p->cf->local_ip;
   p->link_addr = IPA_NONE;
 
-  /* XXXX */
+  /* Lock all channels when in GR recovery mode */
   if (p->p.gr_recovery && p->cf->gr_mode)
   {
     struct bgp_channel *c;
@@ -1344,6 +1340,7 @@ bgp_start(struct proto *P)
   lock->addr = p->cf->remote_ip;
   lock->port = p->cf->remote_port;
   lock->iface = p->cf->iface;
+  lock->vrf = p->cf->iface ? NULL : p->p.vrf;
   lock->type = OBJLOCK_TCP;
   lock->hook = bgp_start_locked;
   lock->data = p;
@@ -1360,6 +1357,10 @@ bgp_shutdown(struct proto *P)
   struct bgp_proto *p = (struct bgp_proto *) P;
   uint subcode = 0;
 
+  char *message = NULL;
+  byte *data = NULL;
+  uint len = 0;
+
   BGP_TRACE(D_EVENTS, "Shutdown requested");
 
   switch (P->down_code)
@@ -1376,10 +1377,12 @@ bgp_shutdown(struct proto *P)
   case PDC_CMD_DISABLE:
   case PDC_CMD_SHUTDOWN:
     subcode = 2; // Errcode 6, 2 - administrative shutdown
+    message = P->message;
     break;
 
   case PDC_CMD_RESTART:
     subcode = 4; // Errcode 6, 4 - administrative reset
+    message = P->message;
     break;
 
   case PDC_RX_LIMIT_HIT:
@@ -1404,8 +1407,22 @@ bgp_shutdown(struct proto *P)
   bgp_store_error(p, NULL, BE_MAN_DOWN, 0);
   p->startup_delay = 0;
 
+  /* RFC 8203 - shutdown communication */
+  if (message)
+  {
+    uint msg_len = strlen(message);
+    msg_len = MIN(msg_len, 128);
+
+    /* Buffer will be freed automatically by protocol shutdown */
+    data = mb_alloc(p->p.pool, msg_len + 1);
+    len = msg_len + 1;
+
+    data[0] = msg_len;
+    memcpy(data+1, message, msg_len);
+  }
+
 done:
-  bgp_stop(p, subcode);
+  bgp_stop(p, subcode, data, len);
   return p->p.proto_state;
 }
 
@@ -1495,6 +1512,20 @@ bgp_channel_start(struct channel *C)
       c->next_hop_addr = src;
   }
 
+  /* Use preferred addresses associated with interface / source address */
+  if (ipa_zero(c->next_hop_addr))
+  {
+    /* We know the iface for single-hop, we make lookup for multihop */
+    struct neighbor *nbr = p->neigh ?: neigh_find2(&p->p, &src, NULL, 0);
+    struct iface *iface = nbr ? nbr->iface : NULL;
+
+    if (bgp_channel_is_ipv4(c) && iface && iface->addr4)
+      c->next_hop_addr = iface->addr4->ip;
+
+    if (bgp_channel_is_ipv6(c) && iface && iface->addr6)
+      c->next_hop_addr = iface->addr6->ip;
+  }
+
   /* Exit if no feasible next hop address is found */
   if (ipa_zero(c->next_hop_addr))
   {
@@ -1523,10 +1554,9 @@ 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;
+  c->packets_to_send = 0;
 }
 
 static void
@@ -1598,6 +1628,10 @@ bgp_postconfig(struct proto_config *CF)
   if (cf->multihop < 0)
     cf->multihop = internal ? 64 : 0;
 
+  /* Link check for single-hop BGP by default */
+  if (cf->check_link < 0)
+    cf->check_link = !cf->multihop;
+
 
   if (!cf->local_as)
     cf_error("Local AS number must be set");
@@ -1796,7 +1830,7 @@ bgp_error(struct bgp_conn *c, uint code, uint subcode, byte *data, int len)
   if (code != 6)
   {
     bgp_update_startup_delay(p);
-    bgp_stop(p, 0);
+    bgp_stop(p, 0, NULL, 0);
   }
 }
 
@@ -2017,17 +2051,18 @@ bgp_show_proto_info(struct proto *P)
     struct bgp_conn *oc = &p->outgoing_conn;
 
     if ((p->start_state < BSS_CONNECT) &&
-	(p->startup_timer->expires))
+	(tm_active(p->startup_timer)))
-      cli_msg(-1006, "    Error wait:       %d/%d",
+      cli_msg(-1006, "    Error wait:       %t/%u",
-	      p->startup_timer->expires - now, p->startup_delay);
+	      tm_remains(p->startup_timer), p->startup_delay);
 
     if ((oc->state == BS_ACTIVE) &&
-	(oc->connect_timer->expires))
+	(tm_active(oc->connect_timer)))
-      cli_msg(-1006, "    Connect delay:    %d/%d",
+      cli_msg(-1006, "    Connect delay:    %t/%u",
-	      oc->connect_timer->expires - now, p->cf->connect_delay_time);
+	      tm_remains(oc->connect_timer), p->cf->connect_delay_time);
 
-    if (p->gr_active_num && p->gr_timer->expires)
+    if (p->gr_active_num && tm_active(p->gr_timer))
-      cli_msg(-1006, "    Restart timer:    %d/-", p->gr_timer->expires - now);
+      cli_msg(-1006, "    Restart timer:    %t/-",
+	      tm_remains(p->gr_timer));
   }
   else if (P->proto_state == PS_UP)
   {
@@ -2036,21 +2071,16 @@ bgp_show_proto_info(struct proto *P)
     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",
-      cli_msg(-1006, "    Session:          %s%s%s%s%s%s%s%s",
+	    p->is_internal ? "internal" : "external",
-	      p->is_internal ? "internal" : "external",
+	    p->cf->multihop ? " multihop" : "",
-	      p->cf->multihop ? " multihop" : "",
+	    p->rr_client ? " route-reflector" : "",
-	      p->rr_client ? " route-reflector" : "",
+	    p->rs_client ? " route-server" : "",
-	      p->rs_client ? " route-server" : "",
+	    p->as4_session ? " AS4" : "");
-	      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",
+    cli_msg(-1006, "    Hold timer:       %t/%u",
 	    tm_remains(p->conn->hold_timer), p->conn->hold_time);
-    cli_msg(-1006, "    Keepalive timer:  %d/%d",
+    cli_msg(-1006, "    Keepalive timer:  %t/%u",
 	    tm_remains(p->conn->keepalive_timer), p->conn->keepalive_time);
   }
 
@@ -2063,12 +2093,19 @@ bgp_show_proto_info(struct proto *P)
   }
 
   {
-    /* XXXX ?? */
     struct bgp_channel *c;
     WALK_LIST(c, p->p.channels)
     {
       channel_show_info(&c->c);
 
+      if (c->c.channel_state == CS_UP)
+      {
+	if (ipa_zero(c->link_addr))
+	  cli_msg(-1006, "    BGP Next hop:   %I", c->next_hop_addr);
+	else
+	  cli_msg(-1006, "    BGP Next hop:   %I %I", c->next_hop_addr, c->link_addr);
+      }
+
       if (c->igp_table_ip4)
 	cli_msg(-1006, "    IGP IPv4 table: %s", c->igp_table_ip4->name);
 

proto/bgp/bgp.h

@@ -108,6 +108,7 @@ struct bgp_config {
   int allow_local_pref;			/* Allow LOCAL_PREF in EBGP sessions */
   int gr_mode;				/* Graceful restart mode (BGP_GR_*) */
   int setkey;				/* Set MD5 password to system SA/SP database */
+  /* Times below are in seconds */
   unsigned gr_time;			/* Graceful restart timeout */
   unsigned connect_delay_time;		/* Minimum delay between connect attempts */
   unsigned connect_retry_time;		/* Timeout for connect attempts */
@@ -210,10 +211,10 @@ struct bgp_conn {
 
   struct bgp_caps *local_caps;
   struct bgp_caps *remote_caps;
-  struct timer *connect_timer;
+  timer *connect_timer;
-  struct timer *hold_timer;
+  timer *hold_timer;
-  struct timer *keepalive_timer;
+  timer *keepalive_timer;
-  struct event *tx_ev;
+  event *tx_ev;
   u32 packets_to_send;			/* Bitmap of packet types to be sent */
   u32 channels_to_send;			/* Bitmap of channels with packets to be sent */
   u8 last_channel;			/* Channel used last time for TX */
@@ -254,11 +255,11 @@ struct bgp_proto {
   struct bfd_request *bfd_req;		/* BFD request, if BFD is used */
   ip_addr source_addr;			/* Local address used as an advertised next hop */
   ip_addr link_addr;			/* Link-local version of source_addr */
-  struct event *event;			/* Event for respawning and shutting process */
+  event *event;				/* Event for respawning and shutting process */
-  struct timer *startup_timer;		/* Timer used to delay protocol startup due to previous errors (startup_delay) */
+  timer *startup_timer;			/* Timer used to delay protocol startup due to previous errors (startup_delay) */
-  struct timer *gr_timer;		/* Timer waiting for reestablishment after graceful restart */
+  timer *gr_timer;			/* Timer waiting for reestablishment after graceful restart */
-  unsigned startup_delay;		/* Time to delay protocol startup by due to errors */
+  uint startup_delay;			/* Delay (in seconds) of protocol startup due to previous errors */
-  bird_clock_t last_proto_error;	/* Time of last error that leads to protocol stop */
+  btime last_proto_error;		/* Time of last error that leads to protocol stop */
   u8 last_error_class; 			/* Error class of last error */
   u32 last_error_code;			/* Error code of last error. BGP protocol errors
 					   are encoded as (bgp_err_code << 16 | bgp_err_subcode) */
@@ -422,7 +423,7 @@ extern struct linpool *bgp_linpool;
 extern struct linpool *bgp_linpool2;
 
 
-void bgp_start_timer(struct timer *t, int value);
+void bgp_start_timer(timer *t, uint value);
 void bgp_check_config(struct bgp_config *c);
 void bgp_error(struct bgp_conn *c, unsigned code, unsigned subcode, byte *data, int len);
 void bgp_close_conn(struct bgp_conn *c);
@@ -436,7 +437,7 @@ void bgp_graceful_restart_done(struct bgp_channel *c);
 void bgp_refresh_begin(struct bgp_channel *c);
 void bgp_refresh_end(struct bgp_channel *c);
 void bgp_store_error(struct bgp_proto *p, struct bgp_conn *c, u8 class, u32 code);
-void bgp_stop(struct bgp_proto *p, unsigned subcode);
+void bgp_stop(struct bgp_proto *p, uint subcode, byte *data, uint len);
 
 struct rte_source *bgp_find_source(struct bgp_proto *p, u32 path_id);
 struct rte_source *bgp_get_source(struct bgp_proto *p, u32 path_id);
@@ -491,11 +492,13 @@ int bgp_encode_attrs(struct bgp_write_state *s, ea_list *attrs, byte *buf, byte
 ea_list * bgp_decode_attrs(struct bgp_parse_state *s, byte *data, uint len);
 
 void bgp_init_bucket_table(struct bgp_channel *c);
+void bgp_free_bucket_table(struct bgp_channel *c);
 void bgp_free_bucket(struct bgp_channel *c, struct bgp_bucket *b);
 void bgp_defer_bucket(struct bgp_channel *c, struct bgp_bucket *b);
 void bgp_withdraw_bucket(struct bgp_channel *c, struct bgp_bucket *b);
 
 void bgp_init_prefix_table(struct bgp_channel *c);
+void bgp_free_prefix_table(struct bgp_channel *c);
 void bgp_free_prefix(struct bgp_channel *c, struct bgp_prefix *bp);
 
 int bgp_rte_better(struct rte *, struct rte *);

proto/bgp/config.Y

@@ -58,6 +58,7 @@ bgp_proto_start: proto_start BGP {
      BGP_CFG->gr_mode = BGP_GR_AWARE;
      BGP_CFG->gr_time = 120;
      BGP_CFG->setkey = 1;
+     BGP_CFG->check_link = -1;
    }
  ;
 

proto/bgp/packets.c

@@ -1483,7 +1483,8 @@ bgp_encode_nlri_vpn6(struct bgp_write_state *s, struct bgp_bucket *buck, byte *b
     ADVANCE(pos, size, 1);
 
     /* Encode MPLS labels */
-    bgp_encode_mpls_labels(s, s->mpls_labels, &pos, &size, pos - 1);
+    if (s->mpls)
+      bgp_encode_mpls_labels(s, s->mpls_labels, &pos, &size, pos - 1);
 
     /* Encode route distinguisher */
     put_u64(pos, net->rd);
@@ -1636,8 +1637,8 @@ bgp_decode_nlri_flow4(struct bgp_parse_state *s, byte *pos, uint len, rta *a)
     uint pxlen = data[1];
 
     // FIXME: Use some generic function
-    memcpy(&px, data, BYTES(pxlen));
+    memcpy(&px, data+2, BYTES(pxlen));
-    px = ip4_and(px, ip4_mkmask(pxlen));
+    px = ip4_and(ip4_ntoh(px), ip4_mkmask(pxlen));
 
     /* Prepare the flow */
     net_addr *n = alloca(sizeof(struct net_addr_flow4) + flen);
@@ -1728,8 +1729,8 @@ bgp_decode_nlri_flow6(struct bgp_parse_state *s, byte *pos, uint len, rta *a)
     uint pxlen = data[1];
 
     // FIXME: Use some generic function
-    memcpy(&px, data, BYTES(pxlen));
+    memcpy(&px, data+2, BYTES(pxlen));
-    px = ip6_and(px, ip6_mkmask(pxlen));
+    px = ip6_and(ip6_ntoh(px), ip6_mkmask(pxlen));
 
     /* Prepare the flow */
     net_addr *n = alloca(sizeof(struct net_addr_flow6) + flen);
@@ -2678,38 +2679,72 @@ bgp_error_dsc(uint code, uint subcode)
   return buff;
 }
 
+/* RFC 8203 - shutdown communication message */
+static int
+bgp_handle_message(struct bgp_proto *p, byte *data, uint len, byte **bp)
+{
+  byte *msg = data + 1;
+  uint msg_len = data[0];
+  uint i;
+
+  /* Handle zero length message */
+  if (msg_len == 0)
+    return 1;
+
+  /* Handle proper message */
+  if ((msg_len > 128) && (msg_len + 1 > len))
+    return 0;
+
+  /* Some elementary cleanup */
+  for (i = 0; i < msg_len; i++)
+    if (msg[i] < ' ')
+      msg[i] = ' ';
+
+  proto_set_message(&p->p, msg, msg_len);
+  *bp += bsprintf(*bp, ": \"%s\"", p->p.message);
+  return 1;
+}
+
 void
 bgp_log_error(struct bgp_proto *p, u8 class, char *msg, uint code, uint subcode, byte *data, uint len)
 {
-  const byte *name;
+  byte argbuf[256], *t = argbuf;
-  byte *t, argbuf[36];
   uint i;
 
   /* Don't report Cease messages generated by myself */
   if (code == 6 && class == BE_BGP_TX)
     return;
 
-  name = bgp_error_dsc(code, subcode);
+  /* Reset shutdown message */
-  t = argbuf;
+  if ((code == 6) && ((subcode == 2) || (subcode == 4)))
+    proto_set_message(&p->p, NULL, 0);
+
   if (len)
     {
-      *t++ = ':';
+      /* Bad peer AS - we would like to print the AS */
-      *t++ = ' ';
-
       if ((code == 2) && (subcode == 2) && ((len == 2) || (len == 4)))
 	{
-	  /* Bad peer AS - we would like to print the AS */
+	  t += bsprintf(t, ": %u", (len == 2) ? get_u16(data) : get_u32(data));
-	  t += bsprintf(t, "%u", (len == 2) ? get_u16(data) : get_u32(data));
 	  goto done;
 	}
+
+      /* RFC 8203 - shutdown communication */
+      if (((code == 6) && ((subcode == 2) || (subcode == 4))))
+	if (bgp_handle_message(p, data, len, &t))
+	  goto done;
+
+      *t++ = ':';
+      *t++ = ' ';
       if (len > 16)
 	len = 16;
       for (i=0; i<len; i++)
 	t += bsprintf(t, "%02x", data[i]);
     }
- done:
+
+done:
   *t = 0;
-  log(L_REMOTE "%s: %s: %s%s", p->p.name, msg, name, argbuf);
+  const byte *dsc = bgp_error_dsc(code, subcode);
+  log(L_REMOTE "%s: %s: %s%s", p->p.name, msg, dsc, argbuf);
 }
 
 static void
@@ -2733,7 +2768,7 @@ bgp_rx_notification(struct bgp_conn *conn, byte *pkt, uint len)
   if (err)
   {
     bgp_update_startup_delay(p);
-    bgp_stop(p, 0);
+    bgp_stop(p, 0, NULL, 0);
   }
 }
 

proto/ospf/config.Y

@@ -78,18 +78,66 @@ static void
 ospf_proto_finish(void)
 {
   struct ospf_config *cf = OSPF_CFG;
-
+  struct ospf_area_config *ac;
-  if (EMPTY_LIST(cf->area_list))
+  struct ospf_iface_patt *ic;
-    cf_error( "No configured areas in OSPF");
 
   /* Define default channel */
   if (EMPTY_LIST(this_proto->channels))
+  {
+    this_proto->net_type = ospf_cfg_is_v2() ? NET_IP4 : NET_IP6;
     channel_config_new(NULL, this_proto->net_type, this_proto);
+  }
+
+  /* Propagate global instance ID to interfaces */
+  if (cf->instance_id_set)
+  {
+    WALK_LIST(ac, cf->area_list)
+      WALK_LIST(ic, ac->patt_list)
+	if (!ic->instance_id_set)
+	{ ic->instance_id = cf->instance_id; ic->instance_id_set = 1; }
+
+    WALK_LIST(ic, cf->vlink_list)
+      if (!ic->instance_id_set)
+      { ic->instance_id = cf->instance_id; ic->instance_id_set = 1; }
+  }
+
+  if (ospf_cfg_is_v3())
+  {
+    uint ipv4 = (this_proto->net_type == NET_IP4);
+    uint base = (ipv4 ? 64 : 0) + (cf->af_mc ? 32 : 0);
+
+    /* RFC 5838 - OSPFv3-AF */
+    if (cf->af_ext)
+    {
+      /* RFC 5838 2.1 - instance IDs based on AFs */
+      WALK_LIST(ac, cf->area_list)
+	WALK_LIST(ic, ac->patt_list)
+	{
+	  if (!ic->instance_id_set)
+	    ic->instance_id = base;
+	  else if (ic->instance_id >= 128)
+	    log(L_WARN "Instance ID %d from unassigned/private range", ic->instance_id);
+	  else if ((ic->instance_id < base) || (ic->instance_id >= (base + 32)))
+	    cf_error("Instance ID %d invalid for given channel type", ic->instance_id);
+	}
+
+      /* RFC 5838 2.8 - vlinks limited to IPv6 unicast */
+      if ((ipv4 || cf->af_mc) && !EMPTY_LIST(cf->vlink_list))
+	cf_error("Vlinks not supported in AFs other than IPv6 unicast");
+    }
+    else
+    {
+      if (ipv4 || cf->af_mc)
+	cf_error("Different channel type");
+    }
+  }
+
+  if (EMPTY_LIST(cf->area_list))
+    cf_error("No configured areas in OSPF");
 
   int areano = 0;
   int backbone = 0;
   int nssa = 0;
-  struct ospf_area_config *ac;
   WALK_LIST(ac, cf->area_list)
   {
     areano++;
@@ -148,10 +196,11 @@ CF_KEYWORDS(ELIGIBLE, POLL, NETWORKS, HIDDEN, VIRTUAL, CHECK, LINK, ONLY, BFD)
 CF_KEYWORDS(RX, BUFFER, LARGE, NORMAL, STUBNET, HIDDEN, SUMMARY, TAG, EXTERNAL)
 CF_KEYWORDS(WAIT, DELAY, LSADB, ECMP, LIMIT, WEIGHT, NSSA, TRANSLATOR, STABILITY)
 CF_KEYWORDS(GLOBAL, LSID, ROUTER, SELF, INSTANCE, REAL, NETMASK, TX, PRIORITY, LENGTH)
-CF_KEYWORDS(SECONDARY, MERGE, LSA, SUPPRESSION)
+CF_KEYWORDS(SECONDARY, MERGE, LSA, SUPPRESSION, MULTICAST, RFC5838)
 
 %type <ld> lsadb_args
-%type <i> ospf_variant nbma_eligible
+%type <i> ospf_variant ospf_af_mc nbma_eligible
+%type <cc> ospf_channel_start ospf_channel
 
 CF_GRAMMAR
 
@@ -166,12 +215,14 @@ ospf_variant:
 ospf_proto_start: proto_start ospf_variant
 {
   this_proto = proto_config_new(&proto_ospf, $1);
-  this_proto->net_type = $2 ? NET_IP4 : NET_IP6;
+  this_proto->net_type = $2 ? NET_IP4 : 0;
 
   init_list(&OSPF_CFG->area_list);
   init_list(&OSPF_CFG->vlink_list);
+  OSPF_CFG->ecmp = rt_default_ecmp;
   OSPF_CFG->tick = OSPF_DEFAULT_TICK;
   OSPF_CFG->ospf2 = $2;
+  OSPF_CFG->af_ext = !$2;
 };
 
 ospf_proto:
@@ -179,16 +230,34 @@ ospf_proto:
  | ospf_proto ospf_proto_item ';'
  ;
 
+ospf_af_mc:
+             { $$ = 0; }
+ | MULTICAST { $$ = 1; }
+ ;
+
+/* We redefine proto_channel to add multicast flag */
+ospf_channel_start: net_type ospf_af_mc
+{
+  $$ = this_channel = channel_config_new(NULL, $1, this_proto);
+
+  /* Save the multicast flag */
+  if (this_channel == proto_cf_main_channel(this_proto))
+    OSPF_CFG->af_mc = $2;
+};
+
+ospf_channel: ospf_channel_start channel_opt_list channel_end;
+
 ospf_proto_item:
    proto_item
- | proto_channel
+ | ospf_channel { this_proto->net_type = $1->net_type; }
  | RFC1583COMPAT bool { OSPF_CFG->rfc1583 = $2; }
+ | RFC5838 bool { OSPF_CFG->af_ext = $2; if (!ospf_cfg_is_v3()) cf_error("RFC5838 option requires OSPFv3"); }
  | STUB ROUTER bool { OSPF_CFG->stub_router = $3; }
  | ECMP bool { OSPF_CFG->ecmp = $2 ? OSPF_DEFAULT_ECMP_LIMIT : 0; }
  | ECMP bool LIMIT expr { OSPF_CFG->ecmp = $2 ? $4 : 0; }
  | MERGE EXTERNAL bool { OSPF_CFG->merge_external = $3; }
  | TICK expr { OSPF_CFG->tick = $2; if($2 <= 0) cf_error("Tick must be greater than zero"); }
- | INSTANCE ID expr { OSPF_CFG->instance_id = $3; if ($3 > 255) cf_error("Instance ID must be in range 0-255"); }
+ | INSTANCE ID expr { OSPF_CFG->instance_id = $3; OSPF_CFG->instance_id_set = 1; if ($3 > 255) cf_error("Instance ID must be in range 0-255"); }
  | ospf_area
  ;
 
@@ -293,7 +362,6 @@ ospf_vlink_start: VIRTUAL LINK idval
   OSPF_PATT->inftransdelay = INFTRANSDELAY_D;
   OSPF_PATT->deadc = DEADC_D;
   OSPF_PATT->type = OSPF_IT_VLINK;
-  OSPF_PATT->instance_id = OSPF_CFG->instance_id;
   init_list(&OSPF_PATT->nbma_list);
   reset_passwords();
  }
@@ -393,8 +461,8 @@ ospf_iface_start:
   OSPF_PATT->priority = PRIORITY_D;
   OSPF_PATT->deadc = DEADC_D;
   OSPF_PATT->type = OSPF_IT_UNDEF;
-  OSPF_PATT->instance_id = OSPF_CFG->instance_id;
   init_list(&OSPF_PATT->nbma_list);
+  OSPF_PATT->check_link = 1;
   OSPF_PATT->ptp_netmask = 2; /* not specified */
   OSPF_PATT->tx_tos = IP_PREC_INTERNET_CONTROL;
   OSPF_PATT->tx_priority = sk_priority_control;
@@ -404,7 +472,7 @@ ospf_iface_start:
 
 ospf_instance_id:
    /* empty */
- | INSTANCE expr { OSPF_PATT->instance_id = $2; if ($2 > 255) cf_error("Instance ID must be in range 0-255"); }
+ | INSTANCE expr { OSPF_PATT->instance_id = $2; OSPF_PATT->instance_id_set = 1; if ($2 > 255) cf_error("Instance ID must be in range 0-255"); }
  ;
 
 ospf_iface_patt_list:
@@ -431,7 +499,7 @@ CF_ADDTO(dynamic_attr, OSPF_TAG { $$ = f_new_dynamic_attr(EAF_TYPE_INT | EAF_TEM
 CF_ADDTO(dynamic_attr, OSPF_ROUTER_ID { $$ = f_new_dynamic_attr(EAF_TYPE_ROUTER_ID | EAF_TEMP, T_QUAD, EA_OSPF_ROUTER_ID); })
 
 CF_CLI_HELP(SHOW OSPF, ..., [[Show information about OSPF protocol]]);
-CF_CLI(SHOW OSPF, optsym, [<name>], [[Show information about OSPF protocol XXX]], CLI_SF_PROTOCOL)
+CF_CLI(SHOW OSPF, optsym, [<name>], [[Show information about OSPF protocol]], CLI_SF_PROTOCOL)
 { ospf_sh(proto_get_named($3, &proto_ospf)); };
 
 CF_CLI(SHOW OSPF NEIGHBORS, optsym opttext, [<name>] [\"<interface>\"], [[Show information about OSPF neighbors]], CLI_SF_PROTOCOL | CLI_SF_INTERFACE | CLI_SF_PARAMETER)

proto/ospf/dbdes.c

@@ -356,7 +356,7 @@ ospf_receive_dbdes(struct ospf_packet *pkt, struct ospf_iface *ifa,
       LOG_PKT_WARN("MTU mismatch with nbr %R on %s (remote %d, local %d)",
 		   n->rid, ifa->ifname, rcv_iface_mtu, ifa->iface->mtu);
 
-    if ((rcv_imms == DBDES_IMMS) &&
+    if (((rcv_imms & DBDES_IMMS) == DBDES_IMMS) &&
 	(n->rid > p->router_id) &&
 	(plen == ospf_dbdes_hdrlen(p)))
     {
@@ -428,7 +428,7 @@ ospf_receive_dbdes(struct ospf_packet *pkt, struct ospf_iface *ifa,
       }
 
       ospf_send_dbdes(p, n);
-      tm_start(n->dbdes_timer, n->ifa->rxmtint);
+      tm_start(n->dbdes_timer, n->ifa->rxmtint S);
     }
     else
     {

proto/ospf/hello.c

@@ -32,10 +32,7 @@ struct ospf_hello3_packet
   struct ospf_packet hdr;
 
   u32 iface_id;
-  u8 priority;
+  u32 options;
-  u8 options3;
-  u8 options2;
-  u8 options;
   u16 helloint;
   u16 deadint;
   u32 dr;
@@ -91,10 +88,7 @@ ospf_send_hello(struct ospf_iface *ifa, int kind, struct ospf_neighbor *dirn)
     struct ospf_hello3_packet *ps = (void *) pkt;
 
     ps->iface_id = htonl(ifa->iface_id);
-    ps->priority = ifa->priority;
+    ps->options = ntohl(ifa->oa->options | (ifa->priority << 24));
-    ps->options3 = ifa->oa->options >> 16;
-    ps->options2 = ifa->oa->options >> 8;
-    ps->options = ifa->oa->options;
     ps->helloint = ntohs(ifa->helloint);
     ps->deadint = htons(ifa->deadint);
     ps->dr = htonl(ifa->drid);
@@ -190,7 +184,8 @@ ospf_receive_hello(struct ospf_packet *pkt, struct ospf_iface *ifa,
   struct ospf_proto *p = ifa->oa->po;
   const char *err_dsc = NULL;
   u32 rcv_iface_id, rcv_helloint, rcv_deadint, rcv_dr, rcv_bdr;
-  u8 rcv_options, rcv_priority;
+  uint rcv_options, rcv_priority;
+  uint loc_options = ifa->oa->options;
   u32 *neighbors;
   u32 neigh_count;
   uint plen, i, err_val = 0;
@@ -245,8 +240,8 @@ ospf_receive_hello(struct ospf_packet *pkt, struct ospf_iface *ifa,
     rcv_deadint = ntohs(ps->deadint);
     rcv_dr = ntohl(ps->dr);
     rcv_bdr = ntohl(ps->bdr);
-    rcv_options = ps->options;
+    rcv_options = ntohl(ps->options) & 0x00FFFFFF;
-    rcv_priority = ps->priority;
+    rcv_priority = ntohl(ps->options) >> 24;
 
     neighbors = ps->neighbors;
     neigh_count = (plen - sizeof(struct ospf_hello3_packet)) / sizeof(u32);
@@ -259,9 +254,13 @@ ospf_receive_hello(struct ospf_packet *pkt, struct ospf_iface *ifa,
     DROP("dead interval mismatch", rcv_deadint);
 
   /* Check whether bits E, N match */
-  if ((rcv_options ^ ifa->oa->options) & (OPT_E | OPT_N))
+  if ((rcv_options ^ loc_options) & (OPT_E | OPT_N))
     DROP("area type mismatch", rcv_options);
 
+  /* RFC 5838 2.4 - AF-bit check unless on IPv6 unicast */
+  if ((loc_options & OPT_AF) && !(loc_options & OPT_V6) && !(rcv_options & OPT_AF))
+    DROP("AF-bit mismatch", rcv_options);
+
   /* Check consistency of existing neighbor entry */
   if (n)
   {

proto/ospf/iface.c

@@ -121,6 +121,7 @@ ospf_sk_open(struct ospf_iface *ifa)
   sk->dport = OSPF_PROTO;
   sk->saddr = ifa->addr->ip;
   sk->iface = ifa->iface;
+  sk->vrf = p->p.vrf;
 
   sk->tos = ifa->cf->tx_tos;
   sk->priority = ifa->cf->tx_priority;
@@ -204,6 +205,7 @@ ospf_open_vlink_sk(struct ospf_proto *p)
   sk->type = SK_IP;
   sk->subtype = ospf_is_v2(p) ? SK_IPV4 : SK_IPV6;
   sk->dport = OSPF_PROTO;
+  sk->vrf = p->p.vrf;
 
   /* FIXME: configurable tos/priority ? */
   sk->tos = IP_PREC_INTERNET_CONTROL;
@@ -396,15 +398,15 @@ ospf_iface_sm(struct ospf_iface *ifa, int event)
 	{
 	  ospf_iface_chstate(ifa, OSPF_IS_WAITING);
 	  if (ifa->wait_timer)
-	    tm_start(ifa->wait_timer, ifa->waitint);
+	    tm_start(ifa->wait_timer, ifa->waitint S);
 	}
       }
 
       if (ifa->hello_timer)
-	tm_start(ifa->hello_timer, ifa->helloint);
+	tm_start(ifa->hello_timer, ifa->helloint S);
 
       if (ifa->poll_timer)
-	tm_start(ifa->poll_timer, ifa->pollint);
+	tm_start(ifa->poll_timer, ifa->pollint S);
 
       ospf_send_hello(ifa, OHS_HELLO, NULL);
     }
@@ -494,13 +496,13 @@ ospf_iface_add(struct object_lock *lock)
 
   if (! ifa->stub)
   {
-    ifa->hello_timer = tm_new_set(ifa->pool, hello_timer_hook, ifa, 0, ifa->helloint);
+    ifa->hello_timer = tm_new_init(ifa->pool, hello_timer_hook, ifa, ifa->helloint S, 0);
 
     if (ifa->type == OSPF_IT_NBMA)
-      ifa->poll_timer = tm_new_set(ifa->pool, poll_timer_hook, ifa, 0, ifa->pollint);
+      ifa->poll_timer = tm_new_init(ifa->pool, poll_timer_hook, ifa, ifa->pollint S, 0);
 
     if ((ifa->type == OSPF_IT_BCAST) || (ifa->type == OSPF_IT_NBMA))
-      ifa->wait_timer = tm_new_set(ifa->pool, wait_timer_hook, ifa, 0, 0);
+      ifa->wait_timer = tm_new_init(ifa->pool, wait_timer_hook, ifa, 0, 0);
 
     ifa->flood_queue_size = ifa_flood_queue_size(ifa);
     ifa->flood_queue = mb_allocz(ifa->pool, ifa->flood_queue_size * sizeof(void *));
@@ -703,7 +705,7 @@ ospf_iface_new_vlink(struct ospf_proto *p, struct ospf_iface_patt *ip)
 
   add_tail(&p->iface_list, NODE ifa);
 
-  ifa->hello_timer = tm_new_set(ifa->pool, hello_timer_hook, ifa, 0, ifa->helloint);
+  ifa->hello_timer = tm_new_init(ifa->pool, hello_timer_hook, ifa, ifa->helloint S, 0);
 
   ifa->flood_queue_size = ifa_flood_queue_size(ifa);
   ifa->flood_queue = mb_allocz(ifa->pool, ifa->flood_queue_size * sizeof(void *));
@@ -715,10 +717,10 @@ ospf_iface_change_timer(timer *tm, uint val)
   if (!tm)
     return;
 
-  tm->recurrent = val;
+  tm->recurrent = val S;
 
-  if (tm->expires)
+  if (tm_active(tm))
-    tm_start(tm, val);
+    tm_start(tm, val S);
 }
 
 static inline void
@@ -801,8 +803,8 @@ ospf_iface_reconfigure(struct ospf_iface *ifa, struct ospf_iface_patt *new)
 	       ifname, ifa->waitint, new->waitint);
 
     ifa->waitint = new->waitint;
-    if (ifa->wait_timer && ifa->wait_timer->expires)
+    if (ifa->wait_timer && tm_active(ifa->wait_timer))
-      tm_start(ifa->wait_timer, ifa->waitint);
+      tm_start(ifa->wait_timer, ifa->waitint S);
   }
 
   /* DEAD TIMER */
@@ -1113,9 +1115,6 @@ ospf_ifa_notify3(struct proto *P, uint flags, struct ifa *a)
 {
   struct ospf_proto *p = (struct ospf_proto *) P;
 
-  if (a->prefix.type != NET_IP6)
-    return;
-
   if (a->flags & IA_SECONDARY)
     return;
 
@@ -1126,6 +1125,9 @@ ospf_ifa_notify3(struct proto *P, uint flags, struct ifa *a)
      other addresses are used for link-LSA. */
   if (a->scope == SCOPE_LINK)
   {
+    if (a->prefix.type != NET_IP6)
+      return;
+
     if (flags & IF_CHANGE_UP)
     {
       struct ospf_mip_walk s = { .iface = a->iface };
@@ -1143,6 +1145,9 @@ ospf_ifa_notify3(struct proto *P, uint flags, struct ifa *a)
   }
   else
   {
+    if (a->prefix.type != ospf_get_af(p))
+      return;
+
     struct ospf_iface *ifa;
     WALK_LIST(ifa, p->iface_list)
       if (ifa->iface == a->iface)

proto/ospf/lsalib.c

@@ -280,7 +280,7 @@ lsa_walk_rt(struct ospf_lsa_rt_walk *rt)
 
 
 void
-lsa_parse_sum_net(struct top_hash_entry *en, int ospf2, net_addr *net, u8 *pxopts, u32 *metric)
+lsa_parse_sum_net(struct top_hash_entry *en, int ospf2, int af, net_addr *net, u8 *pxopts, u32 *metric)
 {
   if (ospf2)
   {
@@ -292,7 +292,7 @@ lsa_parse_sum_net(struct top_hash_entry *en, int ospf2, net_addr *net, u8 *pxopt
   else
   {
     struct ospf_lsa_sum3_net *ls = en->lsa_body;
-    ospf_get_ipv6_prefix(ls->prefix, net, pxopts, NULL);
+    ospf3_get_prefix(ls->prefix, af, net, pxopts, NULL);
     *metric = ls->metric & LSA_METRIC_MASK;
   }
 }
@@ -317,7 +317,7 @@ lsa_parse_sum_rt(struct top_hash_entry *en, int ospf2, u32 *drid, u32 *metric, u
 }
 
 void
-lsa_parse_ext(struct top_hash_entry *en, int ospf2, struct ospf_lsa_ext_local *rt)
+lsa_parse_ext(struct top_hash_entry *en, int ospf2, int af, struct ospf_lsa_ext_local *rt)
 {
   if (ospf2)
   {
@@ -338,13 +338,13 @@ lsa_parse_ext(struct top_hash_entry *en, int ospf2, struct ospf_lsa_ext_local *r
   else
   {
     struct ospf_lsa_ext3 *ext = en->lsa_body;
-    u32 *buf = ospf_get_ipv6_prefix(ext->rest, &rt->net, &rt->pxopts, NULL);
+    u32 *buf = ospf3_get_prefix(ext->rest, af, &rt->net, &rt->pxopts, NULL);
     rt->metric = ext->metric & LSA_METRIC_MASK;
     rt->ebit = ext->metric & LSA_EXT3_EBIT;
 
     rt->fbit = ext->metric & LSA_EXT3_FBIT;
     if (rt->fbit)
-      buf = ospf_get_ipv6_addr(buf, &rt->fwaddr);
+      buf = ospf3_get_addr(buf, af, &rt->fwaddr);
     else
       rt->fwaddr = IPA_NONE;
 

proto/ospf/lsalib.h

@@ -55,9 +55,12 @@ u16 lsa_verify_checksum(const void *lsa_n, int lsa_len);
 int lsa_comp(struct ospf_lsa_header *l1, struct ospf_lsa_header *l2);
 void lsa_walk_rt_init(struct ospf_proto *po, struct top_hash_entry *act, struct ospf_lsa_rt_walk *rt);
 int lsa_walk_rt(struct ospf_lsa_rt_walk *rt);
-void lsa_parse_sum_net(struct top_hash_entry *en, int ospf2, net_addr *net, u8 *pxopts, u32 *metric);
+void lsa_parse_sum_net(struct top_hash_entry *en, int ospf2, int af, net_addr *net, u8 *pxopts, u32 *metric);
 void lsa_parse_sum_rt(struct top_hash_entry *en, int ospf2, u32 *drid, u32 *metric, u32 *options);
-void lsa_parse_ext(struct top_hash_entry *en, int ospf2, struct ospf_lsa_ext_local *rt);
+void lsa_parse_ext(struct top_hash_entry *en, int ospf2, int af, struct ospf_lsa_ext_local *rt);
 int lsa_validate(struct ospf_lsa_header *lsa, u32 lsa_type, int ospf2, void *body);
 
+static inline btime lsa_inst_age(struct top_hash_entry *en)
+{ return current_time() - en->inst_time; }
+
 #endif /* _BIRD_OSPF_LSALIB_H_ */

proto/ospf/lsupd.c

@@ -137,7 +137,7 @@ ospf_lsa_lsrt_up(struct top_hash_entry *en, struct ospf_neighbor *n)
   ret->lsa_body = LSA_BODY_DUMMY;
 
   if (!tm_active(n->lsrt_timer))
-    tm_start(n->lsrt_timer, n->ifa->rxmtint);
+    tm_start(n->lsrt_timer, n->ifa->rxmtint S);
 }
 
 void
@@ -572,7 +572,7 @@ ospf_receive_lsupd(struct ospf_packet *pkt, struct ospf_iface *ifa,
     {
       /* 13. (5a) - enforce minimum time between updates for received LSAs */
       /* We also use this to ratelimit reactions to received self-originated LSAs */
-      if (en && ((now - en->inst_time) < MINLSARRIVAL))
+      if (en && (lsa_inst_age(en) < MINLSARRIVAL))
       {
 	OSPF_TRACE(D_EVENTS, "Skipping LSA received in less that MinLSArrival");
 	continue;
@@ -700,7 +700,7 @@ ospf_receive_lsupd(struct ospf_packet *pkt, struct ospf_iface *ifa,
   if (!EMPTY_SLIST(n->lsrql) && (n->lsrqi == SHEAD(n->lsrql)))
   {
     ospf_send_lsreq(p, n);
-    tm_start(n->lsrq_timer, n->ifa->rxmtint);
+    tm_start(n->lsrq_timer, n->ifa->rxmtint S);
   }
 
   return;

proto/ospf/neighbor.c

@@ -94,11 +94,11 @@ ospf_neighbor_new(struct ospf_iface *ifa)
   init_list(&n->ackl[ACKL_DIRECT]);
   init_list(&n->ackl[ACKL_DELAY]);
 
-  n->inactim = tm_new_set(pool, inactivity_timer_hook, n, 0, 0);
+  n->inactim = tm_new_init(pool, inactivity_timer_hook, n, 0, 0);
-  n->dbdes_timer = tm_new_set(pool, dbdes_timer_hook, n, 0, ifa->rxmtint);
+  n->dbdes_timer = tm_new_init(pool, dbdes_timer_hook, n, ifa->rxmtint S, 0);
-  n->lsrq_timer = tm_new_set(pool, lsrq_timer_hook, n, 0, ifa->rxmtint);
+  n->lsrq_timer = tm_new_init(pool, lsrq_timer_hook, n, ifa->rxmtint S, 0);
-  n->lsrt_timer = tm_new_set(pool, lsrt_timer_hook, n, 0, ifa->rxmtint);
+  n->lsrt_timer = tm_new_init(pool, lsrt_timer_hook, n, ifa->rxmtint S, 0);
-  n->ackd_timer = tm_new_set(pool, ackd_timer_hook, n, 0, ifa->rxmtint / 2);
+  n->ackd_timer = tm_new_init(pool, ackd_timer_hook, n, ifa->rxmtint S / 2, 0);
 
   return (n);
 }
@@ -186,7 +186,7 @@ ospf_neigh_chstate(struct ospf_neighbor *n, u8 state)
     n->myimms = DBDES_IMMS;
 
     tm_start(n->dbdes_timer, 0);
-    tm_start(n->ackd_timer, ifa->rxmtint / 2);
+    tm_start(n->ackd_timer, ifa->rxmtint S / 2);
   }
 
   if (state > NEIGHBOR_EXSTART)
@@ -231,7 +231,7 @@ ospf_neigh_sm(struct ospf_neighbor *n, int event)
       ospf_neigh_chstate(n, NEIGHBOR_INIT);
 
     /* Restart inactivity timer */
-    tm_start(n->inactim, n->ifa->deadint);
+    tm_start(n->inactim, n->ifa->deadint S);
     break;
 
   case INM_2WAYREC:
@@ -651,20 +651,6 @@ ospf_sh_neigh_info(struct ospf_neighbor *n)
 {
   struct ospf_iface *ifa = n->ifa;
   char *pos = "PtP  ";
-  char etime[6];
-  int exp, sec, min;
-
-  exp = n->inactim->expires - now;
-  sec = exp % 60;
-  min = exp / 60;
-  if (min > 59)
-  {
-    bsprintf(etime, "-Inf-");
-  }
-  else
-  {
-    bsprintf(etime, "%02u:%02u", min, sec);
-  }
 
   if ((ifa->type == OSPF_IT_BCAST) || (ifa->type == OSPF_IT_NBMA))
   {
@@ -676,6 +662,7 @@ ospf_sh_neigh_info(struct ospf_neighbor *n)
       pos = "Other";
   }
 
-  cli_msg(-1013, "%-1R\t%3u\t%s/%s\t%-5s\t%-10s %-1I", n->rid, n->priority,
+  cli_msg(-1013, "%-1R\t%3u\t%s/%s\t%7t\t%-10s %-1I",
-	  ospf_ns_names[n->state], pos, etime, ifa->ifname, n->ip);
+	  n->rid, n->priority, ospf_ns_names[n->state], pos,
+	  tm_remains(n->inactim), ifa->ifname, n->ip);
 }

proto/ospf/ospf.c

@@ -92,8 +92,10 @@
  * - RFC 2328 - main OSPFv2 standard
  * - RFC 5340 - main OSPFv3 standard
  * - RFC 3101 - OSPFv2 NSSA areas
- * - RFC 6549 - OSPFv2 multi-instance extensions
+ * - RFC 5709 - OSPFv2 HMAC-SHA Cryptographic Authentication
- * - RFC 6987 - OSPF stub router advertisement
+ * - RFC 5838 - OSPFv3 Support of Address Families
+ * - RFC 6549 - OSPFv2 Multi-Instance Extensions
+ * - RFC 6987 - OSPF Stub Router Advertisement
  */
 
 #include <stdlib.h>
@@ -115,9 +117,9 @@ add_area_nets(struct ospf_area *oa, struct ospf_area_config *ac)
   struct area_net_config *anc;
   struct area_net *an;
 
-  fib_init(&oa->net_fib,  p->p.pool, ospf_is_v2(p) ? NET_IP4 : NET_IP6,
+  fib_init(&oa->net_fib,  p->p.pool, ospf_get_af(p),
 	   sizeof(struct area_net), OFFSETOF(struct area_net, fn), 0, NULL);
-  fib_init(&oa->enet_fib, p->p.pool, ospf_is_v2(p) ? NET_IP4 : NET_IP6,
+  fib_init(&oa->enet_fib, p->p.pool, ospf_get_af(p),
 	   sizeof(struct area_net), OFFSETOF(struct area_net, fn), 0, NULL);
 
   WALK_LIST(anc, ac->net_list)
@@ -134,6 +136,16 @@ add_area_nets(struct ospf_area *oa, struct ospf_area_config *ac)
   }
 }
 
+static inline uint
+ospf_opts(struct ospf_proto *p)
+{
+  if (ospf_is_v2(p))
+    return 0;
+
+  return ((ospf_is_ip6(p) && !p->af_mc) ? OPT_V6 : 0) |
+    (!p->stub_router ? OPT_R : 0) | (p->af_ext ? OPT_AF : 0);
+}
+
 static void
 ospf_area_add(struct ospf_proto *p, struct ospf_area_config *ac)
 {
@@ -155,10 +167,7 @@ ospf_area_add(struct ospf_proto *p, struct ospf_area_config *ac)
   if (oa->areaid == 0)
     p->backbone = oa;
 
-  if (ospf_is_v2(p))
+  oa->options = ac->type | ospf_opts(p);
-    oa->options = ac->type;
-  else
-    oa->options = ac->type | OPT_V6 | (p->stub_router ? 0 : OPT_R);
 
   ospf_notify_rt_lsa(oa);
 }
@@ -224,22 +233,23 @@ ospf_start(struct proto *P)
 
   p->router_id = proto_get_router_id(P->cf);
   p->ospf2 = c->ospf2;
+  p->af_ext = c->af_ext;
+  p->af_mc = c->af_mc;
   p->rfc1583 = c->rfc1583;
   p->stub_router = c->stub_router;
   p->merge_external = c->merge_external;
   p->asbr = c->asbr;
   p->ecmp = c->ecmp;
   p->tick = c->tick;
-  p->disp_timer = tm_new_set(P->pool, ospf_disp, p, 0, p->tick);
+  p->disp_timer = tm_new_init(P->pool, ospf_disp, p, p->tick S, 0);
-  tm_start(p->disp_timer, 1);
+  tm_start(p->disp_timer, 100 MS);
   p->lsab_size = 256;
   p->lsab_used = 0;
   p->lsab = mb_alloc(P->pool, p->lsab_size);
   p->nhpool = lp_new(P->pool, 12*sizeof(struct nexthop));
   init_list(&(p->iface_list));
   init_list(&(p->area_list));
-  fib_init(&p->rtf, P->pool, p->ospf2 ? NET_IP4 : NET_IP6,
+  fib_init(&p->rtf, P->pool, ospf_get_af(p), sizeof(ort), OFFSETOF(ort, fn), 0, NULL);
-	   sizeof(ort), OFFSETOF(ort, fn), 0, NULL);
   if (ospf_is_v3(p))
     idm_init(&p->idm, P->pool, 16);
   p->areano = 0;
@@ -601,11 +611,7 @@ ospf_area_reconfigure(struct ospf_area *oa, struct ospf_area_config *nac)
   struct ospf_iface *ifa;
 
   oa->ac = nac;
-
+  oa->options = nac->type | ospf_opts(p);
-  if (ospf_is_v2(p))
-    oa->options = nac->type;
-  else
-    oa->options = nac->type | OPT_V6 | (p->stub_router ? 0 : OPT_R);
 
   if (nac->type != oac->type)
   {
@@ -659,6 +665,9 @@ ospf_reconfigure(struct proto *P, struct proto_config *CF)
   if (old->abr != new->abr)
     return 0;
 
+  if ((p->af_ext != new->af_ext) || (p->af_mc != new->af_mc))
+    return 0;
+
   if (!proto_configure_channel(P, &P->main_channel, proto_cf_main_channel(CF)))
     return 0;
 
@@ -667,8 +676,8 @@ ospf_reconfigure(struct proto *P, struct proto_config *CF)
   p->asbr = new->asbr;
   p->ecmp = new->ecmp;
   p->tick = new->tick;
-  p->disp_timer->recurrent = p->tick;
+  p->disp_timer->recurrent = p->tick S;
-  tm_start(p->disp_timer, 1);
+  tm_start(p->disp_timer, 100 MS);
 
   /* Mark all areas and ifaces */
   WALK_LIST(oa, p->area_list)
@@ -1073,13 +1082,13 @@ show_lsa_network(struct top_hash_entry *he, int ospf2)
 }
 
 static inline void
-show_lsa_sum_net(struct top_hash_entry *he, int ospf2)
+show_lsa_sum_net(struct top_hash_entry *he, int ospf2, int af)
 {
   net_addr net;
   u8 pxopts;
   u32 metric;
 
-  lsa_parse_sum_net(he, ospf2, &net, &pxopts, &metric);
+  lsa_parse_sum_net(he, ospf2, af, &net, &pxopts, &metric);
   cli_msg(-1016, "\t\txnetwork %N metric %u", &net, metric);
 }
 
@@ -1096,7 +1105,7 @@ show_lsa_sum_rt(struct top_hash_entry *he, int ospf2)
 
 
 static inline void
-show_lsa_external(struct top_hash_entry *he, int ospf2)
+show_lsa_external(struct top_hash_entry *he, int ospf2, int af)
 {
   struct ospf_lsa_ext_local rt;
   char str_via[IPA_MAX_TEXT_LENGTH + 8] = "";
@@ -1105,7 +1114,7 @@ show_lsa_external(struct top_hash_entry *he, int ospf2)
   if (he->lsa_type == LSA_T_EXT)
     he->domain = 0; /* Unmark the LSA */
 
-  lsa_parse_ext(he, ospf2, &rt);
+  lsa_parse_ext(he, ospf2, af, &rt);
 
   if (rt.fbit)
     bsprintf(str_via, " via %I", rt.fwaddr);
@@ -1119,7 +1128,7 @@ show_lsa_external(struct top_hash_entry *he, int ospf2)
 }
 
 static inline void
-show_lsa_prefix(struct top_hash_entry *he, struct top_hash_entry *cnode)
+show_lsa_prefix(struct top_hash_entry *he, struct top_hash_entry *cnode, int af)
 {
   struct ospf_lsa_prefix *px = he->lsa_body;
   u32 *buf;
@@ -1142,7 +1151,7 @@ show_lsa_prefix(struct top_hash_entry *he, struct top_hash_entry *cnode)
     u8 pxopts;
     u16 metric;
 
-    buf = ospf_get_ipv6_prefix(buf, &net, &pxopts, &metric);
+    buf = ospf3_get_prefix(buf, af, &net, &pxopts, &metric);
 
     if (px->ref_type == LSA_T_RT)
       cli_msg(-1016, "\t\tstubnet %N metric %u", &net, metric);
@@ -1156,6 +1165,7 @@ ospf_sh_state(struct proto *P, int verbose, int reachable)
 {
   struct ospf_proto *p = (struct ospf_proto *) P;
   int ospf2 = ospf_is_v2(p);
+  int af = ospf_get_af(p);
   uint i, ix, j1, jx;
   u32 last_area = 0xFFFFFFFF;
 
@@ -1276,7 +1286,7 @@ ospf_sh_state(struct proto *P, int verbose, int reachable)
 
     case LSA_T_SUM_NET:
       if (cnode->lsa_type == LSA_T_RT)
-	show_lsa_sum_net(he, ospf2);
+	show_lsa_sum_net(he, ospf2, af);
       break;
 
     case LSA_T_SUM_RT:
@@ -1286,11 +1296,11 @@ ospf_sh_state(struct proto *P, int verbose, int reachable)
 
     case LSA_T_EXT:
     case LSA_T_NSSA:
-      show_lsa_external(he, ospf2);
+      show_lsa_external(he, ospf2, af);
       break;
 
     case LSA_T_PREFIX:
-      show_lsa_prefix(he, cnode);
+      show_lsa_prefix(he, cnode, af);
       break;
     }
 
@@ -1304,7 +1314,7 @@ ospf_sh_state(struct proto *P, int verbose, int reachable)
 	ix++;
 
       while ((ix < jx) && (hex[ix]->lsa.rt == cnode->lsa.rt))
-	show_lsa_external(hex[ix++], ospf2);
+	show_lsa_external(hex[ix++], ospf2, af);
 
       cnode = NULL;
     }
@@ -1338,7 +1348,7 @@ ospf_sh_state(struct proto *P, int verbose, int reachable)
 	last_rt = he->lsa.rt;
       }
 
-      show_lsa_external(he, ospf2);
+      show_lsa_external(he, ospf2, af);
     }
   }
 

proto/ospf/ospf.h

@@ -18,7 +18,7 @@
 #include "lib/lists.h"
 #include "lib/slists.h"
 #include "lib/socket.h"
-#include "sysdep/unix/timer.h"
+#include "lib/timer.h"
 #include "lib/resource.h"
 #include "nest/protocol.h"
 #include "nest/iface.h"
@@ -58,16 +58,16 @@
   log_rl(&p->log_lsa_tbf, L_REMOTE "%s: " msg, p->p.name, args)
 
 #define LOG_LSA2(msg, args...) \
-  do { if (! p->log_lsa_tbf.mark) \
+  do { if (! p->log_lsa_tbf.drop) \
     log(L_REMOTE "%s: " msg, p->p.name, args); } while(0)
 
 
 #define OSPF_PROTO 89
 
-#define LSREFRESHTIME 1800	/* 30 minutes */
+#define LSREFRESHTIME		1800	/* 30 minutes */
-#define MINLSINTERVAL 5
+#define MINLSINTERVAL		(5 S_)
-#define MINLSARRIVAL 1
+#define MINLSARRIVAL		(1 S_)
-#define LSINFINITY 0xffffff
+#define LSINFINITY		0xffffff
 
 #define OSPF_DEFAULT_TICK 1
 #define OSPF_DEFAULT_STUB_COST 1000
@@ -84,10 +84,13 @@ struct ospf_config
   struct proto_config c;
   uint tick;
   u8 ospf2;
+  u8 af_ext;
+  u8 af_mc;
   u8 rfc1583;
   u8 stub_router;
   u8 merge_external;
   u8 instance_id;
+  u8 instance_id_set;
   u8 abr;
   u8 asbr;
   int ecmp;
@@ -168,9 +171,9 @@ struct ospf_iface_patt
   int tx_priority;
   u16 tx_length;
   u16 rx_buffer;
-
 #define OSPF_RXBUF_MINSIZE 256	/* Minimal allowed size */
   u8 instance_id;
+  u8 instance_id_set;
   u8 autype;			/* OSPF_AUTH_*, not really used in OSPFv3 */
   u8 strictnbma;
   u8 check_link;
@@ -211,12 +214,14 @@ struct ospf_proto
   int padj;			/* Number of neighbors in Exchange or Loading state */
   struct fib rtf;		/* Routing table */
   struct idm idm;		/* OSPFv3 LSA ID map */
-  byte ospf2;			/* OSPF v2 or v3 */
+  u8 ospf2;			/* OSPF v2 or v3 */
-  byte rfc1583;			/* RFC1583 compatibility */
+  u8 af_ext;			/* OSPFv3-AF extension */
-  byte stub_router;		/* Do not forward transit traffic */
+  u8 af_mc;			/* OSPFv3-AF multicast */
-  byte merge_external;		/* Should i merge external routes? */
+  u8 rfc1583;			/* RFC1583 compatibility */
-  byte asbr;			/* May i originate any ext/NSSA lsa? */
+  u8 stub_router;		/* Do not forward transit traffic */
-  byte ecmp;			/* Maximal number of nexthops in ECMP route, or 0 */
+  u8 merge_external;		/* Should i merge external routes? */
+  u8 asbr;			/* May i originate any ext/NSSA lsa? */
+  u8 ecmp;			/* Maximal number of nexthops in ECMP route, or 0 */
   struct ospf_area *backbone;	/* If exists */
   event *flood_event;		/* Event for flooding LS updates */
   void *lsab;			/* LSA buffer used when originating router LSAs */
@@ -264,10 +269,10 @@ struct ospf_iface
   sock *sk;			/* IP socket */
   list neigh_list;		/* List of neighbors (struct ospf_neighbor) */
   u32 cost;			/* Cost of iface */
-  u32 waitint;			/* number of sec before changing state from wait */
+  u32 waitint;			/* Number of seconds before changing state from wait */
-  u32 rxmtint;			/* number of seconds between LSA retransmissions */
+  u32 rxmtint;			/* Number of seconds between LSA retransmissions */
-  u32 pollint;			/* Poll interval */
+  u32 pollint;			/* Poll interval in seconds */
-  u32 deadint;			/* after "deadint" missing hellos is router dead */
+  u32 deadint;			/* After deadint seconds without hellos is router dead */
   u32 iface_id;			/* Interface ID (iface->index or new value for vlinks) */
   u32 vid;			/* ID of peer of virtual link */
   ip_addr vip;			/* IP of peer of virtual link */
@@ -278,8 +283,8 @@ struct ospf_iface
 				   interface.  LSAs contained in the update */
   u16 helloint;			/* number of seconds between hello sending */
   list *passwords;
-  u32 csn;                      /* Last used crypt seq number */
+  u32 csn;			/* Last used crypt seq number */
-  bird_clock_t csn_use;         /* Last time when packet with that CSN was sent */
+  btime csn_use;		/* Last time when packet with that CSN was sent */
   ip_addr all_routers;		/* Multicast (or broadcast) address for all routers */
   ip_addr des_routers;		/* Multicast (or NULL) address for designated routers */
   ip_addr drip;			/* Designated router IP */
@@ -449,14 +454,15 @@ struct ospf_neighbor
 
 
 /* Generic option flags */
-#define OPT_V6		0x01	/* OSPFv3, LSA relevant for IPv6 routing calculation */
+#define OPT_V6		0x0001	/* OSPFv3, LSA relevant for IPv6 routing calculation */
-#define OPT_E		0x02	/* Related to AS-external LSAs */
+#define OPT_E		0x0002	/* Related to AS-external LSAs */
-#define OPT_MC		0x04	/* Related to MOSPF, not used and obsolete */
+#define OPT_MC		0x0004	/* Related to MOSPF, not used and obsolete */
-#define OPT_N		0x08	/* Related to NSSA */
+#define OPT_N		0x0008	/* Related to NSSA */
-#define OPT_P		0x08	/* OSPFv2, flags P and N share position, see NSSA RFC */
+#define OPT_P		0x0008	/* OSPFv2, flags P and N share position, see NSSA RFC */
-#define OPT_EA		0x10	/* OSPFv2, external attributes, not used and obsolete */
+#define OPT_EA		0x0010	/* OSPFv2, external attributes, not used and obsolete */
-#define OPT_R		0x10	/* OSPFv3, originator is active router */
+#define OPT_R		0x0010	/* OSPFv3, originator is active router */
-#define OPT_DC		0x20	/* Related to demand circuits, not used */
+#define OPT_DC		0x0020	/* Related to demand circuits, not used */
+#define OPT_AF		0x0100	/* OSPFv3 Address Families (RFC 5838) */
 
 /* Router-LSA VEB flags are are stored together with links (OSPFv2) or options (OSPFv3) */
 #define OPT_RT_B	(0x01 << 24)
@@ -718,74 +724,96 @@ lsa_net_count(struct ospf_lsa_header *lsa)
 #define IPV6_PREFIX_SPACE(x) ((((x) + 63) / 32) * 4)
 #define IPV6_PREFIX_WORDS(x) (((x) + 63) / 32)
 
-/* FIXME: these functions should be significantly redesigned w.r.t. integration,
-   also should be named as ospf3_* instead of *_ipv6_* */
 
 static inline int
 ospf_valid_prefix(net_addr *n)
 {
-  /* In OSPFv2, prefix is stored as netmask; ip4_masklen() returns 255 for invalid one */
+  /*
-  return n->pxlen <= IP6_MAX_PREFIX_LENGTH;
+   * In OSPFv2, prefix is stored as netmask; ip4_masklen() returns 255 for
+   * invalid one. But OSPFv3-AF may receive IPv4 net with 32 < pxlen < 128.
+   */
+  uint max = (n->type == NET_IP4) ? IP4_MAX_PREFIX_LENGTH : IP6_MAX_PREFIX_LENGTH;
+  return n->pxlen <= max;
 }
 
+/*
+ * In OSPFv3-AF (RFC 5835), IPv4 address is encoded by just placing it in the
+ * first 32 bits of IPv6 address and setting remaining bits to zero. Likewise
+ * for IPv4 prefix, where remaining bits do not matter. We use following
+ * functions to convert between IPv4 and IPv4-in-IPv6 representations:
+ */
+
+static inline ip4_addr ospf3_6to4(ip6_addr a)
+{ return _MI4(_I0(a)); }
+
+static inline ip6_addr ospf3_4to6(ip4_addr a)
+{ return _MI6(_I(a), 0, 0, 0); }
+
+
 static inline u32 *
-ospf_get_ipv6_prefix(u32 *buf, net_addr *N, u8 *pxopts, u16 *rest)
+ospf3_get_prefix(u32 *buf, int af, net_addr *n, u8 *pxopts, u16 *rest)
 {
-  net_addr_ip6 *net = (void *) N;
+  ip6_addr px = IP6_NONE;
-  u8 pxlen = (*buf >> 24);
+  uint pxlen = (*buf >> 24);
   *pxopts = (*buf >> 16) & 0xff;
   if (rest) *rest = *buf & 0xffff;
   buf++;
 
-  *net = NET_ADDR_IP6(IP6_NONE, pxlen);
-
   if (pxlen > 0)
-    _I0(net->prefix) = *buf++;
+    _I0(px) = *buf++;
   if (pxlen > 32)
-    _I1(net->prefix) = *buf++;
+    _I1(px) = *buf++;
   if (pxlen > 64)
-    _I2(net->prefix) = *buf++;
+    _I2(px) = *buf++;
   if (pxlen > 96)
-    _I3(net->prefix) = *buf++;
+    _I3(px) = *buf++;
 
   /* Clean up remaining bits */
   if (pxlen < 128)
-    net->prefix.addr[pxlen / 32] &= u32_mkmask(pxlen % 32);
+    px.addr[pxlen / 32] &= u32_mkmask(pxlen % 32);
+
+  if (af == NET_IP4)
+    net_fill_ip4(n, ospf3_6to4(px), pxlen);
+  else
+    net_fill_ip6(n, px, pxlen);
 
   return buf;
 }
 
 static inline u32 *
-ospf_get_ipv6_addr(u32 *buf, ip_addr *addr)
+ospf3_put_prefix(u32 *buf, net_addr *n, u8 pxopts, u16 rest)
 {
-  *addr = ipa_from_ip6(*(ip6_addr *) buf);
+  ip6_addr px = (n->type == NET_IP4) ? ospf3_4to6(net4_prefix(n)) : net6_prefix(n);
-  return buf + 4;
+  uint pxlen = n->pxlen;
-}
-
-static inline u32 *
-ospf_put_ipv6_prefix(u32 *buf, net_addr *N, u8 pxopts, u16 rest)
-{
-  net_addr_ip6 *net = (void *) N;
-  u32 pxlen = net->pxlen;
 
   *buf++ = ((pxlen << 24) | (pxopts << 16) | rest);
 
   if (pxlen > 0)
-    *buf++ = _I0(net->prefix);
+    *buf++ = _I0(px);
   if (pxlen > 32)
-    *buf++ = _I1(net->prefix);
+    *buf++ = _I1(px);
   if (pxlen > 64)
-    *buf++ = _I2(net->prefix);
+    *buf++ = _I2(px);
   if (pxlen > 96)
-    *buf++ = _I3(net->prefix);
+    *buf++ = _I3(px);
 
   return buf;
 }
 
 static inline u32 *
-ospf_put_ipv6_addr(u32 *buf, ip_addr addr)
+ospf3_get_addr(u32 *buf, int af, ip_addr *addr)
 {
-  *(ip6_addr *) buf = ipa_to_ip6(addr);
+  ip6_addr a;
+  memcpy(&a, buf, 16);
+  *addr = (af == NET_IP4) ? ipa_from_ip4(ospf3_6to4(a)) : ipa_from_ip6(a);
+  return buf + 4;
+}
+
+static inline u32 *
+ospf3_put_addr(u32 *buf, ip_addr addr)
+{
+  ip6_addr a = ipa_is_ip4(addr) ? ospf3_4to6(ipa_to_ip4(addr)) : ipa_to_ip6(addr);
+  memcpy(buf, &a, 16);
   return buf + 4;
 }
 
@@ -838,6 +866,15 @@ static inline int ospf_is_v3(struct ospf_proto *p)
 static inline int ospf_get_version(struct ospf_proto *p)
 { return ospf_is_v2(p) ? 2 : 3; }
 
+static inline int ospf_is_ip4(struct ospf_proto *p)
+{ return p->p.net_type == NET_IP4; }
+
+static inline int ospf_is_ip6(struct ospf_proto *p)
+{ return p->p.net_type == NET_IP6; }
+
+static inline int ospf_get_af(struct ospf_proto *p)
+{ return p->p.net_type; }
+
 struct ospf_area *ospf_find_area(struct ospf_proto *p, u32 aid);
 
 static inline struct ospf_area *ospf_main_area(struct ospf_proto *p)

proto/ospf/packet.c

@@ -77,16 +77,16 @@ ospf_pkt_finalize(struct ospf_iface *ifa, struct ospf_packet *pkt, uint *plen)
        reboot when system does not have independent RTC? */
     if (!ifa->csn)
     {
-      ifa->csn = (u32) now;
+      ifa->csn = (u32) (current_real_time() TO_S);
-      ifa->csn_use = now;
+      ifa->csn_use = current_time();
     }
 
     /* We must have sufficient delay between sending a packet and increasing
        CSN to prevent reordering of packets (in a network) with different CSNs */
-    if ((now - ifa->csn_use) > 1)
+    if ((current_time() - ifa->csn_use) > 1 S)
       ifa->csn++;
 
-    ifa->csn_use = now;
+    ifa->csn_use = current_time();
 
     uint auth_len = mac_type_length(pass->alg);
     byte *auth_tail = ((byte *) pkt + *plen);
@@ -270,9 +270,6 @@ ospf_rx_hook(sock *sk, uint len)
   if (pkt == NULL)
     DROP("bad IP header", len);
 
-  if (ifa->check_ttl && (sk->rcv_ttl < 255))
-    DROP("wrong TTL", sk->rcv_ttl);
-
   if (len < sizeof(struct ospf_packet))
     DROP("too short", len);
 
@@ -379,6 +376,10 @@ found:
   if (ipa_equal(sk->laddr, ifa->des_routers) && (ifa->sk_dr == 0))
     return 1;
 
+  /* TTL check must be done after instance dispatch */
+  if (ifa->check_ttl && (sk->rcv_ttl < 255))
+    DROP("wrong TTL", sk->rcv_ttl);
+
   if (rid == p->router_id)
     DROP1("my own router ID");
 

proto/ospf/rt.c

@@ -10,9 +10,7 @@
 
 #include "ospf.h"
 
-static void add_cand(list * l, struct top_hash_entry *en,
+static void add_cand(struct ospf_area *oa, struct top_hash_entry *en, struct top_hash_entry *par, u32 dist, int i, uint lif, uint nif);
-		     struct top_hash_entry *par, u32 dist,
-		     struct ospf_area *oa, int i);
 static void rt_sync(struct ospf_proto *p);
 
 
@@ -509,7 +507,7 @@ spfa_process_rt(struct ospf_proto *p, struct ospf_area *oa, struct top_hash_entr
       break;
     }
 
-    add_cand(&oa->cand, tmp, act, act->dist + rtl.metric, oa, i);
+    add_cand(oa, tmp, act, act->dist + rtl.metric, i, rtl.lif, rtl.nif);
   }
 }
 
@@ -532,7 +530,7 @@ spfa_process_net(struct ospf_proto *p, struct ospf_area *oa, struct top_hash_ent
   for (i = 0; i < cnt; i++)
   {
     tmp = ospf_hash_find_rt(p->gr, oa->areaid, ln->routers[i]);
-    add_cand(&oa->cand, tmp, act, act->dist, oa, -1);
+    add_cand(oa, tmp, act, act->dist, -1, 0, 0);
   }
 }
 
@@ -576,20 +574,20 @@ spfa_process_prefixes(struct ospf_proto *p, struct ospf_area *oa)
     buf = px->rest;
     for (i = 0; i < px->pxcount; i++)
     {
-      net_addr_ip6 net;
+      net_addr net;
       u8 pxopts;
       u16 metric;
 
-      buf = ospf_get_ipv6_prefix(buf, (net_addr *) &net, &pxopts, &metric);
+      buf = ospf3_get_prefix(buf, ospf_get_af(p), &net, &pxopts, &metric);
 
       if (pxopts & OPT_PX_NU)
 	continue;
 
       /* Store the first global address to use it later as a vlink endpoint */
-      if ((pxopts & OPT_PX_LA) && ipa_zero(src->lb))
+      if ((pxopts & OPT_PX_LA) && (net.type == NET_IP6) && ipa_zero(src->lb))
-	src->lb = ipa_from_ip6(net.prefix);
+	src->lb = ipa_from_ip6(net6_prefix(&net));
 
-      add_network(oa, (net_addr *) &net, src->dist + metric, src, i);
+      add_network(oa, &net, src->dist + metric, src, i);
     }
   }
 }
@@ -651,7 +649,8 @@ ospf_rt_spfa(struct ospf_area *oa)
 }
 
 static int
-link_back(struct ospf_area *oa, struct top_hash_entry *en, struct top_hash_entry *par)
+link_back(struct ospf_area *oa, struct top_hash_entry *en,
+	  struct top_hash_entry *par, uint lif, uint nif)
 {
   struct ospf_proto *p = oa->po;
   struct ospf_lsa_rt_walk rtl;
@@ -689,6 +688,10 @@ link_back(struct ospf_area *oa, struct top_hash_entry *en, struct top_hash_entry
 	tmp = ospf_hash_find_net(p->gr, oa->areaid, rtl.id, rtl.nif);
 	if (tmp == par)
 	{
+	  /*
+	   * Note that there may be multiple matching Rt-fields if router 'en'
+	   * have multiple interfaces to net 'par'. Perhaps we should do ECMP.
+	   */
 	  if (ospf_is_v2(p))
 	    en->lb = ipa_from_u32(rtl.data);
 	  else
@@ -700,7 +703,13 @@ link_back(struct ospf_area *oa, struct top_hash_entry *en, struct top_hash_entry
 
       case LSART_VLNK:
       case LSART_PTP:
-	/* Not necessary the same link, see RFC 2328 [23] */
+	/*
+	 * For OSPFv2, not necessary the same link, see RFC 2328 [23].
+	 * For OSPFv3, we verify that by comparing nif and lif fields.
+	 */
+	if (ospf_is_v3(p) && ((rtl.lif != nif) || (rtl.nif != lif)))
+	  break;
+
 	tmp = ospf_hash_find_rt(p->gr, oa->areaid, rtl.id);
 	if (tmp == par)
 	  return 1;
@@ -761,7 +770,7 @@ ospf_rt_sum(struct ospf_area *oa)
 
     if (en->lsa_type == LSA_T_SUM_NET)
     {
-      lsa_parse_sum_net(en, ospf_is_v2(p), &net, &pxopts, &metric);
+      lsa_parse_sum_net(en, ospf_is_v2(p), ospf_get_af(p), &net, &pxopts, &metric);
 
       if (!ospf_valid_prefix(&net))
       {
@@ -858,7 +867,7 @@ ospf_rt_sum_tr(struct ospf_area *oa)
       net_addr net;
       u8 pxopts;
 
-      lsa_parse_sum_net(en, ospf_is_v2(p), &net, &pxopts, &metric);
+      lsa_parse_sum_net(en, ospf_is_v2(p), ospf_get_af(p), &net, &pxopts, &metric);
 
       if (!ospf_valid_prefix(&net))
       {
@@ -1058,7 +1067,7 @@ decide_nssa_lsa(struct ospf_proto *p, ort *nf, struct ospf_lsa_ext_local *rt)
     return 0;
 
   /* We do not store needed data in struct orta, we have to parse the LSA */
-  lsa_parse_ext(en, ospf_is_v2(p), rt);
+  lsa_parse_ext(en, ospf_is_v2(p), ospf_get_af(p), rt);
 
   if (rt->pxopts & OPT_PX_NU)
     return 0;
@@ -1320,10 +1329,10 @@ ospf_rt_abr2(struct ospf_proto *p)
       if (!translate && (oa->translate == TRANS_ON))
       {
 	if (oa->translator_timer == NULL)
-	  oa->translator_timer = tm_new_set(p->p.pool, translator_timer_hook, oa, 0, 0);
+	  oa->translator_timer = tm_new_init(p->p.pool, translator_timer_hook, oa, 0, 0);
 
 	/* Schedule the end of translation */
-	tm_start(oa->translator_timer, oa->ac->transint);
+	tm_start(oa->translator_timer, oa->ac->transint S);
 	oa->translate = TRANS_WAIT;
       }
     }
@@ -1450,7 +1459,7 @@ ospf_ext_spf(struct ospf_proto *p)
     DBG("%s: Working on LSA. ID: %R, RT: %R, Type: %u\n",
 	p->p.name, en->lsa.id, en->lsa.rt, en->lsa_type);
 
-    lsa_parse_ext(en, ospf_is_v2(p), &rt);
+    lsa_parse_ext(en, ospf_is_v2(p), ospf_get_af(p), &rt);
 
     if (!ospf_valid_prefix(&rt.net))
     {
@@ -1679,13 +1688,27 @@ inherit_nexthops(struct nexthop *pn)
   return pn && (ipa_nonzero(pn->gw) || !pn->iface);
 }
 
+static inline ip_addr
+link_lsa_lladdr(struct ospf_proto *p, struct top_hash_entry *en)
+{
+  struct ospf_lsa_link *link_lsa = en->lsa_body;
+  ip6_addr ll = link_lsa->lladdr;
+
+  if (ip6_zero(ll))
+    return IPA_NONE;
+
+  return ospf_is_ip4(p) ? ipa_from_ip4(ospf3_6to4(ll)) : ipa_from_ip6(ll);
+}
+
 static struct nexthop *
 calc_next_hop(struct ospf_area *oa, struct top_hash_entry *en,
-	      struct top_hash_entry *par, int pos)
+	      struct top_hash_entry *par, int pos, uint lif, uint nif)
 {
   struct ospf_proto *p = oa->po;
   struct nexthop *pn = par->nhs;
-  struct ospf_iface *ifa;
+  struct top_hash_entry *link = NULL;
+  struct ospf_iface *ifa = NULL;
+  ip_addr nh = IPA_NONE;
   u32 rid = en->lsa.rt;
 
   /* 16.1.1. The next hop calculation */
@@ -1710,6 +1733,9 @@ calc_next_hop(struct ospf_area *oa, struct top_hash_entry *en,
     if (!ifa)
       return NULL;
 
+    if (ospf_is_v3(p) && (ifa->iface_id != lif))
+      log(L_WARN "%s: Inconsistent interface ID %u/%u", p->p.name, ifa->iface_id, lif);
+
     return new_nexthop(p, IPA_NONE, ifa->iface, ifa->ecmp_weight);
   }
 
@@ -1720,14 +1746,44 @@ calc_next_hop(struct ospf_area *oa, struct top_hash_entry *en,
     if (!ifa)
       return NULL;
 
+    if (ospf_is_v3(p) && (ifa->iface_id != lif))
+      log(L_WARN "%s: Inconsistent interface ID %u/%u", p->p.name, ifa->iface_id, lif);
+
     if (ifa->type == OSPF_IT_VLINK)
       return new_nexthop(p, IPA_NONE, NULL, 0);
 
-    struct ospf_neighbor *m = find_neigh(ifa, rid);
+    /* FIXME: On physical PtP links we may skip next-hop altogether */
-    if (!m || (m->state != NEIGHBOR_FULL))
-      return NULL;
 
-    return new_nexthop(p, m->ip, ifa->iface, ifa->ecmp_weight);
+    if (ospf_is_v2(p) || ospf_is_ip6(p))
+    {
+      /*
+       * In this case, next-hop is a source address from neighbor's packets.
+       * That is necessary for OSPFv2 and practical for OSPFv3 (as it works even
+       * if neighbor uses LinkLSASuppression), but does not work with OSPFv3-AF
+       * on IPv4 topology, where src is IPv6 but next-hop should be IPv4.
+       */
+      struct ospf_neighbor *m = find_neigh(ifa, rid);
+      if (!m || (m->state != NEIGHBOR_FULL))
+	return NULL;
+
+      nh = m->ip;
+    }
+    else
+    {
+      /*
+       * Next-hop is taken from lladdr field of Link-LSA, based on Neighbor
+       * Iface ID (nif) field in our Router-LSA, which is just nbr->iface_id.
+       */
+      link = ospf_hash_find(p->gr, ifa->iface_id, nif, rid, LSA_T_LINK);
+      if (!link)
+	return NULL;
+
+      nh = link_lsa_lladdr(p, link);
+      if (ipa_zero(nh))
+	return NULL;
+    }
+
+    return new_nexthop(p, nh, ifa->iface, ifa->ecmp_weight);
   }
 
   /* The third case - bcast or nbma neighbor */
@@ -1754,18 +1810,15 @@ calc_next_hop(struct ospf_area *oa, struct top_hash_entry *en,
        * Next-hop is taken from lladdr field of Link-LSA, en->lb_id
        * is computed in link_back().
        */
-      struct top_hash_entry *lhe;
+      link = ospf_hash_find(p->gr, pn->iface->index, en->lb_id, rid, LSA_T_LINK);
-      lhe = ospf_hash_find(p->gr, pn->iface->index, en->lb_id, rid, LSA_T_LINK);
+      if (!link)
-
-      if (!lhe)
 	return NULL;
 
-      struct ospf_lsa_link *llsa = lhe->lsa_body;
+      nh = link_lsa_lladdr(p, link);
-
+      if (ipa_zero(nh))
-      if (ip6_zero(llsa->lladdr))
 	return NULL;
 
-      return new_nexthop(p, ipa_from_ip6(llsa->lladdr), pn->iface, pn->weight);
+      return new_nexthop(p, nh, pn->iface, pn->weight);
     }
   }
 
@@ -1778,8 +1831,8 @@ calc_next_hop(struct ospf_area *oa, struct top_hash_entry *en,
 
 /* Add LSA into list of candidates in Dijkstra's algorithm */
 static void
-add_cand(list * l, struct top_hash_entry *en, struct top_hash_entry *par,
+add_cand(struct ospf_area *oa, struct top_hash_entry *en, struct top_hash_entry *par,
-	 u32 dist, struct ospf_area *oa, int pos)
+	 u32 dist, int pos, uint lif, uint nif)
 {
   struct ospf_proto *p = oa->po;
   node *prev, *n;
@@ -1792,9 +1845,9 @@ add_cand(list * l, struct top_hash_entry *en, struct top_hash_entry *par,
   if (en->lsa.age == LSA_MAXAGE)
     return;
 
-  if (ospf_is_v3(p) && (en->lsa_type == LSA_T_RT))
+  if (ospf_is_v3(p) && (oa->options & OPT_V6) && (en->lsa_type == LSA_T_RT))
   {
-    /* In OSPFv3, check V6 flag */
+    /* In OSPFv3 IPv6 unicast, check V6 flag */
     struct ospf_lsa_rt *rt = en->lsa_body;
     if (!(rt->options & OPT_V6))
       return;
@@ -1809,10 +1862,10 @@ add_cand(list * l, struct top_hash_entry *en, struct top_hash_entry *par,
     return;
 
   /* We should check whether there is a reverse link from en to par, */
-  if (!link_back(oa, en, par))
+  if (!link_back(oa, en, par, lif, nif))
     return;
 
-  struct nexthop *nhs = calc_next_hop(oa, en, par, pos);
+  struct nexthop *nhs = calc_next_hop(oa, en, par, pos, lif, nif);
   if (!nhs)
   {
     log(L_WARN "%s: Cannot find next hop for LSA (Type: %04x, Id: %R, Rt: %R)",
@@ -1869,20 +1922,20 @@ add_cand(list * l, struct top_hash_entry *en, struct top_hash_entry *par,
 
   prev = NULL;
 
-  if (EMPTY_LIST(*l))
+  if (EMPTY_LIST(oa->cand))
   {
-    add_head(l, &en->cn);
+    add_head(&oa->cand, &en->cn);
   }
   else
   {
-    WALK_LIST(n, *l)
+    WALK_LIST(n, oa->cand)
     {
       act = SKIP_BACK(struct top_hash_entry, cn, n);
       if ((act->dist > dist) ||
 	  ((act->dist == dist) && (act->lsa_type == LSA_T_RT)))
       {
 	if (prev == NULL)
-	  add_head(l, &en->cn);
+	  add_head(&oa->cand, &en->cn);
 	else
 	  insert_node(&en->cn, prev);
 	added = 1;
@@ -1893,7 +1946,7 @@ add_cand(list * l, struct top_hash_entry *en, struct top_hash_entry *par,
 
     if (!added)
     {
-      add_tail(l, &en->cn);
+      add_tail(&oa->cand, &en->cn);
     }
   }
 }

proto/ospf/topology.c

@@ -70,7 +70,7 @@ ospf_install_lsa(struct ospf_proto *p, struct ospf_lsa_header *lsa, u32 type, u3
   en->lsa_body = body;
   en->lsa = *lsa;
   en->init_age = en->lsa.age;
-  en->inst_time = now;
+  en->inst_time = current_time();
 
   /*
    * We do not set en->mode. It is either default LSA_M_BASIC, or in a special
@@ -128,7 +128,7 @@ ospf_advance_lsa(struct ospf_proto *p, struct top_hash_entry *en, struct ospf_ls
       en->lsa.sn = lsa->sn + 1;
       en->lsa.age = 0;
       en->init_age = 0;
-      en->inst_time = now;
+      en->inst_time = current_time();
       lsa_generate_checksum(&en->lsa, en->lsa_body);
 
       OSPF_TRACE(D_EVENTS, "Advancing LSA: Type: %04x, Id: %R, Rt: %R, Seq: %08x",
@@ -160,7 +160,7 @@ ospf_advance_lsa(struct ospf_proto *p, struct top_hash_entry *en, struct ospf_ls
       en->lsa = *lsa;
       en->lsa.age = LSA_MAXAGE;
       en->init_age = lsa->age;
-      en->inst_time = now;
+      en->inst_time = current_time();
 
       OSPF_TRACE(D_EVENTS, "Resetting LSA:  Type: %04x, Id: %R, Rt: %R, Seq: %08x",
 		 en->lsa_type, en->lsa.id, en->lsa.rt, en->lsa.sn);
@@ -196,7 +196,7 @@ static int
 ospf_do_originate_lsa(struct ospf_proto *p, struct top_hash_entry *en, void *lsa_body, u16 lsa_blen, u16 lsa_opts)
 {
   /* Enforce MinLSInterval */
-  if ((en->init_age == 0) && en->inst_time && ((en->inst_time + MINLSINTERVAL) > now))
+  if (!en->init_age && en->inst_time && (lsa_inst_age(en) < MINLSINTERVAL))
     return 0;
 
   /* Handle wrapping sequence number */
@@ -237,7 +237,7 @@ ospf_do_originate_lsa(struct ospf_proto *p, struct top_hash_entry *en, void *lsa
   en->lsa.sn++;
   en->lsa.age = 0;
   en->init_age = 0;
-  en->inst_time = now;
+  en->inst_time = current_time();
   lsa_generate_checksum(&en->lsa, en->lsa_body);
 
   OSPF_TRACE(D_EVENTS, "Originating LSA: Type: %04x, Id: %R, Rt: %R, Seq: %08x",
@@ -381,7 +381,7 @@ ospf_refresh_lsa(struct ospf_proto *p, struct top_hash_entry *en)
   en->lsa.sn++;
   en->lsa.age = 0;
   en->init_age = 0;
-  en->inst_time = now;
+  en->inst_time = current_time();
   lsa_generate_checksum(&en->lsa, en->lsa_body);
   ospf_flood_lsa(p, en, NULL);
 }
@@ -476,14 +476,15 @@ void
 ospf_update_lsadb(struct ospf_proto *p)
 {
   struct top_hash_entry *en, *nxt;
-  bird_clock_t real_age;
+  btime now_ = current_time();
+  int real_age;
 
   WALK_SLIST_DELSAFE(en, nxt, p->lsal)
   {
     if (en->next_lsa_body)
       ospf_originate_next_lsa(p, en);
 
-    real_age = en->init_age + (now - en->inst_time);
+    real_age = en->init_age + (now_ - en->inst_time) TO_S;
 
     if (en->lsa.age == LSA_MAXAGE)
     {
@@ -1008,7 +1009,7 @@ prepare_sum3_net_lsa_body(struct ospf_proto *p, ort *nf, u32 metric)
   sum = lsab_allocz(p, sizeof(struct ospf_lsa_sum3_net) +
 		    IPV6_PREFIX_SPACE(nf->fn.addr->pxlen));
   sum->metric = metric;
-  ospf_put_ipv6_prefix(sum->prefix, nf->fn.addr, 0, 0);
+  ospf3_put_prefix(sum->prefix, nf->fn.addr, 0, 0);
 }
 
 static inline void
@@ -1097,7 +1098,7 @@ prepare_ext3_lsa_body(struct ospf_proto *p, ort *nf,
   ext->metric = metric & LSA_METRIC_MASK;
   u32 *buf = ext->rest;
 
-  buf = ospf_put_ipv6_prefix(buf, nf->fn.addr, pbit ? OPT_PX_P : 0, 0);
+  buf = ospf3_put_prefix(buf, nf->fn.addr, pbit ? OPT_PX_P : 0, 0);
 
   if (ebit)
     ext->metric |= LSA_EXT3_EBIT;
@@ -1105,7 +1106,7 @@ prepare_ext3_lsa_body(struct ospf_proto *p, ort *nf,
   if (ipa_nonzero(fwaddr))
   {
     ext->metric |= LSA_EXT3_FBIT;
-    buf = ospf_put_ipv6_addr(buf, fwaddr);
+    buf = ospf3_put_addr(buf, fwaddr);
   }
 
   if (tag)
@@ -1222,7 +1223,7 @@ find_surrogate_fwaddr(struct ospf_proto *p, struct ospf_area *oa)
     {
       WALK_LIST(a, ifa->iface->addrs)
       {
-	if ((a->prefix.type != NET_IP6) ||
+	if ((a->prefix.type != ospf_get_af(p)) ||
 	    (a->flags & IA_SECONDARY) ||
 	    (a->flags & IA_PEER) ||
 	    (a->scope <= SCOPE_LINK))
@@ -1316,39 +1317,47 @@ ospf_rt_notify(struct proto *P, struct channel *ch UNUSED, net *n, rte *new, rte
  */
 
 static inline void
-lsab_put_prefix(struct ospf_proto *p, net_addr *net, u32 cost)
+lsab_put_prefix(struct ospf_proto *p, net_addr *n, u32 cost)
 {
-  void *buf = lsab_alloc(p, IPV6_PREFIX_SPACE(net6_pxlen(net)));
+  void *buf = lsab_alloc(p, IPV6_PREFIX_SPACE(net_pxlen(n)));
-  u8 flags = (net6_pxlen(net) < IP6_MAX_PREFIX_LENGTH) ? 0 : OPT_PX_LA;
+  uint max = (n->type == NET_IP4) ? IP4_MAX_PREFIX_LENGTH : IP6_MAX_PREFIX_LENGTH;
-  ospf_put_ipv6_prefix(buf, net, flags, cost);
+  u8 flags = (net_pxlen(n) < max) ? 0 : OPT_PX_LA;
+  ospf3_put_prefix(buf, n, flags, cost);
 }
 
 static void
 prepare_link_lsa_body(struct ospf_proto *p, struct ospf_iface *ifa)
 {
-  struct ospf_lsa_link *ll;
+  ip_addr nh = ospf_is_ip4(p) ? IPA_NONE : ifa->addr->ip;
   int i = 0;
 
+  /* Preallocating space for header */
   ASSERT(p->lsab_used == 0);
-  ll = lsab_allocz(p, sizeof(struct ospf_lsa_link));
+  lsab_allocz(p, sizeof(struct ospf_lsa_link));
-  ll->options = ifa->oa->options | (ifa->priority << 24);
-  ll->lladdr = ipa_to_ip6(ifa->addr->ip);
-  ll = NULL; /* buffer might be reallocated later */
 
   struct ifa *a;
   WALK_LIST(a, ifa->iface->addrs)
   {
-    if ((a->prefix.type != NET_IP6) ||
+    if ((a->prefix.type != ospf_get_af(p)) ||
 	(a->flags & IA_SECONDARY) ||
 	(a->scope <= SCOPE_LINK))
       continue;
 
+    if (ospf_is_ip4(p) && ipa_zero(nh))
+      nh = a->ip;
+
     lsab_put_prefix(p, &a->prefix, 0);
     i++;
   }
 
-  ll = p->lsab;
+  /* Filling the preallocated header */
+  struct ospf_lsa_link *ll = p->lsab;
+  ll->options = ifa->oa->options | (ifa->priority << 24);
+  ll->lladdr = ospf_is_ip4(p) ? ospf3_4to6(ipa_to_ip4(nh)) : ipa_to_ip6(nh);
   ll->pxcount = i;
+
+  if (ipa_zero(nh))
+    log(L_ERR "%s: Cannot find next hop address for %s", p->p.name, ifa->ifname);
 }
 
 static void
@@ -1410,7 +1419,7 @@ prepare_prefix_rt_lsa_body(struct ospf_proto *p, struct ospf_area *oa)
     struct ifa *a;
     WALK_LIST(a, ifa->iface->addrs)
     {
-      if ((a->prefix.type != NET_IP6) ||
+      if ((a->prefix.type != ospf_get_af(p)) ||
 	  (a->flags & IA_SECONDARY) ||
 	  (a->flags & IA_PEER) ||
 	  (a->scope <= SCOPE_LINK))
@@ -1448,7 +1457,7 @@ prepare_prefix_rt_lsa_body(struct ospf_proto *p, struct ospf_area *oa)
 
   /* If there are some configured vlinks, find some global address
      (even from another area), which will be used as a vlink endpoint. */
-  if (!EMPTY_LIST(cf->vlink_list) && !host_addr)
+  if (!EMPTY_LIST(cf->vlink_list) && !host_addr && ospf_is_ip6(p))
   {
     WALK_LIST(ifa, p->iface_list)
     {
@@ -1571,7 +1580,7 @@ add_link_lsa(struct ospf_proto *p, struct ospf_lsa_link *ll, int offset, int *px
       continue;
 
     /* Skip link-local prefixes */
-    if ((pxlen >= 10) && ((pxb[1] & 0xffc00000) == 0xfe800000))
+    if (ospf_is_ip6(p) && (pxlen >= 10) && ((pxb[1] & 0xffc00000) == 0xfe800000))
       continue;
 
     add_prefix(p, pxb, offset, pxc);
@@ -1628,7 +1637,7 @@ ospf_originate_prefix_net_lsa(struct ospf_proto *p, struct ospf_iface *ifa)
 }
 
 static inline int breaks_minlsinterval(struct top_hash_entry *en)
-{ return en && (en->lsa.age < LSA_MAXAGE) && ((en->inst_time + MINLSINTERVAL) > now); }
+{ return en && (en->lsa.age < LSA_MAXAGE) && (lsa_inst_age(en) < MINLSINTERVAL); }
 
 void
 ospf_update_topology(struct ospf_proto *p)

proto/ospf/topology.h

@@ -26,7 +26,7 @@ struct top_hash_entry
   void *next_lsa_body;		/* For postponed LSA origination */
   u16 next_lsa_blen;		/* For postponed LSA origination */
   u16 next_lsa_opts;		/* For postponed LSA origination */
-  bird_clock_t inst_time;	/* Time of installation into DB */
+  btime inst_time;		/* Time of installation into DB */
   struct ort *nf;		/* Reference fibnode for sum and ext LSAs, NULL for otherwise */
   struct nexthop *nhs;		/* Computed nexthops - valid only in ospf_rt_spf() */
   ip_addr lb;			/* In OSPFv2, link back address. In OSPFv3, any global address in the area useful for vlinks */

proto/radv/config.Y

@@ -27,10 +27,13 @@ static u8 radv_mult_val;	/* Used by radv_mult for second return value */
 CF_DECLS
 
 CF_KEYWORDS(RADV, PREFIX, INTERFACE, MIN, MAX, RA, DELAY, INTERVAL,
-	MANAGED, OTHER, CONFIG, LINK, MTU, REACHABLE, TIME, RETRANS,
+	MANAGED, OTHER, CONFIG, LINGER, LINK, MTU, REACHABLE, TIME, RETRANS,
 	TIMER, CURRENT, HOP, LIMIT, DEFAULT, VALID, PREFERRED, MULT,
 	LIFETIME, SKIP, ONLINK, AUTONOMOUS, RDNSS, DNSSL, NS, DOMAIN,
-	LOCAL, TRIGGER, SENSITIVE, PREFERENCE, LOW, MEDIUM, HIGH)
+	LOCAL, TRIGGER, SENSITIVE, PREFERENCE, LOW, MEDIUM, HIGH, PROPAGATE,
+	ROUTE, ROUTES, RA_PREFERENCE, RA_LIFETIME)
+
+CF_ENUM(T_ENUM_RA_PREFERENCE, RA_PREF_, LOW, MEDIUM, HIGH)
 
 %type<i> radv_mult radv_sensitive radv_preference
 
@@ -56,6 +59,7 @@ radv_proto_item:
  | RDNSS { init_list(&radv_dns_list); } radv_rdnss { add_tail_list(&RADV_CFG->rdnss_list, &radv_dns_list); }
  | DNSSL { init_list(&radv_dns_list); } radv_dnssl { add_tail_list(&RADV_CFG->dnssl_list, &radv_dns_list); }
  | TRIGGER net_ip6 { RADV_CFG->trigger = $2; }
+ | PROPAGATE ROUTES bool { RADV_CFG->propagate_routes = $3; }
  ;
 
 radv_proto_opts:
@@ -76,13 +80,18 @@ radv_iface_start:
   init_list(&RADV_IFACE->rdnss_list);
   init_list(&RADV_IFACE->dnssl_list);
 
-  RADV_IFACE->min_ra_int = -1; /* undefined */
+  RADV_IFACE->min_ra_int = (u32) -1; /* undefined */
   RADV_IFACE->max_ra_int = DEFAULT_MAX_RA_INT;
   RADV_IFACE->min_delay = DEFAULT_MIN_DELAY;
+  RADV_IFACE->prefix_linger_time = (u32) -1;
+  RADV_IFACE->route_linger_time = (u32) -1;
   RADV_IFACE->current_hop_limit = DEFAULT_CURRENT_HOP_LIMIT;
-  RADV_IFACE->default_lifetime = -1;
+  RADV_IFACE->default_lifetime = (u32) -1;
   RADV_IFACE->default_lifetime_sensitive = 1;
   RADV_IFACE->default_preference = RA_PREF_MEDIUM;
+  RADV_IFACE->route_lifetime = (u32) -1;
+  RADV_IFACE->route_lifetime_sensitive = 0;
+  RADV_IFACE->route_preference = RA_PREF_MEDIUM;
 };
 
 radv_iface_item:
@@ -100,7 +109,14 @@ radv_iface_item:
      if ($3 > 9000)  cf_error("Default lifetime must be in range 0-9000");
      if ($4 != (uint) -1) RADV_IFACE->default_lifetime_sensitive = $4;
    }
+ | ROUTE LIFETIME expr radv_sensitive {
+     RADV_IFACE->route_lifetime = $3;
+     if ($4 != (uint) -1) RADV_IFACE->route_lifetime_sensitive = $4;
+   }
  | DEFAULT PREFERENCE radv_preference { RADV_IFACE->default_preference = $3; }
+ | ROUTE PREFERENCE radv_preference { RADV_IFACE->route_preference = $3; }
+ | PREFIX LINGER TIME expr { RADV_IFACE->prefix_linger_time = $4; }
+ | ROUTE LINGER TIME expr { RADV_IFACE->route_linger_time = $4; }
  | PREFIX radv_prefix { add_tail(&RADV_IFACE->pref_list, NODE this_radv_prefix); }
  | RDNSS { init_list(&radv_dns_list); } radv_rdnss { add_tail_list(&RADV_IFACE->rdnss_list, &radv_dns_list); }
  | DNSSL { init_list(&radv_dns_list); } radv_dnssl { add_tail_list(&RADV_IFACE->dnssl_list, &radv_dns_list); }
@@ -123,12 +139,32 @@ radv_iface_finish:
   if (ic->default_lifetime == (u32) -1)
     ic->default_lifetime = 3 * ic->max_ra_int;
 
+  if (ic->route_lifetime == (u32) -1)
+    ic->route_lifetime = 3 * ic->max_ra_int;
+
+  if (ic->prefix_linger_time == (u32) -1)
+    ic->prefix_linger_time = 3 * ic->max_ra_int;
+
+  if (ic->route_linger_time == (u32) -1)
+    ic->route_linger_time = 3 * ic->max_ra_int;
+
   if ((ic->min_ra_int > 3) &&
       (ic->min_ra_int > (ic->max_ra_int * 3 / 4)))
     cf_error("Min RA interval must be at most 3/4 * Max RA interval");
 
   if ((ic->default_lifetime > 0) && (ic->default_lifetime < ic->max_ra_int))
     cf_error("Default lifetime must be either 0 or at least Max RA interval");
+
+  if ((ic->route_lifetime > 0) && (ic->route_lifetime < ic->max_ra_int))
+    cf_error("Route lifetime must be either 0 or at least Max RA interval");
+
+  if ((ic->prefix_linger_time > 0) && (ic->prefix_linger_time < ic->max_ra_int))
+    cf_error("Prefix linger time must be either 0 or at least Max RA interval");
+
+  if ((ic->route_linger_time > 0) && (ic->route_linger_time < ic->max_ra_int))
+    cf_error("Route linger time must be either 0 or at least Max RA interval");
+
+  RADV_CFG->max_linger_time = MAX_(RADV_CFG->max_linger_time, ic->route_linger_time);
 };
 
 
@@ -292,10 +328,13 @@ radv_mult:
  ;
 
 radv_sensitive:
-   /* empty */ { $$ = -1; }
+   /* empty */ { $$ = (uint) -1; }
  | SENSITIVE bool { $$ = $2; }
  ;
 
+CF_ADDTO(dynamic_attr, RA_PREFERENCE { $$ = f_new_dynamic_attr(EAF_TYPE_INT, T_ENUM_RA_PREFERENCE, EA_RA_PREFERENCE); })
+CF_ADDTO(dynamic_attr, RA_LIFETIME { $$ = f_new_dynamic_attr(EAF_TYPE_INT, T_INT, EA_RA_LIFETIME); })
+
 CF_CODE
 
 CF_END

proto/radv/packets.c

@@ -26,6 +26,7 @@ struct radv_ra_packet
 
 #define OPT_PREFIX	3
 #define OPT_MTU		5
+#define OPT_ROUTE	24
 #define OPT_RDNSS	25
 #define OPT_DNSSL	31
 
@@ -52,6 +53,15 @@ struct radv_opt_mtu
   u32 mtu;
 };
 
+struct radv_opt_route {
+  u8 type;
+  u8 length;
+  u8 pxlen;
+  u8 flags;
+  u32 lifetime;
+  u8 prefix[];
+};
+
 struct radv_opt_rdnss
 {
   u8 type;
@@ -70,34 +80,42 @@ struct radv_opt_dnssl
   char domain[];
 };
 
-
+static int
-static struct radv_prefix_config default_prefix = {
+radv_prepare_route(struct radv_iface *ifa, struct radv_route *rt,
-  .onlink = 1,
+		   char **buf, char *bufend)
-  .autonomous = 1,
-  .valid_lifetime = DEFAULT_VALID_LIFETIME,
-  .preferred_lifetime = DEFAULT_PREFERRED_LIFETIME
-};
-
-
-static struct radv_prefix_config *
-radv_prefix_match(struct radv_iface *ifa, struct ifa *a)
 {
-  struct proto *p = &ifa->ra->p;
+  struct radv_proto *p = ifa->ra;
-  struct radv_config *cf = (struct radv_config *) (p->cf);
+  u8 px_blocks = (net6_pxlen(rt->n.addr) + 63) / 64;
-  struct radv_prefix_config *pc;
+  u8 opt_len = 8 * (1 + px_blocks);
 
-  if (a->scope <= SCOPE_LINK)
+  if (*buf + opt_len > bufend)
-    return NULL;
+  {
+    log(L_WARN, "%s: Too many RA options on interface %s",
+	p->p.name, ifa->iface->name);
+    return -1;
+  }
 
-  WALK_LIST(pc, ifa->cf->pref_list)
+  uint preference = rt->preference_set ? rt->preference : ifa->cf->route_preference;
-    if (net_in_netX(&a->prefix, (net_addr *) &pc->prefix))
+  uint lifetime = rt->lifetime_set ? rt->lifetime : ifa->cf->route_lifetime;
-      return pc;
+  uint valid = rt->valid && p->valid && (p->active || !ifa->cf->route_lifetime_sensitive);
 
-  WALK_LIST(pc, cf->pref_list)
+  struct radv_opt_route *opt = (void *) *buf;
-    if (net_in_netX(&a->prefix, (net_addr *) &pc->prefix))
+  *buf += opt_len;
-      return pc;
+  opt->type = OPT_ROUTE;
+  opt->length = 1 + px_blocks;
+  opt->pxlen = net6_pxlen(rt->n.addr);
+  opt->flags = preference;
+  opt->lifetime = valid ? htonl(lifetime) : 0;
 
-  return &default_prefix;
+  /* Copy the relevant part of the prefix */
+  ip6_addr px_addr = ip6_hton(net6_prefix(rt->n.addr));
+  memcpy(opt->prefix, &px_addr, 8 * px_blocks);
+
+  /* Keeping track of first linger timeout */
+  if (!rt->valid)
+    ifa->valid_time = MIN(ifa->valid_time, rt->changed + ifa->cf->route_linger_time S);
+
+  return 0;
 }
 
 static int
@@ -206,7 +224,7 @@ radv_prepare_dnssl(struct radv_iface *ifa, list *dnssl_list, char **buf, char *b
     else
       op->lifetime = htonl(dcf->lifetime);
 
-    while(NODE_VALID(dcf) && 
+    while(NODE_VALID(dcf) &&
 	  (dcf->lifetime == dcf_base->lifetime) &&
 	  (dcf->lifetime_mult == dcf_base->lifetime_mult))
       {
@@ -234,12 +252,47 @@ radv_prepare_dnssl(struct radv_iface *ifa, list *dnssl_list, char **buf, char *b
   return -1;
 }
 
+static int
+radv_prepare_prefix(struct radv_iface *ifa, struct radv_prefix *px,
+		    char **buf, char *bufend)
+{
+  struct radv_prefix_config *pc = px->cf;
+
+  if (*buf + sizeof(struct radv_opt_prefix) > bufend)
+  {
+    log(L_WARN "%s: Too many prefixes on interface %s",
+	ifa->ra->p.name, ifa->iface->name);
+    return -1;
+  }
+
+  struct radv_opt_prefix *op = (void *) *buf;
+  op->type = OPT_PREFIX;
+  op->length = 4;
+  op->pxlen = px->prefix.pxlen;
+  op->flags = (pc->onlink ? OPT_PX_ONLINK : 0) |
+    (pc->autonomous ? OPT_PX_AUTONOMOUS : 0);
+  op->valid_lifetime = (ifa->ra->active || !pc->valid_lifetime_sensitive) ?
+    htonl(pc->valid_lifetime) : 0;
+  op->preferred_lifetime = (ifa->ra->active || !pc->preferred_lifetime_sensitive) ?
+    htonl(pc->preferred_lifetime) : 0;
+  op->reserved = 0;
+  op->prefix = ip6_hton(px->prefix.prefix);
+  *buf += sizeof(*op);
+
+  /* Keeping track of first linger timeout */
+  if (!px->valid)
+    ifa->valid_time = MIN(ifa->valid_time, px->changed + ifa->cf->prefix_linger_time S);
+
+  return 0;
+}
+
 static void
 radv_prepare_ra(struct radv_iface *ifa)
 {
-  struct proto_radv *ra = ifa->ra;
+  struct radv_proto *p = ifa->ra;
-  struct radv_config *cf = (struct radv_config *) (ra->p.cf);
+  struct radv_config *cf = (struct radv_config *) (p->p.cf);
   struct radv_iface_config *ic = ifa->cf;
+  btime now = current_time();
 
   char *buf = ifa->sk->tbuf;
   char *bufstart = buf;
@@ -250,7 +303,7 @@ radv_prepare_ra(struct radv_iface *ifa)
   pkt->code = 0;
   pkt->checksum = 0;
   pkt->current_hop_limit = ic->current_hop_limit;
-  pkt->router_lifetime = (ra->active || !ic->default_lifetime_sensitive) ?
+  pkt->router_lifetime = (p->valid && (p->active || !ic->default_lifetime_sensitive)) ?
     htons(ic->default_lifetime) : 0;
   pkt->flags = (ic->managed ? OPT_RA_MANAGED : 0) |
     (ic->other_config ? OPT_RA_OTHER_CFG : 0) |
@@ -269,37 +322,18 @@ radv_prepare_ra(struct radv_iface *ifa)
     buf += sizeof (*om);
   }
 
-  struct ifa *addr;
+  /* Keeping track of first linger timeout */
-  WALK_LIST(addr, ifa->iface->addrs)
+  ifa->valid_time = TIME_INFINITY;
+
+  struct radv_prefix *px;
+  WALK_LIST(px, ifa->prefixes)
   {
-    if (addr->prefix.type != NET_IP6)
+    /* Skip invalid prefixes that are past linger timeout but still not pruned */
-      continue;
+    if (!px->valid && ((px->changed + ic->prefix_linger_time S) <= now))
+	continue;
 
-    struct radv_prefix_config *pc;
+    if (radv_prepare_prefix(ifa, px, &buf, bufend) < 0)
-    pc = radv_prefix_match(ifa, addr);
-
-    if (!pc || pc->skip)
-      continue;
-
-    if (buf + sizeof(struct radv_opt_prefix) > bufend)
-    {
-      log(L_WARN "%s: Too many prefixes on interface %s", ra->p.name, ifa->iface->name);
       goto done;
-    }
-
-    struct radv_opt_prefix *op = (void *) buf;
-    op->type = OPT_PREFIX;
-    op->length = 4;
-    op->pxlen = net6_pxlen(&addr->prefix);
-    op->flags = (pc->onlink ? OPT_PX_ONLINK : 0) |
-      (pc->autonomous ? OPT_PX_AUTONOMOUS : 0);
-    op->valid_lifetime = (ra->active || !pc->valid_lifetime_sensitive) ?
-      htonl(pc->valid_lifetime) : 0;
-    op->preferred_lifetime = (ra->active || !pc->preferred_lifetime_sensitive) ?
-      htonl(pc->preferred_lifetime) : 0;
-    op->reserved = 0;
-    op->prefix = ip6_hton(net6_prefix(&addr->prefix));
-    buf += sizeof(*op);
   }
 
   if (! ic->rdnss_local)
@@ -316,33 +350,34 @@ radv_prepare_ra(struct radv_iface *ifa)
   if (radv_prepare_dnssl(ifa, &ic->dnssl_list, &buf, bufend) < 0)
     goto done;
 
+  if (p->fib_up)
+  {
+    FIB_WALK(&p->routes, struct radv_route, rt)
+    {
+      /* Skip invalid routes that are past linger timeout but still not pruned */
+      if (!rt->valid && ((rt->changed + ic->route_linger_time S) <= now))
+	continue;
+
+      if (radv_prepare_route(ifa, rt, &buf, bufend) < 0)
+	goto done;
+    }
+    FIB_WALK_END;
+  }
+
  done:
   ifa->plen = buf - bufstart;
 }
 
 
 void
-radv_send_ra(struct radv_iface *ifa, int shutdown)
+radv_send_ra(struct radv_iface *ifa)
 {
-  struct proto_radv *ra = ifa->ra;
+  struct radv_proto *p = ifa->ra;
 
   /* We store prepared RA in tbuf */
   if (!ifa->plen)
     radv_prepare_ra(ifa);
 
-  if (shutdown)
-  {
-    /*
-     * Modify router lifetime to 0, it is not restored because we suppose that
-     * the iface will be removed. The preference value also has to be zeroed.
-     * (RFC 4191 2.2: If router lifetime is 0, the preference value must be 0.)
-     */
-
-    struct radv_ra_packet *pkt = (void *) ifa->sk->tbuf;
-    pkt->router_lifetime = 0;
-    pkt->flags &= ~RA_PREF_MASK;
-  }
-
   RADV_TRACE(D_PACKETS, "Sending RA via %s", ifa->iface->name);
   sk_send_to(ifa->sk, ifa->plen, IP6_ALL_NODES, 0);
 }
@@ -352,13 +387,13 @@ static int
 radv_rx_hook(sock *sk, uint size)
 {
   struct radv_iface *ifa = sk->data;
-  struct proto_radv *ra = ifa->ra;
+  struct radv_proto *p = ifa->ra;
 
   /* We want just packets from sk->iface */
   if (sk->lifindex != sk->iface->index)
     return 1;
 
-  if (ipa_equal(sk->faddr, ifa->addr->ip))
+  if (ipa_equal(sk->faddr, sk->saddr))
     return 1;
 
   if (size < 8)
@@ -408,11 +443,12 @@ radv_err_hook(sock *sk, int err)
 int
 radv_sk_open(struct radv_iface *ifa)
 {
-  sock *sk = sk_new(ifa->ra->p.pool);
+  sock *sk = sk_new(ifa->pool);
   sk->type = SK_IP;
   sk->subtype = SK_IPV6;
   sk->dport = ICMPV6_PROTO;
   sk->saddr = ifa->addr->ip;
+  sk->vrf = ifa->ra->p.vrf;
 
   sk->ttl = 255; /* Mandatory for Neighbor Discovery packets */
   sk->rx_hook = radv_rx_hook;

proto/radv/radv.c

@@ -12,59 +12,206 @@
 /**
  * DOC: Router Advertisements
  *
- * The RAdv protocol is implemented in two files: |radv.c| containing
+ * The RAdv protocol is implemented in two files: |radv.c| containing the
- * the interface with BIRD core and the protocol logic and |packets.c|
+ * interface with BIRD core and the protocol logic and |packets.c| handling low
- * handling low level protocol stuff (RX, TX and packet formats).
+ * level protocol stuff (RX, TX and packet formats). The protocol does not
- * The protocol does not export any routes.
+ * export any routes.
  *
- * The RAdv is structured in the usual way - for each handled interface
+ * The RAdv is structured in the usual way - for each handled interface there is
- * there is a structure &radv_iface that contains a state related to
+ * a structure &radv_iface that contains a state related to that interface
- * that interface together with its resources (a socket, a timer).
+ * together with its resources (a socket, a timer). There is also a prepared RA
- * There is also a prepared RA stored in a TX buffer of the socket
+ * stored in a TX buffer of the socket associated with an iface. These iface
- * associated with an iface. These iface structures are created
+ * structures are created and removed according to iface events from BIRD core
- * and removed according to iface events from BIRD core handled by
+ * handled by radv_if_notify() callback.
- * radv_if_notify() callback.
  *
- * The main logic of RAdv consists of two functions:
+ * The main logic of RAdv consists of two functions: radv_iface_notify(), which
- * radv_iface_notify(), which processes asynchronous events (specified
+ * processes asynchronous events (specified by RA_EV_* codes), and radv_timer(),
- * by RA_EV_* codes), and radv_timer(), which triggers sending RAs and
+ * which triggers sending RAs and computes the next timeout.
- * computes the next timeout.
  *
- * The RAdv protocol could receive routes (through
+ * The RAdv protocol could receive routes (through radv_import_control() and
- * radv_import_control() and radv_rt_notify()), but only the
+ * radv_rt_notify()), but only the configured trigger route is tracked (in
- * configured trigger route is tracked (in &active var).  When a radv
+ * &active var).  When a radv protocol is reconfigured, the connected routing
- * protocol is reconfigured, the connected routing table is examined
+ * table is examined (in radv_check_active()) to have proper &active value in
- * (in radv_check_active()) to have proper &active value in case of
+ * case of the specified trigger prefix was changed.
- * the specified trigger prefix was changed.
  *
  * Supported standards:
  * - RFC 4861 - main RA standard
+ * - RFC 4191 - Default Router Preferences and More-Specific Routes
  * - RFC 6106 - DNS extensions (RDDNS, DNSSL)
- * - RFC 4191 (partial) - Default Router Preference
  */
 
+static void radv_prune_prefixes(struct radv_iface *ifa);
+static void radv_prune_routes(struct radv_proto *p);
+
+/* Invalidate cached RA packet */
+static inline void radv_invalidate(struct radv_iface *ifa)
+{ ifa->plen = 0; }
+
 static void
 radv_timer(timer *tm)
 {
   struct radv_iface *ifa = tm->data;
-  struct proto_radv *ra = ifa->ra;
+  struct radv_proto *p = ifa->ra;
+  btime now = current_time();
 
   RADV_TRACE(D_EVENTS, "Timer fired on %s", ifa->iface->name);
 
-  radv_send_ra(ifa, 0);
+  if (ifa->valid_time <= now)
+    radv_invalidate(ifa);
+
+  if (ifa->prune_time <= now)
+    radv_prune_prefixes(ifa);
+
+  if (p->prune_time <= now)
+    radv_prune_routes(p);
+
+  radv_send_ra(ifa);
 
   /* Update timer */
   ifa->last = now;
-  unsigned after = ifa->cf->min_ra_int;
+  btime t = ifa->cf->min_ra_int S;
-  after += random() % (ifa->cf->max_ra_int - ifa->cf->min_ra_int + 1);
+  btime r = (ifa->cf->max_ra_int - ifa->cf->min_ra_int) S;
+  t += random() % (r + 1);
 
   if (ifa->initial)
+  {
+    t = MIN(t, MAX_INITIAL_RTR_ADVERT_INTERVAL);
     ifa->initial--;
+  }
 
-  if (ifa->initial)
+  tm_start(ifa->timer, t);
-    after = MIN(after, MAX_INITIAL_RTR_ADVERT_INTERVAL);
+}
 
-  tm_start(ifa->timer, after);
+static struct radv_prefix_config default_prefix = {
+  .onlink = 1,
+  .autonomous = 1,
+  .valid_lifetime = DEFAULT_VALID_LIFETIME,
+  .preferred_lifetime = DEFAULT_PREFERRED_LIFETIME
+};
+
+static struct radv_prefix_config dead_prefix = {
+};
+
+/* Find a corresponding config for the given prefix */
+static struct radv_prefix_config *
+radv_prefix_match(struct radv_iface *ifa, net_addr_ip6 *px)
+{
+  struct radv_proto *p = ifa->ra;
+  struct radv_config *cf = (struct radv_config *) (p->p.cf);
+  struct radv_prefix_config *pc;
+
+  WALK_LIST(pc, ifa->cf->pref_list)
+    if (net_in_net_ip6(px, &pc->prefix))
+      return pc;
+
+  WALK_LIST(pc, cf->pref_list)
+    if (net_in_net_ip6(px, &pc->prefix))
+      return pc;
+
+  return &default_prefix;
+}
+
+/*
+ * Go through the list of prefixes, compare them with configs and decide if we
+ * want them or not.
+ */
+static void
+radv_prepare_prefixes(struct radv_iface *ifa)
+{
+  struct radv_proto *p = ifa->ra;
+  struct radv_prefix *pfx, *next;
+  btime now = current_time();
+
+  /* First mark all the prefixes as unused */
+  WALK_LIST(pfx, ifa->prefixes)
+    pfx->mark = 0;
+
+  /* Find all the prefixes we want to use and make sure they are in the list. */
+  struct ifa *addr;
+  WALK_LIST(addr, ifa->iface->addrs)
+  {
+    if ((addr->prefix.type != NET_IP6) ||
+	(addr->scope <= SCOPE_LINK))
+      continue;
+
+    net_addr_ip6 *prefix = (void *) &addr->prefix;
+    struct radv_prefix_config *pc = radv_prefix_match(ifa, prefix);
+
+    if (!pc || pc->skip)
+      continue;
+
+    /* Do we have it already? */
+    struct radv_prefix *existing = NULL;
+    WALK_LIST(pfx, ifa->prefixes)
+      if (net_equal_ip6(&pfx->prefix, prefix))
+      {
+	existing = pfx;
+	break;
+      }
+
+    if (!existing)
+    {
+      RADV_TRACE(D_EVENTS, "Adding new prefix %N on %s",
+		 prefix, ifa->iface->name);
+
+      existing = mb_allocz(ifa->pool, sizeof *existing);
+      net_copy_ip6(&existing->prefix, prefix);
+      add_tail(&ifa->prefixes, NODE existing);
+    }
+
+    /*
+     * Update the information (it may have changed, or even bring a prefix back
+     * to life).
+     */
+    existing->valid = 1;
+    existing->changed = now;
+    existing->mark = 1;
+    existing->cf = pc;
+  }
+
+  WALK_LIST_DELSAFE(pfx, next, ifa->prefixes)
+  {
+    if (pfx->valid && !pfx->mark)
+    {
+      RADV_TRACE(D_EVENTS, "Invalidating prefix %N on %s",
+		 pfx->prefix, ifa->iface->name);
+
+      pfx->valid = 0;
+      pfx->changed = now;
+      pfx->cf = &dead_prefix;
+    }
+  }
+}
+
+static void
+radv_prune_prefixes(struct radv_iface *ifa)
+{
+  struct radv_proto *p = ifa->ra;
+  btime now = current_time();
+  btime next = TIME_INFINITY;
+  btime expires = 0;
+
+  struct radv_prefix *px, *pxn;
+  WALK_LIST_DELSAFE(px, pxn, ifa->prefixes)
+  {
+    if (!px->valid)
+    {
+      expires = px->changed + ifa->cf->prefix_linger_time S;
+
+      if (expires <= now)
+      {
+	RADV_TRACE(D_EVENTS, "Removing prefix %N on %s",
+		   px->prefix, ifa->iface->name);
+
+	rem_node(NODE px);
+	mb_free(px);
+      }
+      else
+	next = MIN(next, expires);
+    }
+  }
+
+  ifa->prune_time = next;
 }
 
 static char* ev_name[] = { NULL, "Init", "Change", "RS" };
@@ -72,7 +219,7 @@ static char* ev_name[] = { NULL, "Init", "Change", "RS" };
 void
 radv_iface_notify(struct radv_iface *ifa, int event)
 {
-  struct proto_radv *ra = ifa->ra;
+  struct radv_proto *p = ifa->ra;
 
   if (!ifa->sk)
     return;
@@ -82,9 +229,11 @@ radv_iface_notify(struct radv_iface *ifa, int event)
   switch (event)
   {
   case RA_EV_CHANGE:
-    ifa->plen = 0;
+    radv_invalidate(ifa);
   case RA_EV_INIT:
     ifa->initial = MAX_INITIAL_RTR_ADVERTISEMENTS;
+    radv_prepare_prefixes(ifa);
+    radv_prune_prefixes(ifa);
     break;
 
   case RA_EV_RS:
@@ -92,31 +241,25 @@ radv_iface_notify(struct radv_iface *ifa, int event)
   }
 
   /* Update timer */
-  unsigned delta = now - ifa->last;
+  btime t = ifa->last + ifa->cf->min_delay S - current_time();
-  unsigned after = 0;
+  tm_start(ifa->timer, t);
-
-  if (delta < ifa->cf->min_delay)
-    after = ifa->cf->min_delay - delta;
-
-  tm_start(ifa->timer, after);
 }
 
 static void
-radv_iface_notify_all(struct proto_radv *ra, int event)
+radv_iface_notify_all(struct radv_proto *p, int event)
 {
   struct radv_iface *ifa;
 
-  WALK_LIST(ifa, ra->iface_list)
+  WALK_LIST(ifa, p->iface_list)
     radv_iface_notify(ifa, event);
 }
 
-
 static struct radv_iface *
-radv_iface_find(struct proto_radv *ra, struct iface *what)
+radv_iface_find(struct radv_proto *p, struct iface *what)
 {
   struct radv_iface *ifa;
 
-  WALK_LIST(ifa, ra->iface_list)
+  WALK_LIST(ifa, p->iface_list)
     if (ifa->iface == what)
       return ifa;
 
@@ -127,59 +270,39 @@ static void
 radv_iface_add(struct object_lock *lock)
 {
   struct radv_iface *ifa = lock->data;
-  struct proto_radv *ra = ifa->ra;
+  struct radv_proto *p = ifa->ra;
 
   if (! radv_sk_open(ifa))
   {
-    log(L_ERR "%s: Socket open failed on interface %s", ra->p.name, ifa->iface->name);
+    log(L_ERR "%s: Socket open failed on interface %s", p->p.name, ifa->iface->name);
     return;
   }
 
   radv_iface_notify(ifa, RA_EV_INIT);
 }
 
-static inline struct ifa *
-find_lladdr(struct iface *iface)
-{
-  struct ifa *a;
-  WALK_LIST(a, iface->addrs)
-    if ((a->prefix.type == NET_IP6) && (a->scope == SCOPE_LINK))
-      return a;
-
-  return NULL;
-}
-
 static void
-radv_iface_new(struct proto_radv *ra, struct iface *iface, struct radv_iface_config *cf)
+radv_iface_new(struct radv_proto *p, struct iface *iface, struct radv_iface_config *cf)
 {
-  pool *pool = ra->p.pool;
   struct radv_iface *ifa;
 
   RADV_TRACE(D_EVENTS, "Adding interface %s", iface->name);
 
+  pool *pool = rp_new(p->p.pool, iface->name);
   ifa = mb_allocz(pool, sizeof(struct radv_iface));
-  ifa->ra = ra;
+  ifa->pool = pool;
+  ifa->ra = p;
   ifa->cf = cf;
   ifa->iface = iface;
+  ifa->addr = iface->llv6;
+  init_list(&ifa->prefixes);
+  ifa->prune_time = TIME_INFINITY;
 
-  add_tail(&ra->iface_list, NODE ifa);
+  add_tail(&p->iface_list, NODE ifa);
 
-  ifa->addr = find_lladdr(iface);
+  ifa->timer = tm_new_init(pool, radv_timer, ifa, 0, 0);
-  if (!ifa->addr)
-  {
-    log(L_ERR "%s: Cannot find link-locad addr on interface %s", ra->p.name, iface->name);
-    return;
-  }
-
-  timer *tm = tm_new(pool);
-  tm->hook = radv_timer;
-  tm->data = ifa;
-  tm->randomize = 0;
-  tm->recurrent = 0;
-  ifa->timer = tm;
 
   struct object_lock *lock = olock_new(pool);
-  lock->addr = IPA_NONE;
   lock->type = OBJLOCK_IP;
   lock->port = ICMPV6_PROTO;
   lock->iface = iface;
@@ -193,39 +316,42 @@ radv_iface_new(struct proto_radv *ra, struct iface *iface, struct radv_iface_con
 static void
 radv_iface_remove(struct radv_iface *ifa)
 {
-  struct proto_radv *ra = ifa->ra;
+  struct radv_proto *p = ifa->ra;
   RADV_TRACE(D_EVENTS, "Removing interface %s", ifa->iface->name);
 
   rem_node(NODE ifa);
 
-  rfree(ifa->sk);
+  rfree(ifa->pool);
-  rfree(ifa->timer);
-  rfree(ifa->lock);
-
-  mb_free(ifa);
 }
 
 static void
-radv_if_notify(struct proto *p, unsigned flags, struct iface *iface)
+radv_if_notify(struct proto *P, unsigned flags, struct iface *iface)
 {
-  struct proto_radv *ra = (struct proto_radv *) p;
+  struct radv_proto *p = (struct radv_proto *) P;
-  struct radv_config *cf = (struct radv_config *) (p->cf);
+  struct radv_config *cf = (struct radv_config *) (P->cf);
 
   if (iface->flags & IF_IGNORE)
     return;
 
   if (flags & IF_CHANGE_UP)
   {
-    struct radv_iface_config *ic = (struct radv_iface_config *)
+    struct radv_iface_config *ic = (void *) iface_patt_find(&cf->patt_list, iface, NULL);
-      iface_patt_find(&cf->patt_list, iface, NULL);
+
+    /* Ignore non-multicast ifaces */
+    if (!(iface->flags & IF_MULTICAST))
+      return;
+
+    /* Ignore ifaces without link-local address */
+    if (!iface->llv6)
+      return;
 
     if (ic)
-      radv_iface_new(ra, iface, ic);
+      radv_iface_new(p, iface, ic);
 
     return;
   }
 
-  struct radv_iface *ifa = radv_iface_find(ra, iface);
+  struct radv_iface *ifa = radv_iface_find(p, iface);
   if (!ifa)
     return;
 
@@ -240,9 +366,9 @@ radv_if_notify(struct proto *p, unsigned flags, struct iface *iface)
 }
 
 static void
-radv_ifa_notify(struct proto *p, unsigned flags UNUSED, struct ifa *a)
+radv_ifa_notify(struct proto *P, unsigned flags UNUSED, struct ifa *a)
 {
-  struct proto_radv *ra = (struct proto_radv *) p;
+  struct radv_proto *p = (struct radv_proto *) P;
 
   if (a->flags & IA_SECONDARY)
     return;
@@ -250,7 +376,7 @@ radv_ifa_notify(struct proto *p, unsigned flags UNUSED, struct ifa *a)
   if (a->scope <= SCOPE_LINK)
     return;
 
-  struct radv_iface *ifa = radv_iface_find(ra, a->iface);
+  struct radv_iface *ifa = radv_iface_find(p, a->iface);
 
   if (ifa)
     radv_iface_notify(ifa, RA_EV_CHANGE);
@@ -269,50 +395,171 @@ radv_net_match_trigger(struct radv_config *cf, net *n)
 }
 
 int
-radv_import_control(struct proto *p, rte **new, ea_list **attrs UNUSED, struct linpool *pool UNUSED)
+radv_import_control(struct proto *P, rte **new, ea_list **attrs UNUSED, struct linpool *pool UNUSED)
 {
-  // struct proto_radv *ra = (struct proto_radv *) p;
+  // struct radv_proto *p = (struct radv_proto *) P;
-  struct radv_config *cf = (struct radv_config *) (p->cf);
+  struct radv_config *cf = (struct radv_config *) (P->cf);
 
   if (radv_net_match_trigger(cf, (*new)->net))
     return RIC_PROCESS;
 
-  return RIC_DROP;
+  if (cf->propagate_routes)
+    return RIC_PROCESS;
+  else
+    return RIC_DROP;
 }
 
 static void
-radv_rt_notify(struct proto *p, struct channel *ch UNUSED, net *n, rte *new, rte *old UNUSED, ea_list *attrs UNUSED)
+radv_rt_notify(struct proto *P, struct channel *ch UNUSED, net *n, rte *new, rte *old UNUSED, ea_list *attrs)
 {
-  struct proto_radv *ra = (struct proto_radv *) p;
+  struct radv_proto *p = (struct radv_proto *) P;
-  struct radv_config *cf = (struct radv_config *) (p->cf);
+  struct radv_config *cf = (struct radv_config *) (P->cf);
+  struct radv_route *rt;
+  eattr *ea;
 
   if (radv_net_match_trigger(cf, n))
   {
-    u8 old_active = ra->active;
+    u8 old_active = p->active;
-    ra->active = !!new;
+    p->active = !!new;
 
-    if (ra->active == old_active)
+    if (p->active == old_active)
       return;
 
-    if (ra->active)
+    if (p->active)
       RADV_TRACE(D_EVENTS, "Triggered");
     else
       RADV_TRACE(D_EVENTS, "Suppressed");
 
-    radv_iface_notify_all(ra, RA_EV_CHANGE);
+    radv_iface_notify_all(p, RA_EV_CHANGE);
+    return;
   }
+
+  if (!cf->propagate_routes)
+    return;
+
+  /*
+   * Some other route we want to send (or stop sending). Update the cache,
+   * with marking a removed one as dead or creating a new one as needed.
+   *
+   * And yes, we exclude the trigger route on purpose.
+   */
+
+  if (new)
+  {
+    /* Update */
+
+    ea = ea_find(attrs, EA_RA_PREFERENCE);
+    uint preference = ea ? ea->u.data : RA_PREF_MEDIUM;
+    uint preference_set = !!ea;
+
+    ea = ea_find(attrs, EA_RA_LIFETIME);
+    uint lifetime = ea ? ea->u.data : 0;
+    uint lifetime_set = !!ea;
+
+    if ((preference != RA_PREF_LOW) &&
+	(preference != RA_PREF_MEDIUM) &&
+	(preference != RA_PREF_HIGH))
+    {
+      log(L_WARN "%s: Invalid ra_preference value %u on route %N",
+	  p->p.name, preference, n->n.addr);
+      preference = RA_PREF_MEDIUM;
+      preference_set = 1;
+      lifetime = 0;
+      lifetime_set = 1;
+    }
+
+    rt = fib_get(&p->routes, n->n.addr);
+
+    /* Ignore update if nothing changed */
+    if (rt->valid &&
+	(rt->preference == preference) &&
+	(rt->preference_set == preference_set) &&
+	(rt->lifetime == lifetime) &&
+	(rt->lifetime_set == lifetime_set))
+      return;
+
+    if (p->routes.entries == 18)
+      log(L_WARN "%s: More than 17 routes exported to RAdv", p->p.name);
+
+    rt->valid = 1;
+    rt->changed = current_time();
+    rt->preference = preference;
+    rt->preference_set = preference_set;
+    rt->lifetime = lifetime;
+    rt->lifetime_set = lifetime_set;
+  }
+  else
+  {
+    /* Withdraw */
+    rt = fib_find(&p->routes, n->n.addr);
+
+    if (!rt || !rt->valid)
+      return;
+
+    /* Invalidate the route */
+    rt->valid = 0;
+    rt->changed = current_time();
+
+    /* Invalidated route will be pruned eventually */
+    btime expires = rt->changed + cf->max_linger_time S;
+    p->prune_time = MIN(p->prune_time, expires);
+  }
+
+  radv_iface_notify_all(p, RA_EV_CHANGE);
+}
+
+/*
+ * Cleans up all the dead routes that expired and schedules itself to be run
+ * again if there are more routes waiting for expiration.
+ */
+static void
+radv_prune_routes(struct radv_proto *p)
+{
+  struct radv_config *cf = (struct radv_config *) (p->p.cf);
+  btime now = current_time();
+  btime next = TIME_INFINITY;
+  btime expires = 0;
+
+  /* Should not happen */
+  if (!p->fib_up)
+    return;
+
+  struct fib_iterator fit;
+  FIB_ITERATE_INIT(&fit, &p->routes);
+
+again:
+  FIB_ITERATE_START(&p->routes, &fit, struct radv_route, rt)
+  {
+    if (!rt->valid)
+    {
+      expires = rt->changed + cf->max_linger_time S;
+
+      /* Delete expired nodes */
+      if (expires <= now)
+      {
+	FIB_ITERATE_PUT(&fit);
+	fib_delete(&p->routes, rt);
+	goto again;
+      }
+      else
+	next = MIN(next, expires);
+    }
+  }
+  FIB_ITERATE_END;
+
+  p->prune_time = next;
 }
 
 static int
-radv_check_active(struct proto_radv *ra)
+radv_check_active(struct radv_proto *p)
 {
-  struct radv_config *cf = (struct radv_config *) (ra->p.cf);
+  struct radv_config *cf = (struct radv_config *) (p->p.cf);
 
   if (!radv_trigger_valid(cf))
     return 1;
 
-  struct channel *c = ra->p.main_channel;
+  struct channel *c = p->p.main_channel;
-  return rt_examine(c->table, &cf->trigger, &ra->p, c->out_filter);
+  return rt_examine(c->table, &cf->trigger, &p->p, c->out_filter);
 }
 
 static void
@@ -328,26 +575,47 @@ radv_postconfig(struct proto_config *CF)
 static struct proto *
 radv_init(struct proto_config *CF)
 {
-  struct proto *p = proto_new(CF);
+  struct proto *P = proto_new(CF);
 
-  p->main_channel = proto_add_channel(p, proto_cf_main_channel(CF));
+  P->main_channel = proto_add_channel(P, proto_cf_main_channel(CF));
 
-  p->import_control = radv_import_control;
+  P->import_control = radv_import_control;
-  p->rt_notify = radv_rt_notify;
+  P->rt_notify = radv_rt_notify;
-  p->if_notify = radv_if_notify;
+  P->if_notify = radv_if_notify;
-  p->ifa_notify = radv_ifa_notify;
+  P->ifa_notify = radv_ifa_notify;
 
-  return p;
+  return P;
+}
+
+static void
+radv_set_fib(struct radv_proto *p, int up)
+{
+  if (up == p->fib_up)
+    return;
+
+  if (up)
+    fib_init(&p->routes, p->p.pool, NET_IP6, sizeof(struct radv_route),
+	     OFFSETOF(struct radv_route, n), 4, NULL);
+  else
+    fib_free(&p->routes);
+
+  p->fib_up = up;
+  p->prune_time = TIME_INFINITY;
 }
 
 static int
-radv_start(struct proto *p)
+radv_start(struct proto *P)
 {
-  struct proto_radv *ra = (struct proto_radv *) p;
+  struct radv_proto *p = (struct radv_proto *) P;
-  struct radv_config *cf = (struct radv_config *) (p->cf);
+  struct radv_config *cf = (struct radv_config *) (P->cf);
 
-  init_list(&(ra->iface_list));
+  init_list(&(p->iface_list));
-  ra->active = !radv_trigger_valid(cf);
+  p->valid = 1;
+  p->active = !radv_trigger_valid(cf);
+
+  p->fib_up = 0;
+  radv_set_fib(p, cf->propagate_routes);
+  p->prune_time = TIME_INFINITY;
 
   return PS_UP;
 }
@@ -356,46 +624,61 @@ static inline void
 radv_iface_shutdown(struct radv_iface *ifa)
 {
   if (ifa->sk)
-    radv_send_ra(ifa, 1);
+  {
+    radv_invalidate(ifa);
+    radv_send_ra(ifa);
+  }
 }
 
 static int
-radv_shutdown(struct proto *p)
+radv_shutdown(struct proto *P)
 {
-  struct proto_radv *ra = (struct proto_radv *) p;
+  struct radv_proto *p = (struct radv_proto *) P;
+
+  p->valid = 0;
 
   struct radv_iface *ifa;
-  WALK_LIST(ifa, ra->iface_list)
+  WALK_LIST(ifa, p->iface_list)
     radv_iface_shutdown(ifa);
 
   return PS_DOWN;
 }
 
 static int
-radv_reconfigure(struct proto *p, struct proto_config *CF)
+radv_reconfigure(struct proto *P, struct proto_config *CF)
 {
-  struct proto_radv *ra = (struct proto_radv *) p;
+  struct radv_proto *p = (struct radv_proto *) P;
-  // struct radv_config *old = (struct radv_config *) (p->cf);
+  struct radv_config *old = (struct radv_config *) (P->cf);
   struct radv_config *new = (struct radv_config *) CF;
 
-  /*
+  if (!proto_configure_channel(P, &P->main_channel, proto_cf_main_channel(CF)))
-   * The question is why there is a reconfigure function for RAdv if
-   * it has almost none internal state so restarting the protocol
-   * would probably suffice. One small reason is that restarting the
-   * protocol would lead to sending a RA with Router Lifetime 0
-   * causing nodes to temporary remove their default routes.
-   */
-
-  if (!proto_configure_channel(p, &p->main_channel, proto_cf_main_channel(CF)))
     return 0;
 
-  p->cf = CF; /* radv_check_active() requires proper p->cf */
+  P->cf = CF; /* radv_check_active() requires proper P->cf */
-  ra->active = radv_check_active(ra);
+  p->active = radv_check_active(p);
+
+  /* Allocate or free FIB */
+  radv_set_fib(p, new->propagate_routes);
+
+  /* We started to accept routes so we need to refeed them */
+  if (!old->propagate_routes && new->propagate_routes)
+    channel_request_feeding(p->p.main_channel);
 
   struct iface *iface;
   WALK_LIST(iface, iface_list)
   {
-    struct radv_iface *ifa = radv_iface_find(ra, iface);
+    if (!(iface->flags & IF_UP))
+      continue;
+
+    /* Ignore non-multicast ifaces */
+    if (!(iface->flags & IF_MULTICAST))
+      continue;
+
+    /* Ignore ifaces without link-local address */
+    if (!iface->llv6)
+      continue;
+
+    struct radv_iface *ifa = radv_iface_find(p, iface);
     struct radv_iface_config *ic = (struct radv_iface_config *)
       iface_patt_find(&new->patt_list, iface, NULL);
 
@@ -415,7 +698,7 @@ radv_reconfigure(struct proto *p, struct proto_config *CF)
     }
 
     if (!ifa && ic)
-      radv_iface_new(ra, iface, ic);
+      radv_iface_new(p, iface, ic);
   }
 
   return 1;
@@ -435,19 +718,53 @@ radv_copy_config(struct proto_config *dest, struct proto_config *src)
 }
 
 static void
-radv_get_status(struct proto *p, byte *buf)
+radv_get_status(struct proto *P, byte *buf)
 {
-  struct proto_radv *ra = (struct proto_radv *) p;
+  struct radv_proto *p = (struct radv_proto *) P;
 
-  if (!ra->active)
+  if (!p->active)
     strcpy(buf, "Suppressed");
 }
 
+static const char *
+radv_pref_str(u32 pref)
+{
+  switch (pref)
+  {
+    case RA_PREF_LOW:
+      return "low";
+    case RA_PREF_MEDIUM:
+      return "medium";
+    case RA_PREF_HIGH:
+      return "high";
+    default:
+      return "??";
+  }
+}
+
+/* The buffer has some minimal size */
+static int
+radv_get_attr(eattr *a, byte *buf, int buflen UNUSED)
+{
+  switch (a->id)
+  {
+  case EA_RA_PREFERENCE:
+    bsprintf(buf, "preference: %s", radv_pref_str(a->u.data));
+    return GA_FULL;
+  case EA_RA_LIFETIME:
+    bsprintf(buf, "lifetime");
+    return GA_NAME;
+  default:
+    return GA_UNKNOWN;
+  }
+}
+
 struct protocol proto_radv = {
   .name =		"RAdv",
   .template =		"radv%d",
+  .attr_class =		EAP_RADV,
   .channel_mask =	NB_IP6,
-  .proto_size =		sizeof(struct proto_radv),
+  .proto_size =		sizeof(struct radv_proto),
   .config_size =	sizeof(struct radv_config),
   .postconfig =		radv_postconfig,
   .init =		radv_init,
@@ -455,5 +772,6 @@ struct protocol proto_radv = {
   .shutdown =		radv_shutdown,
   .reconfigure =	radv_reconfigure,
   .copy_config =	radv_copy_config,
-  .get_status =		radv_get_status
+  .get_status =		radv_get_status,
+  .get_attr =		radv_get_attr
 };

proto/radv/radv.h

@@ -13,7 +13,7 @@
 #include "lib/ip.h"
 #include "lib/lists.h"
 #include "lib/socket.h"
-#include "sysdep/unix/timer.h"
+#include "lib/timer.h"
 #include "lib/resource.h"
 #include "nest/protocol.h"
 #include "nest/iface.h"
@@ -30,7 +30,7 @@
 #define ICMPV6_RA 134
 
 #define MAX_INITIAL_RTR_ADVERTISEMENTS 3
-#define MAX_INITIAL_RTR_ADVERT_INTERVAL 16
+#define MAX_INITIAL_RTR_ADVERT_INTERVAL (16 S_)
 
 #define DEFAULT_MAX_RA_INT 600
 #define DEFAULT_MIN_DELAY 3
@@ -51,6 +51,8 @@ struct radv_config
   list dnssl_list;		/* Global list of DNSSL configs (struct radv_dnssl_config) */
 
   net_addr trigger;		/* Prefix of a trigger route, if defined */
+  u8 propagate_routes;		/* Do we propagate more specific routes (RFC 4191)? */
+  u32 max_linger_time;		/* Maximum of interface route_linger_time */
 };
 
 struct radv_iface_config
@@ -60,22 +62,28 @@ struct radv_iface_config
   list rdnss_list;		/* Local list of RDNSS configs (struct radv_rdnss_config) */
   list dnssl_list;		/* Local list of DNSSL configs (struct radv_dnssl_config) */
 
-  u32 min_ra_int;		/* Standard options from RFC 4261 */
+  u32 min_ra_int;		/* Standard options from RFC 4861 */
   u32 max_ra_int;
   u32 min_delay;
 
+  u32 prefix_linger_time;	/* How long we advertise dead prefixes with lifetime 0 */
+  u32 route_linger_time;	/* How long we advertise dead routes with lifetime 0 */
+
   u8 rdnss_local;		/* Global list is not used for RDNSS */
   u8 dnssl_local;		/* Global list is not used for DNSSL */
 
-  u8 managed;			/* Standard options from RFC 4261 */
+  u8 managed;			/* Standard options from RFC 4861 */
   u8 other_config;
   u32 link_mtu;
   u32 reachable_time;
   u32 retrans_timer;
   u32 current_hop_limit;
   u32 default_lifetime;
+  u32 route_lifetime;		/* Lifetime for the RFC 4191 routes */
   u8 default_lifetime_sensitive; /* Whether default_lifetime depends on trigger */
+  u8 route_lifetime_sensitive;	/* Whether route_lifetime depends on trigger */
   u8 default_preference;	/* Default Router Preference (RFC 4191) */
+  u8 route_preference;		/* Specific Route Preference (RFC 4191) */
 };
 
 struct radv_prefix_config
@@ -84,7 +92,7 @@ struct radv_prefix_config
   net_addr_ip6 prefix;
 
   u8 skip;			/* Do not include this prefix to RA */
-  u8 onlink;			/* Standard options from RFC 4261 */
+  u8 onlink;			/* Standard options from RFC 4861 */
   u8 autonomous;
   u32 valid_lifetime;
   u32 preferred_lifetime;
@@ -110,29 +118,67 @@ struct radv_dnssl_config
   char *domain;			/* Domain for DNS search list, in processed form */
 };
 
+/*
+ * One more specific route as per RFC 4191.
+ *
+ * Note that it does *not* contain the next hop field. The next hop is always
+ * the router sending the advertisment and the more specific route only allows
+ * overriding the preference of the route.
+ */
+struct radv_route
+{
+  u32 lifetime;			/* Lifetime from an attribute */
+  u8 lifetime_set;		/* Whether lifetime is defined */
+  u8 preference;		/* Preference of the route, RA_PREF_* */
+  u8 preference_set;		/* Whether preference is defined */
+  u8 valid;			/* Whethe route is valid or withdrawn */
+  btime changed;		/* Last time when the route changed */
 
-struct proto_radv
+  struct fib_node n;
+};
+
+struct radv_proto
 {
   struct proto p;
   list iface_list;		/* List of active ifaces */
+  u8 valid;			/* Router is valid for forwarding, used for shutdown */
   u8 active;			/* Whether radv is active w.r.t. triggers */
+  u8 fib_up;			/* FIB table (routes) is initialized */
+  struct fib routes;		/* FIB table of specific routes (struct radv_route) */
+  btime prune_time;		/* Next time of route table pruning */
+};
+
+struct radv_prefix		/* One prefix we advertise */
+{
+  node n;
+  net_addr_ip6 prefix;
+
+  u8 valid;			/* Is the prefix valid? If not, we advertise it
+				   with 0 lifetime, so clients stop using it */
+  u8 mark;			/* A temporary mark for processing */
+  btime changed;		/* Last time when the prefix changed */
+  struct radv_prefix_config *cf; /* The config tied to this prefix */
 };
 
 struct radv_iface
 {
   node n;
-  struct proto_radv *ra;
+  struct radv_proto *ra;
   struct radv_iface_config *cf;	/* Related config, must be updated in reconfigure */
   struct iface *iface;
   struct ifa *addr;		/* Link-local address of iface */
+  struct pool *pool;		/* A pool for interface-specific things */
+  list prefixes;		/* The prefixes we advertise (struct radv_prefix) */
+  btime prune_time;		/* Next time of prefix list pruning */
+  btime valid_time;		/* Cached packet is valid until first linger timeout */
 
   timer *timer;
   struct object_lock *lock;
   sock *sk;
 
-  bird_clock_t last;		/* Time of last sending of RA */
+  btime last;			/* Time of last sending of RA */
   u16 plen;			/* Length of prepared RA in tbuf, or 0 if not valid */
-  byte initial;			/* List of active ifaces */
+  byte initial;			/* How many RAs are still to be sent as initial */
 };
 
 #define RA_EV_INIT 1		/* Switch to initial mode */
@@ -145,14 +191,17 @@ struct radv_iface
 #define RA_PREF_HIGH	0x08
 #define RA_PREF_MASK	0x18
 
+/* Attributes */
+#define EA_RA_PREFERENCE	EA_CODE(EAP_RADV, 0)
+#define EA_RA_LIFETIME		EA_CODE(EAP_RADV, 1)
 
 #ifdef LOCAL_DEBUG
 #define RADV_FORCE_DEBUG 1
 #else
 #define RADV_FORCE_DEBUG 0
 #endif
-#define RADV_TRACE(flags, msg, args...) do { if ((ra->p.debug & flags) || RADV_FORCE_DEBUG) \
+#define RADV_TRACE(flags, msg, args...) do { if ((p->p.debug & flags) || RADV_FORCE_DEBUG) \
-	log(L_TRACE "%s: " msg, ra->p.name , ## args ); } while(0)
+	log(L_TRACE "%s: " msg, p->p.name , ## args ); } while(0)
 
 
 /* radv.c */
@@ -160,7 +209,7 @@ void radv_iface_notify(struct radv_iface *ifa, int event);
 
 /* packets.c */
 int radv_process_domain(struct radv_dnssl_config *cf);
-void radv_send_ra(struct radv_iface *ifa, int shutdown);
+void radv_send_ra(struct radv_iface *ifa);
 int radv_sk_open(struct radv_iface *ifa);
 
 

proto/rip/config.Y

@@ -56,9 +56,10 @@ rip_proto_start: proto_start rip_variant
 
   init_list(&RIP_CFG->patt_list);
   RIP_CFG->rip2 = $2;
+  RIP_CFG->ecmp = rt_default_ecmp;
   RIP_CFG->infinity = RIP_DEFAULT_INFINITY;
-  RIP_CFG->min_timeout_time = 60;
+  RIP_CFG->min_timeout_time = 60 S_;
-  RIP_CFG->max_garbage_time = 60;
+  RIP_CFG->max_garbage_time = 60 S_;
 };
 
 rip_proto_item:
@@ -92,6 +93,7 @@ rip_iface_start:
   RIP_IFACE->split_horizon = 1;
   RIP_IFACE->poison_reverse = 1;
   RIP_IFACE->check_zero = 1;
+  RIP_IFACE->check_link = 1;
   RIP_IFACE->ttl_security = rip_cfg_is_v2() ? 0 : 1;
   RIP_IFACE->rx_buffer = rip_cfg_is_v2() ? RIP_MAX_PKT_LENGTH : 0;
   RIP_IFACE->tx_length = rip_cfg_is_v2() ? RIP_MAX_PKT_LENGTH : 0;
@@ -147,9 +149,9 @@ rip_iface_item:
  | SPLIT HORIZON bool	{ RIP_IFACE->split_horizon = $3; }
  | POISON REVERSE bool	{ RIP_IFACE->poison_reverse = $3; }
  | CHECK ZERO bool	{ RIP_IFACE->check_zero = $3; }
- | UPDATE TIME expr	{ RIP_IFACE->update_time = $3; if ($3<=0) cf_error("Update time must be positive"); }
+ | UPDATE TIME expr	{ RIP_IFACE->update_time = $3 S_; if ($3<=0) cf_error("Update time must be positive"); }
- | TIMEOUT TIME expr	{ RIP_IFACE->timeout_time = $3; if ($3<=0) cf_error("Timeout time must be positive"); }
+ | TIMEOUT TIME expr	{ RIP_IFACE->timeout_time = $3 S_; if ($3<=0) cf_error("Timeout time must be positive"); }
- | GARBAGE TIME expr	{ RIP_IFACE->garbage_time = $3; if ($3<=0) cf_error("Garbage time must be positive"); }
+ | GARBAGE TIME expr	{ RIP_IFACE->garbage_time = $3 S_; if ($3<=0) cf_error("Garbage time must be positive"); }
  | ECMP WEIGHT expr	{ RIP_IFACE->ecmp_weight = $3 - 1; if (($3<1) || ($3>256)) cf_error("ECMP weight must be in range 1-256"); }
  | RX BUFFER expr	{ RIP_IFACE->rx_buffer = $3; if (($3<256) || ($3>65535)) cf_error("RX length must be in range 256-65535"); }
  | TX LENGTH expr	{ RIP_IFACE->tx_length = $3; if (($3<256) || ($3>65535)) cf_error("TX length must be in range 256-65535"); }

proto/rip/packets.c

@@ -189,7 +189,10 @@ rip_update_csn(struct rip_proto *p UNUSED, struct rip_iface *ifa)
    * have the same CSN. We are using real time, but enforcing monotonicity.
    */
   if (ifa->cf->auth_type == RIP_AUTH_CRYPTO)
-    ifa->csn = (ifa->csn < (u32) now_real) ? (u32) now_real : ifa->csn + 1;
+  {
+    u32 now_real = (u32) (current_real_time() TO_S);
+    ifa->csn = (ifa->csn < now_real) ? now_real : ifa->csn + 1;
+  }
 }
 
 static void
@@ -434,6 +437,7 @@ rip_send_response(struct rip_proto *p, struct rip_iface *ifa)
   byte *max = rip_tx_buffer(ifa) + ifa->tx_plen -
     (rip_is_v2(p) ? RIP_BLOCK_LENGTH : 2*RIP_BLOCK_LENGTH);
   ip_addr last_next_hop = IPA_NONE;
+  btime now_ = current_time();
   int send = 0;
 
   struct rip_packet *pkt = (void *) pos;
@@ -450,7 +454,7 @@ rip_send_response(struct rip_proto *p, struct rip_iface *ifa)
 
     /* Stale entries that should be removed */
     if ((en->valid == RIP_ENTRY_STALE) &&
-	((en->changed + (bird_clock_t) ifa->cf->garbage_time) <= now))
+	((en->changed + ifa->cf->garbage_time) <= now_))
       goto next_entry;
 
     /* Triggered updates */
@@ -540,7 +544,7 @@ break_loop:
  * activating the new one.
  */
 void
-rip_send_table(struct rip_proto *p, struct rip_iface *ifa, ip_addr addr, bird_clock_t changed)
+rip_send_table(struct rip_proto *p, struct rip_iface *ifa, ip_addr addr, btime changed)
 {
   DBG("RIP: Opening TX session to %I on %s\n", addr, ifa->iface->name);
 
@@ -591,6 +595,7 @@ rip_receive_response(struct rip_proto *p, struct rip_iface *ifa, struct rip_pack
 
   byte *pos = (byte *) pkt + sizeof(struct rip_packet);
   byte *end = (byte *) pkt + plen;
+  btime now_ = current_time();
 
   for (; pos < end; pos += RIP_BLOCK_LENGTH)
   {
@@ -638,7 +643,7 @@ rip_receive_response(struct rip_proto *p, struct rip_iface *ifa, struct rip_pack
 	.next_hop = ipa_nonzero(rte.next_hop) ? rte.next_hop : from->nbr->addr,
 	.metric = rte.metric,
 	.tag = rte.tag,
-	.expires = now + ifa->cf->timeout_time
+	.expires = now_ + ifa->cf->timeout_time
       };
 
       rip_update_rte(p, &rte.net, &new);
@@ -669,8 +674,7 @@ rip_rx_hook(sock *sk, uint len)
       sk->iface->name, sk->faddr, sk->laddr);
 
   /* Silently ignore my own packets */
-  /* FIXME: Better local address check */
+  if (ipa_equal(sk->faddr, sk->saddr))
-  if (ipa_equal(ifa->iface->addr->ip, sk->faddr))
     return 1;
 
   if (rip_is_ng(p) && !ipa_is_link_local(sk->faddr))
@@ -706,7 +710,7 @@ rip_rx_hook(sock *sk, uint len)
   if ((plen - sizeof(struct rip_packet)) % RIP_BLOCK_LENGTH)
     DROP("invalid length", plen);
 
-  n->last_seen = now;
+  n->last_seen = current_time();
   rip_update_bfd(p, n);
 
   switch (pkt->command)
@@ -742,14 +746,8 @@ rip_open_socket(struct rip_iface *ifa)
   sk->sport = ifa->cf->port;
   sk->dport = ifa->cf->port;
   sk->iface = ifa->iface;
-
+  sk->saddr = rip_is_v2(p) ? ifa->iface->addr4->ip : ifa->iface->llv6->ip;
-  /*
+  sk->vrf = p->p.vrf;
-   * For RIPv2, we explicitly choose a primary address, mainly to ensure that
-   * RIP and BFD uses the same one. For RIPng, we left it to kernel, which
-   * should choose some link-local address based on the same scope rule.
-   */
-  if (rip_is_v2(p))
-    sk->saddr = ifa->iface->addr->ip;
 
   sk->rx_hook = rip_rx_hook;
   sk->tx_hook = rip_tx_hook;

proto/rip/rip.c

@@ -364,7 +364,7 @@ rip_rt_notify(struct proto *P, struct channel *ch UNUSED, struct network *net, s
   /* Activate triggered updates */
   if (en->metric != old_metric)
   {
-    en->changed = now;
+    en->changed = current_time();
     rip_trigger_update(p);
   }
 }
@@ -506,10 +506,10 @@ rip_iface_start(struct rip_iface *ifa)
 
   TRACE(D_EVENTS, "Starting interface %s", ifa->iface->name);
 
-  ifa->next_regular = now + (random() % ifa->cf->update_time) + 1;
+  ifa->next_regular = current_time() + (random() % ifa->cf->update_time) + 100 MS;
-  ifa->next_triggered = now;	/* Available immediately */
+  ifa->next_triggered = current_time();	/* Available immediately */
-  ifa->want_triggered = 1;	/* All routes in triggered update */
+  ifa->want_triggered = 1;		/* All routes in triggered update */
-  tm_start(ifa->timer, 1);	/* Or 100 ms */
+  tm_start(ifa->timer, 100 MS);
   ifa->up = 1;
 
   if (!ifa->cf->passive)
@@ -630,13 +630,19 @@ rip_add_iface(struct rip_proto *p, struct iface *iface, struct rip_iface_config
   else if (ic->mode == RIP_IM_MULTICAST)
     ifa->addr = rip_is_v2(p) ? IP4_RIP_ROUTERS : IP6_RIP_ROUTERS;
   else /* Broadcast */
-    ifa->addr = iface->addr->brd;
+    ifa->addr = iface->addr4->brd;
+  /*
+   * The above is just a workaround for BSD as it can't send broadcasts
+   * to 255.255.255.255. BSD systems need the network broadcast address instead.
+   *
+   * TODO: move this to sysdep code
+   */
 
   init_list(&ifa->neigh_list);
 
   add_tail(&p->iface_list, NODE ifa);
 
-  ifa->timer = tm_new_set(p->p.pool, rip_iface_timer, ifa, 0, 0);
+  ifa->timer = tm_new_init(p->p.pool, rip_iface_timer, ifa, 0, 0);
 
   struct object_lock *lock = olock_new(p->p.pool);
   lock->type = OBJLOCK_UDP;
@@ -684,8 +690,8 @@ rip_reconfigure_iface(struct rip_proto *p, struct rip_iface *ifa, struct rip_ifa
 
   rip_iface_update_buffers(ifa);
 
-  if (ifa->next_regular > (now + (bird_clock_t) new->update_time))
+  if (ifa->next_regular > (current_time() + new->update_time))
-    ifa->next_regular = now + (random() % new->update_time) + 1;
+    ifa->next_regular = current_time() + (random() % new->update_time) + 100 MS;
 
   if (new->check_link != old->check_link)
     rip_iface_update_state(ifa);
@@ -706,7 +712,11 @@ rip_reconfigure_ifaces(struct rip_proto *p, struct rip_config *cf)
 
   WALK_LIST(iface, iface_list)
   {
-    if (! (iface->flags & IF_UP))
+    if (!(iface->flags & IF_UP))
+      continue;
+
+    /* Ignore ifaces without appropriate address */
+    if (rip_is_v2(p) ? !iface->addr4 : !iface->llv6)
       continue;
 
     struct rip_iface *ifa = rip_find_iface(p, iface);
@@ -744,6 +754,10 @@ rip_if_notify(struct proto *P, unsigned flags, struct iface *iface)
   {
     struct rip_iface_config *ic = (void *) iface_patt_find(&cf->patt_list, iface, NULL);
 
+    /* Ignore ifaces without appropriate address */
+    if (rip_is_v2(p) ? !iface->addr4 : !iface->llv6)
+      return;
+
     if (ic)
       rip_add_iface(p, iface, ic);
 
@@ -802,8 +816,9 @@ rip_timer(timer *t)
   struct rip_iface *ifa;
   struct rip_neighbor *n, *nn;
   struct fib_iterator fit;
-  bird_clock_t next = now + MIN(cf->min_timeout_time, cf->max_garbage_time);
+  btime now_ = current_time();
-  bird_clock_t expires = 0;
+  btime next = now_ + MIN(cf->min_timeout_time, cf->max_garbage_time);
+  btime expires = 0;
 
   TRACE(D_EVENTS, "Main timer fired");
 
@@ -818,7 +833,7 @@ rip_timer(timer *t)
     /* Checking received routes for timeout and for dead neighbors */
     for (rp = &en->routes; rt = *rp; /* rp = &rt->next */)
     {
-      if (!rip_valid_rte(rt) || (rt->expires <= now))
+      if (!rip_valid_rte(rt) || (rt->expires <= now_))
       {
 	rip_remove_rte(p, rp);
 	changed = 1;
@@ -848,7 +863,7 @@ rip_timer(timer *t)
     {
       expires = en->changed + cf->max_garbage_time;
 
-      if (expires <= now)
+      if (expires <= now_)
       {
 	// TRACE(D_EVENTS, "entry is too old: %N", en->n.addr);
 	en->valid = 0;
@@ -876,20 +891,20 @@ rip_timer(timer *t)
       {
 	expires = n->last_seen + n->ifa->cf->timeout_time;
 
-	if (expires <= now)
+	if (expires <= now_)
 	  rip_remove_neighbor(p, n);
 	else
 	  next = MIN(next, expires);
       }
 
-  tm_start(p->timer, MAX(next - now, 1));
+  tm_start(p->timer, MAX(next - now_, 100 MS));
 }
 
 static inline void
 rip_kick_timer(struct rip_proto *p)
 {
-  if (p->timer->expires > (now + 1))
+  if (p->timer->expires > (current_time() + 100 MS))
-    tm_start(p->timer, 1);	/* Or 100 ms */
+    tm_start(p->timer, 100 MS);
 }
 
 /**
@@ -907,7 +922,8 @@ rip_iface_timer(timer *t)
 {
   struct rip_iface *ifa = t->data;
   struct rip_proto *p = ifa->rip;
-  bird_clock_t period = ifa->cf->update_time;
+  btime now_ = current_time();
+  btime period = ifa->cf->update_time;
 
   if (ifa->cf->passive)
     return;
@@ -916,40 +932,40 @@ rip_iface_timer(timer *t)
 
   if (ifa->tx_active)
   {
-    if (now < (ifa->next_regular + period))
+    if (now_ < (ifa->next_regular + period))
-      { tm_start(ifa->timer, 1); return; }
+    { tm_start(ifa->timer, 100 MS); return; }
 
     /* We are too late, reset is done by rip_send_table() */
     log(L_WARN "%s: Too slow update on %s, resetting", p->p.name, ifa->iface->name);
   }
 
-  if (now >= ifa->next_regular)
+  if (now_ >= ifa->next_regular)
   {
     /* Send regular update, set timer for next period (or following one if necessay) */
     TRACE(D_EVENTS, "Sending regular updates for %s", ifa->iface->name);
     rip_send_table(p, ifa, ifa->addr, 0);
-    ifa->next_regular += period * (1 + ((now - ifa->next_regular) / period));
+    ifa->next_regular += period * (1 + ((now_ - ifa->next_regular) / period));
     ifa->want_triggered = 0;
     p->triggered = 0;
   }
-  else if (ifa->want_triggered && (now >= ifa->next_triggered))
+  else if (ifa->want_triggered && (now_ >= ifa->next_triggered))
   {
     /* Send triggered update, enforce interval between triggered updates */
     TRACE(D_EVENTS, "Sending triggered updates for %s", ifa->iface->name);
     rip_send_table(p, ifa, ifa->addr, ifa->want_triggered);
-    ifa->next_triggered = now + MIN(5, period / 2 + 1);
+    ifa->next_triggered = now_ + MIN(5 S, period / 2);
     ifa->want_triggered = 0;
     p->triggered = 0;
   }
 
-  tm_start(ifa->timer, ifa->want_triggered ? 1 : (ifa->next_regular - now));
+  tm_start(ifa->timer, ifa->want_triggered ? (1 S) : (ifa->next_regular - now_));
 }
 
 static inline void
 rip_iface_kick_timer(struct rip_iface *ifa)
 {
-  if (ifa->timer->expires > (now + 1))
+  if (ifa->timer->expires > (current_time() + 100 MS))
-    tm_start(ifa->timer, 1);	/* Or 100 ms */
+    tm_start(ifa->timer, 100 MS);
 }
 
 static void
@@ -970,7 +986,7 @@ rip_trigger_update(struct rip_proto *p)
       continue;
 
     TRACE(D_EVENTS, "Scheduling triggered updates for %s", ifa->iface->name);
-    ifa->want_triggered = now;
+    ifa->want_triggered = current_time();
     rip_iface_kick_timer(ifa);
   }
 
@@ -1095,7 +1111,7 @@ rip_start(struct proto *P)
   fib_init(&p->rtable, P->pool, cf->rip2 ? NET_IP4 : NET_IP6,
 	   sizeof(struct rip_entry), OFFSETOF(struct rip_entry, n), 0, NULL);
   p->rte_slab = sl_new(P->pool, sizeof(struct rip_rte));
-  p->timer = tm_new_set(P->pool, rip_timer, p, 0, 0);
+  p->timer = tm_new_init(P->pool, rip_timer, p, 0, 0);
 
   p->rip2 = cf->rip2;
   p->ecmp = cf->ecmp;
@@ -1180,7 +1196,7 @@ rip_show_interfaces(struct proto *P, char *iff)
   }
 
   cli_msg(-1021, "%s:", p->p.name);
-  cli_msg(-1021, "%-10s %-6s %6s %6s %6s",
+  cli_msg(-1021, "%-10s %-6s %6s %6s %7s",
 	  "Interface", "State", "Metric", "Nbrs", "Timer");
 
   WALK_LIST(ifa, p->iface_list)
@@ -1193,8 +1209,9 @@ rip_show_interfaces(struct proto *P, char *iff)
       if (n->last_seen)
 	nbrs++;
 
-    int timer = MAX(ifa->next_regular - now, 0);
+    btime now_ = current_time();
-    cli_msg(-1021, "%-10s %-6s %6u %6u %6u",
+    btime timer = (ifa->next_regular > now_) ? (ifa->next_regular - now_) : 0;
+    cli_msg(-1021, "%-10s %-6s %6u %6u %7t",
 	    ifa->iface->name, (ifa->up ? "Up" : "Down"), ifa->cf->metric, nbrs, timer);
   }
 
@@ -1216,7 +1233,7 @@ rip_show_neighbors(struct proto *P, char *iff)
   }
 
   cli_msg(-1022, "%s:", p->p.name);
-  cli_msg(-1022, "%-25s %-10s %6s %6s %6s",
+  cli_msg(-1022, "%-25s %-10s %6s %6s %7s",
 	  "IP address", "Interface", "Metric", "Routes", "Seen");
 
   WALK_LIST(ifa, p->iface_list)
@@ -1229,8 +1246,8 @@ rip_show_neighbors(struct proto *P, char *iff)
       if (!n->last_seen)
 	continue;
 
-      int timer = now - n->last_seen;
+      btime timer = current_time() - n->last_seen;
-      cli_msg(-1022, "%-25I %-10s %6u %6u %6u",
+      cli_msg(-1022, "%-25I %-10s %6u %6u %7t",
 	      n->nbr->addr, ifa->iface->name, ifa->cf->metric, n->uc, timer);
     }
   }
@@ -1248,9 +1265,9 @@ rip_dump(struct proto *P)
   i = 0;
   FIB_WALK(&p->rtable, struct rip_entry, en)
   {
-    debug("RIP: entry #%d: %N via %I dev %s valid %d metric %d age %d s\n",
+    debug("RIP: entry #%d: %N via %I dev %s valid %d metric %d age %t\n",
 	  i++, en->n.addr, en->next_hop, en->iface->name,
-	  en->valid, en->metric, now - en->changed);
+	  en->valid, en->metric, current_time() - en->changed);
   }
   FIB_WALK_END;
 

proto/rip/rip.h

@@ -24,7 +24,7 @@
 #include "lib/resource.h"
 #include "lib/socket.h"
 #include "lib/string.h"
-#include "sysdep/unix/timer.h"
+#include "lib/timer.h"
 
 
 #define RIP_V1			1
@@ -38,9 +38,9 @@
 
 #define RIP_DEFAULT_ECMP_LIMIT	16
 #define RIP_DEFAULT_INFINITY	16
-#define RIP_DEFAULT_UPDATE_TIME	30
+#define RIP_DEFAULT_UPDATE_TIME	  (30 S_)
-#define RIP_DEFAULT_TIMEOUT_TIME 180
+#define RIP_DEFAULT_TIMEOUT_TIME (180 S_)
-#define RIP_DEFAULT_GARBAGE_TIME 120
+#define RIP_DEFAULT_GARBAGE_TIME (120 S_)
 
 
 struct rip_config
@@ -52,8 +52,8 @@ struct rip_config
   u8 ecmp;				/* Maximum number of nexthops in ECMP route, or 0 */
   u8 infinity;				/* Maximum metric value, representing infinity */
 
-  u32 min_timeout_time;			/* Minimum of interface timeout_time */
+  btime min_timeout_time;		/* Minimum of interface timeout_time */
-  u32 max_garbage_time;			/* Maximum of interface garbage_time */
+  btime max_garbage_time;		/* Maximum of interface garbage_time */
 };
 
 struct rip_iface_config
@@ -78,9 +78,9 @@ struct rip_iface_config
   u16 tx_length;			/* TX packet length limit (including headers), 0 for MTU */
   int tx_tos;
   int tx_priority;
-  u32 update_time;			/* Periodic update interval */
+  btime update_time;			/* Periodic update interval */
-  u32 timeout_time;			/* Route expiration timeout */
+  btime timeout_time;			/* Route expiration timeout */
-  u32 garbage_time;			/* Unreachable entry GC timeout */
+  btime garbage_time;			/* Unreachable entry GC timeout */
   list *passwords;			/* Passwords for authentication */
 };
 
@@ -120,14 +120,14 @@ struct rip_iface
   list neigh_list;			/* List of iface neighbors (struct rip_neighbor) */
 
   /* Update scheduling */
-  bird_clock_t next_regular;		/* Next time when regular update should be called */
+  btime next_regular;			/* Next time when regular update should be called */
-  bird_clock_t next_triggered;		/* Next time when triggerd update may be called */
+  btime next_triggered;			/* Next time when triggerd update may be called */
-  bird_clock_t want_triggered;		/* Nonzero if triggered update is scheduled */
+  btime want_triggered;			/* Nonzero if triggered update is scheduled */
 
   /* Active update */
   int tx_active;			/* Update session is active */
   ip_addr tx_addr;			/* Update session destination address */
-  bird_clock_t tx_changed;		/* Minimal changed time for triggered update */
+  btime tx_changed;			/* Minimal changed time for triggered update */
   struct fib_iterator tx_fit;		/* FIB iterator in RIP routing table (p.rtable) */
 };
 
@@ -137,7 +137,7 @@ struct rip_neighbor
   struct rip_iface *ifa;		/* Associated interface, may be NULL if stale */
   struct neighbor *nbr;			/* Associaded core neighbor, may be NULL if stale */
   struct bfd_request *bfd_req;		/* BFD request, if BFD is used */
-  bird_clock_t last_seen;		/* Time of last received and accepted message */
+  btime last_seen;			/* Time of last received and accepted message */
   u32 uc;				/* Use count, number of routes linking the neighbor */
   u32 csn;				/* Last received crypto sequence number */
 };
@@ -153,7 +153,7 @@ struct rip_entry
   struct iface *iface;			/* Outgoing route iface (for next hop) */
   ip_addr next_hop;			/* Outgoing route next hop */
 
-  bird_clock_t changed;			/* Last time when the outgoing route metric changed */
+  btime changed;			/* Last time when the outgoing route metric changed */
 
   struct fib_node n;
 };
@@ -167,7 +167,7 @@ struct rip_rte
   u16 metric;				/* Route metric (after increase) */
   u16 tag;				/* Route tag */
 
-  bird_clock_t expires;			/* Time of route expiration */
+  btime expires;			/* Time of route expiration */
 };
 
 
@@ -211,7 +211,7 @@ void rip_show_neighbors(struct proto *P, char *iff);
 
 /* packets.c */
 void rip_send_request(struct rip_proto *p, struct rip_iface *ifa);
-void rip_send_table(struct rip_proto *p, struct rip_iface *ifa, ip_addr addr, bird_clock_t changed);
+void rip_send_table(struct rip_proto *p, struct rip_iface *ifa, ip_addr addr, btime changed);
 int rip_open_socket(struct rip_iface *ifa);