From a8a3d95be5db1a8a7d5a17e2eb8e233417b1d8c7 Mon Sep 17 00:00:00 2001
From: Ondrej Zajicek <santiago@crfreenet.org>
Date: Sat, 4 Jun 2022 17:34:57 +0200
Subject: [PATCH 1/5] Nest: Improve GC strategy for rtables

Use timer (configurable as 'gc period') to schedule routing table
GC/pruning to ensure that prune is done on time but not too often.

Randomize GC timers to avoid concentration of GC events from different
tables in one loop cycle.

Fix a bug that caused minimum inter-GC interval be 5 us instead of 5 s.

Make default 'gc period' adaptive based on number of routing tables,
from 10 s for small setups to 600 s for large ones.

In marge multi-table RS setup, the patch improved time of flushing
a downed peer from 20-30 min to <2 min and removed 40s latencies.
---
 conf/conf.c     |  1 +
 doc/bird.sgml   | 15 ++++++++++++++
 nest/config.Y   |  4 +++-
 nest/route.h    |  8 +++++---
 nest/rt-table.c | 54 +++++++++++++++++++++++++++++++++++++++++--------
 5 files changed, 70 insertions(+), 12 deletions(-)

diff --git a/conf/conf.c b/conf/conf.c
index a2b01667..025c040e 100644
--- a/conf/conf.c
+++ b/conf/conf.c
@@ -140,6 +140,7 @@ config_parse(struct config *c)
   protos_preconfig(c);
   rt_preconfig(c);
   cf_parse();
+  rt_postconfig(c);
 
   if (EMPTY_LIST(c->protos))
     cf_error("No protocol is specified in the config file");
diff --git a/doc/bird.sgml b/doc/bird.sgml
index 1580facd..326fc7a8 100644
--- a/doc/bird.sgml
+++ b/doc/bird.sgml
@@ -684,6 +684,21 @@ to set options.
 	limit to the settle time from the initial ROA table change even if
 	there are consecutive updates gradually renewing the settle time.
 	Default: 20 s.
+
+	<tag><label id="rtable-gc-threshold">gc threshold <m/number/</tag>
+	Specify a minimum amount of removed networks that triggers a garbage
+	collection (GC) cycle. Default: 1000.
+
+	<tag><label id="rtable-gc-period">gc period <m/time/</tag>
+	Specify a period of time between consecutive GC cycles. When there is a
+	significant amount of route withdraws, GC cycles are executed repeatedly
+	with given period time (with some random factor). When there is just
+	small amount of changes, GC cycles are not executed. In extensive route
+	server setups, running GC on hundreds of full BGP routing tables can
+	take significant amount of time, therefore they should use higher GC
+	periods. Default: adaptive, based on number of routing tables in the
+	configuration. From 10 s (with <= 25 routing tables) up to 600 s (with
+	>= 1500 routing tables).
 </descrip>
 
 
diff --git a/nest/config.Y b/nest/config.Y
index 72bc7930..17999422 100644
--- a/nest/config.Y
+++ b/nest/config.Y
@@ -125,7 +125,7 @@ CF_KEYWORDS(TIMEFORMAT, ISO, SHORT, LONG, ROUTE, PROTOCOL, BASE, LOG, S, MS, US)
 CF_KEYWORDS(GRACEFUL, RESTART, WAIT, MAX, FLUSH, AS)
 CF_KEYWORDS(MIN, IDLE, RX, TX, INTERVAL, MULTIPLIER, PASSIVE)
 CF_KEYWORDS(CHECK, LINK)
-CF_KEYWORDS(SORTED, TRIE, MIN, MAX, SETTLE, TIME)
+CF_KEYWORDS(SORTED, TRIE, MIN, MAX, SETTLE, TIME, GC, THRESHOLD, PERIOD)
 
 /* For r_args_channel */
 CF_KEYWORDS(IPV4, IPV4_MC, IPV4_MPLS, IPV6, IPV6_MC, IPV6_MPLS, IPV6_SADR, VPN4, VPN4_MC, VPN4_MPLS, VPN6, VPN6_MC, VPN6_MPLS, ROA4, ROA6, FLOW4, FLOW6, MPLS, PRI, SEC)
@@ -229,6 +229,8 @@ table_opt:
    }
  | MIN SETTLE TIME expr_us { this_table->min_settle_time = $4; }
  | MAX SETTLE TIME expr_us { this_table->max_settle_time = $4; }
+ | GC THRESHOLD expr { this_table->gc_threshold = $3; }
+ | GC PERIOD expr_us { this_table->gc_period = (uint) $3; if ($3 > 3600 S_) cf_error("GC period must be at most 3600 s"); }
  ;
 
 table_opts:
diff --git a/nest/route.h b/nest/route.h
index 7930058a..395139a3 100644
--- a/nest/route.h
+++ b/nest/route.h
@@ -148,8 +148,8 @@ struct rtable_config {
   struct rtable *table;
   struct proto_config *krt_attached;	/* Kernel syncer attached to this table */
   uint addr_type;			/* Type of address data stored in table (NET_*) */
-  int gc_max_ops;			/* Maximum number of operations before GC is run */
-  int gc_min_time;			/* Minimum time between two consecutive GC runs */
+  uint gc_threshold;			/* Maximum number of operations before GC is run */
+  uint gc_period;			/* Approximate time between two consecutive GC runs */
   byte sorted;				/* Routes of network are sorted according to rte_better() */
   byte internal;			/* Internal table of a protocol */
   byte trie_used;			/* Rtable has attached trie */
@@ -180,10 +180,11 @@ typedef struct rtable {
 					 * obstacle from this routing table.
 					 */
   struct event *rt_event;		/* Routing table event */
+  struct timer *prune_timer;		/* Timer for periodic pruning / GC */
   btime last_rt_change;			/* Last time when route changed */
   btime base_settle_time;		/* Start time of rtable settling interval */
   btime gc_time;			/* Time of last GC */
-  int gc_counter;			/* Number of operations since last GC */
+  uint gc_counter;			/* Number of operations since last GC */
   byte prune_state;			/* Table prune state, 1 -> scheduled, 2-> running */
   byte prune_trie;			/* Prune prefix trie during next table prune */
   byte hcu_scheduled;			/* Hostcache update is scheduled */
@@ -332,6 +333,7 @@ struct config;
 
 void rt_init(void);
 void rt_preconfig(struct config *);
+void rt_postconfig(struct config *);
 void rt_commit(struct config *new, struct config *old);
 void rt_lock_table(rtable *);
 void rt_unlock_table(rtable *);
diff --git a/nest/rt-table.c b/nest/rt-table.c
index c3563171..2cf55421 100644
--- a/nest/rt-table.c
+++ b/nest/rt-table.c
@@ -124,6 +124,7 @@ static void rt_next_hop_update(rtable *tab);
 static inline void rt_prune_table(rtable *tab);
 static inline void rt_schedule_notify(rtable *tab);
 static void rt_flowspec_notify(rtable *tab, net *net);
+static void rt_kick_prune_timer(rtable *tab);
 
 
 static void
@@ -1641,9 +1642,8 @@ rte_recalculate(struct channel *c, net *net, rte *new, struct rte_src *src)
   rte_announce(table, RA_UNDEF, net, new, old, net->routes, old_best);
 
   if (!net->routes &&
-      (table->gc_counter++ >= table->config->gc_max_ops) &&
-      (table->gc_time + table->config->gc_min_time <= current_time()))
-    rt_schedule_prune(table);
+      (table->gc_counter++ >= table->config->gc_threshold))
+    rt_kick_prune_timer(table);
 
   if (old_ok && p->rte_remove)
     p->rte_remove(net, old);
@@ -2098,6 +2098,29 @@ rt_event(void *ptr)
 }
 
 
+static void
+rt_prune_timer(timer *t)
+{
+  rtable *tab = t->data;
+
+  if (tab->gc_counter >= tab->config->gc_threshold)
+    rt_schedule_prune(tab);
+}
+
+static void
+rt_kick_prune_timer(rtable *tab)
+{
+  /* Return if prune is already scheduled */
+  if (tm_active(tab->prune_timer) || (tab->prune_state & 1))
+    return;
+
+  /* Randomize GC period to +/- 50% */
+  btime gc_period = tab->config->gc_period;
+  gc_period = (gc_period / 2) + (random_u32() % (uint) gc_period);
+  tm_start(tab->prune_timer, gc_period);
+}
+
+
 static inline btime
 rt_settled_time(rtable *tab)
 {
@@ -2333,6 +2356,7 @@ rt_setup(pool *pp, struct rtable_config *cf)
     hmap_set(&t->id_map, 0);
 
     t->rt_event = ev_new_init(p, rt_event, t);
+    t->prune_timer = tm_new_init(p, rt_prune_timer, t, 0, 0);
     t->last_rt_change = t->gc_time = current_time();
 
     if (rt_is_flow(t))
@@ -2403,6 +2427,9 @@ rt_prune_table(rtable *tab)
     FIB_ITERATE_INIT(fit, &tab->fib);
     tab->prune_state = 2;
 
+    tab->gc_counter = 0;
+    tab->gc_time = current_time();
+
     if (tab->prune_trie)
     {
       /* Init prefix trie pruning */
@@ -2462,9 +2489,6 @@ again:
   fib_check(&tab->fib);
 #endif
 
-  tab->gc_counter = 0;
-  tab->gc_time = current_time();
-
   /* state change 2->0, 3->1 */
   tab->prune_state &= 1;
 
@@ -2591,6 +2615,20 @@ rt_preconfig(struct config *c)
   rt_new_table(cf_get_symbol("master6"), NET_IP6);
 }
 
+void
+rt_postconfig(struct config *c)
+{
+  uint num_tables = list_length(&c->tables);
+  btime def_gc_period = 400 MS * num_tables;
+  def_gc_period = MAX(def_gc_period, 10 S);
+  def_gc_period = MIN(def_gc_period, 600 S);
+
+  struct rtable_config *rc;
+  WALK_LIST(rc, c->tables)
+    if (rc->gc_period == (uint) -1)
+      rc->gc_period = (uint) def_gc_period;
+}
+
 
 /*
  * Some functions for handing internal next hop updates
@@ -2999,8 +3037,8 @@ rt_new_table(struct symbol *s, uint addr_type)
   cf_define_symbol(s, SYM_TABLE, table, c);
   c->name = s->name;
   c->addr_type = addr_type;
-  c->gc_max_ops = 1000;
-  c->gc_min_time = 5;
+  c->gc_threshold = 1000;
+  c->gc_period = (uint) -1;	/* set in rt_postconfig() */
   c->min_settle_time = 1 S;
   c->max_settle_time = 20 S;
 

From f39e9aa20363eccc38289958f23c789d037c269b Mon Sep 17 00:00:00 2001
From: Ondrej Zajicek <santiago@crfreenet.org>
Date: Sat, 4 Jun 2022 17:54:08 +0200
Subject: [PATCH 2/5] IO: Improve resolution of latency debugging messages

---
 sysdep/unix/io.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/sysdep/unix/io.c b/sysdep/unix/io.c
index 4fd77453..17dc05a3 100644
--- a/sysdep/unix/io.c
+++ b/sysdep/unix/io.c
@@ -2046,8 +2046,8 @@ io_update_time(void)
     event_open->duration = last_time - event_open->timestamp;
 
     if (event_open->duration > config->latency_limit)
-      log(L_WARN "Event 0x%p 0x%p took %d ms",
-	  event_open->hook, event_open->data, (int) (event_open->duration TO_MS));
+      log(L_WARN "Event 0x%p 0x%p took %u.%03u ms",
+	  event_open->hook, event_open->data, (uint) (event_open->duration TO_MS), (uint) (event_open->duration % 1000));
 
     event_open = NULL;
   }
@@ -2151,8 +2151,8 @@ watchdog_stop(void)
 
   btime duration = last_time - loop_time;
   if (duration > config->watchdog_warning)
-    log(L_WARN "I/O loop cycle took %d ms for %d events",
-	(int) (duration TO_MS), event_log_num);
+    log(L_WARN "I/O loop cycle took %u.%03u ms for %d events",
+	(uint) (duration TO_MS), (uint) (duration % 1000), event_log_num);
 }
 
 

From ad686c55c3fad13f39e44ee5732c38296caff782 Mon Sep 17 00:00:00 2001
From: Ondrej Zajicek <santiago@crfreenet.org>
Date: Sun, 5 Jun 2022 04:03:43 +0200
Subject: [PATCH 3/5] Babel: Do not try to remove multicast seqno request
 objects from neighbour list
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

The Babel seqno request code keeps track of which seqno requests are
outstanding for a neighbour by putting them onto a per-neighbour list. When
reusing a seqno request, it will try to remove this node, but if the seqno
request in question was a multicast request with no neighbour attached this
will result in a crash because it tries to remove a list node that wasn't
added to any list.

Fix this by making the list remove conditional. Also fix neighbor removal
which were changing seqno requests to multicast ones instead of removing
them.

Fixes: ebd5751cdeb4 ("Babel: Seqno requests are properly decoupled from
neighbors when the underlying interface disappears").

Based on the patch from Toke Høiland-Jørgensen <toke@toke.dk>,
bug reported by Stefan Haller <stefan.haller@stha.de>, thanks.
---
 proto/babel/babel.c | 9 ++++-----
 1 file changed, 4 insertions(+), 5 deletions(-)

diff --git a/proto/babel/babel.c b/proto/babel/babel.c
index 4a7d550f..71452a6f 100644
--- a/proto/babel/babel.c
+++ b/proto/babel/babel.c
@@ -312,7 +312,9 @@ babel_add_seqno_request(struct babel_proto *p, struct babel_entry *e,
 
       /* Found older */
       rem_node(NODE sr);
-      rem_node(&sr->nbr_node);
+
+      if (sr->nbr)
+        rem_node(&sr->nbr_node);
 
       goto found;
     }
@@ -452,10 +454,7 @@ babel_flush_neighbor(struct babel_proto *p, struct babel_neighbor *nbr)
 
   struct babel_seqno_request *sr;
   WALK_LIST_FIRST2(sr, nbr_node, nbr->requests)
-  {
-    sr->nbr = NULL;
-    rem_node(&sr->nbr_node);
-  }
+    babel_remove_seqno_request(p, sr);
 
   nbr->ifa = NULL;
   rem_node(NODE nbr);

From 141fb51f1a3c22c45025426775c00f66c06406a7 Mon Sep 17 00:00:00 2001
From: Maria Matejka <mq@ucw.cz>
Date: Tue, 7 Jun 2022 10:35:48 +0200
Subject: [PATCH 4/5] IPv4 flowspec literals should reject IPv6 prefices in a
 well-behaved way

When writing flow4 { dst 2001:db8::dead:beef/128; }, BIRD crashed on an
not-well-debuggable segfault as it tried to copy the whole 128-bit
prefix into an IPv4-sized memory.
---
 conf/confbase.Y | 11 ++++++++++-
 conf/flowspec.Y |  2 +-
 2 files changed, 11 insertions(+), 2 deletions(-)

diff --git a/conf/confbase.Y b/conf/confbase.Y
index 6985783b..5f45c507 100644
--- a/conf/confbase.Y
+++ b/conf/confbase.Y
@@ -110,7 +110,7 @@ CF_DECLS
 %type <i> expr bool pxlen4
 %type <time> expr_us time
 %type <a> ipa
-%type <net> net_ip4_ net_ip6_ net_ip6 net_ip_ net_ip net_or_ipa
+%type <net> net_ip4_ 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_ net_ip6_sadr_ net_mpls_
 %type <mls> label_stack_start label_stack
 
@@ -303,6 +303,15 @@ net_:
 
 /* Networks - regular */
 
+net_ip4:
+   net_ip4_
+ | CF_SYM_KNOWN {
+     if (($1->class != (SYM_CONSTANT | T_NET)) || (SYM_VAL($1).net->type != NET_IP4))
+       cf_error("IPv4 network constant expected");
+     $$ = * SYM_VAL($1).net;
+   }
+ ;
+
 net_ip6:
    net_ip6_
  | CF_SYM_KNOWN {
diff --git a/conf/flowspec.Y b/conf/flowspec.Y
index 56a7c5dc..dbdbdda5 100644
--- a/conf/flowspec.Y
+++ b/conf/flowspec.Y
@@ -142,7 +142,7 @@ flow_frag_opts:
  ;
 
 flow4_item:
-   flow_srcdst net_ip {
+   flow_srcdst net_ip4 {
      flow_builder_set_type(this_flow, $1);
      flow_builder4_add_pfx(this_flow, (net_addr_ip4 *) &($2));
    }

From b867c798c3e42c875311a79fe53a36d4a5bfd126 Mon Sep 17 00:00:00 2001
From: Ondrej Zajicek <santiago@crfreenet.org>
Date: Thu, 16 Jun 2022 02:58:37 +0200
Subject: [PATCH 5/5] NEWS and version update

---
 NEWS            | 8 ++++++++
 misc/bird.spec  | 2 +-
 sysdep/config.h | 2 +-
 3 files changed, 10 insertions(+), 2 deletions(-)

diff --git a/NEWS b/NEWS
index c1e51af2..e61a0b74 100644
--- a/NEWS
+++ b/NEWS
@@ -1,3 +1,11 @@
+Version 2.0.10 (2022-06-16)
+  o BGP performance improvements
+  o BFD: New 'strict bind' option
+  o RPKI: VRF support
+  o Allow use of 240.0.0.0/4 as a private range
+  o BIRD client uses exit status to report errors
+  o Important bugfixes
+
 Version 2.0.9 (2022-02-09)
   o BGP: Flowspec validation procedure
   o Babel: MAC authentication support
diff --git a/misc/bird.spec b/misc/bird.spec
index 425a6340..c3fdd7d6 100644
--- a/misc/bird.spec
+++ b/misc/bird.spec
@@ -1,6 +1,6 @@
 Summary: BIRD Internet Routing Daemon
 Name: bird
-Version: 2.0.9
+Version: 2.0.10
 Release: 1
 Copyright: GPL
 Group: Networking/Daemons
diff --git a/sysdep/config.h b/sysdep/config.h
index b0531844..5cdadbb0 100644
--- a/sysdep/config.h
+++ b/sysdep/config.h
@@ -13,7 +13,7 @@
 #ifdef GIT_LABEL
 #define BIRD_VERSION XSTR1(GIT_LABEL)
 #else
-#define BIRD_VERSION "2.0.9"
+#define BIRD_VERSION "2.0.10"
 #endif
 
 /* Include parameters determined by configure script */