From ce413bd37c38d4b3602d11ff59fabbaa722e47e0 Mon Sep 17 00:00:00 2001
From: "suyu.cmj" <mengjuan.cmj@alibaba-inc.com>
Date: Mon, 15 Sep 2025 08:44:17 +0000
Subject: [PATCH] Newly created replication slot may be invalidated by
 checkpoint

Commit 2090edc6f32f652a2c995ca5f7e65748ae1e4c5d introduced a change
that the minimal restart_lsn is obtained at the start of checkpoint
creation. If a replication slot is created and performs a WAL
reservation concurrently, the WAL segment contains the new slot's
restart_lsn could be removed by the ongoing checkpoint. Add a perl
test to reproduce this scenario.
---
 src/backend/replication/slot.c                | 10 +++
 src/backend/utils/misc/injection_point.c      | 15 +++++
 src/include/utils/injection_point.h           |  1 +
 .../recovery/t/049_invalidate_new_slot.pl     | 64 +++++++++++++++++++
 4 files changed, 90 insertions(+)
 create mode 100644 src/test/recovery/t/049_invalidate_new_slot.pl

diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index b9e2b115dab..34cf03b92e7 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -57,6 +57,7 @@
 #include "utils/builtins.h"
 #include "utils/guc_hooks.h"
 #include "utils/varlena.h"
+#include "utils/injection_point.h"
 
 /*
  * Replication slot on-disk data structure.
@@ -1443,6 +1444,15 @@ ReplicationSlotReserveWal(void)
 		slot->data.restart_lsn = restart_lsn;
 		SpinLockRelease(&slot->mutex);
 
+#ifdef USE_INJECTION_POINTS
+		if (InjectionPointFind("delay_slot_reserve_wal"))
+		{
+			elog(LOG, "inject delay before computing required lsn");
+			while (InjectionPointFind("delay_slot_reserve_wal"))
+				pg_usleep(1000000);
+		}
+#endif
+
 		/* prevent WAL removal as fast as possible */
 		ReplicationSlotsComputeRequiredLSN();
 
diff --git a/src/backend/utils/misc/injection_point.c b/src/backend/utils/misc/injection_point.c
index b33cddefabc..76e79b2306d 100644
--- a/src/backend/utils/misc/injection_point.c
+++ b/src/backend/utils/misc/injection_point.c
@@ -535,3 +535,18 @@ InjectionPointRun(const char *name)
 	elog(ERROR, "Injection points are not supported by this build");
 #endif
 }
+
+bool
+InjectionPointFind(const char *name)
+{
+#ifdef USE_INJECTION_POINTS
+	InjectionPointCacheEntry *cache_entry;
+
+	cache_entry = InjectionPointCacheRefresh(name);
+
+	return cache_entry != NULL;
+#else
+	elog(ERROR, "Injection points are not supported by this build");
+	return false;
+#endif
+}
diff --git a/src/include/utils/injection_point.h b/src/include/utils/injection_point.h
index 6e417cedc60..bcc3de3e5b1 100644
--- a/src/include/utils/injection_point.h
+++ b/src/include/utils/injection_point.h
@@ -36,6 +36,7 @@ extern void InjectionPointAttach(const char *name,
 								 int private_data_size);
 extern void InjectionPointRun(const char *name);
 extern bool InjectionPointDetach(const char *name);
+extern bool InjectionPointFind(const char *name);
 
 #ifdef EXEC_BACKEND
 extern PGDLLIMPORT struct InjectionPointsCtl *ActiveInjectionPoints;
diff --git a/src/test/recovery/t/049_invalidate_new_slot.pl b/src/test/recovery/t/049_invalidate_new_slot.pl
new file mode 100644
index 00000000000..178352a4b08
--- /dev/null
+++ b/src/test/recovery/t/049_invalidate_new_slot.pl
@@ -0,0 +1,64 @@
+# This test checks that the new slot maybe invalidated by checkpoint
+#
+
+use strict;
+use warnings;
+use Config;
+use PostgreSQL::Test::Cluster;
+use PostgreSQL::Test::Utils;
+use Test::More;
+
+if ($ENV{enable_injection_points} ne 'yes')
+{
+	plan skip_all => 'Injection points not supported by this build';
+}
+
+# Setup primary node
+my $node = PostgreSQL::Test::Cluster->new('primary');
+$node->init(allows_streaming => 1);
+
+$node->append_conf(
+	'postgresql.conf', qq(
+	log_checkpoints = true
+	checkpoint_timeout = 3000
+    Logging_collector = false
+));
+$node->start;
+
+my $res =
+  $node->safe_psql('postgres', "create extension injection_points;");
+$res = $node->safe_psql('postgres',
+	"select injection_points_attach('delay_slot_reserve_wal', 'notice')");
+
+# create slot and reserve wal
+my $host = $node->host;
+my $port = $node->port;
+`nohup psql -h $host -p $port -c "SELECT pg_create_physical_replication_slot('standby1', true)" >tmp_check/test.file 2>&1 &`;
+
+# switch wal and do checkpoint
+$res = $node->safe_psql('postgres',
+	"select pg_switch_wal()");
+$res = $node->safe_psql('postgres',
+	"create table test(a int)");
+$res = $node->safe_psql('postgres',
+	"INSERT INTO test(a) SELECT generate_series(1,100) as newwal");
+
+`nohup psql -h $host -p $port -c 'checkpoint' >tmp_check/test.file 2>&1 &`;
+sleep 3;
+
+$res = $node->safe_psql('postgres',
+	"select injection_points_detach('delay_slot_reserve_wal')");
+
+my $log_offset = -s $node->logfile;
+$node->wait_for_log(qr/checkpoint complete/, $log_offset);
+print "checkpoint finished\n";
+
+$res = $node->safe_psql('postgres',
+	"select invalidation_reason from pg_replication_slots where slot_name='standby1'");
+print "res:$res\n";
+is($res, "wal_removed", "slot is invalidated by checkpoint");
+
+`rm tmp_check/test.file`;
+$node->stop;
+
+done_testing();
-- 
2.27.0