0001-Reproduce-the-error-in-lock_twophase_recover_11.patch

From 90cedbdcadfe6e68ba2c0e74ba69a626fd402964 Mon Sep 17 00:00:00 2001
From: "suyu.cmj" <mengjuan.cmj@alibaba-inc.com>
Date: Mon, 10 Jul 2023 13:09:09 +0000
Subject: [PATCH] Reproduce the error in lock_twophase_recover()

---
 src/backend/access/transam/xlog.c             | 10 ++++
 src/backend/postmaster/checkpointer.c         | 51 +++++++++++++++++++
 src/include/catalog/pg_proc.dat               |  8 +++
 src/include/postmaster/bgwriter.h             |  3 ++
 ..._add_duplicate_twophase_during_recovery.pl | 50 ++++++++++++++++++
 5 files changed, 122 insertions(+)
 create mode 100644 src/test/recovery/t/034_add_duplicate_twophase_during_recovery.pl

diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 1174dd5beb..9c6f287267 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -9072,6 +9072,11 @@ CreateCheckPoint(int flags)
 
 	TRACE_POSTGRESQL_CHECKPOINT_START(flags);
 
+	/* Test inject fault */
+	while (IsCheckpointDelayed())
+		pg_usleep(1000000);
+	/* Test inject fault end */
+
 	/*
 	 * Get the other info we need for the checkpoint record.
 	 *
@@ -9169,6 +9174,11 @@ CreateCheckPoint(int flags)
 	}
 	pfree(vxids);
 
+	/* Test inject fault */
+	if (IsCheckpointPanicInjected())
+		elog(PANIC, "panic is injected during checkpoint");
+	/* Test inject fault end */
+
 	/*
 	 * Take a snapshot of running transactions and write this to WAL. This
 	 * allows us to reconstruct the state of running transactions during
diff --git a/src/backend/postmaster/checkpointer.c b/src/backend/postmaster/checkpointer.c
index fb912c0381..adcecd902e 100644
--- a/src/backend/postmaster/checkpointer.c
+++ b/src/backend/postmaster/checkpointer.c
@@ -126,6 +126,9 @@ typedef struct
 
 	int			ckpt_flags;		/* checkpoint flags, as defined in xlog.h */
 
+	bool		is_ckpt_delayed;	/* add for test */
+	bool		is_ckpt_panic_injected; /* add for test */
+
 	uint32		num_backend_writes; /* counts user backend buffer writes */
 	uint32		num_backend_fsync;	/* counts user backend fsync calls */
 
@@ -1392,3 +1395,51 @@ FirstCallSinceLastCheckpoint(void)
 
 	return FirstCall;
 }
+
+bool
+IsCheckpointDelayed(void)
+{
+	bool is_delayed = false;
+
+	SpinLockAcquire(&CheckpointerShmem->ckpt_lck);
+	is_delayed = CheckpointerShmem->is_ckpt_delayed;
+	SpinLockRelease(&CheckpointerShmem->ckpt_lck);
+
+	return is_delayed;
+}
+
+bool
+IsCheckpointPanicInjected(void)
+{
+	bool inject_panic = false;
+
+	SpinLockAcquire(&CheckpointerShmem->ckpt_lck);
+	inject_panic = CheckpointerShmem->is_ckpt_panic_injected;
+	SpinLockRelease(&CheckpointerShmem->ckpt_lck);
+
+	return inject_panic;
+}
+
+Datum
+pg_test_inject_checkpoint_delay(PG_FUNCTION_ARGS)
+{
+	bool set_delay = PG_GETARG_BOOL(0);
+
+	SpinLockAcquire(&CheckpointerShmem->ckpt_lck);
+	CheckpointerShmem->is_ckpt_delayed = set_delay;
+	SpinLockRelease(&CheckpointerShmem->ckpt_lck);
+
+	PG_RETURN_VOID();
+}
+
+Datum
+pg_test_inject_checkpoint_panic(PG_FUNCTION_ARGS)
+{
+	bool inject_panic = PG_GETARG_BOOL(0);
+
+	SpinLockAcquire(&CheckpointerShmem->ckpt_lck);
+	CheckpointerShmem->is_ckpt_panic_injected = inject_panic;
+	SpinLockRelease(&CheckpointerShmem->ckpt_lck);
+
+	PG_RETURN_VOID();
+}
\ No newline at end of file
diff --git a/src/include/catalog/pg_proc.dat b/src/include/catalog/pg_proc.dat
index 658229ce77..10ec7b4d32 100644
--- a/src/include/catalog/pg_proc.dat
+++ b/src/include/catalog/pg_proc.dat
@@ -10206,4 +10206,12 @@
   proisstrict => 'f', prorettype => 'bool', proargtypes => 'oid int4 int4 any',
   proargmodes => '{i,i,i,v}', prosrc => 'satisfies_hash_partition' },
 
+{ oid => '5058', descr => 'inject delay during checkpoint',
+  proname => 'pg_test_inject_checkpoint_delay', provolatile => 'v', prorettype => 'void',
+  proargtypes => 'bool', prosrc => 'pg_test_inject_checkpoint_delay' },
+
+{ oid => '5059', descr => 'inject panic during checkpoint',
+  proname => 'pg_test_inject_checkpoint_panic', provolatile => 'v', prorettype => 'void',
+  proargtypes => 'bool', prosrc => 'pg_test_inject_checkpoint_panic' },
+
 ]
diff --git a/src/include/postmaster/bgwriter.h b/src/include/postmaster/bgwriter.h
index 941c6aba7d..40e1eb00f2 100644
--- a/src/include/postmaster/bgwriter.h
+++ b/src/include/postmaster/bgwriter.h
@@ -40,4 +40,7 @@ extern void CheckpointerShmemInit(void);
 
 extern bool FirstCallSinceLastCheckpoint(void);
 
+extern bool IsCheckpointDelayed(void);
+extern bool IsCheckpointPanicInjected(void);
+
 #endif							/* _BGWRITER_H */
diff --git a/src/test/recovery/t/034_add_duplicate_twophase_during_recovery.pl b/src/test/recovery/t/034_add_duplicate_twophase_during_recovery.pl
new file mode 100644
index 0000000000..acd16cb494
--- /dev/null
+++ b/src/test/recovery/t/034_add_duplicate_twophase_during_recovery.pl
@@ -0,0 +1,50 @@
+use strict;
+use warnings;
+
+use PostgresNode;
+use TestLib;
+use Test::More tests => 1;
+
+# Setup primary node
+my $node_primary = get_new_node("primary");
+$node_primary->init;
+$node_primary->append_conf(
+	'postgresql.conf', qq(
+	max_prepared_transactions = 10
+	log_checkpoints = true
+	checkpoint_timeout = 3000
+));
+$node_primary->start;
+
+$node_primary->psql('postgres', "create table t_test_2pc(id int, msg text)");
+$node_primary->psql('postgres', "select pg_test_inject_checkpoint_delay(true)");
+
+# start do checkpoint1
+my $host = $node_primary->host;
+my $port = $node_primary->port;
+`nohup psql -h $host -p $port -c 'checkpoint' >test.file 2>&1 &`;
+
+# prepare transaction during checkpoint1
+$node_primary->psql(
+	'postgres', "
+	BEGIN;
+	INSERT INTO t_test_2pc VALUES (1, 'test add duplicate 2pc');
+	PREPARE TRANSACTION 'xact_test_1';");
+
+# reset delay flag to finish checkpoint1
+$node_primary->psql('postgres', "select pg_test_inject_checkpoint_delay(false)");
+sleep 3;
+
+# inject panic during checkpoint2
+$node_primary->psql('postgres', "select pg_test_inject_checkpoint_panic(true)");
+$node_primary->psql('postgres', "checkpoint");
+
+sleep 3;
+my $logdir = $node_primary->logfile;
+my @res   = `grep -rn "panic is injected during checkpoint" $logdir`;
+my $found = @res;
+ok($found == 1, "inject panic success");
+
+# start will fail due to error in lock_twophase_recover()
+$node_primary->_update_pid(0);
+$node_primary->start;
\ No newline at end of file
-- 
2.19.1.6.gb485710b