0001-Fix-race-condition-that-could-corrupt-pg_control.patch

From 824d2b622eae134cec8f732a29bac69c83f03484 Mon Sep 17 00:00:00 2001
From: Thomas Munro <thomas.munro@gmail.com>
Date: Fri, 22 May 2020 16:23:59 +1200
Subject: [PATCH 1/3] Fix race condition that could corrupt pg_control.

The redo routines for XLOG_CHECKPOINT_{ONLINE,SHUTDOWN} must acquire
ControlFileLock before modifying ControlFile->checkPointCopy, or the
checkpointer could write out a control file with a bad checksum.

Back-patch to all supported releases.

Author: Nathan Bossart
Reviewed-by: Fujii Masao, Thomas Munro
Discussion: https://postgr.es/m/70BF24D6-DC51-443F-B55A-95735803842A%40amazon.com
---
 src/backend/access/transam/xlog.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index ca09d81b08..274b808476 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -9976,7 +9976,9 @@ xlog_redo(XLogReaderState *record)
 		}
 
 		/* ControlFile->checkPointCopy always tracks the latest ckpt XID */
+		LWLockAcquire(ControlFileLock, LW_EXCLUSIVE);
 		ControlFile->checkPointCopy.nextFullXid = checkPoint.nextFullXid;
+		LWLockRelease(ControlFileLock);
 
 		/* Update shared-memory copy of checkpoint XID/epoch */
 		SpinLockAcquire(&XLogCtl->info_lck);
@@ -10033,7 +10035,9 @@ xlog_redo(XLogReaderState *record)
 			SetTransactionIdLimit(checkPoint.oldestXid,
 								  checkPoint.oldestXidDB);
 		/* ControlFile->checkPointCopy always tracks the latest ckpt XID */
+		LWLockAcquire(ControlFileLock, LW_EXCLUSIVE);
 		ControlFile->checkPointCopy.nextFullXid = checkPoint.nextFullXid;
+		LWLockRelease(ControlFileLock);
 
 		/* Update shared-memory copy of checkpoint XID/epoch */
 		SpinLockAcquire(&XLogCtl->info_lck);
-- 
2.16.6