From: Alvaro Herrera <alvherre@alvh.no-ip.org>
Date: Wed, 23 Mar 2022 18:23:51 +0000 (+0100)
Subject: pg_upgrade: Upgrade an Assert to a real 'if' test
X-Git-Tag: REL_13_7~54
X-Git-Url: http://git.ipfire.org/?a=commitdiff_plain;h=ca26c64581b529b2191757dbb27b94c6b0c101be;p=thirdparty%2Fpostgresql.git

pg_upgrade: Upgrade an Assert to a real 'if' test

It seems possible for the condition being tested to be true in
production, and nobody would never know (except when some data
eventually becomes corrupt?).

Author: Álvaro Herrera <alvherre@alvh.no-ip.org>
Discussion: https://postgr.es/m//202109040001.zky3wgv2qeqg@alvherre.pgsql
---

diff --git a/src/bin/pg_rewind/parsexlog.c b/src/bin/pg_rewind/parsexlog.c
index 1a67725ec04..cb1bdc958d3 100644
--- a/src/bin/pg_rewind/parsexlog.c
+++ b/src/bin/pg_rewind/parsexlog.c
@@ -101,7 +101,11 @@ extractPageMap(const char *datadir, XLogRecPtr startpoint, int tliIndex,
 	 * If 'endpoint' didn't point exactly at a record boundary, the caller
 	 * messed up.
 	 */
-	Assert(xlogreader->EndRecPtr == endpoint);
+	if (xlogreader->EndRecPtr != endpoint)
+		pg_fatal("end pointer %X/%X is not a valid end point; expected %X/%X",
+				 (uint32) (endpoint >> 32), (uint32) (endpoint),
+				 (uint32) (xlogreader->EndRecPtr >> 32), (uint32)
+				 (xlogreader->EndRecPtr));
 
 	XLogReaderFree(xlogreader);
 	if (xlogreadfd != -1)