From 1925ac281a238fab50e208548f98daffdaaa0896 Mon Sep 17 00:00:00 2001
From: Tom Lane <tgl@sss.pgh.pa.us>
Date: Sat, 11 Mar 2023 12:15:41 -0500
Subject: [PATCH] Fix misbehavior in contrib/pg_trgm with an unsatisfiable
 regex.

If the regex compiler can see that a regex is unsatisfiable
(for example, '$foo') then it may emit an NFA having no arcs.
pg_trgm's packGraph function did the wrong thing in this case;
it would access off the end of a work array, and with bad luck
could produce a corrupted output data structure causing more
problems later.  This could end with wrong answers or crashes
in queries using a pg_trgm GIN or GiST index with such a regex.

Fix by not trying to de-duplicate if there aren't at least 2 arcs.

Per bug #17830 from Alexander Lakhin.  Back-patch to all supported
branches.

Discussion: https://postgr.es/m/17830-57ff5f89bdb02b09@postgresql.org
---
 contrib/pg_trgm/expected/pg_word_trgm.out |  6 ++++++
 contrib/pg_trgm/sql/pg_word_trgm.sql      |  3 +++
 contrib/pg_trgm/trgm_regexp.c             | 26 ++++++++++++++---------
 3 files changed, 25 insertions(+), 10 deletions(-)

diff --git a/contrib/pg_trgm/expected/pg_word_trgm.out b/contrib/pg_trgm/expected/pg_word_trgm.out
index 936d489390..c66a67f30e 100644
--- a/contrib/pg_trgm/expected/pg_word_trgm.out
+++ b/contrib/pg_trgm/expected/pg_word_trgm.out
@@ -1044,3 +1044,9 @@ select t,word_similarity('Kabankala',t) as sml from test_trgm2 where t %> 'Kaban
  Waikala                          |      0.3
 (89 rows)
 
+-- test unsatisfiable pattern
+select * from test_trgm2 where t ~ '.*$x';
+ t 
+---
+(0 rows)
+
diff --git a/contrib/pg_trgm/sql/pg_word_trgm.sql b/contrib/pg_trgm/sql/pg_word_trgm.sql
index d9fa1c55e5..d2ada49133 100644
--- a/contrib/pg_trgm/sql/pg_word_trgm.sql
+++ b/contrib/pg_trgm/sql/pg_word_trgm.sql
@@ -43,3 +43,6 @@ select t,word_similarity('Baykal',t) as sml from test_trgm2 where 'Baykal' <% t
 select t,word_similarity('Kabankala',t) as sml from test_trgm2 where 'Kabankala' <% t order by sml desc, t;
 select t,word_similarity('Baykal',t) as sml from test_trgm2 where t %> 'Baykal' order by sml desc, t;
 select t,word_similarity('Kabankala',t) as sml from test_trgm2 where t %> 'Kabankala' order by sml desc, t;
+
+-- test unsatisfiable pattern
+select * from test_trgm2 where t ~ '.*$x';
diff --git a/contrib/pg_trgm/trgm_regexp.c b/contrib/pg_trgm/trgm_regexp.c
index 9a00564ae4..06cd3db67b 100644
--- a/contrib/pg_trgm/trgm_regexp.c
+++ b/contrib/pg_trgm/trgm_regexp.c
@@ -1947,9 +1947,7 @@ packGraph(TrgmNFA *trgmNFA, MemoryContext rcontext)
 				arcsCount;
 	HASH_SEQ_STATUS scan_status;
 	TrgmState  *state;
-	TrgmPackArcInfo *arcs,
-			   *p1,
-			   *p2;
+	TrgmPackArcInfo *arcs;
 	TrgmPackedArc *packedArcs;
 	TrgmPackedGraph *result;
 	int			i,
@@ -2021,17 +2019,25 @@ packGraph(TrgmNFA *trgmNFA, MemoryContext rcontext)
 	qsort(arcs, arcIndex, sizeof(TrgmPackArcInfo), packArcInfoCmp);
 
 	/* We could have duplicates because states were merged. Remove them. */
-	/* p1 is probe point, p2 is last known non-duplicate. */
-	p2 = arcs;
-	for (p1 = arcs + 1; p1 < arcs + arcIndex; p1++)
+	if (arcIndex > 1)
 	{
-		if (packArcInfoCmp(p1, p2) > 0)
+		/* p1 is probe point, p2 is last known non-duplicate. */
+		TrgmPackArcInfo *p1,
+				   *p2;
+
+		p2 = arcs;
+		for (p1 = arcs + 1; p1 < arcs + arcIndex; p1++)
 		{
-			p2++;
-			*p2 = *p1;
+			if (packArcInfoCmp(p1, p2) > 0)
+			{
+				p2++;
+				*p2 = *p1;
+			}
 		}
+		arcsCount = (p2 - arcs) + 1;
 	}
-	arcsCount = (p2 - arcs) + 1;
+	else
+		arcsCount = arcIndex;
 
 	/* Create packed representation */
 	result = (TrgmPackedGraph *)