Re: BUG #17558: 15beta2: Endless loop with UNIQUE NULLS NOT DISTINCT and INSERT ... ON CONFLICT

Michael Paquier <michael@paquier.xyz> writes:
> On Tue, Jul 26, 2022 at 11:17:45AM +0800, Richard Guo wrote:
>> Yeah, I can see the same problem. _bt_check_unique() can catch violation
>> of unique index constraint successfully, so insert without 'ON CONFLICT'
>> has no problem. But ExecCheckIndexConstraints() fails to tell the NULL
>> tuple violates unique constraint, and the logic below is quite
>> suspicious to me.

> This is still an open item assigned to Peter, and beta3 is planned for
> the 11th of August.  Would a resolution by the beginning of next week
> be doable?

Since nothing seems to be happening here, I looked into it.  I think
Richard has correctly fingered the buggy code: we need to teach
check_exclusion_or_unique_constraint() that a NULL isn't necessarily
an instant pass.  As attached.

Since the presented test case caused an uninterruptible loop, I also added
a CHECK_FOR_INTERRUPTS() that could stop it.  (I was initially pretty
surprised that we could get through an index scan without hitting any CFI,
but on reflection it's less surprising, because no index scan actually
happens: check_exclusion_or_unique_constraint() short-circuits things
before very much code has been reached.)  I'm kind of inclined to
back-patch the nodeModifyTable.c change all the way, just for safety.

            regards, tom lane

diff --git a/src/backend/executor/execIndexing.c b/src/backend/executor/execIndexing.c
index 0cb0b8f111..6a8735edf7 100644
--- a/src/backend/executor/execIndexing.c
+++ b/src/backend/executor/execIndexing.c
@@ -699,13 +699,19 @@ check_exclusion_or_unique_constraint(Relation heap, Relation index,
     }

     /*
-     * If any of the input values are NULL, the constraint check is assumed to
-     * pass (i.e., we assume the operators are strict).
+     * If any of the input values are NULL, and the index uses the default
+     * nulls-are-distinct mode, the constraint check is assumed to pass (i.e.,
+     * we assume the operators are strict).  Otherwise, we interpret the
+     * constraint as specifying IS NULL for each column whose input value is
+     * NULL.
      */
-    for (i = 0; i < indnkeyatts; i++)
+    if (!indexInfo->ii_NullsNotDistinct)
     {
-        if (isnull[i])
-            return true;
+        for (i = 0; i < indnkeyatts; i++)
+        {
+            if (isnull[i])
+                return true;
+        }
     }

     /*
@@ -717,7 +723,7 @@ check_exclusion_or_unique_constraint(Relation heap, Relation index,
     for (i = 0; i < indnkeyatts; i++)
     {
         ScanKeyEntryInitialize(&scankeys[i],
-                               0,
+                               isnull[i] ? SK_ISNULL | SK_SEARCHNULL : 0,
                                i + 1,
                                constr_strats[i],
                                InvalidOid,
diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index a49c3da5b6..deda321502 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -956,9 +956,11 @@ ExecInsert(ModifyTableContext *context,
              *
              * We loop back here if we find a conflict below, either during
              * the pre-check, or when we re-check after inserting the tuple
-             * speculatively.
+             * speculatively.  Better allow interrupts in case some bug makes
+             * this an infinite loop.
              */
     vlock:
+            CHECK_FOR_INTERRUPTS();
             specConflict = false;
             if (!ExecCheckIndexConstraints(resultRelInfo, slot, estate,
                                            &conflictTid, arbiterIndexes))
diff --git a/src/test/regress/expected/constraints.out b/src/test/regress/expected/constraints.out
index 36ccbb5f15..05b7244e4a 100644
--- a/src/test/regress/expected/constraints.out
+++ b/src/test/regress/expected/constraints.out
@@ -449,15 +449,16 @@ DROP TABLE UNIQUE_TBL;
 CREATE TABLE UNIQUE_TBL (i int UNIQUE NULLS NOT DISTINCT, t text);
 INSERT INTO UNIQUE_TBL VALUES (1, 'one');
 INSERT INTO UNIQUE_TBL VALUES (2, 'two');
-INSERT INTO UNIQUE_TBL VALUES (1, 'three');
+INSERT INTO UNIQUE_TBL VALUES (1, 'three');  -- fail
 ERROR:  duplicate key value violates unique constraint "unique_tbl_i_key"
 DETAIL:  Key (i)=(1) already exists.
 INSERT INTO UNIQUE_TBL VALUES (4, 'four');
 INSERT INTO UNIQUE_TBL VALUES (5, 'one');
 INSERT INTO UNIQUE_TBL (t) VALUES ('six');
-INSERT INTO UNIQUE_TBL (t) VALUES ('seven');
+INSERT INTO UNIQUE_TBL (t) VALUES ('seven');  -- fail
 ERROR:  duplicate key value violates unique constraint "unique_tbl_i_key"
 DETAIL:  Key (i)=(null) already exists.
+INSERT INTO UNIQUE_TBL (t) VALUES ('eight') ON CONFLICT DO NOTHING;  -- no-op
 SELECT * FROM UNIQUE_TBL;
  i |  t
 ---+------
diff --git a/src/test/regress/sql/constraints.sql b/src/test/regress/sql/constraints.sql
index 34de0c969a..833819a32e 100644
--- a/src/test/regress/sql/constraints.sql
+++ b/src/test/regress/sql/constraints.sql
@@ -310,11 +310,12 @@ CREATE TABLE UNIQUE_TBL (i int UNIQUE NULLS NOT DISTINCT, t text);

 INSERT INTO UNIQUE_TBL VALUES (1, 'one');
 INSERT INTO UNIQUE_TBL VALUES (2, 'two');
-INSERT INTO UNIQUE_TBL VALUES (1, 'three');
+INSERT INTO UNIQUE_TBL VALUES (1, 'three');  -- fail
 INSERT INTO UNIQUE_TBL VALUES (4, 'four');
 INSERT INTO UNIQUE_TBL VALUES (5, 'one');
 INSERT INTO UNIQUE_TBL (t) VALUES ('six');
-INSERT INTO UNIQUE_TBL (t) VALUES ('seven');
+INSERT INTO UNIQUE_TBL (t) VALUES ('seven');  -- fail
+INSERT INTO UNIQUE_TBL (t) VALUES ('eight') ON CONFLICT DO NOTHING;  -- no-op

 SELECT * FROM UNIQUE_TBL;