*: remove CorrelatedCol.HashCode to avoid potential bug (#45034)

close #45033

executor/builder.go

@@ -5458,7 +5458,7 @@ func (b *executorBuilder) buildCTE(v *plannercore.PhysicalCTE) Executor {
 
 		var corColHashCodes [][]byte
 		for _, corCol := range corCols {
-			corColHashCodes = append(corColHashCodes, corCol.HashCode(b.ctx.GetSessionVars().StmtCtx))
+			corColHashCodes = append(corColHashCodes, getCorColHashCode(corCol))
 		}
 
 		producer = &cteProducer{

executor/cte.go

@@ -661,12 +661,15 @@ func (p *cteProducer) checkHasDup(probeKey uint64,
 	return false, nil
 }
 
+func getCorColHashCode(corCol *expression.CorrelatedColumn) (res []byte) {
+	return codec.HashCode(res, *corCol.Data)
+}
+
 // Return true if cor col has changed.
 func (p *cteProducer) checkAndUpdateCorColHashCode() bool {
 	var changed bool
 	for i, corCol := range p.corCols {
-		corCol.CleanHashCode()
-		newHashCode := corCol.HashCode(p.ctx.GetSessionVars().StmtCtx)
+		newHashCode := getCorColHashCode(corCol)
 		if !bytes.Equal(newHashCode, p.corColHashCodes[i]) {
 			changed = true
 			p.corColHashCodes[i] = newHashCode

expression/column.go

@@ -37,8 +37,7 @@ import (
 type CorrelatedColumn struct {
 	Column
 
-	Data           *types.Datum
-	columnHashCode []byte
+	Data *types.Datum
 }
 
 // Clone implements Expression interface.
@@ -224,30 +223,6 @@ func (col *CorrelatedColumn) RemapColumn(m map[int64]*Column) (Expression, error
 	}, nil
 }
 
-// HashCode implements Expression interface.
-func (col *CorrelatedColumn) HashCode(sc *stmtctx.StatementContext) []byte {
-	if len(col.columnHashCode) == 0 {
-		col.columnHashCode = make([]byte, 0, 9)
-		col.columnHashCode = append(col.columnHashCode, columnFlag)
-		col.columnHashCode = codec.EncodeInt(col.columnHashCode, col.UniqueID)
-	}
-
-	if len(col.hashcode) < len(col.columnHashCode) {
-		if len(col.hashcode) == 0 {
-			col.hashcode = make([]byte, 0, len(col.columnHashCode))
-		} else {
-			col.hashcode = col.hashcode[:0]
-		}
-		col.hashcode = append(col.hashcode, col.columnHashCode...)
-	}
-
-	// Because col.Data can be changed anytime, so always use newest Datum to calc hash code.
-	if col.Data != nil {
-		col.hashcode = codec.HashCode(col.hashcode, *col.Data)
-	}
-	return col.hashcode
-}
-
 // Column represents a column.
 type Column struct {
 	RetType *types.FieldType

expression/column_test.go

@@ -15,7 +15,6 @@
 package expression
 
 import (
-	"bytes"
 	"fmt"
 	"testing"
 
@@ -264,30 +263,3 @@ func TestGcColumnExprIsTidbShard(t *testing.T) {
 	shardExpr := NewFunctionInternal(ctx, ast.TiDBShard, ft, col)
 	require.True(t, GcColumnExprIsTidbShard(shardExpr))
 }
-
-func TestCorColHashCode(t *testing.T) {
-	ctx := mock.NewContext()
-	sc := ctx.GetSessionVars().StmtCtx
-	col := &Column{UniqueID: 0, ID: 0, RetType: types.NewFieldType(mysql.TypeLonglong)}
-
-	corCol := CorrelatedColumn{
-		Column: *col,
-	}
-
-	oriCorColHashCode := corCol.HashCode(sc)
-	oriColHashCode := col.HashCode(sc)
-	// hash code is same when Data is not set.
-	require.True(t, bytes.Equal(oriColHashCode, oriCorColHashCode))
-
-	// corCol.hashcode changes after datum changed.
-	d1 := types.NewDatum(1)
-	corCol.Data = &d1
-	require.False(t, bytes.Equal(col.HashCode(sc), corCol.HashCode(sc)))
-	d1HashCode := corCol.HashCode(sc)
-	d2 := types.NewFloat64Datum(1.1)
-	corCol.Data = &d2
-	require.False(t, bytes.Equal(d1HashCode, corCol.HashCode(sc)))
-
-	// col.hashcode doesn't change.
-	require.True(t, bytes.Equal(oriColHashCode, col.HashCode(sc)))
-}

planner/core/integration_test.go

@@ -5303,3 +5303,32 @@ func TestIssue43116(t *testing.T) {
 		"  └─TableFullScan 10000.00 cop[tikv] table:a keep order:false, stats:pseudo"))
 	tk.MustQuery("show warnings").Check(testkit.Rows("Warning 1105 Memory capacity of 111 bytes for 'tidb_opt_range_max_size' exceeded when building ranges. Less accurate ranges such as full range are chosen"))
 }
+
+func TestIssue45033(t *testing.T) {
+	store := testkit.CreateMockStore(t)
+	tk := testkit.NewTestKit(t, store)
+	tk.MustExec("use test")
+	tk.MustExec("drop table if exists t1, t2, t3, t4;")
+	tk.MustExec("create table t1 (c1 int, c2 int, c3 int, primary key(c1, c2));")
+	tk.MustExec("create table t2 (c2 int, c1 int, primary key(c2, c1));")
+	tk.MustExec("create table t3 (c4 int, key(c4));")
+	tk.MustExec("create table t4 (c2 varchar(20) , test_col varchar(50), gen_col varchar(50) generated always as(concat(test_col,'')) virtual not null, unique key(gen_col));")
+	tk.MustQuery(`select count(1)
+				 from   (select ( case
+				                    when count(1)
+				                           over(
+				                             partition by a.c2) >= 50 then 1
+				                    else 0
+				                  end ) alias1,
+				                b.c2    as alias_col1
+				         from   t1 a
+				                left join (select c2
+				                           from   t4 f) k
+				                       on k.c2 = a.c2
+				                inner join t2 b
+				                        on b.c1 = a.c3) alias2
+				 where  exists (select 1
+				                from   (select distinct alias3.c4 as c2
+				                        from   t3 alias3) alias4
+				                where  alias4.c2 = alias2.alias_col1);`).Check(testkit.Rows("0"))
+}