diff --git a/executor/benchmark_test.go b/executor/benchmark_test.go
index 3f64164332ce7..ec8e5e0a5a439 100644
--- a/executor/benchmark_test.go
+++ b/executor/benchmark_test.go
@@ -1105,6 +1105,16 @@ func BenchmarkHashJoinExec(b *testing.B) {
 	b.Run(fmt.Sprintf("%v", cas), func(b *testing.B) {
 		benchmarkHashJoinExecWithCase(b, cas)
 	})
+
+	cols = []*types.FieldType{
+		types.NewFieldType(mysql.TypeLonglong),
+	}
+	cas = defaultHashJoinTestCase(cols, 0, false)
+	cas.keyIdx = []int{0}
+	cas.disk = true
+	b.Run(fmt.Sprintf("%v", cas), func(b *testing.B) {
+		benchmarkHashJoinExecWithCase(b, cas)
+	})
 }
 
 func BenchmarkOuterHashJoinExec(b *testing.B) {
diff --git a/executor/hash_table.go b/executor/hash_table.go
index 4afdef58e1973..f5011c5877945 100644
--- a/executor/hash_table.go
+++ b/executor/hash_table.go
@@ -242,12 +242,11 @@ func (c *hashRowContainer) GetMatchedRowsAndPtrs(probeKey uint64, probeRow chunk
 
 	// Some variables used for memTracker.
 	var (
-		matchedDataSize                  = int64(cap(matched))*rowSize + int64(cap(matchedPtrs))*rowPtrSize
-		lastChunkBufPointer *chunk.Chunk = nil
-		memDelta            int64        = 0
-		needTrackMemUsage                = cap(innerPtrs) > signalCheckpointForJoinMask
+		matchedDataSize           = int64(cap(matched))*rowSize + int64(cap(matchedPtrs))*rowPtrSize
+		lastChunkBufPointer       = c.chkBuf
+		memDelta            int64 = 0
+		needTrackMemUsage         = cap(innerPtrs) > signalCheckpointForJoinMask
 	)
-	c.chkBuf = nil
 	c.memTracker.Consume(-c.chkBufSizeForOneProbe)
 	if needTrackMemUsage {
 		c.memTracker.Consume(int64(cap(innerPtrs)) * rowPtrSize)
@@ -265,7 +264,7 @@ func (c *hashRowContainer) GetMatchedRowsAndPtrs(probeKey uint64, probeRow chunk
 		if err != nil {
 			return nil, nil, err
 		}
-		if needTrackMemUsage && c.chkBuf != lastChunkBufPointer && lastChunkBufPointer != nil {
+		if c.chkBuf != lastChunkBufPointer && lastChunkBufPointer != nil {
 			lastChunkSize := lastChunkBufPointer.MemoryUsage()
 			c.chkBufSizeForOneProbe += lastChunkSize
 			memDelta += lastChunkSize