diff --git a/sql/core/src/main/scala/org/apache/spark/sql/execution/Exchange.scala b/sql/core/src/main/scala/org/apache/spark/sql/execution/Exchange.scala
index ba866357f8bf0..d67e5c40a4727 100644
--- a/sql/core/src/main/scala/org/apache/spark/sql/execution/Exchange.scala
+++ b/sql/core/src/main/scala/org/apache/spark/sql/execution/Exchange.scala
@@ -29,7 +29,10 @@ import org.apache.spark.sql.catalyst.rules.Rule
 import org.apache.spark.util.MutablePair
 
 /**
- * :: DeveloperApi ::
+ * Shuffle data according to a new partition rule, and sort inside each partition if necessary.
+ * @param newPartitioning The new partitioning way that required by parent
+ * @param sort            Whether we will sort inside each partition
+ * @param child           Child operator
  */
 @DeveloperApi
 case class Exchange(
diff --git a/sql/core/src/main/scala/org/apache/spark/sql/execution/basicOperators.scala b/sql/core/src/main/scala/org/apache/spark/sql/execution/basicOperators.scala
index c723c502ab90b..cbfcca1ea1546 100644
--- a/sql/core/src/main/scala/org/apache/spark/sql/execution/basicOperators.scala
+++ b/sql/core/src/main/scala/org/apache/spark/sql/execution/basicOperators.scala
@@ -45,11 +45,9 @@ case class Project(projectList: Seq[NamedExpression], child: SparkPlan) extends
   /**
    * outputOrdering of Project is not always same with child's outputOrdering if the certain
    * key is pruned, however, if the key is pruned then we must not require child using this
-   * ordering from upper layer, only if the ordering would not be changed by a negative, there
-   * would be a way to keep the ordering.
-   * TODO: we may utilize this feature later to avoid some unnecessary sorting.
+   * ordering from upper layer, so it is fine to keep it to avoid some unnecessary sorting.
    */
-  override def outputOrdering: Option[Ordering[Row]] = None
+  override def outputOrdering: Option[Ordering[Row]] = child.outputOrdering
 }
 
 /**