diff --git a/dbms/include/DB/DataStreams/LimitByBlockInputStream.h b/dbms/include/DB/DataStreams/LimitByBlockInputStream.h
index 796dd124c36..a9224daecb1 100644
--- a/dbms/include/DB/DataStreams/LimitByBlockInputStream.h
+++ b/dbms/include/DB/DataStreams/LimitByBlockInputStream.h
@@ -9,6 +9,12 @@
 namespace DB
 {
 
+/** Implements LIMIT BY clause witch can be used to obtain a "top N by subgroup".
+  *
+  * For example, if you have table T like this (Num: 1 1 3 3 3 4 4 5 7 7 7 7),
+  * the query SELECT Num FROM T LIMIT 2 BY Num
+  *	will give you the following result: (Num: 1 1 3 3 4 4 5 7 7).
+  */
 class LimitByBlockInputStream : public IProfilingBlockInputStream
 {
 public:
@@ -27,8 +33,8 @@ private:
 private:
 	using MapHashed = HashMap<UInt128, UInt64, UInt128TrivialHash>;
 
-	Names columns_names;
-	size_t group_size;
+	const Names columns_names;
+	const size_t group_size;
 	MapHashed keys_counts;
 };
 
diff --git a/dbms/src/DataStreams/LimitByBlockInputStream.cpp b/dbms/src/DataStreams/LimitByBlockInputStream.cpp
index b0df3e9be36..ea39fb7226c 100644
--- a/dbms/src/DataStreams/LimitByBlockInputStream.cpp
+++ b/dbms/src/DataStreams/LimitByBlockInputStream.cpp
@@ -19,6 +19,8 @@ String LimitByBlockInputStream::getID() const
 
 Block LimitByBlockInputStream::readImpl()
 {
+	/// Execute until end of stream or until
+	/// a block with some new records will be gotten.
 	while (true)
 	{
 		Block block = children[0]->read();
@@ -28,7 +30,7 @@ Block LimitByBlockInputStream::readImpl()
 		const ConstColumnPlainPtrs column_ptrs(getKeyColumns(block));
 		const size_t rows = block.rows();
 		IColumn::Filter filter(rows);
-		bool inserted = false;
+		size_t inserted_count = 0;
 
 		for (size_t i = 0; i < rows; ++i)
 		{
@@ -40,17 +42,22 @@ Block LimitByBlockInputStream::readImpl()
 
 			hash.get128(key.first, key.second);
 
-			const bool valid = (keys_counts[key]++ < group_size);
-			filter[i] = valid;
-			inserted |= valid;
+			if (keys_counts[key]++ < group_size)
+			{
+				inserted_count++;
+				filter[i] = 1;
+			}
+			else
+				filter[i] = 0;
 		}
 
-		if (!inserted)
+		/// Just go to the next block if there isn't any new records in the current one.
+		if (!inserted_count)
 			continue;
 
 		size_t all_columns = block.columns();
 		for (size_t i = 0; i < all_columns; ++i)
-			block.getByPosition(i).column = block.getByPosition(i).column->filter(filter, -1);
+			block.getByPosition(i).column = block.getByPosition(i).column->filter(filter, inserted_count);
 
 		return block;
 	}