[SPARK-18485][CORE] Underlying integer overflow when create...

[SPARK-18485][CORE] Underlying integer overflow when create ChunkedByteBufferOutputStream in MemoryStore

## What changes were proposed in this pull request?

There is an underlying integer overflow when create ChunkedByteBufferOutputStream in MemoryStore. This PR provide a check before cast.

## How was this patch tested?

add new unit test

Author: uncleGen <hustyugm@gmail.com>

Closes #15915 from uncleGen/SPARK-18485.

core/src/main/scala/org/apache/spark/storage/memory/MemoryStore.scala

@@ -331,7 +331,15 @@ private[spark] class MemoryStore(
     var unrollMemoryUsedByThisBlock = 0L
     // Underlying buffer for unrolling the block
     val redirectableStream = new RedirectableOutputStream
-    val bbos = new ChunkedByteBufferOutputStream(initialMemoryThreshold.toInt, allocator)
+    val chunkSize = if (initialMemoryThreshold > Int.MaxValue) {
+      logWarning(s"Initial memory threshold of ${Utils.bytesToString(initialMemoryThreshold)} " +
+        s"is too large to be set as chunk size. Chunk size has been capped to " +
+        s"${Utils.bytesToString(Int.MaxValue)}")
+      Int.MaxValue
+    } else {
+      initialMemoryThreshold.toInt
+    }
+    val bbos = new ChunkedByteBufferOutputStream(chunkSize, allocator)
     redirectableStream.setOutputStream(bbos)
     val serializationStream: SerializationStream = {
       val autoPick = !blockId.isInstanceOf[StreamBlockId]