Use stageId as index for fileSet caches.

core/src/main/scala/spark/scheduler/ShuffleMapTask.scala

@@ -22,6 +22,7 @@ object ShuffleMapTask {
   // expensive on the master node if it needs to launch thousands of tasks.
   val serializedInfoCache = new JHashMap[Int, Array[Byte]]
   val fileSetCache = new JHashMap[Int, Array[Byte]]
+  val jarSetCache = new JHashMap[Int, Array[Byte]]
 
   def serializeInfo(stageId: Int, rdd: RDD[_], dep: ShuffleDependency[_,_,_]): Array[Byte] = {
     synchronized {
@@ -42,8 +43,8 @@ object ShuffleMapTask {
   }
 
   // Since both the JarSet and FileSet have the same format this is used for both.
-  def serializeFileSet(set : HashMap[String, Long]) : Array[Byte] = {
-    val old = fileSetCache.get(set.hashCode)
+  def serializeFileSet(set : HashMap[String, Long], stageId: Int, cache : JHashMap[Int, Array[Byte]]) : Array[Byte] = {
+    val old = cache.get(stageId)
     if (old != null) {
       return old
     } else {
@@ -52,7 +53,7 @@ object ShuffleMapTask {
       objOut.writeObject(set.toArray)
       objOut.close()
       val bytes = out.toByteArray
-      fileSetCache.put(set.hashCode, bytes)
+      cache.put(stageId, bytes)
       return bytes
     }
   }
@@ -84,6 +85,7 @@ object ShuffleMapTask {
     synchronized {
       serializedInfoCache.clear()
       fileSetCache.clear()
+      jarSetCache.clear()
     }
   }
 }
@@ -112,10 +114,10 @@ class ShuffleMapTask(
     out.writeInt(bytes.length)
     out.write(bytes)
 
-    val fileSetBytes = ShuffleMapTask.serializeFileSet(fileSet)
+    val fileSetBytes = ShuffleMapTask.serializeFileSet(fileSet, stageId, ShuffleMapTask.fileSetCache)
     out.writeInt(fileSetBytes.length)
     out.write(fileSetBytes)
-    val jarSetBytes = ShuffleMapTask.serializeFileSet(jarSet)
+    val jarSetBytes = ShuffleMapTask.serializeFileSet(jarSet, stageId, ShuffleMapTask.jarSetCache)
     out.writeInt(jarSetBytes.length)
     out.write(jarSetBytes)