Changing so that the mailbox status is ack:ed after the _whole_ processing of the current batch, which means that Akka only does 1 volatile write per batch (of course the backing mailboxes might do their own volatile writes)

akka-actor/src/main/scala/akka/actor/ActorCell.scala

@@ -443,8 +443,6 @@ private[akka] class ActorCell(
         app.eventHandler.error(e, self, "error while processing " + envelope.message)
         //TODO FIXME How should problems here be handled?
         throw e
-    } finally {
-      mailbox.acknowledgeStatus() //Volatile write
     }
   }
 
@@ -487,8 +485,6 @@ private[akka] class ActorCell(
         messageHandle.channel sendException new ActorKilledException("Actor has been stopped")
         // throwing away message if actor is shut down, no use throwing an exception in receiving actor's thread, isShutdown is enforced on caller side
       }
-    } finally {
-      mailbox.acknowledgeStatus()
     }
   }
 

akka-actor/src/main/scala/akka/dispatch/Mailbox.scala

@@ -128,7 +128,10 @@ abstract class Mailbox extends AbstractMailbox with MessageQueue with SystemMess
      */
     if (s >= Scheduled) {
       updateStatus(s, s & ~Scheduled) || setAsIdle()
-    } else false
+    } else {
+      acknowledgeStatus() // this write is needed to make memory consistent after processMailbox()
+      false
+    }
   }
 
   def shouldBeRegisteredForExecution(hasMessageHint: Boolean, hasSystemMessageHint: Boolean): Boolean = status match {