Explicit notification and removal of lifecycle observers.

Source/core/workers/WorkerThread.cpp

Index: Source/core/workers/WorkerThread.cpp
diff --git a/Source/core/workers/WorkerThread.cpp b/Source/core/workers/WorkerThread.cpp
index c60511544ab3c579e55ce404dcde5ef77705a1b8..e711501aca3a51f6245ac520296fdf3dfdd0b2f1 100644
--- a/Source/core/workers/WorkerThread.cpp
+++ b/Source/core/workers/WorkerThread.cpp
@@ -135,18 +135,20 @@ void WorkerThread::workerThread()
 
     ThreadIdentifier threadID = m_threadID;
 
+#if !ENABLE(OILPAN)
+    ASSERT(m_workerGlobalScope->hasOneRef());
+#endif
     // The below assignment will destroy the context, which will in turn notify messaging proxy.
     // We cannot let any objects survive past thread exit, because no other thread will run GC or otherwise destroy them.
     // If Oilpan is enabled, we detach of the context/global scope, with the final heap cleanup below sweeping it out.
-#if ENABLE(OILPAN)
     m_workerGlobalScope->dispose();
-#else
-    ASSERT(m_workerGlobalScope->hasOneRef());
-#endif
     m_workerGlobalScope = nullptr;
 
     ThreadState::detach();

[haraken, 2014/03/17 01:22:56]

Shall we move ThreadState::detach() at the very end of this method? A thread
should be detached at the very end of the thread execution.

[haraken, 2014/03/17 01:25:45]

As far as I read your comment in
https://codereview.chromium.org/196033004/#msg6, probably do you need to call
ThreadState::detach() before
workerReportingProxy().workerGlobalScopeDestroyed()? Then I might want to have a
comment about that.

[sof, 2014/03/17 07:46:41]

Added a comment, which might address this? (not sure..) Notification has to come
after the (ThreadState) detach.

 
+    // Notify proxy that WorkerGlobalScope is now gone. This can free the WorkerThread object, so do not access it after this.
+    workerReportingProxy().workerGlobalScopeDestroyed();
+
     // Clean up PlatformThreadData before WTF::WTFThreadData goes away!
     PlatformThreadData::current().destroy();