Enable SVGImage to dispatch events in the destructor.

Source/core/svg/graphics/SVGImage.cpp

 /*
  * Copyright (C) 2006 Eric Seidel <eric@webkit.org>
  * Copyright (C) 2008, 2009 Apple Inc. All rights reserved.
  * Copyright (C) Research In Motion Limited 2011. All rights reserved.
  *
  * Redistribution and use in source and binary forms, with or without
  * modification, are permitted provided that the following conditions
  * are met:
  * 1. Redistributions of source code must retain the above copyright
  *    notice, this list of conditions and the following disclaimer.
@@ skipping 46 matching lines @@
 
 namespace blink {
 
 SVGImage::SVGImage(ImageObserver* observer)
     : Image(observer)
 {
 }
 
 SVGImage::~SVGImage()
 {
+    // willBeDestroyed() will fire an unload event against the SVGImage's window object
+    // that can't have listeners so we can call it in EventDispatchForbiddenScope scope.
+    EventDispatchForbiddenScope::AllowUserAgentEvents allowEvents;

[esprehn, 2015/01/09 00:55:06]

Why do you still need this if you have the SVGImage::isInSVGImage check in
closeURL()?

     if (m_page) {
         // Store m_page in a local variable, clearing m_page, so that SVGImageChromeClient knows we're destructed.
         OwnPtrWillBeRawPtr<Page> currentPage = m_page.release();
         // Break both the loader and view references to the frame
         currentPage->willBeDestroyed();
     }
 
     // Verify that page teardown destroyed the Chrome
     ASSERT(!m_chromeClient || !m_chromeClient->image());
 }
@@ skipping 359 matching lines @@
 
     return m_page;
 }
 
 String SVGImage::filenameExtension() const
 {
     return "svg";
 }
 
 }