Oilpan: trace ColorChooserPopupUIController::m_chromeClient

third_party/WebKit/Source/web/ColorChooserPopupUIController.h

 /*
  * Copyright (C) 2011 Google Inc. 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.
  * 2.  Redistributions in binary form must reproduce the above copyright
@@ skipping 46 matching lines @@
     IntSize contentSize() override;
     void writeDocument(SharedBuffer*) override;
     void selectFontsFromOwnerDocument(Document&) override { }
     Locale& locale() override;
     void setValueAndClosePopup(int, const String&) override;
     void setValue(const String&) override;
     void closePopup() override;
     Element& ownerElement() override;
     void didClosePopup() override;
 
+    DEFINE_INLINE_VIRTUAL_TRACE()
+    {
+        visitor->trace(m_chromeClient);
+        ColorChooserUIController::trace(visitor);
+    }
+
 private:
     ColorChooserPopupUIController(LocalFrame*, ChromeClientImpl*, ColorChooserClient*);
 
     void openPopup();
 
-    ChromeClientImpl* m_chromeClient;
+    RawPtrWillBeMember<ChromeClientImpl> m_chromeClient;

[haraken, 2015/11/12 14:17:29]

It seems that the destructor is touching m_chromeClient in closePopup()...

[sof, 2015/11/13 07:35:01]

Good catch.

Hmm, given that ChromeClientImpl is held in a Persistent<> by WebViewImpl, it is
not too likely it will be swept in the same GC iteration. If it is, we're in the
middle of a global shutdown, so having closePopup() inform this client that it
is going away seems irrelevant.

=> Introducing a WeakMember<> here looks like the better option overall. Now
done.

     PagePopup* m_popup;
     Locale& m_locale;
 };
 
 } // namespace blink
 
 #endif // ColorChooserPopupUIController_h