Fixed expando-loss.html test.

third_party/WebKit/Source/core/html/HTMLCanvasElement.h

Index: third_party/WebKit/Source/core/html/HTMLCanvasElement.h
diff --git a/third_party/WebKit/Source/core/html/HTMLCanvasElement.h b/third_party/WebKit/Source/core/html/HTMLCanvasElement.h
index f8b1b19abcd8ac856b20ee4c9e9f5132794baa76..e34301e8b4e55796c3be3c6f575f319193cc5004 100644
--- a/third_party/WebKit/Source/core/html/HTMLCanvasElement.h
+++ b/third_party/WebKit/Source/core/html/HTMLCanvasElement.h
@@ -110,7 +110,8 @@ public:
     // Called by HTMLCanvasElement's V8 bindings.
     ScriptValue getContext(ScriptState*, const String&, const CanvasContextCreationAttributes&);
     // Called by Document::getCSSCanvasContext as well as above getContext().
-    CanvasRenderingContext* getCanvasRenderingContext(const String&, const CanvasContextCreationAttributes&);
+    // The WebGLRenderingContext requires a non-null ScriptState, but the Canvas2DRenderingContext does not.

[haraken, 2015/10/13 02:02:04]

In general, it is not nice to pass a nullptr to ScriptState. If this is a
testing-only issue, you can use V8TestingScope which gives you a ScriptState to
use.

+    CanvasRenderingContext* getCanvasRenderingContext(ScriptState*, const String&, const CanvasContextCreationAttributes&);
 
     bool isPaintable() const;