Run VDA::Decode on GPU IO thread if VDA supports it.

content/common/gpu/media/exynos_video_decode_accelerator.h

Index: content/common/gpu/media/exynos_video_decode_accelerator.h
diff --git a/content/common/gpu/media/exynos_video_decode_accelerator.h b/content/common/gpu/media/exynos_video_decode_accelerator.h
index 77471402ff0d61f259f228cc3d12403c332076e8..4610e239f8ac63e6c7cc10e361daf30e23eb5205 100644
--- a/content/common/gpu/media/exynos_video_decode_accelerator.h
+++ b/content/common/gpu/media/exynos_video_decode_accelerator.h
@@ -61,7 +61,8 @@ class CONTENT_EXPORT ExynosVideoDecodeAccelerator :
       EGLDisplay egl_display,
       EGLContext egl_context,
       Client* client,
-      const base::Callback<bool(void)>& make_context_current);
+      const base::Callback<bool(void)>& make_context_current,
+      const scoped_refptr<base::MessageLoopProxy>& io_message_loop_proxy);
   virtual ~ExynosVideoDecodeAccelerator();
 
   // media::VideoDecodeAccelerator implementation.
@@ -311,6 +312,9 @@ class CONTENT_EXPORT ExynosVideoDecodeAccelerator :
   // Our original calling message loop for the child thread.
   scoped_refptr<base::MessageLoopProxy> child_message_loop_proxy_;
 
+  // Message loop of the IO thread.
+  scoped_refptr<base::MessageLoopProxy> io_message_loop_proxy_;
+
   // WeakPtr<> pointing to |this| for use in posting tasks from the decoder or
   // device worker threads back to the child thread.  Because the worker threads
   // are members of this class, any task running on those threads is guaranteed