[PresentationAPI, Android] Add ChromeMediaRouterDialogController.java and bind it with the native M…

chrome/browser/media/android/router/media_router_dialog_controller_android.cc

Index: chrome/browser/media/android/router/media_router_dialog_controller_android.cc
diff --git a/chrome/browser/media/android/router/media_router_dialog_controller_android.cc b/chrome/browser/media/android/router/media_router_dialog_controller_android.cc
index 88d1fd933eb628f3cc4114cf26302cc3fbd80b7b..ac31a972dff831cf8000d149026d9e7f1bc83223 100644
--- a/chrome/browser/media/android/router/media_router_dialog_controller_android.cc
+++ b/chrome/browser/media/android/router/media_router_dialog_controller_android.cc
@@ -4,9 +4,11 @@
 
 #include "chrome/browser/media/android/router/media_router_dialog_controller_android.h"
 
+#include "base/android/jni_android.h"
 #include "content/public/browser/browser_thread.h"
 #include "content/public/browser/web_contents.h"
 #include "content/public/browser/web_contents_delegate.h"
+#include "jni/ChromeMediaRouterDialogController_jni.h"
 
 DEFINE_WEB_CONTENTS_USER_DATA_KEY(
     media_router::MediaRouterDialogControllerAndroid);
@@ -28,6 +30,19 @@ MediaRouterDialogControllerAndroid::GetOrCreateForWebContents(
 MediaRouterDialogControllerAndroid::MediaRouterDialogControllerAndroid(
     WebContents* web_contents)
     : MediaRouterDialogController(web_contents) {
+  JNIEnv* env = base::android::AttachCurrentThread();
+  CHECK(env);

[Bernhard Bauer, 2015/08/06 15:23:01]

Is this CHECK actually necessary? I would only expect this to fail under very
exceptional circumstances, and then you'd presumably crash anyway.

[whywhat, 2015/08/06 15:27:13]

I'd think in such exceptional circumstances one should crash earlier than later.
IIRC it's the common thing to do after acquiring |env| in Chromium. Have this
practice changed?

[Bernhard Bauer, 2015/08/06 15:40:33]

I have never seen it, TBH. If anything, we should check directly in
AttachCurrentThread(), and we usually do DCHECKs for these sorts of things
(CHECKs are mostly used to turn an otherwise unclean memory access into a clean
crash, for security reasons, but dereferencing a null pointer is okay). And
looking at AttachCurrentThread(), there is already a DCHECK for the result of
JavaVM::AttachCurrentThread().

[whywhat, 2015/08/06 16:29:20]

Ok, CHECK removed.

+  java_dialog_controller_.Reset(Java_ChromeMediaRouterDialogController_create(
+      env,
+      reinterpret_cast<intptr_t>(this),
+      base::android::GetApplicationContext()));
+}
+
+// static
+bool MediaRouterDialogControllerAndroid::Register(JNIEnv* env) {
+  bool ret = RegisterNativesImpl(env);
+  DCHECK(g_ChromeMediaRouterDialogController_clazz);
+  return ret;
 }
 
 MediaRouterDialogControllerAndroid::~MediaRouterDialogControllerAndroid() {