[turbofan] Optimize loads from the global object in JSTypeFeedbackSpecializer.

src/compiler/pipeline.cc

Index: src/compiler/pipeline.cc
diff --git a/src/compiler/pipeline.cc b/src/compiler/pipeline.cc
index 022b6b5d1b2fb4eb24519d825f8548a2dde44d9c..2698c689c45a7e6e645f7a95a1b0fdd2c36830fa 100644
--- a/src/compiler/pipeline.cc
+++ b/src/compiler/pipeline.cc
@@ -542,8 +542,16 @@ struct JSTypeFeedbackPhase {
                               data->info()->unoptimized_code(),
                               data->info()->feedback_vector(), native_context);
     GraphReducer graph_reducer(data->graph(), temp_zone);
-    JSTypeFeedbackSpecializer specializer(data->jsgraph(),
-                                          data->js_type_feedback(), &oracle);
+    Handle<GlobalObject> global_object = Handle<GlobalObject>::null();
+    if (data->info()->has_global_object()) {
+      global_object =
+          Handle<GlobalObject>(data->info()->global_object(), data->isolate());
+    }
+    // TODO(titzer): introduce a specialization mode/flags enum to control
+    // specializing to the global object here.
+    JSTypeFeedbackSpecializer specializer(
+        data->jsgraph(), data->js_type_feedback(), &oracle, global_object,
+        data->info()->dependencies());
     AddReducer(data, &graph_reducer, &specializer);
     graph_reducer.ReduceGraph();
   }