Bug 1509985: Backout diagnostic bug 1366083 - slight slowdown, and it has not found any errors, r=jonco

--HG--
extra : topic : hazard
extra : rebase_source : 5cbcd285bfb0fde14225a84b9107178fb70d5470
extra : amend_source : 0ed3afa45b1e1a9777d0a6e05e1cec46e447eb89

js/public/Value.h

@@ -481,18 +481,6 @@ union alignas(8) Value
     void setObject(JSObject& obj) {
         MOZ_ASSERT(js::gc::IsCellPointerValid(&obj));
 
-        // This should not be possible and is undefined behavior, but some
-        // ObjectValue(nullptr) are sneaking in. Try to catch them here, if
-        // indeed they are going through this code. I tested gcc, and it at
-        // least will *not* elide the null check even though it would be
-        // permitted according to the spec. The temporary is necessary to
-        // prevent gcc from helpfully pointing out that this code makes no
-        // sense.
-#ifdef MOZ_DIAGNOSTIC_ASSERT_ENABLED
-        JSObject* testObj = &obj;
-        MOZ_DIAGNOSTIC_ASSERT(testObj != nullptr);
-#endif
-
 #if defined(JS_PUNBOX64)
         // VisualStudio cannot contain parenthesized C++ style cast and shift
         // inside decltype in template parameter: