[mips64] Fix loading optimization state

The optimization state is a 32-bit value, so we should load it by
lw instruction instead of ld.

Change-Id: I5845341ea9b87568fda08180ed51ad1a5c0635e0
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2507312Reviewed-by: Jakob Gruber <jgruber@chromium.org>
Commit-Queue: Jakob Gruber <jgruber@chromium.org>
Auto-Submit: Liu yu <liuyu@loongson.cn>
Cr-Commit-Position: refs/heads/master@{#70879}

src/builtins/mips64/builtins-mips64.cc

@@ -1033,7 +1033,7 @@ void Builtins::Generate_InterpreterEntryTrampoline(MacroAssembler* masm) {
   // Read off the optimization state in the feedback vector, and if there
   // is optimized code or an optimization marker, call that instead.
   Register optimization_state = a4;
-  __ Ld(optimization_state,
+  __ Lw(optimization_state,
         FieldMemOperand(feedback_vector, FeedbackVector::kFlagsOffset));
 
   // Check if the optimized code slot is not empty or has a optimization marker.