Currently, it restores thread register a bit too late, that results crashes with Shenandoah GC on x86_32 JVM.

diff --git a/src/hotspot/cpu/x86/interp_masm_x86.cpp b/src/hotspot/cpu/x86/interp_masm_x86.cpp
index 9274f298759..9c1376a33c6 100644
--- a/src/hotspot/cpu/x86/interp_masm_x86.cpp
+++ b/src/hotspot/cpu/x86/interp_masm_x86.cpp
@@ -1005,9 +1005,9 @@ void InterpreterMacroAssembler::remove_activation(
   push(state);
   set_last_Java_frame(rthread, noreg, rbp, (address)pc());
   super_call_VM_leaf(CAST_FROM_FN_PTR(address, InterpreterRuntime::at_unwind), rthread);
+ NOT_LP64(get_thread(rthread);) // call_VM clobbered it, restore
   reset_last_Java_frame(rthread, true);
   pop(state);
- NOT_LP64(get_thread(rthread);) // call_VM clobbered it, restore
   bind(fast_path);