[X86] Use OR32mi8Locked instead of LOCK_OR32mi8 in emitLockedStackOp.

They encode the same way, but OR32mi8Locked sets hasUnmodeledSideEffects set
which should be stronger than the mayLoad/mayStore on LOCK_OR32mi8. I think
this makes sense since we are using it as a fence.

This also seems to hide the operation from the speculative load hardening pass
so I've reverted r360511.

llvm-svn: 360747
diff --git a/llvm/lib/Target/X86/X86SpeculativeLoadHardening.cpp b/llvm/lib/Target/X86/X86SpeculativeLoadHardening.cpp
index 7b04337..02f07d8 100644
--- a/llvm/lib/Target/X86/X86SpeculativeLoadHardening.cpp
+++ b/llvm/lib/Target/X86/X86SpeculativeLoadHardening.cpp
@@ -1719,11 +1719,9 @@
 
         // If we have at least one (non-frame-index, non-RIP) register operand,
         // and neither operand is load-dependent, we need to check the load.
-        // Also handle explicit references to RSP as used by idempotent atomic
-        // or with 0.
         unsigned BaseReg = 0, IndexReg = 0;
         if (!BaseMO.isFI() && BaseMO.getReg() != X86::RIP &&
-            BaseMO.getReg() != X86::RSP && BaseMO.getReg() != X86::NoRegister)
+            BaseMO.getReg() != X86::NoRegister)
           BaseReg = BaseMO.getReg();
         if (IndexMO.getReg() != X86::NoRegister)
           IndexReg = IndexMO.getReg();