[llvm] [X86] Prefer `lock or` over mfence (PR #106555)

[Valentin Churavy via llvm-commits]

================
@@ -1429,7 +1444,8 @@ def ProcessorFeatures {
                                          TuningFastScalarShiftMasks,
                                          TuningBranchFusion,
                                          TuningSBBDepBreaking,
-                                         TuningInsertVZEROUPPER];
+                                         TuningInsertVZEROUPPER,
+                                         TuningAvoidMFENCE];
----------------
vchuravy wrote:

Done!

https://github.com/llvm/llvm-project/pull/106555