[llvm] [AMDGPU][gfx1250] Also add a wait on xcnt before volatile accesses (PR #168852)

[Matt Arsenault via llvm-commits]

================
@@ -1872,8 +1872,10 @@ class GCNSubtarget final : public AMDGPUGenSubtargetInfo,
   bool hasClusters() const { return HasClusters; }
 
   /// \returns true if the subtarget requires a wait for xcnt before atomic
-  /// flat/global stores & rmw.
-  bool requiresWaitXCntBeforeAtomicStores() const { return GFX1250Insts; }
+  /// stores and all volatile accesses for all isFLAT operations.
+  bool requiresWaitXCntBeforeAtomicStoreOrVolatileAccesses() const {
----------------
arsenm wrote:

```suggestion
  bool requiresWaitXCntBeforeOrderedAccesses() const {
```

https://github.com/llvm/llvm-project/pull/168852