[clang] [llvm] AMDGPU: Add builtin/intrinsic global_(load|store)_b128 (PR #172090)

[Matt Arsenault via cfe-commits]

================
@@ -344,6 +347,19 @@ bool SemaAMDGPU::checkCoopAtomicFunctionCall(CallExpr *TheCall, bool IsStore) {
   return Fail;
 }
 
+bool SemaAMDGPU::checkScopedMemAccessFunctionCall(CallExpr *TheCall) {
+  bool Fail = false;
+  // Last argument is a string literal
+  Expr *Arg = TheCall->getArg(TheCall->getNumArgs() - 1);
+  auto Scope = dyn_cast<StringLiteral>(Arg->IgnoreParenCasts());
+  if (!Scope) {
+    Fail = true;
----------------
arsenm wrote:

Return and avoid intermediate Fail variable 

https://github.com/llvm/llvm-project/pull/172090