[clang] [llvm] [AArch64] Implement the atomic store with hint intrinsic (PR #198316)

[via cfe-commits]

================
@@ -322,6 +323,111 @@ bool SemaARM::BuiltinARMSpecialReg(unsigned BuiltinID, CallExpr *TheCall,
   return false;
 }
 
+bool SemaARM::BuiltinARMAtomicStoreHintCall(unsigned BuiltinID,
+                                            CallExpr *TheCall) {
+  if (SemaRef.checkArgCount(TheCall, 4))
+    return true;
+
+  // Arg 0 should be the pointer type. The pointee type must be a
+  // scalar integral or floating-point type of 8, 16, 32 or 64 bits.
+  ASTContext &Context = getASTContext();
+  auto PtrArgRes =
+      SemaRef.DefaultFunctionArrayLvalueConversion(TheCall->getArg(0));
+  if (PtrArgRes.isInvalid())
+    return true;
+  auto *PtrArg = PtrArgRes.get();
+  auto *PtrTy = PtrArg->getType()->getAs<PointerType>();
+  if (!PtrTy)
+    return Diag(TheCall->getBeginLoc(),
+                diag::err_atomic_hint_builtin_must_be_pointer)
+           << PtrArg->getType() << 0 << PtrArg->getSourceRange();
+  TheCall->setArg(0, PtrArg);
+
+  QualType PtrQT =
+      Context.getCanonicalType(PtrTy->getPointeeType()).getUnqualifiedType();
----------------
Lukacma wrote:

I think keeping it qualified here will make if statement below simpler and you can get unqualified type when needed.

https://github.com/llvm/llvm-project/pull/198316