[clang] [CIR][AArch64] Add lowering for unpredicated svdup builtins (PR #174433)

[Andrzej Warzyński via cfe-commits]

================
@@ -52,6 +52,51 @@ static mlir::Value genVscaleTimesFactor(mlir::Location loc,
                                builder.getUInt64(scalingFactor, loc));
 }
 
+static bool aarch64SVEIntrinsicsProvenSorted = false;
+
+namespace {
+struct aarc64BuiltinInfo {
+  unsigned builtinID;
+  unsigned LLVMIntrinsic;
+
+  bool operator<(unsigned RHSbuiltinID) const {
+    return builtinID < RHSbuiltinID;
+  }
+  bool operator<(const aarc64BuiltinInfo &TE) const {
+    return builtinID < TE.builtinID;
+  }
+};
+} // end anonymous namespace
+
+#define SVEMAP1(NameBase, LLVMIntrinsic, TypeModifier)                         \
----------------
banach-space wrote:

It's not needed for this specific builtin, but we will use it later. Let me refactor this so that we start using it immediately.

https://github.com/llvm/llvm-project/pull/174433