[CIR][CIRGen][Builtin][Neon] Lower neon_vsubd_s64 and neon_vsubd_u64 (#1256)

Lowering `neon_vsubd_s64` and `neon_vsubd_u64`

- [Clang CGBuiltin
Implementation](https://github.com/llvm/clangir/blob/2b1a638ea07ca10c5727ea835bfbe17b881175cc/clang/lib/CodeGen/CGBuiltin.cpp#L12463-L12465)
- [vaddd_s64 Builtin
definition](https://developer.arm.com/architectures/instruction-sets/intrinsics/vsubd_s64)
- [vaddd_u64 Builtin
definition](https://developer.arm.com/architectures/instruction-sets/intrinsics/vsubd_u64)

Author: AmrDeveloper

clang/lib/CIR/CodeGen/CIRGenBuiltinAArch64.cpp

@@ -3790,7 +3790,7 @@ CIRGenFunction::emitAArch64BuiltinExpr(unsigned BuiltinID, const CallExpr *E,
     return builder.createAdd(Ops[0], emitScalarExpr(E->getArg(1)));
   case NEON::BI__builtin_neon_vsubd_s64:
   case NEON::BI__builtin_neon_vsubd_u64:
-    llvm_unreachable("NEON::BI__builtin_neon_vsubd_u64 NYI");
+    return builder.createSub(Ops[0], emitScalarExpr(E->getArg(1)));
   case NEON::BI__builtin_neon_vqdmlalh_s16:
   case NEON::BI__builtin_neon_vqdmlslh_s16: {
     llvm_unreachable("NEON::BI__builtin_neon_vqdmlslh_s16 NYI");

clang/test/CIR/CodeGen/AArch64/neon.c

@@ -9905,19 +9905,29 @@ uint64_t test_vaddd_u64(uint64_t a, uint64_t b) {
   // LLVM:   ret i64 [[VADDD_I]]
 }
 
-// NYI-LABEL: @test_vsubd_s64(
-// NYI:   [[VSUBD_I:%.*]] = sub i64 %a, %b
-// NYI:   ret i64 [[VSUBD_I]]
-// int64_t test_vsubd_s64(int64_t a, int64_t b) {
-//   return vsubd_s64(a, b);
-// }
+int64_t test_vsubd_s64(int64_t a, int64_t b) {
+  return vsubd_s64(a, b);
 
-// NYI-LABEL: @test_vsubd_u64(
-// NYI:   [[VSUBD_I:%.*]] = sub i64 %a, %b
-// NYI:   ret i64 [[VSUBD_I]]
-// uint64_t test_vsubd_u64(uint64_t a, uint64_t b) {
-//   return vsubd_u64(a, b);
-// }
+  // CIR-LABEL: vsubd_s64
+  // CIR: {{%.*}} = cir.binop(sub, {{%.*}}, {{%.*}}) : !s64i
+
+  // LLVM-LABEL: @test_vsubd_s64
+  // LLVM-SAME: (i64 [[a:%.]], i64 [[b:%.]])
+  // LLVM:   [[VSUBD_I:%.*]]  = sub i64 [[a]], [[b]]
+  // LLVM:   ret i64 [[VSUBD_I]]
+}
+
+uint64_t test_vsubd_u64(uint64_t a, uint64_t b) {
+  return vsubd_u64(a, b);
+
+  // CIR-LABEL: vsubd_u64
+  // CIR: {{%.*}} = cir.binop(sub, {{%.*}}, {{%.*}}) : !u64i
+
+  // LLVM-LABEL: @test_vsubd_u64
+  // LLVM-SAME: (i64 [[a:%.]], i64 [[b:%.]])
+  // LLVM:   [[VSUBD_I:%.*]]  = sub i64 [[a]], [[b]]
+  // LLVM:   ret i64 [[VSUBD_I]]
+}
 
 // NYI-LABEL: @test_vqaddb_s8(
 // NYI:   [[TMP0:%.*]] = insertelement <8 x i8> poison, i8 %a, i64 0