[flang][cuda][NFC] Use the NVVM op for syncwarp (#166695)

Author: clementval

flang/lib/Optimizer/Builder/CUDAIntrinsicCall.cpp

@@ -1122,13 +1122,7 @@ CUDAIntrinsicLibrary::genSyncThreadsOr(mlir::Type resultType,
 void CUDAIntrinsicLibrary::genSyncWarp(
     llvm::ArrayRef<fir::ExtendedValue> args) {
   assert(args.size() == 1);
-  constexpr llvm::StringLiteral funcName = "llvm.nvvm.bar.warp.sync";
-  mlir::Value mask = fir::getBase(args[0]);
-  mlir::FunctionType funcType =
-      mlir::FunctionType::get(builder.getContext(), {mask.getType()}, {});
-  auto funcOp = builder.createFunction(loc, funcName, funcType);
-  llvm::SmallVector<mlir::Value> argsList{mask};
-  fir::CallOp::create(builder, loc, funcOp, argsList);
+  mlir::NVVM::SyncWarpOp::create(builder, loc, fir::getBase(args[0]));
 }
 
 // THIS_GRID

flang/test/Lower/CUDA/cuda-device-proc.cuf

@@ -105,7 +105,7 @@ end
 
 ! CHECK-LABEL: func.func @_QPdevsub() attributes {cuf.proc_attr = #cuf.cuda_proc<global>}
 ! CHECK: nvvm.barrier0
-! CHECK: fir.call @llvm.nvvm.bar.warp.sync(%c1{{.*}}) fastmath<contract> : (i32) -> ()
+! CHECK: nvvm.bar.warp.sync %c1{{.*}} : i32 
 ! CHECK: fir.call @llvm.nvvm.membar.gl() fastmath<contract> : () -> ()
 ! CHECK: fir.call @llvm.nvvm.membar.cta() fastmath<contract> : () -> ()
 ! CHECK: fir.call @llvm.nvvm.membar.sys() fastmath<contract> : () -> ()
@@ -219,7 +219,7 @@ end
 ! CHECK-LABEL: func.func @_QPhost1()
 ! CHECK: cuf.kernel
 ! CHECK: nvvm.barrier0
-! CHECK: fir.call @llvm.nvvm.bar.warp.sync(%c1{{.*}}) fastmath<contract> : (i32) -> ()
+! CHECK: nvvm.bar.warp.sync %c1{{.*}} : i32 
 ! CHECK: fir.call @llvm.nvvm.barrier0.and(%c1{{.*}}) fastmath<contract> : (i32) -> i32
 ! CHECK: fir.call @llvm.nvvm.barrier0.popc(%c1{{.*}}) fastmath<contract> : (i32) -> i32
 ! CHECK: fir.call @llvm.nvvm.barrier0.or(%c1{{.*}}) fastmath<contract> : (i32) -> i32