[clang-tidy] enable modernize-use-override (#55491)

.clang-tidy

@@ -185,7 +185,7 @@ Checks: '
 -modernize-use-equals-delete,
 -modernize-use-noexcept,
 -modernize-use-nullptr,
--modernize-use-override,
+modernize-use-override,
 -modernize-use-transparent-functors,
 -modernize-use-uncaught-exceptions,
 -performance-faster-string-find,

paddle/fluid/framework/ir/coalesce_grad_tensor_pass.cc

@@ -69,7 +69,7 @@ double GetFuseParameterMemorySize() { return FLAGS_fuse_parameter_memory_size; }
 
 class CoalesceGradTensorPass : public ir::Pass {
  protected:
-  void ApplyImpl(ir::Graph *graph) const {
+  void ApplyImpl(ir::Graph *graph) const override {
     if (Get<size_t>(details::kNRanks) <= 1) {
       VLOG(6) << "The number of place is" << Get<size_t>(details::kNRanks)
               << ", there doesn't need apply FuseAllReduceOpPass.";

paddle/fluid/framework/ir/fuse_optimizer_ops_pass/fuse_adam_op_pass.cc

@@ -30,9 +30,9 @@ class Node;
 
 class FuseAdamOpPass : public FuseOptimizerOpPass {
  private:
-  const std::string GetOpType() const { return "adam"; }
+  const std::string GetOpType() const override { return "adam"; }
 
-  const std::vector<std::string> GetAuxiliaryVarNames() const {
+  const std::vector<std::string> GetAuxiliaryVarNames() const override {
     return {"Moment1", "Moment2", "Beta1Pow", "Beta2Pow"};
   }
 
@@ -41,7 +41,7 @@ class FuseAdamOpPass : public FuseOptimizerOpPass {
           &aux_var_set,
       const std::unordered_map<std::string, std::string> &fused_vars_name,
       const std::vector<ir::Node *> &adam_ops,
-      ir::Graph *graph) const {
+      ir::Graph *graph) const override {
     auto fused_adam_node =
         FuseAdamOps(aux_var_set, fused_vars_name, adam_ops, graph);
     return fused_adam_node;

paddle/fluid/framework/ir/fuse_optimizer_ops_pass/fuse_momentum_op_pass.cc

@@ -29,18 +29,18 @@ class Node;
 
 class FuseMomentumOpPass : public FuseOptimizerOpPass {
  private:
-  virtual const std::string GetOpType() const { return "momentum"; }
+  const std::string GetOpType() const override { return "momentum"; }
 
-  virtual const std::vector<std::string> GetAuxiliaryVarNames() const {
+  const std::vector<std::string> GetAuxiliaryVarNames() const override {
     return {"Velocity"};
   }
 
   // Fuse Momentum Ops
-  virtual ir::Node *FuseOptimizerOps(
+  ir::Node *FuseOptimizerOps(
       const std::unordered_map<std::string, std::vector<std::string>> &vars_set,
       const std::unordered_map<std::string, std::string> &fused_vars_name,
       const std::vector<ir::Node *> &momentum_ops,
-      ir::Graph *graph) const {
+      ir::Graph *graph) const override {
     PADDLE_ENFORCE_GT(
         momentum_ops.size(),
         static_cast<size_t>(0),

paddle/fluid/framework/ir/fuse_optimizer_ops_pass/fuse_sgd_op_pass.cc

@@ -28,18 +28,18 @@ class Node;
 
 class FuseSgdOpPass : public FuseOptimizerOpPass {
  private:
-  virtual const std::string GetOpType() const { return "sgd"; }
+  const std::string GetOpType() const override { return "sgd"; }
 
-  virtual const std::vector<std::string> GetAuxiliaryVarNames() const {
+  const std::vector<std::string> GetAuxiliaryVarNames() const override {
     return {};
   }
 
   // Fuse Sgd Ops
-  virtual ir::Node *FuseOptimizerOps(
+  ir::Node *FuseOptimizerOps(
       const std::unordered_map<std::string, std::vector<std::string>> &vars_set,
       const std::unordered_map<std::string, std::string> &fused_vars_name,
       const std::vector<ir::Node *> &sgd_ops,
-      ir::Graph *graph) const {
+      ir::Graph *graph) const override {
     PADDLE_ENFORCE_GT(
         sgd_ops.size(),
         static_cast<size_t>(0),

paddle/fluid/framework/new_executor/workqueue/workqueue.cc

@@ -53,7 +53,7 @@ class WorkQueueImpl : public WorkQueue {
                                        options_.always_spinning);
   }
 
-  virtual ~WorkQueueImpl() {
+  ~WorkQueueImpl() override {
     delete queue_;
     if (tracker_ != nullptr) {
       tracker_->~TaskTracker();
@@ -94,7 +94,7 @@ class WorkQueueGroupImpl : public WorkQueueGroup {
   explicit WorkQueueGroupImpl(
       const std::vector<WorkQueueOptions>& queue_options);
 
-  ~WorkQueueGroupImpl();
+  ~WorkQueueGroupImpl() override;
 
   void AddTask(size_t queue_idx, std::function<void()> fn) override;
 

paddle/fluid/framework/phi_utils.cc

@@ -40,7 +40,7 @@ class KernelArgsNameMakerByOpProto : public KernelArgsNameMaker {
         platform::errors::InvalidArgument("Op proto cannot be nullptr."));
   }
 
-  ~KernelArgsNameMakerByOpProto() {}
+  ~KernelArgsNameMakerByOpProto() override {}
 
   const paddle::small_vector<const char*>& GetInputArgsNames() override;
   const paddle::small_vector<const char*>& GetOutputArgsNames() override;

paddle/fluid/memory/allocation/allocator_facade.cc

@@ -127,7 +127,7 @@ class CUDAGraphAllocator
       : underlying_allocator_(allocator) {}
 
  public:
-  ~CUDAGraphAllocator() {}
+  ~CUDAGraphAllocator() override {}
 
   static std::shared_ptr<Allocator> Create(
       const std::shared_ptr<Allocator>& allocator) {
@@ -135,14 +135,14 @@ class CUDAGraphAllocator
   }
 
  protected:
-  phi::Allocation* AllocateImpl(size_t size) {
+  phi::Allocation* AllocateImpl(size_t size) override {
     VLOG(10) << "Allocate " << size << " for CUDA Graph";
     return new PrivateAllocation(this,
                                  static_unique_ptr_cast<Allocation>(
                                      underlying_allocator_->Allocate(size)));
   }
 
-  void FreeImpl(phi::Allocation* allocation) {
+  void FreeImpl(phi::Allocation* allocation) override {
     VLOG(10) << "delete for CUDA Graph";
     delete allocation;
   }

paddle/fluid/operators/collective/allreduce_op.cc

@@ -36,7 +36,7 @@ class AllReduceDelOp : public framework::OperatorWithKernel {
 
 class AllReduceDelOpMaker : public framework::OpProtoAndCheckerMaker {
  public:
-  void Make() {
+  void Make() override {
     AddInput("X", "(Tensor), tensor to be allreduced.");
     AddOutput("Out", "(Tensor) the result of allreduced.");
     AddAttr<int>("reduce_type", "(int) determine the reduce type.")

paddle/fluid/operators/collective/alltoall_op.cc

@@ -45,7 +45,7 @@ class AllToAllBaseOp : public framework::OperatorWithKernel {
 
 class AllToAllBaseOpMaker : public framework::OpProtoAndCheckerMaker {
  public:
-  void Make() {
+  void Make() override {
     AddInput("X", "(Tensor) tensor send.");
     AddOutput("Out", "(Tensor) the result of alltoall.");
     AddAttr<int>("ring_id", "(int default 0) nccl communication ring id.")

paddle/fluid/operators/collective/barrier_op.cc

@@ -27,7 +27,7 @@ class BarrierOp : public framework::OperatorWithKernel {
 
 class BarrierOpMaker : public framework::OpProtoAndCheckerMaker {
  public:
-  void Make() {
+  void Make() override {
     AddInput("X", "(Tensor) Input data (only used in CUDAKernel).");
     AddOutput("Out", "(Tensor) Output data (only used in CUDAKernel).");
     AddAttr<int>("ring_id", "(int default 0) communication ring id.")

paddle/fluid/operators/collective/c_allgather_op.cc

@@ -44,7 +44,7 @@ class CAllGatherOp : public framework::OperatorWithKernel {
 
 class CAllGatherOpMaker : public framework::OpProtoAndCheckerMaker {
  public:
-  void Make() {
+  void Make() override {
     AddInput("X", "(Tensor) tensor to be allgather");
     AddOutput("Out", "(Tensor) the allgather result");
     AddAttr<int>("ring_id", "(int default 0) communication ring id.")

paddle/fluid/operators/collective/c_broadcast_op.cc

@@ -35,7 +35,7 @@ class CBroadcastOp : public framework::OperatorWithKernel {
 
 class CBroadcastOpMaker : public framework::OpProtoAndCheckerMaker {
  public:
-  void Make() {
+  void Make() override {
     AddInput("X", "(Tensor) tensor to be broadcasted.");
     AddOutput("Out", "(Tensor) the result of broadcast.");
     AddAttr<int>("ring_id", "(int default 0) nccl communication ring id.")

paddle/fluid/operators/collective/c_comm_init_all_op.cc

@@ -38,7 +38,7 @@ namespace operators {
 
 class CCommInitAllInferShape : public framework::InferShapeBase {
  public:
-  ~CCommInitAllInferShape() {}
+  ~CCommInitAllInferShape() override {}
   void operator()(framework::InferShapeContext* ctx) const override{};
 };
 

paddle/fluid/operators/collective/c_comm_init_multitrainer_op.cc

@@ -38,7 +38,7 @@ namespace operators {
 
 class CCommInitMultiTrainerInferShape : public framework::InferShapeBase {
  public:
-  ~CCommInitMultiTrainerInferShape() {}
+  ~CCommInitMultiTrainerInferShape() override {}
   void operator()(framework::InferShapeContext* ctx) const override{};
 };
 

paddle/fluid/operators/collective/c_concat_op.cc

@@ -81,7 +81,7 @@ class CConcatOpGradMaker : public framework::SingleGradOpMaker<T> {
 
 class CConcatOpMaker : public framework::OpProtoAndCheckerMaker {
  public:
-  void Make() {
+  void Make() override {
     AddInput("X", "(Tensor) tensor to be concated.");
     AddOutput("Out", "(Tensor) the result of concat.");
     AddAttr<int>("rank", "(int default 0) rank id.").SetDefault(0);

paddle/fluid/operators/collective/c_identity_op.cc

@@ -44,7 +44,7 @@ class CIdentityOp : public framework::OperatorWithKernel {
 
 class CIdentityOpMaker : public framework::OpProtoAndCheckerMaker {
  public:
-  void Make() {
+  void Make() override {
     AddInput("X", "(Tensor) identity tensor.");
     AddOutput("Out", "(Tensor) identity tensor.");
     AddAttr<int>("ring_id", "(int default 0) nccl communication ring id.")

paddle/fluid/operators/collective/c_reducescatter_op.cc

@@ -42,7 +42,7 @@ class CReduceScatterOp : public framework::OperatorWithKernel {
 
 class CReduceScatterOpMaker : public framework::OpProtoAndCheckerMaker {
  public:
-  void Make() {
+  void Make() override {
     AddInput("X", "(Tensor) tensor to be allgather");
     AddOutput("Out", "(Tensor) the allgather result");
     AddAttr<int>("ring_id", "(int default 0) communication ring id.")

paddle/fluid/operators/collective/c_scatter_op.cc

@@ -61,7 +61,7 @@ class CScatterOp : public framework::OperatorWithKernel {
 
 class CScatterOpMaker : public framework::OpProtoAndCheckerMaker {
  public:
-  void Make() {
+  void Make() override {
     AddInput("X", "(Tensor) tensor to be broadcasted.");
     AddOutput("Out", "(Tensor) the result of broadcast.");
     AddAttr<int>("ring_id", "(int default 0) nccl communication ring id.")

paddle/fluid/operators/collective/c_softmax_with_cross_entropy_op.cc

@@ -83,7 +83,7 @@ class CSoftmaxWithCrossEntropyOp : public framework::OperatorWithKernel {
 class CSoftmaxWithCrossEntropyOpMaker
     : public framework::OpProtoAndCheckerMaker {
  public:
-  void Make() {
+  void Make() override {
     AddInput("Logits",
              "(Tensor, default: Tensor<float>), The input tensor of unscaled "
              "log probabilities, whose dimension :attr:`axis` should be scaled "

paddle/fluid/operators/collective/c_split_op.cc

@@ -89,7 +89,7 @@ class CSplitOpGradMaker : public framework::SingleGradOpMaker<T> {
 
 class CSplitOpMaker : public framework::OpProtoAndCheckerMaker {
  public:
-  void Make() {
+  void Make() override {
     AddInput("X", "(Tensor) tensor to be split.");
     AddOutput("Out", "(Tensor) the result of split.");
     AddAttr<int>("rank", "(int default 0) rank id.").SetDefault(0);

paddle/fluid/operators/collective/c_sync_calc_stream_op.cc

@@ -18,7 +18,7 @@ namespace operators {
 
 class CSyncCalcStreamOpMaker : public framework::OpProtoAndCheckerMaker {
  public:
-  void Make() {
+  void Make() override {
     AddInput("X", "(Tensor) Dependency of the variable need to sync");
     AddOutput("Out", "(Tensor) Dependency of the variable need to sync");
     AddComment(R"DOC(

paddle/fluid/operators/collective/c_sync_comm_stream_op.cc

@@ -31,7 +31,7 @@ class CSyncCommStreamOp : public framework::OperatorWithKernel {
 
 class CSyncCommStreamOpMaker : public framework::OpProtoAndCheckerMaker {
  public:
-  void Make() {
+  void Make() override {
     AddInput("X", "(Tensor) Dependency of the variable need to sync")
         .AsDuplicable();
     AddOutput("Out", "(Tensor) Dependency of the variable need to sync")

paddle/fluid/operators/collective/c_wait_comm_op.cc

@@ -73,7 +73,7 @@ class CWaitCommOp : public framework::OperatorBase {
 
 class CWaitCommOpMaker : public framework::OpProtoAndCheckerMaker {
  public:
-  void Make() {
+  void Make() override {
     AddInput("X", "(Tensor) Dependency of the variable need to sync")
         .AsDuplicable();
     AddOutput("Out", "(Tensor) Dependency of the variable need to sync")

paddle/fluid/operators/collective/c_wait_compute_op.cc

@@ -74,7 +74,7 @@ class CWaitComputeOp : public framework::OperatorBase {
 
 class CWaitComputeOpMaker : public framework::OpProtoAndCheckerMaker {
  public:
-  void Make() {
+  void Make() override {
     AddInput("X", "(Tensor) Dependency of the variable need to sync")
         .AsDuplicable();
     AddOutput("Out", "(Tensor) Dependency of the variable need to sync")

paddle/fluid/operators/collective/global_gather_op.cc

@@ -58,7 +58,7 @@ class GlobalGatherOp : public framework::OperatorWithKernel {
 
 class GlobalGatherOpMaker : public framework::OpProtoAndCheckerMaker {
  public:
-  void Make() {
+  void Make() override {
     AddInput("X", "(Tensor) tensor send.");
     AddInput("local_count",
              "(Tensor) Tensor which has n_expert * world_size elements that "

paddle/fluid/operators/collective/global_scatter_op.cc

@@ -61,7 +61,7 @@ class GlobalScatterOp : public framework::OperatorWithKernel {
 
 class GlobalScatterOpMaker : public framework::OpProtoAndCheckerMaker {
  public:
-  void Make() {
+  void Make() override {
     AddInput("X", "(Tensor) tensor send.");
     AddInput("local_count",
              "(Tensor) Tensor which has n_expert * world_size elements that "

paddle/fluid/operators/collective/mp_allreduce_sum_op.cc

@@ -37,7 +37,7 @@ class MpAllReduceSumOp : public framework::OperatorWithKernel {
 
 class MpAllReduceSumOpMaker : public framework::OpProtoAndCheckerMaker {
  public:
-  void Make() {
+  void Make() override {
     AddInput("X", "(Tensor), tensor to be allreduced in model parallel.");
     AddOutput("Out", "(Tensor) the allreduced result in model parallel.");
     AddAttr<int>("ring_id", "(int default 0) communication ring id.")

paddle/fluid/operators/collective/partial_allgather_op.cc

@@ -45,7 +45,7 @@ class PartialAllGatherOp : public framework::OperatorWithKernel {
 
 class PartialAllGatherOpMaker : public framework::OpProtoAndCheckerMaker {
  public:
-  void Make() {
+  void Make() override {
     AddInput("X", "(Tensor) tensor to be partial allgather");
     AddOutput("Out", "(Tensor) the allgather result");
     AddAttr<int>("ring_id", "(int default 0) communication ring id.")

paddle/fluid/operators/collective/partial_recv_op.cc

@@ -91,7 +91,7 @@ class PartialRecvOp : public framework::OperatorWithKernel {
 
 class PartialRecvOpMaker : public framework::OpProtoAndCheckerMaker {
  public:
-  void Make() {
+  void Make() override {
     AddOutput("Out", "(Tensor) tensor to receive.");
     AddAttr<int>("ring_id", "(int default 0) nccl communication ring id.")
         .SetDefault(0);

paddle/fluid/operators/collective/partial_send_op.cc

@@ -60,7 +60,7 @@ class PartialSendOp : public framework::OperatorWithKernel {
 
 class PartialSendMaker : public framework::OpProtoAndCheckerMaker {
  public:
-  void Make() {
+  void Make() override {
     AddInput("X", "(Tensor) tensor to be sent.");
     AddAttr<int>("ring_id", "(int default 0) nccl communication ring id.")
         .SetDefault(0);

paddle/fluid/operators/collective/recv_v2_op.cc

@@ -80,7 +80,7 @@ class RecvOpV2 : public framework::OperatorWithKernel {
 
 class RecvOpV2Maker : public framework::OpProtoAndCheckerMaker {
  public:
-  void Make() {
+  void Make() override {
     AddOutput("Out", "(Tensor) tensor to receive.");
     AddAttr<int>("ring_id", "(int default 0) nccl communication ring id.")
         .SetDefault(0);

paddle/fluid/operators/collective/send_v2_op.cc

@@ -56,7 +56,7 @@ class SendOpV2 : public framework::OperatorWithKernel {
 
 class SendOpV2Maker : public framework::OpProtoAndCheckerMaker {
  public:
-  void Make() {
+  void Make() override {
     AddInput("X", "(Tensor) tensor to be sent.");
     AddAttr<int>("ring_id", "(int default 0) nccl communication ring id.")
         .SetDefault(0);

paddle/fluid/operators/fused/fused_gemm_epilogue_op.cc

@@ -127,7 +127,7 @@ class FusedGemmEpilogueOp : public framework::OperatorWithKernel {
   }
 
   phi::KernelKey GetExpectedKernelType(
-      const framework::ExecutionContext& ctx) const {
+      const framework::ExecutionContext& ctx) const override {
     auto data_type = OperatorWithKernel::IndicateVarDataType(ctx, "X");
     return phi::KernelKey(data_type, ctx.GetPlace());
   }
@@ -277,7 +277,7 @@ class FusedGemmEpilogueGradOp : public framework::OperatorWithKernel {
   }
 
   phi::KernelKey GetExpectedKernelType(
-      const framework::ExecutionContext& ctx) const {
+      const framework::ExecutionContext& ctx) const override {
     auto data_type = OperatorWithKernel::IndicateVarDataType(ctx, "DOut");
     return phi::KernelKey(data_type, ctx.GetPlace());
   }

paddle/fluid/operators/fused/resnet_unit_op.cc

@@ -35,7 +35,7 @@ class ResNetUnitOp : public framework::OperatorWithKernel {
  public:
   using framework::OperatorWithKernel::OperatorWithKernel;
 
-  void InferShape(framework::InferShapeContext* ctx) const {
+  void InferShape(framework::InferShapeContext* ctx) const override {
     // Check input
     OP_INOUT_CHECK(ctx->HasInput("X"), "Input", "X", "ResNetUnitOp");
     OP_INOUT_CHECK(
@@ -201,7 +201,7 @@ class ResNetUnitOp : public framework::OperatorWithKernel {
 
  protected:
   phi::KernelKey GetExpectedKernelType(
-      const framework::ExecutionContext& ctx) const {
+      const framework::ExecutionContext& ctx) const override {
     auto input_data_type = OperatorWithKernel::IndicateVarDataType(ctx, "X");
     // By default, the type of the scale, bias, mean,
     // and var tensors should be float when input tensor's dtype is float16.
@@ -223,7 +223,7 @@ class ResNetUnitOp : public framework::OperatorWithKernel {
 
 class ResNetUnitOpMaker : public framework::OpProtoAndCheckerMaker {
  public:
-  void Make() {
+  void Make() override {
     AddInput("X", "The input 1 tensor");
     AddInput("FilterX", "Filter tensor of input 1");
     AddInput("ScaleX", "Scale tensor of input 1 used in batchnorm");
@@ -283,7 +283,7 @@ class ResNetUnitGradOp : public framework::OperatorWithKernel {
  public:
   using framework::OperatorWithKernel::OperatorWithKernel;
 
-  void InferShape(framework::InferShapeContext* ctx) const {
+  void InferShape(framework::InferShapeContext* ctx) const override {
     // check input
     OP_INOUT_CHECK(ctx->HasInput("X"), "Input", "X", "ResNetUnitGradOp");
     OP_INOUT_CHECK(
@@ -390,7 +390,7 @@ class ResNetUnitGradOp : public framework::OperatorWithKernel {
 
  protected:
   phi::KernelKey GetExpectedKernelType(
-      const framework::ExecutionContext& ctx) const {
+      const framework::ExecutionContext& ctx) const override {
     PADDLE_ENFORCE_NOT_NULL(
         ctx.InputVar(framework::GradVarName("Y")),
         platform::errors::NotFound(

paddle/fluid/operators/fused/yolo_box_head_op.cc

@@ -21,7 +21,7 @@ class YoloBoxHeadOp : public framework::OperatorWithKernel {
  public:
   using framework::OperatorWithKernel::OperatorWithKernel;
 
-  void InferShape(framework::InferShapeContext* ctx) const {
+  void InferShape(framework::InferShapeContext* ctx) const override {
     OP_INOUT_CHECK(ctx->HasInput("X"), "Input", "X", "yolo_box_head");
     OP_INOUT_CHECK(ctx->HasOutput("Out"), "Output", "Out", "yolo_box_head");
     ctx->SetOutputDim("Out", ctx->GetInputDim("X"));
@@ -30,7 +30,7 @@ class YoloBoxHeadOp : public framework::OperatorWithKernel {
 
 class YoloBoxHeadOpMaker : public framework::OpProtoAndCheckerMaker {
  public:
-  void Make() {
+  void Make() override {
     AddInput("X", "The input tensor");
     AddAttr<std::vector<int>>("anchors",
                               "The anchor width and height, "

paddle/fluid/operators/fused/yolo_box_post_op.cc

@@ -21,7 +21,7 @@ class YoloBoxPostOp : public framework::OperatorWithKernel {
  public:
   using framework::OperatorWithKernel::OperatorWithKernel;
 
-  void InferShape(framework::InferShapeContext* ctx) const {
+  void InferShape(framework::InferShapeContext* ctx) const override {
     OP_INOUT_CHECK(ctx->HasInput("Boxes0"), "Input", "Boxes0", "yolo_box_post");
     OP_INOUT_CHECK(ctx->HasInput("Boxes1"), "Input", "Boxes1", "yolo_box_post");
     OP_INOUT_CHECK(ctx->HasInput("Boxes2"), "Input", "Boxes2", "yolo_box_post");
@@ -37,7 +37,7 @@ class YoloBoxPostOp : public framework::OperatorWithKernel {
 
 class YoloBoxPostOpMaker : public framework::OpProtoAndCheckerMaker {
  public:
-  void Make() {
+  void Make() override {
     AddInput("Boxes0", "The Boxes0 tensor");
     AddInput("Boxes1", "The Boxes1 tensor");
     AddInput("Boxes2", "The Boxes2 tensor");

paddle/fluid/operators/marker_op.cc

@@ -38,7 +38,7 @@ class MarkerOp : public framework::OperatorWithKernel {
 
 class MarkerOpMaker : public framework::OpProtoAndCheckerMaker {
  public:
-  void Make() {
+  void Make() override {
     AddAttr<std::string>("marker_role",
                          "(string, default forward)forward or backward,"
                          " mark different stages of porcess.")

paddle/fluid/operators/mul_op.cc

@@ -32,7 +32,7 @@ class MulOp : public framework::OperatorWithKernel {
   using framework::OperatorWithKernel::OperatorWithKernel;
 
   phi::KernelKey GetExpectedKernelType(
-      const framework::ExecutionContext& ctx) const {
+      const framework::ExecutionContext& ctx) const override {
     auto input_data_type = OperatorWithKernel::IndicateVarDataType(ctx, "X");
     return phi::KernelKey(input_data_type, ctx.GetPlace());
   }
@@ -104,7 +104,7 @@ class MulGradOp : public framework::OperatorWithKernel {
   using framework::OperatorWithKernel::OperatorWithKernel;
 
   phi::KernelKey GetExpectedKernelType(
-      const framework::ExecutionContext& ctx) const {
+      const framework::ExecutionContext& ctx) const override {
     auto input_data_type = OperatorWithKernel::IndicateVarDataType(ctx, "X");
     return phi::KernelKey(input_data_type, ctx.GetPlace());
   }

paddle/fluid/operators/nop_op.cc

@@ -33,7 +33,7 @@ class NopOp : public framework::OperatorWithKernel {
 
 class NopOpMaker : public framework::OpProtoAndCheckerMaker {
  public:
-  void Make() {
+  void Make() override {
     AddInput("X", "(Tensor) The input tensor of nop op.").AsDuplicable();
     AddOutput("Out", "(Tensor) The output tensor of nop op.").AsDuplicable();
     AddComment(R"DOC(

paddle/fluid/operators/optimizers/pow2_decay_with_linear_warmup_op.cc

@@ -40,7 +40,7 @@ class Pow2DecayWithLinearWarmupOp : public framework::OperatorWithKernel {
 class Pow2DecayWithLinearWarmupOpMaker
     : public framework::OpProtoAndCheckerMaker {
  public:
-  void Make() {
+  void Make() override {
     AddInput("LearningRate", "(Tensor) The input learning rate Tensor.");
     AddInput("Step", "(Tensor) The input global step Tensor.");
     AddOutput("LearningRateOut",

paddle/fluid/operators/pscore/fetch_barrier_op.cc

@@ -46,7 +46,7 @@ class FetchBarrierOp : public framework::OperatorBase {
 
 class FetchBarrierOpMaker : public framework::OpProtoAndCheckerMaker {
  public:
-  void Make() {
+  void Make() override {
     AddInput("X", "(Any) Dummy inputs, used for control dependency")
         .AsDispensable()
         .AsDuplicable();

paddle/fluid/operators/pscore/heter_listen_and_serv_op.cc

@@ -177,7 +177,7 @@ void HeterListenAndServOp::RunImpl(const framework::Scope &scope,
 
 class HeterListenAndServOpMaker : public framework::OpProtoAndCheckerMaker {
  public:
-  void Make() {
+  void Make() override {
     AddInput("X", "(Tensor) Variables that server recv.").AsDuplicable();
     AddComment(
         R"DOC(" + "HeterListenAndServ operator" + "\n" + "This operator" +

paddle/fluid/operators/pscore/listen_and_serv_op.cc

@@ -52,7 +52,7 @@ class ListenAndServOp : public framework::OperatorBase {
 
 class ListenAndServOpMaker : public framework::OpProtoAndCheckerMaker {
  public:
-  void Make() {
+  void Make() override {
     AddInput("X", "(Tensor) Variables that server recv.").AsDuplicable();
     AddComment(R"DOC(" + "ListenAndServ operator" + "\n" + "This operator" +
 " will start a RPC server which can receive variables from send_op and send" +

paddle/fluid/operators/pscore/send_and_recv_op.cc

@@ -69,7 +69,7 @@ class SendAndRecvOp : public framework::OperatorWithKernel {
 
 class SendAndRecvOpMaker : public framework::OpProtoAndCheckerMaker {
  public:
-  void Make() {
+  void Make() override {
     AddInput("X", "Tensor Input variable to be sent").AsDuplicable();
     AddOutput("Out", "Tensor Output varibale to be recv").AsDuplicable();
     AddAttr<std::string>("message_name", "");

paddle/fluid/operators/pscore/send_barrier_op.cc

@@ -47,7 +47,7 @@ class SendBarrierOp : public framework::OperatorBase {
 
 class SendBarrierOpMaker : public framework::OpProtoAndCheckerMaker {
  public:
-  void Make() {
+  void Make() override {
     AddInput("X", "(Any) Dummy inputs, used for control dependency")
         .AsDuplicable();
     AddOutput("Out", "(Any) Dummy outputs, used for control dependency")

paddle/fluid/operators/pscore/send_op.cc

@@ -75,7 +75,7 @@ class SendOp : public framework::OperatorBase {
 
 class SendOpMaker : public framework::OpProtoAndCheckerMaker {
  public:
-  void Make() {
+  void Make() override {
     AddInput("X", "(Tensor, SelectedRows) Input variables to be sent")
         .AsDuplicable();
     AddOutput("Out", "(Any) Dummy outputs, used for control dependency")

paddle/fluid/operators/reduce_ops/reduce_mean_op.cc

@@ -94,8 +94,8 @@ DECLARE_NO_NEED_BUFFER_VARS_INFERER(ReduceMeanGradNoNeedBufferVarInferer, "X");
 
 class __reduce_meanMaker__ : public ops::ReduceBaseOpMaker {
  protected:
-  virtual std::string GetName() const { return "reduce_mean"; }
-  virtual std::string GetOpType() const { return "Reduce reduce_mean"; }
+  std::string GetName() const override { return "reduce_mean"; }
+  std::string GetOpType() const override { return "Reduce reduce_mean"; }
 };
 
 DECLARE_INFER_SHAPE_FUNCTOR(

paddle/fluid/operators/tdm_child_op.cc

@@ -24,7 +24,7 @@ namespace paddle {
 namespace operators {
 class TDMChildOpMaker : public framework::OpProtoAndCheckerMaker {
  public:
-  void Make() {
+  void Make() override {
     AddInput("X",
              "X(Tensor), dtype support int32/int64, X variable is the "
              "node id of TDM-Tree");

paddle/fluid/operators/tdm_sampler_op.cc

@@ -25,7 +25,7 @@ namespace operators {
 
 class TDMSamplerOpMaker : public framework::OpProtoAndCheckerMaker {
  public:
-  void Make() {
+  void Make() override {
     AddInput("X",
              "X(Tensor), Input variable which"
              "mapping the leaf node idx of tdm tree,"

paddle/fluid/pybind/imperative.cc

@@ -89,7 +89,7 @@ class PyVariableWrapperHook : public imperative::VariableWrapperHook {
     Py_INCREF(py_func_);
   }
 
-  ~PyVariableWrapperHook() {
+  ~PyVariableWrapperHook() override {
     py::gil_scoped_acquire gil;
     Py_DECREF(py_func_);
   }

paddle/ir/pass/ir_printing.cc

@@ -45,7 +45,7 @@ class IRPrinting : public PassInstrumentation {
   explicit IRPrinting(std::unique_ptr<PassManager::IRPrinterOption> option)
       : option_(std::move(option)) {}
 
-  ~IRPrinting() = default;
+  ~IRPrinting() override = default;
 
   void RunBeforePass(Pass *pass, Operation *op) override {
     if (option_->print_on_change()) {

paddle/ir/pass/pass_timing.cc

@@ -51,7 +51,7 @@ class Timer {
 class PassTimer : public PassInstrumentation {
  public:
   explicit PassTimer(bool print_module) : print_module_(print_module) {}
-  ~PassTimer() = default;
+  ~PassTimer() override = default;
 
   void RunBeforePipeline(ir::Operation* op) override {
     pipeline_timers_[op] = Timer();

paddle/phi/api/profiler/device_tracer.cc

@@ -323,7 +323,7 @@ class DeviceTracerImpl : public DeviceTracer {
 #endif
   }
 
-  void AddAnnotation(uint32_t id, Event *event) {
+  void AddAnnotation(uint32_t id, Event *event) override {
 #ifdef PADDLE_WITH_SW
     std::forward_list<std::pair<uint32_t, Event *>> *local_correlations_pairs =
         nullptr;
@@ -339,7 +339,8 @@ class DeviceTracerImpl : public DeviceTracer {
     local_correlations_pairs->push_front(std::make_pair(id, event));
   }
 
-  void AddAnnotations(const std::map<uint64_t, ThreadEvents> &thr_events) {
+  void AddAnnotations(
+      const std::map<uint64_t, ThreadEvents> &thr_events) override {
     for (auto &tmp : active_kind_records_) {
       for (const ActiveKindRecord &r : tmp) {
         auto iter = thr_events.find(r.thread_id);
@@ -384,7 +385,7 @@ class DeviceTracerImpl : public DeviceTracer {
                      uint64_t start_ns,
                      uint64_t end_ns,
                      int64_t device_id,
-                     uint64_t thread_id) {
+                     uint64_t thread_id) override {
     if (anno.empty()) {
       VLOG(1) << "Empty timeline annotation.";
       return;
@@ -409,7 +410,7 @@ class DeviceTracerImpl : public DeviceTracer {
                      int64_t device_id,
                      int64_t stream_id,
                      uint32_t correlation_id,
-                     uint64_t bytes) {
+                     uint64_t bytes) override {
     // 0 means timestamp information could not be collected for the kernel.
     if (start_ns == 0 || end_ns == 0 || start_ns == end_ns) {
       VLOG(3) << name << " cannot be traced";
@@ -427,7 +428,7 @@ class DeviceTracerImpl : public DeviceTracer {
                         const Place &place,
                         const std::string &alloc_in,
                         const std::string &free_in,
-                        uint64_t thread_id) {
+                        uint64_t thread_id) override {
     if (0 == start_ns || 0 == end_ns) {
       VLOG(3) << alloc_in << ", " << free_in << " Cannot be traced.";
       return;
@@ -452,7 +453,7 @@ class DeviceTracerImpl : public DeviceTracer {
                             uint64_t end_ns,
                             int64_t device_id,
                             uint64_t thread_id,
-                            uint32_t correlation_id) {
+                            uint32_t correlation_id) override {
     if (anno.empty()) {
       VLOG(1) << "Empty timeline annotation.";
       return;
@@ -478,7 +479,7 @@ class DeviceTracerImpl : public DeviceTracer {
                         uint64_t end,
                         int64_t device_id,
                         int64_t stream_id,
-                        uint32_t correlation_id) {
+                        uint32_t correlation_id) override {
     // 0 means timestamp information could not be collected for the kernel.
     if (start == 0 || end == 0 || start == end) {
       VLOG(3) << correlation_id << " cannot be traced";
@@ -490,12 +491,12 @@ class DeviceTracerImpl : public DeviceTracer {
         KernelRecord{name, start, end, device_id, stream_id, correlation_id});
   }
 
-  bool IsEnabled() {
+  bool IsEnabled() override {
     std::lock_guard<std::mutex> l(trace_mu_);
     return enabled_;
   }
 
-  void Enable() {
+  void Enable() override {
     std::lock_guard<std::mutex> l(trace_mu_);
     if (enabled_) {
       return;
@@ -544,7 +545,7 @@ class DeviceTracerImpl : public DeviceTracer {
     enabled_ = true;
   }
 
-  void Reset() {
+  void Reset() override {
 #ifdef PADDLE_WITH_CUPTI
     CUPTI_CALL(
         dynload::cuptiActivityFlushAll(CUPTI_ACTIVITY_FLAG_FLUSH_FORCED));
@@ -559,7 +560,7 @@ class DeviceTracerImpl : public DeviceTracer {
     for (auto &tmp : active_kind_records_) tmp.clear();
   }
 
-  void GenEventKernelCudaElapsedTime() {
+  void GenEventKernelCudaElapsedTime() override {
 #ifdef PADDLE_WITH_CUPTI
     if (correlations_.empty())
       for (auto &tmp : correlations_pairs)
@@ -591,7 +592,7 @@ class DeviceTracerImpl : public DeviceTracer {
 #endif
   }
 
-  proto::Profile GenProfile(const std::string &profile_path) {
+  proto::Profile GenProfile(const std::string &profile_path) override {
     proto::Profile profile_pb = this->GetProfile();
     std::ofstream profile_f;
     profile_f.open(profile_path,
@@ -601,7 +602,7 @@ class DeviceTracerImpl : public DeviceTracer {
     return profile_pb;
   }
 
-  proto::Profile GetProfile() {
+  proto::Profile GetProfile() override {
     int miss = 0, find = 0;
     std::lock_guard<std::mutex> l(trace_mu_);
     proto::Profile profile_pb;
@@ -711,7 +712,7 @@ class DeviceTracerImpl : public DeviceTracer {
     return profile_pb;
   }
 
-  void Disable() {
+  void Disable() override {
 #ifdef PADDLE_WITH_CUPTI
     // flush might cause additional calls to DeviceTracker.
     CUPTI_CALL(