[PIR] standardize the use of value[-5] (#57461)

winter-wang · web-flow · commit 5775b170512c · 2023-09-19T10:00:02.000+08:00
diff --git a/paddle/fluid/framework/new_executor/new_ir_interpreter.cc b/paddle/fluid/framework/new_executor/new_ir_interpreter.cc
@@ -1309,11 +1309,11 @@ void NewIRInterpreter::SolvePersisableVarNames() {
     ::pir::Value value = kv.first;
     const std::string& var_name = kv.second;
     ::pir::OpResult result = value.dyn_cast<::pir::OpResult>();
-    auto* defining_op = value.GetDefiningOp();
+    auto* defining_op = result.owner();
     if (defining_op->HasAttribute(kAttrIsPersisable)) {
-      auto is_persisables = defining_op->attribute(kAttrIsPersisable)
-                                .dyn_cast<::pir::ArrayAttribute>()
-                                .AsVector();
+      auto is_persisables =
+          defining_op->attribute<::pir::ArrayAttribute>(kAttrIsPersisable)
+              .AsVector();
       if (is_persisables[result.index()]
               .dyn_cast<::pir::BoolAttribute>()
               .data()) {
diff --git a/paddle/fluid/pir/dialect/op_generator/op_interface_gen.py b/paddle/fluid/pir/dialect/op_generator/op_interface_gen.py
@@ -27,7 +27,7 @@
 
 OP_VJP_FORWARD_MULTI_INPUT_TEMPLATE = """
     pir::CombineOp combine_op_obj =
-      op_obj.{input_name}().GetDefiningOp()->dyn_cast<pir::CombineOp>();
+      op_obj.{input_name}().dyn_cast<pir::OpResult>().owner()->dyn_cast<pir::CombineOp>();
     std::vector<Tensor> {input_name};
     for (size_t idx = 0; idx < combine_op_obj.inputs().size(); idx++) {{
         {input_name}.emplace_back(
diff --git a/paddle/fluid/pir/phi_kernel_adaptor/phi_kernel_util.cc b/paddle/fluid/pir/phi_kernel_adaptor/phi_kernel_util.cc
@@ -125,7 +125,7 @@ paddle::framework::Variable* CreateVar(
         variable_2_var_name,
     std::map<std::string, int>* var_name_2_id,
     std::vector<paddle::framework::Variable*>* variable_list) {
-  Operation* def_op = value.GetDefiningOp();
+  Operation* def_op = value.dyn_cast<OpResult>().owner();
   bool is_persisable = false;
   if (def_op->isa<::pir::SetParameterOp>()) {
     is_persisable = true;
diff --git a/paddle/fluid/pir/transforms/inplace_pass.cc b/paddle/fluid/pir/transforms/inplace_pass.cc
@@ -37,13 +37,14 @@ static bool CanBeDeleted(pir::Value value) {
       !value.type().isa<paddle::dialect::AllocatedSelectedRowsType>()) {
     return false;
   }
-  if (value.GetDefiningOp()->HasAttribute(kAttrIsPersisable)) {
-    return !(value.GetDefiningOp()
-                 ->attribute(kAttrIsPersisable)
-                 .dyn_cast<pir::ArrayAttribute>()
-                 .AsVector()[value.dyn_cast<pir::OpResult>().index()]
-                 .dyn_cast<pir::BoolAttribute>()
-                 .data());
+  if (auto op_result = value.dyn_cast<pir::OpResult>()) {
+    auto def_op = op_result.owner();
+    if (def_op->HasAttribute(kAttrIsPersisable)) {
+      return !(def_op->attribute<pir::ArrayAttribute>(kAttrIsPersisable)
+                   .AsVector()[op_result.index()]
+                   .dyn_cast<pir::BoolAttribute>()
+                   .data());
+    }
   }
   return true;
 }
diff --git a/paddle/fluid/pir/transforms/pd_op_to_kernel_pass.cc b/paddle/fluid/pir/transforms/pd_op_to_kernel_pass.cc
@@ -248,9 +248,9 @@ pir::OpResult AddPlaceTransferOp(pir::OpResult in,
     pir::Operation* op =
         pir::Operation::Create({in}, op_attribute, {out_type}, op_info);
 
-    if (in.GetDefiningOp()->HasAttribute(kAttrIsPersisable)) {
+    if (in.owner()->HasAttribute(kAttrIsPersisable)) {
       op->set_attribute(kAttrIsPersisable,
-                        in.GetDefiningOp()->attribute(kAttrIsPersisable));
+                        in.owner()->attribute(kAttrIsPersisable));
     }
     block->push_back(op);
 
@@ -527,11 +527,10 @@ phi::KernelKey GetKernelKey(
       if (op->isa<paddle::dialect::UniformOp>()) {
         // try to process uniform, use shape to determin backend
         // TODO(phlrain): shuold support other initilize op
-        auto define_op = op->operand_source(0).GetDefiningOp();
+        auto define_op =
+            op->operand_source(0).dyn_cast<pir::OpResult>().owner();
         if (define_op->isa<paddle::dialect::FullIntArrayOp>()) {
-          auto shape = define_op->attributes()
-                           .at("value")
-                           .dyn_cast<dialect::IntArrayAttribute>()
+          auto shape = define_op->attribute<dialect::IntArrayAttribute>("value")
                            .data()
                            .GetData();
 
@@ -577,13 +576,12 @@ phi::KernelKey GetKernelKey(
       // uses data op outout as inputs. So, we need set kernel backend
       // manually.
       if (op->operand_source(i)
-              .GetDefiningOp()
+              .dyn_cast<pir::OpResult>()
+              .owner()
               ->isa<paddle::dialect::DataOp>()) {
-        auto data_op = op->operand_source(i).GetDefiningOp();
-        auto data_place = data_op->attributes()
-                              .at("place")
-                              .dyn_cast<dialect::PlaceAttribute>()
-                              .data();
+        auto data_op = op->operand_source(i).dyn_cast<pir::OpResult>().owner();
+        auto data_place =
+            data_op->attribute<dialect::PlaceAttribute>("place").data();
 
         auto data_op_backend = paddle::experimental::ParseBackend(data_place);
         if (data_op_backend == phi::Backend::UNDEFINED) {
@@ -592,17 +590,21 @@ phi::KernelKey GetKernelKey(
         kernel_key_parser.key_set.backend_set =
             kernel_key_parser.key_set.backend_set |
             paddle::experimental::BackendSet(data_op_backend);
-      } else if (op->operand_source(i).GetDefiningOp()->name() ==
-                 "builtin.combine") {
-        auto combine_op = op->operand_source(i).GetDefiningOp();
+      } else if (op->operand_source(i)
+                     .dyn_cast<pir::OpResult>()
+                     .owner()
+                     ->isa<pir::CombineOp>()) {
+        auto combine_op =
+            op->operand_source(i).dyn_cast<pir::OpResult>().owner();
         for (size_t j = 0; j < combine_op->num_operands(); ++j) {
-          if (combine_op->operand_source(j).GetDefiningOp()->name() ==
-              "pd_op.data") {
-            auto data_op = combine_op->operand_source(j).GetDefiningOp();
-            auto data_place = data_op->attributes()
-                                  .at("place")
-                                  .dyn_cast<dialect::PlaceAttribute>()
-                                  .data();
+          if (combine_op->operand_source(j)
+                  .dyn_cast<pir::OpResult>()
+                  .owner()
+                  ->isa<DataOp>()) {
+            auto data_op =
+                combine_op->operand_source(j).dyn_cast<pir::OpResult>().owner();
+            auto data_place =
+                data_op->attribute<PlaceAttribute>("place").data();
 
             auto data_op_backend =
                 paddle::experimental::ParseBackend(data_place);
@@ -981,7 +983,7 @@ std::vector<pir::Value> BuildOpInputList(
       } else if (new_in_type.isa<pir::VectorType>()) {
         // [ todo need update here, support combine data transfomer]
         // deal with pre combine op
-        auto pre_define_op = cur_in.GetDefiningOp();
+        auto pre_define_op = cur_in.dyn_cast<pir::OpResult>().owner();
 
         if (pre_define_op->isa<::pir::CombineOp>()) {
           std::vector<pir::Value> inner_inputs;
diff --git a/paddle/fluid/pir/transforms/transform_general_functions.cc b/paddle/fluid/pir/transforms/transform_general_functions.cc
@@ -25,7 +25,7 @@ namespace pir {
 std::pair<std::string, pir::Parameter*> GetParameterFromValue(
     pir::Value value) {
   pir::GetParameterOp op =
-      value.GetDefiningOp()->dyn_cast<pir::GetParameterOp>();
+      value.dyn_cast<OpResult>().owner()->dyn_cast<pir::GetParameterOp>();
   PADDLE_ENFORCE_NOT_NULL(
       op,
       phi::errors::InvalidArgument(
@@ -66,7 +66,7 @@ Operation* GetDefiningOpForInput(Operation* op, uint32_t index) {
       index < op->num_operands(),
       true,
       phi::errors::InvalidArgument("Intput operand's index must be valid."));
-  return op->operand_source(index).GetDefiningOp();
+  return op->operand_source(index).dyn_cast<OpResult>().owner();
 }
 
 Operation* GetFirstUseOperationForOutput(Operation* op, uint32_t index) {
diff --git a/paddle/fluid/primitive/codegen/templates/rule/vjp/generated/generated_vjp.cc.j2 b/paddle/fluid/primitive/codegen/templates/rule/vjp/generated/generated_vjp.cc.j2
@@ -41,7 +41,7 @@ return vjp_res;
 {% macro get_mutable_attribute(attrs, api_name) %}
   {% for i in attrs %}
     {%- if i is mutable_attribute -%}
-auto* {{i.name}}_define_op = std::static_pointer_cast<primitive::LazyTensor>({{i.name~'_'}}.impl())->value().dyn_cast<pir::OpResult>().GetDefiningOp();
+auto* {{i.name}}_define_op = std::static_pointer_cast<primitive::LazyTensor>({{i.name~'_'}}.impl())->value().dyn_cast<pir::OpResult>().owner();
       {% if i.typename is scalar %}
 if({{i.name}}_define_op->name() != "pd_op.full") {
   PADDLE_THROW(platform::errors::Unimplemented(
diff --git a/paddle/fluid/pybind/ir.cc b/paddle/fluid/pybind/ir.cc
@@ -323,9 +323,12 @@ void BindValue(py::module *m) {
 
   )DOC");
   value
-      .def("get_defining_op",
-           &Value::GetDefiningOp,
-           return_value_policy::reference)
+      .def(
+          "get_defining_op",
+          [](const Value &self) {
+            return self.dyn_cast<pir::OpResult>().owner();
+          },
+          return_value_policy::reference)
       .def("first_use", &Value::first_use, return_value_policy::reference)
       .def("has_one_use", &Value::HasOneUse)
       .def("use_empty", &Value::use_empty)
@@ -444,22 +447,19 @@ void BindOpResult(py::module *m) {
            })
       .def("__hash__",
            [](OpResult &self) { return std::hash<pir::Value>{}(self); })
-      .def("get_defining_op",
-           &OpResult::GetDefiningOp,
-           return_value_policy::reference)
+      .def("get_defining_op", &OpResult::owner, return_value_policy::reference)
       .def_property_readonly(
           "block",
-          [](OpResult &self) { return self.GetDefiningOp()->GetParent(); },
+          [](OpResult &self) { return self.owner()->GetParent(); },
           return_value_policy::reference)
       .def_property_readonly(
           "name",
           [](OpResult &self) {
-            if (self.GetDefiningOp()->isa<::pir::GetParameterOp>()) {
-              auto param_name = self.GetDefiningOp()
-                                    ->attributes()
-                                    .at("parameter_name")
-                                    .dyn_cast<pir::StrAttribute>()
-                                    .AsString();
+            if (self.owner()->isa<::pir::GetParameterOp>()) {
+              auto param_name =
+                  self.owner()
+                      ->attribute<pir::StrAttribute>("parameter_name")
+                      .AsString();
               return param_name;
             } else {
               PADDLE_THROW(phi::errors::InvalidArgument(
diff --git a/paddle/pir/core/attribute.h b/paddle/pir/core/attribute.h
@@ -51,6 +51,8 @@ class IR_API Attribute {
 
   bool operator!() const { return storage_ == nullptr; }
 
+  operator const void *() const { return storage_; }
+
   ///
   /// \brief Some Attribute attribute acquisition interfaces.
   ///
@@ -85,8 +87,6 @@ class IR_API Attribute {
     return pir::dyn_cast<U>(*this);
   }
 
-  friend struct std::hash<Attribute>;
-
  protected:
   const Storage *storage_{nullptr};
 };
@@ -98,7 +98,7 @@ namespace std {
 template <>
 struct hash<pir::Attribute> {
   std::size_t operator()(const pir::Attribute &obj) const {
-    return std::hash<const pir::Attribute::Storage *>()(obj.storage_);
+    return std::hash<const void *>()(obj);
   }
 };
 }  // namespace std
diff --git a/paddle/pir/core/builtin_op.cc b/paddle/pir/core/builtin_op.cc
@@ -310,7 +310,7 @@ void SplitOp::PassStopGradients(OperationArgument &argument) {
       for (uint32_t i = 0; i < defining_op->num_operands(); ++i) {
         auto value = defining_op->operand_source(i);
         if (!value) continue;
-        auto *oprand_defining_op = value.GetDefiningOp();
+        auto *oprand_defining_op = value.dyn_cast<OpResult>().owner();
         if (oprand_defining_op->HasAttribute(kStopGradientAttrName)) {
           auto attrs = oprand_defining_op->attribute(kStopGradientAttrName)
                            .dyn_cast<pir::ArrayAttribute>()
diff --git a/paddle/pir/core/op_info.h b/paddle/pir/core/op_info.h
@@ -71,13 +71,13 @@ class IR_API OpInfo {
   template <typename InterfaceT>
   typename InterfaceT::Concept *GetInterfaceImpl() const;
 
+  operator const void *() const { return impl_; }
   void *AsOpaquePointer() const { return impl_; }
   static OpInfo RecoverFromOpaquePointer(void *pointer) {
     return OpInfo(static_cast<OpInfoImpl *>(pointer));
   }
 
   friend class OpInfoImpl;
-  friend struct std::hash<OpInfo>;
 
  private:
   explicit OpInfo(OpInfoImpl *impl) : impl_(impl) {}
@@ -105,7 +105,7 @@ namespace std {
 template <>
 struct hash<pir::OpInfo> {
   std::size_t operator()(const pir::OpInfo &obj) const {
-    return std::hash<const pir::OpInfoImpl *>()(obj.impl_);
+    return std::hash<const void *>()(obj);
   }
 };
 }  // namespace std
diff --git a/paddle/pir/core/type.h b/paddle/pir/core/type.h
@@ -48,7 +48,7 @@ class IR_API Type {
   Type() = default;
 
   Type(const Storage *storage)  // NOLINT
-      : storage_(const_cast<Storage *>(storage)) {}
+      : storage_(storage) {}
 
   Type(const Type &other) = default;
 
@@ -73,11 +73,7 @@ class IR_API Type {
   ///
   /// \brief Support PointerLikeTypeTraits.
   ///
-  ///
-  const void *AsOpaquePointer() const {
-    return static_cast<const void *>(storage_);
-  }
-
+  operator const void *() const { return storage_; }
   static Type RecoverFromOpaquePointer(const void *pointer) {
     return Type(reinterpret_cast<Storage *>(const_cast<void *>(pointer)));
   }
@@ -120,11 +116,6 @@ class IR_API Type {
 
   static Type Parse(std::istream &is, IrContext *ctx);
 
-  ///
-  /// \brief Enable hashing Type.
-  ///
-  friend struct std::hash<Type>;
-
   template <typename U>
   U cast() const {
     return pir::cast<U>(*this);
@@ -189,7 +180,7 @@ namespace std {
 template <>
 struct hash<pir::Type> {
   std::size_t operator()(const pir::Type &obj) const {
-    return std::hash<const pir::Type::Storage *>()(obj.storage_);
+    return std::hash<const void *>()(obj);
   }
 };
 }  // namespace std
diff --git a/paddle/pir/core/type_id.h b/paddle/pir/core/type_id.h
@@ -53,6 +53,7 @@ class TypeId {
   ///
   /// \brief Support PointerLikeTypeTraits.
   ///
+  operator const void *() const { return storage_; }
   void *AsOpaquePointer() const { return storage_; }
   static TypeId RecoverFromOpaquePointer(void *pointer) {
     return TypeId(static_cast<Storage *>(pointer));
@@ -71,11 +72,6 @@ class TypeId {
     return storage_ < other.storage_;
   }
 
-  ///
-  /// \brief Enable hashing TypeId instances.
-  ///
-  friend struct std::hash<TypeId>;
-
  private:
   ///
   /// \brief Construct a TypeId and initialize storage.
@@ -150,7 +146,7 @@ namespace std {
 template <>
 struct hash<pir::TypeId> {
   std::size_t operator()(const pir::TypeId &obj) const {
-    return std::hash<const pir::TypeId::Storage *>()(obj.storage_);
+    return std::hash<const void *>()(obj);
   }
 };
 }  // namespace std
diff --git a/paddle/pir/core/value.cc b/paddle/pir/core/value.cc
@@ -52,11 +52,6 @@ void Value::set_type(pir::Type type) {
   impl_->set_type(type);
 }
 
-Operation *Value::GetDefiningOp() const {
-  if (auto result = dyn_cast<OpResult>()) return result.owner();
-  return nullptr;
-}
-
 std::string Value::PrintUdChain() {
   CHECK_VALUE_NULL_IMPL(PrintUdChain);
   return impl()->PrintUdChain();
diff --git a/paddle/pir/core/value.h b/paddle/pir/core/value.h
@@ -59,8 +59,6 @@ class IR_API Value {
 
   void set_type(Type type);
 
-  Operation *GetDefiningOp() const;
-
   std::string PrintUdChain();
 
   ///
diff --git a/paddle/pir/dialect/shape/utils/shape_utils.cc b/paddle/pir/dialect/shape/utils/shape_utils.cc
@@ -84,7 +84,7 @@ bool SymbolicDimMgr::LoadShapeConstraintGraph() {
   auto build_sym_product = [&](std::vector<Value> range,
                                SymbolicDimProduct& product) {
     for (Value v : range) {
-      auto definingOp = v.GetDefiningOp();
+      auto definingOp = v.dyn_cast<OpResult>().owner();
       if (auto constOp = definingOp->dyn_cast<ConstantOp>()) {
         product.factor *= constOp.value().dyn_cast<Int32Attribute>().data();
         continue;
diff --git a/paddle/pir/pattern_rewrite/pattern_rewrite_driver.cc b/paddle/pir/pattern_rewrite/pattern_rewrite_driver.cc
@@ -173,7 +173,8 @@ class GreedyPatternRewriteDriver : public pir::PatternRewriter {
     // that single use values often have more canonicalization opportunities.
     if (!operand || (!operand.use_empty() && !operand.HasOneUse())) return;
 
-    if (auto* def_op = operand.GetDefiningOp()) AddToWorklist(def_op);
+    if (auto* def_op = operand.dyn_cast<pir::OpResult>().owner())
+      AddToWorklist(def_op);
   }
 
   void AddOperandsToWorklist(const std::vector<pir::Value> operands) {
diff --git a/test/cpp/pir/core/ir_value_test.cc b/test/cpp/pir/core/ir_value_test.cc
diff --git a/test/cpp/pir/shape_dialect/symbolic_op_test.cc b/test/cpp/pir/shape_dialect/symbolic_op_test.cc
