!2845 Fix empty graph dump ir failed

Merge pull request !2845 from zhoufeng/empty-graph-dump-ir

mindspore/ccsrc/session/anf_runtime_algorithm.cc

@@ -694,7 +694,7 @@ void AnfRuntimeAlgorithm::SetOutputInferTypeAndShape(const std::vector<TypeId> &
     MS_LOG(EXCEPTION) << "Types size " << types.size() << "should be same with shapes size " << shapes.size();
   }
   if (shapes.empty()) {
-    MS_LOG(EXCEPTION) << "Illegal empty output_types_shapes";
+    node->set_abstract(std::make_shared<abstract::AbstractNone>());
   } else if (shapes.size() == 1) {
     // single output handle
     std::vector<int> shape_int;
@@ -1012,6 +1012,9 @@ std::vector<KernelGraphPtr> AnfRuntimeAlgorithm::GetCallNodeKernelGraph(const CN
     auto get_switch_kernel_graph = [switch_node](size_t input_index) -> KernelGraphPtr {
       auto partial = switch_node->input(input_index);
       MS_EXCEPTION_IF_NULL(partial);
+      if (IsValueNode<KernelGraph>(partial)) {
+        return GetValueNode<KernelGraphPtr>(partial);
+      }
       auto partial_cnode = partial->cast<CNodePtr>();
       MS_EXCEPTION_IF_NULL(partial_cnode);
       auto graph_node = partial_cnode->input(1);

mindspore/ccsrc/session/ascend_control_parser.cc

@@ -411,8 +411,7 @@ void AscendControlParser::RecurseSwitch(NotNull<KernelGraphPtr> kg, NotNull<CNod
     origin_switch_inputs[kCNodeSwitchCond]};
   for (size_t i = kCNodeSwitchCond + 1; i < kCNodeSwitchLength; ++i) {
     // 3.1 branch kernel graph and args
-    KernelGraphPtr branch_fg;
-    std::tie(std::ignore, branch_fg) = ParsePartial(NOT_NULL(origin_switch_inputs[i]));
+    KernelGraphPtr branch_fg = ParsePartial(NOT_NULL(origin_switch_inputs[i]));
     // 3.2 recurse sub graph
     CNodePtr branch_label = ProcessKernelGraph(NOT_NULL(branch_fg), cur_node, back_label, memo);
     new_switch_inputs.push_back(branch_label);
@@ -456,8 +455,7 @@ void AscendControlParser::RecurseSwitchLayer(NotNull<KernelGraphPtr> kg, NotNull
     origin_switch_inputs[kCNodeSwitchCond]};
   for (size_t i = 0; i < branch_partial.size(); ++i) {
     // 3.1 branch kernel graph and args
-    KernelGraphPtr branch_fg;
-    std::tie(std::ignore, branch_fg) = ParsePartial(NOT_NULL(origin_switch_inputs[i]));
+    KernelGraphPtr branch_fg = ParsePartial(NOT_NULL(origin_switch_inputs[i]));
     // 3.2 recurse sub graph
     CNodePtr branch_label = ProcessKernelGraph(NOT_NULL(branch_fg), cur_node, back_label, memo);
     new_switch_inputs.push_back(branch_label);
@@ -468,8 +466,11 @@ void AscendControlParser::RecurseSwitchLayer(NotNull<KernelGraphPtr> kg, NotNull
   MS_LOG(INFO) << "Succeed processing switch layer " << cur_node->DebugString();
 }
 
-std::tuple<CNodePtr, KernelGraphPtr> AscendControlParser::ParsePartial(NotNull<AnfNodePtr> node) {
+KernelGraphPtr AscendControlParser::ParsePartial(NotNull<AnfNodePtr> node) {
   if (!node.get()->isa<CNode>()) {
+    if (IsValueNode<KernelGraph>(node)) {
+      return GetValueNode<KernelGraphPtr>(node);
+    }
     MS_LOG(EXCEPTION) << "Switch branches must be partial, node: " << node->DebugString();
   }
   // 2.1 branch kernel graph and args
@@ -484,7 +485,7 @@ std::tuple<CNodePtr, KernelGraphPtr> AscendControlParser::ParsePartial(NotNull<A
     MS_LOG(EXCEPTION) << "Index out of range:" << partial_inputs.size() << ".";
   }
   auto branch_kg = GetValueNode<KernelGraphPtr>(partial_inputs[kCNodePartialFunc]);
-  return {partial_cnode, branch_kg};
+  return branch_kg;
 }
 
 void AscendControlParser::InsertMultipleAssignToGraph(NotNull<KernelGraphPtr> from_graph,

mindspore/ccsrc/session/ascend_control_parser.h

@@ -53,7 +53,7 @@ class AscendControlParser {
 
   static void LinkParentGraph(NotNull<KernelGraphPtr> kg, const CNodePtr &from_graph_call_node,
                               const CNodePtr &last_label);
-  static std::tuple<CNodePtr, KernelGraphPtr> ParsePartial(NotNull<AnfNodePtr> node);
+  static KernelGraphPtr ParsePartial(NotNull<AnfNodePtr> node);
 
   static void InsertMultipleAssignToGraph(NotNull<KernelGraphPtr> from_graph, NotNull<KernelGraphPtr> to_graph,
                                           NotNull<AnfNodePtr> from, NotNull<AnfNodePtr> to);

mindspore/ccsrc/session/ascend_session.cc

@@ -255,6 +255,9 @@ static void UpdateRealInput(NotNull<KernelGraphPtr> graph, bool split_flag,
         MS_EXCEPTION_IF_NULL(switch_cnode);
         auto partial = switch_cnode->input(input_index);
         MS_EXCEPTION_IF_NULL(partial);
+        if (IsValueNode<KernelGraph>(partial)) {
+          return {};
+        }
         auto partial_cnode = partial->cast<CNodePtr>();
         MS_EXCEPTION_IF_NULL(partial_cnode);
         auto ret = std::vector<AnfNodePtr>(partial_cnode->inputs().begin() + 2, partial_cnode->inputs().end());

mindspore/ccsrc/session/kernel_graph.cc

@@ -387,18 +387,16 @@ ParameterPtr KernelGraph::NewParameter(const ParameterPtr &parameter) {
     } else {
       kernel_info->SetFeatureMapFlag(true);
     }
-    // if output is a tuple tensor,now can use for loop to handle tuple tensor
-    output_tensor_num = AnfAlgo::GetOutputTensorNum(parameter);
   }
   new_parameter->set_kernel_info(kernel_info);
   // create kernel_build_info for new parameter
   auto kernel_build_info_builder = std::make_shared<kernel::KernelBuildInfo::KernelBuildInfoBuilder>();
   // create init data type,
   std::vector<TypeId> init_data_type = {};
-  for (size_t i = 0; i < output_tensor_num; i++) {
-    TypeId infer_data_type = AnfAlgo::GetOutputInferDataType(new_parameter, i);
-    init_data_type.push_back(AnfAlgo::IsParameterWeight(new_parameter) ? kTypeUnknown : infer_data_type);
-  }
+
+  TypeId infer_data_type = AnfAlgo::GetOutputInferDataType(new_parameter, 0);
+  init_data_type.push_back(AnfAlgo::IsParameterWeight(new_parameter) ? kTypeUnknown : infer_data_type);
+
   // set the format of parameter to DEFAULT_FORMAT
   kernel_build_info_builder->SetOutputsFormat(std::vector<std::string>(output_tensor_num, kOpFormat_DEFAULT));
   // set parameter initaial device data type

tests/ut/cpp/session/anf_runtime_algorithm_test.cc

@@ -590,7 +590,8 @@ TEST_F(AnfRuntimeAlgorithmTest, SetOutputInferTypeAndShape) {
   std::vector<TypeId> none_types = {};
   std::vector<std::vector<size_t>> none_shapes = {};
   EXPECT_THROW(AnfAlgo::SetOutputInferTypeAndShape(none_types, none_shapes, nullptr), std::runtime_error);
-  EXPECT_THROW(AnfAlgo::SetOutputInferTypeAndShape(none_types, none_shapes, add.get()), std::runtime_error);
+  AnfAlgo::SetOutputInferTypeAndShape(none_types, none_shapes, add.get());
+  EXPECT_EQ((*add->abstract()), abstract::AbstractNone());
   // set single input
   std::vector<TypeId> single_types = {kFloat32->type_id()};
   std::vector<std::vector<size_t>> single_shapes = {{2, 32, 224, 224}};