fix bug of ascend when 2 parameter's arg are same

mindspore/ccsrc/backend/common/session/kernel_graph.cc

@@ -770,7 +770,15 @@ void KernelGraph::FrontBackendMapAdd(const AnfNodePtr &front_anf, const AnfNodeP
   if (front_backend_anf_map_.find(front_anf) != front_backend_anf_map_.end()) {
     MS_LOG(EXCEPTION) << "Anf " << front_anf->DebugString() << " has been exist in the front_backend_anf_map_";
   }
+  front_backend_anf_map_[front_anf] = backend_anf;
   if (backend_front_anf_map_.find(backend_anf) != backend_front_anf_map_.end()) {
+    // If def func(x, y) and call as func(arg, arg) ,then the parameter x and y share same param_info "arg".
+    // In this case, parameter is get from param_info and has been exist in the map. So can't add it to map again.
+    if (backend_anf->isa<Parameter>()) {
+      MS_LOG(INFO) << "Backend parameter already exist, backend parameter:" << backend_anf->DebugString()
+                   << ", exist front parameter:" << backend_front_anf_map_[backend_anf]->DebugString();
+      return;
+    }
     auto front_node = front_anf->cast<CNodePtr>();
     MS_EXCEPTION_IF_NULL(front_node);
     auto attr_input = front_node->input(kAnfPrimitiveIndex);
@@ -779,7 +787,6 @@ void KernelGraph::FrontBackendMapAdd(const AnfNodePtr &front_anf, const AnfNodeP
       MS_LOG(EXCEPTION) << "Kernel " << backend_anf->DebugString() << "has been exist in the backend_front_anf_map_";
     }
   }
-  front_backend_anf_map_[front_anf] = backend_anf;
   backend_front_anf_map_[backend_anf] = front_anf;
 }
 

tests/st/construct_input/test_outermost_input.py

@@ -120,6 +120,7 @@ def test_grad_first_input_net(mode):
     Description: Normal input type.
     Expectation: No exception.
     """
+
     class FirstInputTensorNet(nn.Cell):
         def construct(self, tensor_a, tuple_a, list_b, tensor_b, tensor_c, dict_c):
             return tensor_a + tuple_a[2] - list_b[1][1]["y"] + tensor_b - tensor_c + dict_c["y"]
@@ -152,6 +153,17 @@ class GradCellWithParameter(nn.Cell):
         return self.grad(self.net, self.param)(x)
 
 
+class AssignParameterWithCell(nn.Cell):
+    def __init__(self, net):
+        super().__init__()
+        self.net = net
+        self.param = self.net.param
+
+    def construct(self, x):
+        self.param = self.param * 2
+        return x
+
+
 class GradCell(nn.Cell):
     def __init__(self, net):
         super().__init__()
@@ -215,9 +227,10 @@ def test_grad_parameter_as_input_and_fv(mode):
 
 # PyNative run error.
 # Support context.PYNATIVE_MODE later.
-# Support Ascend BE later.
 @pytest.mark.level0
 @pytest.mark.platform_x86_cpu
+@pytest.mark.platform_arm_ascend_training
+@pytest.mark.platform_x86_ascend_training
 @pytest.mark.platform_x86_gpu_training
 @pytest.mark.env_onecard
 @pytest.mark.parametrize('mode', [context.GRAPH_MODE])
@@ -238,6 +251,26 @@ def test_grad_same_parameter_both_input_and_fv(mode):
     assert np.array_equal(a[1].asnumpy(), b[1].asnumpy())
 
 
+@pytest.mark.level0
+@pytest.mark.platform_x86_cpu
+@pytest.mark.platform_arm_ascend_training
+@pytest.mark.platform_x86_ascend_training
+@pytest.mark.platform_x86_gpu_training
+@pytest.mark.env_onecard
+@pytest.mark.parametrize('mode', [context.GRAPH_MODE])
+def test_same_arg_parameter_assign(mode):
+    """
+    Feature: Change value of common parameter.
+    Description: Change value of common parameter.Watch another parameter with same param_obj.
+    Expectation: No exception.
+    """
+    context.set_context(mode=mode)
+    x = Parameter(Tensor(np.array([[1, 2], [3, 4]])), name='input_x')
+    a = AssignParameterWithCell(TestCell(x))(x)
+    print(f'a: {a}')
+    assert np.array_equal(a.asnumpy(), x.asnumpy())
+
+
 class TestCell2(nn.Cell):
     def __init__(self, param1, param2):
         super().__init__()