!34243 DDE coredump: infer public fuction of list append is wrong.

Merge pull request !34243 from lanzhineng/issue

mindspore/ccsrc/pipeline/jit/static_analysis/program_specialize.cc

@@ -618,7 +618,12 @@ void PurifySequenceValueNode(const CNodePtr &cnode, size_t index, ProgramSpecial
   }
   std::vector<size_t> dead_node_positions;
   ValuePtrList elements;
-  for (size_t i = 0; i < (*flags).size(); ++i) {
+  auto sequence_value_size = sequence_value->value().size();
+  if (flags->size() < sequence_value_size) {
+    MS_LOG(EXCEPTION) << "Inner exception. CNode: " << cnode->ToString() << " input: " << old_input->ToString()
+                      << " flags size: " << flags->size() << " values size: " << sequence_value->value().size();
+  }
+  for (size_t i = 0; i < sequence_value_size; ++i) {
     ValuePtr old_sequence_value = sequence_value->value()[i];
     auto old_sequence_str_value = old_sequence_value->cast<StringImmPtr>();
     if (!(*flags)[i]) {
@@ -848,8 +853,8 @@ AnfNodePtr FuncGraphSpecializer::BuildReplacedNode(const AnfNodeConfigPtr &conf)
   AddTodoItem(new_conf->node());
   auto repl = GetReplicatedNode(new_conf->node());
   if (repl->func_graph()) {
-    MS_LOG(DEBUG) << "Set repl: graph(" << repl->func_graph()->ToString() << "), node:" << repl->DebugString()
-                  << ") to replace origin:" << new_conf->node()->DebugString();
+    MS_LOG(DEBUG) << "Set repl: graph(" << repl->func_graph()->ToString() << "), node: " << repl->DebugString()
+                  << ") to replace origin: " << new_conf->node()->DebugString();
   } else {
     MS_LOG(DEBUG) << "Set repl: graph(nullptr), node(" << repl->DebugString()
                   << ") to replace origin: " << new_conf->node()->DebugString();
@@ -1039,7 +1044,7 @@ std::pair<AbstractBasePtrList, AbstractBasePtr> FuncGraphSpecializer::BuildFromB
   std::vector<AbstractBasePtrList> args_vector;
   auto eval_cache_iter = eval_cache_.find(eval);
   if (eval_cache_iter == eval_cache_.end()) {
-    MS_LOG(EXCEPTION) << "Evaluator:" << eval->ToString() << " not exist in cache.";
+    MS_LOG(EXCEPTION) << "Evaluator: " << eval->ToString() << " not exist in cache.";
   }
   auto &origin_eval_cache = eval_cache_iter->second->GetCache();
   for (auto &argvals_map : origin_eval_cache) {

mindspore/core/abstract/abstract_value.cc

@@ -628,10 +628,10 @@ bool AbstractSequence::PurifyElements() {
 
   // Purify the elements.
   auto &elements_use_flags = *elements_use_flags_ptr;
-  if (elements_use_flags.size() != elements_.size()) {
-    MS_LOG(EXCEPTION) << "Elements size should be equal to elements use flags size. " << ToString();
+  if (elements_use_flags.size() < elements_.size()) {
+    MS_LOG(EXCEPTION) << "Elements size should not be greater to elements use flags size. " << ToString();
   }
-  for (size_t i = 0; i < elements_use_flags.size(); ++i) {
+  for (size_t i = 0; i < elements_.size(); ++i) {
     MS_EXCEPTION_IF_NULL(elements_[i]);
     if (!elements_use_flags[i]) {
       const auto unuse_node_none = std::make_shared<AbstractScalar>(std::make_shared<Int32Imm>(0));

mindspore/core/abstract/ops/prim_structures.cc

@@ -309,6 +309,10 @@ AbstractBasePtr InferImplListAppend(const AnalysisEnginePtr &, const PrimitivePt
       }
       // Add one element in flag list.
       auto flags = GetSequenceNodeElementsUseFlags(node);
+      MS_EXCEPTION_IF_NULL(flags);
+      if (flags->size() >= new_list.size()) {
+        continue;
+      }
       (void)flags->emplace_back(false);
     }
   }

tests/st/control/test_list_append.py

@@ -0,0 +1,85 @@
+# Copyright 2021-2022 Huawei Technologies Co., Ltd
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ============================================================================
+""" test list control flow """
+import pytest
+import mindspore.context as context
+from mindspore import Tensor, dtype
+from mindspore.nn import Cell
+import mindspore.ops.operations as P
+
+
+@pytest.mark.skip(reason='Not support list  as parameter in while function yet')
+@pytest.mark.level0
+@pytest.mark.platform_x86_gpu_training
+@pytest.mark.platform_arm_ascend_training
+@pytest.mark.platform_x86_ascend_training
+@pytest.mark.env_onecard
+def test_while_list():
+    """
+    Feature: list in while.
+    Description: Infer list in while.
+    Expectation: Null.
+    """
+
+    class Net(Cell):
+        def __init__(self):
+            super().__init__()
+            self.addn = P.AddN()
+
+        def construct(self, x):
+            y = []
+            for _ in range(3):
+                while x < 10:
+                    y.append(x)
+                    x = self.addn(y)
+            return x
+
+    context.set_context(mode=context.GRAPH_MODE, save_graphs=True, save_graphs_path="./listir")
+    net = Net()
+    x = Tensor([1], dtype.float32)
+    print(net(x))
+
+
+@pytest.mark.level0
+@pytest.mark.platform_x86_gpu_training
+@pytest.mark.platform_arm_ascend_training
+@pytest.mark.platform_x86_ascend_training
+@pytest.mark.env_onecard
+def test_for_list():
+    """
+    Feature: list for.
+    Description: Infer list in for.
+    Expectation: Null.
+    """
+
+    def convert_points_to_homogeneous(points):
+        padding = [[0, 0] for _ in range(len(points.shape))]
+        padding[-1][-1] = 1
+        return padding
+
+    class Net(Cell):
+        def construct(self, x1, x2):
+            y1 = convert_points_to_homogeneous(x1)
+            y2 = convert_points_to_homogeneous(x2)
+            return y1, y2
+
+    context.set_context(mode=context.GRAPH_MODE)
+    x1 = Tensor([[[-1, -1],  # left top
+                  [1, -1],  # right top
+                  [-1, 5],  # left bottom
+                  [1, 5]]], dtype.float32)  # right bottom
+    x2 = Tensor([[0., 0.], [0., 0.], [0., 0.], [0., 0.], [0., 0.]], dtype.float32)
+    net = Net()
+    print(net(x1, x2))