diff --git a/mindspore/lite/nnacl/minimal_filtering_generator.c b/mindspore/lite/nnacl/minimal_filtering_generator.c
index d9a4b3094ce..bac28b4aedd 100644
--- a/mindspore/lite/nnacl/minimal_filtering_generator.c
+++ b/mindspore/lite/nnacl/minimal_filtering_generator.c
@@ -257,6 +257,9 @@ void MatrixMultiplyVec(const float32x4_t *matrix_a, const float32x4_t *matrix_b,
 
 int WinogradWeightTransform(const float *weight_data, float *winograd_data, float *matrix_g, const float *matrix_gt,
                             int oc_block, int input_unit, int kernel_unit, int channel, int batch, bool pack) {
+  if (oc_block == 0) {
+    return NNACL_PARAM_INVALID;
+  }
   // original weight format : ohwi
   int oc_block_num = UP_DIV(batch, oc_block);
   int block_stride = channel * oc_block;
diff --git a/mindspore/lite/src/runtime/kernel/arm/fp32_grad/sparse_softmax_cross_entropy_with_logits.cc b/mindspore/lite/src/runtime/kernel/arm/fp32_grad/sparse_softmax_cross_entropy_with_logits.cc
index 4e29045ef23..01135de869f 100644
--- a/mindspore/lite/src/runtime/kernel/arm/fp32_grad/sparse_softmax_cross_entropy_with_logits.cc
+++ b/mindspore/lite/src/runtime/kernel/arm/fp32_grad/sparse_softmax_cross_entropy_with_logits.cc
@@ -101,7 +101,7 @@ int SparseSoftmaxCrossEntropyWithLogitsCPUKernel::Execute(int task_id) {
   Softmax(ins, losses_, sum_data_, &sm_params_);
   if (is_train()) {
     GradPostExecute(labels, losses_, grads, out);
-  } else if (out != nullptr) {
+  } else {
     ForwardPostExecute(labels, losses_, out);
   }
   return RET_OK;
diff --git a/mindspore/lite/src/runtime/thread_pool.c b/mindspore/lite/src/runtime/thread_pool.c
index 4bb87cec535..dbed7bcb59b 100644
--- a/mindspore/lite/src/runtime/thread_pool.c
+++ b/mindspore/lite/src/runtime/thread_pool.c
@@ -835,6 +835,10 @@ ThreadPool *CreateThreadPool(int thread_num, int mode) {
   }
 #endif
   ThreadPool *thread_pool = (struct ThreadPool *)(malloc(sizeof(ThreadPool)));
+  if (thread_pool == NULL) {
+    LOG_ERROR("Malloc ThreadPool failed");
+    return NULL;
+  }
   thread_pool->thread_num = thread_num > MAX_THREAD_NUM ? MAX_THREAD_NUM : thread_num;
   thread_pool->is_alive = ATOMIC_VAR_INIT(true);
   thread_pool->mode = mode;
diff --git a/mindspore/lite/tools/benchmark/benchmark.cc b/mindspore/lite/tools/benchmark/benchmark.cc
index 0d7f10d634e..3d3bc9ab690 100644
--- a/mindspore/lite/tools/benchmark/benchmark.cc
+++ b/mindspore/lite/tools/benchmark/benchmark.cc
@@ -275,7 +275,7 @@ int Benchmark::CompareStringData(const std::string &name, tensor::MSTensor *tens
         std::cout << "  " << output_strings[i] << std::endl;
       }
       if (calib_strings[i] != output_strings[i]) {
-        MS_LOG(ERROR) << "";
+        MS_LOG(ERROR) << "Compare failed, index: " << i;
         return RET_ERROR;
       }
     }