diff --git a/paddle/fluid/operators/reader/create_custom_reader_op.cc b/paddle/fluid/operators/reader/create_custom_reader_op.cc
index 659f7d595cebcb5b53b0f24c936a06cbd1acbc52..74e6b79a2a3ddd7562d190dcab9dc42324be3cf9 100644
--- a/paddle/fluid/operators/reader/create_custom_reader_op.cc
+++ b/paddle/fluid/operators/reader/create_custom_reader_op.cc
@@ -22,12 +22,11 @@ namespace reader {
 class CustomReader : public framework::DecoratedReader {
  public:
   CustomReader(ReaderBase* reader, const framework::BlockDesc* sub_block,
-               const framework::Scope* scope, const platform::Place& dev_place,
+               const platform::Place& dev_place,
                const std::vector<std::string>& source_var_names,
                const std::vector<std::string>& sink_var_names)
       : DecoratedReader(reader),
         sub_block_(sub_block),
-        scope_(scope),
         exe_(framework::Executor(dev_place)),
         source_var_names_(source_var_names),
         sink_var_names_(sink_var_names) {}
@@ -37,12 +36,10 @@ class CustomReader : public framework::DecoratedReader {
   void UpdateBlockAndScope(const framework::BlockDesc* sub_block,
                            const framework::Scope* scope) {
     sub_block_ = sub_block;
-    scope_ = scope;
   }
 
  private:
   const framework::BlockDesc* sub_block_;
-  const framework::Scope* scope_;
   framework::Executor exe_;
 
   std::vector<std::string> source_var_names_;
@@ -67,7 +64,7 @@ class CreateCustomReaderOp : public framework::OperatorBase {
     const auto& underlying_reader = scope.FindVar(Input("UnderlyingReader"))
                                         ->Get<framework::ReaderHolder>();
     out->Reset(
-        new CustomReader(underlying_reader.Get(), sub_block, &scope, dev_place,
+        new CustomReader(underlying_reader.Get(), sub_block, dev_place,
                          Attr<std::vector<std::string>>("source_var_names"),
                          Attr<std::vector<std::string>>("sink_var_names")));
   }
@@ -150,27 +147,29 @@ void CustomReader::ReadNext(std::vector<framework::LoDTensor>* out) {
       "the size of underlying_outs(%d) are not consistent. Each feeding "
       "element must have its own source and sink variable.",
       source_var_names_.size(), sink_var_names_.size(), underlying_outs.size());
-
-  framework::Scope* exe_scope = &scope_->NewScope();
+  // The scope for CustomReader's sub-block should be independent and shouldn't
+  // be any other computation scope's child. Otherwise, data preprocessing and
+  // compution cannot be concurrent.
+  auto* scope = new framework::Scope();
   // 1. Copy LoDTensors from underlying reader's output to source variables.
   for (size_t i = 0; i < source_var_names_.size(); ++i) {
-    framework::Variable* var = exe_scope->Var(source_var_names_[i]);
+    framework::Variable* var = scope->Var(source_var_names_[i]);
     framework::LoDTensor* tensor = var->GetMutable<framework::LoDTensor>();
     tensor->ShareDataWith(underlying_outs[i]);
     tensor->set_lod(underlying_outs[i].lod());
   }
   // 2. Run the sub-block.
   framework::ProgramDesc* program = sub_block_->Program();
-  exe_.Run(*program, exe_scope, sub_block_->ID(), false, true);
+  exe_.Run(*program, scope, sub_block_->ID(), false, true);
   // 3. Copy LoDTensors from sink variables to out.
   out->resize(sink_var_names_.size());
   for (size_t i = 0; i < sink_var_names_.size(); ++i) {
-    framework::Variable* var = exe_scope->FindVar(sink_var_names_[i]);
+    framework::Variable* var = scope->FindVar(sink_var_names_[i]);
     PADDLE_ENFORCE_NOT_NULL(var);
     const framework::LoDTensor& tensor = var->Get<framework::LoDTensor>();
     framework::TensorCopySync(tensor, platform::CPUPlace(), &(*out)[i]);
   }
-  scope_->DeleteScope(exe_scope);
+  delete scope;
 }
 
 }  // namespace reader
diff --git a/python/paddle/fluid/tests/unittests/test_preprocessor.py b/python/paddle/fluid/tests/unittests/test_preprocessor.py
index 37dd366f3c88eb3aeb4e67f48fd225e1162ff38b..cbf1a7e0c50a87cd43507ffdb94109873cf4e5d9 100644
--- a/python/paddle/fluid/tests/unittests/test_preprocessor.py
+++ b/python/paddle/fluid/tests/unittests/test_preprocessor.py
@@ -74,7 +74,8 @@ class TestPreprocessor(unittest.TestCase):
                 lbl_out = lbl + 1
                 preprocessor.outputs(img_out, lbl_out)
 
-            img, lbl = fluid.layers.io.read_file(preprocessor())
+            data_file = fluid.layers.io.double_buffer(preprocessor())
+            img, lbl = fluid.layers.io.read_file(data_file)
 
             if fluid.core.is_compiled_with_cuda():
                 place = fluid.CUDAPlace(0)