diff --git a/paddle/fluid/operators/data_norm_op.cc b/paddle/fluid/operators/data_norm_op.cc
index 91e8c04a3d3d858efff9ee88f9b0d487095b3aac..a89d2949a2bb4f2cd99be9928009885028edf6c6 100644
--- a/paddle/fluid/operators/data_norm_op.cc
+++ b/paddle/fluid/operators/data_norm_op.cc
@@ -232,7 +232,8 @@ class DataNormOpMaker : public framework::OpProtoAndCheckerMaker {
         .SetDefault(false);
     AddAttr<bool>("use_mkldnn",
                   "(bool, default false) Only used in mkldnn kernel")
-        .SetDefault(false);
+        .SetDefault(false)
+        .AsExtra();
     AddInput("X", "The input tensor");
     AddInput("BatchSize",
              "BatchSize is a 1-dimensional tensor of size C "
diff --git a/python/paddle/fluid/layers/nn.py b/python/paddle/fluid/layers/nn.py
index 4216384b6f8b25b7ad7d09b6cca2cde504b6e1d8..87b7b28f90078dbf831b1916e07f812d403a9007 100755
--- a/python/paddle/fluid/layers/nn.py
+++ b/python/paddle/fluid/layers/nn.py
@@ -3398,6 +3398,7 @@ def data_norm(input,
     }
     attrs = {
         "epsilon": epsilon,
+        "data_layout": data_layout,
         "sync_stats": sync_stats,
         "summary_decay_rate": summary_decay_rate,
     }