diff --git a/test/test_network_ops/test_roll.py b/test/test_network_ops/test_roll.py
index 7584f3deea43cd2c7c4e1f3b51caabe35aed76f5..fe4ab0e8c2149c7310a3b5a0dcf5dd9c8a54ddd0 100644
--- a/test/test_network_ops/test_roll.py
+++ b/test/test_network_ops/test_roll.py
@@ -36,32 +36,32 @@ class TestRoll(TestCase):
         output = output.numpy()
         return output
 
-    def test_roll_3_4_5_float32(self, device="npu"):
+    def test_roll_3_4_5_float32(self):
         input_x1 = self.generate_data(-1, 1, (3, 4, 5), np.float32)
         cpu_output1 = self.cpu_op_exec(input_x1, [2, 1], [0, 1])
         npu_output1 = self.npu_op_exec(input_x1, [2, 1], [0, 1])
         self.assertRtolEqual(cpu_output1, npu_output1)
     
-    def test_roll_3_4_5_float16(self, device="npu"):
+    def test_roll_3_4_5_float16(self):
         input_x1 = self.generate_data(-1, 1, (3, 4, 5), np.float16)
         input_cpu = input_x1.float()
         cpu_output1 = self.cpu_op_exec(input_cpu, [2, 1], [0, 1]).astype(np.float16)
         npu_output1 = self.npu_op_exec(input_x1, [2, 1], [0, 1])
         self.assertRtolEqual(cpu_output1, npu_output1)
     
-    def test_roll_30_40_50_int32(self, device="npu"):
+    def test_roll_30_40_50_int32(self):
         input_x1 = self.generate_data(-1, 1, (30, 40, 50), np.int32)
         cpu_output1 = self.cpu_op_exec(input_x1, [20], [])
         npu_output1 = self.npu_op_exec(input_x1, [20], [])
         self.assertRtolEqual(cpu_output1, npu_output1)
     
-    def test_roll_20_30_40_50_uint8(self, device="npu"):
+    def test_roll_20_30_40_50_uint8(self):
         input_x1 = self.generate_data(-1, 1, (20, 30, 40, 50), np.uint8)
         cpu_output1 = self.cpu_op_exec(input_x1, [-20, 30], [-1, 0])
         npu_output1 = self.npu_op_exec(input_x1, [-20, 30], [-1, 0])
         self.assertRtolEqual(cpu_output1, npu_output1)
     
-    def test_roll_20_30_40_50_flaot32(self, device="npu"):
+    def test_roll_20_30_40_50_flaot32(self):
         input_x1 = self.generate_data(-1, 1, (20, 30, 40, 50), np.float32)
         cpu_output1 = self.cpu_op_exec(input_x1, [30], [3])
         npu_output1 = self.npu_op_exec(input_x1, [30], [3])
diff --git a/test/test_network_ops/test_roll_6d.py b/test/test_network_ops/test_roll_6d.py
index f60236ea738fd75aa9c654ecd362dd546fe1bae7..da6a7a633f6275efc68145e81c79d75ad696ac11 100644
--- a/test/test_network_ops/test_roll_6d.py
+++ b/test/test_network_ops/test_roll_6d.py
@@ -36,7 +36,7 @@ class TestRoll6d(TestCase):
         output1 = output1.numpy()
         return output1
     
-    def test_roll_10_10_10_10_10_10_int8(self, device="npu"):
+    def test_roll_10_10_10_10_10_10_int8(self):
         input1 = self.generate_data(-1, 1, (10, 10, 10, 10, 10, 10), np.int8)
         cpu_output1 = self.cpu_op_exec(input1, [-20, 30, 5], [-3, -4, -5])
         npu_output1 = self.npu_op_exec(input1, [-20, 30, 5], [-3, -4, -5])
diff --git a/torch_npu/csrc/aten/ops/RollKernelNpu.cpp.cpp b/torch_npu/csrc/aten/ops/RollKernelNpu.cpp
similarity index 99%
rename from torch_npu/csrc/aten/ops/RollKernelNpu.cpp.cpp
rename to torch_npu/csrc/aten/ops/RollKernelNpu.cpp
index 2339fe93ecba86f258922d11d5d844bb0e071d62..9dafaff230bb0969f092889945c3c6ece8fa9e0e 100644
--- a/torch_npu/csrc/aten/ops/RollKernelNpu.cpp.cpp
+++ b/torch_npu/csrc/aten/ops/RollKernelNpu.cpp
@@ -108,6 +108,5 @@ at::Tensor NPUNativeFunctions::roll(
   roll_out_npu(result, self, shifts, dims);
   return result;
 }
-
 } // namespace native
 } // namespace at_npu
\ No newline at end of file