Add test to check if importing DALI doesn't break Torch process forking (#3669)

- adds a test with PyTorch to make sure that importing DALI
  doesn't interfere with torch.multiprocessing

Signed-off-by: Janusz Lisiecki <jlisiecki@nvidia.com>

Author: JanuszL

dali/test/python/test_dali_fork_torch.py

@@ -0,0 +1,32 @@
+# Copyright (c) 2022, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+#
+# 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.
+
+import torch
+from torch.multiprocessing import Process
+# we need this import to check if it is safe to import DALI and not touch the CUDA runtime
+# that could crash forked process
+import nvidia.dali as dali
+
+
+def task_function():
+    torch.cuda.set_device(0)
+
+
+def test_actual_proc():
+    phase_process = Process(target=task_function)
+    # phase_process.daemon = True
+    phase_process.start()
+    phase_process.join()
+    assert phase_process.exitcode == 0
+

qa/TL0_python_self_test_frameworks/test_pytorch.sh

@@ -12,6 +12,7 @@ test_body() {
     nosetests --verbose test_external_source_pytorch_dlpack.py
     nosetests --verbose test_external_source_parallel_pytorch.py
     nosetests --verbose test_backend_impl_torch_dlpack.py
+    nosetests --verbose test_dali_fork_torch.py
     nosetests --verbose --attr 'pytorch' test_external_source_impl_utils.py
     nosetests --verbose --attr 'pytorch' test_pipeline_debug.py
 }