Fix mp serialization for integer nn.Parameter on CUDA (#56529)

Summary:
Fixes https://github.com/pytorch/pytorch/issues/56342

Pull Request resolved: https://github.com/pytorch/pytorch/pull/56529

Reviewed By: albanD

Differential Revision: D27896094

Pulled By: ngimel

fbshipit-source-id: fe817781eb7139ea57c78acfd56e7c11b61eb4ed

test/test_multiprocessing.py

@@ -832,14 +832,31 @@ if __name__ == "__main__":
 
     @unittest.skipIf(NO_MULTIPROCESSING_SPAWN, "Disabled for environments that \
                      don't support multiprocessing with spawn start method")
-    def test_integer_parameter_serialization(self):
-        iparam = torch.nn.Parameter(torch.tensor(0, dtype=torch.int64), requires_grad=False)
+    def test_integer_parameter_serialization_cpu(self):
+        self._test_integer_parameter_serialization(device='cpu')
+
+    @unittest.skipIf(NO_MULTIPROCESSING_SPAWN, "Disabled for environments that \
+                     don't support multiprocessing with spawn start method")
+    @unittest.skipIf(not TEST_CUDA_IPC, 'CUDA IPC not available')
+    def test_integer_parameter_serialization_cuda(self):
+        self._test_integer_parameter_serialization(device='cuda')
+
+    def _test_integer_parameter_serialization(self, device):
+        param = torch.nn.Parameter(
+            torch.tensor(0, dtype=torch.int64, device=device),
+            requires_grad=False
+        )
 
         ctx = mp.get_context('spawn')
-        p = ctx.Process(target=integer_parameter_serialization, args=(iparam,))
+        p = ctx.Process(target=integer_parameter_serialization, args=(param,))
         p.start()
         p.join()
 
+        self.assertEqual(
+            0, p.exitcode,
+            msg=f'Failed to serialize successfully for "{device}" device!'
+        )
+
     def test_empty_shared(self):
         t = torch.tensor([])
         t.share_memory_()

torch/multiprocessing/reductions.py

@@ -123,9 +123,14 @@ def rebuild_cuda_tensor(tensor_cls, tensor_size, tensor_stride, tensor_offset,
             storage_cls._release_ipc_counter(ref_counter_handle, ref_counter_offset)
 
     t = torch._utils._rebuild_tensor(storage, tensor_offset, tensor_size, tensor_stride)
+
     if tensor_cls == torch.nn.parameter.Parameter:
-        t = torch.nn.parameter.Parameter(t)
-    t.requires_grad = requires_grad
+        # It is crucial for integer tensors to receive
+        # the requires_grad=False as an argument in the constructor
+        t = torch.nn.parameter.Parameter(t, requires_grad=requires_grad)
+    else:
+        t.requires_grad = requires_grad
+
     return t