Multiprocessing: sharing a zero-sized tensor fails #5719
Description
- PyTorch version: 0.4.0a0+c7611f7
Repro:
import torch
import torch.multiprocessing as mp
q = mp.Queue()
t = torch.tensor([])
q.put(t)Result:
File "/data/users/sgross/python3/lib/python3.5/multiprocessing/queues.py", line 241, in _feed
obj = ForkingPickler.dumps(obj)
File "/data/users/sgross/python3/lib/python3.5/multiprocessing/reduction.py", line 50, in dumps
cls(buf, protocol).dump(obj)
File "/data/users/sgross/pytorch/torch/multiprocessing/reductions.py", line 117, in reduce_storage
df = multiprocessing.reduction.DupFd(fd)
File "/data/users/sgross/python3/lib/python3.5/multiprocessing/reduction.py", line 190, in DupFd
return resource_sharer.DupFd(fd)
File "/data/users/sgross/python3/lib/python3.5/multiprocessing/resource_sharer.py", line 48, in __init__
new_fd = os.dup(fd)
OSError: [Errno 9] Bad file descriptor