nn.VolumetricMaxUnpooling and cudnn.VolumetricMaxPooling don't play with each other

[vfonov]

If a model contains nn.VolumetricMaxPooling module (probably same goes for SpatialMaxPooling), and nn.VolumetricMaxUnpooling and then converted to cudnn using cudnn.convert, then at execution time following message shows up:

/home/vfonov/src/torch/install/share/lua/5.1/nn/THNN.lua:110: bad argument #4 to 'v' (cannot convert 'struct THCudaTensor *' to 'struct THCudaLongTensor *')
stack traceback:
	[C]: in function 'v'
	/home/vfonov/src/torch/install/share/lua/5.1/nn/THNN.lua:110: in function 'VolumetricMaxUnpooling_updateOutput'
	...orch/install/share/lua/5.1/nn/VolumetricMaxUnpooling.lua:31: in function <...orch/install/share/lua/5.1/nn/VolumetricMaxUnpooling.lua:23>
	[C]: in function 'xpcall'
	.../vfonov/src/torch/install/share/lua/5.1/nn/Container.lua:63: in function 'rethrowErrors'
	...vfonov/src/torch/install/share/lua/5.1/nn/Sequential.lua:44: in function 'forward'
	cnn_semisup.lua:271: in function 'opfunc'
	/home/vfonov/src/torch/install/share/lua/5.1/optim/adam.lua:37: in function 'adam'
	cnn_semisup.lua:283: in main chunk
	[C]: in function 'dofile'
	.../src/torch/install/lib/luarocks/rocks/trepl/scm-1/bin/th:145: in main chunk
	[C]: at 0x00405d50

This is due to the fact that cudnn.VolumetricMaxPooling module does not use indices and itime,iheight,iwidth that VolumetricMaxUnpooling tries to grab. Workaround is to disable conversion to cudnn for VolumetricMaxPooling