Regarding Null Move Pruning

[hgm]

If it segfaults you should just check in which node it segfaults, and then determine where exactly in the code by putting some print statements in that node (making sure that what these print is not buffered).

[JohnWoe]

En Passant square must be disabled since a move is passed and it's not legal anymore.
It was a source of bugs when I copied null move logic from Xiphos ... eh developed it from thin air

[syzygy]

I see that before applying the search in the null-move you just change the side to move but you keep the current ply in the search tree. This will overwrite the previous move made and saved in the search stack at the same ply.
I don't know the details of your implementation, but I would verify that first.

Indeed it seems much safer to actually "make" the nullmove. Even if simply switching side works now, when the engine starts to keep track of more state for move ordering and extensions/reductions, it will fail.