> Undefined behavior (or may be just the wrong behavior), we obtain later, > when 
> converting in PyLong_AsVoidPtr() zero integer back to pointer. On
> platforms 
> where converting NULL to an integer doesn't give zero it's a bug.

So maybe the fix should be to special case zero in PyLong_AsVoidPtr, and turn 0L back into NULL there?

I think it's just too risky to change the current behaviour in 3.2 and 3.3, given the number of places that PyLong_FromVoidPtr is used.  Unlike you, I don't have confidence that there are no current or future platforms that don't map NULL to 0.  It might be reasonable to consider a change for 3.4.

> I mean using PyLong_FromUnsignedLongLong((unsigned PY_LONG_LONG)
> ((Py_uintptr_t)p - (Py_uintptr_t)(void *)NULL)) instead ...

Okay, thanks for explaining.