I share your desire to have the check for this shared in core code,
but I'm afraid that just is not going to work.
Okay, so this series is definately not it. Probably the previous one
which exposes helpers is not a bad idea after all. Because it is
ulitmately the driver's decision when to use the helpers...