No subject

lock to use in the clock - mutex or spinlock.

I don't see that having some clocks be one and others another is really
acceptable - think about the resulting mess if you end up with some
parent mux'd clocks which are a mutex and others which are a spinlock.
Can a driver use an atomic call for that?  Sometimes depending on the
mux, sometimes depending on whether a parent clock is already enabled.

What if your clock was enabled while the mux selected another spinlock-
locked clk, but then you switched to a different operating point and the
mux then selected the mutex-locked clock.

We could say that this is illegal - but then we need to have the code
explicitly check this otherwise such situations will be created.