[D-runtime] thread_needLock() and related code

[Martin Nowak]

On Sun, 24 Jun 2012 18:53:23 +0200, Sean Kelly <sean at invisibleduck.org>  
wrote:

> Nope. I think the function is effectively ignored these days anyway.
>
I recently tried to measure performance differences w.r.t. always locking  
in the GC and didn't found any.
Spinlocking an uncontended lock is fairly cheap but I've not tested this  
on Windows.