I don't normally place much credence in wikipedia pages but this one seems reasonably good. And it explains why RCU's are a better alternative to multiple-reader-single-writer locks.
You don't care about this unless you're doing low level work on multiple CPU systems, which I do from time to time. I'm working on several at them moment, each small.
I'm new to rcu's but they look like a very low cost way to manage updates to high(er) contention data items which are updated (note the work "generations" in the wiki). This concept looks a lot like various academic techniques I've read about over they years to create safe software locks.