Tim, have you read the replies to my post on c.p.t regarding ABA bits? Even on this thread someone (I think Helge) argued that even 32 bits may not be enough. Now I'm thinking that maybe "generation counter" solution may not be workable as a general solution.
For ABA, overflow (or wraparound, really) shouldn't be a problem. All you are trying to do is make sure you don't see the same pointer twice, while in a CAS. Seeing the same pointer twice does happen, but not that often. What are the chances that not only will it be the same pointer, but also that the 32 bits have wrapped around back to the same value? I would suspect that the CAS'ing thread must have suspended for quite a while in order for the 32 bits to loop around, and on top of that, to hit the 1 / 2^32 chance of landing on the same value...
Andy.
Tony