RFR: enqueue barrier + some other things

[Roland Westrelin]

> Yes. I think we'd better have first patch that does not move the code, makes the actual code
> changes, maybe renaming methods, but keeping their bodies at the same code positions. Then the
> second patch that moves the code blocks around. It would be much easier to backport this way, as
> actual changes would be visible in first patch, and second patch could be reimplemented by
> reshuffling the existing code.

Here you go:

http://cr.openjdk.java.net/~roland/shenandoah/refactor-mem-utilities/webrev.00/

Only the memory graph utility functions, not moved.

Roland.