Doug Lea reports:

We finally got to a corner case that we left (literally!)
hanging in Phaser -- what to do when multiple sub-phasers register
and deregister such that they try to "steal" each others
registrations. This is a crazy thing for a user to do -- the
outcome will be non-deterministic, and isn't (and cannot
be) spec'ed tightly, but the code was left
as possibly freezing, which is not a nice thing to do.
But now is addressed in CVS update, along with enabling
a previously commented out test method in tck/PhaserTest.