On RT enabled kernels most of the code including spin/rw lock held sections are preemptible, which also makes the tasks migrateable. That violates the per CPU constraints. RT needs therefore a mechanism to control migration independent of preemption.
This series is providing a mostly rewritten version of the RT code based on the newest scheduler bits and includes and depends on Peter's 'let tasks kick themself off the outgoing CPU' feature. While migrate disable/enable are trivial in principle the tricky part is the coordination and serialization against concurrent affinity changes, e.g. sched_setaffinity(2), and CPU hotplug. Thanks, tglx