[PATCH 0/2] sched: migrate_disable() preparations

[Peter Zijlstra]

Hi!

These two patches are the result of Thomas pestering me with migrate_disable()
patches for upstream. The first one is a cleanup/fix for the existing
balance_callback machinery. The second (ab)uses the context_switch() tail
invocation of balance_callbacks() to push away 'undesirables' during CPU
hotplug after we've marked the CPU as !active.

With this in place, Thomas can do his horrible migrate_disable() thing ;-)