DFA cycle_display

[David S. Miller]

I went digging into why -fsched-verbose >= 1 gives me
an abort in verify_local_live_at_start() on the current
mainline with Sparc.

It turns out that, specifically, disabling the emitting
of the cycle_display RTL makes the problem go away. Sparc
and IA64 are the two platforms that actually implement this
thing.

Richard recently fixed another issue with this things,
and noted that cycle_display is quite silly really and
should be implemented differently.

Would anyone have any problem with me killing this thing
off right now?