[Cbe-oss-dev] [PATCH 0:6] scheduler fixes

[Luke Browning]

This patch set implements a different queuing strategy for contexts that
have class 1 faults.  These faults generally take longer to be resolved
and we don't want contexts in this state to be put on the run queue as
it would provide a false indication that are jobs waiting to run.  This
can lead to extra context switching, particularly, in the time slicer as
it puts preempted jobs back onto the run queue.  

This patch set also implements a few other fixes that are not related to
the new queuing strategy.  They are the first N-1 patches.

I built these using Jeremy's internal git.  I couuldn't get the public
one to boot.  I started with Christoph's recent fix, 

spufs: don't requeue victim contex in find_victim if it's not in spu_run

which has a bug.

Regards, Luke