rcu: Cure load woes
authorPeter Zijlstra <peterz@infradead.org>
Mon, 30 May 2011 11:34:51 +0000 (13:34 +0200)
committerIngo Molnar <mingo@elte.hu>
Tue, 31 May 2011 08:01:48 +0000 (10:01 +0200)
commitd72bce0e67e8afc6eb959f656013cbb577426f1e
tree9c93d4df9aa895d6f2f555e0cf50e7ae5ebaded4
parent55922c9d1b84b89cb946c777fddccb3247e7df2c
rcu: Cure load woes

Commit cc3ce5176d83 (rcu: Start RCU kthreads in TASK_INTERRUPTIBLE
state) fudges a sleeping task' state, resulting in the scheduler seeing
a TASK_UNINTERRUPTIBLE task going to sleep, but a TASK_INTERRUPTIBLE
task waking up. The result is unbalanced load calculation.

The problem that patch tried to address is that the RCU threads could
stay in UNINTERRUPTIBLE state for quite a while and triggering the hung
task detector due to on-demand wake-ups.

Cure the problem differently by always giving the tasks at least one
wake-up once the CPU is fully up and running, this will kick them out of
the initial UNINTERRUPTIBLE state and into the regular INTERRUPTIBLE
wait state.

[ The alternative would be teaching kthread_create() to start threads as
  INTERRUPTIBLE but that needs a tad more thought. ]

Reported-by: Damien Wyart <damien.wyart@free.fr>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Acked-by: Paul E. McKenney <paul.mckenney@linaro.org>
Link: http://lkml.kernel.org/r/1306755291.1200.2872.camel@twins
Signed-off-by: Ingo Molnar <mingo@elte.hu>
kernel/rcutree.c
kernel/rcutree_plugin.h