rcu: Allow RCU grace-period cleanup to be preempted
authorPaul E. McKenney <paulmck@linux.vnet.ibm.com>
Thu, 21 Jun 2012 15:19:05 +0000 (08:19 -0700)
committerPaul E. McKenney <paulmck@linux.vnet.ibm.com>
Sun, 23 Sep 2012 14:41:53 +0000 (07:41 -0700)
commitc856bafae7f5b3f59ac1d99279a9b99b3b36ad12
tree2bc16cb69d397440c5082fd451d64db6efa67017
parentcabc49c1ff51baaf1958d501a7a616ce91245c93
rcu: Allow RCU grace-period cleanup to be preempted

RCU grace-period cleanup is currently carried out with interrupts
disabled, which can result in excessive latency spikes on large systems
(many hundreds or thousands of CPUs).  This patch therefore makes the
RCU grace-period cleanup be preemptible, including voluntary preemption
points, which should eliminate those latency spikes.  Similar spikes from
forcing of quiescent states will be dealt with similarly by later patches.

Updated to replace uses of spin_lock_irqsave() with spin_lock_irq(), as
suggested by Peter Zijlstra.

Reported-by: Mike Galbraith <mgalbraith@suse.de>
Reported-by: Dimitri Sivanich <sivanich@sgi.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Reviewed-by: Josh Triplett <josh@joshtriplett.org>
kernel/rcutree.c