rcu: Segregate rcu_state fields to improve cache locality
authorDimitri Sivanich <sivanich@sgi.com>
Fri, 29 Jun 2012 21:17:29 +0000 (14:17 -0700)
committerPaul E. McKenney <paulmck@linux.vnet.ibm.com>
Sun, 23 Sep 2012 14:41:53 +0000 (07:41 -0700)
The fields in the rcu_state structure that are protected by the
root rcu_node structure's ->lock can share a cache line with the
fields protected by ->onofflock.  This can result in excessive
memory contention on large systems, so this commit applies
____cacheline_internodealigned_in_smp to the ->onofflock field in
order to segregate them.

Signed-off-by: Dimitri Sivanich <sivanich@sgi.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Tested-by: Dimitri Sivanich <sivanich@sgi.com>
Reviewed-by: Josh Triplett <josh@joshtriplett.org>
kernel/rcutree.h

index effb2733b7fc1dbac2fdb3ff5a9c98e243c2a405..5d92b80a0a281c58dcc887c377ff84af594714c6 100644 (file)
@@ -394,7 +394,8 @@ struct rcu_state {
 
        /* End of fields guarded by root rcu_node's lock. */
 
-       raw_spinlock_t onofflock;               /* exclude on/offline and */
+       raw_spinlock_t onofflock ____cacheline_internodealigned_in_smp;
+                                               /* exclude on/offline and */
                                                /*  starting new GP. */
        struct rcu_head *orphan_nxtlist;        /* Orphaned callbacks that */
                                                /*  need a grace period. */