linux-kernel - [RFC PATCH 1/1] rcu: Use separate wait queues for leaders and followers

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [thread-next>] [day] [month] [year] [list]

Date:	Mon, 28 Jul 2014 10:58:07 -0400
From:	Pranith Kumar <bobby.prani@...il.com>
To:	"Paul E. McKenney" <paulmck@...ux.vnet.ibm.com>,
	Josh Triplett <josh@...htriplett.org>,
	Steven Rostedt <rostedt@...dmis.org>,
	Mathieu Desnoyers <mathieu.desnoyers@...icios.com>,
	Lai Jiangshan <laijs@...fujitsu.com>,
	linux-kernel@...r.kernel.org (open list:READ-COPY UPDATE...)
Subject: [RFC PATCH 1/1] rcu: Use separate wait queues for leaders and followers

Commit fbce7497ee5a ("rcu: Parallelize and economize NOCB kthread wakeups")

tries to reduce the wake up overhead by creating leader and follower nocb
kthreads. 

One thing overlooked here is that all the kthreads wait on the same wait queue.
When we try to wake up the leader threads on the wait queue, we also try to wake
up the follower threads because of which there is still wake up overhead. 

This commit tries to avoid that by using separate wait queues for the leaders and
followers.

Signed-off-by: Pranith Kumar <bobby.prani@...il.com>
---
 kernel/rcu/tree.h        |  3 ++-
 kernel/rcu/tree_plugin.h | 11 ++++++-----
 2 files changed, 8 insertions(+), 6 deletions(-)

diff --git a/kernel/rcu/tree.h b/kernel/rcu/tree.h
index f703ea8..915ca71 100644
--- a/kernel/rcu/tree.h
+++ b/kernel/rcu/tree.h
@@ -348,7 +348,8 @@ struct rcu_data {
 	atomic_long_t nocb_follower_count_lazy; /*  (approximate). */
 	int nocb_p_count;		/* # CBs being invoked by kthread */
 	int nocb_p_count_lazy;		/*  (approximate). */
-	wait_queue_head_t nocb_wq;	/* For nocb kthreads to sleep on. */
+	wait_queue_head_t nocb_leader_wq; /* leader kthreads sleep on. */
+	wait_queue_head_t nocb_follower_wq; /* follower kthreads sleep on. */
 	struct task_struct *nocb_kthread;
 	bool nocb_defer_wakeup;		/* Defer wakeup of nocb_kthread. */
 
diff --git a/kernel/rcu/tree_plugin.h b/kernel/rcu/tree_plugin.h
index 4242c94..bd6faba 100644
--- a/kernel/rcu/tree_plugin.h
+++ b/kernel/rcu/tree_plugin.h
@@ -2045,7 +2045,7 @@ static void wake_nocb_leader(struct rcu_data *rdp, bool force)
 	if (!ACCESS_ONCE(rdp_leader->nocb_leader_wake) || force) {
 		/* Prior xchg orders against prior callback enqueue. */
 		ACCESS_ONCE(rdp_leader->nocb_leader_wake) = true;
-		wake_up(&rdp_leader->nocb_wq);
+		wake_up(&rdp_leader->nocb_leader_wq);
 	}
 }
 
@@ -2220,7 +2220,7 @@ wait_again:
 	/* Wait for callbacks to appear. */
 	if (!rcu_nocb_poll) {
 		trace_rcu_nocb_wake(my_rdp->rsp->name, my_rdp->cpu, "Sleep");
-		wait_event_interruptible(my_rdp->nocb_wq,
+		wait_event_interruptible(my_rdp->nocb_leader_wq,
 					 ACCESS_ONCE(my_rdp->nocb_leader_wake));
 		/* Memory barrier handled by smp_mb() calls below and repoll. */
 	} else if (firsttime) {
@@ -2300,7 +2300,7 @@ wait_again:
 			 * List was empty, wake up the follower.
 			 * Memory barriers supplied by atomic_long_add().
 			 */
-			wake_up(&rdp->nocb_wq);
+			wake_up(&rdp->nocb_follower_wq);
 		}
 	}
 
@@ -2321,7 +2321,7 @@ static void nocb_follower_wait(struct rcu_data *rdp)
 		if (!rcu_nocb_poll) {
 			trace_rcu_nocb_wake(rdp->rsp->name, rdp->cpu,
 					    "FollowerSleep");
-			wait_event_interruptible(rdp->nocb_wq,
+			wait_event_interruptible(rdp->nocb_follower_wq,
 						 ACCESS_ONCE(rdp->nocb_follower_head));
 		} else if (firsttime) {
 			/* Don't drown trace log with "Poll"! */
@@ -2489,7 +2489,8 @@ void __init rcu_init_nohz(void)
 static void __init rcu_boot_init_nocb_percpu_data(struct rcu_data *rdp)
 {
 	rdp->nocb_tail = &rdp->nocb_head;
-	init_waitqueue_head(&rdp->nocb_wq);
+	init_waitqueue_head(&rdp->nocb_leader_wq);
+	init_waitqueue_head(&rdp->nocb_follower_wq);
 	rdp->nocb_follower_tail = &rdp->nocb_follower_head;
 }
 
-- 
1.9.1

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/