linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
To: linux-kernel@vger.kernel.org
Cc: mingo@elte.hu, laijs@cn.fujitsu.com, dipankar@in.ibm.com,
	akpm@linux-foundation.org, mathieu.desnoyers@polymtl.ca,
	josh@joshtriplett.org, niv@us.ibm.com, tglx@linutronix.de,
	peterz@infradead.org, rostedt@goodmis.org,
	Valdis.Kletnieks@vt.edu, dhowells@redhat.com,
	eric.dumazet@gmail.com, darren@dvhart.com, fweisbec@gmail.com,
	sbw@mit.edu, patches@linaro.org,
	"Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
Subject: [PATCH tip/core/rcu 12/23] rcu: Prevent force_quiescent_state() memory contention
Date: Thu, 30 Aug 2012 11:18:27 -0700	[thread overview]
Message-ID: <1346350718-30937-12-git-send-email-paulmck@linux.vnet.ibm.com> (raw)
In-Reply-To: <1346350718-30937-1-git-send-email-paulmck@linux.vnet.ibm.com>

From: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>

Large systems running RCU_FAST_NO_HZ kernels see extreme memory
contention on the rcu_state structure's ->fqslock field.  This
can be avoided by disabling RCU_FAST_NO_HZ, either at compile time
or at boot time (via the nohz kernel boot parameter), but large
systems will no doubt become sensitive to energy consumption.
This commit therefore uses a combining-tree approach to spread the
memory contention across new cache lines in the leaf rcu_node structures.
This can be thought of as a tournament lock that has only a try-lock
acquisition primitive.

The effect on small systems is minimal, because such systems have
an rcu_node "tree" consisting of a single node.  In addition, this
functionality is not used on fastpaths.

Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
---
 kernel/rcutree.c |   47 +++++++++++++++++++++++++++++++++++++----------
 kernel/rcutree.h |    1 +
 2 files changed, 38 insertions(+), 10 deletions(-)

diff --git a/kernel/rcutree.c b/kernel/rcutree.c
index cce73ff..ed1be62 100644
--- a/kernel/rcutree.c
+++ b/kernel/rcutree.c
@@ -61,6 +61,7 @@
 /* Data structures. */
 
 static struct lock_class_key rcu_node_class[RCU_NUM_LVLS];
+static struct lock_class_key rcu_fqs_class[RCU_NUM_LVLS];
 
 #define RCU_STATE_INITIALIZER(sname, cr) { \
 	.level = { &sname##_state.node[0] }, \
@@ -1824,16 +1825,35 @@ static void force_qs_rnp(struct rcu_state *rsp, int (*f)(struct rcu_data *))
 static void force_quiescent_state(struct rcu_state *rsp)
 {
 	unsigned long flags;
-	struct rcu_node *rnp = rcu_get_root(rsp);
+	bool ret;
+	struct rcu_node *rnp;
+	struct rcu_node *rnp_old = NULL;
+
+	/* Funnel through hierarchy to reduce memory contention. */
+	rnp = per_cpu_ptr(rsp->rda, raw_smp_processor_id())->mynode;
+	for (; rnp != NULL; rnp = rnp->parent) {
+		ret = (ACCESS_ONCE(rsp->gp_flags) & RCU_GP_FLAG_FQS) ||
+		      !raw_spin_trylock(&rnp->fqslock);
+		if (rnp_old != NULL)
+			raw_spin_unlock(&rnp_old->fqslock);
+		if (ret) {
+			rsp->n_force_qs_lh++;
+			return;
+		}
+		rnp_old = rnp;
+	}
+	/* rnp_old == rcu_get_root(rsp), rnp == NULL. */
 
-	if (ACCESS_ONCE(rsp->gp_flags) & RCU_GP_FLAG_FQS)
+	/* Reached the root of the rcu_node tree, acquire lock. */
+	raw_spin_lock_irqsave(&rnp_old->lock, flags);
+	raw_spin_unlock(&rnp_old->fqslock);
+	if (ACCESS_ONCE(rsp->gp_flags) & RCU_GP_FLAG_FQS) {
+		rsp->n_force_qs_lh++;
+		raw_spin_unlock_irqrestore(&rnp_old->lock, flags);
 		return;  /* Someone beat us to it. */
-	if (!raw_spin_trylock_irqsave(&rnp->lock, flags)) {
-		rsp->n_force_qs_lh++; /* Inexact, can lose counts.  Tough! */
-		return;
 	}
 	rsp->gp_flags |= RCU_GP_FLAG_FQS;
-	raw_spin_unlock_irqrestore(&rnp->lock, flags);
+	raw_spin_unlock_irqrestore(&rnp_old->lock, flags);
 	wake_up(&rsp->gp_wq);  /* Memory barrier implied by wake_up() path. */
 }
 
@@ -2721,10 +2741,14 @@ static void __init rcu_init_levelspread(struct rcu_state *rsp)
 static void __init rcu_init_one(struct rcu_state *rsp,
 		struct rcu_data __percpu *rda)
 {
-	static char *buf[] = { "rcu_node_level_0",
-			       "rcu_node_level_1",
-			       "rcu_node_level_2",
-			       "rcu_node_level_3" };  /* Match MAX_RCU_LVLS */
+	static char *buf[] = { "rcu_node_0",
+			       "rcu_node_1",
+			       "rcu_node_2",
+			       "rcu_node_3" };  /* Match MAX_RCU_LVLS */
+	static char *fqs[] = { "rcu_node_fqs_0",
+			       "rcu_node_fqs_1",
+			       "rcu_node_fqs_2",
+			       "rcu_node_fqs_3" };  /* Match MAX_RCU_LVLS */
 	int cpustride = 1;
 	int i;
 	int j;
@@ -2749,6 +2773,9 @@ static void __init rcu_init_one(struct rcu_state *rsp,
 			raw_spin_lock_init(&rnp->lock);
 			lockdep_set_class_and_name(&rnp->lock,
 						   &rcu_node_class[i], buf[i]);
+			raw_spin_lock_init(&rnp->fqslock);
+			lockdep_set_class_and_name(&rnp->fqslock,
+						   &rcu_fqs_class[i], fqs[i]);
 			rnp->gpnum = 0;
 			rnp->qsmask = 0;
 			rnp->qsmaskinit = 0;
diff --git a/kernel/rcutree.h b/kernel/rcutree.h
index 36916df..2d4cc18 100644
--- a/kernel/rcutree.h
+++ b/kernel/rcutree.h
@@ -202,6 +202,7 @@ struct rcu_node {
 				/*  per-CPU kthreads as needed. */
 	unsigned int node_kthread_status;
 				/* State of node_kthread_task for tracing. */
+	raw_spinlock_t fqslock ____cacheline_internodealigned_in_smp;
 } ____cacheline_internodealigned_in_smp;
 
 /*
-- 
1.7.8


  parent reply	other threads:[~2012-08-30 18:19 UTC|newest]

Thread overview: 85+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2012-08-30 18:18 [PATCH tip/core/rcu 0/23] Improvements to RT response on big systems and expedited functions Paul E. McKenney
2012-08-30 18:18 ` [PATCH tip/core/rcu 01/23] rcu: Move RCU grace-period initialization into a kthread Paul E. McKenney
2012-08-30 18:18   ` [PATCH tip/core/rcu 02/23] rcu: Allow RCU grace-period initialization to be preempted Paul E. McKenney
2012-09-02  1:09     ` Josh Triplett
2012-09-05  1:22       ` Paul E. McKenney
2012-08-30 18:18   ` [PATCH tip/core/rcu 03/23] rcu: Move RCU grace-period cleanup into kthread Paul E. McKenney
2012-09-02  1:22     ` Josh Triplett
2012-09-06 13:34     ` Peter Zijlstra
2012-09-06 17:29       ` Paul E. McKenney
2012-08-30 18:18   ` [PATCH tip/core/rcu 04/23] rcu: Allow RCU grace-period cleanup to be preempted Paul E. McKenney
2012-09-02  1:36     ` Josh Triplett
2012-08-30 18:18   ` [PATCH tip/core/rcu 05/23] rcu: Prevent offline CPUs from executing RCU core code Paul E. McKenney
2012-09-02  1:45     ` Josh Triplett
2012-08-30 18:18   ` [PATCH tip/core/rcu 06/23] rcu: Break up rcu_gp_kthread() into subfunctions Paul E. McKenney
2012-09-02  2:11     ` Josh Triplett
2012-09-06 13:39     ` Peter Zijlstra
2012-09-06 17:32       ` Paul E. McKenney
2012-09-06 18:49         ` Josh Triplett
2012-09-06 19:09           ` Peter Zijlstra
2012-09-06 20:30             ` Paul E. McKenney
2012-09-06 20:30           ` Paul E. McKenney
2012-08-30 18:18   ` [PATCH tip/core/rcu 07/23] rcu: Provide OOM handler to motivate lazy RCU callbacks Paul E. McKenney
2012-09-02  2:13     ` Josh Triplett
2012-09-03  9:08     ` Lai Jiangshan
2012-09-05 17:45       ` Paul E. McKenney
2012-09-06 13:46     ` Peter Zijlstra
2012-09-06 13:52       ` Steven Rostedt
2012-09-06 17:41         ` Paul E. McKenney
2012-09-06 17:46           ` Peter Zijlstra
2012-09-06 20:32             ` Paul E. McKenney
2012-08-30 18:18   ` [PATCH tip/core/rcu 08/23] rcu: Segregate rcu_state fields to improve cache locality Paul E. McKenney
2012-09-02  2:51     ` Josh Triplett
2012-08-30 18:18   ` [PATCH tip/core/rcu 09/23] rcu: Move quiescent-state forcing into kthread Paul E. McKenney
2012-08-30 18:18   ` [PATCH tip/core/rcu 10/23] rcu: Allow RCU quiescent-state forcing to be preempted Paul E. McKenney
2012-09-02  5:23     ` Josh Triplett
2012-08-30 18:18   ` [PATCH tip/core/rcu 11/23] rcu: Adjust debugfs tracing for kthread-based quiescent-state forcing Paul E. McKenney
2012-09-02  6:05     ` Josh Triplett
2012-08-30 18:18   ` Paul E. McKenney [this message]
2012-09-02 10:47     ` [PATCH tip/core/rcu 12/23] rcu: Prevent force_quiescent_state() memory contention Josh Triplett
2012-08-30 18:18   ` [PATCH tip/core/rcu 13/23] rcu: Control grace-period duration from sysfs Paul E. McKenney
2012-09-03  9:30     ` Josh Triplett
2012-09-03  9:31       ` Josh Triplett
2012-09-06 14:15     ` Peter Zijlstra
2012-09-06 17:53       ` Paul E. McKenney
2012-09-06 18:28         ` Peter Zijlstra
2012-09-06 20:37           ` Paul E. McKenney
2012-08-30 18:18   ` [PATCH tip/core/rcu 14/23] rcu: Remove now-unused rcu_state fields Paul E. McKenney
2012-09-03  9:31     ` Josh Triplett
2012-09-06 14:17     ` Peter Zijlstra
2012-09-06 18:02       ` Paul E. McKenney
2012-08-30 18:18   ` [PATCH tip/core/rcu 15/23] rcu: Make rcutree module parameters visible in sysfs Paul E. McKenney
2012-09-03  9:32     ` Josh Triplett
2012-08-30 18:18   ` [PATCH tip/core/rcu 16/23] rcu: Prevent initialization-time quiescent-state race Paul E. McKenney
2012-09-03  9:37     ` Josh Triplett
2012-09-05 18:19       ` Paul E. McKenney
2012-09-05 18:55         ` Josh Triplett
2012-09-05 19:49           ` Paul E. McKenney
2012-09-06 14:21         ` Peter Zijlstra
2012-09-06 16:18           ` Paul E. McKenney
2012-09-06 16:22             ` Peter Zijlstra
2012-08-30 18:18   ` [PATCH tip/core/rcu 17/23] rcu: Fix day-zero grace-period initialization/cleanup race Paul E. McKenney
2012-09-03  9:39     ` Josh Triplett
2012-09-06 14:24     ` Peter Zijlstra
2012-09-06 18:06       ` Paul E. McKenney
2012-08-30 18:18   ` [PATCH tip/core/rcu 18/23] rcu: Add random PROVE_RCU_DELAY to grace-period initialization Paul E. McKenney
2012-09-03  9:41     ` Josh Triplett
2012-09-06 14:27     ` Peter Zijlstra
2012-09-06 18:25       ` Paul E. McKenney
2012-08-30 18:18   ` [PATCH tip/core/rcu 19/23] rcu: Adjust for unconditional ->completed assignment Paul E. McKenney
2012-09-03  9:42     ` Josh Triplett
2012-08-30 18:18   ` [PATCH tip/core/rcu 20/23] rcu: Remove callback acceleration from grace-period initialization Paul E. McKenney
2012-09-03  9:42     ` Josh Triplett
2012-08-30 18:18   ` [PATCH tip/core/rcu 21/23] rcu: Eliminate signed overflow in synchronize_rcu_expedited() Paul E. McKenney
2012-09-03  9:43     ` Josh Triplett
2012-08-30 18:18   ` [PATCH tip/core/rcu 22/23] rcu: Reduce synchronize_rcu_expedited() latency Paul E. McKenney
2012-09-03  9:46     ` Josh Triplett
2012-08-30 18:18   ` [PATCH tip/core/rcu 23/23] rcu: Simplify quiescent-state detection Paul E. McKenney
2012-09-03  9:56     ` Josh Triplett
2012-09-06 14:36     ` Peter Zijlstra
2012-09-06 20:01       ` Paul E. McKenney
2012-09-06 21:18         ` Mathieu Desnoyers
2012-09-06 21:31           ` Paul E. McKenney
2012-09-02  1:04   ` [PATCH tip/core/rcu 01/23] rcu: Move RCU grace-period initialization into a kthread Josh Triplett
2012-09-06 13:32   ` Peter Zijlstra
2012-09-06 17:00     ` Paul E. McKenney

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=1346350718-30937-12-git-send-email-paulmck@linux.vnet.ibm.com \
    --to=paulmck@linux.vnet.ibm.com \
    --cc=Valdis.Kletnieks@vt.edu \
    --cc=akpm@linux-foundation.org \
    --cc=darren@dvhart.com \
    --cc=dhowells@redhat.com \
    --cc=dipankar@in.ibm.com \
    --cc=eric.dumazet@gmail.com \
    --cc=fweisbec@gmail.com \
    --cc=josh@joshtriplett.org \
    --cc=laijs@cn.fujitsu.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mathieu.desnoyers@polymtl.ca \
    --cc=mingo@elte.hu \
    --cc=niv@us.ibm.com \
    --cc=patches@linaro.org \
    --cc=peterz@infradead.org \
    --cc=rostedt@goodmis.org \
    --cc=sbw@mit.edu \
    --cc=tglx@linutronix.de \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).