public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
From: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
To: linux-kernel@vger.kernel.org
Cc: mingo@elte.hu, josh@freedesktop.org, dipankar@in.ibm.com,
	akpm@linux-foundation.org
Subject: [PATCH] Force rcutorture tasks to spread over CPUs
Date: Tue, 12 Jun 2007 21:28:04 -0700	[thread overview]
Message-ID: <20070613042804.GA6132@linux.vnet.ibm.com> (raw)

Of late, the scheduler seems to have decided to make things too easy for
RCU -- on some configurations, all of the rcutorture tasks end up on the
same CPU, which doesn't do a very good job of torturing RCU.  This patch
helps the scheduler spread these tasks out by forcing a 20-millisecond
burst of CPU-bound execution on each of rcutorture's tasks, which seems
to work reasonably well in practice.

My challenge for those working on the scheduler is to make this patch
unnecessary.  ;-)

Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
---

 rcutorture.c |   20 ++++++++++++++++++++
 1 file changed, 20 insertions(+)

diff -urpNa -X dontdiff linux-2.6.21.4-rt13/kernel/rcutorture.c linux-2.6.21.4-rt13-rcutorturespread/kernel/rcutorture.c
--- linux-2.6.21.4-rt13/kernel/rcutorture.c	2007-06-12 09:19:02.000000000 -0700
+++ linux-2.6.21.4-rt13-rcutorturespread/kernel/rcutorture.c	2007-06-12 21:05:05.000000000 -0700
@@ -102,6 +102,8 @@ struct rcu_torture {
 };
 
 static int fullstop = 0;	/* stop generating callbacks at test end. */
+static int startwriters;	/* force load-balancing of writers. */
+static int startreaders;	/* force load-balancing of readers. */
 static LIST_HEAD(rcu_torture_freelist);
 static struct rcu_torture *rcu_torture_current = NULL;
 static long rcu_torture_current_version = 0;
@@ -525,6 +527,8 @@ rcu_torture_writer(void *arg)
 	static DEFINE_RCU_RANDOM(rand);
 
 	VERBOSE_PRINTK_STRING("rcu_torture_writer task started");
+	while (!startwriters)
+		barrier();	/* Force scheduler to spread over CPUs. */
 	set_user_nice(current, 19);
 	current->flags |= PF_NOFREEZE;
 
@@ -565,6 +569,8 @@ rcu_torture_fakewriter(void *arg)
 	DEFINE_RCU_RANDOM(rand);
 
 	VERBOSE_PRINTK_STRING("rcu_torture_fakewriter task started");
+	while (!startwriters)
+		barrier();	/* Force scheduler to spread over CPUs. */
 	set_user_nice(current, 19);
 	current->flags |= PF_NOFREEZE;
 
@@ -596,6 +602,8 @@ rcu_torture_reader(void *arg)
 	int pipe_count;
 
 	VERBOSE_PRINTK_STRING("rcu_torture_reader task started");
+	while (!startreaders)
+		barrier();	/* Force scheduler to spread over CPUs. */
 	set_user_nice(current, 19);
 	current->flags |= PF_NOFREEZE;
 
@@ -929,6 +937,8 @@ rcu_torture_init(void)
 
 	/* Start up the kthreads. */
 
+	startwriters = 0;
+	barrier();
 	VERBOSE_PRINTK_STRING("Creating rcu_torture_writer task");
 	writer_task = kthread_run(rcu_torture_writer, NULL,
 				  "rcu_torture_writer");
@@ -956,6 +966,12 @@ rcu_torture_init(void)
 			goto unwind;
 		}
 	}
+	barrier();
+	startwriters = 1;
+	schedule_timeout_interruptible(round_jiffies_relative(HZ/50));
+
+	startreaders = 0;
+	barrier();
 	reader_tasks = kzalloc(nrealreaders * sizeof(reader_tasks[0]),
 			       GFP_KERNEL);
 	if (reader_tasks == NULL) {
@@ -974,6 +990,10 @@ rcu_torture_init(void)
 			goto unwind;
 		}
 	}
+	schedule_timeout_interruptible(round_jiffies_relative(HZ/50));
+	barrier();
+	startreaders = 1;
+
 	if (stat_interval > 0) {
 		VERBOSE_PRINTK_STRING("Creating rcu_torture_stats task");
 		stats_task = kthread_run(rcu_torture_stats, NULL,

             reply	other threads:[~2007-06-13  4:28 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2007-06-13  4:28 Paul E. McKenney [this message]
2007-06-13 16:49 ` [PATCH] Force rcutorture tasks to spread over CPUs Josh Triplett
2007-06-26 23:13 ` Andrew Morton
2007-06-27  6:30   ` Ingo Molnar
2007-06-27 16:30     ` 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=20070613042804.GA6132@linux.vnet.ibm.com \
    --to=paulmck@linux.vnet.ibm.com \
    --cc=akpm@linux-foundation.org \
    --cc=dipankar@in.ibm.com \
    --cc=josh@freedesktop.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mingo@elte.hu \
    /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