[PATCH tip/core/rcu 0/24] Torture-test changes for 4.7

[PATCH tip/core/rcu 0/24] Torture-test changes for 4.7

[Paul E. McKenney]

Hello!

This series contains torture-test changes for RCU, most notably a new
test for performance of RCU grace periods named "rcuperf":

1.	Update scripting to accommodate rcuperf.

2.	Allow for rcupdate.rcu_normal in rcu_torture_writer() decisions.

3.	Expedited-GP batch progress access to torturing for progress
	statistics.

4.	Add RCU grace-period performance tests, AKA "rcuperf".

5.	Documentation for rcuperf kernel parameters.

6.	Bind rcuperf reader/writer kthreads to CPUs.

7.	Set rcuperf writer kthreads to real-time priority.

8.	Print measure of batching efficiency.

9.	Make rcuperf collect expedited event-trace data.

10.	Make scripts analyze rcuperf trace data, if present.

11.	Add rcuperf holdoff boot parameter to reduce interference.

12.	Avoid RCU CPU stall warning and RT throttling.

13.	Add largish-system rcuperf scenario.

14.	Do not wake up shutdown wait queue if "shutdown" is false, courtesy
	of Artem Savkov.

15.	Remove redundant initialization to zero.

16.	Consider FROZEN hotplug notifier transitions, courtesy of
	Anna-Maria Gleixner.

17.	Clarify refusal to run more than one torture test.

18.	Kill qemu, not parent process.

19.	Convert test duration to seconds early.

20.	Add OS-jitter capability.

21.	Don't rebuild identical kernel.

22.	Dump trace buffer upon shutdown.

23.	Add irqs-disabled test for call_rcu().

24.	Add boot-time adjustment of leaf fanout.

								Thanx, Paul

------------------------------------------------------------------------

 Documentation/kernel-parameters.txt                                  |   38 
 include/linux/rcutiny.h                                              |   16 
 include/linux/rcutree.h                                              |    2 
 kernel/rcu/Makefile                                                  |    1 
 kernel/rcu/rcuperf.c                                                 |  661 +++++++++-
 kernel/rcu/rcutorture.c                                              |   12 
 kernel/rcu/tree.c                                                    |   22 
 kernel/torture.c                                                     |    4 
 lib/Kconfig.debug                                                    |   33 
 tools/testing/selftests/rcutorture/bin/jitter.sh                     |   83 +
 tools/testing/selftests/rcutorture/bin/kvm-recheck-rcuperf-ftrace.sh |  121 +
 tools/testing/selftests/rcutorture/bin/kvm-recheck-rcuperf.sh        |  116 +
 tools/testing/selftests/rcutorture/bin/kvm-recheck.sh                |    5 
 tools/testing/selftests/rcutorture/bin/kvm-test-1-run.sh             |   61 
 tools/testing/selftests/rcutorture/bin/kvm.sh                        |   24 
 tools/testing/selftests/rcutorture/configs/rcu/TREE04                |    2 
 tools/testing/selftests/rcutorture/configs/rcu/TREE04.boot           |    2 
 tools/testing/selftests/rcutorture/configs/rcuperf/CFLIST            |    1 
 tools/testing/selftests/rcutorture/configs/rcuperf/CFcommon          |    2 
 tools/testing/selftests/rcutorture/configs/rcuperf/TREE              |   20 
 tools/testing/selftests/rcutorture/configs/rcuperf/TREE54            |   23 
 tools/testing/selftests/rcutorture/configs/rcuperf/ver_functions.sh  |   52 
 22 files changed, 1260 insertions(+), 41 deletions(-)

[PATCH tip/core/rcu 01/24] rcutorture: Update scripting to accommodate rcuperf

[Paul E. McKenney]

This commit adds the scripting changes to add support for the shiny
new rcuperf kernel module.

Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
---
 .../rcutorture/bin/kvm-recheck-rcuperf.sh          | 82 ++++++++++++++++++++++
 .../selftests/rcutorture/bin/kvm-recheck.sh        |  5 +-
 tools/testing/selftests/rcutorture/bin/kvm.sh      |  2 +-
 .../selftests/rcutorture/configs/rcuperf/CFLIST    |  1 +
 .../selftests/rcutorture/configs/rcuperf/CFcommon  |  2 +
 .../selftests/rcutorture/configs/rcuperf/TREE      | 19 +++++
 .../rcutorture/configs/rcuperf/ver_functions.sh    | 52 ++++++++++++++
 7 files changed, 161 insertions(+), 2 deletions(-)
 create mode 100755 tools/testing/selftests/rcutorture/bin/kvm-recheck-rcuperf.sh
 create mode 100644 tools/testing/selftests/rcutorture/configs/rcuperf/CFLIST
 create mode 100644 tools/testing/selftests/rcutorture/configs/rcuperf/CFcommon
 create mode 100644 tools/testing/selftests/rcutorture/configs/rcuperf/TREE
 create mode 100644 tools/testing/selftests/rcutorture/configs/rcuperf/ver_functions.sh

diff --git a/tools/testing/selftests/rcutorture/bin/kvm-recheck-rcuperf.sh b/tools/testing/selftests/rcutorture/bin/kvm-recheck-rcuperf.sh
new file mode 100755
index 000000000000..e5b28174fda0
--- /dev/null
+++ b/tools/testing/selftests/rcutorture/bin/kvm-recheck-rcuperf.sh
@@ -0,0 +1,82 @@
+#!/bin/bash
+#
+# Analyze a given results directory for rcuperf performance measurements.
+#
+# Usage: kvm-recheck-rcuperf.sh resdir
+#
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 2 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program; if not, you can access it online at
+# http://www.gnu.org/licenses/gpl-2.0.html.
+#
+# Copyright (C) IBM Corporation, 2016
+#
+# Authors: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
+
+i="$1"
+if test -d $i
+then
+	:
+else
+	echo Unreadable results directory: $i
+	exit 1
+fi
+. tools/testing/selftests/rcutorture/bin/functions.sh
+
+configfile=`echo $i | sed -e 's/^.*\///'`
+
+grep -e '-perf:.*writer-duration' $i/console.log | sed -e 's/^\[[^]]*]//' |
+awk '
+{
+	gptimes[++n] = $5 / 1000.;
+	sum += $5 / 1000.;
+}
+
+END {
+	if (NR <= 0) {
+		print "No rcuperf records found???"
+		exit;
+	}
+	asort(gptimes);
+	pct50 = int(NR * 50 / 100);
+	if (pct50 < 1)
+		pct50 = 1;
+	pct90 = int(NR * 90 / 100);
+	if (pct90 < 1)
+		pct90 = 1;
+	pct99 = int(NR * 99 / 100);
+	if (pct99 < 1)
+		pct99 = 1;
+	div = 10 ** int(log(gptimes[pct90]) / log(10) + .5) / 100;
+	print "Histogram bucket size: " div;
+	last = gptimes[1] - 10;
+	count = 0;
+	for (i = 1; i <= NR; i++) {
+		current = div * int(gptimes[i] / div);
+		if (last == current) {
+			count++;
+		} else {
+			if (count > 0)
+				print last, count;
+			count = 1;
+			last = current;
+		}
+	}
+	if (count > 0)
+		print last, count;
+	print "Average grace-period duration: " sum / NR " microseconds";
+	print "Minimum grace-period duration: " gptimes[1];
+	print "50th percentile grace-period duration: " gptimes[pct50];
+	print "90th percentile grace-period duration: " gptimes[pct90];
+	print "99th percentile grace-period duration: " gptimes[pct99];
+	print "Maximum grace-period duration: " gptimes[NR];
+}'
diff --git a/tools/testing/selftests/rcutorture/bin/kvm-recheck.sh b/tools/testing/selftests/rcutorture/bin/kvm-recheck.sh
index d86bdd6b6cc2..f659346d3358 100755
--- a/tools/testing/selftests/rcutorture/bin/kvm-recheck.sh
+++ b/tools/testing/selftests/rcutorture/bin/kvm-recheck.sh
@@ -48,7 +48,10 @@ do
 				cat $i/Make.oldconfig.err
 			fi
 			parse-build.sh $i/Make.out $configfile
-			parse-torture.sh $i/console.log $configfile
+			if test "$TORTURE_SUITE" != rcuperf
+			then
+				parse-torture.sh $i/console.log $configfile
+			fi
 			parse-console.sh $i/console.log $configfile
 			if test -r $i/Warnings
 			then
diff --git a/tools/testing/selftests/rcutorture/bin/kvm.sh b/tools/testing/selftests/rcutorture/bin/kvm.sh
index 4a431767f77a..c33cb582b3dc 100755
--- a/tools/testing/selftests/rcutorture/bin/kvm.sh
+++ b/tools/testing/selftests/rcutorture/bin/kvm.sh
@@ -156,7 +156,7 @@ do
 		shift
 		;;
 	--torture)
-		checkarg --torture "(suite name)" "$#" "$2" '^\(lock\|rcu\)$' '^--'
+		checkarg --torture "(suite name)" "$#" "$2" '^\(lock\|rcu\|rcuperf\)$' '^--'
 		TORTURE_SUITE=$2
 		shift
 		;;
diff --git a/tools/testing/selftests/rcutorture/configs/rcuperf/CFLIST b/tools/testing/selftests/rcutorture/configs/rcuperf/CFLIST
new file mode 100644
index 000000000000..c9f56cf20775
--- /dev/null
+++ b/tools/testing/selftests/rcutorture/configs/rcuperf/CFLIST
@@ -0,0 +1 @@
+TREE
diff --git a/tools/testing/selftests/rcutorture/configs/rcuperf/CFcommon b/tools/testing/selftests/rcutorture/configs/rcuperf/CFcommon
new file mode 100644
index 000000000000..a09816b8c0f3
--- /dev/null
+++ b/tools/testing/selftests/rcutorture/configs/rcuperf/CFcommon
@@ -0,0 +1,2 @@
+CONFIG_RCU_PERF_TEST=y
+CONFIG_PRINTK_TIME=y
diff --git a/tools/testing/selftests/rcutorture/configs/rcuperf/TREE b/tools/testing/selftests/rcutorture/configs/rcuperf/TREE
new file mode 100644
index 000000000000..614e107f6db5
--- /dev/null
+++ b/tools/testing/selftests/rcutorture/configs/rcuperf/TREE
@@ -0,0 +1,19 @@
+CONFIG_SMP=y
+CONFIG_PREEMPT_NONE=n
+CONFIG_PREEMPT_VOLUNTARY=n
+CONFIG_PREEMPT=y
+#CHECK#CONFIG_PREEMPT_RCU=y
+CONFIG_HZ_PERIODIC=n
+CONFIG_NO_HZ_IDLE=y
+CONFIG_NO_HZ_FULL=n
+CONFIG_RCU_FAST_NO_HZ=n
+CONFIG_RCU_TRACE=n
+CONFIG_HOTPLUG_CPU=n
+CONFIG_SUSPEND=n
+CONFIG_HIBERNATION=n
+CONFIG_RCU_NOCB_CPU=n
+CONFIG_DEBUG_LOCK_ALLOC=n
+CONFIG_PROVE_LOCKING=n
+CONFIG_RCU_BOOST=n
+CONFIG_DEBUG_OBJECTS_RCU_HEAD=n
+CONFIG_RCU_EXPERT=y
diff --git a/tools/testing/selftests/rcutorture/configs/rcuperf/ver_functions.sh b/tools/testing/selftests/rcutorture/configs/rcuperf/ver_functions.sh
new file mode 100644
index 000000000000..34f2a1b35ee5
--- /dev/null
+++ b/tools/testing/selftests/rcutorture/configs/rcuperf/ver_functions.sh
@@ -0,0 +1,52 @@
+#!/bin/bash
+#
+# Torture-suite-dependent shell functions for the rest of the scripts.
+#
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 2 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program; if not, you can access it online at
+# http://www.gnu.org/licenses/gpl-2.0.html.
+#
+# Copyright (C) IBM Corporation, 2015
+#
+# Authors: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
+
+# rcuperf_param_nreaders bootparam-string
+#
+# Adds nreaders rcuperf module parameter if not already specified.
+rcuperf_param_nreaders () {
+	if ! echo "$1" | grep -q "rcuperf.nreaders"
+	then
+		echo rcuperf.nreaders=-1
+	fi
+}
+
+# rcuperf_param_nwriters bootparam-string
+#
+# Adds nwriters rcuperf module parameter if not already specified.
+rcuperf_param_nwriters () {
+	if ! echo "$1" | grep -q "rcuperf.nwriters"
+	then
+		echo rcuperf.nwriters=-1
+	fi
+}
+
+# per_version_boot_params bootparam-string config-file seconds
+#
+# Adds per-version torture-module parameters to kernels supporting them.
+per_version_boot_params () {
+	echo $1 `rcuperf_param_nreaders "$1"` \
+		`rcuperf_param_nwriters "$1"` \
+		rcuperf.perf_runnable=1 \
+		rcuperf.shutdown=1 \
+		rcuperf.verbose=1
+}
-- 
2.5.2

[PATCH tip/core/rcu 02/24] rcutorture: Allow for rcupdate.rcu_normal

[Paul E. McKenney]

Currently, rcu_torture_writer() checks only for rcu_gp_is_expedited()
when deciding whether or not to do dynamic control of RCU expediting.
This means that if rcupdate.rcu_normal is specified, rcu_torture_writer()
will attempt to dynamically control RCU expediting, but will nonetheless
only test normal RCU grace periods.  This commit therefore adds a check
for !rcu_gp_is_normal(), and prints a message and desists from testing
dynamic control of RCU expediting when doing so is futile.

Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
---
 kernel/rcu/rcutorture.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/kernel/rcu/rcutorture.c b/kernel/rcu/rcutorture.c
index 463867c43221..9234e75b106a 100644
--- a/kernel/rcu/rcutorture.c
+++ b/kernel/rcu/rcutorture.c
@@ -916,7 +916,7 @@ rcu_torture_fqs(void *arg)
 static int
 rcu_torture_writer(void *arg)
 {
-	bool can_expedite = !rcu_gp_is_expedited();
+	bool can_expedite = !rcu_gp_is_expedited() && !rcu_gp_is_normal();
 	int expediting = 0;
 	unsigned long gp_snap;
 	bool gp_cond1 = gp_cond, gp_exp1 = gp_exp, gp_normal1 = gp_normal;
@@ -932,7 +932,7 @@ rcu_torture_writer(void *arg)
 	VERBOSE_TOROUT_STRING("rcu_torture_writer task started");
 	if (!can_expedite) {
 		pr_alert("%s" TORTURE_FLAG
-			 " Grace periods expedited from boot/sysfs for %s,\n",
+			 " GP expediting controlled from boot/sysfs for %s,\n",
 			 torture_type, cur_ops->name);
 		pr_alert("%s" TORTURE_FLAG
 			 " Disabled dynamic grace-period expediting.\n",
-- 
2.5.2

[PATCH tip/core/rcu 03/24] rcutorture: Expedited-GP batch progress access to torturing

[Paul E. McKenney]

This commit provides rcu_exp_batches_completed() and
rcu_exp_batches_completed_sched() functions to allow torture-test modules
to check how many expedited grace period batches have completed.
These are analogous to the existing rcu_batches_completed(),
rcu_batches_completed_bh(), and rcu_batches_completed_sched() functions.

Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
---
 include/linux/rcutiny.h | 16 ++++++++++++++++
 include/linux/rcutree.h |  2 ++
 kernel/rcu/tree.c       | 22 ++++++++++++++++++++++
 3 files changed, 40 insertions(+)

diff --git a/include/linux/rcutiny.h b/include/linux/rcutiny.h
index 64809aea661c..93aea75029fb 100644
--- a/include/linux/rcutiny.h
+++ b/include/linux/rcutiny.h
@@ -149,6 +149,22 @@ static inline unsigned long rcu_batches_completed_sched(void)
 	return 0;
 }
 
+/*
+ * Return the number of expedited grace periods completed.
+ */
+static inline unsigned long rcu_exp_batches_completed(void)
+{
+	return 0;
+}
+
+/*
+ * Return the number of expedited sched grace periods completed.
+ */
+static inline unsigned long rcu_exp_batches_completed_sched(void)
+{
+	return 0;
+}
+
 static inline void rcu_force_quiescent_state(void)
 {
 }
diff --git a/include/linux/rcutree.h b/include/linux/rcutree.h
index ad1eda9fa4da..5043cb823fb2 100644
--- a/include/linux/rcutree.h
+++ b/include/linux/rcutree.h
@@ -87,6 +87,8 @@ unsigned long rcu_batches_started_sched(void);
 unsigned long rcu_batches_completed(void);
 unsigned long rcu_batches_completed_bh(void);
 unsigned long rcu_batches_completed_sched(void);
+unsigned long rcu_exp_batches_completed(void);
+unsigned long rcu_exp_batches_completed_sched(void);
 void show_rcu_gp_kthreads(void);
 
 void rcu_force_quiescent_state(void);
diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c
index 531a328076bd..88df64087dfe 100644
--- a/kernel/rcu/tree.c
+++ b/kernel/rcu/tree.c
@@ -460,6 +460,28 @@ unsigned long rcu_batches_completed_bh(void)
 EXPORT_SYMBOL_GPL(rcu_batches_completed_bh);
 
 /*
+ * Return the number of RCU expedited batches completed thus far for
+ * debug & stats.  Odd numbers mean that a batch is in progress, even
+ * numbers mean idle.  The value returned will thus be roughly double
+ * the cumulative batches since boot.
+ */
+unsigned long rcu_exp_batches_completed(void)
+{
+	return rcu_state_p->expedited_sequence;
+}
+EXPORT_SYMBOL_GPL(rcu_exp_batches_completed);
+
+/*
+ * Return the number of RCU-sched expedited batches completed thus far
+ * for debug & stats.  Similar to rcu_exp_batches_completed().
+ */
+unsigned long rcu_exp_batches_completed_sched(void)
+{
+	return rcu_sched_state.expedited_sequence;
+}
+EXPORT_SYMBOL_GPL(rcu_exp_batches_completed_sched);
+
+/*
  * Force a quiescent state.
  */
 void rcu_force_quiescent_state(void)
-- 
2.5.2

[PATCH tip/core/rcu 04/24] rcutorture: Add RCU grace-period performance tests

[Paul E. McKenney]

This commit adds a new rcuperf module that carries out simple performance
tests of RCU grace periods.

Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
---
 kernel/rcu/Makefile  |   1 +
 kernel/rcu/rcuperf.c | 637 +++++++++++++++++++++++++++++++++++++++++++++++++++
 lib/Kconfig.debug    |  33 +++
 3 files changed, 671 insertions(+)
 create mode 100644 kernel/rcu/rcuperf.c

diff --git a/kernel/rcu/Makefile b/kernel/rcu/Makefile
index 032b2c015beb..18dfc485225c 100644
--- a/kernel/rcu/Makefile
+++ b/kernel/rcu/Makefile
@@ -5,6 +5,7 @@ KCOV_INSTRUMENT := n
 obj-y += update.o sync.o
 obj-$(CONFIG_SRCU) += srcu.o
 obj-$(CONFIG_RCU_TORTURE_TEST) += rcutorture.o
+obj-$(CONFIG_RCU_PERF_TEST) += rcuperf.o
 obj-$(CONFIG_TREE_RCU) += tree.o
 obj-$(CONFIG_PREEMPT_RCU) += tree.o
 obj-$(CONFIG_TREE_RCU_TRACE) += tree_trace.o
diff --git a/kernel/rcu/rcuperf.c b/kernel/rcu/rcuperf.c
new file mode 100644
index 000000000000..9d54a57bee7d
--- /dev/null
+++ b/kernel/rcu/rcuperf.c
@@ -0,0 +1,637 @@
+/*
+ * Read-Copy Update module-based performance-test facility
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, you can access it online at
+ * http://www.gnu.org/licenses/gpl-2.0.html.
+ *
+ * Copyright (C) IBM Corporation, 2015
+ *
+ * Authors: Paul E. McKenney <paulmck@us.ibm.com>
+ */
+#include <linux/types.h>
+#include <linux/kernel.h>
+#include <linux/init.h>
+#include <linux/module.h>
+#include <linux/kthread.h>
+#include <linux/err.h>
+#include <linux/spinlock.h>
+#include <linux/smp.h>
+#include <linux/rcupdate.h>
+#include <linux/interrupt.h>
+#include <linux/sched.h>
+#include <linux/atomic.h>
+#include <linux/bitops.h>
+#include <linux/completion.h>
+#include <linux/moduleparam.h>
+#include <linux/percpu.h>
+#include <linux/notifier.h>
+#include <linux/reboot.h>
+#include <linux/freezer.h>
+#include <linux/cpu.h>
+#include <linux/delay.h>
+#include <linux/stat.h>
+#include <linux/srcu.h>
+#include <linux/slab.h>
+#include <asm/byteorder.h>
+#include <linux/torture.h>
+#include <linux/vmalloc.h>
+
+MODULE_LICENSE("GPL");
+MODULE_AUTHOR("Paul E. McKenney <paulmck@linux.vnet.ibm.com>");
+
+#define PERF_FLAG "-perf:"
+#define PERFOUT_STRING(s) \
+	pr_alert("%s" PERF_FLAG s "\n", perf_type)
+#define VERBOSE_PERFOUT_STRING(s) \
+	do { if (verbose) pr_alert("%s" PERF_FLAG " %s\n", perf_type, s); } while (0)
+#define VERBOSE_PERFOUT_ERRSTRING(s) \
+	do { if (verbose) pr_alert("%s" PERF_FLAG "!!! %s\n", perf_type, s); } while (0)
+
+torture_param(bool, gp_exp, true, "Use expedited GP wait primitives");
+torture_param(int, nreaders, -1, "Number of RCU reader threads");
+torture_param(int, nwriters, -1, "Number of RCU updater threads");
+torture_param(bool, shutdown, false, "Shutdown at end of performance tests.");
+torture_param(bool, verbose, true, "Enable verbose debugging printk()s");
+
+static char *perf_type = "rcu";
+module_param(perf_type, charp, 0444);
+MODULE_PARM_DESC(perf_type, "Type of RCU to performance-test (rcu, rcu_bh, ...)");
+
+static int nrealreaders;
+static int nrealwriters;
+static struct task_struct **writer_tasks;
+static struct task_struct **reader_tasks;
+static struct task_struct *shutdown_task;
+
+static u64 **writer_durations;
+static int *writer_n_durations;
+static atomic_t n_rcu_perf_reader_started;
+static atomic_t n_rcu_perf_writer_started;
+static atomic_t n_rcu_perf_writer_finished;
+static wait_queue_head_t shutdown_wq;
+static u64 t_rcu_perf_writer_started;
+static u64 t_rcu_perf_writer_finished;
+static unsigned long b_rcu_perf_writer_started;
+static unsigned long b_rcu_perf_writer_finished;
+
+static int rcu_perf_writer_state;
+#define RTWS_INIT		0
+#define RTWS_EXP_SYNC		1
+#define RTWS_SYNC		2
+#define RTWS_IDLE		2
+#define RTWS_STOPPING		3
+
+#define MAX_MEAS 10000
+#define MIN_MEAS 100
+
+#if defined(MODULE) || defined(CONFIG_RCU_PERF_TEST_RUNNABLE)
+#define RCUPERF_RUNNABLE_INIT 1
+#else
+#define RCUPERF_RUNNABLE_INIT 0
+#endif
+static int perf_runnable = RCUPERF_RUNNABLE_INIT;
+module_param(perf_runnable, int, 0444);
+MODULE_PARM_DESC(perf_runnable, "Start rcuperf at boot");
+
+/*
+ * Operations vector for selecting different types of tests.
+ */
+
+struct rcu_perf_ops {
+	int ptype;
+	void (*init)(void);
+	void (*cleanup)(void);
+	int (*readlock)(void);
+	void (*readunlock)(int idx);
+	unsigned long (*started)(void);
+	unsigned long (*completed)(void);
+	unsigned long (*exp_completed)(void);
+	void (*sync)(void);
+	void (*exp_sync)(void);
+	const char *name;
+};
+
+static struct rcu_perf_ops *cur_ops;
+
+/*
+ * Definitions for rcu perf testing.
+ */
+
+static int rcu_perf_read_lock(void) __acquires(RCU)
+{
+	rcu_read_lock();
+	return 0;
+}
+
+static void rcu_perf_read_unlock(int idx) __releases(RCU)
+{
+	rcu_read_unlock();
+}
+
+static unsigned long __maybe_unused rcu_no_completed(void)
+{
+	return 0;
+}
+
+static void rcu_sync_perf_init(void)
+{
+}
+
+static struct rcu_perf_ops rcu_ops = {
+	.ptype		= RCU_FLAVOR,
+	.init		= rcu_sync_perf_init,
+	.readlock	= rcu_perf_read_lock,
+	.readunlock	= rcu_perf_read_unlock,
+	.started	= rcu_batches_started,
+	.completed	= rcu_batches_completed,
+	.exp_completed	= rcu_exp_batches_completed,
+	.sync		= synchronize_rcu,
+	.exp_sync	= synchronize_rcu_expedited,
+	.name		= "rcu"
+};
+
+/*
+ * Definitions for rcu_bh perf testing.
+ */
+
+static int rcu_bh_perf_read_lock(void) __acquires(RCU_BH)
+{
+	rcu_read_lock_bh();
+	return 0;
+}
+
+static void rcu_bh_perf_read_unlock(int idx) __releases(RCU_BH)
+{
+	rcu_read_unlock_bh();
+}
+
+static struct rcu_perf_ops rcu_bh_ops = {
+	.ptype		= RCU_BH_FLAVOR,
+	.init		= rcu_sync_perf_init,
+	.readlock	= rcu_bh_perf_read_lock,
+	.readunlock	= rcu_bh_perf_read_unlock,
+	.started	= rcu_batches_started_bh,
+	.completed	= rcu_batches_completed_bh,
+	.exp_completed	= rcu_exp_batches_completed_sched,
+	.sync		= synchronize_rcu_bh,
+	.exp_sync	= synchronize_rcu_bh_expedited,
+	.name		= "rcu_bh"
+};
+
+/*
+ * Definitions for srcu perf testing.
+ */
+
+DEFINE_STATIC_SRCU(srcu_ctl_perf);
+static struct srcu_struct *srcu_ctlp = &srcu_ctl_perf;
+
+static int srcu_perf_read_lock(void) __acquires(srcu_ctlp)
+{
+	return srcu_read_lock(srcu_ctlp);
+}
+
+static void srcu_perf_read_unlock(int idx) __releases(srcu_ctlp)
+{
+	srcu_read_unlock(srcu_ctlp, idx);
+}
+
+static unsigned long srcu_perf_completed(void)
+{
+	return srcu_batches_completed(srcu_ctlp);
+}
+
+static void srcu_perf_synchronize(void)
+{
+	synchronize_srcu(srcu_ctlp);
+}
+
+static void srcu_perf_synchronize_expedited(void)
+{
+	synchronize_srcu_expedited(srcu_ctlp);
+}
+
+static struct rcu_perf_ops srcu_ops = {
+	.ptype		= SRCU_FLAVOR,
+	.init		= rcu_sync_perf_init,
+	.readlock	= srcu_perf_read_lock,
+	.readunlock	= srcu_perf_read_unlock,
+	.started	= NULL,
+	.completed	= srcu_perf_completed,
+	.exp_completed	= srcu_perf_completed,
+	.sync		= srcu_perf_synchronize,
+	.exp_sync	= srcu_perf_synchronize_expedited,
+	.name		= "srcu"
+};
+
+/*
+ * Definitions for sched perf testing.
+ */
+
+static int sched_perf_read_lock(void)
+{
+	preempt_disable();
+	return 0;
+}
+
+static void sched_perf_read_unlock(int idx)
+{
+	preempt_enable();
+}
+
+static struct rcu_perf_ops sched_ops = {
+	.ptype		= RCU_SCHED_FLAVOR,
+	.init		= rcu_sync_perf_init,
+	.readlock	= sched_perf_read_lock,
+	.readunlock	= sched_perf_read_unlock,
+	.started	= rcu_batches_started_sched,
+	.completed	= rcu_batches_completed_sched,
+	.exp_completed	= rcu_exp_batches_completed_sched,
+	.sync		= synchronize_sched,
+	.exp_sync	= synchronize_sched_expedited,
+	.name		= "sched"
+};
+
+#ifdef CONFIG_TASKS_RCU
+
+/*
+ * Definitions for RCU-tasks perf testing.
+ */
+
+static int tasks_perf_read_lock(void)
+{
+	return 0;
+}
+
+static void tasks_perf_read_unlock(int idx)
+{
+}
+
+static struct rcu_perf_ops tasks_ops = {
+	.ptype		= RCU_TASKS_FLAVOR,
+	.init		= rcu_sync_perf_init,
+	.readlock	= tasks_perf_read_lock,
+	.readunlock	= tasks_perf_read_unlock,
+	.started	= rcu_no_completed,
+	.completed	= rcu_no_completed,
+	.sync		= synchronize_rcu_tasks,
+	.exp_sync	= synchronize_rcu_tasks,
+	.name		= "tasks"
+};
+
+#define RCUPERF_TASKS_OPS &tasks_ops,
+
+static bool __maybe_unused torturing_tasks(void)
+{
+	return cur_ops == &tasks_ops;
+}
+
+#else /* #ifdef CONFIG_TASKS_RCU */
+
+#define RCUPERF_TASKS_OPS
+
+static bool __maybe_unused torturing_tasks(void)
+{
+	return false;
+}
+
+#endif /* #else #ifdef CONFIG_TASKS_RCU */
+
+/*
+ * If performance tests complete, wait for shutdown to commence.
+ */
+static void rcu_perf_wait_shutdown(void)
+{
+	cond_resched_rcu_qs();
+	if (atomic_read(&n_rcu_perf_writer_finished) < nrealwriters)
+		return;
+	while (!torture_must_stop())
+		schedule_timeout_uninterruptible(1);
+}
+
+/*
+ * RCU perf reader kthread.  Repeatedly does empty RCU read-side
+ * critical section, minimizing update-side interference.
+ */
+static int
+rcu_perf_reader(void *arg)
+{
+	unsigned long flags;
+	int idx;
+
+	VERBOSE_PERFOUT_STRING("rcu_perf_reader task started");
+	set_user_nice(current, MAX_NICE);
+	atomic_inc(&n_rcu_perf_reader_started);
+
+	do {
+		local_irq_save(flags);
+		idx = cur_ops->readlock();
+		cur_ops->readunlock(idx);
+		local_irq_restore(flags);
+		rcu_perf_wait_shutdown();
+	} while (!torture_must_stop());
+	torture_kthread_stopping("rcu_perf_reader");
+	return 0;
+}
+
+/*
+ * RCU perf writer kthread.  Repeatedly does a grace period.
+ */
+static int
+rcu_perf_writer(void *arg)
+{
+	int i = 0;
+	int i_max;
+	long me = (long)arg;
+	bool started = false, done = false, alldone = false;
+	u64 t;
+	u64 *wdp;
+	u64 *wdpp = writer_durations[me];
+
+	VERBOSE_PERFOUT_STRING("rcu_perf_writer task started");
+	WARN_ON(rcu_gp_is_expedited() && !rcu_gp_is_normal() && !gp_exp);
+	WARN_ON(rcu_gp_is_normal() && gp_exp);
+	WARN_ON(!wdpp);
+	t = ktime_get_mono_fast_ns();
+	if (atomic_inc_return(&n_rcu_perf_writer_started) >= nrealwriters) {
+		t_rcu_perf_writer_started = t;
+		if (gp_exp) {
+			b_rcu_perf_writer_started =
+				cur_ops->exp_completed() / 2;
+		} else {
+			b_rcu_perf_writer_started =
+				cur_ops->completed();
+		}
+	}
+
+	do {
+		wdp = &wdpp[i];
+		*wdp = ktime_get_mono_fast_ns();
+		if (gp_exp) {
+			rcu_perf_writer_state = RTWS_EXP_SYNC;
+			cur_ops->exp_sync();
+		} else {
+			rcu_perf_writer_state = RTWS_SYNC;
+			cur_ops->sync();
+		}
+		rcu_perf_writer_state = RTWS_IDLE;
+		t = ktime_get_mono_fast_ns();
+		*wdp = t - *wdp;
+		i_max = i;
+		if (!started &&
+		    atomic_read(&n_rcu_perf_writer_started) >= nrealwriters)
+			started = true;
+		if (!done && i >= MIN_MEAS) {
+			done = true;
+			pr_alert("%s" PERF_FLAG
+				 "rcu_perf_writer %ld has %d measurements\n",
+				 perf_type, me, MIN_MEAS);
+			if (atomic_inc_return(&n_rcu_perf_writer_finished) >=
+			    nrealwriters) {
+				PERFOUT_STRING("Test complete");
+				t_rcu_perf_writer_finished = t;
+				if (gp_exp) {
+					b_rcu_perf_writer_finished =
+						cur_ops->exp_completed() / 2;
+				} else {
+					b_rcu_perf_writer_finished =
+						cur_ops->completed();
+				}
+				smp_mb(); /* Assign before wake. */
+				wake_up(&shutdown_wq);
+			}
+		}
+		if (done && !alldone &&
+		    atomic_read(&n_rcu_perf_writer_finished) >= nrealwriters)
+			alldone = true;
+		if (started && !alldone && i < MAX_MEAS - 1)
+			i++;
+		rcu_perf_wait_shutdown();
+	} while (!torture_must_stop());
+	rcu_perf_writer_state = RTWS_STOPPING;
+	writer_n_durations[me] = i_max;
+	torture_kthread_stopping("rcu_perf_writer");
+	return 0;
+}
+
+static inline void
+rcu_perf_print_module_parms(struct rcu_perf_ops *cur_ops, const char *tag)
+{
+	pr_alert("%s" PERF_FLAG
+		 "--- %s: nreaders=%d nwriters=%d verbose=%d shutdown=%d\n",
+		 perf_type, tag, nrealreaders, nrealwriters, verbose, shutdown);
+}
+
+static void
+rcu_perf_cleanup(void)
+{
+	int i;
+	int j;
+	int ngps = 0;
+	u64 *wdp;
+	u64 *wdpp;
+
+	if (torture_cleanup_begin())
+		return;
+
+	if (reader_tasks) {
+		for (i = 0; i < nrealreaders; i++)
+			torture_stop_kthread(rcu_perf_reader,
+					     reader_tasks[i]);
+		kfree(reader_tasks);
+	}
+
+	if (writer_tasks) {
+		for (i = 0; i < nrealwriters; i++) {
+			torture_stop_kthread(rcu_perf_writer,
+					     writer_tasks[i]);
+			if (!writer_n_durations)
+				continue;
+			j = writer_n_durations[i];
+			pr_alert("%s%s writer %d gps: %d\n",
+				 perf_type, PERF_FLAG, i, j);
+			ngps += j;
+		}
+		pr_alert("%s%s start: %llu end: %llu duration: %llu gps: %d batches: %ld\n",
+			 perf_type, PERF_FLAG,
+			 t_rcu_perf_writer_started, t_rcu_perf_writer_finished,
+			 t_rcu_perf_writer_finished -
+			 t_rcu_perf_writer_started,
+			 ngps,
+			 b_rcu_perf_writer_finished -
+			 b_rcu_perf_writer_started);
+		for (i = 0; i < nrealwriters; i++) {
+			if (!writer_durations)
+				break;
+			if (!writer_n_durations)
+				continue;
+			wdpp = writer_durations[i];
+			if (!wdpp)
+				continue;
+			for (j = 0; j <= writer_n_durations[i]; j++) {
+				wdp = &wdpp[j];
+				pr_alert("%s%s %4d writer-duration: %5d %llu\n",
+					perf_type, PERF_FLAG,
+					i, j, *wdp);
+				if (j % 100 == 0)
+					schedule_timeout_uninterruptible(1);
+			}
+			kfree(writer_durations[i]);
+		}
+		kfree(writer_tasks);
+		kfree(writer_durations);
+		kfree(writer_n_durations);
+	}
+
+	/* Do flavor-specific cleanup operations.  */
+	if (cur_ops->cleanup != NULL)
+		cur_ops->cleanup();
+
+	torture_cleanup_end();
+}
+
+/*
+ * Return the number if non-negative.  If -1, the number of CPUs.
+ * If less than -1, that much less than the number of CPUs, but
+ * at least one.
+ */
+static int compute_real(int n)
+{
+	int nr;
+
+	if (n >= 0) {
+		nr = n;
+	} else {
+		nr = num_online_cpus() + 1 + n;
+		if (nr <= 0)
+			nr = 1;
+	}
+	return nr;
+}
+
+/*
+ * RCU perf shutdown kthread.  Just waits to be awakened, then shuts
+ * down system.
+ */
+static int
+rcu_perf_shutdown(void *arg)
+{
+	do {
+		wait_event(shutdown_wq,
+			   atomic_read(&n_rcu_perf_writer_finished) >=
+			   nrealwriters);
+	} while (atomic_read(&n_rcu_perf_writer_finished) < nrealwriters);
+	smp_mb(); /* Wake before output. */
+	rcu_perf_cleanup();
+	kernel_power_off();
+	return -EINVAL;
+}
+
+static int __init
+rcu_perf_init(void)
+{
+	long i;
+	int firsterr = 0;
+	static struct rcu_perf_ops *perf_ops[] = {
+		&rcu_ops, &rcu_bh_ops, &srcu_ops, &sched_ops,
+		RCUPERF_TASKS_OPS
+	};
+
+	if (!torture_init_begin(perf_type, verbose, &perf_runnable))
+		return -EBUSY;
+
+	/* Process args and tell the world that the perf'er is on the job. */
+	for (i = 0; i < ARRAY_SIZE(perf_ops); i++) {
+		cur_ops = perf_ops[i];
+		if (strcmp(perf_type, cur_ops->name) == 0)
+			break;
+	}
+	if (i == ARRAY_SIZE(perf_ops)) {
+		pr_alert("rcu-perf: invalid perf type: \"%s\"\n",
+			 perf_type);
+		pr_alert("rcu-perf types:");
+		for (i = 0; i < ARRAY_SIZE(perf_ops); i++)
+			pr_alert(" %s", perf_ops[i]->name);
+		pr_alert("\n");
+		firsterr = -EINVAL;
+		goto unwind;
+	}
+	if (cur_ops->init)
+		cur_ops->init();
+
+	nrealwriters = compute_real(nwriters);
+	nrealreaders = compute_real(nreaders);
+	atomic_set(&n_rcu_perf_reader_started, 0);
+	atomic_set(&n_rcu_perf_writer_started, 0);
+	atomic_set(&n_rcu_perf_writer_finished, 0);
+	rcu_perf_print_module_parms(cur_ops, "Start of test");
+
+	/* Start up the kthreads. */
+
+	if (shutdown) {
+		init_waitqueue_head(&shutdown_wq);
+		firsterr = torture_create_kthread(rcu_perf_shutdown, NULL,
+						  shutdown_task);
+		if (firsterr)
+			goto unwind;
+		schedule_timeout_uninterruptible(1);
+	}
+	reader_tasks = kcalloc(nrealreaders, sizeof(reader_tasks[0]),
+			       GFP_KERNEL);
+	if (reader_tasks == NULL) {
+		VERBOSE_PERFOUT_ERRSTRING("out of memory");
+		firsterr = -ENOMEM;
+		goto unwind;
+	}
+	for (i = 0; i < nrealreaders; i++) {
+		firsterr = torture_create_kthread(rcu_perf_reader, NULL,
+						  reader_tasks[i]);
+		if (firsterr)
+			goto unwind;
+	}
+	while (atomic_read(&n_rcu_perf_reader_started) < nrealreaders)
+		schedule_timeout_uninterruptible(1);
+	writer_tasks = kcalloc(nrealwriters, sizeof(reader_tasks[0]),
+			       GFP_KERNEL);
+	writer_durations = kcalloc(nrealwriters, sizeof(*writer_durations),
+				   GFP_KERNEL);
+	writer_n_durations =
+		kcalloc(nrealwriters, sizeof(*writer_n_durations),
+			GFP_KERNEL);
+	if (!writer_tasks || !writer_durations || !writer_n_durations) {
+		VERBOSE_PERFOUT_ERRSTRING("out of memory");
+		firsterr = -ENOMEM;
+		goto unwind;
+	}
+	for (i = 0; i < nrealwriters; i++) {
+		writer_durations[i] =
+			kcalloc(MAX_MEAS, sizeof(*writer_durations[i]),
+				GFP_KERNEL);
+		if (!writer_durations[i])
+			goto unwind;
+		firsterr = torture_create_kthread(rcu_perf_writer, (void *)i,
+						  writer_tasks[i]);
+		if (firsterr)
+			goto unwind;
+	}
+	torture_init_end();
+	return 0;
+
+unwind:
+	torture_init_end();
+	rcu_perf_cleanup();
+	return firsterr;
+}
+
+module_init(rcu_perf_init);
+module_exit(rcu_perf_cleanup);
diff --git a/lib/Kconfig.debug b/lib/Kconfig.debug
index 1e9a607534ca..f4b797a690ba 100644
--- a/lib/Kconfig.debug
+++ b/lib/Kconfig.debug
@@ -1289,6 +1289,39 @@ config TORTURE_TEST
 	tristate
 	default n
 
+config RCU_PERF_TEST
+	tristate "performance tests for RCU"
+	depends on DEBUG_KERNEL
+	select TORTURE_TEST
+	select SRCU
+	select TASKS_RCU
+	default n
+	help
+	  This option provides a kernel module that runs performance
+	  tests on the RCU infrastructure.  The kernel module may be built
+	  after the fact on the running kernel to be tested, if desired.
+
+	  Say Y here if you want RCU performance tests to be built into
+	  the kernel.
+	  Say M if you want the RCU performance tests to build as a module.
+	  Say N if you are unsure.
+
+config RCU_PERF_TEST_RUNNABLE
+	bool "performance tests for RCU runnable by default"
+	depends on RCU_PERF_TEST = y
+	default n
+	help
+	  This option provides a way to build the RCU performance tests
+	  directly into the kernel without them starting up at boot time.
+	  You can use /sys/module to manually override this setting.
+	  This /proc file is available only when the RCU performance
+	  tests have been built into the kernel.
+
+	  Say Y here if you want the RCU performance tests to start during
+	  boot (you probably don't).
+	  Say N here if you want the RCU performance tests to start only
+	  after being manually enabled via /sys/module.
+
 config RCU_TORTURE_TEST
 	tristate "torture tests for RCU"
 	depends on DEBUG_KERNEL
-- 
2.5.2

[PATCH tip/core/rcu 05/24] rcutorture: Documentation for rcuperf kernel parameters

[Paul E. McKenney]

This commit adds documentation for the new rcuperf module's kernel
boot parameters.

Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
---
 Documentation/kernel-parameters.txt | 32 ++++++++++++++++++++++++++++++++
 1 file changed, 32 insertions(+)

diff --git a/Documentation/kernel-parameters.txt b/Documentation/kernel-parameters.txt
index ecc74fa4bfde..951af481da5a 100644
--- a/Documentation/kernel-parameters.txt
+++ b/Documentation/kernel-parameters.txt
@@ -3284,6 +3284,38 @@ bytes respectively. Such letter suffixes can also be entirely omitted.
 			Lazy RCU callbacks are those which RCU can
 			prove do nothing more than free memory.
 
+	rcuperf.gp_exp= [KNL]
+			Measure performance of expedited synchronous
+			grace-period primitives.
+
+	rcuperf.nreaders= [KNL]
+			Set number of RCU readers.  The value -1 selects
+			N, where N is the number of CPUs.  A value
+			"n" less than -1 selects N-n+1, where N is again
+			the number of CPUs.  For example, -2 selects N
+			(the number of CPUs), -3 selects N+1, and so on.
+			A value of "n" less than or equal to -N selects
+			a single reader.
+
+	rcuperf.nwriters= [KNL]
+			Set number of RCU writers.  The values operate
+			the same as for rcuperf.nreaders.
+			N, where N is the number of CPUs
+
+	rcuperf.perf_runnable= [BOOT]
+			Start rcuperf running at boot time.
+
+	rcuperf.shutdown= [KNL]
+			Shut the system down after performance tests
+			complete.  This is useful for hands-off automated
+			testing.
+
+	rcuperf.perf_type= [KNL]
+			Specify the RCU implementation to test.
+
+	rcuperf.verbose= [KNL]
+			Enable additional printk() statements.
+
 	rcutorture.cbflood_inter_holdoff= [KNL]
 			Set holdoff time (jiffies) between successive
 			callback-flood tests.
-- 
2.5.2

[PATCH tip/core/rcu 06/24] rcutorture: Bind rcuperf reader/writer kthreads to CPUs

[Paul E. McKenney]

This commit forces more deterministic behavior by binding rcuperf's
rcu_perf_reader() and rcu_perf_writer() kthreads to their respective
CPUs.

Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
---
 kernel/rcu/rcuperf.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/kernel/rcu/rcuperf.c b/kernel/rcu/rcuperf.c
index 9d54a57bee7d..7a1edf417d18 100644
--- a/kernel/rcu/rcuperf.c
+++ b/kernel/rcu/rcuperf.c
@@ -328,8 +328,10 @@ rcu_perf_reader(void *arg)
 {
 	unsigned long flags;
 	int idx;
+	long me = (long)arg;
 
 	VERBOSE_PERFOUT_STRING("rcu_perf_reader task started");
+	set_cpus_allowed_ptr(current, cpumask_of(me % nr_cpu_ids));
 	set_user_nice(current, MAX_NICE);
 	atomic_inc(&n_rcu_perf_reader_started);
 
@@ -362,6 +364,7 @@ rcu_perf_writer(void *arg)
 	WARN_ON(rcu_gp_is_expedited() && !rcu_gp_is_normal() && !gp_exp);
 	WARN_ON(rcu_gp_is_normal() && gp_exp);
 	WARN_ON(!wdpp);
+	set_cpus_allowed_ptr(current, cpumask_of(me % nr_cpu_ids));
 	t = ktime_get_mono_fast_ns();
 	if (atomic_inc_return(&n_rcu_perf_writer_started) >= nrealwriters) {
 		t_rcu_perf_writer_started = t;
@@ -594,7 +597,7 @@ rcu_perf_init(void)
 		goto unwind;
 	}
 	for (i = 0; i < nrealreaders; i++) {
-		firsterr = torture_create_kthread(rcu_perf_reader, NULL,
+		firsterr = torture_create_kthread(rcu_perf_reader, (void *)i,
 						  reader_tasks[i]);
 		if (firsterr)
 			goto unwind;
-- 
2.5.2

[PATCH tip/core/rcu 07/24] rcutorture: Set rcuperf writer kthreads to real-time priority

[Paul E. McKenney]

This commit forces more deterministic update-side behavior by setting
rcuperf's rcu_perf_writer() kthreads to real-time priority.

Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
---
 kernel/rcu/rcuperf.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/kernel/rcu/rcuperf.c b/kernel/rcu/rcuperf.c
index 7a1edf417d18..e18d016a9888 100644
--- a/kernel/rcu/rcuperf.c
+++ b/kernel/rcu/rcuperf.c
@@ -355,6 +355,7 @@ rcu_perf_writer(void *arg)
 	int i = 0;
 	int i_max;
 	long me = (long)arg;
+	struct sched_param sp;
 	bool started = false, done = false, alldone = false;
 	u64 t;
 	u64 *wdp;
@@ -365,6 +366,8 @@ rcu_perf_writer(void *arg)
 	WARN_ON(rcu_gp_is_normal() && gp_exp);
 	WARN_ON(!wdpp);
 	set_cpus_allowed_ptr(current, cpumask_of(me % nr_cpu_ids));
+	sp.sched_priority = 1;
+	sched_setscheduler_nocheck(current, SCHED_FIFO, &sp);
 	t = ktime_get_mono_fast_ns();
 	if (atomic_inc_return(&n_rcu_perf_writer_started) >= nrealwriters) {
 		t_rcu_perf_writer_started = t;
-- 
2.5.2

[PATCH tip/core/rcu 08/24] rcutorture: Print measure of batching efficiency

[Paul E. McKenney]

This commit adds a line giving the number of grace periods, the number
of batches, and the ratio.  The larger the ratio, the greater the
batching efficiency.

Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
---
 .../rcutorture/bin/kvm-recheck-rcuperf.sh          | 26 +++++++++++++---------
 1 file changed, 16 insertions(+), 10 deletions(-)

diff --git a/tools/testing/selftests/rcutorture/bin/kvm-recheck-rcuperf.sh b/tools/testing/selftests/rcutorture/bin/kvm-recheck-rcuperf.sh
index e5b28174fda0..1f72df8eedc7 100755
--- a/tools/testing/selftests/rcutorture/bin/kvm-recheck-rcuperf.sh
+++ b/tools/testing/selftests/rcutorture/bin/kvm-recheck-rcuperf.sh
@@ -34,33 +34,38 @@ fi
 
 configfile=`echo $i | sed -e 's/^.*\///'`
 
-grep -e '-perf:.*writer-duration' $i/console.log | sed -e 's/^\[[^]]*]//' |
+sed -e 's/^\[[^]]*]//' < $i/console.log |
 awk '
-{
+/-perf: .* gps: .* batches:/ {
+	ngps = $9;
+	nbatches = $11;
+}
+
+/-perf: .*writer-duration/ {
 	gptimes[++n] = $5 / 1000.;
 	sum += $5 / 1000.;
 }
 
 END {
-	if (NR <= 0) {
+	newNR = asort(gptimes);
+	if (newNR <= 0) {
 		print "No rcuperf records found???"
 		exit;
 	}
-	asort(gptimes);
-	pct50 = int(NR * 50 / 100);
+	pct50 = int(newNR * 50 / 100);
 	if (pct50 < 1)
 		pct50 = 1;
-	pct90 = int(NR * 90 / 100);
+	pct90 = int(newNR * 90 / 100);
 	if (pct90 < 1)
 		pct90 = 1;
-	pct99 = int(NR * 99 / 100);
+	pct99 = int(newNR * 99 / 100);
 	if (pct99 < 1)
 		pct99 = 1;
 	div = 10 ** int(log(gptimes[pct90]) / log(10) + .5) / 100;
 	print "Histogram bucket size: " div;
 	last = gptimes[1] - 10;
 	count = 0;
-	for (i = 1; i <= NR; i++) {
+	for (i = 1; i <= newNR; i++) {
 		current = div * int(gptimes[i] / div);
 		if (last == current) {
 			count++;
@@ -73,10 +78,11 @@ END {
 	}
 	if (count > 0)
 		print last, count;
-	print "Average grace-period duration: " sum / NR " microseconds";
+	print "Average grace-period duration: " sum / newNR " microseconds";
 	print "Minimum grace-period duration: " gptimes[1];
 	print "50th percentile grace-period duration: " gptimes[pct50];
 	print "90th percentile grace-period duration: " gptimes[pct90];
 	print "99th percentile grace-period duration: " gptimes[pct99];
-	print "Maximum grace-period duration: " gptimes[NR];
+	print "Maximum grace-period duration: " gptimes[newNR];
+	print "Grace periods: " ngps + 0 " Batches: " nbatches + 0 " Ratio: " ngps / nbatches;
 }'
-- 
2.5.2

[PATCH tip/core/rcu 09/24] rcutorture: Make rcuperf collect expedited event-trace data

[Paul E. McKenney]

This commit enables ftrace in the rcuperf TREE kernel build and adds
an ftrace_dump() at the end of rcuperf processing.  This data will be
used to measure the actual durations of the expedited grace periods
without the added delays inherent in the kernel-module measurements.

Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
---
 kernel/rcu/rcuperf.c                                    | 1 +
 tools/testing/selftests/rcutorture/configs/rcuperf/TREE | 1 +
 2 files changed, 2 insertions(+)

diff --git a/kernel/rcu/rcuperf.c b/kernel/rcu/rcuperf.c
index e18d016a9888..12561f96f0a2 100644
--- a/kernel/rcu/rcuperf.c
+++ b/kernel/rcu/rcuperf.c
@@ -404,6 +404,7 @@ rcu_perf_writer(void *arg)
 				 perf_type, me, MIN_MEAS);
 			if (atomic_inc_return(&n_rcu_perf_writer_finished) >=
 			    nrealwriters) {
+				rcu_ftrace_dump(DUMP_ALL);
 				PERFOUT_STRING("Test complete");
 				t_rcu_perf_writer_finished = t;
 				if (gp_exp) {
diff --git a/tools/testing/selftests/rcutorture/configs/rcuperf/TREE b/tools/testing/selftests/rcutorture/configs/rcuperf/TREE
index 614e107f6db5..a312f671a29a 100644
--- a/tools/testing/selftests/rcutorture/configs/rcuperf/TREE
+++ b/tools/testing/selftests/rcutorture/configs/rcuperf/TREE
@@ -17,3 +17,4 @@ CONFIG_PROVE_LOCKING=n
 CONFIG_RCU_BOOST=n
 CONFIG_DEBUG_OBJECTS_RCU_HEAD=n
 CONFIG_RCU_EXPERT=y
+CONFIG_RCU_TRACE=y
-- 
2.5.2

[PATCH tip/core/rcu 10/24] rcutorture: Make scripts analyze rcuperf trace data, if present

[Paul E. McKenney]

The rcuperf event-trace data is more accurate than are the rcuperf
printk()s because locking keeps things ordered.  This commit therefore
parses and analyzes this event-trace data if present, and falls back on
the printk()s otherwise.

Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
---
 .../rcutorture/bin/kvm-recheck-rcuperf-ftrace.sh   | 121 +++++++++++++++++++++
 .../rcutorture/bin/kvm-recheck-rcuperf.sh          |   8 ++
 2 files changed, 129 insertions(+)
 create mode 100755 tools/testing/selftests/rcutorture/bin/kvm-recheck-rcuperf-ftrace.sh

diff --git a/tools/testing/selftests/rcutorture/bin/kvm-recheck-rcuperf-ftrace.sh b/tools/testing/selftests/rcutorture/bin/kvm-recheck-rcuperf-ftrace.sh
new file mode 100755
index 000000000000..f79b0e9e84fc
--- /dev/null
+++ b/tools/testing/selftests/rcutorture/bin/kvm-recheck-rcuperf-ftrace.sh
@@ -0,0 +1,121 @@
+#!/bin/bash
+#
+# Analyze a given results directory for rcuperf performance measurements,
+# looking for ftrace data.  Exits with 0 if data was found, analyzed, and
+# printed.  Intended to be invoked from kvm-recheck-rcuperf.sh after
+# argument checking.
+#
+# Usage: kvm-recheck-rcuperf-ftrace.sh resdir
+#
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 2 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program; if not, you can access it online at
+# http://www.gnu.org/licenses/gpl-2.0.html.
+#
+# Copyright (C) IBM Corporation, 2016
+#
+# Authors: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
+
+i="$1"
+. tools/testing/selftests/rcutorture/bin/functions.sh
+
+if test "`grep -c 'rcu_exp_grace_period.*start' < $i/console.log`" -lt 100
+then
+	exit 10
+fi
+
+sed -e 's/^\[[^]]*]//' < $i/console.log |
+grep 'us : rcu_exp_grace_period' |
+sed -e 's/us : / : /' |
+tr -d '\015' |
+awk '
+$8 == "start" {
+	if (starttask != "")
+		nlost++;
+	starttask = $1;
+	starttime = $3;
+	startseq = $7;
+}
+
+$8 == "end" {
+	if (starttask == $1 && startseq == $7) {
+		curgpdur = $3 - starttime;
+		gptimes[++n] = curgpdur;
+		gptaskcnt[starttask]++;
+		sum += curgpdur;
+		if (curgpdur > 1000)
+			print "Long GP " starttime "us to " $3 "us (" curgpdur "us)";
+		starttask = "";
+	} else {
+		# Lost a message or some such, reset.
+		starttask = "";
+		nlost++;
+	}
+}
+
+$8 == "done" {
+	piggybackcnt[$1]++;
+}
+
+END {
+	newNR = asort(gptimes);
+	if (newNR <= 0) {
+		print "No ftrace records found???"
+		exit 10;
+	}
+	pct50 = int(newNR * 50 / 100);
+	if (pct50 < 1)
+		pct50 = 1;
+	pct90 = int(newNR * 90 / 100);
+	if (pct90 < 1)
+		pct90 = 1;
+	pct99 = int(newNR * 99 / 100);
+	if (pct99 < 1)
+		pct99 = 1;
+	div = 10 ** int(log(gptimes[pct90]) / log(10) + .5) / 100;
+	print "Histogram bucket size: " div;
+	last = gptimes[1] - 10;
+	count = 0;
+	for (i = 1; i <= newNR; i++) {
+		current = div * int(gptimes[i] / div);
+		if (last == current) {
+			count++;
+		} else {
+			if (count > 0)
+				print last, count;
+			count = 1;
+			last = current;
+		}
+	}
+	if (count > 0)
+		print last, count;
+	print "Distribution of grace periods across tasks:";
+	for (i in gptaskcnt) {
+		print "\t" i, gptaskcnt[i];
+		nbatches += gptaskcnt[i];
+	}
+	ngps = nbatches;
+	print "Distribution of piggybacking across tasks:";
+	for (i in piggybackcnt) {
+		print "\t" i, piggybackcnt[i];
+		ngps += piggybackcnt[i];
+	}
+	print "Average grace-period duration: " sum / newNR " microseconds";
+	print "Minimum grace-period duration: " gptimes[1];
+	print "50th percentile grace-period duration: " gptimes[pct50];
+	print "90th percentile grace-period duration: " gptimes[pct90];
+	print "99th percentile grace-period duration: " gptimes[pct99];
+	print "Maximum grace-period duration: " gptimes[newNR];
+	print "Grace periods: " ngps + 0 " Batches: " nbatches + 0 " Ratio: " ngps / nbatches " Lost: " nlost + 0;
+	print "Computed from ftrace data.";
+}'
+exit 0
diff --git a/tools/testing/selftests/rcutorture/bin/kvm-recheck-rcuperf.sh b/tools/testing/selftests/rcutorture/bin/kvm-recheck-rcuperf.sh
index 1f72df8eedc7..8f3121afc716 100755
--- a/tools/testing/selftests/rcutorture/bin/kvm-recheck-rcuperf.sh
+++ b/tools/testing/selftests/rcutorture/bin/kvm-recheck-rcuperf.sh
@@ -30,8 +30,15 @@ else
 	echo Unreadable results directory: $i
 	exit 1
 fi
+PATH=`pwd`/tools/testing/selftests/rcutorture/bin:$PATH; export PATH
 . tools/testing/selftests/rcutorture/bin/functions.sh
 
+if kvm-recheck-rcuperf-ftrace.sh $i
+then
+	# ftrace data was successfully analyzed, call it good!
+	exit 0
+fi
+
 configfile=`echo $i | sed -e 's/^.*\///'`
 
 sed -e 's/^\[[^]]*]//' < $i/console.log |
@@ -85,4 +92,5 @@ END {
 	print "99th percentile grace-period duration: " gptimes[pct99];
 	print "Maximum grace-period duration: " gptimes[newNR];
 	print "Grace periods: " ngps + 0 " Batches: " nbatches + 0 " Ratio: " ngps / nbatches;
+	print "Computed from rcuperf printk output.";
 }'
-- 
2.5.2

[PATCH tip/core/rcu 11/24] rcutorture: Add rcuperf holdoff boot parameter to reduce interference

[Paul E. McKenney]

Boot-time activity can legitimately grab CPUs for extended time periods,
so the commit adds a boot parameter to delay the start of the performance
test until boot has completed.  Defaults to 10 seconds.

Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
---
 Documentation/kernel-parameters.txt | 6 ++++++
 kernel/rcu/rcuperf.c                | 5 +++++
 2 files changed, 11 insertions(+)

diff --git a/Documentation/kernel-parameters.txt b/Documentation/kernel-parameters.txt
index 951af481da5a..da9ee466789b 100644
--- a/Documentation/kernel-parameters.txt
+++ b/Documentation/kernel-parameters.txt
@@ -3288,6 +3288,12 @@ bytes respectively. Such letter suffixes can also be entirely omitted.
 			Measure performance of expedited synchronous
 			grace-period primitives.
 
+	rcuperf.holdoff= [KNL]
+			Set test-start holdoff period.  The purpose of
+			this parameter is to delay the start of the
+			test until boot completes in order to avoid
+			interference.
+
 	rcuperf.nreaders= [KNL]
 			Set number of RCU readers.  The value -1 selects
 			N, where N is the number of CPUs.  A value
diff --git a/kernel/rcu/rcuperf.c b/kernel/rcu/rcuperf.c
index 12561f96f0a2..278600143bb6 100644
--- a/kernel/rcu/rcuperf.c
+++ b/kernel/rcu/rcuperf.c
@@ -59,6 +59,7 @@ MODULE_AUTHOR("Paul E. McKenney <paulmck@linux.vnet.ibm.com>");
 	do { if (verbose) pr_alert("%s" PERF_FLAG "!!! %s\n", perf_type, s); } while (0)
 
 torture_param(bool, gp_exp, true, "Use expedited GP wait primitives");
+torture_param(int, holdoff, 10, "Holdoff time before test start (s)");
 torture_param(int, nreaders, -1, "Number of RCU reader threads");
 torture_param(int, nwriters, -1, "Number of RCU updater threads");
 torture_param(bool, shutdown, false, "Shutdown at end of performance tests.");
@@ -368,6 +369,10 @@ rcu_perf_writer(void *arg)
 	set_cpus_allowed_ptr(current, cpumask_of(me % nr_cpu_ids));
 	sp.sched_priority = 1;
 	sched_setscheduler_nocheck(current, SCHED_FIFO, &sp);
+
+	if (holdoff)
+		schedule_timeout_uninterruptible(holdoff * HZ);
+
 	t = ktime_get_mono_fast_ns();
 	if (atomic_inc_return(&n_rcu_perf_writer_started) >= nrealwriters) {
 		t_rcu_perf_writer_started = t;
-- 
2.5.2

[PATCH tip/core/rcu 12/24] rcutorture: Avoid RCU CPU stall warning and RT throttling

[Paul E. McKenney]

Running rcuperf can result in RCU CPU stall warnings and RT throttling.
These occur because on of the real-time writer processes does
ftrace_dump() while still running at real-time priority.  This commit
therefore prevents these problems by setting the writer thread back to
SCHED_NORMAL (AKA SCHED_OTHER) before doing ftrace_dump().

In addition, this commit adds a small fixed delay before dumping ftrace
buffer in order to decrease the probability that this dumping will
interfere with other writers' grace periods.

Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
---
 kernel/rcu/rcuperf.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/kernel/rcu/rcuperf.c b/kernel/rcu/rcuperf.c
index 278600143bb6..4c0572859ff0 100644
--- a/kernel/rcu/rcuperf.c
+++ b/kernel/rcu/rcuperf.c
@@ -404,11 +404,15 @@ rcu_perf_writer(void *arg)
 			started = true;
 		if (!done && i >= MIN_MEAS) {
 			done = true;
+			sp.sched_priority = 0;
+			sched_setscheduler_nocheck(current,
+						   SCHED_NORMAL, &sp);
 			pr_alert("%s" PERF_FLAG
 				 "rcu_perf_writer %ld has %d measurements\n",
 				 perf_type, me, MIN_MEAS);
 			if (atomic_inc_return(&n_rcu_perf_writer_finished) >=
 			    nrealwriters) {
+				schedule_timeout_interruptible(10);
 				rcu_ftrace_dump(DUMP_ALL);
 				PERFOUT_STRING("Test complete");
 				t_rcu_perf_writer_finished = t;
-- 
2.5.2

[PATCH tip/core/rcu 13/24] rcutorture: Add largish-system rcuperf scenario

[Paul E. McKenney]

This commit adds an rcuperf scenario named TREE54 that uses 54 CPUs and
provides a four-level rcu_node combining tree.

Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
---
 .../selftests/rcutorture/configs/rcuperf/TREE54    | 23 ++++++++++++++++++++++
 1 file changed, 23 insertions(+)
 create mode 100644 tools/testing/selftests/rcutorture/configs/rcuperf/TREE54

diff --git a/tools/testing/selftests/rcutorture/configs/rcuperf/TREE54 b/tools/testing/selftests/rcutorture/configs/rcuperf/TREE54
new file mode 100644
index 000000000000..985fb170d13c
--- /dev/null
+++ b/tools/testing/selftests/rcutorture/configs/rcuperf/TREE54
@@ -0,0 +1,23 @@
+CONFIG_SMP=y
+CONFIG_NR_CPUS=54
+CONFIG_PREEMPT_NONE=n
+CONFIG_PREEMPT_VOLUNTARY=n
+CONFIG_PREEMPT=y
+#CHECK#CONFIG_PREEMPT_RCU=y
+CONFIG_HZ_PERIODIC=n
+CONFIG_NO_HZ_IDLE=y
+CONFIG_NO_HZ_FULL=n
+CONFIG_RCU_FAST_NO_HZ=n
+CONFIG_RCU_TRACE=n
+CONFIG_HOTPLUG_CPU=n
+CONFIG_SUSPEND=n
+CONFIG_HIBERNATION=n
+CONFIG_RCU_FANOUT=3
+CONFIG_RCU_FANOUT_LEAF=2
+CONFIG_RCU_NOCB_CPU=n
+CONFIG_DEBUG_LOCK_ALLOC=n
+CONFIG_PROVE_LOCKING=n
+CONFIG_RCU_BOOST=n
+CONFIG_DEBUG_OBJECTS_RCU_HEAD=n
+CONFIG_RCU_EXPERT=y
+CONFIG_RCU_TRACE=y
-- 
2.5.2

[PATCH tip/core/rcu 14/24] rcuperf: Do not wake up shutdown wait queue if "shutdown" is false.

[Paul E. McKenney]

From: Artem Savkov <artem.savkov@gmail.com>

After finishing its tests rcuperf tries to wake up shutdown_wq even if
"shutdown" param is set to false, resulting in a wake_up() call on an
unitialized wait_queue_head_t which leads to "BUG: spinlock bad magic" and
"BUG: unable to handle kernel NULL pointer dereference".

Fix by checking "shutdown" param before waking up the queue.

Signed-off-by: Artem Savkov <artem.savkov@gmail.com>
---
 kernel/rcu/rcuperf.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/kernel/rcu/rcuperf.c b/kernel/rcu/rcuperf.c
index 4c0572859ff0..3cee0d8393ed 100644
--- a/kernel/rcu/rcuperf.c
+++ b/kernel/rcu/rcuperf.c
@@ -423,8 +423,10 @@ rcu_perf_writer(void *arg)
 					b_rcu_perf_writer_finished =
 						cur_ops->completed();
 				}
-				smp_mb(); /* Assign before wake. */
-				wake_up(&shutdown_wq);
+				if (shutdown) {
+					smp_mb(); /* Assign before wake. */
+					wake_up(&shutdown_wq);
+				}
 			}
 		}
 		if (done && !alldone &&
-- 
2.5.2

[PATCH tip/core/rcu 15/24] rcutorture: Remove redundant initialization to zero

[Paul E. McKenney]

The current code initializes the global per-CPU variables
rcu_torture_count and rcu_torture_batch to zero.  However, C does this
initialization by default, and explicit initialization of per-CPU
variables now needs a different syntax if "make tags" is to work.
This commit therefore removes the initialization.

Reported-by: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
---
 kernel/rcu/rcutorture.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/kernel/rcu/rcutorture.c b/kernel/rcu/rcutorture.c
index 9234e75b106a..52b49fe90919 100644
--- a/kernel/rcu/rcutorture.c
+++ b/kernel/rcu/rcutorture.c
@@ -130,8 +130,8 @@ static struct rcu_torture __rcu *rcu_torture_current;
 static unsigned long rcu_torture_current_version;
 static struct rcu_torture rcu_tortures[10 * RCU_TORTURE_PIPE_LEN];
 static DEFINE_SPINLOCK(rcu_torture_lock);
-static DEFINE_PER_CPU(long [RCU_TORTURE_PIPE_LEN + 1], rcu_torture_count) = { 0 };
-static DEFINE_PER_CPU(long [RCU_TORTURE_PIPE_LEN + 1], rcu_torture_batch) = { 0 };
+static DEFINE_PER_CPU(long [RCU_TORTURE_PIPE_LEN + 1], rcu_torture_count);
+static DEFINE_PER_CPU(long [RCU_TORTURE_PIPE_LEN + 1], rcu_torture_batch);
 static atomic_t rcu_torture_wcount[RCU_TORTURE_PIPE_LEN + 1];
 static atomic_t n_rcu_torture_alloc;
 static atomic_t n_rcu_torture_alloc_fail;
-- 
2.5.2

[PATCH tip/core/rcu 16/24] rcutorture: Consider FROZEN hotplug notifier transitions

[Paul E. McKenney]

From: Anna-Maria Gleixner <anna-maria@linutronix.de>

The hotplug notifier rcutorture_cpu_notify() doesn't consider the
corresponding CPU_XXX_FROZEN transitions. They occur on
suspend/resume and are usually handled the same way as the
corresponding non frozen transitions.

Mask the switch case action argument with '~CPU_TASKS_FROZEN' to map
CPU_XXX_FROZEN hotplug transitions on corresponding non-frozen
transitions.

Cc: Josh Triplett <josh@joshtriplett.org>
Cc: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
---
 kernel/rcu/rcutorture.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/kernel/rcu/rcutorture.c b/kernel/rcu/rcutorture.c
index 52b49fe90919..633a68a09440 100644
--- a/kernel/rcu/rcutorture.c
+++ b/kernel/rcu/rcutorture.c
@@ -1585,7 +1585,7 @@ static int rcutorture_cpu_notify(struct notifier_block *self,
 {
 	long cpu = (long)hcpu;
 
-	switch (action) {
+	switch (action & ~CPU_TASKS_FROZEN) {
 	case CPU_ONLINE:
 	case CPU_DOWN_FAILED:
 		(void)rcutorture_booster_init(cpu);
-- 
2.5.2

[PATCH tip/core/rcu 17/24] torture: Clarify refusal to run more than one torture test

[Paul E. McKenney]

This commit clarifies error messages -- you only get to run one torture
test at a time!

Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
---
 kernel/torture.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/kernel/torture.c b/kernel/torture.c
index 44aa462d033f..e912ccd960f0 100644
--- a/kernel/torture.c
+++ b/kernel/torture.c
@@ -602,8 +602,9 @@ bool torture_init_begin(char *ttype, bool v, int *runnable)
 {
 	mutex_lock(&fullstop_mutex);
 	if (torture_type != NULL) {
-		pr_alert("torture_init_begin: refusing %s init: %s running",
+		pr_alert("torture_init_begin: Refusing %s init: %s running.\n",
 			 ttype, torture_type);
+		pr_alert("torture_init_begin: One torture test at a time!\n");
 		mutex_unlock(&fullstop_mutex);
 		return false;
 	}
-- 
2.5.2

[PATCH tip/core/rcu 18/24] torture: Kill qemu, not parent process

[Paul E. McKenney]

The current hang-check machinery in the rcutorture scripts uses "$!" of
a parenthesized bash statement to capture the pid.  Unfortunately, this
captures not qemu's pid, but rather that of its parent that implements
the parenthesized statement.  This commit therefore adjusts things so as
to capture qemu's actual pid, which then allows the script to actually
kill qemu in event of a kernel hang.

Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
---
 .../selftests/rcutorture/bin/kvm-test-1-run.sh     | 30 +++++++++++++++++-----
 1 file changed, 24 insertions(+), 6 deletions(-)

diff --git a/tools/testing/selftests/rcutorture/bin/kvm-test-1-run.sh b/tools/testing/selftests/rcutorture/bin/kvm-test-1-run.sh
index 0f80eefb0bfd..2eb8fefbe7d9 100755
--- a/tools/testing/selftests/rcutorture/bin/kvm-test-1-run.sh
+++ b/tools/testing/selftests/rcutorture/bin/kvm-test-1-run.sh
@@ -168,14 +168,25 @@ then
 fi
 echo "NOTE: $QEMU either did not run or was interactive" > $resdir/console.log
 echo $QEMU $qemu_args -m 512 -kernel $resdir/bzImage -append \"$qemu_append $boot_args\" > $resdir/qemu-cmd
-( $QEMU $qemu_args -m 512 -kernel $resdir/bzImage -append "$qemu_append $boot_args"; echo $? > $resdir/qemu-retval ) &
-qemu_pid=$!
+( $QEMU $qemu_args -m 512 -kernel $resdir/bzImage -append "$qemu_append $boot_args"& echo $! > $resdir/qemu_pid; wait `cat  $resdir/qemu_pid`; echo $? > $resdir/qemu-retval ) &
 commandcompleted=0
-echo Monitoring qemu job at pid $qemu_pid
+sleep 10 # Give qemu's pid a chance to reach the file
+if test -s "$resdir/qemu_pid"
+then
+	qemu_pid=`cat "$resdir/qemu_pid"`
+	echo Monitoring qemu job at pid $qemu_pid
+else
+	qemu_pid=""
+	echo Monitoring qemu job at yet-as-unknown pid
+fi
 while :
 do
+	if test -z "$qemu_pid" -a -s "$resdir/qemu_pid"
+	then
+		qemu_pid=`cat "$resdir/qemu_pid"`
+	fi
 	kruntime=`awk 'BEGIN { print systime() - '"$kstarttime"' }' < /dev/null`
-	if kill -0 $qemu_pid > /dev/null 2>&1
+	if test -z "$qemu_pid" || kill -0 "$qemu_pid" > /dev/null 2>&1
 	then
 		if test $kruntime -ge $seconds
 		then
@@ -195,12 +206,16 @@ do
 				ps -fp $killpid >> $resdir/Warnings 2>&1
 			fi
 		else
-			echo ' ---' `date`: Kernel done
+			echo ' ---' `date`: "Kernel done"
 		fi
 		break
 	fi
 done
-if test $commandcompleted -eq 0
+if test -z "$qemu_pid" -a -s "$resdir/qemu_pid"
+then
+	qemu_pid=`cat "$resdir/qemu_pid"`
+fi
+if test $commandcompleted -eq 0 -a -n "$qemu_pid"
 then
 	echo Grace period for qemu job at pid $qemu_pid
 	while :
@@ -220,6 +235,9 @@ then
 		fi
 		sleep 1
 	done
+elif test -z "$qemu_pid"
+then
+	echo Unknown PID, cannot kill qemu command
 fi
 
 parse-torture.sh $resdir/console.log $title
-- 
2.5.2

[PATCH tip/core/rcu 19/24] rcutorture: Convert test duration to seconds early

[Paul E. McKenney]

This commit converts test duration from minutes to seconds early on
in order to prepare for upcoming OS-jitter-injection changes.

Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
---
 tools/testing/selftests/rcutorture/bin/kvm-test-1-run.sh | 5 ++---
 tools/testing/selftests/rcutorture/bin/kvm.sh            | 4 ++--
 2 files changed, 4 insertions(+), 5 deletions(-)

diff --git a/tools/testing/selftests/rcutorture/bin/kvm-test-1-run.sh b/tools/testing/selftests/rcutorture/bin/kvm-test-1-run.sh
index 2eb8fefbe7d9..73a265668421 100755
--- a/tools/testing/selftests/rcutorture/bin/kvm-test-1-run.sh
+++ b/tools/testing/selftests/rcutorture/bin/kvm-test-1-run.sh
@@ -6,7 +6,7 @@
 # Execute this in the source tree.  Do not run it as a background task
 # because qemu does not seem to like that much.
 #
-# Usage: kvm-test-1-run.sh config builddir resdir minutes qemu-args boot_args
+# Usage: kvm-test-1-run.sh config builddir resdir seconds qemu-args boot_args
 #
 # qemu-args defaults to "-enable-kvm -soundhw pcspk -nographic", along with
 #			arguments specifying the number of CPUs and other
@@ -123,8 +123,7 @@ while test -f $builddir.ready
 do
 	sleep 1
 done
-minutes=$4
-seconds=$(($minutes * 60))
+seconds=$4
 qemu_args=$5
 boot_args=$6
 
diff --git a/tools/testing/selftests/rcutorture/bin/kvm.sh b/tools/testing/selftests/rcutorture/bin/kvm.sh
index c33cb582b3dc..704e219f67a7 100755
--- a/tools/testing/selftests/rcutorture/bin/kvm.sh
+++ b/tools/testing/selftests/rcutorture/bin/kvm.sh
@@ -34,7 +34,7 @@ T=/tmp/kvm.sh.$$
 trap 'rm -rf $T' 0
 mkdir $T
 
-dur=30
+dur=$((30*60))
 dryrun=""
 KVM="`pwd`/tools/testing/selftests/rcutorture"; export KVM
 PATH=${KVM}/bin:$PATH; export PATH
@@ -116,7 +116,7 @@ do
 		;;
 	--duration)
 		checkarg --duration "(minutes)" $# "$2" '^[0-9]*$' '^error'
-		dur=$2
+		dur=$(($2*60))
 		shift
 		;;
 	--interactive)
-- 
2.5.2

[PATCH tip/core/rcu 20/24] rcutorture: Add OS-jitter capability

[Paul E. McKenney]

This commit adds a --jitter OS-jitter capability to expose bugs based
on no-delay assumptions.

Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
---
 tools/testing/selftests/rcutorture/bin/jitter.sh | 83 ++++++++++++++++++++++++
 tools/testing/selftests/rcutorture/bin/kvm.sh    | 18 +++++
 2 files changed, 101 insertions(+)
 create mode 100755 tools/testing/selftests/rcutorture/bin/jitter.sh

diff --git a/tools/testing/selftests/rcutorture/bin/jitter.sh b/tools/testing/selftests/rcutorture/bin/jitter.sh
new file mode 100755
index 000000000000..276d6b93f9ec
--- /dev/null
+++ b/tools/testing/selftests/rcutorture/bin/jitter.sh
@@ -0,0 +1,83 @@
+#!/bin/bash
+#
+# Alternate sleeping and spinning on randomly selected CPUs.  The purpose
+# of this script is to inflict random OS jitter on a concurrently running
+# test.
+#
+# Usage: jitter.sh me duration [ sleepmax [ spinmax ] ]
+#
+# me: Random-number-generator seed salt.
+# duration: Time to run in seconds.
+# sleepmax: Maximum microseconds to sleep, defaults to one second.
+# spinmax: Maximum microseconds to spin, defaults to one millisecond.
+#
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 2 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program; if not, you can access it online at
+# http://www.gnu.org/licenses/gpl-2.0.html.
+#
+# Copyright (C) IBM Corporation, 2016
+#
+# Authors: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
+
+me=$(($1 * 1000))
+duration=$2
+sleepmax=${3-1000000}
+spinmax=${4-1000}
+
+n=1
+
+starttime=`awk 'BEGIN { print systime(); }' < /dev/null`
+
+while :
+do
+	# Check for done.
+	t=`awk -v s=$starttime 'BEGIN { print systime() - s; }' < /dev/null`
+	if test "$t" -gt "$duration"
+	then
+		exit 0;
+	fi
+
+	# Set affinity to randomly selected CPU
+	cpus=`ls /sys/devices/system/cpu/*/online |
+		sed -e 's,/[^/]*$,,' -e 's/^[^0-9]*//' |
+		grep -v '^0*$'`
+	cpumask=`awk -v cpus="$cpus" -v me=$me -v n=$n 'BEGIN {
+		srand(n + me + systime());
+		ncpus = split(cpus, ca);
+		curcpu = ca[int(rand() * ncpus + 1)];
+		printf("%#x\n", lshift(1, curcpu));
+	}' < /dev/null`
+	n=$(($n+1))
+	taskset -p $cpumask $$ > /dev/null
+
+	# Sleep a random duration
+	sleeptime=`awk -v me=$me -v n=$n -v sleepmax=$sleepmax 'BEGIN {
+		srand(n + me + systime());
+		printf("%06d", int(rand() * sleepmax));
+	}' < /dev/null`
+	n=$(($n+1))
+	sleep .$sleeptime
+
+	# Spin a random duration
+	limit=`awk -v me=$me -v n=$n -v spinmax=$spinmax 'BEGIN {
+		srand(n + me + systime());
+		printf("%06d", int(rand() * spinmax));
+	}' < /dev/null`
+	n=$(($n+1))
+	for i in {1..$limit}
+	do
+		echo > /dev/null
+	done
+done
+
+exit 1
diff --git a/tools/testing/selftests/rcutorture/bin/kvm.sh b/tools/testing/selftests/rcutorture/bin/kvm.sh
index 704e219f67a7..0d598145873e 100755
--- a/tools/testing/selftests/rcutorture/bin/kvm.sh
+++ b/tools/testing/selftests/rcutorture/bin/kvm.sh
@@ -48,6 +48,7 @@ resdir=""
 configs=""
 cpus=0
 ds=`date +%Y.%m.%d-%H:%M:%S`
+jitter=0
 
 . functions.sh
 
@@ -63,6 +64,7 @@ usage () {
 	echo "       --dryrun sched|script"
 	echo "       --duration minutes"
 	echo "       --interactive"
+	echo "       --jitter N [ maxsleep (us) [ maxspin (us) ] ]"
 	echo "       --kmake-arg kernel-make-arguments"
 	echo "       --mac nn:nn:nn:nn:nn:nn"
 	echo "       --no-initrd"
@@ -122,6 +124,11 @@ do
 	--interactive)
 		TORTURE_QEMU_INTERACTIVE=1; export TORTURE_QEMU_INTERACTIVE
 		;;
+	--jitter)
+		checkarg --jitter "(# threads [ sleep [ spin ] ])" $# "$2" '^-\{,1\}[0-9]\+\( \+[0-9]\+\)\{,2\} *$' '^error$'
+		jitter="$2"
+		shift
+		;;
 	--kmake-arg)
 		checkarg --kmake-arg "(kernel make arguments)" $# "$2" '.*' '^error$'
 		TORTURE_KMAKE_ARG="$2"
@@ -299,6 +306,7 @@ awk < $T/cfgcpu.pack \
 	-v CONFIGDIR="$CONFIGFRAG/" \
 	-v KVM="$KVM" \
 	-v ncpus=$cpus \
+	-v jitter="$jitter" \
 	-v rd=$resdir/$ds/ \
 	-v dur=$dur \
 	-v TORTURE_QEMU_ARG="$TORTURE_QEMU_ARG" \
@@ -359,6 +367,16 @@ function dump(first, pastlast, batchnum)
 		print "\techo ----", cfr[j], cpusr[j] ovf ": Starting kernel. `date` >> " rd "/log";
 		print "fi"
 	}
+	njitter = 0;
+	split(jitter, ja);
+	if (ja[1] == -1 && ncpus == 0)
+		njitter = 1;
+	else if (ja[1] == -1)
+		njitter = ncpus;
+	else
+		njitter = ja[1];
+	for (j = 0; j < njitter; j++)
+		print "jitter.sh " j " " dur " " ja[2] " " ja[3] "&"
 	print "wait"
 	print "if test -z \"$TORTURE_BUILDONLY\""
 	print "then"
-- 
2.5.2

[PATCH tip/core/rcu 21/24] rcutorture: Don't rebuild identical kernel

[Paul E. McKenney]

Currently, if the user specifies multiple runs of a given test
configuration, the scripting does multiple kernel builds.  This wastes
both time and disk space, so this commit makes the scripting use the
first build for all runs of a given test configuration.

Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
---
 .../selftests/rcutorture/bin/kvm-test-1-run.sh     | 26 ++++++++++++++++------
 1 file changed, 19 insertions(+), 7 deletions(-)

diff --git a/tools/testing/selftests/rcutorture/bin/kvm-test-1-run.sh b/tools/testing/selftests/rcutorture/bin/kvm-test-1-run.sh
index 73a265668421..4109f306d855 100755
--- a/tools/testing/selftests/rcutorture/bin/kvm-test-1-run.sh
+++ b/tools/testing/selftests/rcutorture/bin/kvm-test-1-run.sh
@@ -91,25 +91,33 @@ fi
 # CONFIG_PCMCIA=n
 # CONFIG_CARDBUS=n
 # CONFIG_YENTA=n
-if kvm-build.sh $config_template $builddir $T
+base_resdir=`echo $resdir | sed -e 's/\.[0-9]\+$//'`
+if test "$base_resdir" != "$resdir" -a -f $base_resdir/bzImage -a -f $base_resdir/vmlinux
 then
+	# Rerunning previous test, so use that test's kernel.
+	QEMU="`identify_qemu $base_resdir/vmlinux`"
+	KERNEL=$base_resdir/bzImage
+	ln -s $base_resdir/Make*.out $resdir  # for kvm-recheck.sh
+	ln -s $base_resdir/.config $resdir  # for kvm-recheck.sh
+elif kvm-build.sh $config_template $builddir $T
+then
+	# Had to build a kernel for this test.
 	QEMU="`identify_qemu $builddir/vmlinux`"
 	BOOT_IMAGE="`identify_boot_image $QEMU`"
 	cp $builddir/Make*.out $resdir
+	cp $builddir/vmlinux $resdir
 	cp $builddir/.config $resdir
 	if test -n "$BOOT_IMAGE"
 	then
 		cp $builddir/$BOOT_IMAGE $resdir
+		KERNEL=$resdir/bzImage
 	else
 		echo No identifiable boot image, not running KVM, see $resdir.
 		echo Do the torture scripts know about your architecture?
 	fi
 	parse-build.sh $resdir/Make.out $title
-	if test -f $builddir.wait
-	then
-		mv $builddir.wait $builddir.ready
-	fi
 else
+	# Build failed.
 	cp $builddir/Make*.out $resdir
 	cp $builddir/.config $resdir || :
 	echo Build failed, not running KVM, see $resdir.
@@ -119,6 +127,10 @@ else
 	fi
 	exit 1
 fi
+if test -f $builddir.wait
+then
+	mv $builddir.wait $builddir.ready
+fi
 while test -f $builddir.ready
 do
 	sleep 1
@@ -166,8 +178,8 @@ then
 	exit 0
 fi
 echo "NOTE: $QEMU either did not run or was interactive" > $resdir/console.log
-echo $QEMU $qemu_args -m 512 -kernel $resdir/bzImage -append \"$qemu_append $boot_args\" > $resdir/qemu-cmd
-( $QEMU $qemu_args -m 512 -kernel $resdir/bzImage -append "$qemu_append $boot_args"& echo $! > $resdir/qemu_pid; wait `cat  $resdir/qemu_pid`; echo $? > $resdir/qemu-retval ) &
+echo $QEMU $qemu_args -m 512 -kernel $KERNEL -append \"$qemu_append $boot_args\" > $resdir/qemu-cmd
+( $QEMU $qemu_args -m 512 -kernel $KERNEL -append "$qemu_append $boot_args"& echo $! > $resdir/qemu_pid; wait `cat  $resdir/qemu_pid`; echo $? > $resdir/qemu-retval ) &
 commandcompleted=0
 sleep 10 # Give qemu's pid a chance to reach the file
 if test -s "$resdir/qemu_pid"
-- 
2.5.2

[PATCH tip/core/rcu 22/24] rcutorture: Dump trace buffer upon shutdown

[Paul E. McKenney]

When running from the scripts, rcutorture is completely headless,
so there is no way to to manually dump the trace buffer.  This commit
therefore unconditionally dumps the trace buffer upon timed shutdown.
However, if you are using rmmod to end the test, it is still up to you
to manually dump the trace buffer.

Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
---
 kernel/torture.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/kernel/torture.c b/kernel/torture.c
index e912ccd960f0..fa0bdeee17ac 100644
--- a/kernel/torture.c
+++ b/kernel/torture.c
@@ -451,6 +451,7 @@ static int torture_shutdown(void *arg)
 		torture_shutdown_hook();
 	else
 		VERBOSE_TOROUT_STRING("No torture_shutdown_hook(), skipping.");
+	ftrace_dump(DUMP_ALL);
 	kernel_power_off();	/* Shut down the system. */
 	return 0;
 }
-- 
2.5.2

[PATCH tip/core/rcu 23/24] rcutorture: Add irqs-disabled test for call_rcu()

[Paul E. McKenney]

Mutation testing carried out by Iftekhar Ahmed of Oregon State
University showed that rcutorture is failing to test invocations
of call_rcu() having interrupts disabled.  This commit therefore
adds interrupt disabling around one of the existing invocations
of call_rcu() (and friends).

Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
---
 kernel/rcu/rcutorture.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/kernel/rcu/rcutorture.c b/kernel/rcu/rcutorture.c
index 633a68a09440..084a28a732eb 100644
--- a/kernel/rcu/rcutorture.c
+++ b/kernel/rcu/rcutorture.c
@@ -1478,7 +1478,9 @@ static int rcu_torture_barrier_cbs(void *arg)
 		 * The above smp_load_acquire() ensures barrier_phase load
 		 * is ordered before the folloiwng ->call().
 		 */
+		local_irq_disable(); /* Just to test no-irq call_rcu(). */
 		cur_ops->call(&rcu, rcu_torture_barrier_cbf);
+		local_irq_enable();
 		if (atomic_dec_and_test(&barrier_cbs_count))
 			wake_up(&barrier_wq);
 	} while (!torture_must_stop());
-- 
2.5.2

[PATCH tip/core/rcu 24/24] rcutorture: Add boot-time adjustment of leaf fanout

[Paul E. McKenney]

Currently, the rcutorture scripts do not test boot-time adjustment of
leaf fanout (via the rcutree.rcu_fanout_leaf boot parameter), as was noted
during testing carried out by Iftekhar Ahmed of Oregon State University.
This commit therefore adjusts TREE04's CONFIG_RCU_FANOUT_LEAF from 4 to 3,
and also adds rcutree.rcu_fanout_leaf=4 to its boot parameters.  This
change forces RCU's boot-time geometry-change code to be exercised.

Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
---
 tools/testing/selftests/rcutorture/configs/rcu/TREE04      | 2 +-
 tools/testing/selftests/rcutorture/configs/rcu/TREE04.boot | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/tools/testing/selftests/rcutorture/configs/rcu/TREE04 b/tools/testing/selftests/rcutorture/configs/rcu/TREE04
index 39a2c6d7d7ec..17cbe098b115 100644
--- a/tools/testing/selftests/rcutorture/configs/rcu/TREE04
+++ b/tools/testing/selftests/rcutorture/configs/rcu/TREE04
@@ -14,7 +14,7 @@ CONFIG_HOTPLUG_CPU=n
 CONFIG_SUSPEND=n
 CONFIG_HIBERNATION=n
 CONFIG_RCU_FANOUT=4
-CONFIG_RCU_FANOUT_LEAF=4
+CONFIG_RCU_FANOUT_LEAF=3
 CONFIG_RCU_NOCB_CPU=n
 CONFIG_DEBUG_LOCK_ALLOC=n
 CONFIG_DEBUG_OBJECTS_RCU_HEAD=n
diff --git a/tools/testing/selftests/rcutorture/configs/rcu/TREE04.boot b/tools/testing/selftests/rcutorture/configs/rcu/TREE04.boot
index 0fc8a3428938..e34c33430447 100644
--- a/tools/testing/selftests/rcutorture/configs/rcu/TREE04.boot
+++ b/tools/testing/selftests/rcutorture/configs/rcu/TREE04.boot
@@ -1 +1 @@
-rcutorture.torture_type=rcu_bh
+rcutorture.torture_type=rcu_bh rcutree.rcu_fanout_leaf=4
-- 
2.5.2