[RFC PATCH 0/6] RCU: Preemptible-RCU

[RFC PATCH 0/6] RCU: Preemptible-RCU

[Gautham R Shenoy]

Hello everyone, 

This patchset is an updated version of the preemptible RCU patchset that
Paul McKenney had posted it in September earlier this year that can be
found here --> http://lkml.org/lkml/2007/9/10/213

This patchset incorporates the review comments from Oleg Nesterov
and Steven Rostedt.

The testing report of the patchset is as follows:
====================================================================
Patch-stack:  	2.6.23-rc3 + cpu-hotplug patches from 
		http://lkml.org/lkml/2007/11/15/239 + Preempt-RCU
		patches.
Test:		RCU-Torture running parallelly with CPU-Hotplug
		operations.
Duration:	24 hours.
Architectures:	x86,x86_64, ppc64.
====================================================================


Currently it is based against the latest linux-2.6-sched-devel.git

Awaiting your feedback!

Thanks and Regards
gautham.
-- 
Gautham R Shenoy
Linux Technology Center
IBM India.
"Freedom comes with a price tag of responsibility, which is still a bargain,
because Freedom is priceless!"

[RFC PATCH 1/6] Preempt-RCU: Use softirq instead of tasklets for RCU

[Gautham R Shenoy]

Preempt-RCU: Use softirq instead of tasklets for RCU

From: Dipankar Sarma <dipankar@in.ibm.com>

This patch makes RCU use softirq instead of tasklets.

It also adds a memory barrier after raising the softirq
inorder to ensure that the cpu sees the most recently updated
value of rcu->cur while processing callbacks.
The discussion of the related theoretical race pointed out
by James Huang can be found here --> http://lkml.org/lkml/2007/11/20/603

Signed-off-by: Gautham R Shenoy <ego@in.ibm.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Dipankar Sarma <dipankar@in.ibm.com>
---

 include/linux/interrupt.h |    1 +
 kernel/rcupdate.c         |   25 +++++++++++++++++--------
 2 files changed, 18 insertions(+), 8 deletions(-)

diff --git a/include/linux/interrupt.h b/include/linux/interrupt.h
index 2306920..c3db4a0 100644
--- a/include/linux/interrupt.h
+++ b/include/linux/interrupt.h
@@ -256,6 +256,7 @@ enum
 #ifdef CONFIG_HIGH_RES_TIMERS
 	HRTIMER_SOFTIRQ,
 #endif
+	RCU_SOFTIRQ, 	/* Preferable RCU should always be the last softirq */
 };
 
 /* softirq mask and active fields moved to irq_cpustat_t in
diff --git a/kernel/rcupdate.c b/kernel/rcupdate.c
index a66d4d1..8658948 100644
--- a/kernel/rcupdate.c
+++ b/kernel/rcupdate.c
@@ -73,8 +73,6 @@ static struct rcu_ctrlblk rcu_bh_ctrlblk = {
 DEFINE_PER_CPU(struct rcu_data, rcu_data) = { 0L };
 DEFINE_PER_CPU(struct rcu_data, rcu_bh_data) = { 0L };
 
-/* Fake initialization required by compiler */
-static DEFINE_PER_CPU(struct tasklet_struct, rcu_tasklet) = {NULL};
 static int blimit = 10;
 static int qhimark = 10000;
 static int qlowmark = 100;
@@ -231,6 +229,18 @@ void rcu_barrier(void)
 }
 EXPORT_SYMBOL_GPL(rcu_barrier);
 
+/* Raises the softirq for processing rcu_callbacks. */
+static inline void raise_rcu_softirq(void)
+{
+	raise_softirq(RCU_SOFTIRQ);
+	/*
+	 * The smp_mb() here is required to ensure that this cpu's
+	 * __rcu_process_callbacks() reads the most recently updated
+	 * value of rcu->cur.
+	 */
+	smp_mb();
+}
+
 /*
  * Invoke the completed RCU callbacks. They are expected to be in
  * a per-cpu list.
@@ -260,7 +270,7 @@ static void rcu_do_batch(struct rcu_data *rdp)
 	if (!rdp->donelist)
 		rdp->donetail = &rdp->donelist;
 	else
-		tasklet_schedule(&per_cpu(rcu_tasklet, rdp->cpu));
+		raise_rcu_softirq();
 }
 
 /*
@@ -412,7 +422,6 @@ static void rcu_offline_cpu(int cpu)
 					&per_cpu(rcu_bh_data, cpu));
 	put_cpu_var(rcu_data);
 	put_cpu_var(rcu_bh_data);
-	tasklet_kill_immediate(&per_cpu(rcu_tasklet, cpu), cpu);
 }
 
 #else
@@ -424,7 +433,7 @@ static void rcu_offline_cpu(int cpu)
 #endif
 
 /*
- * This does the RCU processing work from tasklet context. 
+ * This does the RCU processing work from softirq context.
  */
 static void __rcu_process_callbacks(struct rcu_ctrlblk *rcp,
 					struct rcu_data *rdp)
@@ -469,7 +478,7 @@ static void __rcu_process_callbacks(struct rcu_ctrlblk *rcp,
 		rcu_do_batch(rdp);
 }
 
-static void rcu_process_callbacks(unsigned long unused)
+static void rcu_process_callbacks(struct softirq_action *unused)
 {
 	__rcu_process_callbacks(&rcu_ctrlblk, &__get_cpu_var(rcu_data));
 	__rcu_process_callbacks(&rcu_bh_ctrlblk, &__get_cpu_var(rcu_bh_data));
@@ -533,7 +542,7 @@ void rcu_check_callbacks(int cpu, int user)
 		rcu_bh_qsctr_inc(cpu);
 	} else if (!in_softirq())
 		rcu_bh_qsctr_inc(cpu);
-	tasklet_schedule(&per_cpu(rcu_tasklet, cpu));
+	raise_rcu_softirq();
 }
 
 static void rcu_init_percpu_data(int cpu, struct rcu_ctrlblk *rcp,
@@ -556,7 +565,7 @@ static void __devinit rcu_online_cpu(int cpu)
 
 	rcu_init_percpu_data(cpu, &rcu_ctrlblk, rdp);
 	rcu_init_percpu_data(cpu, &rcu_bh_ctrlblk, bh_rdp);
-	tasklet_init(&per_cpu(rcu_tasklet, cpu), rcu_process_callbacks, 0UL);
+	open_softirq(RCU_SOFTIRQ, rcu_process_callbacks, NULL);
 }
 
 static int __cpuinit rcu_cpu_notify(struct notifier_block *self,
-- 
Gautham R Shenoy
Linux Technology Center
IBM India.
"Freedom comes with a price tag of responsibility, which is still a bargain,
because Freedom is priceless!"

[RFC PATCH 2/6] Preempt-RCU: Reorganize RCU code into rcuclassic.c and rcupdate.c

[Gautham R Shenoy]

Preempt-RCU: Reorganize RCU code into rcuclassic.c and rcupdate.c

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

This patch re-organizes the RCU code to enable multiple implementations
of RCU. Users of RCU continues to include rcupdate.h and the
RCU interfaces remain the same. This is in preparation for
subsequently merging the preemptible RCU implementation.

Signed-off-by: Gautham R Shenoy <ego@in.ibm.com>
Signed-off-by: Dipankar Sarma <dipankar@in.ibm.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
---

 include/linux/rcuclassic.h |  161 ++++++++++++
 include/linux/rcupdate.h   |  168 ++++---------
 kernel/Makefile            |    2 
 kernel/rcuclassic.c        |  576 ++++++++++++++++++++++++++++++++++++++++++++
 kernel/rcupdate.c          |  575 ++------------------------------------------
 5 files changed, 812 insertions(+), 670 deletions(-)

diff --git a/include/linux/rcuclassic.h b/include/linux/rcuclassic.h
new file mode 100644
index 0000000..2b8b045
--- /dev/null
+++ b/include/linux/rcuclassic.h
@@ -0,0 +1,161 @@
+/*
+ * Read-Copy Update mechanism for mutual exclusion (classic version)
+ *
+ * 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, write to the Free Software
+ * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
+ *
+ * Copyright IBM Corporation, 2001
+ *
+ * Author: Dipankar Sarma <dipankar@in.ibm.com>
+ *
+ * Based on the original work by Paul McKenney <paulmck@us.ibm.com>
+ * and inputs from Rusty Russell, Andrea Arcangeli and Andi Kleen.
+ * Papers:
+ * http://www.rdrop.com/users/paulmck/paper/rclockpdcsproof.pdf
+ * http://lse.sourceforge.net/locking/rclock_OLS.2001.05.01c.sc.pdf (OLS2001)
+ *
+ * For detailed explanation of Read-Copy Update mechanism see -
+ * 		Documentation/RCU
+ *
+ */
+
+#ifndef __LINUX_RCUCLASSIC_H
+#define __LINUX_RCUCLASSIC_H
+
+#ifdef __KERNEL__
+
+#include <linux/cache.h>
+#include <linux/spinlock.h>
+#include <linux/threads.h>
+#include <linux/percpu.h>
+#include <linux/cpumask.h>
+#include <linux/seqlock.h>
+
+
+/* Global control variables for rcupdate callback mechanism. */
+struct rcu_ctrlblk {
+	long	cur;		/* Current batch number.                      */
+	long	completed;	/* Number of the last completed batch         */
+	int	next_pending;	/* Is the next batch already waiting?         */
+
+	int	signaled;
+
+	spinlock_t	lock	____cacheline_internodealigned_in_smp;
+	cpumask_t	cpumask; /* CPUs that need to switch in order    */
+				 /* for current batch to proceed.        */
+} ____cacheline_internodealigned_in_smp;
+
+/* Is batch a before batch b ? */
+static inline int rcu_batch_before(long a, long b)
+{
+	return (a - b) < 0;
+}
+
+/* Is batch a after batch b ? */
+static inline int rcu_batch_after(long a, long b)
+{
+	return (a - b) > 0;
+}
+
+/*
+ * Per-CPU data for Read-Copy UPdate.
+ * nxtlist - new callbacks are added here
+ * curlist - current batch for which quiescent cycle started if any
+ */
+struct rcu_data {
+	/* 1) quiescent state handling : */
+	long		quiescbatch;     /* Batch # for grace period */
+	int		passed_quiesc;	 /* User-mode/idle loop etc. */
+	int		qs_pending;	 /* core waits for quiesc state */
+
+	/* 2) batch handling */
+	long  	       	batch;           /* Batch # for current RCU batch */
+	struct rcu_head *nxtlist;
+	struct rcu_head **nxttail;
+	long            qlen; 	 	 /* # of queued callbacks */
+	struct rcu_head *curlist;
+	struct rcu_head **curtail;
+	struct rcu_head *donelist;
+	struct rcu_head **donetail;
+	long		blimit;		 /* Upper limit on a processed batch */
+	int cpu;
+	struct rcu_head barrier;
+};
+
+DECLARE_PER_CPU(struct rcu_data, rcu_data);
+DECLARE_PER_CPU(struct rcu_data, rcu_bh_data);
+
+/*
+ * Increment the quiescent state counter.
+ * The counter is a bit degenerated: We do not need to know
+ * how many quiescent states passed, just if there was at least
+ * one since the start of the grace period. Thus just a flag.
+ */
+static inline void rcu_qsctr_inc(int cpu)
+{
+	struct rcu_data *rdp = &per_cpu(rcu_data, cpu);
+	rdp->passed_quiesc = 1;
+}
+static inline void rcu_bh_qsctr_inc(int cpu)
+{
+	struct rcu_data *rdp = &per_cpu(rcu_bh_data, cpu);
+	rdp->passed_quiesc = 1;
+}
+
+extern int rcu_pending(int cpu);
+extern int rcu_needs_cpu(int cpu);
+
+#ifdef CONFIG_DEBUG_LOCK_ALLOC
+extern struct lockdep_map rcu_lock_map;
+# define rcu_read_acquire()	\
+			lock_acquire(&rcu_lock_map, 0, 0, 2, 1, _THIS_IP_)
+# define rcu_read_release()	lock_release(&rcu_lock_map, 1, _THIS_IP_)
+#else
+# define rcu_read_acquire()	do { } while (0)
+# define rcu_read_release()	do { } while (0)
+#endif
+
+#define __rcu_read_lock() \
+	do { \
+		preempt_disable(); \
+		__acquire(RCU); \
+		rcu_read_acquire(); \
+	} while (0)
+#define __rcu_read_unlock() \
+	do { \
+		rcu_read_release(); \
+		__release(RCU); \
+		preempt_enable(); \
+	} while (0)
+#define __rcu_read_lock_bh() \
+	do { \
+		local_bh_disable(); \
+		__acquire(RCU_BH); \
+		rcu_read_acquire(); \
+	} while (0)
+#define __rcu_read_unlock_bh() \
+	do { \
+		rcu_read_release(); \
+		__release(RCU_BH); \
+		local_bh_enable(); \
+	} while (0)
+
+#define __synchronize_sched() synchronize_rcu()
+
+extern void __rcu_init(void);
+extern void rcu_check_callbacks(int cpu, int user);
+extern void rcu_restart_cpu(int cpu);
+
+#endif /* __KERNEL__ */
+#endif /* __LINUX_RCUCLASSIC_H */
diff --git a/include/linux/rcupdate.h b/include/linux/rcupdate.h
index cc24a01..12aa13e 100644
--- a/include/linux/rcupdate.h
+++ b/include/linux/rcupdate.h
@@ -15,7 +15,7 @@
  * along with this program; if not, write to the Free Software
  * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
  *
- * Copyright (C) IBM Corporation, 2001
+ * Copyright IBM Corporation, 2001
  *
  * Author: Dipankar Sarma <dipankar@in.ibm.com>
  * 
@@ -53,96 +53,14 @@ struct rcu_head {
 	void (*func)(struct rcu_head *head);
 };
 
+#include <linux/rcuclassic.h>
+
 #define RCU_HEAD_INIT 	{ .next = NULL, .func = NULL }
 #define RCU_HEAD(head) struct rcu_head head = RCU_HEAD_INIT
 #define INIT_RCU_HEAD(ptr) do { \
        (ptr)->next = NULL; (ptr)->func = NULL; \
 } while (0)
 
-
-
-/* Global control variables for rcupdate callback mechanism. */
-struct rcu_ctrlblk {
-	long	cur;		/* Current batch number.                      */
-	long	completed;	/* Number of the last completed batch         */
-	int	next_pending;	/* Is the next batch already waiting?         */
-
-	int	signaled;
-
-	spinlock_t	lock	____cacheline_internodealigned_in_smp;
-	cpumask_t	cpumask; /* CPUs that need to switch in order    */
-	                         /* for current batch to proceed.        */
-} ____cacheline_internodealigned_in_smp;
-
-/* Is batch a before batch b ? */
-static inline int rcu_batch_before(long a, long b)
-{
-        return (a - b) < 0;
-}
-
-/* Is batch a after batch b ? */
-static inline int rcu_batch_after(long a, long b)
-{
-        return (a - b) > 0;
-}
-
-/*
- * Per-CPU data for Read-Copy UPdate.
- * nxtlist - new callbacks are added here
- * curlist - current batch for which quiescent cycle started if any
- */
-struct rcu_data {
-	/* 1) quiescent state handling : */
-	long		quiescbatch;     /* Batch # for grace period */
-	int		passed_quiesc;	 /* User-mode/idle loop etc. */
-	int		qs_pending;	 /* core waits for quiesc state */
-
-	/* 2) batch handling */
-	long  	       	batch;           /* Batch # for current RCU batch */
-	struct rcu_head *nxtlist;
-	struct rcu_head **nxttail;
-	long            qlen; 	 	 /* # of queued callbacks */
-	struct rcu_head *curlist;
-	struct rcu_head **curtail;
-	struct rcu_head *donelist;
-	struct rcu_head **donetail;
-	long		blimit;		 /* Upper limit on a processed batch */
-	int cpu;
-	struct rcu_head barrier;
-};
-
-DECLARE_PER_CPU(struct rcu_data, rcu_data);
-DECLARE_PER_CPU(struct rcu_data, rcu_bh_data);
-
-/*
- * Increment the quiescent state counter.
- * The counter is a bit degenerated: We do not need to know
- * how many quiescent states passed, just if there was at least
- * one since the start of the grace period. Thus just a flag.
- */
-static inline void rcu_qsctr_inc(int cpu)
-{
-	struct rcu_data *rdp = &per_cpu(rcu_data, cpu);
-	rdp->passed_quiesc = 1;
-}
-static inline void rcu_bh_qsctr_inc(int cpu)
-{
-	struct rcu_data *rdp = &per_cpu(rcu_bh_data, cpu);
-	rdp->passed_quiesc = 1;
-}
-
-extern int rcu_pending(int cpu);
-extern int rcu_needs_cpu(int cpu);
-
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
-extern struct lockdep_map rcu_lock_map;
-# define rcu_read_acquire()	lock_acquire(&rcu_lock_map, 0, 0, 2, 1, _THIS_IP_)
-# define rcu_read_release()	lock_release(&rcu_lock_map, 1, _THIS_IP_)
-#else
-# define rcu_read_acquire()	do { } while (0)
-# define rcu_read_release()	do { } while (0)
-#endif
-
 /**
  * rcu_read_lock - mark the beginning of an RCU read-side critical section.
  *
@@ -172,24 +90,13 @@ extern struct lockdep_map rcu_lock_map;
  *
  * It is illegal to block while in an RCU read-side critical section.
  */
-#define rcu_read_lock() \
-	do { \
-		preempt_disable(); \
-		__acquire(RCU); \
-		rcu_read_acquire(); \
-	} while(0)
+#define rcu_read_lock() __rcu_read_lock()
 
 /**
  * rcu_read_unlock - marks the end of an RCU read-side critical section.
  *
  * See rcu_read_lock() for more information.
  */
-#define rcu_read_unlock() \
-	do { \
-		rcu_read_release(); \
-		__release(RCU); \
-		preempt_enable(); \
-	} while(0)
 
 /*
  * So where is rcu_write_lock()?  It does not exist, as there is no
@@ -200,6 +107,7 @@ extern struct lockdep_map rcu_lock_map;
  * used as well.  RCU does not care how the writers keep out of each
  * others' way, as long as they do so.
  */
+#define rcu_read_unlock() __rcu_read_unlock()
 
 /**
  * rcu_read_lock_bh - mark the beginning of a softirq-only RCU critical section
@@ -212,24 +120,14 @@ extern struct lockdep_map rcu_lock_map;
  * can use just rcu_read_lock().
  *
  */
-#define rcu_read_lock_bh() \
-	do { \
-		local_bh_disable(); \
-		__acquire(RCU_BH); \
-		rcu_read_acquire(); \
-	} while(0)
+#define rcu_read_lock_bh() __rcu_read_lock_bh()
 
 /*
  * rcu_read_unlock_bh - marks the end of a softirq-only RCU critical section
  *
  * See rcu_read_lock_bh() for more information.
  */
-#define rcu_read_unlock_bh() \
-	do { \
-		rcu_read_release(); \
-		__release(RCU_BH); \
-		local_bh_enable(); \
-	} while(0)
+#define rcu_read_unlock_bh() __rcu_read_unlock_bh()
 
 /*
  * Prevent the compiler from merging or refetching accesses.  The compiler
@@ -293,21 +191,53 @@ extern struct lockdep_map rcu_lock_map;
  * In "classic RCU", these two guarantees happen to be one and
  * the same, but can differ in realtime RCU implementations.
  */
-#define synchronize_sched() synchronize_rcu()
+#define synchronize_sched() __synchronize_sched()
+
+/**
+ * call_rcu - Queue an RCU callback for invocation after a grace period.
+ * @head: structure to be used for queueing the RCU updates.
+ * @func: actual update function to be invoked after the grace period
+ *
+ * The update function will be invoked some time after a full grace
+ * period elapses, in other words after all currently executing RCU
+ * read-side critical sections have completed.  RCU read-side critical
+ * sections are delimited by rcu_read_lock() and rcu_read_unlock(),
+ * and may be nested.
+ */
+extern void call_rcu(struct rcu_head *head,
+			      void (*func)(struct rcu_head *head));
+
+/**
+ * call_rcu_bh - Queue an RCU for invocation after a quicker grace period.
+ * @head: structure to be used for queueing the RCU updates.
+ * @func: actual update function to be invoked after the grace period
+ *
+ * The update function will be invoked some time after a full grace
+ * period elapses, in other words after all currently executing RCU
+ * read-side critical sections have completed. call_rcu_bh() assumes
+ * that the read-side critical sections end on completion of a softirq
+ * handler. This means that read-side critical sections in process
+ * context must not be interrupted by softirqs. This interface is to be
+ * used when most of the read-side critical sections are in softirq context.
+ * RCU read-side critical sections are delimited by :
+ *  - rcu_read_lock() and  rcu_read_unlock(), if in interrupt context.
+ *  OR
+ *  - rcu_read_lock_bh() and rcu_read_unlock_bh(), if in process context.
+ *  These may be nested.
+ */
+extern void call_rcu_bh(struct rcu_head *head,
+			void (*func)(struct rcu_head *head));
+
+/* Exported common interfaces */
+extern void synchronize_rcu(void);
+extern void rcu_barrier(void);
 
+/* Internal to kernel */
 extern void rcu_init(void);
 extern void rcu_check_callbacks(int cpu, int user);
-extern void rcu_restart_cpu(int cpu);
+
 extern long rcu_batches_completed(void);
 extern long rcu_batches_completed_bh(void);
 
-/* Exported interfaces */
-extern void FASTCALL(call_rcu(struct rcu_head *head, 
-				void (*func)(struct rcu_head *head)));
-extern void FASTCALL(call_rcu_bh(struct rcu_head *head,
-				void (*func)(struct rcu_head *head)));
-extern void synchronize_rcu(void);
-extern void rcu_barrier(void);
-
 #endif /* __KERNEL__ */
 #endif /* __LINUX_RCUPDATE_H */
diff --git a/kernel/Makefile b/kernel/Makefile
index dfa9695..def5dd6 100644
--- a/kernel/Makefile
+++ b/kernel/Makefile
@@ -6,7 +6,7 @@ obj-y     = sched.o fork.o exec_domain.o panic.o printk.o profile.o \
 	    exit.o itimer.o time.o softirq.o resource.o \
 	    sysctl.o capability.o ptrace.o timer.o user.o user_namespace.o \
 	    signal.o sys.o kmod.o workqueue.o pid.o \
-	    rcupdate.o extable.o params.o posix-timers.o \
+	    rcupdate.o rcuclassic.o extable.o params.o posix-timers.o \
 	    kthread.o wait.o kfifo.o sys_ni.o posix-cpu-timers.o mutex.o \
 	    hrtimer.o rwsem.o latency.o nsproxy.o srcu.o \
 	    utsname.o notifier.o
diff --git a/kernel/rcuclassic.c b/kernel/rcuclassic.c
new file mode 100644
index 0000000..11c16aa
--- /dev/null
+++ b/kernel/rcuclassic.c
@@ -0,0 +1,576 @@
+/*
+ * Read-Copy Update mechanism for mutual exclusion
+ *
+ * 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, write to the Free Software
+ * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
+ *
+ * Copyright IBM Corporation, 2001
+ *
+ * Authors: Dipankar Sarma <dipankar@in.ibm.com>
+ *	    Manfred Spraul <manfred@colorfullife.com>
+ *
+ * Based on the original work by Paul McKenney <paulmck@us.ibm.com>
+ * and inputs from Rusty Russell, Andrea Arcangeli and Andi Kleen.
+ * Papers:
+ * http://www.rdrop.com/users/paulmck/paper/rclockpdcsproof.pdf
+ * http://lse.sourceforge.net/locking/rclock_OLS.2001.05.01c.sc.pdf (OLS2001)
+ *
+ * For detailed explanation of Read-Copy Update mechanism see -
+ * 		Documentation/RCU
+ *
+ */
+#include <linux/types.h>
+#include <linux/kernel.h>
+#include <linux/init.h>
+#include <linux/spinlock.h>
+#include <linux/smp.h>
+#include <linux/rcupdate.h>
+#include <linux/interrupt.h>
+#include <linux/sched.h>
+#include <asm/atomic.h>
+#include <linux/bitops.h>
+#include <linux/module.h>
+#include <linux/completion.h>
+#include <linux/moduleparam.h>
+#include <linux/percpu.h>
+#include <linux/notifier.h>
+/* #include <linux/rcupdate.h> @@@ */
+#include <linux/cpu.h>
+#include <linux/mutex.h>
+
+#ifdef CONFIG_DEBUG_LOCK_ALLOC
+static struct lock_class_key rcu_lock_key;
+struct lockdep_map rcu_lock_map =
+	STATIC_LOCKDEP_MAP_INIT("rcu_read_lock", &rcu_lock_key);
+EXPORT_SYMBOL_GPL(rcu_lock_map);
+#endif
+
+
+/* Definition for rcupdate control block. */
+static struct rcu_ctrlblk rcu_ctrlblk = {
+	.cur = -300,
+	.completed = -300,
+	.lock = __SPIN_LOCK_UNLOCKED(&rcu_ctrlblk.lock),
+	.cpumask = CPU_MASK_NONE,
+};
+static struct rcu_ctrlblk rcu_bh_ctrlblk = {
+	.cur = -300,
+	.completed = -300,
+	.lock = __SPIN_LOCK_UNLOCKED(&rcu_bh_ctrlblk.lock),
+	.cpumask = CPU_MASK_NONE,
+};
+
+DEFINE_PER_CPU(struct rcu_data, rcu_data) = { 0L };
+DEFINE_PER_CPU(struct rcu_data, rcu_bh_data) = { 0L };
+
+static int blimit = 10;
+static int qhimark = 10000;
+static int qlowmark = 100;
+
+#ifdef CONFIG_SMP
+static void force_quiescent_state(struct rcu_data *rdp,
+			struct rcu_ctrlblk *rcp)
+{
+	int cpu;
+	cpumask_t cpumask;
+	set_need_resched();
+	if (unlikely(!rcp->signaled)) {
+		rcp->signaled = 1;
+		/*
+		 * Don't send IPI to itself. With irqs disabled,
+		 * rdp->cpu is the current cpu.
+		 */
+		cpumask = rcp->cpumask;
+		cpu_clear(rdp->cpu, cpumask);
+		for_each_cpu_mask(cpu, cpumask)
+			smp_send_reschedule(cpu);
+	}
+}
+#else
+static inline void force_quiescent_state(struct rcu_data *rdp,
+			struct rcu_ctrlblk *rcp)
+{
+	set_need_resched();
+}
+#endif
+
+/**
+ * call_rcu - Queue an RCU callback for invocation after a grace period.
+ * @head: structure to be used for queueing the RCU updates.
+ * @func: actual update function to be invoked after the grace period
+ *
+ * The update function will be invoked some time after a full grace
+ * period elapses, in other words after all currently executing RCU
+ * read-side critical sections have completed.  RCU read-side critical
+ * sections are delimited by rcu_read_lock() and rcu_read_unlock(),
+ * and may be nested.
+ */
+void call_rcu(struct rcu_head *head,
+				void (*func)(struct rcu_head *rcu))
+{
+	unsigned long flags;
+	struct rcu_data *rdp;
+
+	head->func = func;
+	head->next = NULL;
+	local_irq_save(flags);
+	rdp = &__get_cpu_var(rcu_data);
+	*rdp->nxttail = head;
+	rdp->nxttail = &head->next;
+	if (unlikely(++rdp->qlen > qhimark)) {
+		rdp->blimit = INT_MAX;
+		force_quiescent_state(rdp, &rcu_ctrlblk);
+	}
+	local_irq_restore(flags);
+}
+EXPORT_SYMBOL_GPL(call_rcu);
+
+/**
+ * call_rcu_bh - Queue an RCU for invocation after a quicker grace period.
+ * @head: structure to be used for queueing the RCU updates.
+ * @func: actual update function to be invoked after the grace period
+ *
+ * The update function will be invoked some time after a full grace
+ * period elapses, in other words after all currently executing RCU
+ * read-side critical sections have completed. call_rcu_bh() assumes
+ * that the read-side critical sections end on completion of a softirq
+ * handler. This means that read-side critical sections in process
+ * context must not be interrupted by softirqs. This interface is to be
+ * used when most of the read-side critical sections are in softirq context.
+ * RCU read-side critical sections are delimited by rcu_read_lock() and
+ * rcu_read_unlock(), * if in interrupt context or rcu_read_lock_bh()
+ * and rcu_read_unlock_bh(), if in process context. These may be nested.
+ */
+void call_rcu_bh(struct rcu_head *head,
+				void (*func)(struct rcu_head *rcu))
+{
+	unsigned long flags;
+	struct rcu_data *rdp;
+
+	head->func = func;
+	head->next = NULL;
+	local_irq_save(flags);
+	rdp = &__get_cpu_var(rcu_bh_data);
+	*rdp->nxttail = head;
+	rdp->nxttail = &head->next;
+
+	if (unlikely(++rdp->qlen > qhimark)) {
+		rdp->blimit = INT_MAX;
+		force_quiescent_state(rdp, &rcu_bh_ctrlblk);
+	}
+
+	local_irq_restore(flags);
+}
+EXPORT_SYMBOL_GPL(call_rcu_bh);
+
+/*
+ * Return the number of RCU batches processed thus far.  Useful
+ * for debug and statistics.
+ */
+long rcu_batches_completed(void)
+{
+	return rcu_ctrlblk.completed;
+}
+EXPORT_SYMBOL_GPL(rcu_batches_completed);
+
+/*
+ * Return the number of RCU batches processed thus far.  Useful
+ * for debug and statistics.
+ */
+long rcu_batches_completed_bh(void)
+{
+	return rcu_bh_ctrlblk.completed;
+}
+EXPORT_SYMBOL_GPL(rcu_batches_completed_bh);
+
+/* Raises the softirq for processing rcu_callbacks. */
+static inline void raise_rcu_softirq(void)
+{
+	raise_softirq(RCU_SOFTIRQ);
+	/*
+	 * The smp_mb() here is required to ensure that this cpu's
+	 * __rcu_process_callbacks() reads the most recently updated
+	 * value of rcu->cur.
+	 */
+	smp_mb();
+}
+
+/*
+ * Invoke the completed RCU callbacks. They are expected to be in
+ * a per-cpu list.
+ */
+static void rcu_do_batch(struct rcu_data *rdp)
+{
+	struct rcu_head *next, *list;
+	int count = 0;
+
+	list = rdp->donelist;
+	while (list) {
+		next = list->next;
+		prefetch(next);
+		list->func(list);
+		list = next;
+		if (++count >= rdp->blimit)
+			break;
+	}
+	rdp->donelist = list;
+
+	local_irq_disable();
+	rdp->qlen -= count;
+	local_irq_enable();
+	if (rdp->blimit == INT_MAX && rdp->qlen <= qlowmark)
+		rdp->blimit = blimit;
+
+	if (!rdp->donelist)
+		rdp->donetail = &rdp->donelist;
+	else
+		raise_rcu_softirq();
+}
+
+/*
+ * Grace period handling:
+ * The grace period handling consists out of two steps:
+ * - A new grace period is started.
+ *   This is done by rcu_start_batch. The start is not broadcasted to
+ *   all cpus, they must pick this up by comparing rcp->cur with
+ *   rdp->quiescbatch. All cpus are recorded  in the
+ *   rcu_ctrlblk.cpumask bitmap.
+ * - All cpus must go through a quiescent state.
+ *   Since the start of the grace period is not broadcasted, at least two
+ *   calls to rcu_check_quiescent_state are required:
+ *   The first call just notices that a new grace period is running. The
+ *   following calls check if there was a quiescent state since the beginning
+ *   of the grace period. If so, it updates rcu_ctrlblk.cpumask. If
+ *   the bitmap is empty, then the grace period is completed.
+ *   rcu_check_quiescent_state calls rcu_start_batch(0) to start the next grace
+ *   period (if necessary).
+ */
+/*
+ * Register a new batch of callbacks, and start it up if there is currently no
+ * active batch and the batch to be registered has not already occurred.
+ * Caller must hold rcu_ctrlblk.lock.
+ */
+static void rcu_start_batch(struct rcu_ctrlblk *rcp)
+{
+	if (rcp->next_pending &&
+			rcp->completed == rcp->cur) {
+		rcp->next_pending = 0;
+		/*
+		 * next_pending == 0 must be visible in
+		 * __rcu_process_callbacks() before it can see new value of cur.
+		 */
+		smp_wmb();
+		rcp->cur++;
+
+		/*
+		 * Accessing nohz_cpu_mask before incrementing rcp->cur needs a
+		 * Barrier  Otherwise it can cause tickless idle CPUs to be
+		 * included in rcp->cpumask, which will extend graceperiods
+		 * unnecessarily.
+		 */
+		smp_mb();
+		cpus_andnot(rcp->cpumask, cpu_online_map, nohz_cpu_mask);
+
+		rcp->signaled = 0;
+	}
+}
+
+/*
+ * cpu went through a quiescent state since the beginning of the grace period.
+ * Clear it from the cpu mask and complete the grace period if it was the last
+ * cpu. Start another grace period if someone has further entries pending
+ */
+static void cpu_quiet(int cpu, struct rcu_ctrlblk *rcp)
+{
+	cpu_clear(cpu, rcp->cpumask);
+	if (cpus_empty(rcp->cpumask)) {
+		/* batch completed ! */
+		rcp->completed = rcp->cur;
+		rcu_start_batch(rcp);
+	}
+}
+
+/*
+ * Check if the cpu has gone through a quiescent state (say context
+ * switch). If so and if it already hasn't done so in this RCU
+ * quiescent cycle, then indicate that it has done so.
+ */
+static void rcu_check_quiescent_state(struct rcu_ctrlblk *rcp,
+					struct rcu_data *rdp)
+{
+	if (rdp->quiescbatch != rcp->cur) {
+		/* start new grace period: */
+		rdp->qs_pending = 1;
+		rdp->passed_quiesc = 0;
+		rdp->quiescbatch = rcp->cur;
+		return;
+	}
+
+	/* Grace period already completed for this cpu?
+	 * qs_pending is checked instead of the actual bitmap to avoid
+	 * cacheline trashing.
+	 */
+	if (!rdp->qs_pending)
+		return;
+
+	/*
+	 * Was there a quiescent state since the beginning of the grace
+	 * period? If no, then exit and wait for the next call.
+	 */
+	if (!rdp->passed_quiesc)
+		return;
+	rdp->qs_pending = 0;
+
+	spin_lock(&rcp->lock);
+	/*
+	 * rdp->quiescbatch/rcp->cur and the cpu bitmap can come out of sync
+	 * during cpu startup. Ignore the quiescent state.
+	 */
+	if (likely(rdp->quiescbatch == rcp->cur))
+		cpu_quiet(rdp->cpu, rcp);
+
+	spin_unlock(&rcp->lock);
+}
+
+
+#ifdef CONFIG_HOTPLUG_CPU
+
+/* warning! helper for rcu_offline_cpu. do not use elsewhere without reviewing
+ * locking requirements, the list it's pulling from has to belong to a cpu
+ * which is dead and hence not processing interrupts.
+ */
+static void rcu_move_batch(struct rcu_data *this_rdp, struct rcu_head *list,
+				struct rcu_head **tail)
+{
+	local_irq_disable();
+	*this_rdp->nxttail = list;
+	if (list)
+		this_rdp->nxttail = tail;
+	local_irq_enable();
+}
+
+static void __rcu_offline_cpu(struct rcu_data *this_rdp,
+				struct rcu_ctrlblk *rcp, struct rcu_data *rdp)
+{
+	/* if the cpu going offline owns the grace period
+	 * we can block indefinitely waiting for it, so flush
+	 * it here
+	 */
+	spin_lock_bh(&rcp->lock);
+	if (rcp->cur != rcp->completed)
+		cpu_quiet(rdp->cpu, rcp);
+	spin_unlock_bh(&rcp->lock);
+	rcu_move_batch(this_rdp, rdp->curlist, rdp->curtail);
+	rcu_move_batch(this_rdp, rdp->nxtlist, rdp->nxttail);
+	rcu_move_batch(this_rdp, rdp->donelist, rdp->donetail);
+}
+
+static void rcu_offline_cpu(int cpu)
+{
+	struct rcu_data *this_rdp = &get_cpu_var(rcu_data);
+	struct rcu_data *this_bh_rdp = &get_cpu_var(rcu_bh_data);
+
+	__rcu_offline_cpu(this_rdp, &rcu_ctrlblk,
+					&per_cpu(rcu_data, cpu));
+	__rcu_offline_cpu(this_bh_rdp, &rcu_bh_ctrlblk,
+					&per_cpu(rcu_bh_data, cpu));
+	put_cpu_var(rcu_data);
+	put_cpu_var(rcu_bh_data);
+}
+
+#else
+
+static void rcu_offline_cpu(int cpu)
+{
+}
+
+#endif
+
+/*
+ * This does the RCU processing work from softirq context.
+ */
+static void __rcu_process_callbacks(struct rcu_ctrlblk *rcp,
+					struct rcu_data *rdp)
+{
+	if (rdp->curlist && !rcu_batch_before(rcp->completed, rdp->batch)) {
+		*rdp->donetail = rdp->curlist;
+		rdp->donetail = rdp->curtail;
+		rdp->curlist = NULL;
+		rdp->curtail = &rdp->curlist;
+	}
+
+	if (rdp->nxtlist && !rdp->curlist) {
+		local_irq_disable();
+		rdp->curlist = rdp->nxtlist;
+		rdp->curtail = rdp->nxttail;
+		rdp->nxtlist = NULL;
+		rdp->nxttail = &rdp->nxtlist;
+		local_irq_enable();
+
+		/*
+		 * start the next batch of callbacks
+		 */
+
+		/* determine batch number */
+		rdp->batch = rcp->cur + 1;
+		/* see the comment and corresponding wmb() in
+		 * the rcu_start_batch()
+		 */
+		smp_rmb();
+
+		if (!rcp->next_pending) {
+			/* and start it/schedule start if it's a new batch */
+			spin_lock(&rcp->lock);
+			rcp->next_pending = 1;
+			rcu_start_batch(rcp);
+			spin_unlock(&rcp->lock);
+		}
+	}
+
+	rcu_check_quiescent_state(rcp, rdp);
+	if (rdp->donelist)
+		rcu_do_batch(rdp);
+}
+
+static void rcu_process_callbacks(struct softirq_action *unused)
+{
+	__rcu_process_callbacks(&rcu_ctrlblk, &__get_cpu_var(rcu_data));
+	__rcu_process_callbacks(&rcu_bh_ctrlblk, &__get_cpu_var(rcu_bh_data));
+}
+
+static int __rcu_pending(struct rcu_ctrlblk *rcp, struct rcu_data *rdp)
+{
+	/* This cpu has pending rcu entries and the grace period
+	 * for them has completed.
+	 */
+	if (rdp->curlist && !rcu_batch_before(rcp->completed, rdp->batch))
+		return 1;
+
+	/* This cpu has no pending entries, but there are new entries */
+	if (!rdp->curlist && rdp->nxtlist)
+		return 1;
+
+	/* This cpu has finished callbacks to invoke */
+	if (rdp->donelist)
+		return 1;
+
+	/* The rcu core waits for a quiescent state from the cpu */
+	if (rdp->quiescbatch != rcp->cur || rdp->qs_pending)
+		return 1;
+
+	/* nothing to do */
+	return 0;
+}
+
+/*
+ * Check to see if there is any immediate RCU-related work to be done
+ * by the current CPU, returning 1 if so.  This function is part of the
+ * RCU implementation; it is -not- an exported member of the RCU API.
+ */
+int rcu_pending(int cpu)
+{
+	return __rcu_pending(&rcu_ctrlblk, &per_cpu(rcu_data, cpu)) ||
+		__rcu_pending(&rcu_bh_ctrlblk, &per_cpu(rcu_bh_data, cpu));
+}
+
+/*
+ * Check to see if any future RCU-related work will need to be done
+ * by the current CPU, even if none need be done immediately, returning
+ * 1 if so.  This function is part of the RCU implementation; it is -not-
+ * an exported member of the RCU API.
+ */
+int rcu_needs_cpu(int cpu)
+{
+	struct rcu_data *rdp = &per_cpu(rcu_data, cpu);
+	struct rcu_data *rdp_bh = &per_cpu(rcu_bh_data, cpu);
+
+	return (!!rdp->curlist || !!rdp_bh->curlist || rcu_pending(cpu));
+}
+
+void rcu_check_callbacks(int cpu, int user)
+{
+	if (user ||
+	    (idle_cpu(cpu) && !in_softirq() &&
+				hardirq_count() <= (1 << HARDIRQ_SHIFT))) {
+		rcu_qsctr_inc(cpu);
+		rcu_bh_qsctr_inc(cpu);
+	} else if (!in_softirq())
+		rcu_bh_qsctr_inc(cpu);
+	raise_rcu_softirq();
+}
+
+static void rcu_init_percpu_data(int cpu, struct rcu_ctrlblk *rcp,
+						struct rcu_data *rdp)
+{
+	memset(rdp, 0, sizeof(*rdp));
+	rdp->curtail = &rdp->curlist;
+	rdp->nxttail = &rdp->nxtlist;
+	rdp->donetail = &rdp->donelist;
+	rdp->quiescbatch = rcp->completed;
+	rdp->qs_pending = 0;
+	rdp->cpu = cpu;
+	rdp->blimit = blimit;
+}
+
+static void __devinit rcu_online_cpu(int cpu)
+{
+	struct rcu_data *rdp = &per_cpu(rcu_data, cpu);
+	struct rcu_data *bh_rdp = &per_cpu(rcu_bh_data, cpu);
+
+	rcu_init_percpu_data(cpu, &rcu_ctrlblk, rdp);
+	rcu_init_percpu_data(cpu, &rcu_bh_ctrlblk, bh_rdp);
+	open_softirq(RCU_SOFTIRQ, rcu_process_callbacks, NULL);
+}
+
+static int __cpuinit rcu_cpu_notify(struct notifier_block *self,
+				unsigned long action, void *hcpu)
+{
+	long cpu = (long)hcpu;
+
+	switch (action) {
+	case CPU_UP_PREPARE:
+	case CPU_UP_PREPARE_FROZEN:
+		rcu_online_cpu(cpu);
+		break;
+	case CPU_DEAD:
+	case CPU_DEAD_FROZEN:
+		rcu_offline_cpu(cpu);
+		break;
+	default:
+		break;
+	}
+	return NOTIFY_OK;
+}
+
+static struct notifier_block __cpuinitdata rcu_nb = {
+	.notifier_call	= rcu_cpu_notify,
+};
+
+/*
+ * Initializes rcu mechanism.  Assumed to be called early.
+ * That is before local timer(SMP) or jiffie timer (uniproc) is setup.
+ * Note that rcu_qsctr and friends are implicitly
+ * initialized due to the choice of ``0'' for RCU_CTR_INVALID.
+ */
+void __init __rcu_init(void)
+{
+	rcu_cpu_notify(&rcu_nb, CPU_UP_PREPARE,
+			(void *)(long)smp_processor_id());
+	/* Register notifier for non-boot CPUs */
+	register_cpu_notifier(&rcu_nb);
+}
+
+module_param(blimit, int, 0);
+module_param(qhimark, int, 0);
+module_param(qlowmark, int, 0);
diff --git a/kernel/rcupdate.c b/kernel/rcupdate.c
index 8658948..0ccd009 100644
--- a/kernel/rcupdate.c
+++ b/kernel/rcupdate.c
@@ -15,7 +15,7 @@
  * along with this program; if not, write to the Free Software
  * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
  *
- * Copyright (C) IBM Corporation, 2001
+ * Copyright IBM Corporation, 2001
  *
  * Authors: Dipankar Sarma <dipankar@in.ibm.com>
  *	    Manfred Spraul <manfred@colorfullife.com>
@@ -35,163 +35,57 @@
 #include <linux/init.h>
 #include <linux/spinlock.h>
 #include <linux/smp.h>
-#include <linux/rcupdate.h>
 #include <linux/interrupt.h>
 #include <linux/sched.h>
 #include <asm/atomic.h>
 #include <linux/bitops.h>
-#include <linux/module.h>
 #include <linux/completion.h>
-#include <linux/moduleparam.h>
 #include <linux/percpu.h>
 #include <linux/notifier.h>
 #include <linux/cpu.h>
 #include <linux/mutex.h>
+#include <linux/module.h>
 
-#ifdef CONFIG_DEBUG_LOCK_ALLOC
-static struct lock_class_key rcu_lock_key;
-struct lockdep_map rcu_lock_map =
-	STATIC_LOCKDEP_MAP_INIT("rcu_read_lock", &rcu_lock_key);
-
-EXPORT_SYMBOL_GPL(rcu_lock_map);
-#endif
-
-/* Definition for rcupdate control block. */
-static struct rcu_ctrlblk rcu_ctrlblk = {
-	.cur = -300,
-	.completed = -300,
-	.lock = __SPIN_LOCK_UNLOCKED(&rcu_ctrlblk.lock),
-	.cpumask = CPU_MASK_NONE,
-};
-static struct rcu_ctrlblk rcu_bh_ctrlblk = {
-	.cur = -300,
-	.completed = -300,
-	.lock = __SPIN_LOCK_UNLOCKED(&rcu_bh_ctrlblk.lock),
-	.cpumask = CPU_MASK_NONE,
+struct rcu_synchronize {
+	struct rcu_head head;
+	struct completion completion;
 };
 
-DEFINE_PER_CPU(struct rcu_data, rcu_data) = { 0L };
-DEFINE_PER_CPU(struct rcu_data, rcu_bh_data) = { 0L };
-
-static int blimit = 10;
-static int qhimark = 10000;
-static int qlowmark = 100;
-
+static DEFINE_PER_CPU(struct rcu_head, rcu_barrier_head) = {NULL};
 static atomic_t rcu_barrier_cpu_count;
 static DEFINE_MUTEX(rcu_barrier_mutex);
 static struct completion rcu_barrier_completion;
 
-#ifdef CONFIG_SMP
-static void force_quiescent_state(struct rcu_data *rdp,
-			struct rcu_ctrlblk *rcp)
-{
-	int cpu;
-	cpumask_t cpumask;
-	set_need_resched();
-	if (unlikely(!rcp->signaled)) {
-		rcp->signaled = 1;
-		/*
-		 * Don't send IPI to itself. With irqs disabled,
-		 * rdp->cpu is the current cpu.
-		 */
-		cpumask = rcp->cpumask;
-		cpu_clear(rdp->cpu, cpumask);
-		for_each_cpu_mask(cpu, cpumask)
-			smp_send_reschedule(cpu);
-	}
-}
-#else
-static inline void force_quiescent_state(struct rcu_data *rdp,
-			struct rcu_ctrlblk *rcp)
+/* Because of FASTCALL declaration of complete, we use this wrapper */
+static void wakeme_after_rcu(struct rcu_head  *head)
 {
-	set_need_resched();
+	struct rcu_synchronize *rcu;
+
+	rcu = container_of(head, struct rcu_synchronize, head);
+	complete(&rcu->completion);
 }
-#endif
 
 /**
- * call_rcu - Queue an RCU callback for invocation after a grace period.
- * @head: structure to be used for queueing the RCU updates.
- * @func: actual update function to be invoked after the grace period
+ * synchronize_rcu - wait until a grace period has elapsed.
  *
- * The update function will be invoked some time after a full grace
- * period elapses, in other words after all currently executing RCU
+ * Control will return to the caller some time after a full grace
+ * period has elapsed, in other words after all currently executing RCU
  * read-side critical sections have completed.  RCU read-side critical
  * sections are delimited by rcu_read_lock() and rcu_read_unlock(),
  * and may be nested.
  */
-void fastcall call_rcu(struct rcu_head *head,
-				void (*func)(struct rcu_head *rcu))
-{
-	unsigned long flags;
-	struct rcu_data *rdp;
-
-	head->func = func;
-	head->next = NULL;
-	local_irq_save(flags);
-	rdp = &__get_cpu_var(rcu_data);
-	*rdp->nxttail = head;
-	rdp->nxttail = &head->next;
-	if (unlikely(++rdp->qlen > qhimark)) {
-		rdp->blimit = INT_MAX;
-		force_quiescent_state(rdp, &rcu_ctrlblk);
-	}
-	local_irq_restore(flags);
-}
-
-/**
- * call_rcu_bh - Queue an RCU for invocation after a quicker grace period.
- * @head: structure to be used for queueing the RCU updates.
- * @func: actual update function to be invoked after the grace period
- *
- * The update function will be invoked some time after a full grace
- * period elapses, in other words after all currently executing RCU
- * read-side critical sections have completed. call_rcu_bh() assumes
- * that the read-side critical sections end on completion of a softirq
- * handler. This means that read-side critical sections in process
- * context must not be interrupted by softirqs. This interface is to be
- * used when most of the read-side critical sections are in softirq context.
- * RCU read-side critical sections are delimited by rcu_read_lock() and
- * rcu_read_unlock(), * if in interrupt context or rcu_read_lock_bh()
- * and rcu_read_unlock_bh(), if in process context. These may be nested.
- */
-void fastcall call_rcu_bh(struct rcu_head *head,
-				void (*func)(struct rcu_head *rcu))
+void synchronize_rcu(void)
 {
-	unsigned long flags;
-	struct rcu_data *rdp;
-
-	head->func = func;
-	head->next = NULL;
-	local_irq_save(flags);
-	rdp = &__get_cpu_var(rcu_bh_data);
-	*rdp->nxttail = head;
-	rdp->nxttail = &head->next;
-
-	if (unlikely(++rdp->qlen > qhimark)) {
-		rdp->blimit = INT_MAX;
-		force_quiescent_state(rdp, &rcu_bh_ctrlblk);
-	}
-
-	local_irq_restore(flags);
-}
+	struct rcu_synchronize rcu;
 
-/*
- * Return the number of RCU batches processed thus far.  Useful
- * for debug and statistics.
- */
-long rcu_batches_completed(void)
-{
-	return rcu_ctrlblk.completed;
-}
+	init_completion(&rcu.completion);
+	/* Will wake me after RCU finished */
+	call_rcu(&rcu.head, wakeme_after_rcu);
 
-/*
- * Return the number of RCU batches processed thus far.  Useful
- * for debug and statistics.
- */
-long rcu_batches_completed_bh(void)
-{
-	return rcu_bh_ctrlblk.completed;
+	/* Wait for it */
+	wait_for_completion(&rcu.completion);
 }
+EXPORT_SYMBOL_GPL(synchronize_rcu);
 
 static void rcu_barrier_callback(struct rcu_head *notused)
 {
@@ -205,10 +99,8 @@ static void rcu_barrier_callback(struct rcu_head *notused)
 static void rcu_barrier_func(void *notused)
 {
 	int cpu = smp_processor_id();
-	struct rcu_data *rdp = &per_cpu(rcu_data, cpu);
-	struct rcu_head *head;
+	struct rcu_head *head = &per_cpu(rcu_barrier_head, cpu);
 
-	head = &rdp->barrier;
 	atomic_inc(&rcu_barrier_cpu_count);
 	call_rcu(head, rcu_barrier_callback);
 }
@@ -229,425 +121,8 @@ void rcu_barrier(void)
 }
 EXPORT_SYMBOL_GPL(rcu_barrier);
 
-/* Raises the softirq for processing rcu_callbacks. */
-static inline void raise_rcu_softirq(void)
-{
-	raise_softirq(RCU_SOFTIRQ);
-	/*
-	 * The smp_mb() here is required to ensure that this cpu's
-	 * __rcu_process_callbacks() reads the most recently updated
-	 * value of rcu->cur.
-	 */
-	smp_mb();
-}
-
-/*
- * Invoke the completed RCU callbacks. They are expected to be in
- * a per-cpu list.
- */
-static void rcu_do_batch(struct rcu_data *rdp)
-{
-	struct rcu_head *next, *list;
-	int count = 0;
-
-	list = rdp->donelist;
-	while (list) {
-		next = list->next;
-		prefetch(next);
-		list->func(list);
-		list = next;
-		if (++count >= rdp->blimit)
-			break;
-	}
-	rdp->donelist = list;
-
-	local_irq_disable();
-	rdp->qlen -= count;
-	local_irq_enable();
-	if (rdp->blimit == INT_MAX && rdp->qlen <= qlowmark)
-		rdp->blimit = blimit;
-
-	if (!rdp->donelist)
-		rdp->donetail = &rdp->donelist;
-	else
-		raise_rcu_softirq();
-}
-
-/*
- * Grace period handling:
- * The grace period handling consists out of two steps:
- * - A new grace period is started.
- *   This is done by rcu_start_batch. The start is not broadcasted to
- *   all cpus, they must pick this up by comparing rcp->cur with
- *   rdp->quiescbatch. All cpus are recorded  in the
- *   rcu_ctrlblk.cpumask bitmap.
- * - All cpus must go through a quiescent state.
- *   Since the start of the grace period is not broadcasted, at least two
- *   calls to rcu_check_quiescent_state are required:
- *   The first call just notices that a new grace period is running. The
- *   following calls check if there was a quiescent state since the beginning
- *   of the grace period. If so, it updates rcu_ctrlblk.cpumask. If
- *   the bitmap is empty, then the grace period is completed.
- *   rcu_check_quiescent_state calls rcu_start_batch(0) to start the next grace
- *   period (if necessary).
- */
-/*
- * Register a new batch of callbacks, and start it up if there is currently no
- * active batch and the batch to be registered has not already occurred.
- * Caller must hold rcu_ctrlblk.lock.
- */
-static void rcu_start_batch(struct rcu_ctrlblk *rcp)
-{
-	if (rcp->next_pending &&
-			rcp->completed == rcp->cur) {
-		rcp->next_pending = 0;
-		/*
-		 * next_pending == 0 must be visible in
-		 * __rcu_process_callbacks() before it can see new value of cur.
-		 */
-		smp_wmb();
-		rcp->cur++;
-
-		/*
-		 * Accessing nohz_cpu_mask before incrementing rcp->cur needs a
-		 * Barrier  Otherwise it can cause tickless idle CPUs to be
-		 * included in rcp->cpumask, which will extend graceperiods
-		 * unnecessarily.
-		 */
-		smp_mb();
-		cpus_andnot(rcp->cpumask, cpu_online_map, nohz_cpu_mask);
-
-		rcp->signaled = 0;
-	}
-}
-
-/*
- * cpu went through a quiescent state since the beginning of the grace period.
- * Clear it from the cpu mask and complete the grace period if it was the last
- * cpu. Start another grace period if someone has further entries pending
- */
-static void cpu_quiet(int cpu, struct rcu_ctrlblk *rcp)
-{
-	cpu_clear(cpu, rcp->cpumask);
-	if (cpus_empty(rcp->cpumask)) {
-		/* batch completed ! */
-		rcp->completed = rcp->cur;
-		rcu_start_batch(rcp);
-	}
-}
-
-/*
- * Check if the cpu has gone through a quiescent state (say context
- * switch). If so and if it already hasn't done so in this RCU
- * quiescent cycle, then indicate that it has done so.
- */
-static void rcu_check_quiescent_state(struct rcu_ctrlblk *rcp,
-					struct rcu_data *rdp)
-{
-	if (rdp->quiescbatch != rcp->cur) {
-		/* start new grace period: */
-		rdp->qs_pending = 1;
-		rdp->passed_quiesc = 0;
-		rdp->quiescbatch = rcp->cur;
-		return;
-	}
-
-	/* Grace period already completed for this cpu?
-	 * qs_pending is checked instead of the actual bitmap to avoid
-	 * cacheline trashing.
-	 */
-	if (!rdp->qs_pending)
-		return;
-
-	/* 
-	 * Was there a quiescent state since the beginning of the grace
-	 * period? If no, then exit and wait for the next call.
-	 */
-	if (!rdp->passed_quiesc)
-		return;
-	rdp->qs_pending = 0;
-
-	spin_lock(&rcp->lock);
-	/*
-	 * rdp->quiescbatch/rcp->cur and the cpu bitmap can come out of sync
-	 * during cpu startup. Ignore the quiescent state.
-	 */
-	if (likely(rdp->quiescbatch == rcp->cur))
-		cpu_quiet(rdp->cpu, rcp);
-
-	spin_unlock(&rcp->lock);
-}
-
-
-#ifdef CONFIG_HOTPLUG_CPU
-
-/* warning! helper for rcu_offline_cpu. do not use elsewhere without reviewing
- * locking requirements, the list it's pulling from has to belong to a cpu
- * which is dead and hence not processing interrupts.
- */
-static void rcu_move_batch(struct rcu_data *this_rdp, struct rcu_head *list,
-				struct rcu_head **tail)
-{
-	local_irq_disable();
-	*this_rdp->nxttail = list;
-	if (list)
-		this_rdp->nxttail = tail;
-	local_irq_enable();
-}
-
-static void __rcu_offline_cpu(struct rcu_data *this_rdp,
-				struct rcu_ctrlblk *rcp, struct rcu_data *rdp)
-{
-	/* if the cpu going offline owns the grace period
-	 * we can block indefinitely waiting for it, so flush
-	 * it here
-	 */
-	spin_lock_bh(&rcp->lock);
-	if (rcp->cur != rcp->completed)
-		cpu_quiet(rdp->cpu, rcp);
-	spin_unlock_bh(&rcp->lock);
-	rcu_move_batch(this_rdp, rdp->curlist, rdp->curtail);
-	rcu_move_batch(this_rdp, rdp->nxtlist, rdp->nxttail);
-	rcu_move_batch(this_rdp, rdp->donelist, rdp->donetail);
-}
-
-static void rcu_offline_cpu(int cpu)
-{
-	struct rcu_data *this_rdp = &get_cpu_var(rcu_data);
-	struct rcu_data *this_bh_rdp = &get_cpu_var(rcu_bh_data);
-
-	__rcu_offline_cpu(this_rdp, &rcu_ctrlblk,
-					&per_cpu(rcu_data, cpu));
-	__rcu_offline_cpu(this_bh_rdp, &rcu_bh_ctrlblk,
-					&per_cpu(rcu_bh_data, cpu));
-	put_cpu_var(rcu_data);
-	put_cpu_var(rcu_bh_data);
-}
-
-#else
-
-static void rcu_offline_cpu(int cpu)
-{
-}
-
-#endif
-
-/*
- * This does the RCU processing work from softirq context.
- */
-static void __rcu_process_callbacks(struct rcu_ctrlblk *rcp,
-					struct rcu_data *rdp)
-{
-	if (rdp->curlist && !rcu_batch_before(rcp->completed, rdp->batch)) {
-		*rdp->donetail = rdp->curlist;
-		rdp->donetail = rdp->curtail;
-		rdp->curlist = NULL;
-		rdp->curtail = &rdp->curlist;
-	}
-
-	if (rdp->nxtlist && !rdp->curlist) {
-		local_irq_disable();
-		rdp->curlist = rdp->nxtlist;
-		rdp->curtail = rdp->nxttail;
-		rdp->nxtlist = NULL;
-		rdp->nxttail = &rdp->nxtlist;
-		local_irq_enable();
-
-		/*
-		 * start the next batch of callbacks
-		 */
-
-		/* determine batch number */
-		rdp->batch = rcp->cur + 1;
-		/* see the comment and corresponding wmb() in
-		 * the rcu_start_batch()
-		 */
-		smp_rmb();
-
-		if (!rcp->next_pending) {
-			/* and start it/schedule start if it's a new batch */
-			spin_lock(&rcp->lock);
-			rcp->next_pending = 1;
-			rcu_start_batch(rcp);
-			spin_unlock(&rcp->lock);
-		}
-	}
-
-	rcu_check_quiescent_state(rcp, rdp);
-	if (rdp->donelist)
-		rcu_do_batch(rdp);
-}
-
-static void rcu_process_callbacks(struct softirq_action *unused)
-{
-	__rcu_process_callbacks(&rcu_ctrlblk, &__get_cpu_var(rcu_data));
-	__rcu_process_callbacks(&rcu_bh_ctrlblk, &__get_cpu_var(rcu_bh_data));
-}
-
-static int __rcu_pending(struct rcu_ctrlblk *rcp, struct rcu_data *rdp)
-{
-	/* This cpu has pending rcu entries and the grace period
-	 * for them has completed.
-	 */
-	if (rdp->curlist && !rcu_batch_before(rcp->completed, rdp->batch))
-		return 1;
-
-	/* This cpu has no pending entries, but there are new entries */
-	if (!rdp->curlist && rdp->nxtlist)
-		return 1;
-
-	/* This cpu has finished callbacks to invoke */
-	if (rdp->donelist)
-		return 1;
-
-	/* The rcu core waits for a quiescent state from the cpu */
-	if (rdp->quiescbatch != rcp->cur || rdp->qs_pending)
-		return 1;
-
-	/* nothing to do */
-	return 0;
-}
-
-/*
- * Check to see if there is any immediate RCU-related work to be done
- * by the current CPU, returning 1 if so.  This function is part of the
- * RCU implementation; it is -not- an exported member of the RCU API.
- */
-int rcu_pending(int cpu)
-{
-	return __rcu_pending(&rcu_ctrlblk, &per_cpu(rcu_data, cpu)) ||
-		__rcu_pending(&rcu_bh_ctrlblk, &per_cpu(rcu_bh_data, cpu));
-}
-
-/*
- * Check to see if any future RCU-related work will need to be done
- * by the current CPU, even if none need be done immediately, returning
- * 1 if so.  This function is part of the RCU implementation; it is -not-
- * an exported member of the RCU API.
- */
-int rcu_needs_cpu(int cpu)
-{
-	struct rcu_data *rdp = &per_cpu(rcu_data, cpu);
-	struct rcu_data *rdp_bh = &per_cpu(rcu_bh_data, cpu);
-
-	return (!!rdp->curlist || !!rdp_bh->curlist || rcu_pending(cpu));
-}
-
-void rcu_check_callbacks(int cpu, int user)
-{
-	if (user || 
-	    (idle_cpu(cpu) && !in_softirq() && 
-				hardirq_count() <= (1 << HARDIRQ_SHIFT))) {
-		rcu_qsctr_inc(cpu);
-		rcu_bh_qsctr_inc(cpu);
-	} else if (!in_softirq())
-		rcu_bh_qsctr_inc(cpu);
-	raise_rcu_softirq();
-}
-
-static void rcu_init_percpu_data(int cpu, struct rcu_ctrlblk *rcp,
-						struct rcu_data *rdp)
-{
-	memset(rdp, 0, sizeof(*rdp));
-	rdp->curtail = &rdp->curlist;
-	rdp->nxttail = &rdp->nxtlist;
-	rdp->donetail = &rdp->donelist;
-	rdp->quiescbatch = rcp->completed;
-	rdp->qs_pending = 0;
-	rdp->cpu = cpu;
-	rdp->blimit = blimit;
-}
-
-static void __devinit rcu_online_cpu(int cpu)
-{
-	struct rcu_data *rdp = &per_cpu(rcu_data, cpu);
-	struct rcu_data *bh_rdp = &per_cpu(rcu_bh_data, cpu);
-
-	rcu_init_percpu_data(cpu, &rcu_ctrlblk, rdp);
-	rcu_init_percpu_data(cpu, &rcu_bh_ctrlblk, bh_rdp);
-	open_softirq(RCU_SOFTIRQ, rcu_process_callbacks, NULL);
-}
-
-static int __cpuinit rcu_cpu_notify(struct notifier_block *self,
-				unsigned long action, void *hcpu)
-{
-	long cpu = (long)hcpu;
-	switch (action) {
-	case CPU_UP_PREPARE:
-	case CPU_UP_PREPARE_FROZEN:
-		rcu_online_cpu(cpu);
-		break;
-	case CPU_DEAD:
-	case CPU_DEAD_FROZEN:
-		rcu_offline_cpu(cpu);
-		break;
-	default:
-		break;
-	}
-	return NOTIFY_OK;
-}
-
-static struct notifier_block __cpuinitdata rcu_nb = {
-	.notifier_call	= rcu_cpu_notify,
-};
-
-/*
- * Initializes rcu mechanism.  Assumed to be called early.
- * That is before local timer(SMP) or jiffie timer (uniproc) is setup.
- * Note that rcu_qsctr and friends are implicitly
- * initialized due to the choice of ``0'' for RCU_CTR_INVALID.
- */
 void __init rcu_init(void)
 {
-	rcu_cpu_notify(&rcu_nb, CPU_UP_PREPARE,
-			(void *)(long)smp_processor_id());
-	/* Register notifier for non-boot CPUs */
-	register_cpu_notifier(&rcu_nb);
-}
-
-struct rcu_synchronize {
-	struct rcu_head head;
-	struct completion completion;
-};
-
-/* Because of FASTCALL declaration of complete, we use this wrapper */
-static void wakeme_after_rcu(struct rcu_head  *head)
-{
-	struct rcu_synchronize *rcu;
-
-	rcu = container_of(head, struct rcu_synchronize, head);
-	complete(&rcu->completion);
-}
-
-/**
- * synchronize_rcu - wait until a grace period has elapsed.
- *
- * Control will return to the caller some time after a full grace
- * period has elapsed, in other words after all currently executing RCU
- * read-side critical sections have completed.  RCU read-side critical
- * sections are delimited by rcu_read_lock() and rcu_read_unlock(),
- * and may be nested.
- *
- * If your read-side code is not protected by rcu_read_lock(), do -not-
- * use synchronize_rcu().
- */
-void synchronize_rcu(void)
-{
-	struct rcu_synchronize rcu;
-
-	init_completion(&rcu.completion);
-	/* Will wake me after RCU finished */
-	call_rcu(&rcu.head, wakeme_after_rcu);
-
-	/* Wait for it */
-	wait_for_completion(&rcu.completion);
+	__rcu_init();
 }
 
-module_param(blimit, int, 0);
-module_param(qhimark, int, 0);
-module_param(qlowmark, int, 0);
-EXPORT_SYMBOL_GPL(rcu_batches_completed);
-EXPORT_SYMBOL_GPL(rcu_batches_completed_bh);
-EXPORT_SYMBOL_GPL(call_rcu);
-EXPORT_SYMBOL_GPL(call_rcu_bh);
-EXPORT_SYMBOL_GPL(synchronize_rcu);
-- 
Gautham R Shenoy
Linux Technology Center
IBM India.
"Freedom comes with a price tag of responsibility, which is still a bargain,
because Freedom is priceless!"

[RFC PATCH 3/6] Preempt-RCU: Fix rcu_barrier for preemptive environment

[Gautham R Shenoy]

Preempt-RCU: Fix rcu_barrier for preemptive environment.

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

Fix rcu_barrier() to work properly in preemptive kernel environment.
Also, the ordering of callback must be preserved while moving
callbacks to another CPU during CPU hotplug.

Signed-off-by: Gautham R Shenoy <ego@in.ibm.com>
Signed-off-by: Dipankar Sarma <dipankar@in.ibm.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
---

 kernel/rcuclassic.c |    2 +-
 kernel/rcupdate.c   |   10 ++++++++++
 2 files changed, 11 insertions(+), 1 deletions(-)

diff --git a/kernel/rcuclassic.c b/kernel/rcuclassic.c
index 11c16aa..deb2acc 100644
--- a/kernel/rcuclassic.c
+++ b/kernel/rcuclassic.c
@@ -371,9 +371,9 @@ static void __rcu_offline_cpu(struct rcu_data *this_rdp,
 	if (rcp->cur != rcp->completed)
 		cpu_quiet(rdp->cpu, rcp);
 	spin_unlock_bh(&rcp->lock);
+	rcu_move_batch(this_rdp, rdp->donelist, rdp->donetail);
 	rcu_move_batch(this_rdp, rdp->curlist, rdp->curtail);
 	rcu_move_batch(this_rdp, rdp->nxtlist, rdp->nxttail);
-	rcu_move_batch(this_rdp, rdp->donelist, rdp->donetail);
 }
 
 static void rcu_offline_cpu(int cpu)
diff --git a/kernel/rcupdate.c b/kernel/rcupdate.c
index 0ccd009..760dfc2 100644
--- a/kernel/rcupdate.c
+++ b/kernel/rcupdate.c
@@ -115,7 +115,17 @@ void rcu_barrier(void)
 	mutex_lock(&rcu_barrier_mutex);
 	init_completion(&rcu_barrier_completion);
 	atomic_set(&rcu_barrier_cpu_count, 0);
+	/*
+	 * The queueing of callbacks in all CPUs must be atomic with
+	 * respect to RCU, otherwise one CPU may queue a callback,
+	 * wait for a grace period, decrement barrier count and call
+	 * complete(), while other CPUs have not yet queued anything.
+	 * So, we need to make sure that grace periods cannot complete
+	 * until all the callbacks are queued.
+	 */
+	rcu_read_lock();
 	on_each_cpu(rcu_barrier_func, NULL, 0, 1);
+	rcu_read_unlock();
 	wait_for_completion(&rcu_barrier_completion);
 	mutex_unlock(&rcu_barrier_mutex);
 }
-- 
Gautham R Shenoy
Linux Technology Center
IBM India.
"Freedom comes with a price tag of responsibility, which is still a bargain,
because Freedom is priceless!"

[RFC PATCH 4/6] Preempt-RCU: Implementation

[Gautham R Shenoy]

Preempt-RCU: Implementation

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

This patch implements a new version of RCU which allows its read-side
critical sections to be preempted. It uses a set of counter pairs
to keep track of the read-side critical sections and flips them
when all tasks exit read-side critical section. The details
of this implementation can be found in this paper -

	http://www.rdrop.com/users/paulmck/RCU/OLSrtRCU.2006.08.11a.pdf

and the article-

	http://lwn.net/Articles/253651/

This patch was developed as a part of the -rt kernel development and
meant to provide better latencies when read-side critical sections of
RCU don't disable preemption.  As a consequence of keeping track of RCU
readers, the readers have a slight overhead (optimizations in the paper).
This implementation co-exists with the "classic" RCU implementations
and can be switched to at compiler.

Also includes RCU tracing summarized in debugfs.

Signed-off-by: Gautham R Shenoy <ego@in.ibm.com>
Signed-off-by: Dipankar Sarma <dipankar@in.ibm.com>
Signed-off-by: Paul E. McKenney <paulmck@us.ibm.com>
---

 include/linux/rcuclassic.h       |    3 
 include/linux/rcupdate.h         |   11 -
 include/linux/rcupreempt.h       |   86 ++++
 include/linux/rcupreempt_trace.h |   99 +++++
 include/linux/sched.h            |    5 
 kernel/Kconfig.preempt           |   38 ++
 kernel/Makefile                  |    7 
 kernel/fork.c                    |    4 
 kernel/rcuclassic.c              |    1 
 kernel/rcupreempt.c              |  816 ++++++++++++++++++++++++++++++++++++++
 kernel/rcupreempt_trace.c        |  330 +++++++++++++++
 11 files changed, 1394 insertions(+), 6 deletions(-)

diff --git a/include/linux/rcuclassic.h b/include/linux/rcuclassic.h
index 2b8b045..4d66242 100644
--- a/include/linux/rcuclassic.h
+++ b/include/linux/rcuclassic.h
@@ -157,5 +157,8 @@ extern void __rcu_init(void);
 extern void rcu_check_callbacks(int cpu, int user);
 extern void rcu_restart_cpu(int cpu);
 
+extern long rcu_batches_completed(void);
+extern long rcu_batches_completed_bh(void);
+
 #endif /* __KERNEL__ */
 #endif /* __LINUX_RCUCLASSIC_H */
diff --git a/include/linux/rcupdate.h b/include/linux/rcupdate.h
index 12aa13e..d32c14d 100644
--- a/include/linux/rcupdate.h
+++ b/include/linux/rcupdate.h
@@ -53,7 +53,11 @@ struct rcu_head {
 	void (*func)(struct rcu_head *head);
 };
 
+#ifdef CONFIG_CLASSIC_RCU
 #include <linux/rcuclassic.h>
+#else /* #ifdef CONFIG_CLASSIC_RCU */
+#include <linux/rcupreempt.h>
+#endif /* #else #ifdef CONFIG_CLASSIC_RCU */
 
 #define RCU_HEAD_INIT 	{ .next = NULL, .func = NULL }
 #define RCU_HEAD(head) struct rcu_head head = RCU_HEAD_INIT
@@ -231,13 +235,12 @@ extern void call_rcu_bh(struct rcu_head *head,
 /* Exported common interfaces */
 extern void synchronize_rcu(void);
 extern void rcu_barrier(void);
+extern long rcu_batches_completed(void);
+extern long rcu_batches_completed_bh(void);
 
 /* Internal to kernel */
 extern void rcu_init(void);
-extern void rcu_check_callbacks(int cpu, int user);
-
-extern long rcu_batches_completed(void);
-extern long rcu_batches_completed_bh(void);
+extern int rcu_needs_cpu(int cpu);
 
 #endif /* __KERNEL__ */
 #endif /* __LINUX_RCUPDATE_H */
diff --git a/include/linux/rcupreempt.h b/include/linux/rcupreempt.h
new file mode 100644
index 0000000..ece8eb3
--- /dev/null
+++ b/include/linux/rcupreempt.h
@@ -0,0 +1,86 @@
+/*
+ * Read-Copy Update mechanism for mutual exclusion (RT implementation)
+ *
+ * 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, write to the Free Software
+ * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
+ *
+ * Copyright (C) IBM Corporation, 2006
+ *
+ * Author:  Paul McKenney <paulmck@us.ibm.com>
+ *
+ * Based on the original work by Paul McKenney <paul.mckenney@us.ibm.com>
+ * and inputs from Rusty Russell, Andrea Arcangeli and Andi Kleen.
+ * Papers:
+ * http://www.rdrop.com/users/paulmck/paper/rclockpdcsproof.pdf
+ * http://lse.sourceforge.net/locking/rclock_OLS.2001.05.01c.sc.pdf (OLS2001)
+ *
+ * For detailed explanation of Read-Copy Update mechanism see -
+ * 		Documentation/RCU
+ *
+ */
+
+#ifndef __LINUX_RCUPREEMPT_H
+#define __LINUX_RCUPREEMPT_H
+
+#ifdef __KERNEL__
+
+#include <linux/cache.h>
+#include <linux/spinlock.h>
+#include <linux/threads.h>
+#include <linux/percpu.h>
+#include <linux/cpumask.h>
+#include <linux/seqlock.h>
+
+#define rcu_qsctr_inc(cpu)
+#define rcu_bh_qsctr_inc(cpu)
+#define call_rcu_bh(head, rcu) call_rcu(head, rcu)
+
+extern void __rcu_read_lock(void);
+extern void __rcu_read_unlock(void);
+extern int rcu_pending(int cpu);
+extern int rcu_needs_cpu(int cpu);
+
+#define __rcu_read_lock_bh()	{ rcu_read_lock(); local_bh_disable(); }
+#define __rcu_read_unlock_bh()	{ local_bh_enable(); rcu_read_unlock(); }
+
+extern void __synchronize_sched(void);
+
+extern void __rcu_init(void);
+extern void rcu_check_callbacks(int cpu, int user);
+extern void rcu_restart_cpu(int cpu);
+extern long rcu_batches_completed(void);
+
+/*
+ * Return the number of RCU batches processed thus far. Useful for debug
+ * and statistic. The _bh variant is identifcal to straight RCU
+ */
+static inline long rcu_batches_completed_bh(void)
+{
+	return rcu_batches_completed();
+}
+
+#ifdef CONFIG_RCU_TRACE
+struct rcupreempt_trace;
+extern long *rcupreempt_flipctr(int cpu);
+extern long rcupreempt_data_completed(void);
+extern int rcupreempt_flip_flag(int cpu);
+extern int rcupreempt_mb_flag(int cpu);
+extern char *rcupreempt_try_flip_state_name(void);
+extern struct rcupreempt_trace *rcupreempt_trace_cpu(int cpu);
+#endif
+
+struct softirq_action;
+
+#endif /* __KERNEL__ */
+#endif /* __LINUX_RCUPREEMPT_H */
diff --git a/include/linux/rcupreempt_trace.h b/include/linux/rcupreempt_trace.h
new file mode 100644
index 0000000..21cd6b2
--- /dev/null
+++ b/include/linux/rcupreempt_trace.h
@@ -0,0 +1,99 @@
+/*
+ * Read-Copy Update mechanism for mutual exclusion (RT implementation)
+ *
+ * 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, write to the Free Software
+ * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
+ *
+ * Copyright (C) IBM Corporation, 2006
+ *
+ * Author:  Paul McKenney <paulmck@us.ibm.com>
+ *
+ * Based on the original work by Paul McKenney <paul.mckenney@us.ibm.com>
+ * and inputs from Rusty Russell, Andrea Arcangeli and Andi Kleen.
+ * Papers:
+ * http://www.rdrop.com/users/paulmck/paper/rclockpdcsproof.pdf
+ * http://lse.sourceforge.net/locking/rclock_OLS.2001.05.01c.sc.pdf (OLS2001)
+ *
+ * For detailed explanation of the Preemptible Read-Copy Update mechanism see -
+ * 		 http://lwn.net/Articles/253651/
+ */
+
+#ifndef __LINUX_RCUPREEMPT_TRACE_H
+#define __LINUX_RCUPREEMPT_TRACE_H
+
+#ifdef __KERNEL__
+#include <linux/types.h>
+#include <linux/kernel.h>
+
+#include <asm/atomic.h>
+
+/*
+ * PREEMPT_RCU data structures.
+ */
+
+struct rcupreempt_trace {
+	long		next_length;
+	long		next_add;
+	long		wait_length;
+	long		wait_add;
+	long		done_length;
+	long		done_add;
+	long		done_remove;
+	atomic_t	done_invoked;
+	long		rcu_check_callbacks;
+	atomic_t	rcu_try_flip_1;
+	atomic_t	rcu_try_flip_e1;
+	long		rcu_try_flip_i1;
+	long		rcu_try_flip_ie1;
+	long		rcu_try_flip_g1;
+	long		rcu_try_flip_a1;
+	long		rcu_try_flip_ae1;
+	long		rcu_try_flip_a2;
+	long		rcu_try_flip_z1;
+	long		rcu_try_flip_ze1;
+	long		rcu_try_flip_z2;
+	long		rcu_try_flip_m1;
+	long		rcu_try_flip_me1;
+	long		rcu_try_flip_m2;
+};
+
+#ifdef CONFIG_RCU_TRACE
+#define RCU_TRACE(fn, arg) 	fn(arg);
+#else
+#define RCU_TRACE(fn, arg)
+#endif
+
+extern void rcupreempt_trace_move2done(struct rcupreempt_trace *trace);
+extern void rcupreempt_trace_move2wait(struct rcupreempt_trace *trace);
+extern void rcupreempt_trace_try_flip_1(struct rcupreempt_trace *trace);
+extern void rcupreempt_trace_try_flip_e1(struct rcupreempt_trace *trace);
+extern void rcupreempt_trace_try_flip_i1(struct rcupreempt_trace *trace);
+extern void rcupreempt_trace_try_flip_ie1(struct rcupreempt_trace *trace);
+extern void rcupreempt_trace_try_flip_g1(struct rcupreempt_trace *trace);
+extern void rcupreempt_trace_try_flip_a1(struct rcupreempt_trace *trace);
+extern void rcupreempt_trace_try_flip_ae1(struct rcupreempt_trace *trace);
+extern void rcupreempt_trace_try_flip_a2(struct rcupreempt_trace *trace);
+extern void rcupreempt_trace_try_flip_z1(struct rcupreempt_trace *trace);
+extern void rcupreempt_trace_try_flip_ze1(struct rcupreempt_trace *trace);
+extern void rcupreempt_trace_try_flip_z2(struct rcupreempt_trace *trace);
+extern void rcupreempt_trace_try_flip_m1(struct rcupreempt_trace *trace);
+extern void rcupreempt_trace_try_flip_me1(struct rcupreempt_trace *trace);
+extern void rcupreempt_trace_try_flip_m2(struct rcupreempt_trace *trace);
+extern void rcupreempt_trace_check_callbacks(struct rcupreempt_trace *trace);
+extern void rcupreempt_trace_done_remove(struct rcupreempt_trace *trace);
+extern void rcupreempt_trace_invoke(struct rcupreempt_trace *trace);
+extern void rcupreempt_trace_next_add(struct rcupreempt_trace *trace);
+
+#endif /* __KERNEL__ */
+#endif /* __LINUX_RCUPREEMPT_TRACE_H */
diff --git a/include/linux/sched.h b/include/linux/sched.h
index 15481e2..0416752 100644
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -979,6 +979,11 @@ struct task_struct {
 	int nr_cpus_allowed;
 	unsigned int time_slice;
 
+#ifdef CONFIG_PREEMPT_RCU
+	int rcu_read_lock_nesting;
+	int rcu_flipctr_idx;
+#endif /* #ifdef CONFIG_PREEMPT_RCU */
+
 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
 	struct sched_info sched_info;
 #endif
diff --git a/kernel/Kconfig.preempt b/kernel/Kconfig.preempt
index c64ce9c..06cafcc 100644
--- a/kernel/Kconfig.preempt
+++ b/kernel/Kconfig.preempt
@@ -63,3 +63,41 @@ config PREEMPT_BKL
 	  Say Y here if you are building a kernel for a desktop system.
 	  Say N if you are unsure.
 
+choice
+	prompt "RCU implementation type:"
+	default CLASSIC_RCU
+
+config CLASSIC_RCU
+	bool "Classic RCU"
+	help
+	  This option selects the classic RCU implementation that is
+	  designed for best read-side performance on non-realtime
+	  systems.
+
+	  Say Y if you are unsure.
+
+config PREEMPT_RCU
+	bool "Preemptible RCU"
+	depends on PREEMPT
+	help
+	  This option reduces the latency of the kernel by making certain
+	  RCU sections preemptible. Normally RCU code is non-preemptible, if
+	  this option is selected then read-only RCU sections become
+	  preemptible. This helps latency, but may expose bugs due to
+	  now-naive assumptions about each RCU read-side critical section
+	  remaining on a given CPU through its execution.
+
+	  Say N if you are unsure.
+
+endchoice
+
+config RCU_TRACE
+	bool "Enable tracing for RCU - currently stats in debugfs"
+	select DEBUG_FS
+	default y
+	help
+	  This option provides tracing in RCU which presents stats
+	  in debugfs for debugging RCU implementation.
+
+	  Say Y here if you want to enable RCU tracing
+	  Say N if you are unsure.
diff --git a/kernel/Makefile b/kernel/Makefile
index def5dd6..68755cd 100644
--- a/kernel/Makefile
+++ b/kernel/Makefile
@@ -6,7 +6,7 @@ obj-y     = sched.o fork.o exec_domain.o panic.o printk.o profile.o \
 	    exit.o itimer.o time.o softirq.o resource.o \
 	    sysctl.o capability.o ptrace.o timer.o user.o user_namespace.o \
 	    signal.o sys.o kmod.o workqueue.o pid.o \
-	    rcupdate.o rcuclassic.o extable.o params.o posix-timers.o \
+	    rcupdate.o extable.o params.o posix-timers.o \
 	    kthread.o wait.o kfifo.o sys_ni.o posix-cpu-timers.o mutex.o \
 	    hrtimer.o rwsem.o latency.o nsproxy.o srcu.o \
 	    utsname.o notifier.o
@@ -52,6 +52,11 @@ obj-$(CONFIG_DETECT_SOFTLOCKUP) += softlockup.o
 obj-$(CONFIG_GENERIC_HARDIRQS) += irq/
 obj-$(CONFIG_SECCOMP) += seccomp.o
 obj-$(CONFIG_RCU_TORTURE_TEST) += rcutorture.o
+obj-$(CONFIG_CLASSIC_RCU) += rcuclassic.o
+obj-$(CONFIG_PREEMPT_RCU) += rcupreempt.o
+ifeq ($(CONFIG_PREEMPT_RCU),y)
+obj-$(CONFIG_RCU_TRACE) += rcupreempt_trace.o
+endif
 obj-$(CONFIG_RELAY) += relay.o
 obj-$(CONFIG_SYSCTL) += utsname_sysctl.o
 obj-$(CONFIG_TASK_DELAY_ACCT) += delayacct.o
diff --git a/kernel/fork.c b/kernel/fork.c
index 930c518..9f8ef32 100644
--- a/kernel/fork.c
+++ b/kernel/fork.c
@@ -1045,6 +1045,10 @@ static struct task_struct *copy_process(unsigned long clone_flags,
 	copy_flags(clone_flags, p);
 	INIT_LIST_HEAD(&p->children);
 	INIT_LIST_HEAD(&p->sibling);
+#ifdef CONFIG_PREEMPT_RCU
+	p->rcu_read_lock_nesting = 0;
+	p->rcu_flipctr_idx = 0;
+#endif /* #ifdef CONFIG_PREEMPT_RCU */
 	p->vfork_done = NULL;
 	spin_lock_init(&p->alloc_lock);
 
diff --git a/kernel/rcuclassic.c b/kernel/rcuclassic.c
index deb2acc..c611537 100644
--- a/kernel/rcuclassic.c
+++ b/kernel/rcuclassic.c
@@ -45,7 +45,6 @@
 #include <linux/moduleparam.h>
 #include <linux/percpu.h>
 #include <linux/notifier.h>
-/* #include <linux/rcupdate.h> @@@ */
 #include <linux/cpu.h>
 #include <linux/mutex.h>
 
diff --git a/kernel/rcupreempt.c b/kernel/rcupreempt.c
new file mode 100644
index 0000000..a5aabb1
--- /dev/null
+++ b/kernel/rcupreempt.c
@@ -0,0 +1,816 @@
+/*
+ * Read-Copy Update mechanism for mutual exclusion, realtime implementation
+ *
+ * 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, write to the Free Software
+ * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
+ *
+ * Copyright IBM Corporation, 2006
+ *
+ * Authors: Paul E. McKenney <paulmck@us.ibm.com>
+ *		With thanks to Esben Nielsen, Bill Huey, and Ingo Molnar
+ *		for pushing me away from locks and towards counters, and
+ *		to Suparna Bhattacharya for pushing me completely away
+ *		from atomic instructions on the read side.
+ *
+ * Papers:  http://www.rdrop.com/users/paulmck/RCU
+ *
+ * Design Document: http://lwn.net/Articles/253651/
+ *
+ * For detailed explanation of Read-Copy Update mechanism see -
+ * 		Documentation/RCU/ *.txt
+ *
+ */
+#include <linux/types.h>
+#include <linux/kernel.h>
+#include <linux/init.h>
+#include <linux/spinlock.h>
+#include <linux/smp.h>
+#include <linux/rcupdate.h>
+#include <linux/interrupt.h>
+#include <linux/sched.h>
+#include <asm/atomic.h>
+#include <linux/bitops.h>
+#include <linux/module.h>
+#include <linux/completion.h>
+#include <linux/moduleparam.h>
+#include <linux/percpu.h>
+#include <linux/notifier.h>
+#include <linux/rcupdate.h>
+#include <linux/cpu.h>
+#include <linux/random.h>
+#include <linux/delay.h>
+#include <linux/byteorder/swabb.h>
+#include <linux/cpumask.h>
+#include <linux/rcupreempt_trace.h>
+
+/*
+ * Macro that prevents the compiler from reordering accesses, but does
+ * absolutely -nothing- to prevent CPUs from reordering.  This is used
+ * only to mediate communication between mainline code and hardware
+ * interrupt and NMI handlers.
+ */
+#define ACCESS_ONCE(x) (*(volatile typeof(x) *)&(x))
+
+/*
+ * PREEMPT_RCU data structures.
+ */
+
+/*
+ * GP_STAGES specifies the number of times the state machine has
+ * to go through the all the rcu_try_flip_states (see below)
+ * in a single Grace Period.
+ *
+ * GP in GP_STAGES stands for Grace Period ;)
+ */
+#define GP_STAGES    2
+struct rcu_data {
+	spinlock_t	lock;		/* Protect rcu_data fields. */
+	long		completed;	/* Number of last completed batch. */
+	int		waitlistcount;
+	struct tasklet_struct rcu_tasklet;
+	struct rcu_head *nextlist;
+	struct rcu_head **nexttail;
+	struct rcu_head *waitlist[GP_STAGES];
+	struct rcu_head **waittail[GP_STAGES];
+	struct rcu_head *donelist;
+	struct rcu_head **donetail;
+	long rcu_flipctr[2];
+#ifdef CONFIG_RCU_TRACE
+	struct rcupreempt_trace trace;
+#endif /* #ifdef CONFIG_RCU_TRACE */
+};
+
+/*
+ * States for rcu_try_flip() and friends.
+ */
+
+enum rcu_try_flip_states {
+
+	/*
+	 * Stay here if nothing is happening. Flip the counter if somthing
+	 * starts happening. Denoted by "I"
+	 */
+	rcu_try_flip_idle_state,
+
+	/*
+	 * Wait here for all CPUs to notice that the counter has flipped. This
+	 * prevents the old set of counters from ever being incremented once
+	 * we leave this state, which in turn is necessary because we cannot
+	 * test any individual counter for zero -- we can only check the sum.
+	 * Denoted by "A".
+	 */
+	rcu_try_flip_waitack_state,
+
+	/*
+	 * Wait here for the sum of the old per-CPU counters to reach zero.
+	 * Denoted by "Z".
+	 */
+	rcu_try_flip_waitzero_state,
+
+	/*
+	 * Wait here for each of the other CPUs to execute a memory barrier.
+	 * This is necessary to ensure that these other CPUs really have
+	 * completed executing their RCU read-side critical sections, despite
+	 * their CPUs wildly reordering memory. Denoted by "M".
+	 */
+	rcu_try_flip_waitmb_state,
+};
+
+struct rcu_ctrlblk {
+	spinlock_t	fliplock;	/* Protect state-machine transitions. */
+	long		completed;	/* Number of last completed batch. */
+	enum rcu_try_flip_states rcu_try_flip_state; /* The current state of
+							the rcu state machine */
+};
+
+static DEFINE_PER_CPU(struct rcu_data, rcu_data);
+static struct rcu_ctrlblk rcu_ctrlblk = {
+	.fliplock = __SPIN_LOCK_UNLOCKED(rcu_ctrlblk.fliplock),
+	.completed = 0,
+	.rcu_try_flip_state = rcu_try_flip_idle_state,
+};
+
+
+#ifdef CONFIG_RCU_TRACE
+static char *rcu_try_flip_state_names[] =
+	{ "idle", "waitack", "waitzero", "waitmb" };
+#endif /* #ifdef CONFIG_RCU_TRACE */
+
+/*
+ * Enum and per-CPU flag to determine when each CPU has seen
+ * the most recent counter flip.
+ */
+
+enum rcu_flip_flag_values {
+	rcu_flip_seen,		/* Steady/initial state, last flip seen. */
+				/* Only GP detector can update. */
+	rcu_flipped		/* Flip just completed, need confirmation. */
+				/* Only corresponding CPU can update. */
+};
+static DEFINE_PER_CPU_SHARED_ALIGNED(enum rcu_flip_flag_values, rcu_flip_flag)
+								= rcu_flip_seen;
+
+/*
+ * Enum and per-CPU flag to determine when each CPU has executed the
+ * needed memory barrier to fence in memory references from its last RCU
+ * read-side critical section in the just-completed grace period.
+ */
+
+enum rcu_mb_flag_values {
+	rcu_mb_done,		/* Steady/initial state, no mb()s required. */
+				/* Only GP detector can update. */
+	rcu_mb_needed		/* Flip just completed, need an mb(). */
+				/* Only corresponding CPU can update. */
+};
+static DEFINE_PER_CPU_SHARED_ALIGNED(enum rcu_mb_flag_values, rcu_mb_flag)
+								= rcu_mb_done;
+
+/*
+ * RCU_DATA_ME: find the current CPU's rcu_data structure.
+ * RCU_DATA_CPU: find the specified CPU's rcu_data structure.
+ */
+#define RCU_DATA_ME()		(&__get_cpu_var(rcu_data))
+#define RCU_DATA_CPU(cpu)	(&per_cpu(rcu_data, cpu))
+
+/*
+ * Helper macro for tracing when the appropriate rcu_data is not
+ * cached in a local variable, but where the CPU number is so cached.
+ */
+#define RCU_TRACE_CPU(f, cpu) RCU_TRACE(f, &(RCU_DATA_CPU(cpu)->trace));
+
+/*
+ * Helper macro for tracing when the appropriate rcu_data is not
+ * cached in a local variable.
+ */
+#define RCU_TRACE_ME(f) RCU_TRACE(f, &(RCU_DATA_ME()->trace));
+
+/*
+ * Helper macro for tracing when the appropriate rcu_data is pointed
+ * to by a local variable.
+ */
+#define RCU_TRACE_RDP(f, rdp) RCU_TRACE(f, &((rdp)->trace));
+
+/*
+ * Return the number of RCU batches processed thus far.  Useful
+ * for debug and statistics.
+ */
+long rcu_batches_completed(void)
+{
+	return rcu_ctrlblk.completed;
+}
+EXPORT_SYMBOL_GPL(rcu_batches_completed);
+
+EXPORT_SYMBOL_GPL(rcu_batches_completed_bh);
+
+void __rcu_read_lock(void)
+{
+	int idx;
+	struct task_struct *t = current;
+	int nesting;
+
+	nesting = ACCESS_ONCE(t->rcu_read_lock_nesting);
+	if (nesting != 0) {
+
+		/* An earlier rcu_read_lock() covers us, just count it. */
+
+		t->rcu_read_lock_nesting = nesting + 1;
+
+	} else {
+		unsigned long flags;
+
+		/*
+		 * We disable interrupts for the following reasons:
+		 * - If we get scheduling clock interrupt here, and we
+		 *   end up acking the counter flip, it's like a promise
+		 *   that we will never increment the old counter again.
+		 *   Thus we will break that promise if that
+		 *   scheduling clock interrupt happens between the time
+		 *   we pick the .completed field and the time that we
+		 *   increment our counter.
+		 *
+		 * - We don't want to be preempted out here.
+		 *
+		 * NMIs can still occur, of course, and might themselves
+		 * contain rcu_read_lock().
+		 */
+
+		local_irq_save(flags);
+
+		/*
+		 * Outermost nesting of rcu_read_lock(), so increment
+		 * the current counter for the current CPU.  Use volatile
+		 * casts to prevent the compiler from reordering.
+		 */
+
+		idx = ACCESS_ONCE(rcu_ctrlblk.completed) & 0x1;
+		ACCESS_ONCE(RCU_DATA_ME()->rcu_flipctr[idx])++;
+
+		/*
+		 * Now that the per-CPU counter has been incremented, we
+		 * are protected from races with rcu_read_lock() invoked
+		 * from NMI handlers on this CPU.  We can therefore safely
+		 * increment the nesting counter, relieving further NMIs
+		 * of the need to increment the per-CPU counter.
+		 */
+
+		ACCESS_ONCE(t->rcu_read_lock_nesting) = nesting + 1;
+
+		/*
+		 * Now that we have preventing any NMIs from storing
+		 * to the ->rcu_flipctr_idx, we can safely use it to
+		 * remember which counter to decrement in the matching
+		 * rcu_read_unlock().
+		 */
+
+		ACCESS_ONCE(t->rcu_flipctr_idx) = idx;
+		local_irq_restore(flags);
+	}
+}
+EXPORT_SYMBOL_GPL(__rcu_read_lock);
+
+void __rcu_read_unlock(void)
+{
+	int idx;
+	struct task_struct *t = current;
+	int nesting;
+
+	nesting = ACCESS_ONCE(t->rcu_read_lock_nesting);
+	if (nesting > 1) {
+
+		/*
+		 * We are still protected by the enclosing rcu_read_lock(),
+		 * so simply decrement the counter.
+		 */
+
+		t->rcu_read_lock_nesting = nesting - 1;
+
+	} else {
+		unsigned long flags;
+
+		/*
+		 * Disable local interrupts to prevent the grace-period
+		 * detection state machine from seeing us half-done.
+		 * NMIs can still occur, of course, and might themselves
+		 * contain rcu_read_lock() and rcu_read_unlock().
+		 */
+
+		local_irq_save(flags);
+
+		/*
+		 * Outermost nesting of rcu_read_unlock(), so we must
+		 * decrement the current counter for the current CPU.
+		 * This must be done carefully, because NMIs can
+		 * occur at any point in this code, and any rcu_read_lock()
+		 * and rcu_read_unlock() pairs in the NMI handlers
+		 * must interact non-destructively with this code.
+		 * Lots of volatile casts, and -very- careful ordering.
+		 *
+		 * Changes to this code, including this one, must be
+		 * inspected, validated, and tested extremely carefully!!!
+		 */
+
+		/*
+		 * First, pick up the index.
+		 */
+
+		idx = ACCESS_ONCE(t->rcu_flipctr_idx);
+
+		/*
+		 * Now that we have fetched the counter index, it is
+		 * safe to decrement the per-task RCU nesting counter.
+		 * After this, any interrupts or NMIs will increment and
+		 * decrement the per-CPU counters.
+		 */
+		ACCESS_ONCE(t->rcu_read_lock_nesting) = nesting - 1;
+
+		/*
+		 * It is now safe to decrement this task's nesting count.
+		 * NMIs that occur after this statement will route their
+		 * rcu_read_lock() calls through this "else" clause, and
+		 * will thus start incrementing the per-CPU counter on
+		 * their own.  They will also clobber ->rcu_flipctr_idx,
+		 * but that is OK, since we have already fetched it.
+		 */
+
+		ACCESS_ONCE(RCU_DATA_ME()->rcu_flipctr[idx])--;
+		local_irq_restore(flags);
+	}
+}
+EXPORT_SYMBOL_GPL(__rcu_read_unlock);
+
+/*
+ * If a global counter flip has occurred since the last time that we
+ * advanced callbacks, advance them.  Hardware interrupts must be
+ * disabled when calling this function.
+ */
+static void __rcu_advance_callbacks(struct rcu_data *rdp)
+{
+	int cpu;
+	int i;
+	int wlc = 0;
+
+	if (rdp->completed != rcu_ctrlblk.completed) {
+		if (rdp->waitlist[GP_STAGES - 1] != NULL) {
+			*rdp->donetail = rdp->waitlist[GP_STAGES - 1];
+			rdp->donetail = rdp->waittail[GP_STAGES - 1];
+			RCU_TRACE_RDP(rcupreempt_trace_move2done, rdp);
+		}
+		for (i = GP_STAGES - 2; i >= 0; i--) {
+			if (rdp->waitlist[i] != NULL) {
+				rdp->waitlist[i + 1] = rdp->waitlist[i];
+				rdp->waittail[i + 1] = rdp->waittail[i];
+				wlc++;
+			} else {
+				rdp->waitlist[i + 1] = NULL;
+				rdp->waittail[i + 1] =
+					&rdp->waitlist[i + 1];
+			}
+		}
+		if (rdp->nextlist != NULL) {
+			rdp->waitlist[0] = rdp->nextlist;
+			rdp->waittail[0] = rdp->nexttail;
+			wlc++;
+			rdp->nextlist = NULL;
+			rdp->nexttail = &rdp->nextlist;
+			RCU_TRACE_RDP(rcupreempt_trace_move2wait, rdp);
+		} else {
+			rdp->waitlist[0] = NULL;
+			rdp->waittail[0] = &rdp->waitlist[0];
+		}
+		rdp->waitlistcount = wlc;
+		rdp->completed = rcu_ctrlblk.completed;
+	}
+
+	/*
+	 * Check to see if this CPU needs to report that it has seen
+	 * the most recent counter flip, thereby declaring that all
+	 * subsequent rcu_read_lock() invocations will respect this flip.
+	 */
+
+	cpu = raw_smp_processor_id();
+	if (per_cpu(rcu_flip_flag, cpu) == rcu_flipped) {
+		smp_mb();  /* Subsequent counter accesses must see new value */
+		per_cpu(rcu_flip_flag, cpu) = rcu_flip_seen;
+		smp_mb();  /* Subsequent RCU read-side critical sections */
+			   /*  seen -after- acknowledgement. */
+	}
+}
+
+/*
+ * Get here when RCU is idle.  Decide whether we need to
+ * move out of idle state, and return non-zero if so.
+ * "Straightforward" approach for the moment, might later
+ * use callback-list lengths, grace-period duration, or
+ * some such to determine when to exit idle state.
+ * Might also need a pre-idle test that does not acquire
+ * the lock, but let's get the simple case working first...
+ */
+
+static int
+rcu_try_flip_idle(void)
+{
+	int cpu;
+
+	RCU_TRACE_ME(rcupreempt_trace_try_flip_i1);
+	if (!rcu_pending(smp_processor_id())) {
+		RCU_TRACE_ME(rcupreempt_trace_try_flip_ie1);
+		return 0;
+	}
+
+	/*
+	 * Do the flip.
+	 */
+
+	RCU_TRACE_ME(rcupreempt_trace_try_flip_g1);
+	rcu_ctrlblk.completed++;  /* stands in for rcu_try_flip_g2 */
+
+	/*
+	 * Need a memory barrier so that other CPUs see the new
+	 * counter value before they see the subsequent change of all
+	 * the rcu_flip_flag instances to rcu_flipped.
+	 */
+
+	smp_mb();	/* see above block comment. */
+
+	/* Now ask each CPU for acknowledgement of the flip. */
+
+	for_each_possible_cpu(cpu)
+		per_cpu(rcu_flip_flag, cpu) = rcu_flipped;
+
+	return 1;
+}
+
+/*
+ * Wait for CPUs to acknowledge the flip.
+ */
+
+static int
+rcu_try_flip_waitack(void)
+{
+	int cpu;
+
+	RCU_TRACE_ME(rcupreempt_trace_try_flip_a1);
+	for_each_possible_cpu(cpu)
+		if (per_cpu(rcu_flip_flag, cpu) != rcu_flip_seen) {
+			RCU_TRACE_ME(rcupreempt_trace_try_flip_ae1);
+			return 0;
+		}
+
+	/*
+	 * Make sure our checks above don't bleed into subsequent
+	 * waiting for the sum of the counters to reach zero.
+	 */
+
+	smp_mb();	/* see above block comment. */
+	RCU_TRACE_ME(rcupreempt_trace_try_flip_a2);
+	return 1;
+}
+
+/*
+ * Wait for collective ``last'' counter to reach zero,
+ * then tell all CPUs to do an end-of-grace-period memory barrier.
+ */
+
+static int
+rcu_try_flip_waitzero(void)
+{
+	int cpu;
+	int lastidx = !(rcu_ctrlblk.completed & 0x1);
+	int sum = 0;
+
+	/* Check to see if the sum of the "last" counters is zero. */
+
+	RCU_TRACE_ME(rcupreempt_trace_try_flip_z1);
+	for_each_possible_cpu(cpu)
+		sum += RCU_DATA_CPU(cpu)->rcu_flipctr[lastidx];
+	if (sum != 0) {
+		RCU_TRACE_ME(rcupreempt_trace_try_flip_ze1);
+		return 0;
+	}
+
+	/*
+	 * This ensures that the other CPUs see the call for
+	 * memory barriers -after- the sum to zero has been
+	 * detected here
+	 */
+	smp_mb();  /*  ^^^^^^^^^^^^ */
+
+	/* Call for a memory barrier from each CPU. */
+	for_each_possible_cpu(cpu)
+		per_cpu(rcu_mb_flag, cpu) = rcu_mb_needed;
+
+	RCU_TRACE_ME(rcupreempt_trace_try_flip_z2);
+	return 1;
+}
+
+/*
+ * Wait for all CPUs to do their end-of-grace-period memory barrier.
+ * Return 0 once all CPUs have done so.
+ */
+
+static int
+rcu_try_flip_waitmb(void)
+{
+	int cpu;
+
+	RCU_TRACE_ME(rcupreempt_trace_try_flip_m1);
+	for_each_possible_cpu(cpu)
+		if (per_cpu(rcu_mb_flag, cpu) != rcu_mb_done) {
+			RCU_TRACE_ME(rcupreempt_trace_try_flip_me1);
+			return 0;
+		}
+
+	smp_mb(); /* Ensure that the above checks precede any following flip. */
+	RCU_TRACE_ME(rcupreempt_trace_try_flip_m2);
+	return 1;
+}
+
+/*
+ * Attempt a single flip of the counters.  Remember, a single flip does
+ * -not- constitute a grace period.  Instead, the interval between
+ * at least GP_STAGES consecutive flips is a grace period.
+ *
+ * If anyone is nuts enough to run this CONFIG_PREEMPT_RCU implementation
+ * on a large SMP, they might want to use a hierarchical organization of
+ * the per-CPU-counter pairs.
+ */
+static void rcu_try_flip(void)
+{
+	unsigned long flags;
+
+	RCU_TRACE_ME(rcupreempt_trace_try_flip_1);
+	if (unlikely(!spin_trylock_irqsave(&rcu_ctrlblk.fliplock, flags))) {
+		RCU_TRACE_ME(rcupreempt_trace_try_flip_e1);
+		return;
+	}
+
+	/*
+	 * Take the next transition(s) through the RCU grace-period
+	 * flip-counter state machine.
+	 */
+
+	switch (rcu_ctrlblk.rcu_try_flip_state) {
+	case rcu_try_flip_idle_state:
+		if (rcu_try_flip_idle())
+			rcu_ctrlblk.rcu_try_flip_state =
+				rcu_try_flip_waitack_state;
+		break;
+	case rcu_try_flip_waitack_state:
+		if (rcu_try_flip_waitack())
+			rcu_ctrlblk.rcu_try_flip_state =
+				rcu_try_flip_waitzero_state;
+		break;
+	case rcu_try_flip_waitzero_state:
+		if (rcu_try_flip_waitzero())
+			rcu_ctrlblk.rcu_try_flip_state =
+				rcu_try_flip_waitmb_state;
+		break;
+	case rcu_try_flip_waitmb_state:
+		if (rcu_try_flip_waitmb())
+			rcu_ctrlblk.rcu_try_flip_state =
+				rcu_try_flip_idle_state;
+	}
+	spin_unlock_irqrestore(&rcu_ctrlblk.fliplock, flags);
+}
+
+/*
+ * Check to see if this CPU needs to do a memory barrier in order to
+ * ensure that any prior RCU read-side critical sections have committed
+ * their counter manipulations and critical-section memory references
+ * before declaring the grace period to be completed.
+ */
+static void rcu_check_mb(int cpu)
+{
+	if (per_cpu(rcu_mb_flag, cpu) == rcu_mb_needed) {
+		smp_mb();  /* Ensure RCU read-side accesses are visible. */
+		per_cpu(rcu_mb_flag, cpu) = rcu_mb_done;
+	}
+}
+
+void rcu_check_callbacks(int cpu, int user)
+{
+	unsigned long flags;
+	struct rcu_data *rdp = RCU_DATA_CPU(cpu);
+
+	rcu_check_mb(cpu);
+	if (rcu_ctrlblk.completed == rdp->completed)
+		rcu_try_flip();
+	spin_lock_irqsave(&rdp->lock, flags);
+	RCU_TRACE_RDP(rcupreempt_trace_check_callbacks, rdp);
+	__rcu_advance_callbacks(rdp);
+	if (rdp->donelist == NULL) {
+		spin_unlock_irqrestore(&rdp->lock, flags);
+	} else {
+		spin_unlock_irqrestore(&rdp->lock, flags);
+		raise_softirq(RCU_SOFTIRQ);
+	}
+}
+
+/*
+ * Needed by dynticks, to make sure all RCU processing has finished
+ * when we go idle:
+ */
+void rcu_advance_callbacks(int cpu, int user)
+{
+	unsigned long flags;
+	struct rcu_data *rdp = RCU_DATA_CPU(cpu);
+
+	if (rcu_ctrlblk.completed == rdp->completed) {
+		rcu_try_flip();
+		if (rcu_ctrlblk.completed == rdp->completed)
+			return;
+	}
+	spin_lock_irqsave(&rdp->lock, flags);
+	RCU_TRACE_RDP(rcupreempt_trace_check_callbacks, rdp);
+	__rcu_advance_callbacks(rdp);
+	spin_unlock_irqrestore(&rdp->lock, flags);
+}
+
+static void rcu_process_callbacks(struct softirq_action *unused)
+{
+	unsigned long flags;
+	struct rcu_head *next, *list;
+	struct rcu_data *rdp = RCU_DATA_ME();
+
+	spin_lock_irqsave(&rdp->lock, flags);
+	list = rdp->donelist;
+	if (list == NULL) {
+		spin_unlock_irqrestore(&rdp->lock, flags);
+		return;
+	}
+	rdp->donelist = NULL;
+	rdp->donetail = &rdp->donelist;
+	RCU_TRACE_RDP(rcupreempt_trace_done_remove, rdp);
+	spin_unlock_irqrestore(&rdp->lock, flags);
+	while (list) {
+		next = list->next;
+		list->func(list);
+		list = next;
+		RCU_TRACE_ME(rcupreempt_trace_invoke);
+	}
+}
+
+void call_rcu(struct rcu_head *head, void (*func)(struct rcu_head *rcu))
+{
+	unsigned long flags;
+	struct rcu_data *rdp;
+
+	head->func = func;
+	head->next = NULL;
+	local_irq_save(flags);
+	rdp = RCU_DATA_ME();
+	spin_lock(&rdp->lock);
+	__rcu_advance_callbacks(rdp);
+	*rdp->nexttail = head;
+	rdp->nexttail = &head->next;
+	RCU_TRACE_RDP(rcupreempt_trace_next_add, rdp);
+	spin_unlock(&rdp->lock);
+	local_irq_restore(flags);
+}
+EXPORT_SYMBOL_GPL(call_rcu);
+
+/*
+ * Wait until all currently running preempt_disable() code segments
+ * (including hardware-irq-disable segments) complete.  Note that
+ * in -rt this does -not- necessarily result in all currently executing
+ * interrupt -handlers- having completed.
+ */
+void __synchronize_sched(void)
+{
+	cpumask_t oldmask;
+	int cpu;
+
+	if (sched_getaffinity(0, &oldmask) < 0)
+		oldmask = cpu_possible_map;
+	for_each_online_cpu(cpu) {
+		sched_setaffinity(0, cpumask_of_cpu(cpu));
+		schedule();
+	}
+	sched_setaffinity(0, oldmask);
+}
+EXPORT_SYMBOL_GPL(__synchronize_sched);
+
+/*
+ * Check to see if any future RCU-related work will need to be done
+ * by the current CPU, even if none need be done immediately, returning
+ * 1 if so.  Assumes that notifiers would take care of handling any
+ * outstanding requests from the RCU core.
+ *
+ * This function is part of the RCU implementation; it is -not-
+ * an exported member of the RCU API.
+ */
+int rcu_needs_cpu(int cpu)
+{
+	struct rcu_data *rdp = RCU_DATA_CPU(cpu);
+
+	return (rdp->donelist != NULL ||
+		!!rdp->waitlistcount ||
+		rdp->nextlist != NULL);
+}
+
+int rcu_pending(int cpu)
+{
+	struct rcu_data *rdp = RCU_DATA_CPU(cpu);
+
+	/* The CPU has at least one callback queued somewhere. */
+
+	if (rdp->donelist != NULL ||
+	    !!rdp->waitlistcount ||
+	    rdp->nextlist != NULL)
+		return 1;
+
+	/* The RCU core needs an acknowledgement from this CPU. */
+
+	if ((per_cpu(rcu_flip_flag, cpu) == rcu_flipped) ||
+	    (per_cpu(rcu_mb_flag, cpu) == rcu_mb_needed))
+		return 1;
+
+	/* This CPU has fallen behind the global grace-period number. */
+
+	if (rdp->completed != rcu_ctrlblk.completed)
+		return 1;
+
+	/* Nothing needed from this CPU. */
+
+	return 0;
+}
+
+void __init __rcu_init(void)
+{
+	int cpu;
+	int i;
+	struct rcu_data *rdp;
+
+	printk(KERN_NOTICE "Preemptible RCU implementation.\n");
+	for_each_possible_cpu(cpu) {
+		rdp = RCU_DATA_CPU(cpu);
+		spin_lock_init(&rdp->lock);
+		rdp->completed = 0;
+		rdp->waitlistcount = 0;
+		rdp->nextlist = NULL;
+		rdp->nexttail = &rdp->nextlist;
+		for (i = 0; i < GP_STAGES; i++) {
+			rdp->waitlist[i] = NULL;
+			rdp->waittail[i] = &rdp->waitlist[i];
+		}
+		rdp->donelist = NULL;
+		rdp->donetail = &rdp->donelist;
+		rdp->rcu_flipctr[0] = 0;
+		rdp->rcu_flipctr[1] = 0;
+	}
+	open_softirq(RCU_SOFTIRQ, rcu_process_callbacks, NULL);
+}
+
+/*
+ * Deprecated, use synchronize_rcu() or synchronize_sched() instead.
+ */
+void synchronize_kernel(void)
+{
+	synchronize_rcu();
+}
+
+#ifdef CONFIG_RCU_TRACE
+long *rcupreempt_flipctr(int cpu)
+{
+	return &RCU_DATA_CPU(cpu)->rcu_flipctr[0];
+}
+EXPORT_SYMBOL_GPL(rcupreempt_flipctr);
+
+int rcupreempt_flip_flag(int cpu)
+{
+	return per_cpu(rcu_flip_flag, cpu);
+}
+EXPORT_SYMBOL_GPL(rcupreempt_flip_flag);
+
+int rcupreempt_mb_flag(int cpu)
+{
+	return per_cpu(rcu_mb_flag, cpu);
+}
+EXPORT_SYMBOL_GPL(rcupreempt_mb_flag);
+
+char *rcupreempt_try_flip_state_name(void)
+{
+	return rcu_try_flip_state_names[rcu_ctrlblk.rcu_try_flip_state];
+}
+EXPORT_SYMBOL_GPL(rcupreempt_try_flip_state_name);
+
+struct rcupreempt_trace *rcupreempt_trace_cpu(int cpu)
+{
+	struct rcu_data *rdp = RCU_DATA_CPU(cpu);
+
+	return &rdp->trace;
+}
+EXPORT_SYMBOL_GPL(rcupreempt_trace_cpu);
+
+#endif /* #ifdef RCU_TRACE */
diff --git a/kernel/rcupreempt_trace.c b/kernel/rcupreempt_trace.c
new file mode 100644
index 0000000..49ac494
--- /dev/null
+++ b/kernel/rcupreempt_trace.c
@@ -0,0 +1,330 @@
+/*
+ * Read-Copy Update tracing for realtime implementation
+ *
+ * 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, write to the Free Software
+ * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
+ *
+ * Copyright IBM Corporation, 2006
+ *
+ * Papers:  http://www.rdrop.com/users/paulmck/RCU
+ *
+ * For detailed explanation of Read-Copy Update mechanism see -
+ * 		Documentation/RCU/ *.txt
+ *
+ */
+#include <linux/types.h>
+#include <linux/kernel.h>
+#include <linux/init.h>
+#include <linux/spinlock.h>
+#include <linux/smp.h>
+#include <linux/rcupdate.h>
+#include <linux/interrupt.h>
+#include <linux/sched.h>
+#include <asm/atomic.h>
+#include <linux/bitops.h>
+#include <linux/module.h>
+#include <linux/completion.h>
+#include <linux/moduleparam.h>
+#include <linux/percpu.h>
+#include <linux/notifier.h>
+#include <linux/rcupdate.h>
+#include <linux/cpu.h>
+#include <linux/mutex.h>
+#include <linux/rcupreempt_trace.h>
+#include <linux/debugfs.h>
+
+static struct mutex rcupreempt_trace_mutex;
+static char *rcupreempt_trace_buf;
+#define RCUPREEMPT_TRACE_BUF_SIZE 4096
+
+void rcupreempt_trace_move2done(struct rcupreempt_trace *trace)
+{
+	trace->done_length += trace->wait_length;
+	trace->done_add += trace->wait_length;
+	trace->wait_length = 0;
+}
+void rcupreempt_trace_move2wait(struct rcupreempt_trace *trace)
+{
+	trace->wait_length += trace->next_length;
+	trace->wait_add += trace->next_length;
+	trace->next_length = 0;
+}
+void rcupreempt_trace_try_flip_1(struct rcupreempt_trace *trace)
+{
+	atomic_inc(&trace->rcu_try_flip_1);
+}
+void rcupreempt_trace_try_flip_e1(struct rcupreempt_trace *trace)
+{
+	atomic_inc(&trace->rcu_try_flip_e1);
+}
+void rcupreempt_trace_try_flip_i1(struct rcupreempt_trace *trace)
+{
+	trace->rcu_try_flip_i1++;
+}
+void rcupreempt_trace_try_flip_ie1(struct rcupreempt_trace *trace)
+{
+	trace->rcu_try_flip_ie1++;
+}
+void rcupreempt_trace_try_flip_g1(struct rcupreempt_trace *trace)
+{
+	trace->rcu_try_flip_g1++;
+}
+void rcupreempt_trace_try_flip_a1(struct rcupreempt_trace *trace)
+{
+	trace->rcu_try_flip_a1++;
+}
+void rcupreempt_trace_try_flip_ae1(struct rcupreempt_trace *trace)
+{
+	trace->rcu_try_flip_ae1++;
+}
+void rcupreempt_trace_try_flip_a2(struct rcupreempt_trace *trace)
+{
+	trace->rcu_try_flip_a2++;
+}
+void rcupreempt_trace_try_flip_z1(struct rcupreempt_trace *trace)
+{
+	trace->rcu_try_flip_z1++;
+}
+void rcupreempt_trace_try_flip_ze1(struct rcupreempt_trace *trace)
+{
+	trace->rcu_try_flip_ze1++;
+}
+void rcupreempt_trace_try_flip_z2(struct rcupreempt_trace *trace)
+{
+	trace->rcu_try_flip_z2++;
+}
+void rcupreempt_trace_try_flip_m1(struct rcupreempt_trace *trace)
+{
+	trace->rcu_try_flip_m1++;
+}
+void rcupreempt_trace_try_flip_me1(struct rcupreempt_trace *trace)
+{
+	trace->rcu_try_flip_me1++;
+}
+void rcupreempt_trace_try_flip_m2(struct rcupreempt_trace *trace)
+{
+	trace->rcu_try_flip_m2++;
+}
+void rcupreempt_trace_check_callbacks(struct rcupreempt_trace *trace)
+{
+	trace->rcu_check_callbacks++;
+}
+void rcupreempt_trace_done_remove(struct rcupreempt_trace *trace)
+{
+	trace->done_remove += trace->done_length;
+	trace->done_length = 0;
+}
+void rcupreempt_trace_invoke(struct rcupreempt_trace *trace)
+{
+	atomic_inc(&trace->done_invoked);
+}
+void rcupreempt_trace_next_add(struct rcupreempt_trace *trace)
+{
+	trace->next_add++;
+	trace->next_length++;
+}
+
+static void rcupreempt_trace_sum(struct rcupreempt_trace *sp)
+{
+	struct rcupreempt_trace *cp;
+	int cpu;
+
+	memset(sp, 0, sizeof(*sp));
+	for_each_possible_cpu(cpu) {
+		cp = rcupreempt_trace_cpu(cpu);
+		sp->next_length += cp->next_length;
+		sp->next_add += cp->next_add;
+		sp->wait_length += cp->wait_length;
+		sp->wait_add += cp->wait_add;
+		sp->done_length += cp->done_length;
+		sp->done_add += cp->done_add;
+		sp->done_remove += cp->done_remove;
+		atomic_set(&sp->done_invoked, atomic_read(&cp->done_invoked));
+		sp->rcu_check_callbacks += cp->rcu_check_callbacks;
+		atomic_set(&sp->rcu_try_flip_1,
+			   atomic_read(&cp->rcu_try_flip_1));
+		atomic_set(&sp->rcu_try_flip_e1,
+			   atomic_read(&cp->rcu_try_flip_e1));
+		sp->rcu_try_flip_i1 += cp->rcu_try_flip_i1;
+		sp->rcu_try_flip_ie1 += cp->rcu_try_flip_ie1;
+		sp->rcu_try_flip_g1 += cp->rcu_try_flip_g1;
+		sp->rcu_try_flip_a1 += cp->rcu_try_flip_a1;
+		sp->rcu_try_flip_ae1 += cp->rcu_try_flip_ae1;
+		sp->rcu_try_flip_a2 += cp->rcu_try_flip_a2;
+		sp->rcu_try_flip_z1 += cp->rcu_try_flip_z1;
+		sp->rcu_try_flip_ze1 += cp->rcu_try_flip_ze1;
+		sp->rcu_try_flip_z2 += cp->rcu_try_flip_z2;
+		sp->rcu_try_flip_m1 += cp->rcu_try_flip_m1;
+		sp->rcu_try_flip_me1 += cp->rcu_try_flip_me1;
+		sp->rcu_try_flip_m2 += cp->rcu_try_flip_m2;
+	}
+}
+
+static ssize_t rcustats_read(struct file *filp, char __user *buffer,
+				size_t count, loff_t *ppos)
+{
+	struct rcupreempt_trace trace;
+	ssize_t bcount;
+	int cnt = 0;
+
+	rcupreempt_trace_sum(&trace);
+	mutex_lock(&rcupreempt_trace_mutex);
+	snprintf(&rcupreempt_trace_buf[cnt], RCUPREEMPT_TRACE_BUF_SIZE - cnt,
+		 "ggp=%ld rcc=%ld\n",
+		 rcu_batches_completed(),
+		 trace.rcu_check_callbacks);
+	snprintf(&rcupreempt_trace_buf[cnt], RCUPREEMPT_TRACE_BUF_SIZE - cnt,
+		 "na=%ld nl=%ld wa=%ld wl=%ld da=%ld dl=%ld dr=%ld di=%d\n"
+		 "1=%d e1=%d i1=%ld ie1=%ld g1=%ld a1=%ld ae1=%ld a2=%ld\n"
+		 "z1=%ld ze1=%ld z2=%ld m1=%ld me1=%ld m2=%ld\n",
+
+		 trace.next_add, trace.next_length,
+		 trace.wait_add, trace.wait_length,
+		 trace.done_add, trace.done_length,
+		 trace.done_remove, atomic_read(&trace.done_invoked),
+		 atomic_read(&trace.rcu_try_flip_1),
+		 atomic_read(&trace.rcu_try_flip_e1),
+		 trace.rcu_try_flip_i1, trace.rcu_try_flip_ie1,
+		 trace.rcu_try_flip_g1,
+		 trace.rcu_try_flip_a1, trace.rcu_try_flip_ae1,
+			 trace.rcu_try_flip_a2,
+		 trace.rcu_try_flip_z1, trace.rcu_try_flip_ze1,
+			 trace.rcu_try_flip_z2,
+		 trace.rcu_try_flip_m1, trace.rcu_try_flip_me1,
+			trace.rcu_try_flip_m2);
+	bcount = simple_read_from_buffer(buffer, count, ppos,
+			rcupreempt_trace_buf, strlen(rcupreempt_trace_buf));
+	mutex_unlock(&rcupreempt_trace_mutex);
+	return bcount;
+}
+
+static ssize_t rcugp_read(struct file *filp, char __user *buffer,
+				size_t count, loff_t *ppos)
+{
+	long oldgp = rcu_batches_completed();
+	ssize_t bcount;
+
+	mutex_lock(&rcupreempt_trace_mutex);
+	synchronize_rcu();
+	snprintf(rcupreempt_trace_buf, RCUPREEMPT_TRACE_BUF_SIZE,
+		"oldggp=%ld  newggp=%ld\n", oldgp, rcu_batches_completed());
+	bcount = simple_read_from_buffer(buffer, count, ppos,
+			rcupreempt_trace_buf, strlen(rcupreempt_trace_buf));
+	mutex_unlock(&rcupreempt_trace_mutex);
+	return bcount;
+}
+
+static ssize_t rcuctrs_read(struct file *filp, char __user *buffer,
+				size_t count, loff_t *ppos)
+{
+	int cnt = 0;
+	int cpu;
+	int f = rcu_batches_completed() & 0x1;
+	ssize_t bcount;
+
+	mutex_lock(&rcupreempt_trace_mutex);
+
+	cnt += snprintf(&rcupreempt_trace_buf[cnt], RCUPREEMPT_TRACE_BUF_SIZE,
+				"CPU last cur F M\n");
+	for_each_online_cpu(cpu) {
+		long *flipctr = rcupreempt_flipctr(cpu);
+		cnt += snprintf(&rcupreempt_trace_buf[cnt],
+				RCUPREEMPT_TRACE_BUF_SIZE - cnt,
+					"%3d %4ld %3ld %d %d\n",
+			       cpu,
+			       flipctr[!f],
+			       flipctr[f],
+			       rcupreempt_flip_flag(cpu),
+			       rcupreempt_mb_flag(cpu));
+	}
+	cnt += snprintf(&rcupreempt_trace_buf[cnt],
+			RCUPREEMPT_TRACE_BUF_SIZE - cnt,
+			"ggp = %ld, state = %s\n",
+			rcu_batches_completed(),
+			rcupreempt_try_flip_state_name());
+	cnt += snprintf(&rcupreempt_trace_buf[cnt],
+			RCUPREEMPT_TRACE_BUF_SIZE - cnt,
+			"\n");
+	bcount = simple_read_from_buffer(buffer, count, ppos,
+			rcupreempt_trace_buf, strlen(rcupreempt_trace_buf));
+	mutex_unlock(&rcupreempt_trace_mutex);
+	return bcount;
+}
+
+static struct file_operations rcustats_fops = {
+	.owner = THIS_MODULE,
+	.read = rcustats_read,
+};
+
+static struct file_operations rcugp_fops = {
+	.owner = THIS_MODULE,
+	.read = rcugp_read,
+};
+
+static struct file_operations rcuctrs_fops = {
+	.owner = THIS_MODULE,
+	.read = rcuctrs_read,
+};
+
+static struct dentry *rcudir, *statdir, *ctrsdir, *gpdir;
+static int rcupreempt_debugfs_init(void)
+{
+	rcudir = debugfs_create_dir("rcu", NULL);
+	if (!rcudir)
+		goto out;
+	statdir = debugfs_create_file("rcustats", 0444, rcudir,
+						NULL, &rcustats_fops);
+	if (!statdir)
+		goto free_out;
+
+	gpdir = debugfs_create_file("rcugp", 0444, rcudir, NULL, &rcugp_fops);
+	if (!gpdir)
+		goto free_out;
+
+	ctrsdir = debugfs_create_file("rcuctrs", 0444, rcudir,
+						NULL, &rcuctrs_fops);
+	if (!ctrsdir)
+		goto free_out;
+	return 0;
+free_out:
+	if (statdir)
+		debugfs_remove(statdir);
+	if (gpdir)
+		debugfs_remove(gpdir);
+	debugfs_remove(rcudir);
+out:
+	return 1;
+}
+
+static int __init rcupreempt_trace_init(void)
+{
+	mutex_init(&rcupreempt_trace_mutex);
+	rcupreempt_trace_buf = kmalloc(RCUPREEMPT_TRACE_BUF_SIZE, GFP_KERNEL);
+	if (!rcupreempt_trace_buf)
+		return 1;
+	return rcupreempt_debugfs_init();
+}
+
+static void __exit rcupreempt_trace_cleanup(void)
+{
+	debugfs_remove(statdir);
+	debugfs_remove(gpdir);
+	debugfs_remove(ctrsdir);
+	debugfs_remove(rcudir);
+	kfree(rcupreempt_trace_buf);
+}
+
+
+module_init(rcupreempt_trace_init);
+module_exit(rcupreempt_trace_cleanup);
-- 
Gautham R Shenoy
Linux Technology Center
IBM India.
"Freedom comes with a price tag of responsibility, which is still a bargain,
because Freedom is priceless!"

[RFC PATCH 5/6] Preempt-RCU: CPU Hotplug handling

[Gautham R Shenoy]

Preempt-RCU: CPU Hotplug handling

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

This patch allows preemptible RCU to tolerate CPU-hotplug operations.
It accomplishes this by maintaining a local copy of a map of online
CPUs, which it accesses under its own lock.

Signed-off-by: Gautham R Shenoy <ego@in.ibm.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
---

 kernel/rcupreempt.c |  147 +++++++++++++++++++++++++++++++++++++++++++++++++--
 1 files changed, 142 insertions(+), 5 deletions(-)

diff --git a/kernel/rcupreempt.c b/kernel/rcupreempt.c
index a5aabb1..987cfb7 100644
--- a/kernel/rcupreempt.c
+++ b/kernel/rcupreempt.c
@@ -147,6 +147,8 @@ static char *rcu_try_flip_state_names[] =
 	{ "idle", "waitack", "waitzero", "waitmb" };
 #endif /* #ifdef CONFIG_RCU_TRACE */
 
+static cpumask_t rcu_cpu_online_map __read_mostly = CPU_MASK_NONE;
+
 /*
  * Enum and per-CPU flag to determine when each CPU has seen
  * the most recent counter flip.
@@ -445,7 +447,7 @@ rcu_try_flip_idle(void)
 
 	/* Now ask each CPU for acknowledgement of the flip. */
 
-	for_each_possible_cpu(cpu)
+	for_each_cpu_mask(cpu, rcu_cpu_online_map)
 		per_cpu(rcu_flip_flag, cpu) = rcu_flipped;
 
 	return 1;
@@ -461,7 +463,7 @@ rcu_try_flip_waitack(void)
 	int cpu;
 
 	RCU_TRACE_ME(rcupreempt_trace_try_flip_a1);
-	for_each_possible_cpu(cpu)
+	for_each_cpu_mask(cpu, rcu_cpu_online_map)
 		if (per_cpu(rcu_flip_flag, cpu) != rcu_flip_seen) {
 			RCU_TRACE_ME(rcupreempt_trace_try_flip_ae1);
 			return 0;
@@ -492,7 +494,7 @@ rcu_try_flip_waitzero(void)
 	/* Check to see if the sum of the "last" counters is zero. */
 
 	RCU_TRACE_ME(rcupreempt_trace_try_flip_z1);
-	for_each_possible_cpu(cpu)
+	for_each_cpu_mask(cpu, rcu_cpu_online_map)
 		sum += RCU_DATA_CPU(cpu)->rcu_flipctr[lastidx];
 	if (sum != 0) {
 		RCU_TRACE_ME(rcupreempt_trace_try_flip_ze1);
@@ -507,7 +509,7 @@ rcu_try_flip_waitzero(void)
 	smp_mb();  /*  ^^^^^^^^^^^^ */
 
 	/* Call for a memory barrier from each CPU. */
-	for_each_possible_cpu(cpu)
+	for_each_cpu_mask(cpu, rcu_cpu_online_map)
 		per_cpu(rcu_mb_flag, cpu) = rcu_mb_needed;
 
 	RCU_TRACE_ME(rcupreempt_trace_try_flip_z2);
@@ -525,7 +527,7 @@ rcu_try_flip_waitmb(void)
 	int cpu;
 
 	RCU_TRACE_ME(rcupreempt_trace_try_flip_m1);
-	for_each_possible_cpu(cpu)
+	for_each_cpu_mask(cpu, rcu_cpu_online_map)
 		if (per_cpu(rcu_mb_flag, cpu) != rcu_mb_done) {
 			RCU_TRACE_ME(rcupreempt_trace_try_flip_me1);
 			return 0;
@@ -637,6 +639,98 @@ void rcu_advance_callbacks(int cpu, int user)
 	spin_unlock_irqrestore(&rdp->lock, flags);
 }
 
+#ifdef CONFIG_HOTPLUG_CPU
+#define rcu_offline_cpu_enqueue(srclist, srctail, dstlist, dsttail) do { \
+		*dsttail = srclist; \
+		if (srclist != NULL) { \
+			dsttail = srctail; \
+			srclist = NULL; \
+			srctail = &srclist;\
+		} \
+	} while (0)
+
+void rcu_offline_cpu(int cpu)
+{
+	int i;
+	struct rcu_head *list = NULL;
+	unsigned long flags;
+	struct rcu_data *rdp = RCU_DATA_CPU(cpu);
+	struct rcu_head **tail = &list;
+
+	/*
+	 * Remove all callbacks from the newly dead CPU, retaining order.
+	 * Otherwise rcu_barrier() will fail
+	 */
+
+	spin_lock_irqsave(&rdp->lock, flags);
+	rcu_offline_cpu_enqueue(rdp->donelist, rdp->donetail, list, tail);
+	for (i = GP_STAGES - 1; i >= 0; i--)
+		rcu_offline_cpu_enqueue(rdp->waitlist[i], rdp->waittail[i],
+						list, tail);
+	rcu_offline_cpu_enqueue(rdp->nextlist, rdp->nexttail, list, tail);
+	spin_unlock_irqrestore(&rdp->lock, flags);
+	rdp->waitlistcount = 0;
+
+	/* Disengage the newly dead CPU from the grace-period computation. */
+
+	spin_lock_irqsave(&rcu_ctrlblk.fliplock, flags);
+	rcu_check_mb(cpu);
+	if (per_cpu(rcu_flip_flag, cpu) == rcu_flipped) {
+		smp_mb();  /* Subsequent counter accesses must see new value */
+		per_cpu(rcu_flip_flag, cpu) = rcu_flip_seen;
+		smp_mb();  /* Subsequent RCU read-side critical sections */
+			   /*  seen -after- acknowledgement. */
+	}
+
+	RCU_DATA_ME()->rcu_flipctr[0] += RCU_DATA_CPU(cpu)->rcu_flipctr[0];
+	RCU_DATA_ME()->rcu_flipctr[1] += RCU_DATA_CPU(cpu)->rcu_flipctr[1];
+
+	RCU_DATA_CPU(cpu)->rcu_flipctr[0] = 0;
+	RCU_DATA_CPU(cpu)->rcu_flipctr[1] = 0;
+
+	cpu_clear(cpu, rcu_cpu_online_map);
+
+	spin_unlock_irqrestore(&rcu_ctrlblk.fliplock, flags);
+
+	/*
+	 * Place the removed callbacks on the current CPU's queue.
+	 * Make them all start a new grace period: simple approach,
+	 * in theory could starve a given set of callbacks, but
+	 * you would need to be doing some serious CPU hotplugging
+	 * to make this happen.  If this becomes a problem, adding
+	 * a synchronize_rcu() to the hotplug path would be a simple
+	 * fix.
+	 */
+
+	rdp = RCU_DATA_ME();
+	spin_lock_irqsave(&rdp->lock, flags);
+	*rdp->nexttail = list;
+	if (list)
+		rdp->nexttail = tail;
+	spin_unlock_irqrestore(&rdp->lock, flags);
+}
+
+void __devinit rcu_online_cpu(int cpu)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&rcu_ctrlblk.fliplock, flags);
+	cpu_set(cpu, rcu_cpu_online_map);
+	spin_unlock_irqrestore(&rcu_ctrlblk.fliplock, flags);
+}
+
+#else /* #ifdef CONFIG_HOTPLUG_CPU */
+
+void rcu_offline_cpu(int cpu)
+{
+}
+
+void __devinit rcu_online_cpu(int cpu)
+{
+}
+
+#endif /* #else #ifdef CONFIG_HOTPLUG_CPU */
+
 static void rcu_process_callbacks(struct softirq_action *unused)
 {
 	unsigned long flags;
@@ -746,6 +840,32 @@ int rcu_pending(int cpu)
 	return 0;
 }
 
+static int __cpuinit rcu_cpu_notify(struct notifier_block *self,
+				unsigned long action, void *hcpu)
+{
+	long cpu = (long)hcpu;
+
+	switch (action) {
+	case CPU_UP_PREPARE:
+	case CPU_UP_PREPARE_FROZEN:
+		rcu_online_cpu(cpu);
+		break;
+	case CPU_UP_CANCELED:
+	case CPU_UP_CANCELED_FROZEN:
+	case CPU_DEAD:
+	case CPU_DEAD_FROZEN:
+		rcu_offline_cpu(cpu);
+		break;
+	default:
+		break;
+	}
+	return NOTIFY_OK;
+}
+
+static struct notifier_block __cpuinitdata rcu_nb = {
+	.notifier_call = rcu_cpu_notify,
+};
+
 void __init __rcu_init(void)
 {
 	int cpu;
@@ -769,6 +889,23 @@ void __init __rcu_init(void)
 		rdp->rcu_flipctr[0] = 0;
 		rdp->rcu_flipctr[1] = 0;
 	}
+	register_cpu_notifier(&rcu_nb);
+
+	/*
+	 * We don't need protection against CPU-Hotplug here
+	 * since
+	 * a) If a CPU comes online while we are iterating over the
+	 *    cpu_online_map below, we would only end up making a
+	 *    duplicate call to rcu_online_cpu() which sets the corresponding
+	 *    CPU's mask in the rcu_cpu_online_map.
+	 *
+	 * b) A CPU cannot go offline at this point in time since the user
+	 *    does not have access to the sysfs interface, nor do we
+	 *    suspend the system.
+	 */
+	for_each_online_cpu(cpu)
+		rcu_cpu_notify(&rcu_nb, CPU_UP_PREPARE,	(void *)(long) cpu);
+
 	open_softirq(RCU_SOFTIRQ, rcu_process_callbacks, NULL);
 }
 
-- 
Gautham R Shenoy
Linux Technology Center
IBM India.
"Freedom comes with a price tag of responsibility, which is still a bargain,
because Freedom is priceless!"

[RFC PATCH 6/6] Preempt-RCU: Update RCU Documentation.

[Gautham R Shenoy]

Preempt-RCU: Update RCU Documentation.

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

This patch updates the RCU documentation to reflect preemptible RCU as
well as recent publications.

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

 Documentation/RCU/RTFP.txt    |  210 +++++++++++++++++++++++++++++++++++++++--
 Documentation/RCU/rcu.txt     |   19 +++-
 Documentation/RCU/torture.txt |   11 +-
 3 files changed, 221 insertions(+), 19 deletions(-)

diff --git a/Documentation/RCU/RTFP.txt b/Documentation/RCU/RTFP.txt
index 6221464..39ad8f5 100644
--- a/Documentation/RCU/RTFP.txt
+++ b/Documentation/RCU/RTFP.txt
@@ -9,8 +9,8 @@ The first thing resembling RCU was published in 1980, when Kung and Lehman
 [Kung80] recommended use of a garbage collector to defer destruction
 of nodes in a parallel binary search tree in order to simplify its
 implementation.  This works well in environments that have garbage
-collectors, but current production garbage collectors incur significant
-read-side overhead.
+collectors, but most production garbage collectors incur significant
+overhead.
 
 In 1982, Manber and Ladner [Manber82,Manber84] recommended deferring
 destruction until all threads running at that time have terminated, again
@@ -99,16 +99,25 @@ locking, reduces contention, reduces memory latency for readers, and
 parallelizes pipeline stalls and memory latency for writers.  However,
 these techniques still impose significant read-side overhead in the
 form of memory barriers.  Researchers at Sun worked along similar lines
-in the same timeframe [HerlihyLM02,HerlihyLMS03].  These techniques
-can be thought of as inside-out reference counts, where the count is
-represented by the number of hazard pointers referencing a given data
-structure (rather than the more conventional counter field within the
-data structure itself).
+in the same timeframe [HerlihyLM02].  These techniques can be thought
+of as inside-out reference counts, where the count is represented by the
+number of hazard pointers referencing a given data structure (rather than
+the more conventional counter field within the data structure itself).
+
+By the same token, RCU can be thought of as a "bulk reference count",
+where some form of reference counter covers all reference by a given CPU
+or thread during a set timeframe.  This timeframe is related to, but
+not necessarily exactly the same as, an RCU grace period.  In classic
+RCU, the reference counter is the per-CPU bit in the "bitmask" field,
+and each such bit covers all references that might have been made by
+the corresponding CPU during the prior grace period.  Of course, RCU
+can be thought of in other terms as well.
 
 In 2003, the K42 group described how RCU could be used to create
-hot-pluggable implementations of operating-system functions.  Later that
-year saw a paper describing an RCU implementation of System V IPC
-[Arcangeli03], and an introduction to RCU in Linux Journal [McKenney03a].
+hot-pluggable implementations of operating-system functions [Appavoo03a].
+Later that year saw a paper describing an RCU implementation of System
+V IPC [Arcangeli03], and an introduction to RCU in Linux Journal
+[McKenney03a].
 
 2004 has seen a Linux-Journal article on use of RCU in dcache
 [McKenney04a], a performance comparison of locking to RCU on several
@@ -117,10 +126,19 @@ number of operating-system kernels [PaulEdwardMcKenneyPhD], a paper
 describing how to make RCU safe for soft-realtime applications [Sarma04c],
 and a paper describing SELinux performance with RCU [JamesMorris04b].
 
-2005 has seen further adaptation of RCU to realtime use, permitting
+2005 brought further adaptation of RCU to realtime use, permitting
 preemption of RCU realtime critical sections [PaulMcKenney05a,
 PaulMcKenney05b].
 
+2006 saw the first best-paper award for an RCU paper [ThomasEHart2006a],
+as well as further work on efficient implementations of preemptible
+RCU [PaulEMcKenney2006b], but priority-boosting of RCU read-side critical
+sections proved elusive.  An RCU implementation permitting general
+blocking in read-side critical sections appeared [PaulEMcKenney2006c],
+Robert Olsson described an RCU-protected trie-hash combination
+[RobertOlsson2006a].
+
+
 Bibtex Entries
 
 @article{Kung80
@@ -203,6 +221,41 @@ Bibtex Entries
 ,Address="New Orleans, LA"
 }
 
+@conference{Pu95a,
+Author = "Calton Pu and Tito Autrey and Andrew Black and Charles Consel and
+Crispin Cowan and Jon Inouye and Lakshmi Kethana and Jonathan Walpole and
+Ke Zhang",
+Title = "Optimistic Incremental Specialization: Streamlining a Commercial
+Operating System",
+Booktitle = "15\textsuperscript{th} ACM Symposium on
+Operating Systems Principles (SOSP'95)",
+address = "Copper Mountain, CO",
+month="December",
+year="1995",
+pages="314-321",
+annotation="
+	Uses a replugger, but with a flag to signal when people are
+	using the resource at hand.  Only one reader at a time.
+"
+}
+
+@conference{Cowan96a,
+Author = "Crispin Cowan and Tito Autrey and Charles Krasic and
+Calton Pu and Jonathan Walpole",
+Title = "Fast Concurrent Dynamic Linking for an Adaptive Operating System",
+Booktitle = "International Conference on Configurable Distributed Systems
+(ICCDS'96)",
+address = "Annapolis, MD",
+month="May",
+year="1996",
+pages="108",
+isbn="0-8186-7395-8",
+annotation="
+	Uses a replugger, but with a counter to signal when people are
+	using the resource at hand.  Allows multiple readers.
+"
+}
+
 @techreport{Slingwine95
 ,author="John D. Slingwine and Paul E. McKenney"
 ,title="Apparatus and Method for Achieving Reduced Overhead Mutual
@@ -312,6 +365,49 @@ Andrea Arcangeli and Andi Kleen and Orran Krieger and Rusty Russell"
 [Viewed June 23, 2004]"
 }
 
+@conference{Michael02a
+,author="Maged M. Michael"
+,title="Safe Memory Reclamation for Dynamic Lock-Free Objects Using Atomic
+Reads and Writes"
+,Year="2002"
+,Month="August"
+,booktitle="{Proceedings of the 21\textsuperscript{st} Annual ACM
+Symposium on Principles of Distributed Computing}"
+,pages="21-30"
+,annotation="
+	Each thread keeps an array of pointers to items that it is
+	currently referencing.	Sort of an inside-out garbage collection
+	mechanism, but one that requires the accessing code to explicitly
+	state its needs.  Also requires read-side memory barriers on
+	most architectures.
+"
+}
+
+@conference{Michael02b
+,author="Maged M. Michael"
+,title="High Performance Dynamic Lock-Free Hash Tables and List-Based Sets"
+,Year="2002"
+,Month="August"
+,booktitle="{Proceedings of the 14\textsuperscript{th} Annual ACM
+Symposium on Parallel
+Algorithms and Architecture}"
+,pages="73-82"
+,annotation="
+	Like the title says...
+"
+}
+
+@InProceedings{HerlihyLM02
+,author={Maurice Herlihy and Victor Luchangco and Mark Moir}
+,title="The Repeat Offender Problem: A Mechanism for Supporting Dynamic-Sized,
+Lock-Free Data Structures"
+,booktitle={Proceedings of 16\textsuperscript{th} International
+Symposium on Distributed Computing}
+,year=2002
+,month="October"
+,pages="339-353"
+}
+
 @article{Appavoo03a
 ,author="J. Appavoo and K. Hui and C. A. N. Soules and R. W. Wisniewski and
 D. M. {Da Silva} and O. Krieger and M. A. Auslander and D. J. Edelsohn and
@@ -447,3 +543,95 @@ Oregon Health and Sciences University"
 	Realtime turns into making RCU yet more realtime friendly.
 "
 }
+
+@conference{ThomasEHart2006a
+,Author="Thomas E. Hart and Paul E. McKenney and Angela Demke Brown"
+,Title="Making Lockless Synchronization Fast: Performance Implications
+of Memory Reclamation"
+,Booktitle="20\textsuperscript{th} {IEEE} International Parallel and
+Distributed Processing Symposium"
+,month="April"
+,year="2006"
+,day="25-29"
+,address="Rhodes, Greece"
+,annotation="
+	Compares QSBR (AKA "classic RCU"), HPBR, EBR, and lock-free
+	reference counting.
+"
+}
+
+@Conference{PaulEMcKenney2006b
+,Author="Paul E. McKenney and Dipankar Sarma and Ingo Molnar and
+Suparna Bhattacharya"
+,Title="Extending RCU for Realtime and Embedded Workloads"
+,Booktitle="{Ottawa Linux Symposium}"
+,Month="July"
+,Year="2006"
+,pages="v2 123-138"
+,note="Available:
+\url{http://www.linuxsymposium.org/2006/view_abstract.php?content_key=184}
+\url{http://www.rdrop.com/users/paulmck/RCU/OLSrtRCU.2006.08.11a.pdf}
+[Viewed January 1, 2007]"
+,annotation="
+	Described how to improve the -rt implementation of realtime RCU.
+"
+}
+
+@unpublished{PaulEMcKenney2006c
+,Author="Paul E. McKenney"
+,Title="Sleepable {RCU}"
+,month="October"
+,day="9"
+,year="2006"
+,note="Available:
+\url{http://lwn.net/Articles/202847/}
+Revised:
+\url{http://www.rdrop.com/users/paulmck/RCU/srcu.2007.01.14a.pdf}
+[Viewed August 21, 2006]"
+,annotation="
+	LWN article introducing SRCU.
+"
+}
+
+@unpublished{RobertOlsson2006a
+,Author="Robert Olsson and Stefan Nilsson"
+,Title="{TRASH}: A dynamic {LC}-trie and hash data structure"
+,month="August"
+,day="18"
+,year="2006"
+,note="Available:
+\url{http://www.nada.kth.se/~snilsson/public/papers/trash/trash.pdf}
+[Viewed February 24, 2007]"
+,annotation="
+	RCU-protected dynamic trie-hash combination.
+"
+}
+
+@unpublished{ThomasEHart2007a
+,Author="Thomas E. Hart and Paul E. McKenney and Angela Demke Brown and Jonathan Walpole"
+,Title="Performance of memory reclamation for lockless synchronization"
+,journal="J. Parallel Distrib. Comput."
+,year="2007"
+,note="To appear in J. Parallel Distrib. Comput.
+       \url{doi=10.1016/j.jpdc.2007.04.010}"
+,annotation={
+	Compares QSBR (AKA "classic RCU"), HPBR, EBR, and lock-free
+	reference counting.  Journal version of ThomasEHart2006a.
+}
+}
+
+@unpublished{PaulEMcKenney2007QRCUspin
+,Author="Paul E. McKenney"
+,Title="Using Promela and Spin to verify parallel algorithms"
+,month="August"
+,day="1"
+,year="2007"
+,note="Available:
+\url{http://lwn.net/Articles/243851/}
+[Viewed September 8, 2007]"
+,annotation="
+	LWN article describing Promela and spin, and also using Oleg
+	Nesterov's QRCU as an example (with Paul McKenney's fastpath).
+"
+}
+
diff --git a/Documentation/RCU/rcu.txt b/Documentation/RCU/rcu.txt
index f84407c..95821a2 100644
--- a/Documentation/RCU/rcu.txt
+++ b/Documentation/RCU/rcu.txt
@@ -36,6 +36,14 @@ o	How can the updater tell when a grace period has completed
 	executed in user mode, or executed in the idle loop, we can
 	safely free up that item.
 
+	Preemptible variants of RCU (CONFIG_PREEMPT_RCU) get the
+	same effect, but require that the readers manipulate CPU-local
+	counters.  These counters allow limited types of blocking
+	within RCU read-side critical sections.  SRCU also uses
+	CPU-local counters, and permits general blocking within
+	RCU read-side critical sections.  These two variants of
+	RCU detect grace periods by sampling these counters.
+
 o	If I am running on a uniprocessor kernel, which can only do one
 	thing at a time, why should I wait for a grace period?
 
@@ -46,7 +54,10 @@ o	How can I see where RCU is currently used in the Linux kernel?
 	Search for "rcu_read_lock", "rcu_read_unlock", "call_rcu",
 	"rcu_read_lock_bh", "rcu_read_unlock_bh", "call_rcu_bh",
 	"srcu_read_lock", "srcu_read_unlock", "synchronize_rcu",
-	"synchronize_net", and "synchronize_srcu".
+	"synchronize_net", "synchronize_srcu", and the other RCU
+	primitives.  Or grab one of the cscope databases from:
+
+	http://www.rdrop.com/users/paulmck/RCU/linuxusage/rculocktab.html
 
 o	What guidelines should I follow when writing code that uses RCU?
 
@@ -67,7 +78,11 @@ o	I hear that RCU is patented?  What is with that?
 
 o	I hear that RCU needs work in order to support realtime kernels?
 
-	Yes, work in progress.
+	This work is largely completed.  Realtime-friendly RCU can be
+	enabled via the CONFIG_PREEMPT_RCU kernel configuration parameter.
+	However, work is in progress for enabling priority boosting of
+	preempted RCU read-side critical sections.This is needed if you
+	have CPU-bound realtime threads.
 
 o	Where can I find more information on RCU?
 
diff --git a/Documentation/RCU/torture.txt b/Documentation/RCU/torture.txt
index 25a3c3f..2967a65 100644
--- a/Documentation/RCU/torture.txt
+++ b/Documentation/RCU/torture.txt
@@ -46,12 +46,13 @@ stat_interval	The number of seconds between output of torture
 
 shuffle_interval
 		The number of seconds to keep the test threads affinitied
-		to a particular subset of the CPUs.  Used in conjunction
-		with test_no_idle_hz.
+		to a particular subset of the CPUs, defaults to 5 seconds.
+		Used in conjunction with test_no_idle_hz.
 
 test_no_idle_hz	Whether or not to test the ability of RCU to operate in
 		a kernel that disables the scheduling-clock interrupt to
 		idle CPUs.  Boolean parameter, "1" to test, "0" otherwise.
+		Defaults to omitting this test.
 
 torture_type	The type of RCU to test: "rcu" for the rcu_read_lock() API,
 		"rcu_sync" for rcu_read_lock() with synchronous reclamation,
@@ -82,8 +83,6 @@ be evident.  ;-)
 
 The entries are as follows:
 
-o	"ggp": The number of counter flips (or batches) since boot.
-
 o	"rtc": The hexadecimal address of the structure currently visible
 	to readers.
 
@@ -117,8 +116,8 @@ o	"Reader Pipe": Histogram of "ages" of structures seen by readers.
 o	"Reader Batch": Another histogram of "ages" of structures seen
 	by readers, but in terms of counter flips (or batches) rather
 	than in terms of grace periods.  The legal number of non-zero
-	entries is again two.  The reason for this separate view is
-	that it is easier to get the third entry to show up in the
+	entries is again two.  The reason for this separate view is that
+	it is sometimes easier to get the third entry to show up in the
 	"Reader Batch" list than in the "Reader Pipe" list.
 
 o	"Free-Block Circulation": Shows the number of torture structures
-- 
Gautham R Shenoy
Linux Technology Center
IBM India.
"Freedom comes with a price tag of responsibility, which is still a bargain,
because Freedom is priceless!"

Re: [RFC PATCH 0/6] RCU: Preemptible-RCU

[Steven Rostedt]

On Thu, 13 Dec 2007, Gautham R Shenoy wrote:

>
> Currently it is based against the latest linux-2.6-sched-devel.git
>
> Awaiting your feedback!

Hi Gautham,

Thanks for posting this. I believe this is the same version of preempt RCU
as we have in the RT patch. It seems to be very stable. I ran the RT patch
version of the RCU Preempt (just the Preempt RCU patches without RT on
latest git) on a 64way box and the results seems just as good (if not
slightly better) than classic RCU!  I'll rerun this patch series on that
box and post the results.

>From what I'm seening with this, is that it is ready for mainline. These
patches should probably go into -mm and be ready for 2.6.25.  If Andrew
wants to wait for my results, I'll run them tonight.

Thanks Gautham, Paul and Dipankar for all this great work!

-- Steve

Re: [RFC PATCH 0/6] RCU: Preemptible-RCU

[Ingo Molnar]

* Gautham R Shenoy <ego@in.ibm.com> wrote:

> Hello everyone,
> 
> This patchset is an updated version of the preemptible RCU patchset 
> that Paul McKenney had posted it in September earlier this year that 
> can be found here --> http://lkml.org/lkml/2007/9/10/213
> 
> This patchset incorporates the review comments from Oleg Nesterov and 
> Steven Rostedt.
> 
> The testing report of the patchset is as follows:
> ====================================================================
> Patch-stack:  	2.6.23-rc3 + cpu-hotplug patches from 
> 		http://lkml.org/lkml/2007/11/15/239 + Preempt-RCU
> 		patches.
> Test:		RCU-Torture running parallelly with CPU-Hotplug
> 		operations.
> Duration:	24 hours.
> Architectures:	x86,x86_64, ppc64.
> ====================================================================
> 
> 
> Currently it is based against the latest linux-2.6-sched-devel.git
> 
> Awaiting your feedback!

thanks Gautham, the patchset from you and Paul looks good to me and i've 
applied it to sched-devel.git to get it tested and reviewed some more.

from the Nitpicking Police, there are a couple of minor style 
problems/warnings with the code, you can see it via:

  scripts/checkpatch.pl --file kernel/rcu*.c

nothing serious - RCU is still one of the cleanest subsystems in the 
kernel:

                                      errors   lines of code   errors/KLOC
kernel/rcuclassic.c                        0             575             0
kernel/rcupdate.c                          1             138           7.2
kernel/rcupreempt.c                        0             953             0
kernel/rcupreempt_trace.c                  0             330             0
kernel/rcutorture.c                        8             995           8.0

the eventual goal would be to match:

   scripts/checkpatch.pl --file kernel/sched*.[ch]

output ;-)

	Ingo

Re: [RFC PATCH 0/6] RCU: Preemptible-RCU

[Ingo Molnar]

* Steven Rostedt <rostedt@goodmis.org> wrote:

> On Thu, 13 Dec 2007, Gautham R Shenoy wrote:
> 
> >
> > Currently it is based against the latest linux-2.6-sched-devel.git
> >
> > Awaiting your feedback!
> 
> Hi Gautham,
> 
> Thanks for posting this. I believe this is the same version of preempt 
> RCU as we have in the RT patch. It seems to be very stable. I ran the 
> RT patch version of the RCU Preempt (just the Preempt RCU patches 
> without RT on latest git) on a 64way box and the results seems just as 
> good (if not slightly better) than classic RCU!  I'll rerun this patch 
> series on that box and post the results.
> 
> >From what I'm seening with this, is that it is ready for mainline. 
> >These
> patches should probably go into -mm and be ready for 2.6.25.  If Andrew
> wants to wait for my results, I'll run them tonight.
> 
> Thanks Gautham, Paul and Dipankar for all this great work!

Steve, if you went with a fine comb over the code and have done a 
thorough review, could you send me your Reviewed-by line for this 
patchset?

	Ingo

Re: [RFC PATCH 6/6] Preempt-RCU: Update RCU Documentation.

[Ingo Molnar]

* Gautham R Shenoy <ego@in.ibm.com> wrote:

> Preempt-RCU: Update RCU Documentation.
> 
> From: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
> 
> This patch updates the RCU documentation to reflect preemptible RCU as
> well as recent publications.
> 
> Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>

Gautham, this patch is missing your SoB line - could you please send it?

	Ingo

Re: [RFC PATCH 0/6] RCU: Preemptible-RCU

[Steven Rostedt]

On Thu, 13 Dec 2007, Ingo Molnar wrote:
> Steve, if you went with a fine comb over the code and have done a
> thorough review, could you send me your Reviewed-by line for this
> patchset?

Sure thing. And here's the thread that contains a lot of my review (and I
had some with Paul on IRC).

http://lkml.org/lkml/2007/9/10/213

Reviewed-by: Steven Rostedt <srostedt@redhat.com>

Re: [RFC PATCH 6/6] Preempt-RCU: Update RCU Documentation.

[Gautham R Shenoy]

On Thu, Dec 13, 2007 at 09:42:53PM +0100, Ingo Molnar wrote:
> 
> * Gautham R Shenoy <ego@in.ibm.com> wrote:
> 
> > Preempt-RCU: Update RCU Documentation.
> > 
> > From: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
> > 
> > This patch updates the RCU documentation to reflect preemptible RCU as
> > well as recent publications.
> > 
> > Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
> 
> Gautham, this patch is missing your SoB line - could you please send it?

Sorry about that. I do take responsibility of this patch too!

Patch Appended below.
> 
> 	Ingo

Thanks and Regards
gautham.

---->
Preempt-RCU: Update RCU Documentation.

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

This patch updates the RCU documentation to reflect preemptible RCU as
well as recent publications.

Signed-off-by: Gautham R Shenoy <ego@in.ibm.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
---

 Documentation/RCU/RTFP.txt    |  210 +++++++++++++++++++++++++++++++++++++++--
 Documentation/RCU/rcu.txt     |   19 +++-
 Documentation/RCU/torture.txt |   11 +-
 3 files changed, 221 insertions(+), 19 deletions(-)

diff --git a/Documentation/RCU/RTFP.txt b/Documentation/RCU/RTFP.txt
index 6221464..39ad8f5 100644
--- a/Documentation/RCU/RTFP.txt
+++ b/Documentation/RCU/RTFP.txt
@@ -9,8 +9,8 @@ The first thing resembling RCU was published in 1980, when Kung and Lehman
 [Kung80] recommended use of a garbage collector to defer destruction
 of nodes in a parallel binary search tree in order to simplify its
 implementation.  This works well in environments that have garbage
-collectors, but current production garbage collectors incur significant
-read-side overhead.
+collectors, but most production garbage collectors incur significant
+overhead.
 
 In 1982, Manber and Ladner [Manber82,Manber84] recommended deferring
 destruction until all threads running at that time have terminated, again
@@ -99,16 +99,25 @@ locking, reduces contention, reduces memory latency for readers, and
 parallelizes pipeline stalls and memory latency for writers.  However,
 these techniques still impose significant read-side overhead in the
 form of memory barriers.  Researchers at Sun worked along similar lines
-in the same timeframe [HerlihyLM02,HerlihyLMS03].  These techniques
-can be thought of as inside-out reference counts, where the count is
-represented by the number of hazard pointers referencing a given data
-structure (rather than the more conventional counter field within the
-data structure itself).
+in the same timeframe [HerlihyLM02].  These techniques can be thought
+of as inside-out reference counts, where the count is represented by the
+number of hazard pointers referencing a given data structure (rather than
+the more conventional counter field within the data structure itself).
+
+By the same token, RCU can be thought of as a "bulk reference count",
+where some form of reference counter covers all reference by a given CPU
+or thread during a set timeframe.  This timeframe is related to, but
+not necessarily exactly the same as, an RCU grace period.  In classic
+RCU, the reference counter is the per-CPU bit in the "bitmask" field,
+and each such bit covers all references that might have been made by
+the corresponding CPU during the prior grace period.  Of course, RCU
+can be thought of in other terms as well.
 
 In 2003, the K42 group described how RCU could be used to create
-hot-pluggable implementations of operating-system functions.  Later that
-year saw a paper describing an RCU implementation of System V IPC
-[Arcangeli03], and an introduction to RCU in Linux Journal [McKenney03a].
+hot-pluggable implementations of operating-system functions [Appavoo03a].
+Later that year saw a paper describing an RCU implementation of System
+V IPC [Arcangeli03], and an introduction to RCU in Linux Journal
+[McKenney03a].
 
 2004 has seen a Linux-Journal article on use of RCU in dcache
 [McKenney04a], a performance comparison of locking to RCU on several
@@ -117,10 +126,19 @@ number of operating-system kernels [PaulEdwardMcKenneyPhD], a paper
 describing how to make RCU safe for soft-realtime applications [Sarma04c],
 and a paper describing SELinux performance with RCU [JamesMorris04b].
 
-2005 has seen further adaptation of RCU to realtime use, permitting
+2005 brought further adaptation of RCU to realtime use, permitting
 preemption of RCU realtime critical sections [PaulMcKenney05a,
 PaulMcKenney05b].
 
+2006 saw the first best-paper award for an RCU paper [ThomasEHart2006a],
+as well as further work on efficient implementations of preemptible
+RCU [PaulEMcKenney2006b], but priority-boosting of RCU read-side critical
+sections proved elusive.  An RCU implementation permitting general
+blocking in read-side critical sections appeared [PaulEMcKenney2006c],
+Robert Olsson described an RCU-protected trie-hash combination
+[RobertOlsson2006a].
+
+
 Bibtex Entries
 
 @article{Kung80
@@ -203,6 +221,41 @@ Bibtex Entries
 ,Address="New Orleans, LA"
 }
 
+@conference{Pu95a,
+Author = "Calton Pu and Tito Autrey and Andrew Black and Charles Consel and
+Crispin Cowan and Jon Inouye and Lakshmi Kethana and Jonathan Walpole and
+Ke Zhang",
+Title = "Optimistic Incremental Specialization: Streamlining a Commercial
+Operating System",
+Booktitle = "15\textsuperscript{th} ACM Symposium on
+Operating Systems Principles (SOSP'95)",
+address = "Copper Mountain, CO",
+month="December",
+year="1995",
+pages="314-321",
+annotation="
+	Uses a replugger, but with a flag to signal when people are
+	using the resource at hand.  Only one reader at a time.
+"
+}
+
+@conference{Cowan96a,
+Author = "Crispin Cowan and Tito Autrey and Charles Krasic and
+Calton Pu and Jonathan Walpole",
+Title = "Fast Concurrent Dynamic Linking for an Adaptive Operating System",
+Booktitle = "International Conference on Configurable Distributed Systems
+(ICCDS'96)",
+address = "Annapolis, MD",
+month="May",
+year="1996",
+pages="108",
+isbn="0-8186-7395-8",
+annotation="
+	Uses a replugger, but with a counter to signal when people are
+	using the resource at hand.  Allows multiple readers.
+"
+}
+
 @techreport{Slingwine95
 ,author="John D. Slingwine and Paul E. McKenney"
 ,title="Apparatus and Method for Achieving Reduced Overhead Mutual
@@ -312,6 +365,49 @@ Andrea Arcangeli and Andi Kleen and Orran Krieger and Rusty Russell"
 [Viewed June 23, 2004]"
 }
 
+@conference{Michael02a
+,author="Maged M. Michael"
+,title="Safe Memory Reclamation for Dynamic Lock-Free Objects Using Atomic
+Reads and Writes"
+,Year="2002"
+,Month="August"
+,booktitle="{Proceedings of the 21\textsuperscript{st} Annual ACM
+Symposium on Principles of Distributed Computing}"
+,pages="21-30"
+,annotation="
+	Each thread keeps an array of pointers to items that it is
+	currently referencing.	Sort of an inside-out garbage collection
+	mechanism, but one that requires the accessing code to explicitly
+	state its needs.  Also requires read-side memory barriers on
+	most architectures.
+"
+}
+
+@conference{Michael02b
+,author="Maged M. Michael"
+,title="High Performance Dynamic Lock-Free Hash Tables and List-Based Sets"
+,Year="2002"
+,Month="August"
+,booktitle="{Proceedings of the 14\textsuperscript{th} Annual ACM
+Symposium on Parallel
+Algorithms and Architecture}"
+,pages="73-82"
+,annotation="
+	Like the title says...
+"
+}
+
+@InProceedings{HerlihyLM02
+,author={Maurice Herlihy and Victor Luchangco and Mark Moir}
+,title="The Repeat Offender Problem: A Mechanism for Supporting Dynamic-Sized,
+Lock-Free Data Structures"
+,booktitle={Proceedings of 16\textsuperscript{th} International
+Symposium on Distributed Computing}
+,year=2002
+,month="October"
+,pages="339-353"
+}
+
 @article{Appavoo03a
 ,author="J. Appavoo and K. Hui and C. A. N. Soules and R. W. Wisniewski and
 D. M. {Da Silva} and O. Krieger and M. A. Auslander and D. J. Edelsohn and
@@ -447,3 +543,95 @@ Oregon Health and Sciences University"
 	Realtime turns into making RCU yet more realtime friendly.
 "
 }
+
+@conference{ThomasEHart2006a
+,Author="Thomas E. Hart and Paul E. McKenney and Angela Demke Brown"
+,Title="Making Lockless Synchronization Fast: Performance Implications
+of Memory Reclamation"
+,Booktitle="20\textsuperscript{th} {IEEE} International Parallel and
+Distributed Processing Symposium"
+,month="April"
+,year="2006"
+,day="25-29"
+,address="Rhodes, Greece"
+,annotation="
+	Compares QSBR (AKA "classic RCU"), HPBR, EBR, and lock-free
+	reference counting.
+"
+}
+
+@Conference{PaulEMcKenney2006b
+,Author="Paul E. McKenney and Dipankar Sarma and Ingo Molnar and
+Suparna Bhattacharya"
+,Title="Extending RCU for Realtime and Embedded Workloads"
+,Booktitle="{Ottawa Linux Symposium}"
+,Month="July"
+,Year="2006"
+,pages="v2 123-138"
+,note="Available:
+\url{http://www.linuxsymposium.org/2006/view_abstract.php?content_key=184}
+\url{http://www.rdrop.com/users/paulmck/RCU/OLSrtRCU.2006.08.11a.pdf}
+[Viewed January 1, 2007]"
+,annotation="
+	Described how to improve the -rt implementation of realtime RCU.
+"
+}
+
+@unpublished{PaulEMcKenney2006c
+,Author="Paul E. McKenney"
+,Title="Sleepable {RCU}"
+,month="October"
+,day="9"
+,year="2006"
+,note="Available:
+\url{http://lwn.net/Articles/202847/}
+Revised:
+\url{http://www.rdrop.com/users/paulmck/RCU/srcu.2007.01.14a.pdf}
+[Viewed August 21, 2006]"
+,annotation="
+	LWN article introducing SRCU.
+"
+}
+
+@unpublished{RobertOlsson2006a
+,Author="Robert Olsson and Stefan Nilsson"
+,Title="{TRASH}: A dynamic {LC}-trie and hash data structure"
+,month="August"
+,day="18"
+,year="2006"
+,note="Available:
+\url{http://www.nada.kth.se/~snilsson/public/papers/trash/trash.pdf}
+[Viewed February 24, 2007]"
+,annotation="
+	RCU-protected dynamic trie-hash combination.
+"
+}
+
+@unpublished{ThomasEHart2007a
+,Author="Thomas E. Hart and Paul E. McKenney and Angela Demke Brown and Jonathan Walpole"
+,Title="Performance of memory reclamation for lockless synchronization"
+,journal="J. Parallel Distrib. Comput."
+,year="2007"
+,note="To appear in J. Parallel Distrib. Comput.
+       \url{doi=10.1016/j.jpdc.2007.04.010}"
+,annotation={
+	Compares QSBR (AKA "classic RCU"), HPBR, EBR, and lock-free
+	reference counting.  Journal version of ThomasEHart2006a.
+}
+}
+
+@unpublished{PaulEMcKenney2007QRCUspin
+,Author="Paul E. McKenney"
+,Title="Using Promela and Spin to verify parallel algorithms"
+,month="August"
+,day="1"
+,year="2007"
+,note="Available:
+\url{http://lwn.net/Articles/243851/}
+[Viewed September 8, 2007]"
+,annotation="
+	LWN article describing Promela and spin, and also using Oleg
+	Nesterov's QRCU as an example (with Paul McKenney's fastpath).
+"
+}
+
diff --git a/Documentation/RCU/rcu.txt b/Documentation/RCU/rcu.txt
index f84407c..95821a2 100644
--- a/Documentation/RCU/rcu.txt
+++ b/Documentation/RCU/rcu.txt
@@ -36,6 +36,14 @@ o	How can the updater tell when a grace period has completed
 	executed in user mode, or executed in the idle loop, we can
 	safely free up that item.
 
+	Preemptible variants of RCU (CONFIG_PREEMPT_RCU) get the
+	same effect, but require that the readers manipulate CPU-local
+	counters.  These counters allow limited types of blocking
+	within RCU read-side critical sections.  SRCU also uses
+	CPU-local counters, and permits general blocking within
+	RCU read-side critical sections.  These two variants of
+	RCU detect grace periods by sampling these counters.
+
 o	If I am running on a uniprocessor kernel, which can only do one
 	thing at a time, why should I wait for a grace period?
 
@@ -46,7 +54,10 @@ o	How can I see where RCU is currently used in the Linux kernel?
 	Search for "rcu_read_lock", "rcu_read_unlock", "call_rcu",
 	"rcu_read_lock_bh", "rcu_read_unlock_bh", "call_rcu_bh",
 	"srcu_read_lock", "srcu_read_unlock", "synchronize_rcu",
-	"synchronize_net", and "synchronize_srcu".
+	"synchronize_net", "synchronize_srcu", and the other RCU
+	primitives.  Or grab one of the cscope databases from:
+
+	http://www.rdrop.com/users/paulmck/RCU/linuxusage/rculocktab.html
 
 o	What guidelines should I follow when writing code that uses RCU?
 
@@ -67,7 +78,11 @@ o	I hear that RCU is patented?  What is with that?
 
 o	I hear that RCU needs work in order to support realtime kernels?
 
-	Yes, work in progress.
+	This work is largely completed.  Realtime-friendly RCU can be
+	enabled via the CONFIG_PREEMPT_RCU kernel configuration parameter.
+	However, work is in progress for enabling priority boosting of
+	preempted RCU read-side critical sections.This is needed if you
+	have CPU-bound realtime threads.
 
 o	Where can I find more information on RCU?
 
diff --git a/Documentation/RCU/torture.txt b/Documentation/RCU/torture.txt
index 25a3c3f..2967a65 100644
--- a/Documentation/RCU/torture.txt
+++ b/Documentation/RCU/torture.txt
@@ -46,12 +46,13 @@ stat_interval	The number of seconds between output of torture
 
 shuffle_interval
 		The number of seconds to keep the test threads affinitied
-		to a particular subset of the CPUs.  Used in conjunction
-		with test_no_idle_hz.
+		to a particular subset of the CPUs, defaults to 5 seconds.
+		Used in conjunction with test_no_idle_hz.
 
 test_no_idle_hz	Whether or not to test the ability of RCU to operate in
 		a kernel that disables the scheduling-clock interrupt to
 		idle CPUs.  Boolean parameter, "1" to test, "0" otherwise.
+		Defaults to omitting this test.
 
 torture_type	The type of RCU to test: "rcu" for the rcu_read_lock() API,
 		"rcu_sync" for rcu_read_lock() with synchronous reclamation,
@@ -82,8 +83,6 @@ be evident.  ;-)
 
 The entries are as follows:
 
-o	"ggp": The number of counter flips (or batches) since boot.
-
 o	"rtc": The hexadecimal address of the structure currently visible
 	to readers.
 
@@ -117,8 +116,8 @@ o	"Reader Pipe": Histogram of "ages" of structures seen by readers.
 o	"Reader Batch": Another histogram of "ages" of structures seen
 	by readers, but in terms of counter flips (or batches) rather
 	than in terms of grace periods.  The legal number of non-zero
-	entries is again two.  The reason for this separate view is
-	that it is easier to get the third entry to show up in the
+	entries is again two.  The reason for this separate view is that
+	it is sometimes easier to get the third entry to show up in the
 	"Reader Batch" list than in the "Reader Pipe" list.
 
 o	"Free-Block Circulation": Shows the number of torture structures
-- 
Gautham R Shenoy
Linux Technology Center
IBM India.
"Freedom comes with a price tag of responsibility, which is still a bargain,
because Freedom is priceless!"

Re: [RFC PATCH 0/6] RCU: Preemptible-RCU

[Gautham R Shenoy]

Hi Steve, 

On Thu, Dec 13, 2007 at 12:36:47PM -0500, Steven Rostedt wrote:
> On Thu, 13 Dec 2007, Gautham R Shenoy wrote:
> 
> >
> > Currently it is based against the latest linux-2.6-sched-devel.git
> >
> > Awaiting your feedback!
> 
> Hi Gautham,
> 
> Thanks for posting this. I believe this is the same version of preempt RCU
> as we have in the RT patch. It seems to be very stable. I ran the RT patch
> version of the RCU Preempt (just the Preempt RCU patches without RT on
> latest git) on a 64way box and the results seems just as good (if not
> slightly better) than classic RCU!  I'll rerun this patch series on that
> box and post the results.

The version of synchronize_sched() that's implemented in this patchset
requires these CPU-Hotplug patches -->
http://lkml.org/lkml/2007/11/15/239.

This is the reason why I rebased it against linux-2.6-sched-devel.git
since it already has those patches.

> 
> >From what I'm seening with this, is that it is ready for mainline. These
> patches should probably go into -mm and be ready for 2.6.25.  If Andrew
> wants to wait for my results, I'll run them tonight.
> 
> Thanks Gautham, Paul and Dipankar for all this great work!
> 

Anytime! That probably the only way I could teach myself how RCU works!

> -- Steve
> 

Thanks and Regards
gautham.
-- 
Gautham R Shenoy
Linux Technology Center
IBM India.
"Freedom comes with a price tag of responsibility, which is still a bargain,
because Freedom is priceless!"

Re: [RFC PATCH 0/6] RCU: Preemptible-RCU

[Paul E. McKenney]

On Thu, Dec 13, 2007 at 09:38:04PM +0100, Ingo Molnar wrote:
> 
> * Gautham R Shenoy <ego@in.ibm.com> wrote:
> 
> > Hello everyone,
> > 
> > This patchset is an updated version of the preemptible RCU patchset 
> > that Paul McKenney had posted it in September earlier this year that 
> > can be found here --> http://lkml.org/lkml/2007/9/10/213
> > 
> > This patchset incorporates the review comments from Oleg Nesterov and 
> > Steven Rostedt.
> > 
> > The testing report of the patchset is as follows:
> > ====================================================================
> > Patch-stack:  	2.6.23-rc3 + cpu-hotplug patches from 
> > 		http://lkml.org/lkml/2007/11/15/239 + Preempt-RCU
> > 		patches.
> > Test:		RCU-Torture running parallelly with CPU-Hotplug
> > 		operations.
> > Duration:	24 hours.
> > Architectures:	x86,x86_64, ppc64.
> > ====================================================================
> > 
> > 
> > Currently it is based against the latest linux-2.6-sched-devel.git
> > 
> > Awaiting your feedback!
> 
> thanks Gautham, the patchset from you and Paul looks good to me and i've 
> applied it to sched-devel.git to get it tested and reviewed some more.
> 
> from the Nitpicking Police, there are a couple of minor style 
> problems/warnings with the code, you can see it via:
> 
>   scripts/checkpatch.pl --file kernel/rcu*.c
> 
> nothing serious - RCU is still one of the cleanest subsystems in the 
> kernel:
> 
>                                       errors   lines of code   errors/KLOC
> kernel/rcuclassic.c                        0             575             0
> kernel/rcupdate.c                          1             138           7.2

This one is the exception to checkpatch.pl's rule against "volatile".  ;-)

The volatile declaration is within the ACCESS_ONCE() macro that is used
within rcu_read_lock() and rcu_read_unlock() to force the compiler to
maintain ordering with respect to interrupt handler running only on that
same CPU.

> kernel/rcupreempt.c                        0             953             0
> kernel/rcupreempt_trace.c                  0             330             0
> kernel/rcutorture.c                        8             995           8.0

Hmmm...  Definitely some old whitespace issues here...

> the eventual goal would be to match:
> 
>    scripts/checkpatch.pl --file kernel/sched*.[ch]
> 
> output ;-)

We should be able to make some progress in that direction.  ;-)

							Thanx, Paul

Re: [RFC PATCH 2/6] Preempt-RCU: Reorganize RCU code into rcuclassic.c and rcupdate.c

[Johannes Weiner]

Hi,

Gautham R Shenoy <ego@in.ibm.com> writes:

> diff --git a/kernel/rcuclassic.c b/kernel/rcuclassic.c
> new file mode 100644
> index 0000000..11c16aa
> --- /dev/null
> +++ b/kernel/rcuclassic.c
> +/**
> + * call_rcu - Queue an RCU callback for invocation after a grace period.
> + * @head: structure to be used for queueing the RCU updates.
> + * @func: actual update function to be invoked after the grace period
> + *
> + * The update function will be invoked some time after a full grace
> + * period elapses, in other words after all currently executing RCU
> + * read-side critical sections have completed.  RCU read-side critical
> + * sections are delimited by rcu_read_lock() and rcu_read_unlock(),
> + * and may be nested.
> + */
> +void call_rcu(struct rcu_head *head,
> +				void (*func)(struct rcu_head *rcu))
> +{
> +	unsigned long flags;
> +	struct rcu_data *rdp;
> +
> +	head->func = func;
> +	head->next = NULL;
> +	local_irq_save(flags);
> +	rdp = &__get_cpu_var(rcu_data);
> +	*rdp->nxttail = head;
> +	rdp->nxttail = &head->next;
> +	if (unlikely(++rdp->qlen > qhimark)) {
> +		rdp->blimit = INT_MAX;
> +		force_quiescent_state(rdp, &rcu_ctrlblk);
> +	}
> +	local_irq_restore(flags);
> +}
> +EXPORT_SYMBOL_GPL(call_rcu);
> +
> +/**
> + * call_rcu_bh - Queue an RCU for invocation after a quicker grace period.
> + * @head: structure to be used for queueing the RCU updates.
> + * @func: actual update function to be invoked after the grace period
> + *
> + * The update function will be invoked some time after a full grace
> + * period elapses, in other words after all currently executing RCU
> + * read-side critical sections have completed. call_rcu_bh() assumes
> + * that the read-side critical sections end on completion of a softirq
> + * handler. This means that read-side critical sections in process
> + * context must not be interrupted by softirqs. This interface is to be
> + * used when most of the read-side critical sections are in softirq context.
> + * RCU read-side critical sections are delimited by rcu_read_lock() and
> + * rcu_read_unlock(), * if in interrupt context or rcu_read_lock_bh()
> + * and rcu_read_unlock_bh(), if in process context. These may be nested.
> + */
> +void call_rcu_bh(struct rcu_head *head,
> +				void (*func)(struct rcu_head *rcu))
> +{
> +	unsigned long flags;
> +	struct rcu_data *rdp;
> +
> +	head->func = func;
> +	head->next = NULL;
> +	local_irq_save(flags);
> +	rdp = &__get_cpu_var(rcu_bh_data);
> +	*rdp->nxttail = head;
> +	rdp->nxttail = &head->next;
> +
> +	if (unlikely(++rdp->qlen > qhimark)) {
> +		rdp->blimit = INT_MAX;
> +		force_quiescent_state(rdp, &rcu_bh_ctrlblk);
> +	}
> +
> +	local_irq_restore(flags);
> +}
> +EXPORT_SYMBOL_GPL(call_rcu_bh);

Those two functions are identical beside the difference of `rcu_data'
<-> `rcu_bh_data' and `rcu_ctrlblk' <-> `rcu_bh_ctrlblk'.

Is there a way to collapse the code into one helper function or would
that effect performance too much?

	Hannes

Re: [RFC PATCH 2/6] Preempt-RCU: Reorganize RCU code into rcuclassic.c and rcupdate.c

[Paul E. McKenney]

On Fri, Dec 14, 2007 at 03:51:14PM +0100, Johannes Weiner wrote:
> Hi,
> 
> Gautham R Shenoy <ego@in.ibm.com> writes:
> 
> > diff --git a/kernel/rcuclassic.c b/kernel/rcuclassic.c
> > new file mode 100644
> > index 0000000..11c16aa
> > --- /dev/null
> > +++ b/kernel/rcuclassic.c
> > +/**
> > + * call_rcu - Queue an RCU callback for invocation after a grace period.
> > + * @head: structure to be used for queueing the RCU updates.
> > + * @func: actual update function to be invoked after the grace period
> > + *
> > + * The update function will be invoked some time after a full grace
> > + * period elapses, in other words after all currently executing RCU
> > + * read-side critical sections have completed.  RCU read-side critical
> > + * sections are delimited by rcu_read_lock() and rcu_read_unlock(),
> > + * and may be nested.
> > + */
> > +void call_rcu(struct rcu_head *head,
> > +				void (*func)(struct rcu_head *rcu))
> > +{
> > +	unsigned long flags;
> > +	struct rcu_data *rdp;
> > +
> > +	head->func = func;
> > +	head->next = NULL;
> > +	local_irq_save(flags);
> > +	rdp = &__get_cpu_var(rcu_data);
> > +	*rdp->nxttail = head;
> > +	rdp->nxttail = &head->next;
> > +	if (unlikely(++rdp->qlen > qhimark)) {
> > +		rdp->blimit = INT_MAX;
> > +		force_quiescent_state(rdp, &rcu_ctrlblk);
> > +	}
> > +	local_irq_restore(flags);
> > +}
> > +EXPORT_SYMBOL_GPL(call_rcu);
> > +
> > +/**
> > + * call_rcu_bh - Queue an RCU for invocation after a quicker grace period.
> > + * @head: structure to be used for queueing the RCU updates.
> > + * @func: actual update function to be invoked after the grace period
> > + *
> > + * The update function will be invoked some time after a full grace
> > + * period elapses, in other words after all currently executing RCU
> > + * read-side critical sections have completed. call_rcu_bh() assumes
> > + * that the read-side critical sections end on completion of a softirq
> > + * handler. This means that read-side critical sections in process
> > + * context must not be interrupted by softirqs. This interface is to be
> > + * used when most of the read-side critical sections are in softirq context.
> > + * RCU read-side critical sections are delimited by rcu_read_lock() and
> > + * rcu_read_unlock(), * if in interrupt context or rcu_read_lock_bh()
> > + * and rcu_read_unlock_bh(), if in process context. These may be nested.
> > + */
> > +void call_rcu_bh(struct rcu_head *head,
> > +				void (*func)(struct rcu_head *rcu))
> > +{
> > +	unsigned long flags;
> > +	struct rcu_data *rdp;
> > +
> > +	head->func = func;
> > +	head->next = NULL;
> > +	local_irq_save(flags);
> > +	rdp = &__get_cpu_var(rcu_bh_data);
> > +	*rdp->nxttail = head;
> > +	rdp->nxttail = &head->next;
> > +
> > +	if (unlikely(++rdp->qlen > qhimark)) {
> > +		rdp->blimit = INT_MAX;
> > +		force_quiescent_state(rdp, &rcu_bh_ctrlblk);
> > +	}
> > +
> > +	local_irq_restore(flags);
> > +}
> > +EXPORT_SYMBOL_GPL(call_rcu_bh);
> 
> Those two functions are identical beside the difference of `rcu_data'
> <-> `rcu_bh_data' and `rcu_ctrlblk' <-> `rcu_bh_ctrlblk'.
> 
> Is there a way to collapse the code into one helper function or would
> that effect performance too much?

This code is performance critical, but on the other hand, perhaps gcc
has gotten smarter in the years since this was written.  (This patch is
just moving this code from one file to another, rather than writing it
from scratch.)  Would current gcc implementations be able to do the
inlining required?

							Thanx, Paul

Re: [RFC PATCH 4/6] Preempt-RCU: Implementation

[Roman Zippel]

Hi,

On Thursday 13. December 2007, Gautham R Shenoy wrote:

> diff --git a/kernel/Kconfig.preempt b/kernel/Kconfig.preempt
> index c64ce9c..06cafcc 100644
> --- a/kernel/Kconfig.preempt
> +++ b/kernel/Kconfig.preempt
> @@ -63,3 +63,41 @@ config PREEMPT_BKL
>  	  Say Y here if you are building a kernel for a desktop system.
>  	  Say N if you are unsure.
>
> +choice
> +	prompt "RCU implementation type:"
> +	default CLASSIC_RCU
> +
> +config CLASSIC_RCU
> +	bool "Classic RCU"
> +	help
> +	  This option selects the classic RCU implementation that is
> +	  designed for best read-side performance on non-realtime
> +	  systems.
> +
> +	  Say Y if you are unsure.
> +
> +config PREEMPT_RCU
> +	bool "Preemptible RCU"
> +	depends on PREEMPT
> +	help
> +	  This option reduces the latency of the kernel by making certain
> +	  RCU sections preemptible. Normally RCU code is non-preemptible, if
> +	  this option is selected then read-only RCU sections become
> +	  preemptible. This helps latency, but may expose bugs due to
> +	  now-naive assumptions about each RCU read-side critical section
> +	  remaining on a given CPU through its execution.
> +
> +	  Say N if you are unsure.
> +
> +endchoice

Why got this moved into init/Kconfig?
Now it's somewhere in the root menu, not really belonging to anything.
Also why is this a choice? Are more RCU types planned?

bye, Roman

Re: [RFC PATCH 4/6] Preempt-RCU: Implementation

[Paul E. McKenney]

On Fri, Feb 29, 2008 at 05:34:54AM +0100, Roman Zippel wrote:
> Hi,
> 
> On Thursday 13. December 2007, Gautham R Shenoy wrote:
> 
> > diff --git a/kernel/Kconfig.preempt b/kernel/Kconfig.preempt
> > index c64ce9c..06cafcc 100644
> > --- a/kernel/Kconfig.preempt
> > +++ b/kernel/Kconfig.preempt
> > @@ -63,3 +63,41 @@ config PREEMPT_BKL
> >  	  Say Y here if you are building a kernel for a desktop system.
> >  	  Say N if you are unsure.
> >
> > +choice
> > +	prompt "RCU implementation type:"
> > +	default CLASSIC_RCU
> > +
> > +config CLASSIC_RCU
> > +	bool "Classic RCU"
> > +	help
> > +	  This option selects the classic RCU implementation that is
> > +	  designed for best read-side performance on non-realtime
> > +	  systems.
> > +
> > +	  Say Y if you are unsure.
> > +
> > +config PREEMPT_RCU
> > +	bool "Preemptible RCU"
> > +	depends on PREEMPT
> > +	help
> > +	  This option reduces the latency of the kernel by making certain
> > +	  RCU sections preemptible. Normally RCU code is non-preemptible, if
> > +	  this option is selected then read-only RCU sections become
> > +	  preemptible. This helps latency, but may expose bugs due to
> > +	  now-naive assumptions about each RCU read-side critical section
> > +	  remaining on a given CPU through its execution.
> > +
> > +	  Say N if you are unsure.
> > +
> > +endchoice
> 
> Why got this moved into init/Kconfig?

Because there are some arches that don't have kernel/Kconfig.preempt,
its earlier home.  Therefore, putting it into kernel/Kconfig.preempt
broke those arches' builds by supplying neither PREEMPT_RCU nor
CLASSIC_RCU.

> Now it's somewhere in the root menu, not really belonging to anything.

Do you have a suggested location?

> Also why is this a choice? Are more RCU types planned?

I don't expect additional drop-in replacements for RCU, though people
are certainly free to experiment if they wish.  It is a choice because
this gives people a very clear idea of the two options and because
it makes the implementation a bit cleaner.

							Thanx, Paul

Re: [RFC PATCH 4/6] Preempt-RCU: Implementation

[Roman Zippel]

Hi

On Thu, 28 Feb 2008, Paul E. McKenney wrote:

> > Why got this moved into init/Kconfig?
> 
> Because there are some arches that don't have kernel/Kconfig.preempt,
> its earlier home.  Therefore, putting it into kernel/Kconfig.preempt
> broke those arches' builds by supplying neither PREEMPT_RCU nor
> CLASSIC_RCU.
> 
> > Now it's somewhere in the root menu, not really belonging to anything.
> 
> Do you have a suggested location?
> 
> > Also why is this a choice? Are more RCU types planned?
> 
> I don't expect additional drop-in replacements for RCU, though people
> are certainly free to experiment if they wish.  It is a choice because
> this gives people a very clear idea of the two options and because
> it makes the implementation a bit cleaner.

I'd suggest to move PREEMPT_RCU back to Kconfig.preempt and if you really 
need the second symbol leave this behind (maybe with a comment):

config CLASSIC_RCU
	def_bool !PREEMPT_RCU

Once there are more options, we can still look for a better place...

Also could you please add a proper dependency to RCU_TRACE on PREEMPT_RCU, 
so that this condition isn't needed anymore:

ifeq ($(CONFIG_PREEMPT_RCU),y)
obj-$(CONFIG_RCU_TRACE) += rcupreempt_trace.o
endif

Thanks.

bye, Roman

Re: [RFC PATCH 4/6] Preempt-RCU: Implementation

[Steven Rostedt]

On Fri, 29 Feb 2008, Roman Zippel wrote:
>
> Also why is this a choice? Are more RCU types planned?

Why shouldn't it be a choice? You can have CLASSIC_RCU or PREEMPT_RCU, one
or the other and not both. Sounds perfect for being a choice.

-- Steve

Re: [RFC PATCH 4/6] Preempt-RCU: Implementation

[Steven Rostedt]

On Fri, 29 Feb 2008, Roman Zippel wrote:
>
> I'd suggest to move PREEMPT_RCU back to Kconfig.preempt and if you really
> need the second symbol leave this behind (maybe with a comment):
>
> config CLASSIC_RCU
> 	def_bool !PREEMPT_RCU

Ah, OK, you are saying that we should only show the option for
PREEMPT_RCU? If selected, we use that and if not, we default to
CLASSIC_RCU.

I guess that can work.

-- Steve

>
> Once there are more options, we can still look for a better place...

Re: [RFC PATCH 4/6] Preempt-RCU: Implementation

[Roman Zippel]

Hi,

On Fri, 29 Feb 2008, Steven Rostedt wrote:

> > Also why is this a choice? Are more RCU types planned?
> 
> Why shouldn't it be a choice? You can have CLASSIC_RCU or PREEMPT_RCU, one
> or the other and not both. Sounds perfect for being a choice.

With this logic almost every bool could be implemented via choice.
A simple bool perfectly gives you two choices too.

bye, Roman

Re: [RFC PATCH 4/6] Preempt-RCU: Implementation

[Paul E. McKenney]

On Fri, Feb 29, 2008 at 01:38:15PM +0100, Roman Zippel wrote:
> Hi
> 
> On Thu, 28 Feb 2008, Paul E. McKenney wrote:
> 
> > > Why got this moved into init/Kconfig?
> > 
> > Because there are some arches that don't have kernel/Kconfig.preempt,
> > its earlier home.  Therefore, putting it into kernel/Kconfig.preempt
> > broke those arches' builds by supplying neither PREEMPT_RCU nor
> > CLASSIC_RCU.
> > 
> > > Now it's somewhere in the root menu, not really belonging to anything.
> > 
> > Do you have a suggested location?
> > 
> > > Also why is this a choice? Are more RCU types planned?
> > 
> > I don't expect additional drop-in replacements for RCU, though people
> > are certainly free to experiment if they wish.  It is a choice because
> > this gives people a very clear idea of the two options and because
> > it makes the implementation a bit cleaner.
> 
> I'd suggest to move PREEMPT_RCU back to Kconfig.preempt and if you really 
> need the second symbol leave this behind (maybe with a comment):
> 
> config CLASSIC_RCU
> 	def_bool !PREEMPT_RCU
> 
> Once there are more options, we can still look for a better place...
> 
> Also could you please add a proper dependency to RCU_TRACE on PREEMPT_RCU, 
> so that this condition isn't needed anymore:
> 
> ifeq ($(CONFIG_PREEMPT_RCU),y)
> obj-$(CONFIG_RCU_TRACE) += rcupreempt_trace.o
> endif

Is this what you had in mind?  I don't have any way to test on a
system not supporting CONFIG_PREEMPT, but seems to work on x86.

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

 init/Kconfig           |   34 +++-------------------------------
 kernel/Kconfig.preempt |   15 +++++++++++++++
 2 files changed, 18 insertions(+), 31 deletions(-)

diff -urpNa -X dontdiff linux-2.6.25-rc3/init/Kconfig linux-2.6.25-rc3-preempt_rcu/init/Kconfig
--- linux-2.6.25-rc3/init/Kconfig	2008-02-26 16:58:42.000000000 -0800
+++ linux-2.6.25-rc3-preempt_rcu/init/Kconfig	2008-03-01 11:30:59.000000000 -0800
@@ -860,38 +860,10 @@ source "block/Kconfig"
 config PREEMPT_NOTIFIERS
 	bool
 
-choice
-	prompt "RCU implementation type:"
-	default CLASSIC_RCU
-	help
-	  This allows you to choose either the classic RCU implementation
-	  that is designed for best read-side performance on non-realtime
-	  systems, or the preemptible RCU implementation for best latency
-	  on realtime systems.  Note that some kernel preemption modes
-	  will restrict your choice.
-
-	  Select the default if you are unsure.
-
 config CLASSIC_RCU
-	bool "Classic RCU"
+	def_bool !PREEMPT_RCU
 	help
 	  This option selects the classic RCU implementation that is
 	  designed for best read-side performance on non-realtime
-	  systems.
-
-	  Say Y if you are unsure.
-
-config PREEMPT_RCU
-	bool "Preemptible RCU"
-	depends on PREEMPT
-	help
-	  This option reduces the latency of the kernel by making certain
-	  RCU sections preemptible. Normally RCU code is non-preemptible, if
-	  this option is selected then read-only RCU sections become
-	  preemptible. This helps latency, but may expose bugs due to
-	  now-naive assumptions about each RCU read-side critical section
-	  remaining on a given CPU through its execution.
-
-	  Say N if you are unsure.
-
-endchoice
+	  systems.  Classic RCU is the default.  Note that the
+	  PREEMPT_RCU symbol is used to select/deselect this option.
diff -urpNa -X dontdiff linux-2.6.25-rc3/kernel/Kconfig.preempt linux-2.6.25-rc3-preempt_rcu/kernel/Kconfig.preempt
--- linux-2.6.25-rc3/kernel/Kconfig.preempt	2008-02-26 16:58:42.000000000 -0800
+++ linux-2.6.25-rc3-preempt_rcu/kernel/Kconfig.preempt	2008-03-01 11:35:39.000000000 -0800
@@ -52,8 +52,23 @@ config PREEMPT
 
 endchoice
 
+config PREEMPT_RCU
+	bool "Preemptible RCU"
+	depends on PREEMPT
+	default n
+	help
+	  This option reduces the latency of the kernel by making certain
+	  RCU sections preemptible. Normally RCU code is non-preemptible, if
+	  this option is selected then read-only RCU sections become
+	  preemptible. This helps latency, but may expose bugs due to
+	  now-naive assumptions about each RCU read-side critical section
+	  remaining on a given CPU through its execution.
+
+	  Say N if you are unsure.
+
 config RCU_TRACE
 	bool "Enable tracing for RCU - currently stats in debugfs"
+	depends on PREEMPT_RCU
 	select DEBUG_FS
 	default y
 	help

Re: [RFC PATCH 4/6] Preempt-RCU: Implementation

[Steven Rostedt]

On Sat, 1 Mar 2008, Paul E. McKenney wrote:
> Is this what you had in mind?  I don't have any way to test on a
> system not supporting CONFIG_PREEMPT, but seems to work on x86.
>
> Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
> ---
>
>  init/Kconfig           |   34 +++-------------------------------
>  kernel/Kconfig.preempt |   15 +++++++++++++++
>  2 files changed, 18 insertions(+), 31 deletions(-)
>
> diff -urpNa -X dontdiff linux-2.6.25-rc3/init/Kconfig linux-2.6.25-rc3-preempt_rcu/init/Kconfig
> --- linux-2.6.25-rc3/init/Kconfig	2008-02-26 16:58:42.000000000 -0800
> +++ linux-2.6.25-rc3-preempt_rcu/init/Kconfig	2008-03-01 11:30:59.000000000 -0800
> @@ -860,38 +860,10 @@ source "block/Kconfig"
>  config PREEMPT_NOTIFIERS
>  	bool
>
> -choice
> -	prompt "RCU implementation type:"
> -	default CLASSIC_RCU
> -	help
> -	  This allows you to choose either the classic RCU implementation
> -	  that is designed for best read-side performance on non-realtime
> -	  systems, or the preemptible RCU implementation for best latency
> -	  on realtime systems.  Note that some kernel preemption modes
> -	  will restrict your choice.
> -
> -	  Select the default if you are unsure.
> -
>  config CLASSIC_RCU
> -	bool "Classic RCU"
> +	def_bool !PREEMPT_RCU
>  	help
>  	  This option selects the classic RCU implementation that is
>  	  designed for best read-side performance on non-realtime
> -	  systems.

Actually Paul, you don't need to make this an "option". Simply the default
if PREEMPT_RCU is not selected. In otherwords, no prompt.

	config CLASSIC_RCU
	  bool
	  depends on !PREEMPT_RCU
	  default y

Then, have the PREEMPT_RCU defined down below (as you already did), and
have it be the only option. That is, you can have normal RCU or select
to have PREEMPT_RCU.

If the user selects PREEMPT_RCU, CLASSIC_RCU will be unselected.

-- Steve

> -
> -	  Say Y if you are unsure.
> -
> -config PREEMPT_RCU
> -	bool "Preemptible RCU"
> -	depends on PREEMPT
> -	help
> -	  This option reduces the latency of the kernel by making certain
> -	  RCU sections preemptible. Normally RCU code is non-preemptible, if
> -	  this option is selected then read-only RCU sections become
> -	  preemptible. This helps latency, but may expose bugs due to
> -	  now-naive assumptions about each RCU read-side critical section
> -	  remaining on a given CPU through its execution.
> -
> -	  Say N if you are unsure.
> -
> -endchoice
> +	  systems.  Classic RCU is the default.  Note that the
> +	  PREEMPT_RCU symbol is used to select/deselect this option.
> diff -urpNa -X dontdiff linux-2.6.25-rc3/kernel/Kconfig.preempt linux-2.6.25-rc3-preempt_rcu/kernel/Kconfig.preempt
> --- linux-2.6.25-rc3/kernel/Kconfig.preempt	2008-02-26 16:58:42.000000000 -0800
> +++ linux-2.6.25-rc3-preempt_rcu/kernel/Kconfig.preempt	2008-03-01 11:35:39.000000000 -0800
> @@ -52,8 +52,23 @@ config PREEMPT
>
>  endchoice
>
> +config PREEMPT_RCU
> +	bool "Preemptible RCU"
> +	depends on PREEMPT
> +	default n
> +	help
> +	  This option reduces the latency of the kernel by making certain
> +	  RCU sections preemptible. Normally RCU code is non-preemptible, if
> +	  this option is selected then read-only RCU sections become
> +	  preemptible. This helps latency, but may expose bugs due to
> +	  now-naive assumptions about each RCU read-side critical section
> +	  remaining on a given CPU through its execution.
> +
> +	  Say N if you are unsure.
> +
>  config RCU_TRACE
>  	bool "Enable tracing for RCU - currently stats in debugfs"
> +	depends on PREEMPT_RCU
>  	select DEBUG_FS
>  	default y
>  	help
>

Re: [RFC PATCH 4/6] Preempt-RCU: Implementation

[Roman Zippel]

Hi,

On Sat, 1 Mar 2008, Paul E. McKenney wrote:

> Is this what you had in mind?  I don't have any way to test on a
> system not supporting CONFIG_PREEMPT, but seems to work on x86.

Yes, looks fine.

> +config PREEMPT_RCU
> +	bool "Preemptible RCU"
> +	depends on PREEMPT
> +	default n

"default n" isn't really necessary, it's already the default.

bye, Roman

Re: [RFC PATCH 4/6] Preempt-RCU: Implementation

[Roman Zippel]

Hi,

On Sat, 1 Mar 2008, Steven Rostedt wrote:

> >  config CLASSIC_RCU
> > -	bool "Classic RCU"
> > +	def_bool !PREEMPT_RCU
> >  	help
> >  	  This option selects the classic RCU implementation that is
> >  	  designed for best read-side performance on non-realtime
> > -	  systems.
> 
> Actually Paul, you don't need to make this an "option". Simply the default
> if PREEMPT_RCU is not selected. In otherwords, no prompt.
> 
> 	config CLASSIC_RCU
> 	  bool
> 	  depends on !PREEMPT_RCU
> 	  default y

There is no prompt. :)
"def_bool !PREEMPT_RCU" does the same as this.

bye, Roman

Re: [RFC PATCH 4/6] Preempt-RCU: Implementation

[Paul E. McKenney]

On Sun, Mar 02, 2008 at 04:06:10AM +0100, Roman Zippel wrote:
> Hi,
> 
> On Sat, 1 Mar 2008, Paul E. McKenney wrote:
> 
> > Is this what you had in mind?  I don't have any way to test on a
> > system not supporting CONFIG_PREEMPT, but seems to work on x86.
> 
> Yes, looks fine.
> 
> > +config PREEMPT_RCU
> > +	bool "Preemptible RCU"
> > +	depends on PREEMPT
> > +	default n
> 
> "default n" isn't really necessary, it's already the default.

Fair enough.  But something like 125 Kconfig files in 2.6.25-rc3 have
at least one "default n" in them, so is it worth getting rid of it?
Seems to me that the explicit "default n" has some substantial readability
advantages.

						Thanx, Paul

[PATCH] move PREEMPT_RCU config option back under PREEMPT

[Paul E. McKenney]

Hello!

The original preemptible-RCU patch put the choice between classic and
preemptible RCU into kernel/Kconfig.preempt, which resulted in build
failures on machines not supporting CONFIG_PREEMPT.  This choice was
therefore moved to init/Kconfig, which worked, but placed the choice
between classic and preemptible RCU at the top level, a very obtuse choice
indeed.  This patch changes from the Kconfig "choice" mechanism to a pair
of booleans, only one of which (CONFIG_PREEMPT_RCU) is user-visible,
and is located in kernel/Kconfig.preempt, where one would expect it
to be.  The other (CONFIG_CLASSIC_RCU) is in init/Kconfig so that it
is available to all architectures, hopefully avoiding build breakage.
Thanks to Roman Zippel for suggesting this approach.

I have tested this, but sadly do not have access to a machine that does
not support CONFIG_PREEMPT.  However, I did edit my config in an attempt
to simulate this situation.

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

 init/Kconfig           |   34 +++-------------------------------
 kernel/Kconfig.preempt |   15 +++++++++++++++
 2 files changed, 18 insertions(+), 31 deletions(-)

diff -urpNa -X dontdiff linux-2.6.25-rc3/init/Kconfig linux-2.6.25-rc3-preempt_rcu/init/Kconfig
--- linux-2.6.25-rc3/init/Kconfig	2008-02-26 16:58:42.000000000 -0800
+++ linux-2.6.25-rc3-preempt_rcu/init/Kconfig	2008-03-01 11:30:59.000000000 -0800
@@ -860,38 +860,10 @@ source "block/Kconfig"
 config PREEMPT_NOTIFIERS
 	bool
 
-choice
-	prompt "RCU implementation type:"
-	default CLASSIC_RCU
-	help
-	  This allows you to choose either the classic RCU implementation
-	  that is designed for best read-side performance on non-realtime
-	  systems, or the preemptible RCU implementation for best latency
-	  on realtime systems.  Note that some kernel preemption modes
-	  will restrict your choice.
-
-	  Select the default if you are unsure.
-
 config CLASSIC_RCU
-	bool "Classic RCU"
+	def_bool !PREEMPT_RCU
 	help
 	  This option selects the classic RCU implementation that is
 	  designed for best read-side performance on non-realtime
-	  systems.
-
-	  Say Y if you are unsure.
-
-config PREEMPT_RCU
-	bool "Preemptible RCU"
-	depends on PREEMPT
-	help
-	  This option reduces the latency of the kernel by making certain
-	  RCU sections preemptible. Normally RCU code is non-preemptible, if
-	  this option is selected then read-only RCU sections become
-	  preemptible. This helps latency, but may expose bugs due to
-	  now-naive assumptions about each RCU read-side critical section
-	  remaining on a given CPU through its execution.
-
-	  Say N if you are unsure.
-
-endchoice
+	  systems.  Classic RCU is the default.  Note that the
+	  PREEMPT_RCU symbol is used to select/deselect this option.
diff -urpNa -X dontdiff linux-2.6.25-rc3/kernel/Kconfig.preempt linux-2.6.25-rc3-preempt_rcu/kernel/Kconfig.preempt
--- linux-2.6.25-rc3/kernel/Kconfig.preempt	2008-02-26 16:58:42.000000000 -0800
+++ linux-2.6.25-rc3-preempt_rcu/kernel/Kconfig.preempt	2008-03-01 11:35:39.000000000 -0800
@@ -52,8 +52,23 @@ config PREEMPT
 
 endchoice
 
+config PREEMPT_RCU
+	bool "Preemptible RCU"
+	depends on PREEMPT
+	default n
+	help
+	  This option reduces the latency of the kernel by making certain
+	  RCU sections preemptible. Normally RCU code is non-preemptible, if
+	  this option is selected then read-only RCU sections become
+	  preemptible. This helps latency, but may expose bugs due to
+	  now-naive assumptions about each RCU read-side critical section
+	  remaining on a given CPU through its execution.
+
+	  Say N if you are unsure.
+
 config RCU_TRACE
 	bool "Enable tracing for RCU - currently stats in debugfs"
+	depends on PREEMPT_RCU
 	select DEBUG_FS
 	default y
 	help

Re: [RFC PATCH 4/6] Preempt-RCU: Implementation

[Roman Zippel]

Hi,

On Mon, 3 Mar 2008, Paul E. McKenney wrote:

> > "default n" isn't really necessary, it's already the default.
> 
> Fair enough.  But something like 125 Kconfig files in 2.6.25-rc3 have
> at least one "default n" in them, so is it worth getting rid of it?
> Seems to me that the explicit "default n" has some substantial readability
> advantages.

The inverse would mean all the other configs have a readability 
disadvantage.
In most cases they can be simply removed, only in form of 'def_bool n' it
makes somewhat sense.

bye, Roman

Re: [PATCH] move PREEMPT_RCU config option back under PREEMPT

[Steven Rostedt]

On Tue, 4 Mar 2008, Paul E. McKenney wrote:
>
> The original preemptible-RCU patch put the choice between classic and
> preemptible RCU into kernel/Kconfig.preempt, which resulted in build
> failures on machines not supporting CONFIG_PREEMPT.  This choice was
> therefore moved to init/Kconfig, which worked, but placed the choice
> between classic and preemptible RCU at the top level, a very obtuse choice
> indeed.  This patch changes from the Kconfig "choice" mechanism to a pair
> of booleans, only one of which (CONFIG_PREEMPT_RCU) is user-visible,
> and is located in kernel/Kconfig.preempt, where one would expect it
> to be.  The other (CONFIG_CLASSIC_RCU) is in init/Kconfig so that it
> is available to all architectures, hopefully avoiding build breakage.
> Thanks to Roman Zippel for suggesting this approach.
>
> I have tested this, but sadly do not have access to a machine that does
> not support CONFIG_PREEMPT.  However, I did edit my config in an attempt
> to simulate this situation.

One minor comment, but otherwise:

Acked-by: Steven Rostedt <srostedt@redhat.com>

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


>  config CLASSIC_RCU
> -	bool "Classic RCU"
> +	def_bool !PREEMPT_RCU
>  	help
>  	  This option selects the classic RCU implementation that is
>  	  designed for best read-side performance on non-realtime
> -	  systems.
> -
> -	  Say Y if you are unsure.
> -

You can get rid of the "help" part since it isn't visible to users.

-- Steve

Re: [PATCH] move PREEMPT_RCU config option back under PREEMPT

[Paul E. McKenney]

On Tue, Mar 04, 2008 at 03:55:42PM -0500, Steven Rostedt wrote:
> 
> On Tue, 4 Mar 2008, Paul E. McKenney wrote:
> >
> > The original preemptible-RCU patch put the choice between classic and
> > preemptible RCU into kernel/Kconfig.preempt, which resulted in build
> > failures on machines not supporting CONFIG_PREEMPT.  This choice was
> > therefore moved to init/Kconfig, which worked, but placed the choice
> > between classic and preemptible RCU at the top level, a very obtuse choice
> > indeed.  This patch changes from the Kconfig "choice" mechanism to a pair
> > of booleans, only one of which (CONFIG_PREEMPT_RCU) is user-visible,
> > and is located in kernel/Kconfig.preempt, where one would expect it
> > to be.  The other (CONFIG_CLASSIC_RCU) is in init/Kconfig so that it
> > is available to all architectures, hopefully avoiding build breakage.
> > Thanks to Roman Zippel for suggesting this approach.
> >
> > I have tested this, but sadly do not have access to a machine that does
> > not support CONFIG_PREEMPT.  However, I did edit my config in an attempt
> > to simulate this situation.
> 
> One minor comment, but otherwise:
> 
> Acked-by: Steven Rostedt <srostedt@redhat.com>
> 
> >
> > Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
> 
> 
> >  config CLASSIC_RCU
> > -	bool "Classic RCU"
> > +	def_bool !PREEMPT_RCU
> >  	help
> >  	  This option selects the classic RCU implementation that is
> >  	  designed for best read-side performance on non-realtime
> > -	  systems.
> > -
> > -	  Say Y if you are unsure.
> > -
> 
> You can get rid of the "help" part since it isn't visible to users.

So how about if I replace it with comment lines (starting with "#",
not with "comment")?

						Thanx, Paul

Re: [PATCH] move PREEMPT_RCU config option back under PREEMPT

[Roman Zippel]

Hi,

On Tue, 4 Mar 2008, Paul E. McKenney wrote:

> > > -	bool "Classic RCU"
> > > +	def_bool !PREEMPT_RCU
> > >  	help
> > >  	  This option selects the classic RCU implementation that is
> > >  	  designed for best read-side performance on non-realtime
> > > -	  systems.
> > > -
> > > -	  Say Y if you are unsure.
> > > -
> > 
> > You can get rid of the "help" part since it isn't visible to users.
> 
> So how about if I replace it with comment lines (starting with "#",
> not with "comment")?

Actually xconfig can display this text, so using the help like this is 
fine.

bye, Roman

Re: [PATCH] move PREEMPT_RCU config option back under PREEMPT

[Paul E. McKenney]

On Wed, Mar 05, 2008 at 03:06:32AM +0100, Roman Zippel wrote:
> Hi,
> 
> On Tue, 4 Mar 2008, Paul E. McKenney wrote:
> 
> > > > -	bool "Classic RCU"
> > > > +	def_bool !PREEMPT_RCU
> > > >  	help
> > > >  	  This option selects the classic RCU implementation that is
> > > >  	  designed for best read-side performance on non-realtime
> > > > -	  systems.
> > > > -
> > > > -	  Say Y if you are unsure.
> > > > -
> > > 
> > > You can get rid of the "help" part since it isn't visible to users.
> > 
> > So how about if I replace it with comment lines (starting with "#",
> > not with "comment")?
> 
> Actually xconfig can display this text, so using the help like this is 
> fine.

Cool, I will stick with the "help" clause, then.

							Thanx, Paul