diff --git a/include/linux/rcuclassic.h b/include/linux/rcuclassic.h
index 8c77490..04c7281 100644
--- a/include/linux/rcuclassic.h
+++ b/include/linux/rcuclassic.h
@@ -45,7 +45,7 @@
 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?         */
+	long	pending;	/* Number of the last pending batch           */
 
 	int	signaled;
 
@@ -66,11 +66,7 @@
 	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
- */
+/* Per-CPU data for Read-Copy UPdate. */
 struct rcu_data {
 	/* 1) quiescent state handling : */
 	long		quiescbatch;     /* Batch # for grace period */
@@ -78,12 +74,24 @@
 	int		qs_pending;	 /* core waits for quiesc state */
 
 	/* 2) batch handling */
-	long  	       	batch;           /* Batch # for current RCU batch */
+	/*
+	 * if nxtlist is not NULL, then:
+	 * batch:
+	 *	The batch # for the last entry of nxtlist
+	 * [*nxttail[1], NULL = *nxttail[2]):
+	 *	Entries that batch # <= batch
+	 * [*nxttail[0], *nxttail[1]):
+	 *	Entries that batch # <= batch - 1
+	 * [nxtlist, *nxttail[0]):
+	 *	Entries that batch # <= batch - 2
+	 *	The grace period for these entries has completed, and
+	 *	the other grace-period-completed entries may be moved
+	 *	here temporarily in rcu_process_callbacks().
+	 */
+	long  	       	batch;
 	struct rcu_head *nxtlist;
-	struct rcu_head **nxttail;
+	struct rcu_head **nxttail[3];
 	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 */
diff --git a/kernel/rcuclassic.c b/kernel/rcuclassic.c
index aad93cd..d427114 100644
--- a/kernel/rcuclassic.c
+++ b/kernel/rcuclassic.c
@@ -60,12 +60,14 @@
 static struct rcu_ctrlblk rcu_ctrlblk = {
 	.cur = -300,
 	.completed = -300,
+	.pending = -300,
 	.lock = __SPIN_LOCK_UNLOCKED(&rcu_ctrlblk.lock),
 	.cpumask = CPU_MASK_NONE,
 };
 static struct rcu_ctrlblk rcu_bh_ctrlblk = {
 	.cur = -300,
 	.completed = -300,
+	.pending = -300,
 	.lock = __SPIN_LOCK_UNLOCKED(&rcu_bh_ctrlblk.lock),
 	.cpumask = CPU_MASK_NONE,
 };
@@ -118,6 +120,43 @@
 }
 #endif
 
+static void __call_rcu(struct rcu_head *head, struct rcu_ctrlblk *rcp,
+		struct rcu_data *rdp)
+{
+	long batch;
+	smp_mb(); /* reads the most recently updated value of rcu->cur. */
+
+	/*
+	 * Determine the batch number of this callback.
+	 *
+	 * Using ACCESS_ONCE to avoid the following error when gcc eliminates
+	 * local variable "batch" and emits codes like this:
+	 *	1) rdp->batch = rcp->cur + 1 # gets old value
+	 *	......
+	 *	2)rcu_batch_after(rcp->cur + 1, rdp->batch) # gets new value
+	 * then [*nxttail[0], *nxttail[1]) may contain callbacks
+	 * that batch# = rdp->batch, see the comment of struct rcu_data.
+	 */
+	batch = ACCESS_ONCE(rcp->cur) + 1;
+
+	if (rdp->nxtlist && rcu_batch_after(batch, rdp->batch)) {
+		/* process callbacks */
+		rdp->nxttail[0] = rdp->nxttail[1];
+		rdp->nxttail[1] = rdp->nxttail[2];
+		if (rcu_batch_after(batch - 1, rdp->batch))
+			rdp->nxttail[0] = rdp->nxttail[2];
+	}
+
+	rdp->batch = batch;
+	*rdp->nxttail[2] = head;
+	rdp->nxttail[2] = &head->next;
+
+	if (unlikely(++rdp->qlen > qhimark)) {
+		rdp->blimit = INT_MAX;
+		force_quiescent_state(rdp, &rcu_ctrlblk);
+	}
+}
+
 /**
  * call_rcu - Queue an RCU callback for invocation after a grace period.
  * @head: structure to be used for queueing the RCU updates.
@@ -133,18 +172,11 @@
 				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);
-	}
+	__call_rcu(head, &rcu_ctrlblk, &__get_cpu_var(rcu_data));
 	local_irq_restore(flags);
 }
 EXPORT_SYMBOL_GPL(call_rcu);
@@ -169,20 +201,11 @@
 				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);
-	}
-
+	__call_rcu(head, &rcu_bh_ctrlblk, &__get_cpu_var(rcu_bh_data));
 	local_irq_restore(flags);
 }
 EXPORT_SYMBOL_GPL(call_rcu_bh);
@@ -211,12 +234,6 @@
 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();
 }
 
 /*
@@ -276,14 +293,8 @@
  */
 static void rcu_start_batch(struct rcu_ctrlblk *rcp)
 {
-	if (rcp->next_pending &&
+	if (rcp->cur != rcp->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++;
 
 		/*
@@ -364,13 +375,15 @@
  * 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)
+				struct rcu_head **tail, long batch)
 {
-	local_irq_disable();
-	*this_rdp->nxttail = list;
-	if (list)
-		this_rdp->nxttail = tail;
-	local_irq_enable();
+	if (list) {
+		local_irq_disable();
+		this_rdp->batch = batch;
+		*this_rdp->nxttail[2] = list;
+		this_rdp->nxttail[2] = tail;
+		local_irq_enable();
+	}
 }
 
 static void __rcu_offline_cpu(struct rcu_data *this_rdp,
@@ -384,9 +397,9 @@
 	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);
+	/* spin_lock implies smp_mb() */
+	rcu_move_batch(this_rdp, rdp->donelist, rdp->donetail, rcp->cur + 1);
+	rcu_move_batch(this_rdp, rdp->nxtlist, rdp->nxttail[2], rcp->cur + 1);
 
 	local_irq_disable();
 	this_rdp->qlen += rdp->qlen;
@@ -420,37 +433,45 @@
 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) {
+	if (rdp->nxtlist) {
 		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
+		 * move the other grace-period-completed entries to
+		 * [rdp->nxtlist, *rdp->nxttail[0]) temporarily
 		 */
+		if (!rcu_batch_before(rcp->completed, rdp->batch))
+			rdp->nxttail[0] = rdp->nxttail[1] = rdp->nxttail[2];
+		else if (!rcu_batch_before(rcp->completed, rdp->batch - 1))
+			rdp->nxttail[0] = rdp->nxttail[1];
 
-		/* determine batch number */
-		rdp->batch = rcp->cur + 1;
-		/* see the comment and corresponding wmb() in
-		 * the rcu_start_batch()
+		/*
+		 * the grace period for entries in
+		 * [rdp->nxtlist, *rdp->nxttail[0]) has completed and
+		 * move these entries to donelist
 		 */
-		smp_rmb();
+		if (rdp->nxttail[0] != &rdp->nxtlist) {
+			*rdp->donetail = rdp->nxtlist;
+			rdp->donetail = rdp->nxttail[0];
+			rdp->nxtlist = *rdp->nxttail[0];
+			*rdp->donetail = NULL;
 
-		if (!rcp->next_pending) {
+			if (rdp->nxttail[1] == rdp->nxttail[0])
+				rdp->nxttail[1] = &rdp->nxtlist;
+			if (rdp->nxttail[2] == rdp->nxttail[0])
+				rdp->nxttail[2] = &rdp->nxtlist;
+			rdp->nxttail[0] = &rdp->nxtlist;
+		}
+
+		local_irq_enable();
+
+		if (rcu_batch_after(rdp->batch, rcp->pending)) {
 			/* and start it/schedule start if it's a new batch */
 			spin_lock(&rcp->lock);
-			rcp->next_pending = 1;
-			rcu_start_batch(rcp);
+			if (rcu_batch_after(rdp->batch, rcp->pending)) {
+				rcp->pending = rdp->batch;
+				rcu_start_batch(rcp);
+			}
 			spin_unlock(&rcp->lock);
 		}
 	}
@@ -468,15 +489,26 @@
 
 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;
+	if (rdp->nxtlist) {
+		/*
+		 * This cpu has pending rcu entries and the grace period
+		 * for them has completed.
+		 */
+		if (!rcu_batch_before(rcp->completed, rdp->batch))
+			return 1;
+		if (!rcu_batch_before(rcp->completed, rdp->batch - 1) &&
+				rdp->nxttail[0] != rdp->nxttail[1])
+			return 1;
+		if (rdp->nxttail[0] != &rdp->nxtlist)
+			return 1;
 
-	/* This cpu has no pending entries, but there are new entries */
-	if (!rdp->curlist && rdp->nxtlist)
-		return 1;
+		/*
+		 * This cpu has pending rcu entries and the new batch
+		 * for then hasn't been started nor scheduled start
+		 */
+		if (rcu_batch_after(rdp->batch, rcp->pending))
+			return 1;
+	}
 
 	/* This cpu has finished callbacks to invoke */
 	if (rdp->donelist)
@@ -512,7 +544,7 @@
 	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));
+	return !!rdp->nxtlist || !!rdp_bh->nxtlist || rcu_pending(cpu);
 }
 
 void rcu_check_callbacks(int cpu, int user)
@@ -559,8 +591,7 @@
 						struct rcu_data *rdp)
 {
 	memset(rdp, 0, sizeof(*rdp));
-	rdp->curtail = &rdp->curlist;
-	rdp->nxttail = &rdp->nxtlist;
+	rdp->nxttail[0] = rdp->nxttail[1] = rdp->nxttail[2] = &rdp->nxtlist;
 	rdp->donetail = &rdp->donelist;
 	rdp->quiescbatch = rcp->completed;
 	rdp->qs_pending = 0;