Merge davem@nuts.davemloft.net:/disk1/BK/net-2.6

into kernel.bkbits.net:/home/davem/net-2.6

10 files changed, 207 insertions, 138 deletions

diff --git a/kernel/Makefile b/kernel/Makefile
index 2b7b352b1a38..47f98594e9e5 100644
--- a/kernel/Makefile
+++ b/kernel/Makefile
@@ -33,23 +33,7 @@ ifneq ($(CONFIG_IA64),y)
 CFLAGS_sched.o := $(PROFILING) -fno-omit-frame-pointer
 endif
 
-# configs.o uses generated files - dependecies must be listed explicitly
-$(obj)/configs.o: $(obj)/ikconfig.h
-
-ifdef CONFIG_IKCONFIG_PROC
 $(obj)/configs.o: $(obj)/config_data.h
-endif
-
-# ikconfig.h contains all the selected config entries - generated
-# from top-level Makefile and .config. Info from ikconfig.h can
-# be extracted from the kernel binary.
-
-quiet_cmd_ikconfig = IKCFG   $@
-      cmd_ikconfig = $(CONFIG_SHELL) $< .config $(srctree)/Makefile > $@
-
-targets += ikconfig.h
-$(obj)/ikconfig.h: scripts/mkconfigs .config $(srctree)/Makefile FORCE
-	$(call if_changed,ikconfig)
 
 # config_data.h contains the same information as ikconfig.h but gzipped.
 # Info from config_data can be extracted from /proc/config*
@@ -58,7 +42,7 @@ $(obj)/config_data.gz: .config FORCE
 	$(call if_changed,gzip)
 
 quiet_cmd_ikconfiggz = IKCFG   $@
-      cmd_ikconfiggz = cat $< | scripts/bin2c kernel_config_data > $@
+      cmd_ikconfiggz = (echo "const char kernel_config_data[] = MAGIC_START"; cat $< | scripts/bin2c; echo "MAGIC_END;") > $@
 targets += config_data.h
 $(obj)/config_data.h: $(obj)/config_data.gz FORCE
 	$(call if_changed,ikconfiggz)
diff --git a/kernel/auditsc.c b/kernel/auditsc.c
index 342b57141fd9..e688c73f6a9e 100644
--- a/kernel/auditsc.c
+++ b/kernel/auditsc.c
@@ -177,9 +177,10 @@ static inline int audit_add_rule(struct audit_entry *entry,
 	return 0;
 }
 
-static void audit_free_rule(void *arg)
+static void audit_free_rule(struct rcu_head *head)
 {
-	kfree(arg);
+	struct audit_entry *e = container_of(head, struct audit_entry, rcu);
+	kfree(e);
 }
 
 /* Note that audit_add_rule and audit_del_rule are called via
@@ -195,7 +196,7 @@ static inline int audit_del_rule(struct audit_rule *rule,
 	list_for_each_entry(e, list, list) {
 		if (!audit_compare_rule(rule, &e->rule)) {
 			list_del_rcu(&e->list);
-			call_rcu(&e->rcu, audit_free_rule, e);
+			call_rcu(&e->rcu, audit_free_rule);
 			return 0;
 		}
 	}
diff --git a/kernel/configs.c b/kernel/configs.c
index 326ab7b214f6..d18a944ad249 100644
--- a/kernel/configs.c
+++ b/kernel/configs.c
@@ -34,13 +34,26 @@
 /**************************************************/
 /* the actual current config file                 */
 
-/* This one is for extraction from the kernel binary file image. */
-#include "ikconfig.h"
+/*
+ * Define kernel_config_data and kernel_config_data_size, which contains the
+ * wrapped and compressed configuration file.  The file is first compressed
+ * with gzip and then bounded by two eight byte magic numbers to allow
+ * extraction from a binary kernel image:
+ *
+ *   IKCFG_ST
+ *   <image>
+ *   IKCFG_ED
+ */
+#define MAGIC_START	"IKCFG_ST"
+#define MAGIC_END	"IKCFG_ED"
+#include "config_data.h"
 
-#ifdef CONFIG_IKCONFIG_PROC
 
-/* This is the data that can be read from /proc/config.gz. */
-#include "config_data.h"
+#define MAGIC_SIZE (sizeof(MAGIC_START) - 1)
+#define kernel_config_data_size \
+	(sizeof(kernel_config_data) - 1 - MAGIC_SIZE * 2)
+
+#ifdef CONFIG_IKCONFIG_PROC
 
 /**************************************************/
 /* globals and useful constants                   */
@@ -58,7 +71,7 @@ ikconfig_read_current(struct file *file, char __user *buf,
 		return 0;
 
 	count = min(len, (size_t)(kernel_config_data_size - pos));
-	if(copy_to_user(buf, kernel_config_data + pos, count))
+	if (copy_to_user(buf, kernel_config_data + MAGIC_SIZE + pos, count))
 		return -EFAULT;
 
 	*offset += count;
diff --git a/kernel/cpu.c b/kernel/cpu.c
index 72b984c67eb3..083521327e64 100644
--- a/kernel/cpu.c
+++ b/kernel/cpu.c
@@ -20,14 +20,6 @@
 DECLARE_MUTEX(cpucontrol);
 
 static struct notifier_block *cpu_chain;
-/*
- * Represents all cpu's present in the system
- * In systems capable of hotplug, this map could dynamically grow
- * as new cpu's are detected in the system via any platform specific
- * method, such as ACPI for e.g.
- */
-cpumask_t	cpu_present_map;
-EXPORT_SYMBOL(cpu_present_map);
 
 /* Need to know about CPUs going up/down? */
 int register_cpu_notifier(struct notifier_block *nb)
diff --git a/kernel/kmod.c b/kernel/kmod.c
index ea62192b7597..579269c38a3b 100644
--- a/kernel/kmod.c
+++ b/kernel/kmod.c
@@ -154,7 +154,6 @@ static int ____call_usermodehelper(void *data)
 {
 	struct subprocess_info *sub_info = data;
 	int retval;
-	cpumask_t mask = CPU_MASK_ALL;
 
 	/* Unblock all signals. */
 	flush_signals(current);
@@ -165,7 +164,7 @@ static int ____call_usermodehelper(void *data)
 	spin_unlock_irq(&current->sighand->siglock);
 
 	/* We can run anywhere, unlike our parent keventd(). */
-	set_cpus_allowed(current, mask);
+	set_cpus_allowed(current, CPU_MASK_ALL);
 
 	retval = -EPERM;
 	if (current->fs->root)
diff --git a/kernel/kthread.c b/kernel/kthread.c
index da0ec5b25cdf..5689ebb1a250 100644
--- a/kernel/kthread.c
+++ b/kernel/kthread.c
@@ -65,7 +65,6 @@ static int kthread(void *_create)
 	void *data;
 	sigset_t blocked;
 	int ret = -EINTR;
-	cpumask_t mask = CPU_MASK_ALL;
 
 	kthread_exit_files();
 
@@ -79,7 +78,7 @@ static int kthread(void *_create)
 	flush_signals(current);
 
 	/* By default we can run anywhere, unlike keventd. */
-	set_cpus_allowed(current, mask);
+	set_cpus_allowed(current, CPU_MASK_ALL);
 
 	/* OK, tell user we're spawned, wait for stop or wakeup */
 	__set_current_state(TASK_INTERRUPTIBLE);
diff --git a/kernel/power/process.c b/kernel/power/process.c
index 347435415eaf..0d1f63ec7287 100644
--- a/kernel/power/process.c
+++ b/kernel/power/process.c
@@ -109,7 +109,6 @@ void thaw_processes(void)
 			wake_up_process(p);
 		} else
 			printk(KERN_INFO " Strange, %s not stopped\n", p->comm );
-		wake_up_process(p);
 	} while_each_thread(g, p);
 
 	read_unlock(&tasklist_lock);
diff --git a/kernel/rcupdate.c b/kernel/rcupdate.c
index 13a1b5a5825f..f35f944abe3d 100644
--- a/kernel/rcupdate.c
+++ b/kernel/rcupdate.c
@@ -47,8 +47,17 @@
 
 /* Definition for rcupdate control block. */
 struct rcu_ctrlblk rcu_ctrlblk = 
-	{ .mutex = SPIN_LOCK_UNLOCKED, .curbatch = 1, 
-	  .maxbatch = 1, .rcu_cpu_mask = CPU_MASK_NONE };
+	{ .cur = -300, .completed = -300 , .lock = SEQCNT_ZERO };
+
+/* Bookkeeping of the progress of the grace period */
+struct {
+	spinlock_t	mutex; /* Guard this struct and writes to rcu_ctrlblk */
+	cpumask_t	rcu_cpu_mask; /* CPUs that need to switch in order    */
+	                              /* for current batch to proceed.        */
+} rcu_state ____cacheline_maxaligned_in_smp =
+	  {.mutex = SPIN_LOCK_UNLOCKED, .rcu_cpu_mask = CPU_MASK_NONE };
+
+
 DEFINE_PER_CPU(struct rcu_data, rcu_data) = { 0L };
 
 /* Fake initialization required by compiler */
@@ -59,23 +68,24 @@ static DEFINE_PER_CPU(struct tasklet_struct, rcu_tasklet) = {NULL};
  * call_rcu - Queue an RCU update request.
  * @head: structure to be used for queueing the RCU updates.
  * @func: actual update function to be invoked after the grace period
- * @arg: argument to be passed to the update function
  *
  * The update function will be invoked as soon as all CPUs have performed 
  * a context switch or been seen in the idle loop or in a user process. 
  * The read-side of critical section that use call_rcu() for updation must 
  * be protected by rcu_read_lock()/rcu_read_unlock().
  */
-void fastcall call_rcu(struct rcu_head *head, void (*func)(void *arg), void *arg)
+void fastcall call_rcu(struct rcu_head *head,
+				void (*func)(struct rcu_head *rcu))
 {
 	int cpu;
 	unsigned long flags;
 
 	head->func = func;
-	head->arg = arg;
+	head->next = NULL;
 	local_irq_save(flags);
 	cpu = smp_processor_id();
-	list_add_tail(&head->list, &RCU_nxtlist(cpu));
+	*RCU_nxttail(cpu) = head;
+	RCU_nxttail(cpu) = &head->next;
 	local_irq_restore(flags);
 }
 
@@ -83,39 +93,70 @@ void fastcall call_rcu(struct rcu_head *head, void (*func)(void *arg), void *arg
  * Invoke the completed RCU callbacks. They are expected to be in
  * a per-cpu list.
  */
-static void rcu_do_batch(struct list_head *list)
+static void rcu_do_batch(struct rcu_head *list)
 {
-	struct list_head *entry;
-	struct rcu_head *head;
+	struct rcu_head *next;
 
-	while (!list_empty(list)) {
-		entry = list->next;
-		list_del(entry);
-		head = list_entry(entry, struct rcu_head, list);
-		head->func(head->arg);
+	while (list) {
+		next = list->next;
+		list->func(list);
+		list = next;
 	}
 }
 
 /*
+ * 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 rcu_ctrlblk.cur with
+ *   RCU_quiescbatch(cpu). All cpus are recorded  in the
+ *   rcu_state.rcu_cpu_mask 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_state.rcu_cpu_mask. 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 the rcu_ctrlblk lock.
+ * Caller must hold rcu_state.mutex.
  */
-static void rcu_start_batch(long newbatch)
+static void rcu_start_batch(int next_pending)
 {
-	cpumask_t active;
-
-	if (rcu_batch_before(rcu_ctrlblk.maxbatch, newbatch)) {
-		rcu_ctrlblk.maxbatch = newbatch;
+	if (next_pending)
+		rcu_ctrlblk.next_pending = 1;
+
+	if (rcu_ctrlblk.next_pending &&
+			rcu_ctrlblk.completed == rcu_ctrlblk.cur) {
+		/* Can't change, since spin lock held. */
+		cpus_andnot(rcu_state.rcu_cpu_mask, cpu_online_map,
+							nohz_cpu_mask);
+		write_seqcount_begin(&rcu_ctrlblk.lock);
+		rcu_ctrlblk.next_pending = 0;
+		rcu_ctrlblk.cur++;
+		write_seqcount_end(&rcu_ctrlblk.lock);
 	}
-	if (rcu_batch_before(rcu_ctrlblk.maxbatch, rcu_ctrlblk.curbatch) ||
-	    !cpus_empty(rcu_ctrlblk.rcu_cpu_mask)) {
-		return;
+}
+
+/*
+ * 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)
+{
+	cpu_clear(cpu, rcu_state.rcu_cpu_mask);
+	if (cpus_empty(rcu_state.rcu_cpu_mask)) {
+		/* batch completed ! */
+		rcu_ctrlblk.completed = rcu_ctrlblk.cur;
+		rcu_start_batch(0);
 	}
-	/* Can't change, since spin lock held. */
-	active = nohz_cpu_mask;
-	cpus_complement(active);
-	cpus_and(rcu_ctrlblk.rcu_cpu_mask, cpu_online_map, active);
 }
 
 /*
@@ -127,7 +168,19 @@ static void rcu_check_quiescent_state(void)
 {
 	int cpu = smp_processor_id();
 
-	if (!cpu_isset(cpu, rcu_ctrlblk.rcu_cpu_mask))
+	if (RCU_quiescbatch(cpu) != rcu_ctrlblk.cur) {
+		/* new grace period: record qsctr value. */
+		RCU_qs_pending(cpu) = 1;
+		RCU_last_qsctr(cpu) = RCU_qsctr(cpu);
+		RCU_quiescbatch(cpu) = rcu_ctrlblk.cur;
+		return;
+	}
+
+	/* Grace period already completed for this cpu?
+	 * qs_pending is checked instead of the actual bitmap to avoid
+	 * cacheline trashing.
+	 */
+	if (!RCU_qs_pending(cpu))
 		return;
 
 	/* 
@@ -135,27 +188,19 @@ static void rcu_check_quiescent_state(void)
 	 * we may miss one quiescent state of that CPU. That is
 	 * tolerable. So no need to disable interrupts.
 	 */
-	if (RCU_last_qsctr(cpu) == RCU_QSCTR_INVALID) {
-		RCU_last_qsctr(cpu) = RCU_qsctr(cpu);
-		return;
-	}
 	if (RCU_qsctr(cpu) == RCU_last_qsctr(cpu))
 		return;
+	RCU_qs_pending(cpu) = 0;
 
-	spin_lock(&rcu_ctrlblk.mutex);
-	if (!cpu_isset(cpu, rcu_ctrlblk.rcu_cpu_mask))
-		goto out_unlock;
-
-	cpu_clear(cpu, rcu_ctrlblk.rcu_cpu_mask);
-	RCU_last_qsctr(cpu) = RCU_QSCTR_INVALID;
-	if (!cpus_empty(rcu_ctrlblk.rcu_cpu_mask))
-		goto out_unlock;
-
-	rcu_ctrlblk.curbatch++;
-	rcu_start_batch(rcu_ctrlblk.maxbatch);
+	spin_lock(&rcu_state.mutex);
+	/*
+	 * RCU_quiescbatch/batch.cur and the cpu bitmap can come out of sync
+	 * during cpu startup. Ignore the quiescent state.
+	 */
+	if (likely(RCU_quiescbatch(cpu) == rcu_ctrlblk.cur))
+		cpu_quiet(cpu);
 
-out_unlock:
-	spin_unlock(&rcu_ctrlblk.mutex);
+	spin_unlock(&rcu_state.mutex);
 }
 
 
@@ -185,25 +230,11 @@ static void rcu_offline_cpu(int cpu)
 	 * we can block indefinitely waiting for it, so flush
 	 * it here
 	 */
-	spin_lock_irq(&rcu_ctrlblk.mutex);
-	if (cpus_empty(rcu_ctrlblk.rcu_cpu_mask))
-		goto unlock;
-
-	cpu_clear(cpu, rcu_ctrlblk.rcu_cpu_mask);
-	if (cpus_empty(rcu_ctrlblk.rcu_cpu_mask)) {
-		rcu_ctrlblk.curbatch++;
-		/* We may avoid calling start batch if
-		 * we are starting the batch only
-		 * because of the DEAD CPU (the current
-		 * CPU will start a new batch anyway for
-		 * the callbacks we will move to current CPU).
-		 * However, we will avoid this optimisation
-		 * for now.
-		 */
-		rcu_start_batch(rcu_ctrlblk.maxbatch);
-	}
+	spin_lock_bh(&rcu_state.mutex);
+	if (rcu_ctrlblk.cur != rcu_ctrlblk.completed)
+		cpu_quiet(cpu);
 unlock:
-	spin_unlock_irq(&rcu_ctrlblk.mutex);
+	spin_unlock_bh(&rcu_state.mutex);
 
 	rcu_move_batch(&RCU_curlist(cpu));
 	rcu_move_batch(&RCU_nxtlist(cpu));
@@ -213,39 +244,59 @@ unlock:
 
 #endif
 
+void rcu_restart_cpu(int cpu)
+{
+	spin_lock_bh(&rcu_state.mutex);
+	RCU_quiescbatch(cpu) = rcu_ctrlblk.completed;
+	RCU_qs_pending(cpu) = 0;
+	spin_unlock_bh(&rcu_state.mutex);
+}
+
 /*
  * This does the RCU processing work from tasklet context. 
  */
 static void rcu_process_callbacks(unsigned long unused)
 {
 	int cpu = smp_processor_id();
-	LIST_HEAD(list);
+	struct rcu_head *rcu_list = NULL;
 
-	if (!list_empty(&RCU_curlist(cpu)) &&
-	    rcu_batch_after(rcu_ctrlblk.curbatch, RCU_batch(cpu))) {
-		__list_splice(&RCU_curlist(cpu), &list);
-		INIT_LIST_HEAD(&RCU_curlist(cpu));
+	if (RCU_curlist(cpu) &&
+	    !rcu_batch_before(rcu_ctrlblk.completed, RCU_batch(cpu))) {
+		rcu_list = RCU_curlist(cpu);
+		RCU_curlist(cpu) = NULL;
 	}
 
 	local_irq_disable();
-	if (!list_empty(&RCU_nxtlist(cpu)) && list_empty(&RCU_curlist(cpu))) {
-		__list_splice(&RCU_nxtlist(cpu), &RCU_curlist(cpu));
-		INIT_LIST_HEAD(&RCU_nxtlist(cpu));
+	if (RCU_nxtlist(cpu) && !RCU_curlist(cpu)) {
+		int next_pending, seq;
+
+		RCU_curlist(cpu) = RCU_nxtlist(cpu);
+		RCU_nxtlist(cpu) = NULL;
+		RCU_nxttail(cpu) = &RCU_nxtlist(cpu);
 		local_irq_enable();
 
 		/*
 		 * start the next batch of callbacks
 		 */
-		spin_lock(&rcu_ctrlblk.mutex);
-		RCU_batch(cpu) = rcu_ctrlblk.curbatch + 1;
-		rcu_start_batch(RCU_batch(cpu));
-		spin_unlock(&rcu_ctrlblk.mutex);
+		do {
+			seq = read_seqcount_begin(&rcu_ctrlblk.lock);
+			/* determine batch number */
+			RCU_batch(cpu) = rcu_ctrlblk.cur + 1;
+			next_pending = rcu_ctrlblk.next_pending;
+		} while (read_seqcount_retry(&rcu_ctrlblk.lock, seq));
+
+		if (!next_pending) {
+			/* and start it/schedule start if it's a new batch */
+			spin_lock(&rcu_state.mutex);
+			rcu_start_batch(1);
+			spin_unlock(&rcu_state.mutex);
+		}
 	} else {
 		local_irq_enable();
 	}
 	rcu_check_quiescent_state();
-	if (!list_empty(&list))
-		rcu_do_batch(&list);
+	if (rcu_list)
+		rcu_do_batch(rcu_list);
 }
 
 void rcu_check_callbacks(int cpu, int user)
@@ -261,8 +312,9 @@ static void __devinit rcu_online_cpu(int cpu)
 {
 	memset(&per_cpu(rcu_data, cpu), 0, sizeof(struct rcu_data));
 	tasklet_init(&RCU_tasklet(cpu), rcu_process_callbacks, 0UL);
-	INIT_LIST_HEAD(&RCU_nxtlist(cpu));
-	INIT_LIST_HEAD(&RCU_curlist(cpu));
+	RCU_nxttail(cpu) = &RCU_nxtlist(cpu);
+	RCU_quiescbatch(cpu) = rcu_ctrlblk.completed;
+	RCU_qs_pending(cpu) = 0;
 }
 
 static int __devinit rcu_cpu_notify(struct notifier_block *self, 
@@ -302,11 +354,18 @@ void __init rcu_init(void)
 	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(void *completion)
+static void wakeme_after_rcu(struct rcu_head  *head)
 {
-	complete(completion);
+	struct rcu_synchronize *rcu;
+
+	rcu = container_of(head, struct rcu_synchronize, head);
+	complete(&rcu->completion);
 }
 
 /**
@@ -315,14 +374,14 @@ static void wakeme_after_rcu(void *completion)
  */
 void synchronize_kernel(void)
 {
-	struct rcu_head rcu;
-	DECLARE_COMPLETION(completion);
+	struct rcu_synchronize rcu;
 
+	init_completion(&rcu.completion);
 	/* Will wake me after RCU finished */
-	call_rcu(&rcu, wakeme_after_rcu, &completion);
+	call_rcu(&rcu.head, wakeme_after_rcu);
 
 	/* Wait for it */
-	wait_for_completion(&completion);
+	wait_for_completion(&rcu.completion);
 }
 
 
diff --git a/kernel/sched.c b/kernel/sched.c
index 8f49ba1202c3..9e676db0267b 100644
--- a/kernel/sched.c
+++ b/kernel/sched.c
@@ -696,10 +696,9 @@ static int wake_idle(int cpu, task_t *p)
 		return cpu;
 
 	cpus_and(tmp, sd->span, cpu_online_map);
-	for_each_cpu_mask(i, tmp) {
-		if (!cpu_isset(i, p->cpus_allowed))
-			continue;
+	cpus_and(tmp, tmp, p->cpus_allowed);
 
+	for_each_cpu_mask(i, tmp) {
 		if (idle_cpu(i))
 			return i;
 	}
@@ -2941,6 +2940,21 @@ out_unlock:
 	return retval;
 }
 
+/*
+ * Represents all cpu's present in the system
+ * In systems capable of hotplug, this map could dynamically grow
+ * as new cpu's are detected in the system via any platform specific
+ * method, such as ACPI for e.g.
+ */
+
+cpumask_t cpu_present_map;
+EXPORT_SYMBOL(cpu_present_map);
+
+#ifndef CONFIG_SMP
+cpumask_t cpu_online_map = CPU_MASK_ALL;
+cpumask_t cpu_possible_map = CPU_MASK_ALL;
+#endif
+
 /**
  * sys_sched_getaffinity - get the cpu affinity of a process
  * @pid: pid of the process
@@ -3320,7 +3334,7 @@ int set_cpus_allowed(task_t *p, cpumask_t new_mask)
 	runqueue_t *rq;
 
 	rq = task_rq_lock(p, &flags);
-	if (any_online_cpu(new_mask) == NR_CPUS) {
+	if (!cpus_intersects(new_mask, cpu_online_map)) {
 		ret = -EINVAL;
 		goto out;
 	}
@@ -3495,8 +3509,7 @@ static void migrate_all_tasks(int src_cpu)
 		if (dest_cpu == NR_CPUS)
 			dest_cpu = any_online_cpu(tsk->cpus_allowed);
 		if (dest_cpu == NR_CPUS) {
-			cpus_clear(tsk->cpus_allowed);
-			cpus_complement(tsk->cpus_allowed);
+			cpus_setall(tsk->cpus_allowed);
 			dest_cpu = any_online_cpu(tsk->cpus_allowed);
 
 			/* Don't tell them about moving exiting tasks
@@ -3558,6 +3571,7 @@ static int migration_call(struct notifier_block *nfb, unsigned long action,
 		p = kthread_create(migration_thread, hcpu, "migration/%d",cpu);
 		if (IS_ERR(p))
 			return NOTIFY_BAD;
+		p->flags |= PF_NOFREEZE;
 		kthread_bind(p, cpu);
 		/* Must be high prio: stop_machine expects to yield to it. */
 		rq = task_rq_lock(p, &flags);
@@ -3812,7 +3826,7 @@ void sched_domain_debug(void)
 			int j;
 			char str[NR_CPUS];
 			struct sched_group *group = sd->groups;
-			cpumask_t groupmask, tmp;
+			cpumask_t groupmask;
 
 			cpumask_scnprintf(str, NR_CPUS, sd->span);
 			cpus_clear(groupmask);
@@ -3842,8 +3856,7 @@ void sched_domain_debug(void)
 				if (!cpus_weight(group->cpumask))
 					printk(" ERROR empty group:");
 
-				cpus_and(tmp, groupmask, group->cpumask);
-				if (cpus_weight(tmp) > 0)
+				if (cpus_intersects(groupmask, group->cpumask))
 					printk(" ERROR repeated CPUs:");
 
 				cpus_or(groupmask, groupmask, group->cpumask);
@@ -3862,8 +3875,7 @@ void sched_domain_debug(void)
 			sd = sd->parent;
 
 			if (sd) {
-				cpus_and(tmp, groupmask, sd->span);
-				if (!cpus_equal(tmp, groupmask))
+				if (!cpus_subset(groupmask, sd->span))
 					printk(KERN_DEBUG "ERROR parent span is not a superset of domain->span\n");
 			}
 
@@ -3902,16 +3914,15 @@ void __init sched_init(void)
 	/* Set up an initial dummy domain for early boot */
 	static struct sched_domain sched_domain_init;
 	static struct sched_group sched_group_init;
-	cpumask_t cpu_mask_all = CPU_MASK_ALL;
 
 	memset(&sched_domain_init, 0, sizeof(struct sched_domain));
-	sched_domain_init.span = cpu_mask_all;
+	sched_domain_init.span = CPU_MASK_ALL;
 	sched_domain_init.groups = &sched_group_init;
 	sched_domain_init.last_balance = jiffies;
 	sched_domain_init.balance_interval = INT_MAX; /* Don't balance */
 
 	memset(&sched_group_init, 0, sizeof(struct sched_group));
-	sched_group_init.cpumask = cpu_mask_all;
+	sched_group_init.cpumask = CPU_MASK_ALL;
 	sched_group_init.next = &sched_group_init;
 	sched_group_init.cpu_power = SCHED_LOAD_SCALE;
 #endif
diff --git a/kernel/sysctl.c b/kernel/sysctl.c
index 7dca63a88ea2..641727bab22f 100644
--- a/kernel/sysctl.c
+++ b/kernel/sysctl.c
@@ -39,6 +39,8 @@
 #include <linux/initrd.h>
 #include <linux/times.h>
 #include <linux/limits.h>
+#include <linux/dcache.h>
+
 #include <asm/uaccess.h>
 
 #ifdef CONFIG_ROOT_NFS
@@ -777,6 +779,16 @@ static ctl_table vm_table[] = {
 		.strategy	= &sysctl_intvec,
 		.extra1		= &zero,
 	},
+	{
+		.ctl_name	= VM_VFS_CACHE_PRESSURE,
+		.procname	= "vfs_cache_pressure",
+		.data		= &sysctl_vfs_cache_pressure,
+		.maxlen		= sizeof(sysctl_vfs_cache_pressure),
+		.mode		= 0644,
+		.proc_handler	= &proc_dointvec,
+		.strategy	= &sysctl_intvec,
+		.extra1		= &zero,
+	},
 	{ .ctl_name = 0 }
 };