[linux.git] / kernel / cpu.c

/* CPU control.
 * (C) 2001, 2002, 2003, 2004 Rusty Russell
 *
 * This code is licenced under the GPL.
 */
#include <linux/proc_fs.h>
#include <linux/smp.h>
#include <linux/init.h>
#include <linux/notifier.h>
#include <linux/sched.h>
#include <linux/unistd.h>
#include <linux/cpu.h>
#include <linux/module.h>
#include <linux/kthread.h>
#include <linux/stop_machine.h>
#include <linux/mutex.h>

/*
 * 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 __read_mostly;
EXPORT_SYMBOL(cpu_present_map);

#ifndef CONFIG_SMP

/*
 * Represents all cpu's that are currently online.
 */
cpumask_t cpu_online_map __read_mostly = CPU_MASK_ALL;
EXPORT_SYMBOL(cpu_online_map);

cpumask_t cpu_possible_map __read_mostly = CPU_MASK_ALL;
EXPORT_SYMBOL(cpu_possible_map);

#else /* CONFIG_SMP */

/* Serializes the updates to cpu_online_map, cpu_present_map */
static DEFINE_MUTEX(cpu_add_remove_lock);

static __cpuinitdata RAW_NOTIFIER_HEAD(cpu_chain);

/* If set, cpu_up and cpu_down will return -EBUSY and do nothing.
 * Should always be manipulated under cpu_add_remove_lock
 */
static int cpu_hotplug_disabled;

static struct {
	struct task_struct *active_writer;
	struct mutex lock; /* Synchronizes accesses to refcount, */
	/*
	 * Also blocks the new readers during
	 * an ongoing cpu hotplug operation.
	 */
	int refcount;
} cpu_hotplug;

void __init cpu_hotplug_init(void)
{
	cpu_hotplug.active_writer = NULL;
	mutex_init(&cpu_hotplug.lock);
	cpu_hotplug.refcount = 0;
}

cpumask_t cpu_active_map;

#ifdef CONFIG_HOTPLUG_CPU

void get_online_cpus(void)
{
	might_sleep();
	if (cpu_hotplug.active_writer == current)
		return;
	mutex_lock(&cpu_hotplug.lock);
	cpu_hotplug.refcount++;
	mutex_unlock(&cpu_hotplug.lock);

}
EXPORT_SYMBOL_GPL(get_online_cpus);

void put_online_cpus(void)
{
	if (cpu_hotplug.active_writer == current)
		return;
	mutex_lock(&cpu_hotplug.lock);
	if (!--cpu_hotplug.refcount && unlikely(cpu_hotplug.active_writer))
		wake_up_process(cpu_hotplug.active_writer);
	mutex_unlock(&cpu_hotplug.lock);

}
EXPORT_SYMBOL_GPL(put_online_cpus);

#endif	/* CONFIG_HOTPLUG_CPU */

/*
 * The following two API's must be used when attempting
 * to serialize the updates to cpu_online_map, cpu_present_map.
 */
void cpu_maps_update_begin(void)
{
	mutex_lock(&cpu_add_remove_lock);
}

void cpu_maps_update_done(void)
{
	mutex_unlock(&cpu_add_remove_lock);
}

/*
 * This ensures that the hotplug operation can begin only when the
 * refcount goes to zero.
 *
 * Note that during a cpu-hotplug operation, the new readers, if any,
 * will be blocked by the cpu_hotplug.lock
 *
 * Since cpu_hotplug_begin() is always called after invoking
 * cpu_maps_update_begin(), we can be sure that only one writer is active.
 *
 * Note that theoretically, there is a possibility of a livelock:
 * - Refcount goes to zero, last reader wakes up the sleeping
 *   writer.
 * - Last reader unlocks the cpu_hotplug.lock.
 * - A new reader arrives at this moment, bumps up the refcount.
 * - The writer acquires the cpu_hotplug.lock finds the refcount
 *   non zero and goes to sleep again.
 *
 * However, this is very difficult to achieve in practice since
 * get_online_cpus() not an api which is called all that often.
 *
 */
static void cpu_hotplug_begin(void)
{
	cpu_hotplug.active_writer = current;

	for (;;) {
		mutex_lock(&cpu_hotplug.lock);
		if (likely(!cpu_hotplug.refcount))
			break;
		__set_current_state(TASK_UNINTERRUPTIBLE);
		mutex_unlock(&cpu_hotplug.lock);
		schedule();
	}
}

static void cpu_hotplug_done(void)
{
	cpu_hotplug.active_writer = NULL;
	mutex_unlock(&cpu_hotplug.lock);
}
/* Need to know about CPUs going up/down? */
int __ref register_cpu_notifier(struct notifier_block *nb)
{
	int ret;
	cpu_maps_update_begin();
	ret = raw_notifier_chain_register(&cpu_chain, nb);
	cpu_maps_update_done();
	return ret;
}

#ifdef CONFIG_HOTPLUG_CPU

EXPORT_SYMBOL(register_cpu_notifier);

void __ref unregister_cpu_notifier(struct notifier_block *nb)
{
	cpu_maps_update_begin();
	raw_notifier_chain_unregister(&cpu_chain, nb);
	cpu_maps_update_done();
}
EXPORT_SYMBOL(unregister_cpu_notifier);

static inline void check_for_tasks(int cpu)
{
	struct task_struct *p;

	write_lock_irq(&tasklist_lock);
	for_each_process(p) {
		if (task_cpu(p) == cpu &&
		    (!cputime_eq(p->utime, cputime_zero) ||
		     !cputime_eq(p->stime, cputime_zero)))
			printk(KERN_WARNING "Task %s (pid = %d) is on cpu %d\
				(state = %ld, flags = %x) \n",
				 p->comm, task_pid_nr(p), cpu,
				 p->state, p->flags);
	}
	write_unlock_irq(&tasklist_lock);
}

struct take_cpu_down_param {
	unsigned long mod;
	void *hcpu;
};

/* Take this CPU down. */
static int __ref take_cpu_down(void *_param)
{
	struct take_cpu_down_param *param = _param;
	int err;

	raw_notifier_call_chain(&cpu_chain, CPU_DYING | param->mod,
				param->hcpu);
	/* Ensure this CPU doesn't handle any more interrupts. */
	err = __cpu_disable();
	if (err < 0)
		return err;

	/* Force idle task to run as soon as we yield: it should
	   immediately notice cpu is offline and die quickly. */
	sched_idle_next();
	return 0;
}

/* Requires cpu_add_remove_lock to be held */
static int __ref _cpu_down(unsigned int cpu, int tasks_frozen)
{
	int err, nr_calls = 0;
	cpumask_t old_allowed, tmp;
	void *hcpu = (void *)(long)cpu;
	unsigned long mod = tasks_frozen ? CPU_TASKS_FROZEN : 0;
	struct take_cpu_down_param tcd_param = {
		.mod = mod,
		.hcpu = hcpu,
	};

	if (num_online_cpus() == 1)
		return -EBUSY;

	if (!cpu_online(cpu))
		return -EINVAL;

	cpu_hotplug_begin();
	err = __raw_notifier_call_chain(&cpu_chain, CPU_DOWN_PREPARE | mod,
					hcpu, -1, &nr_calls);
	if (err == NOTIFY_BAD) {
		nr_calls--;
		__raw_notifier_call_chain(&cpu_chain, CPU_DOWN_FAILED | mod,
					  hcpu, nr_calls, NULL);
		printk("%s: attempt to take down CPU %u failed\n",
				__func__, cpu);
		err = -EINVAL;
		goto out_release;
	}

	/* Ensure that we are not runnable on dying cpu */
	old_allowed = current->cpus_allowed;
	cpus_setall(tmp);
	cpu_clear(cpu, tmp);
	set_cpus_allowed_ptr(current, &tmp);
	tmp = cpumask_of_cpu(cpu);

	err = __stop_machine(take_cpu_down, &tcd_param, &tmp);
	if (err) {
		/* CPU didn't die: tell everyone.  Can't complain. */
		if (raw_notifier_call_chain(&cpu_chain, CPU_DOWN_FAILED | mod,
					    hcpu) == NOTIFY_BAD)
			BUG();

		goto out_allowed;
	}
	BUG_ON(cpu_online(cpu));

	/* Wait for it to sleep (leaving idle task). */
	while (!idle_cpu(cpu))
		yield();

	/* This actually kills the CPU. */
	__cpu_die(cpu);

	/* CPU is completely dead: tell everyone.  Too late to complain. */
	if (raw_notifier_call_chain(&cpu_chain, CPU_DEAD | mod,
				    hcpu) == NOTIFY_BAD)
		BUG();

	check_for_tasks(cpu);

out_allowed:
	set_cpus_allowed_ptr(current, &old_allowed);
out_release:
	cpu_hotplug_done();
	if (!err) {
		if (raw_notifier_call_chain(&cpu_chain, CPU_POST_DEAD | mod,
					    hcpu) == NOTIFY_BAD)
			BUG();
	}
	return err;
}

int __ref cpu_down(unsigned int cpu)
{
	int err = 0;

	cpu_maps_update_begin();

	if (cpu_hotplug_disabled) {
		err = -EBUSY;
		goto out;
	}

	cpu_clear(cpu, cpu_active_map);

	/*
	 * Make sure the all cpus did the reschedule and are not
	 * using stale version of the cpu_active_map.
	 * This is not strictly necessary becuase stop_machine()
	 * that we run down the line already provides the required
	 * synchronization. But it's really a side effect and we do not
	 * want to depend on the innards of the stop_machine here.
	 */
	synchronize_sched();

	err = _cpu_down(cpu, 0);

	if (cpu_online(cpu))
		cpu_set(cpu, cpu_active_map);

out:
	cpu_maps_update_done();
	return err;
}
EXPORT_SYMBOL(cpu_down);
#endif /*CONFIG_HOTPLUG_CPU*/

/* Requires cpu_add_remove_lock to be held */
static int __cpuinit _cpu_up(unsigned int cpu, int tasks_frozen)
{
	int ret, nr_calls = 0;
	void *hcpu = (void *)(long)cpu;
	unsigned long mod = tasks_frozen ? CPU_TASKS_FROZEN : 0;

	if (cpu_online(cpu) || !cpu_present(cpu))
		return -EINVAL;

	cpu_hotplug_begin();
	ret = __raw_notifier_call_chain(&cpu_chain, CPU_UP_PREPARE | mod, hcpu,
							-1, &nr_calls);
	if (ret == NOTIFY_BAD) {
		nr_calls--;
		printk("%s: attempt to bring up CPU %u failed\n",
				__func__, cpu);
		ret = -EINVAL;
		goto out_notify;
	}

	/* Arch-specific enabling code. */
	ret = __cpu_up(cpu);
	if (ret != 0)
		goto out_notify;
	BUG_ON(!cpu_online(cpu));

	cpu_set(cpu, cpu_active_map);

	/* Now call notifier in preparation. */
	raw_notifier_call_chain(&cpu_chain, CPU_ONLINE | mod, hcpu);

out_notify:
	if (ret != 0)
		__raw_notifier_call_chain(&cpu_chain,
				CPU_UP_CANCELED | mod, hcpu, nr_calls, NULL);
	cpu_hotplug_done();

	return ret;
}

int __cpuinit cpu_up(unsigned int cpu)
{
	int err = 0;
	if (!cpu_isset(cpu, cpu_possible_map)) {
		printk(KERN_ERR "can't online cpu %d because it is not "
			"configured as may-hotadd at boot time\n", cpu);
#if defined(CONFIG_IA64) || defined(CONFIG_X86_64) || defined(CONFIG_S390)
		printk(KERN_ERR "please check additional_cpus= boot "
				"parameter\n");
#endif
		return -EINVAL;
	}

	cpu_maps_update_begin();

	if (cpu_hotplug_disabled) {
		err = -EBUSY;
		goto out;
	}

	err = _cpu_up(cpu, 0);

out:
	cpu_maps_update_done();
	return err;
}

#ifdef CONFIG_PM_SLEEP_SMP
static cpumask_t frozen_cpus;

int disable_nonboot_cpus(void)
{
	int cpu, first_cpu, error = 0;

	cpu_maps_update_begin();
	first_cpu = first_cpu(cpu_online_map);
	/* We take down all of the non-boot CPUs in one shot to avoid races
	 * with the userspace trying to use the CPU hotplug at the same time
	 */
	cpus_clear(frozen_cpus);
	printk("Disabling non-boot CPUs ...\n");
	for_each_online_cpu(cpu) {
		if (cpu == first_cpu)
			continue;
		error = _cpu_down(cpu, 1);
		if (!error) {
			cpu_set(cpu, frozen_cpus);
			printk("CPU%d is down\n", cpu);
		} else {
			printk(KERN_ERR "Error taking CPU%d down: %d\n",
				cpu, error);
			break;
		}
	}
	if (!error) {
		BUG_ON(num_online_cpus() > 1);
		/* Make sure the CPUs won't be enabled by someone else */
		cpu_hotplug_disabled = 1;
	} else {
		printk(KERN_ERR "Non-boot CPUs are not disabled\n");
	}
	cpu_maps_update_done();
	return error;
}

void __ref enable_nonboot_cpus(void)
{
	int cpu, error;

	/* Allow everyone to use the CPU hotplug again */
	cpu_maps_update_begin();
	cpu_hotplug_disabled = 0;
	if (cpus_empty(frozen_cpus))
		goto out;

	printk("Enabling non-boot CPUs ...\n");
	for_each_cpu_mask_nr(cpu, frozen_cpus) {
		error = _cpu_up(cpu, 1);
		if (!error) {
			printk("CPU%d is up\n", cpu);
			continue;
		}
		printk(KERN_WARNING "Error taking CPU%d up: %d\n", cpu, error);
	}
	cpus_clear(frozen_cpus);
out:
	cpu_maps_update_done();
}
#endif /* CONFIG_PM_SLEEP_SMP */

#endif /* CONFIG_SMP */

/*
 * cpu_bit_bitmap[] is a special, "compressed" data structure that
 * represents all NR_CPUS bits binary values of 1<<nr.
 *
 * It is used by cpumask_of_cpu() to get a constant address to a CPU
 * mask value that has a single bit set only.
 */

/* cpu_bit_bitmap[0] is empty - so we can back into it */
#define MASK_DECLARE_1(x)	[x+1][0] = 1UL << (x)
#define MASK_DECLARE_2(x)	MASK_DECLARE_1(x), MASK_DECLARE_1(x+1)
#define MASK_DECLARE_4(x)	MASK_DECLARE_2(x), MASK_DECLARE_2(x+2)
#define MASK_DECLARE_8(x)	MASK_DECLARE_4(x), MASK_DECLARE_4(x+4)

const unsigned long cpu_bit_bitmap[BITS_PER_LONG+1][BITS_TO_LONGS(NR_CPUS)] = {

	MASK_DECLARE_8(0),	MASK_DECLARE_8(8),
	MASK_DECLARE_8(16),	MASK_DECLARE_8(24),
#if BITS_PER_LONG > 32
	MASK_DECLARE_8(32),	MASK_DECLARE_8(40),
	MASK_DECLARE_8(48),	MASK_DECLARE_8(56),
#endif
};
EXPORT_SYMBOL_GPL(cpu_bit_bitmap);
Commit	Line	Data
1da177e4 LT	1	/* CPU control.
	2	* (C) 2001, 2002, 2003, 2004 Rusty Russell
	3	*
	4	* This code is licenced under the GPL.
	5	*/
	6	#include <linux/proc_fs.h>
	7	#include <linux/smp.h>
	8	#include <linux/init.h>
	9	#include <linux/notifier.h>
	10	#include <linux/sched.h>
	11	#include <linux/unistd.h>
	12	#include <linux/cpu.h>
	13	#include <linux/module.h>
	14	#include <linux/kthread.h>
	15	#include <linux/stop_machine.h>
81615b62	16	#include <linux/mutex.h>
1da177e4	17
68f4f1ec MK	18	/*
	19	* Represents all cpu's present in the system
	20	* In systems capable of hotplug, this map could dynamically grow
	21	* as new cpu's are detected in the system via any platform specific
	22	* method, such as ACPI for e.g.
	23	*/
	24	cpumask_t cpu_present_map __read_mostly;
	25	EXPORT_SYMBOL(cpu_present_map);
	26
	27	#ifndef CONFIG_SMP
	28
	29	/*
	30	* Represents all cpu's that are currently online.
	31	*/
	32	cpumask_t cpu_online_map __read_mostly = CPU_MASK_ALL;
	33	EXPORT_SYMBOL(cpu_online_map);
	34
	35	cpumask_t cpu_possible_map __read_mostly = CPU_MASK_ALL;
	36	EXPORT_SYMBOL(cpu_possible_map);
	37
	38	#else /* CONFIG_SMP */
	39
d221938c	40	/* Serializes the updates to cpu_online_map, cpu_present_map */
aa953877	41	static DEFINE_MUTEX(cpu_add_remove_lock);
1da177e4	42
bd5349cf	43	static __cpuinitdata RAW_NOTIFIER_HEAD(cpu_chain);
1da177e4	44
e3920fb4 RW	45	/* If set, cpu_up and cpu_down will return -EBUSY and do nothing.
	46	* Should always be manipulated under cpu_add_remove_lock
	47	*/
	48	static int cpu_hotplug_disabled;
	49
d221938c GS	50	static struct {
	51	struct task_struct *active_writer;
	52	struct mutex lock; /* Synchronizes accesses to refcount, */
	53	/*
	54	* Also blocks the new readers during
	55	* an ongoing cpu hotplug operation.
	56	*/
	57	int refcount;
d221938c	58	} cpu_hotplug;
90d45d17	59
d221938c GS	60	void __init cpu_hotplug_init(void)
	61	{
	62	cpu_hotplug.active_writer = NULL;
	63	mutex_init(&cpu_hotplug.lock);
	64	cpu_hotplug.refcount = 0;
d221938c GS	65	}
d221938c GS	66
e761b772 MK	67	cpumask_t cpu_active_map;
e761b772 MK	68
d221938c	69	#ifdef CONFIG_HOTPLUG_CPU
90d45d17	70
86ef5c9a	71	void get_online_cpus(void)
a9d9baa1	72	{
d221938c GS	73	might_sleep();
d221938c GS	74	if (cpu_hotplug.active_writer == current)
aa953877	75	return;
d221938c GS	76	mutex_lock(&cpu_hotplug.lock);
	77	cpu_hotplug.refcount++;
	78	mutex_unlock(&cpu_hotplug.lock);
	79
a9d9baa1	80	}
86ef5c9a	81	EXPORT_SYMBOL_GPL(get_online_cpus);
90d45d17	82
86ef5c9a	83	void put_online_cpus(void)
a9d9baa1	84	{
d221938c	85	if (cpu_hotplug.active_writer == current)
aa953877	86	return;
d221938c	87	mutex_lock(&cpu_hotplug.lock);
d2ba7e2a ON	88	if (!--cpu_hotplug.refcount && unlikely(cpu_hotplug.active_writer))
d2ba7e2a ON	89	wake_up_process(cpu_hotplug.active_writer);
d221938c GS	90	mutex_unlock(&cpu_hotplug.lock);
d221938c GS	91
a9d9baa1	92	}
86ef5c9a	93	EXPORT_SYMBOL_GPL(put_online_cpus);
a9d9baa1	94
a9d9baa1	95	#endif /* CONFIG_HOTPLUG_CPU */
90d45d17	96
d221938c GS	97	/*
	98	* The following two API's must be used when attempting
	99	* to serialize the updates to cpu_online_map, cpu_present_map.
	100	*/
	101	void cpu_maps_update_begin(void)
	102	{
	103	mutex_lock(&cpu_add_remove_lock);
	104	}
	105
	106	void cpu_maps_update_done(void)
	107	{
	108	mutex_unlock(&cpu_add_remove_lock);
	109	}
	110
	111	/*
	112	* This ensures that the hotplug operation can begin only when the
	113	* refcount goes to zero.
	114	*
	115	* Note that during a cpu-hotplug operation, the new readers, if any,
	116	* will be blocked by the cpu_hotplug.lock
	117	*
d2ba7e2a ON	118	* Since cpu_hotplug_begin() is always called after invoking
d2ba7e2a ON	119	* cpu_maps_update_begin(), we can be sure that only one writer is active.
d221938c GS	120	*
	121	* Note that theoretically, there is a possibility of a livelock:
	122	* - Refcount goes to zero, last reader wakes up the sleeping
	123	* writer.
	124	* - Last reader unlocks the cpu_hotplug.lock.
	125	* - A new reader arrives at this moment, bumps up the refcount.
	126	* - The writer acquires the cpu_hotplug.lock finds the refcount
	127	* non zero and goes to sleep again.
	128	*
	129	* However, this is very difficult to achieve in practice since
86ef5c9a	130	* get_online_cpus() not an api which is called all that often.
d221938c GS	131	*
	132	*/
	133	static void cpu_hotplug_begin(void)
	134	{
d221938c	135	cpu_hotplug.active_writer = current;
d2ba7e2a ON	136
	137	for (;;) {
	138	mutex_lock(&cpu_hotplug.lock);
	139	if (likely(!cpu_hotplug.refcount))
	140	break;
	141	__set_current_state(TASK_UNINTERRUPTIBLE);
d221938c GS	142	mutex_unlock(&cpu_hotplug.lock);
d221938c GS	143	schedule();
d221938c	144	}
d221938c GS	145	}
	146
	147	static void cpu_hotplug_done(void)
	148	{
	149	cpu_hotplug.active_writer = NULL;
	150	mutex_unlock(&cpu_hotplug.lock);
	151	}
1da177e4	152	/* Need to know about CPUs going up/down? */
f7b16c10	153	int __ref register_cpu_notifier(struct notifier_block *nb)
1da177e4	154	{
bd5349cf	155	int ret;
d221938c	156	cpu_maps_update_begin();
bd5349cf	157	ret = raw_notifier_chain_register(&cpu_chain, nb);
d221938c	158	cpu_maps_update_done();
bd5349cf	159	return ret;
1da177e4	160	}
65edc68c CS	161
	162	#ifdef CONFIG_HOTPLUG_CPU
	163
1da177e4 LT	164	EXPORT_SYMBOL(register_cpu_notifier);
1da177e4 LT	165
9647155f	166	void __ref unregister_cpu_notifier(struct notifier_block *nb)
1da177e4	167	{
d221938c	168	cpu_maps_update_begin();
bd5349cf	169	raw_notifier_chain_unregister(&cpu_chain, nb);
d221938c	170	cpu_maps_update_done();
1da177e4 LT	171	}
	172	EXPORT_SYMBOL(unregister_cpu_notifier);
	173
1da177e4 LT	174	static inline void check_for_tasks(int cpu)
	175	{
	176	struct task_struct *p;
	177
	178	write_lock_irq(&tasklist_lock);
	179	for_each_process(p) {
	180	if (task_cpu(p) == cpu &&
	181	(!cputime_eq(p->utime, cputime_zero) \|\|
	182	!cputime_eq(p->stime, cputime_zero)))
	183	printk(KERN_WARNING "Task %s (pid = %d) is on cpu %d\
e7407dcc	184	(state = %ld, flags = %x) \n",
ba25f9dc PE	185	p->comm, task_pid_nr(p), cpu,
ba25f9dc PE	186	p->state, p->flags);
1da177e4 LT	187	}
	188	write_unlock_irq(&tasklist_lock);
	189	}
	190
db912f96 AK	191	struct take_cpu_down_param {
	192	unsigned long mod;
	193	void *hcpu;
	194	};
	195
1da177e4	196	/* Take this CPU down. */
514a20a5	197	static int __ref take_cpu_down(void *_param)
1da177e4	198	{
db912f96	199	struct take_cpu_down_param *param = _param;
1da177e4 LT	200	int err;
1da177e4 LT	201
db912f96 AK	202	raw_notifier_call_chain(&cpu_chain, CPU_DYING \| param->mod,
db912f96 AK	203	param->hcpu);
1da177e4 LT	204	/* Ensure this CPU doesn't handle any more interrupts. */
	205	err = __cpu_disable();
	206	if (err < 0)
f3705136	207	return err;
1da177e4	208
f3705136 ZM	209	/* Force idle task to run as soon as we yield: it should
	210	immediately notice cpu is offline and die quickly. */
	211	sched_idle_next();
	212	return 0;
1da177e4 LT	213	}
1da177e4 LT	214
e3920fb4	215	/* Requires cpu_add_remove_lock to be held */
514a20a5	216	static int __ref _cpu_down(unsigned int cpu, int tasks_frozen)
1da177e4	217	{
e7407dcc	218	int err, nr_calls = 0;
1da177e4	219	cpumask_t old_allowed, tmp;
e7407dcc	220	void hcpu = (void )(long)cpu;
8bb78442	221	unsigned long mod = tasks_frozen ? CPU_TASKS_FROZEN : 0;
db912f96 AK	222	struct take_cpu_down_param tcd_param = {
	223	.mod = mod,
	224	.hcpu = hcpu,
	225	};
1da177e4	226
e3920fb4 RW	227	if (num_online_cpus() == 1)
e3920fb4 RW	228	return -EBUSY;
1da177e4	229
e3920fb4 RW	230	if (!cpu_online(cpu))
e3920fb4 RW	231	return -EINVAL;
1da177e4	232
d221938c	233	cpu_hotplug_begin();
8bb78442	234	err = __raw_notifier_call_chain(&cpu_chain, CPU_DOWN_PREPARE \| mod,
e7407dcc	235	hcpu, -1, &nr_calls);
1da177e4	236	if (err == NOTIFY_BAD) {
a0d8cdb6	237	nr_calls--;
8bb78442 RW	238	__raw_notifier_call_chain(&cpu_chain, CPU_DOWN_FAILED \| mod,
8bb78442 RW	239	hcpu, nr_calls, NULL);
1da177e4	240	printk("%s: attempt to take down CPU %u failed\n",
af1f16d0	241	__func__, cpu);
baaca49f GS	242	err = -EINVAL;
baaca49f GS	243	goto out_release;
1da177e4 LT	244	}
	245
	246	/* Ensure that we are not runnable on dying cpu */
	247	old_allowed = current->cpus_allowed;
f70316da	248	cpus_setall(tmp);
1da177e4	249	cpu_clear(cpu, tmp);
f70316da	250	set_cpus_allowed_ptr(current, &tmp);
eeec4fad	251	tmp = cpumask_of_cpu(cpu);
1da177e4	252
eeec4fad	253	err = __stop_machine(take_cpu_down, &tcd_param, &tmp);
04321587	254	if (err) {
1da177e4	255	/* CPU didn't die: tell everyone. Can't complain. */
8bb78442	256	if (raw_notifier_call_chain(&cpu_chain, CPU_DOWN_FAILED \| mod,
e7407dcc	257	hcpu) == NOTIFY_BAD)
1da177e4 LT	258	BUG();
1da177e4 LT	259
ffdb5976	260	goto out_allowed;
8fa1d7d3	261	}
04321587	262	BUG_ON(cpu_online(cpu));
1da177e4 LT	263
	264	/* Wait for it to sleep (leaving idle task). */
	265	while (!idle_cpu(cpu))
	266	yield();
	267
	268	/* This actually kills the CPU. */
	269	__cpu_die(cpu);
	270
1da177e4	271	/* CPU is completely dead: tell everyone. Too late to complain. */
8bb78442 RW	272	if (raw_notifier_call_chain(&cpu_chain, CPU_DEAD \| mod,
8bb78442 RW	273	hcpu) == NOTIFY_BAD)
1da177e4 LT	274	BUG();
	275
	276	check_for_tasks(cpu);
	277
1da177e4	278	out_allowed:
f70316da	279	set_cpus_allowed_ptr(current, &old_allowed);
baaca49f	280	out_release:
d221938c	281	cpu_hotplug_done();
3da1c84c ON	282	if (!err) {
	283	if (raw_notifier_call_chain(&cpu_chain, CPU_POST_DEAD \| mod,
	284	hcpu) == NOTIFY_BAD)
	285	BUG();
	286	}
e3920fb4 RW	287	return err;
	288	}
	289
514a20a5	290	int __ref cpu_down(unsigned int cpu)
e3920fb4 RW	291	{
	292	int err = 0;
	293
d221938c	294	cpu_maps_update_begin();
e761b772 MK	295
e761b772 MK	296	if (cpu_hotplug_disabled) {
e3920fb4	297	err = -EBUSY;
e761b772 MK	298	goto out;
	299	}
	300
	301	cpu_clear(cpu, cpu_active_map);
	302
39b0fad7 MK	303	/*
	304	* Make sure the all cpus did the reschedule and are not
	305	* using stale version of the cpu_active_map.
	306	* This is not strictly necessary becuase stop_machine()
	307	* that we run down the line already provides the required
	308	* synchronization. But it's really a side effect and we do not
	309	* want to depend on the innards of the stop_machine here.
	310	*/
	311	synchronize_sched();
e3920fb4	312
e761b772	313	err = _cpu_down(cpu, 0);
e3920fb4	314
e761b772 MK	315	if (cpu_online(cpu))
	316	cpu_set(cpu, cpu_active_map);
	317
	318	out:
d221938c	319	cpu_maps_update_done();
1da177e4 LT	320	return err;
1da177e4 LT	321	}
b62b8ef9	322	EXPORT_SYMBOL(cpu_down);
1da177e4 LT	323	#endif /CONFIG_HOTPLUG_CPU/
1da177e4 LT	324
e3920fb4	325	/* Requires cpu_add_remove_lock to be held */
8bb78442	326	static int __cpuinit _cpu_up(unsigned int cpu, int tasks_frozen)
1da177e4	327	{
baaca49f	328	int ret, nr_calls = 0;
1da177e4	329	void hcpu = (void )(long)cpu;
8bb78442	330	unsigned long mod = tasks_frozen ? CPU_TASKS_FROZEN : 0;
1da177e4	331
e3920fb4 RW	332	if (cpu_online(cpu) \|\| !cpu_present(cpu))
e3920fb4 RW	333	return -EINVAL;
90d45d17	334
d221938c	335	cpu_hotplug_begin();
8bb78442	336	ret = __raw_notifier_call_chain(&cpu_chain, CPU_UP_PREPARE \| mod, hcpu,
baaca49f	337	-1, &nr_calls);
1da177e4	338	if (ret == NOTIFY_BAD) {
a0d8cdb6	339	nr_calls--;
1da177e4	340	printk("%s: attempt to bring up CPU %u failed\n",
af1f16d0	341	__func__, cpu);
1da177e4 LT	342	ret = -EINVAL;
	343	goto out_notify;
	344	}
	345
	346	/* Arch-specific enabling code. */
	347	ret = __cpu_up(cpu);
	348	if (ret != 0)
	349	goto out_notify;
6978c705	350	BUG_ON(!cpu_online(cpu));
1da177e4	351
279ef6bb DA	352	cpu_set(cpu, cpu_active_map);
279ef6bb DA	353
1da177e4	354	/* Now call notifier in preparation. */
8bb78442	355	raw_notifier_call_chain(&cpu_chain, CPU_ONLINE \| mod, hcpu);
1da177e4 LT	356
	357	out_notify:
	358	if (ret != 0)
baaca49f	359	__raw_notifier_call_chain(&cpu_chain,
8bb78442	360	CPU_UP_CANCELED \| mod, hcpu, nr_calls, NULL);
d221938c	361	cpu_hotplug_done();
e3920fb4 RW	362
	363	return ret;
	364	}
	365
b282b6f8	366	int __cpuinit cpu_up(unsigned int cpu)
e3920fb4 RW	367	{
e3920fb4 RW	368	int err = 0;
73e753a5 KH	369	if (!cpu_isset(cpu, cpu_possible_map)) {
	370	printk(KERN_ERR "can't online cpu %d because it is not "
	371	"configured as may-hotadd at boot time\n", cpu);
	372	#if defined(CONFIG_IA64) \|\| defined(CONFIG_X86_64) \|\| defined(CONFIG_S390)
	373	printk(KERN_ERR "please check additional_cpus= boot "
	374	"parameter\n");
	375	#endif
	376	return -EINVAL;
	377	}
e3920fb4	378
d221938c	379	cpu_maps_update_begin();
e761b772 MK	380
e761b772 MK	381	if (cpu_hotplug_disabled) {
e3920fb4	382	err = -EBUSY;
e761b772 MK	383	goto out;
	384	}
	385
	386	err = _cpu_up(cpu, 0);
	387
e761b772	388	out:
d221938c	389	cpu_maps_update_done();
e3920fb4 RW	390	return err;
	391	}
	392
f3de4be9	393	#ifdef CONFIG_PM_SLEEP_SMP
e3920fb4 RW	394	static cpumask_t frozen_cpus;
	395
	396	int disable_nonboot_cpus(void)
	397	{
e1d9fd2e	398	int cpu, first_cpu, error = 0;
e3920fb4	399
d221938c	400	cpu_maps_update_begin();
1d64b9cb	401	first_cpu = first_cpu(cpu_online_map);
e3920fb4 RW	402	/* We take down all of the non-boot CPUs in one shot to avoid races
	403	* with the userspace trying to use the CPU hotplug at the same time
	404	*/
	405	cpus_clear(frozen_cpus);
	406	printk("Disabling non-boot CPUs ...\n");
	407	for_each_online_cpu(cpu) {
	408	if (cpu == first_cpu)
	409	continue;
8bb78442	410	error = _cpu_down(cpu, 1);
e3920fb4 RW	411	if (!error) {
	412	cpu_set(cpu, frozen_cpus);
	413	printk("CPU%d is down\n", cpu);
	414	} else {
	415	printk(KERN_ERR "Error taking CPU%d down: %d\n",
	416	cpu, error);
	417	break;
	418	}
	419	}
	420	if (!error) {
	421	BUG_ON(num_online_cpus() > 1);
	422	/* Make sure the CPUs won't be enabled by someone else */
	423	cpu_hotplug_disabled = 1;
	424	} else {
e1d9fd2e	425	printk(KERN_ERR "Non-boot CPUs are not disabled\n");
e3920fb4	426	}
d221938c	427	cpu_maps_update_done();
e3920fb4 RW	428	return error;
	429	}
	430
fa7303e2	431	void __ref enable_nonboot_cpus(void)
e3920fb4 RW	432	{
	433	int cpu, error;
	434
	435	/* Allow everyone to use the CPU hotplug again */
d221938c	436	cpu_maps_update_begin();
e3920fb4	437	cpu_hotplug_disabled = 0;
ed746e3b	438	if (cpus_empty(frozen_cpus))
1d64b9cb	439	goto out;
e3920fb4 RW	440
e3920fb4 RW	441	printk("Enabling non-boot CPUs ...\n");
363ab6f1	442	for_each_cpu_mask_nr(cpu, frozen_cpus) {
8bb78442	443	error = _cpu_up(cpu, 1);
e3920fb4 RW	444	if (!error) {
	445	printk("CPU%d is up\n", cpu);
	446	continue;
	447	}
1d64b9cb	448	printk(KERN_WARNING "Error taking CPU%d up: %d\n", cpu, error);
e3920fb4 RW	449	}
e3920fb4 RW	450	cpus_clear(frozen_cpus);
1d64b9cb	451	out:
d221938c	452	cpu_maps_update_done();
1da177e4	453	}
f3de4be9	454	#endif /* CONFIG_PM_SLEEP_SMP */
68f4f1ec MK	455
68f4f1ec MK	456	#endif /* CONFIG_SMP */
b8d317d1	457
e56b3bc7 LT	458	/*
	459	* cpu_bit_bitmap[] is a special, "compressed" data structure that
	460	* represents all NR_CPUS bits binary values of 1<<nr.
	461	*
	462	* It is used by cpumask_of_cpu() to get a constant address to a CPU
	463	* mask value that has a single bit set only.
	464	*/
b8d317d1	465
e56b3bc7 LT	466	/* cpu_bit_bitmap[0] is empty - so we can back into it */
	467	#define MASK_DECLARE_1(x) [x+1][0] = 1UL << (x)
	468	#define MASK_DECLARE_2(x) MASK_DECLARE_1(x), MASK_DECLARE_1(x+1)
	469	#define MASK_DECLARE_4(x) MASK_DECLARE_2(x), MASK_DECLARE_2(x+2)
	470	#define MASK_DECLARE_8(x) MASK_DECLARE_4(x), MASK_DECLARE_4(x+4)
b8d317d1	471
e56b3bc7 LT	472	const unsigned long cpu_bit_bitmap[BITS_PER_LONG+1][BITS_TO_LONGS(NR_CPUS)] = {
	473
	474	MASK_DECLARE_8(0), MASK_DECLARE_8(8),
	475	MASK_DECLARE_8(16), MASK_DECLARE_8(24),
	476	#if BITS_PER_LONG > 32
	477	MASK_DECLARE_8(32), MASK_DECLARE_8(40),
	478	MASK_DECLARE_8(48), MASK_DECLARE_8(56),
b8d317d1 MT	479	#endif
b8d317d1 MT	480	};
e56b3bc7	481	EXPORT_SYMBOL_GPL(cpu_bit_bitmap);