[linux-2.6-block.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>

/* This protects CPUs going up and down... */
static DEFINE_MUTEX(cpu_add_remove_lock);
static DEFINE_MUTEX(cpu_bitmask_lock);

static __cpuinitdata BLOCKING_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;

#ifdef CONFIG_HOTPLUG_CPU

/* Crappy recursive lock-takers in cpufreq! Complain loudly about idiots */
static struct task_struct *recursive;
static int recursive_depth;

void lock_cpu_hotplug(void)
{
	struct task_struct *tsk = current;

	if (tsk == recursive) {
		static int warnings = 10;
		if (warnings) {
			printk(KERN_ERR "Lukewarm IQ detected in hotplug locking\n");
			WARN_ON(1);
			warnings--;
		}
		recursive_depth++;
		return;
	}
	mutex_lock(&cpu_bitmask_lock);
	recursive = tsk;
}
EXPORT_SYMBOL_GPL(lock_cpu_hotplug);

void unlock_cpu_hotplug(void)
{
	WARN_ON(recursive != current);
	if (recursive_depth) {
		recursive_depth--;
		return;
	}
	mutex_unlock(&cpu_bitmask_lock);
	recursive = NULL;
}
EXPORT_SYMBOL_GPL(unlock_cpu_hotplug);

#endif	/* CONFIG_HOTPLUG_CPU */

/* Need to know about CPUs going up/down? */
int __cpuinit register_cpu_notifier(struct notifier_block *nb)
{
	return blocking_notifier_chain_register(&cpu_chain, nb);
}

#ifdef CONFIG_HOTPLUG_CPU

EXPORT_SYMBOL(register_cpu_notifier);

void unregister_cpu_notifier(struct notifier_block *nb)
{
	blocking_notifier_chain_unregister(&cpu_chain, nb);
}
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 = %lx) \n",
				 p->comm, p->pid, cpu, p->state, p->flags);
	}
	write_unlock_irq(&tasklist_lock);
}

/* Take this CPU down. */
static int take_cpu_down(void *unused)
{
	int err;

	/* 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 _cpu_down(unsigned int cpu)
{
	int err;
	struct task_struct *p;
	cpumask_t old_allowed, tmp;

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

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

	err = blocking_notifier_call_chain(&cpu_chain, CPU_DOWN_PREPARE,
						(void *)(long)cpu);
	if (err == NOTIFY_BAD) {
		printk("%s: attempt to take down CPU %u failed\n",
				__FUNCTION__, cpu);
		return -EINVAL;
	}

	/* Ensure that we are not runnable on dying cpu */
	old_allowed = current->cpus_allowed;
	tmp = CPU_MASK_ALL;
	cpu_clear(cpu, tmp);
	set_cpus_allowed(current, tmp);

	mutex_lock(&cpu_bitmask_lock);
	p = __stop_machine_run(take_cpu_down, NULL, cpu);
	mutex_unlock(&cpu_bitmask_lock);

	if (IS_ERR(p)) {
		/* CPU didn't die: tell everyone.  Can't complain. */
		if (blocking_notifier_call_chain(&cpu_chain, CPU_DOWN_FAILED,
				(void *)(long)cpu) == NOTIFY_BAD)
			BUG();

		err = PTR_ERR(p);
		goto out_allowed;
	}

	if (cpu_online(cpu))
		goto out_thread;

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

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

	/* Move it here so it can run. */
	kthread_bind(p, get_cpu());
	put_cpu();

	/* CPU is completely dead: tell everyone.  Too late to complain. */
	if (blocking_notifier_call_chain(&cpu_chain, CPU_DEAD,
			(void *)(long)cpu) == NOTIFY_BAD)
		BUG();

	check_for_tasks(cpu);

out_thread:
	err = kthread_stop(p);
out_allowed:
	set_cpus_allowed(current, old_allowed);
	return err;
}

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

	mutex_lock(&cpu_add_remove_lock);
	if (cpu_hotplug_disabled)
		err = -EBUSY;
	else
		err = _cpu_down(cpu);

	mutex_unlock(&cpu_add_remove_lock);
	return err;
}
#endif /*CONFIG_HOTPLUG_CPU*/

/* Requires cpu_add_remove_lock to be held */
static int __devinit _cpu_up(unsigned int cpu)
{
	int ret;
	void *hcpu = (void *)(long)cpu;

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

	ret = blocking_notifier_call_chain(&cpu_chain, CPU_UP_PREPARE, hcpu);
	if (ret == NOTIFY_BAD) {
		printk("%s: attempt to bring up CPU %u failed\n",
				__FUNCTION__, cpu);
		ret = -EINVAL;
		goto out_notify;
	}

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

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

out_notify:
	if (ret != 0)
		blocking_notifier_call_chain(&cpu_chain,
				CPU_UP_CANCELED, hcpu);

	return ret;
}

int __devinit cpu_up(unsigned int cpu)
{
	int err = 0;

	mutex_lock(&cpu_add_remove_lock);
	if (cpu_hotplug_disabled)
		err = -EBUSY;
	else
		err = _cpu_up(cpu);

	mutex_unlock(&cpu_add_remove_lock);
	return err;
}

#ifdef CONFIG_SUSPEND_SMP
static cpumask_t frozen_cpus;

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

	mutex_lock(&cpu_add_remove_lock);
	first_cpu = first_cpu(cpu_present_map);
	if (!cpu_online(first_cpu)) {
		error = _cpu_up(first_cpu);
		if (error) {
			printk(KERN_ERR "Could not bring CPU%d up.\n",
				first_cpu);
			goto out;
		}
	}
	error = set_cpus_allowed(current, cpumask_of_cpu(first_cpu));
	if (error) {
		printk(KERN_ERR "Could not run on CPU%d\n", first_cpu);
		goto out;
	}
	/* 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);
		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");
	}
out:
	mutex_unlock(&cpu_add_remove_lock);
	return error;
}

void enable_nonboot_cpus(void)
{
	int cpu, error;

	/* Allow everyone to use the CPU hotplug again */
	mutex_lock(&cpu_add_remove_lock);
	cpu_hotplug_disabled = 0;
	mutex_unlock(&cpu_add_remove_lock);

	printk("Enabling non-boot CPUs ...\n");
	for_each_cpu_mask(cpu, frozen_cpus) {
		error = cpu_up(cpu);
		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);
}
#endif
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 LT	17
1da177e4 LT	18	/* This protects CPUs going up and down... */
aa953877 LT	19	static DEFINE_MUTEX(cpu_add_remove_lock);
aa953877 LT	20	static DEFINE_MUTEX(cpu_bitmask_lock);
1da177e4	21
65edc68c	22	static __cpuinitdata BLOCKING_NOTIFIER_HEAD(cpu_chain);
1da177e4	23
e3920fb4 RW	24	/* If set, cpu_up and cpu_down will return -EBUSY and do nothing.
	25	* Should always be manipulated under cpu_add_remove_lock
	26	*/
	27	static int cpu_hotplug_disabled;
	28
a9d9baa1	29	#ifdef CONFIG_HOTPLUG_CPU
90d45d17	30
aa953877 LT	31	/* Crappy recursive lock-takers in cpufreq! Complain loudly about idiots */
	32	static struct task_struct *recursive;
	33	static int recursive_depth;
90d45d17	34
a9d9baa1 AR	35	void lock_cpu_hotplug(void)
a9d9baa1 AR	36	{
aa953877 LT	37	struct task_struct *tsk = current;
	38
	39	if (tsk == recursive) {
	40	static int warnings = 10;
	41	if (warnings) {
	42	printk(KERN_ERR "Lukewarm IQ detected in hotplug locking\n");
	43	WARN_ON(1);
	44	warnings--;
	45	}
	46	recursive_depth++;
	47	return;
	48	}
	49	mutex_lock(&cpu_bitmask_lock);
	50	recursive = tsk;
a9d9baa1 AR	51	}
a9d9baa1 AR	52	EXPORT_SYMBOL_GPL(lock_cpu_hotplug);
90d45d17	53
a9d9baa1 AR	54	void unlock_cpu_hotplug(void)
a9d9baa1 AR	55	{
aa953877 LT	56	WARN_ON(recursive != current);
	57	if (recursive_depth) {
	58	recursive_depth--;
	59	return;
a9d9baa1	60	}
aa953877 LT	61	mutex_unlock(&cpu_bitmask_lock);
aa953877 LT	62	recursive = NULL;
a9d9baa1 AR	63	}
	64	EXPORT_SYMBOL_GPL(unlock_cpu_hotplug);
	65
a9d9baa1	66	#endif /* CONFIG_HOTPLUG_CPU */
90d45d17	67
1da177e4	68	/* Need to know about CPUs going up/down? */
65edc68c	69	int __cpuinit register_cpu_notifier(struct notifier_block *nb)
1da177e4	70	{
e041c683	71	return blocking_notifier_chain_register(&cpu_chain, nb);
1da177e4	72	}
65edc68c CS	73
	74	#ifdef CONFIG_HOTPLUG_CPU
	75
1da177e4 LT	76	EXPORT_SYMBOL(register_cpu_notifier);
	77
	78	void unregister_cpu_notifier(struct notifier_block *nb)
	79	{
e041c683	80	blocking_notifier_chain_unregister(&cpu_chain, nb);
1da177e4 LT	81	}
	82	EXPORT_SYMBOL(unregister_cpu_notifier);
	83
1da177e4 LT	84	static inline void check_for_tasks(int cpu)
	85	{
	86	struct task_struct *p;
	87
	88	write_lock_irq(&tasklist_lock);
	89	for_each_process(p) {
	90	if (task_cpu(p) == cpu &&
	91	(!cputime_eq(p->utime, cputime_zero) \|\|
	92	!cputime_eq(p->stime, cputime_zero)))
	93	printk(KERN_WARNING "Task %s (pid = %d) is on cpu %d\
	94	(state = %ld, flags = %lx) \n",
	95	p->comm, p->pid, cpu, p->state, p->flags);
	96	}
	97	write_unlock_irq(&tasklist_lock);
	98	}
	99
	100	/* Take this CPU down. */
	101	static int take_cpu_down(void *unused)
	102	{
	103	int err;
	104
1da177e4 LT	105	/* Ensure this CPU doesn't handle any more interrupts. */
	106	err = __cpu_disable();
	107	if (err < 0)
f3705136	108	return err;
1da177e4	109
f3705136 ZM	110	/* Force idle task to run as soon as we yield: it should
	111	immediately notice cpu is offline and die quickly. */
	112	sched_idle_next();
	113	return 0;
1da177e4 LT	114	}
1da177e4 LT	115
e3920fb4 RW	116	/* Requires cpu_add_remove_lock to be held */
e3920fb4 RW	117	static int _cpu_down(unsigned int cpu)
1da177e4 LT	118	{
	119	int err;
	120	struct task_struct *p;
	121	cpumask_t old_allowed, tmp;
	122
e3920fb4 RW	123	if (num_online_cpus() == 1)
e3920fb4 RW	124	return -EBUSY;
1da177e4	125
e3920fb4 RW	126	if (!cpu_online(cpu))
e3920fb4 RW	127	return -EINVAL;
1da177e4	128
e041c683	129	err = blocking_notifier_call_chain(&cpu_chain, CPU_DOWN_PREPARE,
1da177e4 LT	130	(void *)(long)cpu);
	131	if (err == NOTIFY_BAD) {
	132	printk("%s: attempt to take down CPU %u failed\n",
	133	__FUNCTION__, cpu);
e3920fb4	134	return -EINVAL;
1da177e4 LT	135	}
	136
	137	/* Ensure that we are not runnable on dying cpu */
	138	old_allowed = current->cpus_allowed;
	139	tmp = CPU_MASK_ALL;
	140	cpu_clear(cpu, tmp);
	141	set_cpus_allowed(current, tmp);
	142
aa953877	143	mutex_lock(&cpu_bitmask_lock);
1da177e4	144	p = __stop_machine_run(take_cpu_down, NULL, cpu);
aa953877 LT	145	mutex_unlock(&cpu_bitmask_lock);
aa953877 LT	146
1da177e4 LT	147	if (IS_ERR(p)) {
1da177e4 LT	148	/* CPU didn't die: tell everyone. Can't complain. */
e041c683	149	if (blocking_notifier_call_chain(&cpu_chain, CPU_DOWN_FAILED,
1da177e4 LT	150	(void *)(long)cpu) == NOTIFY_BAD)
	151	BUG();
	152
	153	err = PTR_ERR(p);
	154	goto out_allowed;
	155	}
	156
	157	if (cpu_online(cpu))
	158	goto out_thread;
	159
	160	/* Wait for it to sleep (leaving idle task). */
	161	while (!idle_cpu(cpu))
	162	yield();
	163
	164	/* This actually kills the CPU. */
	165	__cpu_die(cpu);
	166
	167	/* Move it here so it can run. */
	168	kthread_bind(p, get_cpu());
	169	put_cpu();
	170
	171	/* CPU is completely dead: tell everyone. Too late to complain. */
e041c683 AS	172	if (blocking_notifier_call_chain(&cpu_chain, CPU_DEAD,
e041c683 AS	173	(void *)(long)cpu) == NOTIFY_BAD)
1da177e4 LT	174	BUG();
	175
	176	check_for_tasks(cpu);
	177
	178	out_thread:
	179	err = kthread_stop(p);
	180	out_allowed:
	181	set_cpus_allowed(current, old_allowed);
e3920fb4 RW	182	return err;
	183	}
	184
	185	int cpu_down(unsigned int cpu)
	186	{
	187	int err = 0;
	188
	189	mutex_lock(&cpu_add_remove_lock);
	190	if (cpu_hotplug_disabled)
	191	err = -EBUSY;
	192	else
	193	err = _cpu_down(cpu);
	194
aa953877	195	mutex_unlock(&cpu_add_remove_lock);
1da177e4 LT	196	return err;
	197	}
	198	#endif /CONFIG_HOTPLUG_CPU/
	199
e3920fb4 RW	200	/* Requires cpu_add_remove_lock to be held */
e3920fb4 RW	201	static int __devinit _cpu_up(unsigned int cpu)
1da177e4 LT	202	{
	203	int ret;
	204	void hcpu = (void )(long)cpu;
	205
e3920fb4 RW	206	if (cpu_online(cpu) \|\| !cpu_present(cpu))
e3920fb4 RW	207	return -EINVAL;
90d45d17	208
e041c683	209	ret = blocking_notifier_call_chain(&cpu_chain, CPU_UP_PREPARE, hcpu);
1da177e4 LT	210	if (ret == NOTIFY_BAD) {
	211	printk("%s: attempt to bring up CPU %u failed\n",
	212	__FUNCTION__, cpu);
	213	ret = -EINVAL;
	214	goto out_notify;
	215	}
	216
	217	/* Arch-specific enabling code. */
aa953877	218	mutex_lock(&cpu_bitmask_lock);
1da177e4	219	ret = __cpu_up(cpu);
aa953877	220	mutex_unlock(&cpu_bitmask_lock);
1da177e4 LT	221	if (ret != 0)
1da177e4 LT	222	goto out_notify;
6978c705	223	BUG_ON(!cpu_online(cpu));
1da177e4 LT	224
1da177e4 LT	225	/* Now call notifier in preparation. */
e041c683	226	blocking_notifier_call_chain(&cpu_chain, CPU_ONLINE, hcpu);
1da177e4 LT	227
	228	out_notify:
	229	if (ret != 0)
e041c683 AS	230	blocking_notifier_call_chain(&cpu_chain,
e041c683 AS	231	CPU_UP_CANCELED, hcpu);
e3920fb4 RW	232
	233	return ret;
	234	}
	235
	236	int __devinit cpu_up(unsigned int cpu)
	237	{
	238	int err = 0;
	239
	240	mutex_lock(&cpu_add_remove_lock);
	241	if (cpu_hotplug_disabled)
	242	err = -EBUSY;
	243	else
	244	err = _cpu_up(cpu);
	245
	246	mutex_unlock(&cpu_add_remove_lock);
	247	return err;
	248	}
	249
	250	#ifdef CONFIG_SUSPEND_SMP
	251	static cpumask_t frozen_cpus;
	252
	253	int disable_nonboot_cpus(void)
	254	{
	255	int cpu, first_cpu, error;
	256
	257	mutex_lock(&cpu_add_remove_lock);
	258	first_cpu = first_cpu(cpu_present_map);
	259	if (!cpu_online(first_cpu)) {
	260	error = _cpu_up(first_cpu);
	261	if (error) {
	262	printk(KERN_ERR "Could not bring CPU%d up.\n",
	263	first_cpu);
	264	goto out;
	265	}
	266	}
	267	error = set_cpus_allowed(current, cpumask_of_cpu(first_cpu));
	268	if (error) {
	269	printk(KERN_ERR "Could not run on CPU%d\n", first_cpu);
	270	goto out;
	271	}
	272	/* We take down all of the non-boot CPUs in one shot to avoid races
	273	* with the userspace trying to use the CPU hotplug at the same time
	274	*/
	275	cpus_clear(frozen_cpus);
	276	printk("Disabling non-boot CPUs ...\n");
	277	for_each_online_cpu(cpu) {
	278	if (cpu == first_cpu)
	279	continue;
	280	error = _cpu_down(cpu);
	281	if (!error) {
	282	cpu_set(cpu, frozen_cpus);
	283	printk("CPU%d is down\n", cpu);
	284	} else {
	285	printk(KERN_ERR "Error taking CPU%d down: %d\n",
	286	cpu, error);
	287	break;
	288	}
	289	}
	290	if (!error) {
	291	BUG_ON(num_online_cpus() > 1);
	292	/* Make sure the CPUs won't be enabled by someone else */
	293	cpu_hotplug_disabled = 1;
	294	} else {
	295	printk(KERN_ERR "Non-boot CPUs are not disabled");
296	}
1da177e4	297	out:
aa953877	298	mutex_unlock(&cpu_add_remove_lock);
e3920fb4 RW	299	return error;
	300	}
	301
	302	void enable_nonboot_cpus(void)
	303	{
	304	int cpu, error;
	305
	306	/* Allow everyone to use the CPU hotplug again */
	307	mutex_lock(&cpu_add_remove_lock);
	308	cpu_hotplug_disabled = 0;
	309	mutex_unlock(&cpu_add_remove_lock);
	310
	311	printk("Enabling non-boot CPUs ...\n");
	312	for_each_cpu_mask(cpu, frozen_cpus) {
	313	error = cpu_up(cpu);
	314	if (!error) {
	315	printk("CPU%d is up\n", cpu);
	316	continue;
	317	}
	318	printk(KERN_WARNING "Error taking CPU%d up: %d\n",
	319	cpu, error);
	320	}
	321	cpus_clear(frozen_cpus);
1da177e4	322	}
e3920fb4	323	#endif