[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 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;

#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)
{
	int ret;
	mutex_lock(&cpu_add_remove_lock);
	ret = raw_notifier_chain_register(&cpu_chain, nb);
	mutex_unlock(&cpu_add_remove_lock);
	return ret;
}

#ifdef CONFIG_HOTPLUG_CPU

EXPORT_SYMBOL(register_cpu_notifier);

void unregister_cpu_notifier(struct notifier_block *nb)
{
	mutex_lock(&cpu_add_remove_lock);
	raw_notifier_chain_unregister(&cpu_chain, nb);
	mutex_unlock(&cpu_add_remove_lock);
}
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 = raw_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 (raw_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 (raw_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 = raw_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. */
	raw_notifier_call_chain(&cpu_chain, CPU_ONLINE, hcpu);

out_notify:
	if (ret != 0)
		raw_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
bd5349cf	22	static __cpuinitdata RAW_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	{
bd5349cf NB	71	int ret;
	72	mutex_lock(&cpu_add_remove_lock);
	73	ret = raw_notifier_chain_register(&cpu_chain, nb);
	74	mutex_unlock(&cpu_add_remove_lock);
	75	return ret;
1da177e4	76	}
65edc68c CS	77
	78	#ifdef CONFIG_HOTPLUG_CPU
	79
1da177e4 LT	80	EXPORT_SYMBOL(register_cpu_notifier);
	81
	82	void unregister_cpu_notifier(struct notifier_block *nb)
	83	{
bd5349cf NB	84	mutex_lock(&cpu_add_remove_lock);
	85	raw_notifier_chain_unregister(&cpu_chain, nb);
	86	mutex_unlock(&cpu_add_remove_lock);
1da177e4 LT	87	}
	88	EXPORT_SYMBOL(unregister_cpu_notifier);
	89
1da177e4 LT	90	static inline void check_for_tasks(int cpu)
	91	{
	92	struct task_struct *p;
	93
	94	write_lock_irq(&tasklist_lock);
	95	for_each_process(p) {
	96	if (task_cpu(p) == cpu &&
	97	(!cputime_eq(p->utime, cputime_zero) \|\|
	98	!cputime_eq(p->stime, cputime_zero)))
	99	printk(KERN_WARNING "Task %s (pid = %d) is on cpu %d\
	100	(state = %ld, flags = %lx) \n",
	101	p->comm, p->pid, cpu, p->state, p->flags);
	102	}
	103	write_unlock_irq(&tasklist_lock);
	104	}
	105
	106	/* Take this CPU down. */
	107	static int take_cpu_down(void *unused)
	108	{
	109	int err;
	110
1da177e4 LT	111	/* Ensure this CPU doesn't handle any more interrupts. */
	112	err = __cpu_disable();
	113	if (err < 0)
f3705136	114	return err;
1da177e4	115
f3705136 ZM	116	/* Force idle task to run as soon as we yield: it should
	117	immediately notice cpu is offline and die quickly. */
	118	sched_idle_next();
	119	return 0;
1da177e4 LT	120	}
1da177e4 LT	121
e3920fb4 RW	122	/* Requires cpu_add_remove_lock to be held */
e3920fb4 RW	123	static int _cpu_down(unsigned int cpu)
1da177e4 LT	124	{
	125	int err;
	126	struct task_struct *p;
	127	cpumask_t old_allowed, tmp;
	128
e3920fb4 RW	129	if (num_online_cpus() == 1)
e3920fb4 RW	130	return -EBUSY;
1da177e4	131
e3920fb4 RW	132	if (!cpu_online(cpu))
e3920fb4 RW	133	return -EINVAL;
1da177e4	134
bd5349cf	135	err = raw_notifier_call_chain(&cpu_chain, CPU_DOWN_PREPARE,
1da177e4 LT	136	(void *)(long)cpu);
	137	if (err == NOTIFY_BAD) {
	138	printk("%s: attempt to take down CPU %u failed\n",
	139	__FUNCTION__, cpu);
e3920fb4	140	return -EINVAL;
1da177e4 LT	141	}
	142
	143	/* Ensure that we are not runnable on dying cpu */
	144	old_allowed = current->cpus_allowed;
	145	tmp = CPU_MASK_ALL;
	146	cpu_clear(cpu, tmp);
	147	set_cpus_allowed(current, tmp);
	148
aa953877	149	mutex_lock(&cpu_bitmask_lock);
1da177e4	150	p = __stop_machine_run(take_cpu_down, NULL, cpu);
aa953877 LT	151	mutex_unlock(&cpu_bitmask_lock);
aa953877 LT	152
1da177e4 LT	153	if (IS_ERR(p)) {
1da177e4 LT	154	/* CPU didn't die: tell everyone. Can't complain. */
bd5349cf	155	if (raw_notifier_call_chain(&cpu_chain, CPU_DOWN_FAILED,
1da177e4 LT	156	(void *)(long)cpu) == NOTIFY_BAD)
	157	BUG();
	158
	159	err = PTR_ERR(p);
	160	goto out_allowed;
	161	}
	162
	163	if (cpu_online(cpu))
	164	goto out_thread;
	165
	166	/* Wait for it to sleep (leaving idle task). */
	167	while (!idle_cpu(cpu))
	168	yield();
	169
	170	/* This actually kills the CPU. */
	171	__cpu_die(cpu);
	172
	173	/* Move it here so it can run. */
	174	kthread_bind(p, get_cpu());
	175	put_cpu();
	176
	177	/* CPU is completely dead: tell everyone. Too late to complain. */
bd5349cf	178	if (raw_notifier_call_chain(&cpu_chain, CPU_DEAD,
e041c683	179	(void *)(long)cpu) == NOTIFY_BAD)
1da177e4 LT	180	BUG();
	181
	182	check_for_tasks(cpu);
	183
	184	out_thread:
	185	err = kthread_stop(p);
	186	out_allowed:
	187	set_cpus_allowed(current, old_allowed);
e3920fb4 RW	188	return err;
	189	}
	190
	191	int cpu_down(unsigned int cpu)
	192	{
	193	int err = 0;
	194
	195	mutex_lock(&cpu_add_remove_lock);
	196	if (cpu_hotplug_disabled)
	197	err = -EBUSY;
	198	else
	199	err = _cpu_down(cpu);
	200
aa953877	201	mutex_unlock(&cpu_add_remove_lock);
1da177e4 LT	202	return err;
	203	}
	204	#endif /CONFIG_HOTPLUG_CPU/
	205
e3920fb4 RW	206	/* Requires cpu_add_remove_lock to be held */
e3920fb4 RW	207	static int __devinit _cpu_up(unsigned int cpu)
1da177e4 LT	208	{
	209	int ret;
	210	void hcpu = (void )(long)cpu;
	211
e3920fb4 RW	212	if (cpu_online(cpu) \|\| !cpu_present(cpu))
e3920fb4 RW	213	return -EINVAL;
90d45d17	214
bd5349cf	215	ret = raw_notifier_call_chain(&cpu_chain, CPU_UP_PREPARE, hcpu);
1da177e4 LT	216	if (ret == NOTIFY_BAD) {
	217	printk("%s: attempt to bring up CPU %u failed\n",
	218	__FUNCTION__, cpu);
	219	ret = -EINVAL;
	220	goto out_notify;
	221	}
	222
	223	/* Arch-specific enabling code. */
aa953877	224	mutex_lock(&cpu_bitmask_lock);
1da177e4	225	ret = __cpu_up(cpu);
aa953877	226	mutex_unlock(&cpu_bitmask_lock);
1da177e4 LT	227	if (ret != 0)
1da177e4 LT	228	goto out_notify;
6978c705	229	BUG_ON(!cpu_online(cpu));
1da177e4 LT	230
1da177e4 LT	231	/* Now call notifier in preparation. */
bd5349cf	232	raw_notifier_call_chain(&cpu_chain, CPU_ONLINE, hcpu);
1da177e4 LT	233
	234	out_notify:
	235	if (ret != 0)
bd5349cf	236	raw_notifier_call_chain(&cpu_chain,
e041c683	237	CPU_UP_CANCELED, hcpu);
e3920fb4 RW	238
	239	return ret;
	240	}
	241
	242	int __devinit cpu_up(unsigned int cpu)
	243	{
	244	int err = 0;
	245
	246	mutex_lock(&cpu_add_remove_lock);
	247	if (cpu_hotplug_disabled)
	248	err = -EBUSY;
	249	else
	250	err = _cpu_up(cpu);
	251
	252	mutex_unlock(&cpu_add_remove_lock);
	253	return err;
	254	}
	255
	256	#ifdef CONFIG_SUSPEND_SMP
	257	static cpumask_t frozen_cpus;
	258
	259	int disable_nonboot_cpus(void)
	260	{
	261	int cpu, first_cpu, error;
	262
	263	mutex_lock(&cpu_add_remove_lock);
	264	first_cpu = first_cpu(cpu_present_map);
	265	if (!cpu_online(first_cpu)) {
	266	error = _cpu_up(first_cpu);
	267	if (error) {
	268	printk(KERN_ERR "Could not bring CPU%d up.\n",
	269	first_cpu);
	270	goto out;
	271	}
	272	}
	273	error = set_cpus_allowed(current, cpumask_of_cpu(first_cpu));
	274	if (error) {
	275	printk(KERN_ERR "Could not run on CPU%d\n", first_cpu);
	276	goto out;
	277	}
	278	/* We take down all of the non-boot CPUs in one shot to avoid races
	279	* with the userspace trying to use the CPU hotplug at the same time
	280	*/
	281	cpus_clear(frozen_cpus);
	282	printk("Disabling non-boot CPUs ...\n");
	283	for_each_online_cpu(cpu) {
	284	if (cpu == first_cpu)
	285	continue;
	286	error = _cpu_down(cpu);
	287	if (!error) {
	288	cpu_set(cpu, frozen_cpus);
	289	printk("CPU%d is down\n", cpu);
	290	} else {
	291	printk(KERN_ERR "Error taking CPU%d down: %d\n",
	292	cpu, error);
	293	break;
	294	}
	295	}
	296	if (!error) {
	297	BUG_ON(num_online_cpus() > 1);
	298	/* Make sure the CPUs won't be enabled by someone else */
	299	cpu_hotplug_disabled = 1;
	300	} else {
	301	printk(KERN_ERR "Non-boot CPUs are not disabled");
302	}
1da177e4	303	out:
aa953877	304	mutex_unlock(&cpu_add_remove_lock);
e3920fb4 RW	305	return error;
	306	}
	307
	308	void enable_nonboot_cpus(void)
	309	{
	310	int cpu, error;
	311
	312	/* Allow everyone to use the CPU hotplug again */
	313	mutex_lock(&cpu_add_remove_lock);
	314	cpu_hotplug_disabled = 0;
	315	mutex_unlock(&cpu_add_remove_lock);
	316
	317	printk("Enabling non-boot CPUs ...\n");
	318	for_each_cpu_mask(cpu, frozen_cpus) {
	319	error = cpu_up(cpu);
	320	if (!error) {
	321	printk("CPU%d is up\n", cpu);
	322	continue;
	323	}
	324	printk(KERN_WARNING "Error taking CPU%d up: %d\n",
	325	cpu, error);
	326	}
	327	cpus_clear(frozen_cpus);
1da177e4	328	}
e3920fb4	329	#endif