[linux-2.6-block.git] / drivers / base / cpu.c

/*
 * CPU subsystem support
 */

#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/init.h>
#include <linux/sched.h>
#include <linux/cpu.h>
#include <linux/topology.h>
#include <linux/device.h>
#include <linux/node.h>
#include <linux/gfp.h>
#include <linux/percpu.h>

#include "base.h"

struct bus_type cpu_subsys = {
	.name = "cpu",
	.dev_name = "cpu",
};
EXPORT_SYMBOL_GPL(cpu_subsys);

static DEFINE_PER_CPU(struct device *, cpu_sys_devices);

#ifdef CONFIG_HOTPLUG_CPU
static ssize_t show_online(struct device *dev,
			   struct device_attribute *attr,
			   char *buf)
{
	struct cpu *cpu = container_of(dev, struct cpu, dev);

	return sprintf(buf, "%u\n", !!cpu_online(cpu->dev.id));
}

static ssize_t __ref store_online(struct device *dev,
				  struct device_attribute *attr,
				  const char *buf, size_t count)
{
	struct cpu *cpu = container_of(dev, struct cpu, dev);
	ssize_t ret;

	cpu_hotplug_driver_lock();
	switch (buf[0]) {
	case '0':
		ret = cpu_down(cpu->dev.id);
		if (!ret)
			kobject_uevent(&dev->kobj, KOBJ_OFFLINE);
		break;
	case '1':
		ret = cpu_up(cpu->dev.id);
		if (!ret)
			kobject_uevent(&dev->kobj, KOBJ_ONLINE);
		break;
	default:
		ret = -EINVAL;
	}
	cpu_hotplug_driver_unlock();

	if (ret >= 0)
		ret = count;
	return ret;
}
static DEVICE_ATTR(online, 0644, show_online, store_online);

static void __cpuinit register_cpu_control(struct cpu *cpu)
{
	device_create_file(&cpu->dev, &dev_attr_online);
}
void unregister_cpu(struct cpu *cpu)
{
	int logical_cpu = cpu->dev.id;

	unregister_cpu_under_node(logical_cpu, cpu_to_node(logical_cpu));

	device_remove_file(&cpu->dev, &dev_attr_online);

	device_unregister(&cpu->dev);
	per_cpu(cpu_sys_devices, logical_cpu) = NULL;
	return;
}

#ifdef CONFIG_ARCH_CPU_PROBE_RELEASE
static ssize_t cpu_probe_store(struct device *dev,
			       struct device_attribute *attr,
			       const char *buf,
			       size_t count)
{
	return arch_cpu_probe(buf, count);
}

static ssize_t cpu_release_store(struct device *dev,
				 struct device_attribute *attr,
				 const char *buf,
				 size_t count)
{
	return arch_cpu_release(buf, count);
}

static DEVICE_ATTR(probe, S_IWUSR, NULL, cpu_probe_store);
static DEVICE_ATTR(release, S_IWUSR, NULL, cpu_release_store);
#endif /* CONFIG_ARCH_CPU_PROBE_RELEASE */

#else /* ... !CONFIG_HOTPLUG_CPU */
static inline void register_cpu_control(struct cpu *cpu)
{
}
#endif /* CONFIG_HOTPLUG_CPU */

#ifdef CONFIG_KEXEC
#include <linux/kexec.h>

static ssize_t show_crash_notes(struct device *dev, struct device_attribute *attr,
				char *buf)
{
	struct cpu *cpu = container_of(dev, struct cpu, dev);
	ssize_t rc;
	unsigned long long addr;
	int cpunum;

	cpunum = cpu->dev.id;

	/*
	 * Might be reading other cpu's data based on which cpu read thread
	 * has been scheduled. But cpu data (memory) is allocated once during
	 * boot up and this data does not change there after. Hence this
	 * operation should be safe. No locking required.
	 */
	addr = per_cpu_ptr_to_phys(per_cpu_ptr(crash_notes, cpunum));
	rc = sprintf(buf, "%Lx\n", addr);
	return rc;
}
static DEVICE_ATTR(crash_notes, 0400, show_crash_notes, NULL);
#endif

/*
 * Print cpu online, possible, present, and system maps
 */

struct cpu_attr {
	struct device_attribute attr;
	const struct cpumask *const * const map;
};

static ssize_t show_cpus_attr(struct device *dev,
			      struct device_attribute *attr,
			      char *buf)
{
	struct cpu_attr *ca = container_of(attr, struct cpu_attr, attr);
	int n = cpulist_scnprintf(buf, PAGE_SIZE-2, *(ca->map));

	buf[n++] = '\n';
	buf[n] = '\0';
	return n;
}

#define _CPU_ATTR(name, map) \
	{ __ATTR(name, 0444, show_cpus_attr, NULL), map }

/* Keep in sync with cpu_subsys_attrs */
static struct cpu_attr cpu_attrs[] = {
	_CPU_ATTR(online, &cpu_online_mask),
	_CPU_ATTR(possible, &cpu_possible_mask),
	_CPU_ATTR(present, &cpu_present_mask),
};

/*
 * Print values for NR_CPUS and offlined cpus
 */
static ssize_t print_cpus_kernel_max(struct device *dev,
				     struct device_attribute *attr, char *buf)
{
	int n = snprintf(buf, PAGE_SIZE-2, "%d\n", NR_CPUS - 1);
	return n;
}
static DEVICE_ATTR(kernel_max, 0444, print_cpus_kernel_max, NULL);

/* arch-optional setting to enable display of offline cpus >= nr_cpu_ids */
unsigned int total_cpus;

static ssize_t print_cpus_offline(struct device *dev,
				  struct device_attribute *attr, char *buf)
{
	int n = 0, len = PAGE_SIZE-2;
	cpumask_var_t offline;

	/* display offline cpus < nr_cpu_ids */
	if (!alloc_cpumask_var(&offline, GFP_KERNEL))
		return -ENOMEM;
	cpumask_andnot(offline, cpu_possible_mask, cpu_online_mask);
	n = cpulist_scnprintf(buf, len, offline);
	free_cpumask_var(offline);

	/* display offline cpus >= nr_cpu_ids */
	if (total_cpus && nr_cpu_ids < total_cpus) {
		if (n && n < len)
			buf[n++] = ',';

		if (nr_cpu_ids == total_cpus-1)
			n += snprintf(&buf[n], len - n, "%d", nr_cpu_ids);
		else
			n += snprintf(&buf[n], len - n, "%d-%d",
						      nr_cpu_ids, total_cpus-1);
	}

	n += snprintf(&buf[n], len - n, "\n");
	return n;
}
static DEVICE_ATTR(offline, 0444, print_cpus_offline, NULL);

static void cpu_device_release(struct device *dev)
{
	/*
	 * This is an empty function to prevent the driver core from spitting a
	 * warning at us.  Yes, I know this is directly opposite of what the
	 * documentation for the driver core and kobjects say, and the author
	 * of this code has already been publically ridiculed for doing
	 * something as foolish as this.  However, at this point in time, it is
	 * the only way to handle the issue of statically allocated cpu
	 * devices.  The different architectures will have their cpu device
	 * code reworked to properly handle this in the near future, so this
	 * function will then be changed to correctly free up the memory held
	 * by the cpu device.
	 *
	 * Never copy this way of doing things, or you too will be made fun of
	 * on the linux-kerenl list, you have been warned.
	 */
}

/*
 * register_cpu - Setup a sysfs device for a CPU.
 * @cpu - cpu->hotpluggable field set to 1 will generate a control file in
 *	  sysfs for this CPU.
 * @num - CPU number to use when creating the device.
 *
 * Initialize and register the CPU device.
 */
int __cpuinit register_cpu(struct cpu *cpu, int num)
{
	int error;

	cpu->node_id = cpu_to_node(num);
	cpu->dev.id = num;
	cpu->dev.bus = &cpu_subsys;
	cpu->dev.release = cpu_device_release;
	error = device_register(&cpu->dev);
	if (!error && cpu->hotpluggable)
		register_cpu_control(cpu);
	if (!error)
		per_cpu(cpu_sys_devices, num) = &cpu->dev;
	if (!error)
		register_cpu_under_node(num, cpu_to_node(num));

#ifdef CONFIG_KEXEC
	if (!error)
		error = device_create_file(&cpu->dev, &dev_attr_crash_notes);
#endif
	return error;
}

struct device *get_cpu_device(unsigned cpu)
{
	if (cpu < nr_cpu_ids && cpu_possible(cpu))
		return per_cpu(cpu_sys_devices, cpu);
	else
		return NULL;
}
EXPORT_SYMBOL_GPL(get_cpu_device);

static struct attribute *cpu_root_attrs[] = {
#ifdef CONFIG_ARCH_CPU_PROBE_RELEASE
	&dev_attr_probe.attr,
	&dev_attr_release.attr,
#endif
	&cpu_attrs[0].attr.attr,
	&cpu_attrs[1].attr.attr,
	&cpu_attrs[2].attr.attr,
	&dev_attr_kernel_max.attr,
	&dev_attr_offline.attr,
	NULL
};

static struct attribute_group cpu_root_attr_group = {
	.attrs = cpu_root_attrs,
};

static const struct attribute_group *cpu_root_attr_groups[] = {
	&cpu_root_attr_group,
	NULL,
};

bool cpu_is_hotpluggable(unsigned cpu)
{
	struct device *dev = get_cpu_device(cpu);
	return dev && container_of(dev, struct cpu, dev)->hotpluggable;
}
EXPORT_SYMBOL_GPL(cpu_is_hotpluggable);

#ifdef CONFIG_GENERIC_CPU_DEVICES
static DEFINE_PER_CPU(struct cpu, cpu_devices);
#endif

static void __init cpu_dev_register_generic(void)
{
#ifdef CONFIG_GENERIC_CPU_DEVICES
	int i;

	for_each_possible_cpu(i) {
		if (register_cpu(&per_cpu(cpu_devices, i), i))
			panic("Failed to register CPU device");
	}
#endif
}

void __init cpu_dev_init(void)
{
	if (subsys_system_register(&cpu_subsys, cpu_root_attr_groups))
		panic("Failed to register CPU subsystem");

	cpu_dev_register_generic();

#if defined(CONFIG_SCHED_MC) || defined(CONFIG_SCHED_SMT)
	sched_create_sysfs_power_savings_entries(cpu_subsys.dev_root);
#endif
}
Commit	Line	Data
	1	/*
	2	* CPU subsystem support
	3	*/
	4
	5	#include <linux/kernel.h>
	6	#include <linux/module.h>
	7	#include <linux/init.h>
	8	#include <linux/sched.h>
	9	#include <linux/cpu.h>
	10	#include <linux/topology.h>
	11	#include <linux/device.h>
	12	#include <linux/node.h>
	13	#include <linux/gfp.h>
	14	#include <linux/percpu.h>
	15
	16	#include "base.h"
	17
	18	struct bus_type cpu_subsys = {
	19	.name = "cpu",
	20	.dev_name = "cpu",
	21	};
	22	EXPORT_SYMBOL_GPL(cpu_subsys);
	23
	24	static DEFINE_PER_CPU(struct device *, cpu_sys_devices);
	25
	26	#ifdef CONFIG_HOTPLUG_CPU
	27	static ssize_t show_online(struct device *dev,
	28	struct device_attribute *attr,
	29	char *buf)
	30	{
	31	struct cpu *cpu = container_of(dev, struct cpu, dev);
	32
	33	return sprintf(buf, "%u\n", !!cpu_online(cpu->dev.id));
	34	}
	35
	36	static ssize_t __ref store_online(struct device *dev,
	37	struct device_attribute *attr,
	38	const char *buf, size_t count)
	39	{
	40	struct cpu *cpu = container_of(dev, struct cpu, dev);
	41	ssize_t ret;
	42
	43	cpu_hotplug_driver_lock();
	44	switch (buf[0]) {
	45	case '0':
	46	ret = cpu_down(cpu->dev.id);
	47	if (!ret)
	48	kobject_uevent(&dev->kobj, KOBJ_OFFLINE);
	49	break;
	50	case '1':
	51	ret = cpu_up(cpu->dev.id);
	52	if (!ret)
	53	kobject_uevent(&dev->kobj, KOBJ_ONLINE);
	54	break;
	55	default:
	56	ret = -EINVAL;
	57	}
	58	cpu_hotplug_driver_unlock();
	59
	60	if (ret >= 0)
	61	ret = count;
	62	return ret;
	63	}
	64	static DEVICE_ATTR(online, 0644, show_online, store_online);
	65
	66	static void __cpuinit register_cpu_control(struct cpu *cpu)
	67	{
	68	device_create_file(&cpu->dev, &dev_attr_online);
	69	}
	70	void unregister_cpu(struct cpu *cpu)
	71	{
	72	int logical_cpu = cpu->dev.id;
	73
	74	unregister_cpu_under_node(logical_cpu, cpu_to_node(logical_cpu));
	75
	76	device_remove_file(&cpu->dev, &dev_attr_online);
	77
	78	device_unregister(&cpu->dev);
	79	per_cpu(cpu_sys_devices, logical_cpu) = NULL;
	80	return;
	81	}
	82
	83	#ifdef CONFIG_ARCH_CPU_PROBE_RELEASE
	84	static ssize_t cpu_probe_store(struct device *dev,
	85	struct device_attribute *attr,
	86	const char *buf,
	87	size_t count)
	88	{
	89	return arch_cpu_probe(buf, count);
	90	}
	91
	92	static ssize_t cpu_release_store(struct device *dev,
	93	struct device_attribute *attr,
	94	const char *buf,
	95	size_t count)
	96	{
	97	return arch_cpu_release(buf, count);
	98	}
	99
	100	static DEVICE_ATTR(probe, S_IWUSR, NULL, cpu_probe_store);
	101	static DEVICE_ATTR(release, S_IWUSR, NULL, cpu_release_store);
	102	#endif /* CONFIG_ARCH_CPU_PROBE_RELEASE */
	103
	104	#else /* ... !CONFIG_HOTPLUG_CPU */
	105	static inline void register_cpu_control(struct cpu *cpu)
	106	{
	107	}
	108	#endif /* CONFIG_HOTPLUG_CPU */
	109
	110	#ifdef CONFIG_KEXEC
	111	#include <linux/kexec.h>
	112
	113	static ssize_t show_crash_notes(struct device dev, struct device_attribute attr,
	114	char *buf)
	115	{
	116	struct cpu *cpu = container_of(dev, struct cpu, dev);
	117	ssize_t rc;
	118	unsigned long long addr;
	119	int cpunum;
	120
	121	cpunum = cpu->dev.id;
	122
	123	/*
	124	* Might be reading other cpu's data based on which cpu read thread
	125	* has been scheduled. But cpu data (memory) is allocated once during
	126	* boot up and this data does not change there after. Hence this
	127	* operation should be safe. No locking required.
	128	*/
	129	addr = per_cpu_ptr_to_phys(per_cpu_ptr(crash_notes, cpunum));
	130	rc = sprintf(buf, "%Lx\n", addr);
	131	return rc;
	132	}
	133	static DEVICE_ATTR(crash_notes, 0400, show_crash_notes, NULL);
	134	#endif
	135
	136	/*
	137	* Print cpu online, possible, present, and system maps
	138	*/
	139
	140	struct cpu_attr {
	141	struct device_attribute attr;
	142	const struct cpumask const const map;
	143	};
	144
	145	static ssize_t show_cpus_attr(struct device *dev,
	146	struct device_attribute *attr,
	147	char *buf)
	148	{
	149	struct cpu_attr *ca = container_of(attr, struct cpu_attr, attr);
	150	int n = cpulist_scnprintf(buf, PAGE_SIZE-2, *(ca->map));
	151
	152	buf[n++] = '\n';
	153	buf[n] = '\0';
	154	return n;
	155	}
	156
	157	#define _CPU_ATTR(name, map) \
	158	{ __ATTR(name, 0444, show_cpus_attr, NULL), map }
	159
	160	/* Keep in sync with cpu_subsys_attrs */
	161	static struct cpu_attr cpu_attrs[] = {
	162	_CPU_ATTR(online, &cpu_online_mask),
	163	_CPU_ATTR(possible, &cpu_possible_mask),
	164	_CPU_ATTR(present, &cpu_present_mask),
	165	};
	166
	167	/*
	168	* Print values for NR_CPUS and offlined cpus
	169	*/
	170	static ssize_t print_cpus_kernel_max(struct device *dev,
	171	struct device_attribute attr, char buf)
	172	{
	173	int n = snprintf(buf, PAGE_SIZE-2, "%d\n", NR_CPUS - 1);
	174	return n;
	175	}
	176	static DEVICE_ATTR(kernel_max, 0444, print_cpus_kernel_max, NULL);
	177
	178	/* arch-optional setting to enable display of offline cpus >= nr_cpu_ids */
	179	unsigned int total_cpus;
	180
	181	static ssize_t print_cpus_offline(struct device *dev,
	182	struct device_attribute attr, char buf)
	183	{
	184	int n = 0, len = PAGE_SIZE-2;
	185	cpumask_var_t offline;
	186
	187	/* display offline cpus < nr_cpu_ids */
	188	if (!alloc_cpumask_var(&offline, GFP_KERNEL))
	189	return -ENOMEM;
	190	cpumask_andnot(offline, cpu_possible_mask, cpu_online_mask);
	191	n = cpulist_scnprintf(buf, len, offline);
	192	free_cpumask_var(offline);
	193
	194	/* display offline cpus >= nr_cpu_ids */
	195	if (total_cpus && nr_cpu_ids < total_cpus) {
	196	if (n && n < len)
	197	buf[n++] = ',';
	198
	199	if (nr_cpu_ids == total_cpus-1)
	200	n += snprintf(&buf[n], len - n, "%d", nr_cpu_ids);
	201	else
	202	n += snprintf(&buf[n], len - n, "%d-%d",
	203	nr_cpu_ids, total_cpus-1);
	204	}
	205
	206	n += snprintf(&buf[n], len - n, "\n");
	207	return n;
	208	}
	209	static DEVICE_ATTR(offline, 0444, print_cpus_offline, NULL);
	210
	211	static void cpu_device_release(struct device *dev)
	212	{
	213	/*
	214	* This is an empty function to prevent the driver core from spitting a
	215	* warning at us. Yes, I know this is directly opposite of what the
	216	* documentation for the driver core and kobjects say, and the author
	217	* of this code has already been publically ridiculed for doing
	218	* something as foolish as this. However, at this point in time, it is
	219	* the only way to handle the issue of statically allocated cpu
	220	* devices. The different architectures will have their cpu device
	221	* code reworked to properly handle this in the near future, so this
	222	* function will then be changed to correctly free up the memory held
	223	* by the cpu device.
	224	*
	225	* Never copy this way of doing things, or you too will be made fun of
	226	* on the linux-kerenl list, you have been warned.
	227	*/
	228	}
	229
	230	/*
	231	* register_cpu - Setup a sysfs device for a CPU.
	232	* @cpu - cpu->hotpluggable field set to 1 will generate a control file in
	233	* sysfs for this CPU.
	234	* @num - CPU number to use when creating the device.
	235	*
	236	* Initialize and register the CPU device.
	237	*/
	238	int __cpuinit register_cpu(struct cpu *cpu, int num)
	239	{
	240	int error;
	241
	242	cpu->node_id = cpu_to_node(num);
	243	cpu->dev.id = num;
	244	cpu->dev.bus = &cpu_subsys;
	245	cpu->dev.release = cpu_device_release;
	246	error = device_register(&cpu->dev);
	247	if (!error && cpu->hotpluggable)
	248	register_cpu_control(cpu);
	249	if (!error)
	250	per_cpu(cpu_sys_devices, num) = &cpu->dev;
	251	if (!error)
	252	register_cpu_under_node(num, cpu_to_node(num));
	253
	254	#ifdef CONFIG_KEXEC
	255	if (!error)
	256	error = device_create_file(&cpu->dev, &dev_attr_crash_notes);
	257	#endif
	258	return error;
	259	}
	260
	261	struct device *get_cpu_device(unsigned cpu)
	262	{
	263	if (cpu < nr_cpu_ids && cpu_possible(cpu))
	264	return per_cpu(cpu_sys_devices, cpu);
	265	else
	266	return NULL;
	267	}
	268	EXPORT_SYMBOL_GPL(get_cpu_device);
	269
	270	static struct attribute *cpu_root_attrs[] = {
	271	#ifdef CONFIG_ARCH_CPU_PROBE_RELEASE
	272	&dev_attr_probe.attr,
	273	&dev_attr_release.attr,
	274	#endif
	275	&cpu_attrs[0].attr.attr,
	276	&cpu_attrs[1].attr.attr,
	277	&cpu_attrs[2].attr.attr,
	278	&dev_attr_kernel_max.attr,
	279	&dev_attr_offline.attr,
	280	NULL
	281	};
	282
	283	static struct attribute_group cpu_root_attr_group = {
	284	.attrs = cpu_root_attrs,
	285	};
	286
	287	static const struct attribute_group *cpu_root_attr_groups[] = {
	288	&cpu_root_attr_group,
	289	NULL,
	290	};
	291
	292	bool cpu_is_hotpluggable(unsigned cpu)
	293	{
	294	struct device *dev = get_cpu_device(cpu);
	295	return dev && container_of(dev, struct cpu, dev)->hotpluggable;
	296	}
	297	EXPORT_SYMBOL_GPL(cpu_is_hotpluggable);
	298
	299	#ifdef CONFIG_GENERIC_CPU_DEVICES
	300	static DEFINE_PER_CPU(struct cpu, cpu_devices);
	301	#endif
	302
	303	static void __init cpu_dev_register_generic(void)
	304	{
	305	#ifdef CONFIG_GENERIC_CPU_DEVICES
	306	int i;
	307
	308	for_each_possible_cpu(i) {
	309	if (register_cpu(&per_cpu(cpu_devices, i), i))
	310	panic("Failed to register CPU device");
	311	}
	312	#endif
	313	}
	314
	315	void __init cpu_dev_init(void)
	316	{
	317	if (subsys_system_register(&cpu_subsys, cpu_root_attr_groups))
	318	panic("Failed to register CPU subsystem");
	319
	320	cpu_dev_register_generic();
	321
	322	#if defined(CONFIG_SCHED_MC) \|\| defined(CONFIG_SCHED_SMT)
	323	sched_create_sysfs_power_savings_entries(cpu_subsys.dev_root);
	324	#endif
	325	}