[net-next-2.6.git] / drivers / base / memory.c

/*
 * drivers/base/memory.c - basic Memory class support
 *
 * Written by Matt Tolentino <matthew.e.tolentino@intel.com>
 *            Dave Hansen <haveblue@us.ibm.com>
 *
 * This file provides the necessary infrastructure to represent
 * a SPARSEMEM-memory-model system's physical memory in /sysfs.
 * All arch-independent code that assumes MEMORY_HOTPLUG requires
 * SPARSEMEM should be contained here, or in mm/memory_hotplug.c.
 */

#include <linux/sysdev.h>
#include <linux/module.h>
#include <linux/init.h>
#include <linux/topology.h>
#include <linux/capability.h>
#include <linux/device.h>
#include <linux/memory.h>
#include <linux/kobject.h>
#include <linux/memory_hotplug.h>
#include <linux/mm.h>
#include <linux/mutex.h>
#include <linux/stat.h>
#include <linux/slab.h>

#include <asm/atomic.h>
#include <asm/uaccess.h>

#define MEMORY_CLASS_NAME	"memory"

static struct sysdev_class memory_sysdev_class = {
	.name = MEMORY_CLASS_NAME,
};

static const char *memory_uevent_name(struct kset *kset, struct kobject *kobj)
{
	return MEMORY_CLASS_NAME;
}

static int memory_uevent(struct kset *kset, struct kobject *obj, struct kobj_uevent_env *env)
{
	int retval = 0;

	return retval;
}

static const struct kset_uevent_ops memory_uevent_ops = {
	.name		= memory_uevent_name,
	.uevent		= memory_uevent,
};

static BLOCKING_NOTIFIER_HEAD(memory_chain);

int register_memory_notifier(struct notifier_block *nb)
{
        return blocking_notifier_chain_register(&memory_chain, nb);
}
EXPORT_SYMBOL(register_memory_notifier);

void unregister_memory_notifier(struct notifier_block *nb)
{
        blocking_notifier_chain_unregister(&memory_chain, nb);
}
EXPORT_SYMBOL(unregister_memory_notifier);

static ATOMIC_NOTIFIER_HEAD(memory_isolate_chain);

int register_memory_isolate_notifier(struct notifier_block *nb)
{
	return atomic_notifier_chain_register(&memory_isolate_chain, nb);
}
EXPORT_SYMBOL(register_memory_isolate_notifier);

void unregister_memory_isolate_notifier(struct notifier_block *nb)
{
	atomic_notifier_chain_unregister(&memory_isolate_chain, nb);
}
EXPORT_SYMBOL(unregister_memory_isolate_notifier);

/*
 * register_memory - Setup a sysfs device for a memory block
 */
static
int register_memory(struct memory_block *memory, struct mem_section *section)
{
	int error;

	memory->sysdev.cls = &memory_sysdev_class;
	memory->sysdev.id = __section_nr(section);

	error = sysdev_register(&memory->sysdev);
	return error;
}

static void
unregister_memory(struct memory_block *memory, struct mem_section *section)
{
	BUG_ON(memory->sysdev.cls != &memory_sysdev_class);
	BUG_ON(memory->sysdev.id != __section_nr(section));

	/* drop the ref. we got in remove_memory_block() */
	kobject_put(&memory->sysdev.kobj);
	sysdev_unregister(&memory->sysdev);
}

/*
 * use this as the physical section index that this memsection
 * uses.
 */

static ssize_t show_mem_phys_index(struct sys_device *dev,
			struct sysdev_attribute *attr, char *buf)
{
	struct memory_block *mem =
		container_of(dev, struct memory_block, sysdev);
	return sprintf(buf, "%08lx\n", mem->phys_index);
}

/*
 * Show whether the section of memory is likely to be hot-removable
 */
static ssize_t show_mem_removable(struct sys_device *dev,
			struct sysdev_attribute *attr, char *buf)
{
	unsigned long start_pfn;
	int ret;
	struct memory_block *mem =
		container_of(dev, struct memory_block, sysdev);

	start_pfn = section_nr_to_pfn(mem->phys_index);
	ret = is_mem_section_removable(start_pfn, PAGES_PER_SECTION);
	return sprintf(buf, "%d\n", ret);
}

/*
 * online, offline, going offline, etc.
 */
static ssize_t show_mem_state(struct sys_device *dev,
			struct sysdev_attribute *attr, char *buf)
{
	struct memory_block *mem =
		container_of(dev, struct memory_block, sysdev);
	ssize_t len = 0;

	/*
	 * We can probably put these states in a nice little array
	 * so that they're not open-coded
	 */
	switch (mem->state) {
		case MEM_ONLINE:
			len = sprintf(buf, "online\n");
			break;
		case MEM_OFFLINE:
			len = sprintf(buf, "offline\n");
			break;
		case MEM_GOING_OFFLINE:
			len = sprintf(buf, "going-offline\n");
			break;
		default:
			len = sprintf(buf, "ERROR-UNKNOWN-%ld\n",
					mem->state);
			WARN_ON(1);
			break;
	}

	return len;
}

int memory_notify(unsigned long val, void *v)
{
	return blocking_notifier_call_chain(&memory_chain, val, v);
}

int memory_isolate_notify(unsigned long val, void *v)
{
	return atomic_notifier_call_chain(&memory_isolate_chain, val, v);
}

/*
 * MEMORY_HOTPLUG depends on SPARSEMEM in mm/Kconfig, so it is
 * OK to have direct references to sparsemem variables in here.
 */
static int
memory_block_action(struct memory_block *mem, unsigned long action)
{
	int i;
	unsigned long psection;
	unsigned long start_pfn, start_paddr;
	struct page *first_page;
	int ret;
	int old_state = mem->state;

	psection = mem->phys_index;
	first_page = pfn_to_page(psection << PFN_SECTION_SHIFT);

	/*
	 * The probe routines leave the pages reserved, just
	 * as the bootmem code does.  Make sure they're still
	 * that way.
	 */
	if (action == MEM_ONLINE) {
		for (i = 0; i < PAGES_PER_SECTION; i++) {
			if (PageReserved(first_page+i))
				continue;

			printk(KERN_WARNING "section number %ld page number %d "
				"not reserved, was it already online? \n",
				psection, i);
			return -EBUSY;
		}
	}

	switch (action) {
		case MEM_ONLINE:
			start_pfn = page_to_pfn(first_page);
			ret = online_pages(start_pfn, PAGES_PER_SECTION);
			break;
		case MEM_OFFLINE:
			mem->state = MEM_GOING_OFFLINE;
			start_paddr = page_to_pfn(first_page) << PAGE_SHIFT;
			ret = remove_memory(start_paddr,
					    PAGES_PER_SECTION << PAGE_SHIFT);
			if (ret) {
				mem->state = old_state;
				break;
			}
			break;
		default:
			WARN(1, KERN_WARNING "%s(%p, %ld) unknown action: %ld\n",
					__func__, mem, action, action);
			ret = -EINVAL;
	}

	return ret;
}

static int memory_block_change_state(struct memory_block *mem,
		unsigned long to_state, unsigned long from_state_req)
{
	int ret = 0;
	mutex_lock(&mem->state_mutex);

	if (mem->state != from_state_req) {
		ret = -EINVAL;
		goto out;
	}

	ret = memory_block_action(mem, to_state);
	if (!ret)
		mem->state = to_state;

out:
	mutex_unlock(&mem->state_mutex);
	return ret;
}

static ssize_t
store_mem_state(struct sys_device *dev,
		struct sysdev_attribute *attr, const char *buf, size_t count)
{
	struct memory_block *mem;
	unsigned int phys_section_nr;
	int ret = -EINVAL;

	mem = container_of(dev, struct memory_block, sysdev);
	phys_section_nr = mem->phys_index;

	if (!present_section_nr(phys_section_nr))
		goto out;

	if (!strncmp(buf, "online", min((int)count, 6)))
		ret = memory_block_change_state(mem, MEM_ONLINE, MEM_OFFLINE);
	else if(!strncmp(buf, "offline", min((int)count, 7)))
		ret = memory_block_change_state(mem, MEM_OFFLINE, MEM_ONLINE);
out:
	if (ret)
		return ret;
	return count;
}

/*
 * phys_device is a bad name for this.  What I really want
 * is a way to differentiate between memory ranges that
 * are part of physical devices that constitute
 * a complete removable unit or fru.
 * i.e. do these ranges belong to the same physical device,
 * s.t. if I offline all of these sections I can then
 * remove the physical device?
 */
static ssize_t show_phys_device(struct sys_device *dev,
				struct sysdev_attribute *attr, char *buf)
{
	struct memory_block *mem =
		container_of(dev, struct memory_block, sysdev);
	return sprintf(buf, "%d\n", mem->phys_device);
}

static SYSDEV_ATTR(phys_index, 0444, show_mem_phys_index, NULL);
static SYSDEV_ATTR(state, 0644, show_mem_state, store_mem_state);
static SYSDEV_ATTR(phys_device, 0444, show_phys_device, NULL);
static SYSDEV_ATTR(removable, 0444, show_mem_removable, NULL);

#define mem_create_simple_file(mem, attr_name)	\
	sysdev_create_file(&mem->sysdev, &attr_##attr_name)
#define mem_remove_simple_file(mem, attr_name)	\
	sysdev_remove_file(&mem->sysdev, &attr_##attr_name)

/*
 * Block size attribute stuff
 */
static ssize_t
print_block_size(struct sysdev_class *class, struct sysdev_class_attribute *attr,
		 char *buf)
{
	return sprintf(buf, "%#lx\n", (unsigned long)PAGES_PER_SECTION * PAGE_SIZE);
}

static SYSDEV_CLASS_ATTR(block_size_bytes, 0444, print_block_size, NULL);

static int block_size_init(void)
{
	return sysfs_create_file(&memory_sysdev_class.kset.kobj,
				&attr_block_size_bytes.attr);
}

/*
 * Some architectures will have custom drivers to do this, and
 * will not need to do it from userspace.  The fake hot-add code
 * as well as ppc64 will do all of their discovery in userspace
 * and will require this interface.
 */
#ifdef CONFIG_ARCH_MEMORY_PROBE
static ssize_t
memory_probe_store(struct class *class, struct class_attribute *attr,
		   const char *buf, size_t count)
{
	u64 phys_addr;
	int nid;
	int ret;

	phys_addr = simple_strtoull(buf, NULL, 0);

	nid = memory_add_physaddr_to_nid(phys_addr);
	ret = add_memory(nid, phys_addr, PAGES_PER_SECTION << PAGE_SHIFT);

	if (ret)
		count = ret;

	return count;
}
static CLASS_ATTR(probe, S_IWUSR, NULL, memory_probe_store);

static int memory_probe_init(void)
{
	return sysfs_create_file(&memory_sysdev_class.kset.kobj,
				&class_attr_probe.attr);
}
#else
static inline int memory_probe_init(void)
{
	return 0;
}
#endif

#ifdef CONFIG_MEMORY_FAILURE
/*
 * Support for offlining pages of memory
 */

/* Soft offline a page */
static ssize_t
store_soft_offline_page(struct class *class,
			struct class_attribute *attr,
			const char *buf, size_t count)
{
	int ret;
	u64 pfn;
	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;
	if (strict_strtoull(buf, 0, &pfn) < 0)
		return -EINVAL;
	pfn >>= PAGE_SHIFT;
	if (!pfn_valid(pfn))
		return -ENXIO;
	ret = soft_offline_page(pfn_to_page(pfn), 0);
	return ret == 0 ? count : ret;
}

/* Forcibly offline a page, including killing processes. */
static ssize_t
store_hard_offline_page(struct class *class,
			struct class_attribute *attr,
			const char *buf, size_t count)
{
	int ret;
	u64 pfn;
	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;
	if (strict_strtoull(buf, 0, &pfn) < 0)
		return -EINVAL;
	pfn >>= PAGE_SHIFT;
	ret = __memory_failure(pfn, 0, 0);
	return ret ? ret : count;
}

static CLASS_ATTR(soft_offline_page, 0644, NULL, store_soft_offline_page);
static CLASS_ATTR(hard_offline_page, 0644, NULL, store_hard_offline_page);

static __init int memory_fail_init(void)
{
	int err;

	err = sysfs_create_file(&memory_sysdev_class.kset.kobj,
				&class_attr_soft_offline_page.attr);
	if (!err)
		err = sysfs_create_file(&memory_sysdev_class.kset.kobj,
				&class_attr_hard_offline_page.attr);
	return err;
}
#else
static inline int memory_fail_init(void)
{
	return 0;
}
#endif

/*
 * Note that phys_device is optional.  It is here to allow for
 * differentiation between which *physical* devices each
 * section belongs to...
 */
int __weak arch_get_memory_phys_device(unsigned long start_pfn)
{
	return 0;
}

static int add_memory_block(int nid, struct mem_section *section,
			unsigned long state, enum mem_add_context context)
{
	struct memory_block *mem = kzalloc(sizeof(*mem), GFP_KERNEL);
	unsigned long start_pfn;
	int ret = 0;

	if (!mem)
		return -ENOMEM;

	mem->phys_index = __section_nr(section);
	mem->state = state;
	mutex_init(&mem->state_mutex);
	start_pfn = section_nr_to_pfn(mem->phys_index);
	mem->phys_device = arch_get_memory_phys_device(start_pfn);

	ret = register_memory(mem, section);
	if (!ret)
		ret = mem_create_simple_file(mem, phys_index);
	if (!ret)
		ret = mem_create_simple_file(mem, state);
	if (!ret)
		ret = mem_create_simple_file(mem, phys_device);
	if (!ret)
		ret = mem_create_simple_file(mem, removable);
	if (!ret) {
		if (context == HOTPLUG)
			ret = register_mem_sect_under_node(mem, nid);
	}

	return ret;
}

/*
 * For now, we have a linear search to go find the appropriate
 * memory_block corresponding to a particular phys_index. If
 * this gets to be a real problem, we can always use a radix
 * tree or something here.
 *
 * This could be made generic for all sysdev classes.
 */
struct memory_block *find_memory_block(struct mem_section *section)
{
	struct kobject *kobj;
	struct sys_device *sysdev;
	struct memory_block *mem;
	char name[sizeof(MEMORY_CLASS_NAME) + 9 + 1];

	/*
	 * This only works because we know that section == sysdev->id
	 * slightly redundant with sysdev_register()
	 */
	sprintf(&name[0], "%s%d", MEMORY_CLASS_NAME, __section_nr(section));

	kobj = kset_find_obj(&memory_sysdev_class.kset, name);
	if (!kobj)
		return NULL;

	sysdev = container_of(kobj, struct sys_device, kobj);
	mem = container_of(sysdev, struct memory_block, sysdev);

	return mem;
}

int remove_memory_block(unsigned long node_id, struct mem_section *section,
		int phys_device)
{
	struct memory_block *mem;

	mem = find_memory_block(section);
	unregister_mem_sect_under_nodes(mem);
	mem_remove_simple_file(mem, phys_index);
	mem_remove_simple_file(mem, state);
	mem_remove_simple_file(mem, phys_device);
	mem_remove_simple_file(mem, removable);
	unregister_memory(mem, section);

	return 0;
}

/*
 * need an interface for the VM to add new memory regions,
 * but without onlining it.
 */
int register_new_memory(int nid, struct mem_section *section)
{
	return add_memory_block(nid, section, MEM_OFFLINE, HOTPLUG);
}

int unregister_memory_section(struct mem_section *section)
{
	if (!present_section(section))
		return -EINVAL;

	return remove_memory_block(0, section, 0);
}

/*
 * Initialize the sysfs support for memory devices...
 */
int __init memory_dev_init(void)
{
	unsigned int i;
	int ret;
	int err;

	memory_sysdev_class.kset.uevent_ops = &memory_uevent_ops;
	ret = sysdev_class_register(&memory_sysdev_class);
	if (ret)
		goto out;

	/*
	 * Create entries for memory sections that were found
	 * during boot and have been initialized
	 */
	for (i = 0; i < NR_MEM_SECTIONS; i++) {
		if (!present_section_nr(i))
			continue;
		err = add_memory_block(0, __nr_to_section(i), MEM_ONLINE,
				       BOOT);
		if (!ret)
			ret = err;
	}

	err = memory_probe_init();
	if (!ret)
		ret = err;
	err = memory_fail_init();
	if (!ret)
		ret = err;
	err = block_size_init();
	if (!ret)
		ret = err;
out:
	if (ret)
		printk(KERN_ERR "%s() failed: %d\n", __func__, ret);
	return ret;
}
Commit	Line	Data
3947be19 DH	1	/*
	2	* drivers/base/memory.c - basic Memory class support
	3	*
	4	* Written by Matt Tolentino <matthew.e.tolentino@intel.com>
	5	* Dave Hansen <haveblue@us.ibm.com>
	6	*
	7	* This file provides the necessary infrastructure to represent
	8	* a SPARSEMEM-memory-model system's physical memory in /sysfs.
	9	* All arch-independent code that assumes MEMORY_HOTPLUG requires
	10	* SPARSEMEM should be contained here, or in mm/memory_hotplug.c.
	11	*/
	12
	13	#include <linux/sysdev.h>
	14	#include <linux/module.h>
	15	#include <linux/init.h>
3947be19	16	#include <linux/topology.h>
c59ede7b	17	#include <linux/capability.h>
3947be19 DH	18	#include <linux/device.h>
	19	#include <linux/memory.h>
	20	#include <linux/kobject.h>
	21	#include <linux/memory_hotplug.h>
	22	#include <linux/mm.h>
da19cbcf	23	#include <linux/mutex.h>
9f1b16a5	24	#include <linux/stat.h>
5a0e3ad6	25	#include <linux/slab.h>
9f1b16a5	26
3947be19 DH	27	#include <asm/atomic.h>
	28	#include <asm/uaccess.h>
	29
	30	#define MEMORY_CLASS_NAME "memory"
	31
	32	static struct sysdev_class memory_sysdev_class = {
af5ca3f4	33	.name = MEMORY_CLASS_NAME,
3947be19	34	};
3947be19	35
312c004d	36	static const char memory_uevent_name(struct kset kset, struct kobject *kobj)
3947be19 DH	37	{
	38	return MEMORY_CLASS_NAME;
	39	}
	40
9ec0fd4e	41	static int memory_uevent(struct kset kset, struct kobject obj, struct kobj_uevent_env *env)
3947be19 DH	42	{
	43	int retval = 0;
	44
	45	return retval;
	46	}
	47
9cd43611	48	static const struct kset_uevent_ops memory_uevent_ops = {
312c004d KS	49	.name = memory_uevent_name,
312c004d KS	50	.uevent = memory_uevent,
3947be19 DH	51	};
3947be19 DH	52
e041c683	53	static BLOCKING_NOTIFIER_HEAD(memory_chain);
3947be19	54
98a38ebd	55	int register_memory_notifier(struct notifier_block *nb)
3947be19	56	{
e041c683	57	return blocking_notifier_chain_register(&memory_chain, nb);
3947be19	58	}
3c82c30c	59	EXPORT_SYMBOL(register_memory_notifier);
3947be19	60
98a38ebd	61	void unregister_memory_notifier(struct notifier_block *nb)
3947be19	62	{
e041c683	63	blocking_notifier_chain_unregister(&memory_chain, nb);
3947be19	64	}
3c82c30c	65	EXPORT_SYMBOL(unregister_memory_notifier);
3947be19	66
925cc71e RJ	67	static ATOMIC_NOTIFIER_HEAD(memory_isolate_chain);
	68
	69	int register_memory_isolate_notifier(struct notifier_block *nb)
	70	{
	71	return atomic_notifier_chain_register(&memory_isolate_chain, nb);
	72	}
	73	EXPORT_SYMBOL(register_memory_isolate_notifier);
	74
	75	void unregister_memory_isolate_notifier(struct notifier_block *nb)
	76	{
	77	atomic_notifier_chain_unregister(&memory_isolate_chain, nb);
	78	}
	79	EXPORT_SYMBOL(unregister_memory_isolate_notifier);
	80
3947be19 DH	81	/*
	82	* register_memory - Setup a sysfs device for a memory block
	83	*/
00a41db5 BP	84	static
00a41db5 BP	85	int register_memory(struct memory_block memory, struct mem_section section)
3947be19 DH	86	{
	87	int error;
	88
	89	memory->sysdev.cls = &memory_sysdev_class;
	90	memory->sysdev.id = __section_nr(section);
	91
	92	error = sysdev_register(&memory->sysdev);
3947be19 DH	93	return error;
	94	}
	95
	96	static void
00a41db5	97	unregister_memory(struct memory_block memory, struct mem_section section)
3947be19 DH	98	{
	99	BUG_ON(memory->sysdev.cls != &memory_sysdev_class);
	100	BUG_ON(memory->sysdev.id != __section_nr(section));
	101
00a41db5 BP	102	/* drop the ref. we got in remove_memory_block() */
00a41db5 BP	103	kobject_put(&memory->sysdev.kobj);
3947be19	104	sysdev_unregister(&memory->sysdev);
3947be19 DH	105	}
	106
	107	/*
	108	* use this as the physical section index that this memsection
	109	* uses.
	110	*/
	111
4a0b2b4d AK	112	static ssize_t show_mem_phys_index(struct sys_device *dev,
4a0b2b4d AK	113	struct sysdev_attribute attr, char buf)
3947be19 DH	114	{
	115	struct memory_block *mem =
	116	container_of(dev, struct memory_block, sysdev);
	117	return sprintf(buf, "%08lx\n", mem->phys_index);
	118	}
	119
5c755e9f BP	120	/*
	121	* Show whether the section of memory is likely to be hot-removable
	122	*/
1f07be1c SR	123	static ssize_t show_mem_removable(struct sys_device *dev,
1f07be1c SR	124	struct sysdev_attribute attr, char buf)
5c755e9f BP	125	{
	126	unsigned long start_pfn;
	127	int ret;
	128	struct memory_block *mem =
	129	container_of(dev, struct memory_block, sysdev);
	130
	131	start_pfn = section_nr_to_pfn(mem->phys_index);
	132	ret = is_mem_section_removable(start_pfn, PAGES_PER_SECTION);
	133	return sprintf(buf, "%d\n", ret);
	134	}
	135
3947be19 DH	136	/*
	137	* online, offline, going offline, etc.
	138	*/
4a0b2b4d AK	139	static ssize_t show_mem_state(struct sys_device *dev,
4a0b2b4d AK	140	struct sysdev_attribute attr, char buf)
3947be19 DH	141	{
	142	struct memory_block *mem =
	143	container_of(dev, struct memory_block, sysdev);
	144	ssize_t len = 0;
	145
	146	/*
	147	* We can probably put these states in a nice little array
	148	* so that they're not open-coded
	149	*/
	150	switch (mem->state) {
	151	case MEM_ONLINE:
	152	len = sprintf(buf, "online\n");
	153	break;
	154	case MEM_OFFLINE:
	155	len = sprintf(buf, "offline\n");
	156	break;
	157	case MEM_GOING_OFFLINE:
	158	len = sprintf(buf, "going-offline\n");
	159	break;
	160	default:
	161	len = sprintf(buf, "ERROR-UNKNOWN-%ld\n",
	162	mem->state);
	163	WARN_ON(1);
	164	break;
	165	}
	166
	167	return len;
	168	}
	169
7b78d335	170	int memory_notify(unsigned long val, void *v)
3947be19	171	{
e041c683	172	return blocking_notifier_call_chain(&memory_chain, val, v);
3947be19 DH	173	}
3947be19 DH	174
925cc71e RJ	175	int memory_isolate_notify(unsigned long val, void *v)
	176	{
	177	return atomic_notifier_call_chain(&memory_isolate_chain, val, v);
	178	}
	179
3947be19 DH	180	/*
	181	* MEMORY_HOTPLUG depends on SPARSEMEM in mm/Kconfig, so it is
	182	* OK to have direct references to sparsemem variables in here.
	183	*/
	184	static int
	185	memory_block_action(struct memory_block *mem, unsigned long action)
	186	{
	187	int i;
	188	unsigned long psection;
	189	unsigned long start_pfn, start_paddr;
	190	struct page *first_page;
	191	int ret;
	192	int old_state = mem->state;
	193
	194	psection = mem->phys_index;
	195	first_page = pfn_to_page(psection << PFN_SECTION_SHIFT);
	196
	197	/*
	198	* The probe routines leave the pages reserved, just
	199	* as the bootmem code does. Make sure they're still
	200	* that way.
	201	*/
	202	if (action == MEM_ONLINE) {
	203	for (i = 0; i < PAGES_PER_SECTION; i++) {
	204	if (PageReserved(first_page+i))
	205	continue;
	206
	207	printk(KERN_WARNING "section number %ld page number %d "
	208	"not reserved, was it already online? \n",
	209	psection, i);
	210	return -EBUSY;
	211	}
	212	}
	213
	214	switch (action) {
	215	case MEM_ONLINE:
	216	start_pfn = page_to_pfn(first_page);
	217	ret = online_pages(start_pfn, PAGES_PER_SECTION);
	218	break;
	219	case MEM_OFFLINE:
	220	mem->state = MEM_GOING_OFFLINE;
3947be19 DH	221	start_paddr = page_to_pfn(first_page) << PAGE_SHIFT;
	222	ret = remove_memory(start_paddr,
	223	PAGES_PER_SECTION << PAGE_SHIFT);
	224	if (ret) {
	225	mem->state = old_state;
	226	break;
	227	}
3947be19 DH	228	break;
3947be19 DH	229	default:
f810a5cf	230	WARN(1, KERN_WARNING "%s(%p, %ld) unknown action: %ld\n",
2b3a302a	231	__func__, mem, action, action);
3947be19 DH	232	ret = -EINVAL;
3947be19 DH	233	}
3947be19 DH	234
	235	return ret;
	236	}
	237
	238	static int memory_block_change_state(struct memory_block *mem,
	239	unsigned long to_state, unsigned long from_state_req)
	240	{
	241	int ret = 0;
da19cbcf	242	mutex_lock(&mem->state_mutex);
3947be19 DH	243
	244	if (mem->state != from_state_req) {
	245	ret = -EINVAL;
	246	goto out;
	247	}
	248
	249	ret = memory_block_action(mem, to_state);
	250	if (!ret)
	251	mem->state = to_state;
	252
	253	out:
da19cbcf	254	mutex_unlock(&mem->state_mutex);
3947be19 DH	255	return ret;
	256	}
	257
	258	static ssize_t
4a0b2b4d AK	259	store_mem_state(struct sys_device *dev,
4a0b2b4d AK	260	struct sysdev_attribute attr, const char buf, size_t count)
3947be19 DH	261	{
	262	struct memory_block *mem;
	263	unsigned int phys_section_nr;
	264	int ret = -EINVAL;
	265
	266	mem = container_of(dev, struct memory_block, sysdev);
	267	phys_section_nr = mem->phys_index;
	268
540557b9	269	if (!present_section_nr(phys_section_nr))
3947be19 DH	270	goto out;
	271
	272	if (!strncmp(buf, "online", min((int)count, 6)))
	273	ret = memory_block_change_state(mem, MEM_ONLINE, MEM_OFFLINE);
	274	else if(!strncmp(buf, "offline", min((int)count, 7)))
	275	ret = memory_block_change_state(mem, MEM_OFFLINE, MEM_ONLINE);
	276	out:
	277	if (ret)
	278	return ret;
	279	return count;
	280	}
	281
	282	/*
	283	* phys_device is a bad name for this. What I really want
	284	* is a way to differentiate between memory ranges that
	285	* are part of physical devices that constitute
	286	* a complete removable unit or fru.
	287	* i.e. do these ranges belong to the same physical device,
	288	* s.t. if I offline all of these sections I can then
	289	* remove the physical device?
	290	*/
4a0b2b4d AK	291	static ssize_t show_phys_device(struct sys_device *dev,
4a0b2b4d AK	292	struct sysdev_attribute attr, char buf)
3947be19 DH	293	{
	294	struct memory_block *mem =
	295	container_of(dev, struct memory_block, sysdev);
	296	return sprintf(buf, "%d\n", mem->phys_device);
	297	}
	298
	299	static SYSDEV_ATTR(phys_index, 0444, show_mem_phys_index, NULL);
	300	static SYSDEV_ATTR(state, 0644, show_mem_state, store_mem_state);
	301	static SYSDEV_ATTR(phys_device, 0444, show_phys_device, NULL);
5c755e9f	302	static SYSDEV_ATTR(removable, 0444, show_mem_removable, NULL);
3947be19 DH	303
	304	#define mem_create_simple_file(mem, attr_name) \
	305	sysdev_create_file(&mem->sysdev, &attr_##attr_name)
	306	#define mem_remove_simple_file(mem, attr_name) \
	307	sysdev_remove_file(&mem->sysdev, &attr_##attr_name)
	308
	309	/*
	310	* Block size attribute stuff
	311	*/
	312	static ssize_t
8564a6c1 AK	313	print_block_size(struct sysdev_class class, struct sysdev_class_attribute attr,
8564a6c1 AK	314	char *buf)
3947be19	315	{
ba168fc3	316	return sprintf(buf, "%#lx\n", (unsigned long)PAGES_PER_SECTION * PAGE_SIZE);
3947be19 DH	317	}
3947be19 DH	318
8564a6c1	319	static SYSDEV_CLASS_ATTR(block_size_bytes, 0444, print_block_size, NULL);
3947be19 DH	320
	321	static int block_size_init(void)
	322	{
28ec24e2	323	return sysfs_create_file(&memory_sysdev_class.kset.kobj,
8564a6c1	324	&attr_block_size_bytes.attr);
3947be19 DH	325	}
	326
	327	/*
	328	* Some architectures will have custom drivers to do this, and
	329	* will not need to do it from userspace. The fake hot-add code
	330	* as well as ppc64 will do all of their discovery in userspace
	331	* and will require this interface.
	332	*/
	333	#ifdef CONFIG_ARCH_MEMORY_PROBE
	334	static ssize_t
28812fe1 AK	335	memory_probe_store(struct class class, struct class_attribute attr,
28812fe1 AK	336	const char *buf, size_t count)
3947be19 DH	337	{
3947be19 DH	338	u64 phys_addr;
bc02af93	339	int nid;
3947be19 DH	340	int ret;
	341
	342	phys_addr = simple_strtoull(buf, NULL, 0);
	343
bc02af93 YG	344	nid = memory_add_physaddr_to_nid(phys_addr);
bc02af93 YG	345	ret = add_memory(nid, phys_addr, PAGES_PER_SECTION << PAGE_SHIFT);
3947be19 DH	346
	347	if (ret)
	348	count = ret;
	349
	350	return count;
	351	}
bd796671	352	static CLASS_ATTR(probe, S_IWUSR, NULL, memory_probe_store);
3947be19 DH	353
	354	static int memory_probe_init(void)
	355	{
28ec24e2	356	return sysfs_create_file(&memory_sysdev_class.kset.kobj,
bd796671	357	&class_attr_probe.attr);
3947be19 DH	358	}
3947be19 DH	359	#else
28ec24e2 AM	360	static inline int memory_probe_init(void)
	361	{
	362	return 0;
	363	}
3947be19 DH	364	#endif
3947be19 DH	365
facb6011 AK	366	#ifdef CONFIG_MEMORY_FAILURE
	367	/*
	368	* Support for offlining pages of memory
	369	*/
	370
	371	/* Soft offline a page */
	372	static ssize_t
28812fe1 AK	373	store_soft_offline_page(struct class *class,
	374	struct class_attribute *attr,
	375	const char *buf, size_t count)
facb6011 AK	376	{
	377	int ret;
	378	u64 pfn;
	379	if (!capable(CAP_SYS_ADMIN))
	380	return -EPERM;
	381	if (strict_strtoull(buf, 0, &pfn) < 0)
	382	return -EINVAL;
	383	pfn >>= PAGE_SHIFT;
	384	if (!pfn_valid(pfn))
	385	return -ENXIO;
	386	ret = soft_offline_page(pfn_to_page(pfn), 0);
	387	return ret == 0 ? count : ret;
	388	}
	389
	390	/* Forcibly offline a page, including killing processes. */
	391	static ssize_t
28812fe1 AK	392	store_hard_offline_page(struct class *class,
	393	struct class_attribute *attr,
	394	const char *buf, size_t count)
facb6011 AK	395	{
	396	int ret;
	397	u64 pfn;
	398	if (!capable(CAP_SYS_ADMIN))
	399	return -EPERM;
	400	if (strict_strtoull(buf, 0, &pfn) < 0)
	401	return -EINVAL;
	402	pfn >>= PAGE_SHIFT;
	403	ret = __memory_failure(pfn, 0, 0);
	404	return ret ? ret : count;
	405	}
	406
bd796671 GKH	407	static CLASS_ATTR(soft_offline_page, 0644, NULL, store_soft_offline_page);
bd796671 GKH	408	static CLASS_ATTR(hard_offline_page, 0644, NULL, store_hard_offline_page);
facb6011 AK	409
	410	static __init int memory_fail_init(void)
	411	{
	412	int err;
	413
	414	err = sysfs_create_file(&memory_sysdev_class.kset.kobj,
bd796671	415	&class_attr_soft_offline_page.attr);
facb6011 AK	416	if (!err)
facb6011 AK	417	err = sysfs_create_file(&memory_sysdev_class.kset.kobj,
bd796671	418	&class_attr_hard_offline_page.attr);
facb6011 AK	419	return err;
	420	}
	421	#else
	422	static inline int memory_fail_init(void)
	423	{
	424	return 0;
	425	}
	426	#endif
	427
3947be19 DH	428	/*
	429	* Note that phys_device is optional. It is here to allow for
	430	* differentiation between which physical devices each
	431	* section belongs to...
	432	*/
bc32df00 HC	433	int __weak arch_get_memory_phys_device(unsigned long start_pfn)
	434	{
	435	return 0;
	436	}
3947be19	437
c04fc586	438	static int add_memory_block(int nid, struct mem_section *section,
bc32df00	439	unsigned long state, enum mem_add_context context)
3947be19	440	{
0b0acbec	441	struct memory_block mem = kzalloc(sizeof(mem), GFP_KERNEL);
bc32df00	442	unsigned long start_pfn;
3947be19 DH	443	int ret = 0;
	444
	445	if (!mem)
	446	return -ENOMEM;
	447
3947be19 DH	448	mem->phys_index = __section_nr(section);
3947be19 DH	449	mem->state = state;
da19cbcf	450	mutex_init(&mem->state_mutex);
bc32df00 HC	451	start_pfn = section_nr_to_pfn(mem->phys_index);
bc32df00 HC	452	mem->phys_device = arch_get_memory_phys_device(start_pfn);
3947be19	453
00a41db5	454	ret = register_memory(mem, section);
3947be19 DH	455	if (!ret)
	456	ret = mem_create_simple_file(mem, phys_index);
	457	if (!ret)
	458	ret = mem_create_simple_file(mem, state);
	459	if (!ret)
	460	ret = mem_create_simple_file(mem, phys_device);
5c755e9f BP	461	if (!ret)
5c755e9f BP	462	ret = mem_create_simple_file(mem, removable);
c04fc586 GH	463	if (!ret) {
	464	if (context == HOTPLUG)
	465	ret = register_mem_sect_under_node(mem, nid);
	466	}
3947be19 DH	467
	468	return ret;
	469	}
	470
	471	/*
	472	* For now, we have a linear search to go find the appropriate
	473	* memory_block corresponding to a particular phys_index. If
	474	* this gets to be a real problem, we can always use a radix
	475	* tree or something here.
	476	*
	477	* This could be made generic for all sysdev classes.
	478	*/
c04fc586	479	struct memory_block find_memory_block(struct mem_section section)
3947be19 DH	480	{
	481	struct kobject *kobj;
	482	struct sys_device *sysdev;
	483	struct memory_block *mem;
	484	char name[sizeof(MEMORY_CLASS_NAME) + 9 + 1];
	485
	486	/*
	487	* This only works because we know that section == sysdev->id
	488	* slightly redundant with sysdev_register()
	489	*/
	490	sprintf(&name[0], "%s%d", MEMORY_CLASS_NAME, __section_nr(section));
	491
	492	kobj = kset_find_obj(&memory_sysdev_class.kset, name);
	493	if (!kobj)
	494	return NULL;
	495
	496	sysdev = container_of(kobj, struct sys_device, kobj);
	497	mem = container_of(sysdev, struct memory_block, sysdev);
	498
	499	return mem;
	500	}
	501
	502	int remove_memory_block(unsigned long node_id, struct mem_section *section,
	503	int phys_device)
	504	{
	505	struct memory_block *mem;
	506
	507	mem = find_memory_block(section);
c04fc586	508	unregister_mem_sect_under_nodes(mem);
3947be19 DH	509	mem_remove_simple_file(mem, phys_index);
	510	mem_remove_simple_file(mem, state);
	511	mem_remove_simple_file(mem, phys_device);
5c755e9f	512	mem_remove_simple_file(mem, removable);
00a41db5	513	unregister_memory(mem, section);
3947be19 DH	514
	515	return 0;
	516	}
	517
	518	/*
	519	* need an interface for the VM to add new memory regions,
	520	* but without onlining it.
	521	*/
c04fc586	522	int register_new_memory(int nid, struct mem_section *section)
3947be19	523	{
bc32df00	524	return add_memory_block(nid, section, MEM_OFFLINE, HOTPLUG);
3947be19 DH	525	}
	526
	527	int unregister_memory_section(struct mem_section *section)
	528	{
540557b9	529	if (!present_section(section))
3947be19 DH	530	return -EINVAL;
	531
	532	return remove_memory_block(0, section, 0);
	533	}
	534
	535	/*
	536	* Initialize the sysfs support for memory devices...
	537	*/
	538	int __init memory_dev_init(void)
	539	{
	540	unsigned int i;
	541	int ret;
28ec24e2	542	int err;
3947be19	543
312c004d	544	memory_sysdev_class.kset.uevent_ops = &memory_uevent_ops;
3947be19	545	ret = sysdev_class_register(&memory_sysdev_class);
28ec24e2 AM	546	if (ret)
28ec24e2 AM	547	goto out;
3947be19 DH	548
	549	/*
	550	* Create entries for memory sections that were found
	551	* during boot and have been initialized
	552	*/
	553	for (i = 0; i < NR_MEM_SECTIONS; i++) {
540557b9	554	if (!present_section_nr(i))
3947be19	555	continue;
c04fc586	556	err = add_memory_block(0, __nr_to_section(i), MEM_ONLINE,
bc32df00	557	BOOT);
28ec24e2 AM	558	if (!ret)
28ec24e2 AM	559	ret = err;
3947be19 DH	560	}
3947be19 DH	561
28ec24e2	562	err = memory_probe_init();
facb6011 AK	563	if (!ret)
	564	ret = err;
	565	err = memory_fail_init();
28ec24e2 AM	566	if (!ret)
	567	ret = err;
	568	err = block_size_init();
	569	if (!ret)
	570	ret = err;
	571	out:
	572	if (ret)
2b3a302a	573	printk(KERN_ERR "%s() failed: %d\n", __func__, ret);
3947be19 DH	574	return ret;
3947be19 DH	575	}