[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>

static DEFINE_MUTEX(mem_sysfs_mutex);

#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;
}

struct memory_block *find_memory_block_hinted(struct mem_section *section,
					      struct memory_block *hint)
{
	struct kobject *kobj;
	struct sys_device *sysdev;
	struct memory_block *mem;
	char name[sizeof(MEMORY_CLASS_NAME) + 9 + 1];

	kobj = hint ? &hint->sysdev.kobj : NULL;

	/*
	 * 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_hinted(&memory_sysdev_class.kset, name, kobj);
	if (!kobj)
		return NULL;

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

	return mem;
}

/*
 * 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)
{
	return find_memory_block_hinted(section, NULL);
}

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;

	mutex_lock(&mem_sysfs_mutex);

	mem->phys_index = __section_nr(section);
	mem->state = state;
	mem->section_count++;
	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);
	}

	mutex_unlock(&mem_sysfs_mutex);
	return ret;
}

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

	mutex_lock(&mem_sysfs_mutex);
	mem = find_memory_block(section);

	mem->section_count--;
	if (mem->section_count == 0) {
		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);
	}

	mutex_unlock(&mem_sysfs_mutex);
	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
2938ffbd NF	30	static DEFINE_MUTEX(mem_sysfs_mutex);
2938ffbd NF	31
3947be19 DH	32	#define MEMORY_CLASS_NAME "memory"
	33
	34	static struct sysdev_class memory_sysdev_class = {
af5ca3f4	35	.name = MEMORY_CLASS_NAME,
3947be19	36	};
3947be19	37
312c004d	38	static const char memory_uevent_name(struct kset kset, struct kobject *kobj)
3947be19 DH	39	{
	40	return MEMORY_CLASS_NAME;
	41	}
	42
9ec0fd4e	43	static int memory_uevent(struct kset kset, struct kobject obj, struct kobj_uevent_env *env)
3947be19 DH	44	{
	45	int retval = 0;
	46
	47	return retval;
	48	}
	49
9cd43611	50	static const struct kset_uevent_ops memory_uevent_ops = {
312c004d KS	51	.name = memory_uevent_name,
312c004d KS	52	.uevent = memory_uevent,
3947be19 DH	53	};
3947be19 DH	54
e041c683	55	static BLOCKING_NOTIFIER_HEAD(memory_chain);
3947be19	56
98a38ebd	57	int register_memory_notifier(struct notifier_block *nb)
3947be19	58	{
e041c683	59	return blocking_notifier_chain_register(&memory_chain, nb);
3947be19	60	}
3c82c30c	61	EXPORT_SYMBOL(register_memory_notifier);
3947be19	62
98a38ebd	63	void unregister_memory_notifier(struct notifier_block *nb)
3947be19	64	{
e041c683	65	blocking_notifier_chain_unregister(&memory_chain, nb);
3947be19	66	}
3c82c30c	67	EXPORT_SYMBOL(unregister_memory_notifier);
3947be19	68
925cc71e RJ	69	static ATOMIC_NOTIFIER_HEAD(memory_isolate_chain);
	70
	71	int register_memory_isolate_notifier(struct notifier_block *nb)
	72	{
	73	return atomic_notifier_chain_register(&memory_isolate_chain, nb);
	74	}
	75	EXPORT_SYMBOL(register_memory_isolate_notifier);
	76
	77	void unregister_memory_isolate_notifier(struct notifier_block *nb)
	78	{
	79	atomic_notifier_chain_unregister(&memory_isolate_chain, nb);
	80	}
	81	EXPORT_SYMBOL(unregister_memory_isolate_notifier);
	82
3947be19 DH	83	/*
	84	* register_memory - Setup a sysfs device for a memory block
	85	*/
00a41db5 BP	86	static
00a41db5 BP	87	int register_memory(struct memory_block memory, struct mem_section section)
3947be19 DH	88	{
	89	int error;
	90
	91	memory->sysdev.cls = &memory_sysdev_class;
	92	memory->sysdev.id = __section_nr(section);
	93
	94	error = sysdev_register(&memory->sysdev);
3947be19 DH	95	return error;
	96	}
	97
	98	static void
00a41db5	99	unregister_memory(struct memory_block memory, struct mem_section section)
3947be19 DH	100	{
	101	BUG_ON(memory->sysdev.cls != &memory_sysdev_class);
	102	BUG_ON(memory->sysdev.id != __section_nr(section));
	103
00a41db5 BP	104	/* drop the ref. we got in remove_memory_block() */
00a41db5 BP	105	kobject_put(&memory->sysdev.kobj);
3947be19	106	sysdev_unregister(&memory->sysdev);
3947be19 DH	107	}
	108
	109	/*
	110	* use this as the physical section index that this memsection
	111	* uses.
	112	*/
	113
4a0b2b4d AK	114	static ssize_t show_mem_phys_index(struct sys_device *dev,
4a0b2b4d AK	115	struct sysdev_attribute attr, char buf)
3947be19 DH	116	{
	117	struct memory_block *mem =
	118	container_of(dev, struct memory_block, sysdev);
	119	return sprintf(buf, "%08lx\n", mem->phys_index);
	120	}
	121
5c755e9f BP	122	/*
	123	* Show whether the section of memory is likely to be hot-removable
	124	*/
1f07be1c SR	125	static ssize_t show_mem_removable(struct sys_device *dev,
1f07be1c SR	126	struct sysdev_attribute attr, char buf)
5c755e9f BP	127	{
	128	unsigned long start_pfn;
	129	int ret;
	130	struct memory_block *mem =
	131	container_of(dev, struct memory_block, sysdev);
	132
	133	start_pfn = section_nr_to_pfn(mem->phys_index);
	134	ret = is_mem_section_removable(start_pfn, PAGES_PER_SECTION);
	135	return sprintf(buf, "%d\n", ret);
	136	}
	137
3947be19 DH	138	/*
	139	* online, offline, going offline, etc.
	140	*/
4a0b2b4d AK	141	static ssize_t show_mem_state(struct sys_device *dev,
4a0b2b4d AK	142	struct sysdev_attribute attr, char buf)
3947be19 DH	143	{
	144	struct memory_block *mem =
	145	container_of(dev, struct memory_block, sysdev);
	146	ssize_t len = 0;
	147
	148	/*
	149	* We can probably put these states in a nice little array
	150	* so that they're not open-coded
	151	*/
	152	switch (mem->state) {
	153	case MEM_ONLINE:
	154	len = sprintf(buf, "online\n");
	155	break;
	156	case MEM_OFFLINE:
	157	len = sprintf(buf, "offline\n");
	158	break;
	159	case MEM_GOING_OFFLINE:
	160	len = sprintf(buf, "going-offline\n");
	161	break;
	162	default:
	163	len = sprintf(buf, "ERROR-UNKNOWN-%ld\n",
	164	mem->state);
	165	WARN_ON(1);
	166	break;
	167	}
	168
	169	return len;
	170	}
	171
7b78d335	172	int memory_notify(unsigned long val, void *v)
3947be19	173	{
e041c683	174	return blocking_notifier_call_chain(&memory_chain, val, v);
3947be19 DH	175	}
3947be19 DH	176
925cc71e RJ	177	int memory_isolate_notify(unsigned long val, void *v)
	178	{
	179	return atomic_notifier_call_chain(&memory_isolate_chain, val, v);
	180	}
	181
3947be19 DH	182	/*
	183	* MEMORY_HOTPLUG depends on SPARSEMEM in mm/Kconfig, so it is
	184	* OK to have direct references to sparsemem variables in here.
	185	*/
	186	static int
	187	memory_block_action(struct memory_block *mem, unsigned long action)
	188	{
	189	int i;
	190	unsigned long psection;
	191	unsigned long start_pfn, start_paddr;
	192	struct page *first_page;
	193	int ret;
	194	int old_state = mem->state;
	195
	196	psection = mem->phys_index;
	197	first_page = pfn_to_page(psection << PFN_SECTION_SHIFT);
	198
	199	/*
	200	* The probe routines leave the pages reserved, just
	201	* as the bootmem code does. Make sure they're still
	202	* that way.
	203	*/
	204	if (action == MEM_ONLINE) {
	205	for (i = 0; i < PAGES_PER_SECTION; i++) {
	206	if (PageReserved(first_page+i))
	207	continue;
	208
	209	printk(KERN_WARNING "section number %ld page number %d "
	210	"not reserved, was it already online? \n",
	211	psection, i);
	212	return -EBUSY;
	213	}
	214	}
	215
	216	switch (action) {
	217	case MEM_ONLINE:
	218	start_pfn = page_to_pfn(first_page);
	219	ret = online_pages(start_pfn, PAGES_PER_SECTION);
	220	break;
	221	case MEM_OFFLINE:
	222	mem->state = MEM_GOING_OFFLINE;
3947be19 DH	223	start_paddr = page_to_pfn(first_page) << PAGE_SHIFT;
	224	ret = remove_memory(start_paddr,
	225	PAGES_PER_SECTION << PAGE_SHIFT);
	226	if (ret) {
	227	mem->state = old_state;
	228	break;
	229	}
3947be19 DH	230	break;
3947be19 DH	231	default:
f810a5cf	232	WARN(1, KERN_WARNING "%s(%p, %ld) unknown action: %ld\n",
2b3a302a	233	__func__, mem, action, action);
3947be19 DH	234	ret = -EINVAL;
3947be19 DH	235	}
3947be19 DH	236
	237	return ret;
	238	}
	239
	240	static int memory_block_change_state(struct memory_block *mem,
	241	unsigned long to_state, unsigned long from_state_req)
	242	{
	243	int ret = 0;
da19cbcf	244	mutex_lock(&mem->state_mutex);
3947be19 DH	245
	246	if (mem->state != from_state_req) {
	247	ret = -EINVAL;
	248	goto out;
	249	}
	250
	251	ret = memory_block_action(mem, to_state);
	252	if (!ret)
	253	mem->state = to_state;
	254
	255	out:
da19cbcf	256	mutex_unlock(&mem->state_mutex);
3947be19 DH	257	return ret;
	258	}
	259
	260	static ssize_t
4a0b2b4d AK	261	store_mem_state(struct sys_device *dev,
4a0b2b4d AK	262	struct sysdev_attribute attr, const char buf, size_t count)
3947be19 DH	263	{
	264	struct memory_block *mem;
	265	unsigned int phys_section_nr;
	266	int ret = -EINVAL;
	267
	268	mem = container_of(dev, struct memory_block, sysdev);
	269	phys_section_nr = mem->phys_index;
	270
540557b9	271	if (!present_section_nr(phys_section_nr))
3947be19 DH	272	goto out;
	273
	274	if (!strncmp(buf, "online", min((int)count, 6)))
	275	ret = memory_block_change_state(mem, MEM_ONLINE, MEM_OFFLINE);
	276	else if(!strncmp(buf, "offline", min((int)count, 7)))
	277	ret = memory_block_change_state(mem, MEM_OFFLINE, MEM_ONLINE);
	278	out:
	279	if (ret)
	280	return ret;
	281	return count;
	282	}
	283
	284	/*
	285	* phys_device is a bad name for this. What I really want
	286	* is a way to differentiate between memory ranges that
	287	* are part of physical devices that constitute
	288	* a complete removable unit or fru.
	289	* i.e. do these ranges belong to the same physical device,
	290	* s.t. if I offline all of these sections I can then
	291	* remove the physical device?
	292	*/
4a0b2b4d AK	293	static ssize_t show_phys_device(struct sys_device *dev,
4a0b2b4d AK	294	struct sysdev_attribute attr, char buf)
3947be19 DH	295	{
	296	struct memory_block *mem =
	297	container_of(dev, struct memory_block, sysdev);
	298	return sprintf(buf, "%d\n", mem->phys_device);
	299	}
	300
	301	static SYSDEV_ATTR(phys_index, 0444, show_mem_phys_index, NULL);
	302	static SYSDEV_ATTR(state, 0644, show_mem_state, store_mem_state);
	303	static SYSDEV_ATTR(phys_device, 0444, show_phys_device, NULL);
5c755e9f	304	static SYSDEV_ATTR(removable, 0444, show_mem_removable, NULL);
3947be19 DH	305
	306	#define mem_create_simple_file(mem, attr_name) \
	307	sysdev_create_file(&mem->sysdev, &attr_##attr_name)
	308	#define mem_remove_simple_file(mem, attr_name) \
	309	sysdev_remove_file(&mem->sysdev, &attr_##attr_name)
	310
	311	/*
	312	* Block size attribute stuff
	313	*/
	314	static ssize_t
8564a6c1 AK	315	print_block_size(struct sysdev_class class, struct sysdev_class_attribute attr,
8564a6c1 AK	316	char *buf)
3947be19	317	{
4dc86ae1	318	return sprintf(buf, "%lx\n", (unsigned long)PAGES_PER_SECTION * PAGE_SIZE);
3947be19 DH	319	}
3947be19 DH	320
8564a6c1	321	static SYSDEV_CLASS_ATTR(block_size_bytes, 0444, print_block_size, NULL);
3947be19 DH	322
	323	static int block_size_init(void)
	324	{
28ec24e2	325	return sysfs_create_file(&memory_sysdev_class.kset.kobj,
8564a6c1	326	&attr_block_size_bytes.attr);
3947be19 DH	327	}
	328
	329	/*
	330	* Some architectures will have custom drivers to do this, and
	331	* will not need to do it from userspace. The fake hot-add code
	332	* as well as ppc64 will do all of their discovery in userspace
	333	* and will require this interface.
	334	*/
	335	#ifdef CONFIG_ARCH_MEMORY_PROBE
	336	static ssize_t
28812fe1 AK	337	memory_probe_store(struct class class, struct class_attribute attr,
28812fe1 AK	338	const char *buf, size_t count)
3947be19 DH	339	{
3947be19 DH	340	u64 phys_addr;
bc02af93	341	int nid;
3947be19 DH	342	int ret;
	343
	344	phys_addr = simple_strtoull(buf, NULL, 0);
	345
bc02af93 YG	346	nid = memory_add_physaddr_to_nid(phys_addr);
bc02af93 YG	347	ret = add_memory(nid, phys_addr, PAGES_PER_SECTION << PAGE_SHIFT);
3947be19 DH	348
	349	if (ret)
	350	count = ret;
	351
	352	return count;
	353	}
bd796671	354	static CLASS_ATTR(probe, S_IWUSR, NULL, memory_probe_store);
3947be19 DH	355
	356	static int memory_probe_init(void)
	357	{
28ec24e2	358	return sysfs_create_file(&memory_sysdev_class.kset.kobj,
bd796671	359	&class_attr_probe.attr);
3947be19 DH	360	}
3947be19 DH	361	#else
28ec24e2 AM	362	static inline int memory_probe_init(void)
	363	{
	364	return 0;
	365	}
3947be19 DH	366	#endif
3947be19 DH	367
facb6011 AK	368	#ifdef CONFIG_MEMORY_FAILURE
	369	/*
	370	* Support for offlining pages of memory
	371	*/
	372
	373	/* Soft offline a page */
	374	static ssize_t
28812fe1 AK	375	store_soft_offline_page(struct class *class,
	376	struct class_attribute *attr,
	377	const char *buf, size_t count)
facb6011 AK	378	{
	379	int ret;
	380	u64 pfn;
	381	if (!capable(CAP_SYS_ADMIN))
	382	return -EPERM;
	383	if (strict_strtoull(buf, 0, &pfn) < 0)
	384	return -EINVAL;
	385	pfn >>= PAGE_SHIFT;
	386	if (!pfn_valid(pfn))
	387	return -ENXIO;
	388	ret = soft_offline_page(pfn_to_page(pfn), 0);
	389	return ret == 0 ? count : ret;
	390	}
	391
	392	/* Forcibly offline a page, including killing processes. */
	393	static ssize_t
28812fe1 AK	394	store_hard_offline_page(struct class *class,
	395	struct class_attribute *attr,
	396	const char *buf, size_t count)
facb6011 AK	397	{
	398	int ret;
	399	u64 pfn;
	400	if (!capable(CAP_SYS_ADMIN))
	401	return -EPERM;
	402	if (strict_strtoull(buf, 0, &pfn) < 0)
	403	return -EINVAL;
	404	pfn >>= PAGE_SHIFT;
	405	ret = __memory_failure(pfn, 0, 0);
	406	return ret ? ret : count;
	407	}
	408
bd796671 GKH	409	static CLASS_ATTR(soft_offline_page, 0644, NULL, store_soft_offline_page);
bd796671 GKH	410	static CLASS_ATTR(hard_offline_page, 0644, NULL, store_hard_offline_page);
facb6011 AK	411
	412	static __init int memory_fail_init(void)
	413	{
	414	int err;
	415
	416	err = sysfs_create_file(&memory_sysdev_class.kset.kobj,
bd796671	417	&class_attr_soft_offline_page.attr);
facb6011 AK	418	if (!err)
facb6011 AK	419	err = sysfs_create_file(&memory_sysdev_class.kset.kobj,
bd796671	420	&class_attr_hard_offline_page.attr);
facb6011 AK	421	return err;
	422	}
	423	#else
	424	static inline int memory_fail_init(void)
	425	{
	426	return 0;
	427	}
	428	#endif
	429
3947be19 DH	430	/*
	431	* Note that phys_device is optional. It is here to allow for
	432	* differentiation between which physical devices each
	433	* section belongs to...
	434	*/
bc32df00 HC	435	int __weak arch_get_memory_phys_device(unsigned long start_pfn)
	436	{
	437	return 0;
	438	}
3947be19	439
98383031 RH	440	struct memory_block find_memory_block_hinted(struct mem_section section,
98383031 RH	441	struct memory_block *hint)
3947be19 DH	442	{
	443	struct kobject *kobj;
	444	struct sys_device *sysdev;
	445	struct memory_block *mem;
	446	char name[sizeof(MEMORY_CLASS_NAME) + 9 + 1];
	447
98383031 RH	448	kobj = hint ? &hint->sysdev.kobj : NULL;
98383031 RH	449
3947be19 DH	450	/*
	451	* This only works because we know that section == sysdev->id
	452	* slightly redundant with sysdev_register()
	453	*/
	454	sprintf(&name[0], "%s%d", MEMORY_CLASS_NAME, __section_nr(section));
	455
98383031	456	kobj = kset_find_obj_hinted(&memory_sysdev_class.kset, name, kobj);
3947be19 DH	457	if (!kobj)
	458	return NULL;
	459
	460	sysdev = container_of(kobj, struct sys_device, kobj);
	461	mem = container_of(sysdev, struct memory_block, sysdev);
	462
	463	return mem;
	464	}
	465
98383031 RH	466	/*
	467	* For now, we have a linear search to go find the appropriate
	468	* memory_block corresponding to a particular phys_index. If
	469	* this gets to be a real problem, we can always use a radix
	470	* tree or something here.
	471	*
	472	* This could be made generic for all sysdev classes.
	473	*/
	474	struct memory_block find_memory_block(struct mem_section section)
	475	{
	476	return find_memory_block_hinted(section, NULL);
	477	}
	478
e4619c85 NF	479	static int add_memory_block(int nid, struct mem_section *section,
	480	unsigned long state, enum mem_add_context context)
	481	{
	482	struct memory_block mem = kzalloc(sizeof(mem), GFP_KERNEL);
	483	unsigned long start_pfn;
	484	int ret = 0;
	485
	486	if (!mem)
	487	return -ENOMEM;
	488
2938ffbd NF	489	mutex_lock(&mem_sysfs_mutex);
2938ffbd NF	490
e4619c85 NF	491	mem->phys_index = __section_nr(section);
e4619c85 NF	492	mem->state = state;
07681215	493	mem->section_count++;
e4619c85 NF	494	mutex_init(&mem->state_mutex);
	495	start_pfn = section_nr_to_pfn(mem->phys_index);
	496	mem->phys_device = arch_get_memory_phys_device(start_pfn);
	497
	498	ret = register_memory(mem, section);
	499	if (!ret)
	500	ret = mem_create_simple_file(mem, phys_index);
	501	if (!ret)
	502	ret = mem_create_simple_file(mem, state);
	503	if (!ret)
	504	ret = mem_create_simple_file(mem, phys_device);
	505	if (!ret)
	506	ret = mem_create_simple_file(mem, removable);
	507	if (!ret) {
	508	if (context == HOTPLUG)
	509	ret = register_mem_sect_under_node(mem, nid);
	510	}
	511
2938ffbd	512	mutex_unlock(&mem_sysfs_mutex);
e4619c85 NF	513	return ret;
	514	}
	515
3947be19 DH	516	int remove_memory_block(unsigned long node_id, struct mem_section *section,
	517	int phys_device)
	518	{
	519	struct memory_block *mem;
	520
2938ffbd	521	mutex_lock(&mem_sysfs_mutex);
3947be19	522	mem = find_memory_block(section);
07681215 NF	523
	524	mem->section_count--;
	525	if (mem->section_count == 0) {
	526	unregister_mem_sect_under_nodes(mem);
	527	mem_remove_simple_file(mem, phys_index);
	528	mem_remove_simple_file(mem, state);
	529	mem_remove_simple_file(mem, phys_device);
	530	mem_remove_simple_file(mem, removable);
	531	unregister_memory(mem, section);
	532	}
3947be19	533
2938ffbd	534	mutex_unlock(&mem_sysfs_mutex);
3947be19 DH	535	return 0;
	536	}
	537
	538	/*
	539	* need an interface for the VM to add new memory regions,
	540	* but without onlining it.
	541	*/
c04fc586	542	int register_new_memory(int nid, struct mem_section *section)
3947be19	543	{
bc32df00	544	return add_memory_block(nid, section, MEM_OFFLINE, HOTPLUG);
3947be19 DH	545	}
	546
	547	int unregister_memory_section(struct mem_section *section)
	548	{
540557b9	549	if (!present_section(section))
3947be19 DH	550	return -EINVAL;
	551
	552	return remove_memory_block(0, section, 0);
	553	}
	554
	555	/*
	556	* Initialize the sysfs support for memory devices...
	557	*/
	558	int __init memory_dev_init(void)
	559	{
	560	unsigned int i;
	561	int ret;
28ec24e2	562	int err;
3947be19	563
312c004d	564	memory_sysdev_class.kset.uevent_ops = &memory_uevent_ops;
3947be19	565	ret = sysdev_class_register(&memory_sysdev_class);
28ec24e2 AM	566	if (ret)
28ec24e2 AM	567	goto out;
3947be19 DH	568
	569	/*
	570	* Create entries for memory sections that were found
	571	* during boot and have been initialized
	572	*/
	573	for (i = 0; i < NR_MEM_SECTIONS; i++) {
540557b9	574	if (!present_section_nr(i))
3947be19	575	continue;
c04fc586	576	err = add_memory_block(0, __nr_to_section(i), MEM_ONLINE,
bc32df00	577	BOOT);
28ec24e2 AM	578	if (!ret)
28ec24e2 AM	579	ret = err;
3947be19 DH	580	}
3947be19 DH	581
28ec24e2	582	err = memory_probe_init();
facb6011 AK	583	if (!ret)
	584	ret = err;
	585	err = memory_fail_init();
28ec24e2 AM	586	if (!ret)
	587	ret = err;
	588	err = block_size_init();
	589	if (!ret)
	590	ret = err;
	591	out:
	592	if (ret)
2b3a302a	593	printk(KERN_ERR "%s() failed: %d\n", __func__, ret);
3947be19 DH	594	return ret;
3947be19 DH	595	}