[mirror_ubuntu-artful-kernel.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 <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 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);

/*
 * 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, 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);
}

/*
 * 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 class *class, char *buf)
{
	return sprintf(buf, "%lx\n", (unsigned long)PAGES_PER_SECTION * PAGE_SIZE);
}

static 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,
				&class_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, 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, 0700, 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

/*
 * Note that phys_device is optional.  It is here to allow for
 * differentiation between which *physical* devices each
 * section belongs to...
 */

static int add_memory_block(unsigned long node_id, struct mem_section *section,
		     unsigned long state, int phys_device)
{
	struct memory_block *mem = kzalloc(sizeof(*mem), GFP_KERNEL);
	int ret = 0;

	if (!mem)
		return -ENOMEM;

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

	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);

	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.
 */
static 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);
	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(struct mem_section *section)
{
	return add_memory_block(0, section, MEM_OFFLINE, 0);
}

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, 0);
		if (!ret)
			ret = err;
	}

	err = memory_probe_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>
3947be19 DH	24	#include <asm/atomic.h>
	25	#include <asm/uaccess.h>
	26
	27	#define MEMORY_CLASS_NAME "memory"
	28
	29	static struct sysdev_class memory_sysdev_class = {
af5ca3f4	30	.name = MEMORY_CLASS_NAME,
3947be19	31	};
3947be19	32
312c004d	33	static const char memory_uevent_name(struct kset kset, struct kobject *kobj)
3947be19 DH	34	{
	35	return MEMORY_CLASS_NAME;
	36	}
	37
9ec0fd4e	38	static int memory_uevent(struct kset kset, struct kobject obj, struct kobj_uevent_env *env)
3947be19 DH	39	{
	40	int retval = 0;
	41
	42	return retval;
	43	}
	44
312c004d KS	45	static struct kset_uevent_ops memory_uevent_ops = {
	46	.name = memory_uevent_name,
	47	.uevent = memory_uevent,
3947be19 DH	48	};
3947be19 DH	49
e041c683	50	static BLOCKING_NOTIFIER_HEAD(memory_chain);
3947be19	51
98a38ebd	52	int register_memory_notifier(struct notifier_block *nb)
3947be19	53	{
e041c683	54	return blocking_notifier_chain_register(&memory_chain, nb);
3947be19	55	}
3c82c30c	56	EXPORT_SYMBOL(register_memory_notifier);
3947be19	57
98a38ebd	58	void unregister_memory_notifier(struct notifier_block *nb)
3947be19	59	{
e041c683	60	blocking_notifier_chain_unregister(&memory_chain, nb);
3947be19	61	}
3c82c30c	62	EXPORT_SYMBOL(unregister_memory_notifier);
3947be19 DH	63
	64	/*
	65	* register_memory - Setup a sysfs device for a memory block
	66	*/
00a41db5 BP	67	static
00a41db5 BP	68	int register_memory(struct memory_block memory, struct mem_section section)
3947be19 DH	69	{
	70	int error;
	71
	72	memory->sysdev.cls = &memory_sysdev_class;
	73	memory->sysdev.id = __section_nr(section);
	74
	75	error = sysdev_register(&memory->sysdev);
3947be19 DH	76	return error;
	77	}
	78
	79	static void
00a41db5	80	unregister_memory(struct memory_block memory, struct mem_section section)
3947be19 DH	81	{
	82	BUG_ON(memory->sysdev.cls != &memory_sysdev_class);
	83	BUG_ON(memory->sysdev.id != __section_nr(section));
	84
00a41db5 BP	85	/* drop the ref. we got in remove_memory_block() */
00a41db5 BP	86	kobject_put(&memory->sysdev.kobj);
3947be19	87	sysdev_unregister(&memory->sysdev);
3947be19 DH	88	}
	89
	90	/*
	91	* use this as the physical section index that this memsection
	92	* uses.
	93	*/
	94
4a0b2b4d AK	95	static ssize_t show_mem_phys_index(struct sys_device *dev,
4a0b2b4d AK	96	struct sysdev_attribute attr, char buf)
3947be19 DH	97	{
	98	struct memory_block *mem =
	99	container_of(dev, struct memory_block, sysdev);
	100	return sprintf(buf, "%08lx\n", mem->phys_index);
	101	}
	102
5c755e9f BP	103	/*
	104	* Show whether the section of memory is likely to be hot-removable
	105	*/
	106	static ssize_t show_mem_removable(struct sys_device dev, char buf)
	107	{
	108	unsigned long start_pfn;
	109	int ret;
	110	struct memory_block *mem =
	111	container_of(dev, struct memory_block, sysdev);
	112
	113	start_pfn = section_nr_to_pfn(mem->phys_index);
	114	ret = is_mem_section_removable(start_pfn, PAGES_PER_SECTION);
	115	return sprintf(buf, "%d\n", ret);
	116	}
	117
3947be19 DH	118	/*
	119	* online, offline, going offline, etc.
	120	*/
4a0b2b4d AK	121	static ssize_t show_mem_state(struct sys_device *dev,
4a0b2b4d AK	122	struct sysdev_attribute attr, char buf)
3947be19 DH	123	{
	124	struct memory_block *mem =
	125	container_of(dev, struct memory_block, sysdev);
	126	ssize_t len = 0;
	127
	128	/*
	129	* We can probably put these states in a nice little array
	130	* so that they're not open-coded
	131	*/
	132	switch (mem->state) {
	133	case MEM_ONLINE:
	134	len = sprintf(buf, "online\n");
	135	break;
	136	case MEM_OFFLINE:
	137	len = sprintf(buf, "offline\n");
	138	break;
	139	case MEM_GOING_OFFLINE:
	140	len = sprintf(buf, "going-offline\n");
	141	break;
	142	default:
	143	len = sprintf(buf, "ERROR-UNKNOWN-%ld\n",
	144	mem->state);
	145	WARN_ON(1);
	146	break;
	147	}
	148
	149	return len;
	150	}
	151
7b78d335	152	int memory_notify(unsigned long val, void *v)
3947be19	153	{
e041c683	154	return blocking_notifier_call_chain(&memory_chain, val, v);
3947be19 DH	155	}
	156
	157	/*
	158	* MEMORY_HOTPLUG depends on SPARSEMEM in mm/Kconfig, so it is
	159	* OK to have direct references to sparsemem variables in here.
	160	*/
	161	static int
	162	memory_block_action(struct memory_block *mem, unsigned long action)
	163	{
	164	int i;
	165	unsigned long psection;
	166	unsigned long start_pfn, start_paddr;
	167	struct page *first_page;
	168	int ret;
	169	int old_state = mem->state;
	170
	171	psection = mem->phys_index;
	172	first_page = pfn_to_page(psection << PFN_SECTION_SHIFT);
	173
	174	/*
	175	* The probe routines leave the pages reserved, just
	176	* as the bootmem code does. Make sure they're still
	177	* that way.
	178	*/
	179	if (action == MEM_ONLINE) {
	180	for (i = 0; i < PAGES_PER_SECTION; i++) {
	181	if (PageReserved(first_page+i))
	182	continue;
	183
	184	printk(KERN_WARNING "section number %ld page number %d "
	185	"not reserved, was it already online? \n",
	186	psection, i);
	187	return -EBUSY;
	188	}
	189	}
	190
	191	switch (action) {
	192	case MEM_ONLINE:
	193	start_pfn = page_to_pfn(first_page);
	194	ret = online_pages(start_pfn, PAGES_PER_SECTION);
	195	break;
	196	case MEM_OFFLINE:
	197	mem->state = MEM_GOING_OFFLINE;
3947be19 DH	198	start_paddr = page_to_pfn(first_page) << PAGE_SHIFT;
	199	ret = remove_memory(start_paddr,
	200	PAGES_PER_SECTION << PAGE_SHIFT);
	201	if (ret) {
	202	mem->state = old_state;
	203	break;
	204	}
3947be19 DH	205	break;
3947be19 DH	206	default:
f810a5cf	207	WARN(1, KERN_WARNING "%s(%p, %ld) unknown action: %ld\n",
2b3a302a	208	__func__, mem, action, action);
3947be19 DH	209	ret = -EINVAL;
3947be19 DH	210	}
3947be19 DH	211
	212	return ret;
	213	}
	214
	215	static int memory_block_change_state(struct memory_block *mem,
	216	unsigned long to_state, unsigned long from_state_req)
	217	{
	218	int ret = 0;
da19cbcf	219	mutex_lock(&mem->state_mutex);
3947be19 DH	220
	221	if (mem->state != from_state_req) {
	222	ret = -EINVAL;
	223	goto out;
	224	}
	225
	226	ret = memory_block_action(mem, to_state);
	227	if (!ret)
	228	mem->state = to_state;
	229
	230	out:
da19cbcf	231	mutex_unlock(&mem->state_mutex);
3947be19 DH	232	return ret;
	233	}
	234
	235	static ssize_t
4a0b2b4d AK	236	store_mem_state(struct sys_device *dev,
4a0b2b4d AK	237	struct sysdev_attribute attr, const char buf, size_t count)
3947be19 DH	238	{
	239	struct memory_block *mem;
	240	unsigned int phys_section_nr;
	241	int ret = -EINVAL;
	242
	243	mem = container_of(dev, struct memory_block, sysdev);
	244	phys_section_nr = mem->phys_index;
	245
540557b9	246	if (!present_section_nr(phys_section_nr))
3947be19 DH	247	goto out;
	248
	249	if (!strncmp(buf, "online", min((int)count, 6)))
	250	ret = memory_block_change_state(mem, MEM_ONLINE, MEM_OFFLINE);
	251	else if(!strncmp(buf, "offline", min((int)count, 7)))
	252	ret = memory_block_change_state(mem, MEM_OFFLINE, MEM_ONLINE);
	253	out:
	254	if (ret)
	255	return ret;
	256	return count;
	257	}
	258
	259	/*
	260	* phys_device is a bad name for this. What I really want
	261	* is a way to differentiate between memory ranges that
	262	* are part of physical devices that constitute
	263	* a complete removable unit or fru.
	264	* i.e. do these ranges belong to the same physical device,
	265	* s.t. if I offline all of these sections I can then
	266	* remove the physical device?
	267	*/
4a0b2b4d AK	268	static ssize_t show_phys_device(struct sys_device *dev,
4a0b2b4d AK	269	struct sysdev_attribute attr, char buf)
3947be19 DH	270	{
	271	struct memory_block *mem =
	272	container_of(dev, struct memory_block, sysdev);
	273	return sprintf(buf, "%d\n", mem->phys_device);
	274	}
	275
	276	static SYSDEV_ATTR(phys_index, 0444, show_mem_phys_index, NULL);
	277	static SYSDEV_ATTR(state, 0644, show_mem_state, store_mem_state);
	278	static SYSDEV_ATTR(phys_device, 0444, show_phys_device, NULL);
5c755e9f	279	static SYSDEV_ATTR(removable, 0444, show_mem_removable, NULL);
3947be19 DH	280
	281	#define mem_create_simple_file(mem, attr_name) \
	282	sysdev_create_file(&mem->sysdev, &attr_##attr_name)
	283	#define mem_remove_simple_file(mem, attr_name) \
	284	sysdev_remove_file(&mem->sysdev, &attr_##attr_name)
	285
	286	/*
	287	* Block size attribute stuff
	288	*/
	289	static ssize_t
	290	print_block_size(struct class class, char buf)
	291	{
	292	return sprintf(buf, "%lx\n", (unsigned long)PAGES_PER_SECTION * PAGE_SIZE);
	293	}
	294
	295	static CLASS_ATTR(block_size_bytes, 0444, print_block_size, NULL);
	296
	297	static int block_size_init(void)
	298	{
28ec24e2 AM	299	return sysfs_create_file(&memory_sysdev_class.kset.kobj,
28ec24e2 AM	300	&class_attr_block_size_bytes.attr);
3947be19 DH	301	}
	302
	303	/*
	304	* Some architectures will have custom drivers to do this, and
	305	* will not need to do it from userspace. The fake hot-add code
	306	* as well as ppc64 will do all of their discovery in userspace
	307	* and will require this interface.
	308	*/
	309	#ifdef CONFIG_ARCH_MEMORY_PROBE
	310	static ssize_t
be7ee9b2	311	memory_probe_store(struct class class, const char buf, size_t count)
3947be19 DH	312	{
3947be19 DH	313	u64 phys_addr;
bc02af93	314	int nid;
3947be19 DH	315	int ret;
	316
	317	phys_addr = simple_strtoull(buf, NULL, 0);
	318
bc02af93 YG	319	nid = memory_add_physaddr_to_nid(phys_addr);
bc02af93 YG	320	ret = add_memory(nid, phys_addr, PAGES_PER_SECTION << PAGE_SHIFT);
3947be19 DH	321
	322	if (ret)
	323	count = ret;
	324
	325	return count;
	326	}
	327	static CLASS_ATTR(probe, 0700, NULL, memory_probe_store);
	328
	329	static int memory_probe_init(void)
	330	{
28ec24e2 AM	331	return sysfs_create_file(&memory_sysdev_class.kset.kobj,
28ec24e2 AM	332	&class_attr_probe.attr);
3947be19 DH	333	}
3947be19 DH	334	#else
28ec24e2 AM	335	static inline int memory_probe_init(void)
	336	{
	337	return 0;
	338	}
3947be19 DH	339	#endif
	340
	341	/*
	342	* Note that phys_device is optional. It is here to allow for
	343	* differentiation between which physical devices each
	344	* section belongs to...
	345	*/
	346
	347	static int add_memory_block(unsigned long node_id, struct mem_section *section,
	348	unsigned long state, int phys_device)
	349	{
0b0acbec	350	struct memory_block mem = kzalloc(sizeof(mem), GFP_KERNEL);
3947be19 DH	351	int ret = 0;
	352
	353	if (!mem)
	354	return -ENOMEM;
	355
3947be19 DH	356	mem->phys_index = __section_nr(section);
3947be19 DH	357	mem->state = state;
da19cbcf	358	mutex_init(&mem->state_mutex);
3947be19 DH	359	mem->phys_device = phys_device;
3947be19 DH	360
00a41db5	361	ret = register_memory(mem, section);
3947be19 DH	362	if (!ret)
	363	ret = mem_create_simple_file(mem, phys_index);
	364	if (!ret)
	365	ret = mem_create_simple_file(mem, state);
	366	if (!ret)
	367	ret = mem_create_simple_file(mem, phys_device);
5c755e9f BP	368	if (!ret)
5c755e9f BP	369	ret = mem_create_simple_file(mem, removable);
3947be19 DH	370
	371	return ret;
	372	}
	373
	374	/*
	375	* For now, we have a linear search to go find the appropriate
	376	* memory_block corresponding to a particular phys_index. If
	377	* this gets to be a real problem, we can always use a radix
	378	* tree or something here.
	379	*
	380	* This could be made generic for all sysdev classes.
	381	*/
	382	static struct memory_block find_memory_block(struct mem_section section)
	383	{
	384	struct kobject *kobj;
	385	struct sys_device *sysdev;
	386	struct memory_block *mem;
	387	char name[sizeof(MEMORY_CLASS_NAME) + 9 + 1];
	388
	389	/*
	390	* This only works because we know that section == sysdev->id
	391	* slightly redundant with sysdev_register()
	392	*/
	393	sprintf(&name[0], "%s%d", MEMORY_CLASS_NAME, __section_nr(section));
	394
	395	kobj = kset_find_obj(&memory_sysdev_class.kset, name);
	396	if (!kobj)
	397	return NULL;
	398
	399	sysdev = container_of(kobj, struct sys_device, kobj);
	400	mem = container_of(sysdev, struct memory_block, sysdev);
	401
	402	return mem;
	403	}
	404
	405	int remove_memory_block(unsigned long node_id, struct mem_section *section,
	406	int phys_device)
	407	{
	408	struct memory_block *mem;
	409
	410	mem = find_memory_block(section);
	411	mem_remove_simple_file(mem, phys_index);
	412	mem_remove_simple_file(mem, state);
	413	mem_remove_simple_file(mem, phys_device);
5c755e9f	414	mem_remove_simple_file(mem, removable);
00a41db5	415	unregister_memory(mem, section);
3947be19 DH	416
	417	return 0;
	418	}
	419
	420	/*
	421	* need an interface for the VM to add new memory regions,
	422	* but without onlining it.
	423	*/
	424	int register_new_memory(struct mem_section *section)
	425	{
	426	return add_memory_block(0, section, MEM_OFFLINE, 0);
	427	}
	428
	429	int unregister_memory_section(struct mem_section *section)
	430	{
540557b9	431	if (!present_section(section))
3947be19 DH	432	return -EINVAL;
	433
	434	return remove_memory_block(0, section, 0);
	435	}
	436
	437	/*
	438	* Initialize the sysfs support for memory devices...
	439	*/
	440	int __init memory_dev_init(void)
	441	{
	442	unsigned int i;
	443	int ret;
28ec24e2	444	int err;
3947be19	445
312c004d	446	memory_sysdev_class.kset.uevent_ops = &memory_uevent_ops;
3947be19	447	ret = sysdev_class_register(&memory_sysdev_class);
28ec24e2 AM	448	if (ret)
28ec24e2 AM	449	goto out;
3947be19 DH	450
	451	/*
	452	* Create entries for memory sections that were found
	453	* during boot and have been initialized
	454	*/
	455	for (i = 0; i < NR_MEM_SECTIONS; i++) {
540557b9	456	if (!present_section_nr(i))
3947be19	457	continue;
28ec24e2 AM	458	err = add_memory_block(0, __nr_to_section(i), MEM_ONLINE, 0);
	459	if (!ret)
	460	ret = err;
3947be19 DH	461	}
3947be19 DH	462
28ec24e2 AM	463	err = memory_probe_init();
	464	if (!ret)
	465	ret = err;
	466	err = block_size_init();
	467	if (!ret)
	468	ret = err;
	469	out:
	470	if (ret)
2b3a302a	471	printk(KERN_ERR "%s() failed: %d\n", __func__, ret);
3947be19 DH	472	return ret;
3947be19 DH	473	}