[thirdparty/kernel/stable.git] / drivers / base / node.c

/*
 * drivers/base/node.c - basic Node class support
 */

#include <linux/sysdev.h>
#include <linux/module.h>
#include <linux/init.h>
#include <linux/mm.h>
#include <linux/memory.h>
#include <linux/node.h>
#include <linux/hugetlb.h>
#include <linux/cpumask.h>
#include <linux/topology.h>
#include <linux/nodemask.h>
#include <linux/cpu.h>
#include <linux/device.h>
#include <linux/swap.h>
#include <linux/slab.h>

static struct sysdev_class_attribute *node_state_attrs[];

static struct sysdev_class node_class = {
	.name = "node",
	.attrs = node_state_attrs,
};


static ssize_t node_read_cpumap(struct sys_device *dev, int type, char *buf)
{
	struct node *node_dev = to_node(dev);
	const struct cpumask *mask = cpumask_of_node(node_dev->sysdev.id);
	int len;

	/* 2008/04/07: buf currently PAGE_SIZE, need 9 chars per 32 bits. */
	BUILD_BUG_ON((NR_CPUS/32 * 9) > (PAGE_SIZE-1));

	len = type?
		cpulist_scnprintf(buf, PAGE_SIZE-2, mask) :
		cpumask_scnprintf(buf, PAGE_SIZE-2, mask);
 	buf[len++] = '\n';
 	buf[len] = '\0';
	return len;
}

static inline ssize_t node_read_cpumask(struct sys_device *dev,
				struct sysdev_attribute *attr, char *buf)
{
	return node_read_cpumap(dev, 0, buf);
}
static inline ssize_t node_read_cpulist(struct sys_device *dev,
				struct sysdev_attribute *attr, char *buf)
{
	return node_read_cpumap(dev, 1, buf);
}

static SYSDEV_ATTR(cpumap,  S_IRUGO, node_read_cpumask, NULL);
static SYSDEV_ATTR(cpulist, S_IRUGO, node_read_cpulist, NULL);

#define K(x) ((x) << (PAGE_SHIFT - 10))
static ssize_t node_read_meminfo(struct sys_device * dev,
			struct sysdev_attribute *attr, char * buf)
{
	int n;
	int nid = dev->id;
	struct sysinfo i;

	si_meminfo_node(&i, nid);

	n = sprintf(buf, "\n"
		       "Node %d MemTotal:       %8lu kB\n"
		       "Node %d MemFree:        %8lu kB\n"
		       "Node %d MemUsed:        %8lu kB\n"
		       "Node %d Active:         %8lu kB\n"
		       "Node %d Inactive:       %8lu kB\n"
		       "Node %d Active(anon):   %8lu kB\n"
		       "Node %d Inactive(anon): %8lu kB\n"
		       "Node %d Active(file):   %8lu kB\n"
		       "Node %d Inactive(file): %8lu kB\n"
		       "Node %d Unevictable:    %8lu kB\n"
		       "Node %d Mlocked:        %8lu kB\n"
#ifdef CONFIG_HIGHMEM
		       "Node %d HighTotal:      %8lu kB\n"
		       "Node %d HighFree:       %8lu kB\n"
		       "Node %d LowTotal:       %8lu kB\n"
		       "Node %d LowFree:        %8lu kB\n"
#endif
		       "Node %d Dirty:          %8lu kB\n"
		       "Node %d Writeback:      %8lu kB\n"
		       "Node %d FilePages:      %8lu kB\n"
		       "Node %d Mapped:         %8lu kB\n"
		       "Node %d AnonPages:      %8lu kB\n"
		       "Node %d Shmem:          %8lu kB\n"
		       "Node %d KernelStack:    %8lu kB\n"
		       "Node %d PageTables:     %8lu kB\n"
		       "Node %d NFS_Unstable:   %8lu kB\n"
		       "Node %d Bounce:         %8lu kB\n"
		       "Node %d WritebackTmp:   %8lu kB\n"
		       "Node %d Slab:           %8lu kB\n"
		       "Node %d SReclaimable:   %8lu kB\n"
		       "Node %d SUnreclaim:     %8lu kB\n",
		       nid, K(i.totalram),
		       nid, K(i.freeram),
		       nid, K(i.totalram - i.freeram),
		       nid, K(node_page_state(nid, NR_ACTIVE_ANON) +
				node_page_state(nid, NR_ACTIVE_FILE)),
		       nid, K(node_page_state(nid, NR_INACTIVE_ANON) +
				node_page_state(nid, NR_INACTIVE_FILE)),
		       nid, K(node_page_state(nid, NR_ACTIVE_ANON)),
		       nid, K(node_page_state(nid, NR_INACTIVE_ANON)),
		       nid, K(node_page_state(nid, NR_ACTIVE_FILE)),
		       nid, K(node_page_state(nid, NR_INACTIVE_FILE)),
		       nid, K(node_page_state(nid, NR_UNEVICTABLE)),
		       nid, K(node_page_state(nid, NR_MLOCK)),
#ifdef CONFIG_HIGHMEM
		       nid, K(i.totalhigh),
		       nid, K(i.freehigh),
		       nid, K(i.totalram - i.totalhigh),
		       nid, K(i.freeram - i.freehigh),
#endif
		       nid, K(node_page_state(nid, NR_FILE_DIRTY)),
		       nid, K(node_page_state(nid, NR_WRITEBACK)),
		       nid, K(node_page_state(nid, NR_FILE_PAGES)),
		       nid, K(node_page_state(nid, NR_FILE_MAPPED)),
		       nid, K(node_page_state(nid, NR_ANON_PAGES)),
		       nid, K(node_page_state(nid, NR_SHMEM)),
		       nid, node_page_state(nid, NR_KERNEL_STACK) *
				THREAD_SIZE / 1024,
		       nid, K(node_page_state(nid, NR_PAGETABLE)),
		       nid, K(node_page_state(nid, NR_UNSTABLE_NFS)),
		       nid, K(node_page_state(nid, NR_BOUNCE)),
		       nid, K(node_page_state(nid, NR_WRITEBACK_TEMP)),
		       nid, K(node_page_state(nid, NR_SLAB_RECLAIMABLE) +
				node_page_state(nid, NR_SLAB_UNRECLAIMABLE)),
		       nid, K(node_page_state(nid, NR_SLAB_RECLAIMABLE)),
		       nid, K(node_page_state(nid, NR_SLAB_UNRECLAIMABLE)));
	n += hugetlb_report_node_meminfo(nid, buf + n);
	return n;
}

#undef K
static SYSDEV_ATTR(meminfo, S_IRUGO, node_read_meminfo, NULL);

static ssize_t node_read_numastat(struct sys_device * dev,
				struct sysdev_attribute *attr, char * buf)
{
	return sprintf(buf,
		       "numa_hit %lu\n"
		       "numa_miss %lu\n"
		       "numa_foreign %lu\n"
		       "interleave_hit %lu\n"
		       "local_node %lu\n"
		       "other_node %lu\n",
		       node_page_state(dev->id, NUMA_HIT),
		       node_page_state(dev->id, NUMA_MISS),
		       node_page_state(dev->id, NUMA_FOREIGN),
		       node_page_state(dev->id, NUMA_INTERLEAVE_HIT),
		       node_page_state(dev->id, NUMA_LOCAL),
		       node_page_state(dev->id, NUMA_OTHER));
}
static SYSDEV_ATTR(numastat, S_IRUGO, node_read_numastat, NULL);

static ssize_t node_read_distance(struct sys_device * dev,
			struct sysdev_attribute *attr, char * buf)
{
	int nid = dev->id;
	int len = 0;
	int i;

	/*
	 * buf is currently PAGE_SIZE in length and each node needs 4 chars
	 * at the most (distance + space or newline).
	 */
	BUILD_BUG_ON(MAX_NUMNODES * 4 > PAGE_SIZE);

	for_each_online_node(i)
		len += sprintf(buf + len, "%s%d", i ? " " : "", node_distance(nid, i));

	len += sprintf(buf + len, "\n");
	return len;
}
static SYSDEV_ATTR(distance, S_IRUGO, node_read_distance, NULL);

#ifdef CONFIG_HUGETLBFS
/*
 * hugetlbfs per node attributes registration interface:
 * When/if hugetlb[fs] subsystem initializes [sometime after this module],
 * it will register its per node attributes for all online nodes with
 * memory.  It will also call register_hugetlbfs_with_node(), below, to
 * register its attribute registration functions with this node driver.
 * Once these hooks have been initialized, the node driver will call into
 * the hugetlb module to [un]register attributes for hot-plugged nodes.
 */
static node_registration_func_t __hugetlb_register_node;
static node_registration_func_t __hugetlb_unregister_node;

static inline bool hugetlb_register_node(struct node *node)
{
	if (__hugetlb_register_node &&
			node_state(node->sysdev.id, N_HIGH_MEMORY)) {
		__hugetlb_register_node(node);
		return true;
	}
	return false;
}

static inline void hugetlb_unregister_node(struct node *node)
{
	if (__hugetlb_unregister_node)
		__hugetlb_unregister_node(node);
}

void register_hugetlbfs_with_node(node_registration_func_t doregister,
				  node_registration_func_t unregister)
{
	__hugetlb_register_node   = doregister;
	__hugetlb_unregister_node = unregister;
}
#else
static inline void hugetlb_register_node(struct node *node) {}

static inline void hugetlb_unregister_node(struct node *node) {}
#endif


/*
 * register_node - Setup a sysfs device for a node.
 * @num - Node number to use when creating the device.
 *
 * Initialize and register the node device.
 */
int register_node(struct node *node, int num, struct node *parent)
{
	int error;

	node->sysdev.id = num;
	node->sysdev.cls = &node_class;
	error = sysdev_register(&node->sysdev);

	if (!error){
		sysdev_create_file(&node->sysdev, &attr_cpumap);
		sysdev_create_file(&node->sysdev, &attr_cpulist);
		sysdev_create_file(&node->sysdev, &attr_meminfo);
		sysdev_create_file(&node->sysdev, &attr_numastat);
		sysdev_create_file(&node->sysdev, &attr_distance);

		scan_unevictable_register_node(node);

		hugetlb_register_node(node);
	}
	return error;
}

/**
 * unregister_node - unregister a node device
 * @node: node going away
 *
 * Unregisters a node device @node.  All the devices on the node must be
 * unregistered before calling this function.
 */
void unregister_node(struct node *node)
{
	sysdev_remove_file(&node->sysdev, &attr_cpumap);
	sysdev_remove_file(&node->sysdev, &attr_cpulist);
	sysdev_remove_file(&node->sysdev, &attr_meminfo);
	sysdev_remove_file(&node->sysdev, &attr_numastat);
	sysdev_remove_file(&node->sysdev, &attr_distance);

	scan_unevictable_unregister_node(node);
	hugetlb_unregister_node(node);		/* no-op, if memoryless node */

	sysdev_unregister(&node->sysdev);
}

struct node node_devices[MAX_NUMNODES];

/*
 * register cpu under node
 */
int register_cpu_under_node(unsigned int cpu, unsigned int nid)
{
	int ret;
	struct sys_device *obj;

	if (!node_online(nid))
		return 0;

	obj = get_cpu_sysdev(cpu);
	if (!obj)
		return 0;

	ret = sysfs_create_link(&node_devices[nid].sysdev.kobj,
				&obj->kobj,
				kobject_name(&obj->kobj));
	if (ret)
		return ret;

	return sysfs_create_link(&obj->kobj,
				 &node_devices[nid].sysdev.kobj,
				 kobject_name(&node_devices[nid].sysdev.kobj));
}

int unregister_cpu_under_node(unsigned int cpu, unsigned int nid)
{
	struct sys_device *obj;

	if (!node_online(nid))
		return 0;

	obj = get_cpu_sysdev(cpu);
	if (!obj)
		return 0;

	sysfs_remove_link(&node_devices[nid].sysdev.kobj,
			  kobject_name(&obj->kobj));
	sysfs_remove_link(&obj->kobj,
			  kobject_name(&node_devices[nid].sysdev.kobj));

	return 0;
}

#ifdef CONFIG_MEMORY_HOTPLUG_SPARSE
#define page_initialized(page)  (page->lru.next)

static int get_nid_for_pfn(unsigned long pfn)
{
	struct page *page;

	if (!pfn_valid_within(pfn))
		return -1;
	page = pfn_to_page(pfn);
	if (!page_initialized(page))
		return -1;
	return pfn_to_nid(pfn);
}

/* register memory section under specified node if it spans that node */
int register_mem_sect_under_node(struct memory_block *mem_blk, int nid)
{
	int ret;
	unsigned long pfn, sect_start_pfn, sect_end_pfn;

	if (!mem_blk)
		return -EFAULT;
	if (!node_online(nid))
		return 0;
	sect_start_pfn = section_nr_to_pfn(mem_blk->phys_index);
	sect_end_pfn = sect_start_pfn + PAGES_PER_SECTION - 1;
	for (pfn = sect_start_pfn; pfn <= sect_end_pfn; pfn++) {
		int page_nid;

		page_nid = get_nid_for_pfn(pfn);
		if (page_nid < 0)
			continue;
		if (page_nid != nid)
			continue;
		ret = sysfs_create_link_nowarn(&node_devices[nid].sysdev.kobj,
					&mem_blk->sysdev.kobj,
					kobject_name(&mem_blk->sysdev.kobj));
		if (ret)
			return ret;

		return sysfs_create_link_nowarn(&mem_blk->sysdev.kobj,
				&node_devices[nid].sysdev.kobj,
				kobject_name(&node_devices[nid].sysdev.kobj));
	}
	/* mem section does not span the specified node */
	return 0;
}

/* unregister memory section under all nodes that it spans */
int unregister_mem_sect_under_nodes(struct memory_block *mem_blk)
{
	NODEMASK_ALLOC(nodemask_t, unlinked_nodes, GFP_KERNEL);
	unsigned long pfn, sect_start_pfn, sect_end_pfn;

	if (!mem_blk) {
		NODEMASK_FREE(unlinked_nodes);
		return -EFAULT;
	}
	if (!unlinked_nodes)
		return -ENOMEM;
	nodes_clear(*unlinked_nodes);
	sect_start_pfn = section_nr_to_pfn(mem_blk->phys_index);
	sect_end_pfn = sect_start_pfn + PAGES_PER_SECTION - 1;
	for (pfn = sect_start_pfn; pfn <= sect_end_pfn; pfn++) {
		int nid;

		nid = get_nid_for_pfn(pfn);
		if (nid < 0)
			continue;
		if (!node_online(nid))
			continue;
		if (node_test_and_set(nid, *unlinked_nodes))
			continue;
		sysfs_remove_link(&node_devices[nid].sysdev.kobj,
			 kobject_name(&mem_blk->sysdev.kobj));
		sysfs_remove_link(&mem_blk->sysdev.kobj,
			 kobject_name(&node_devices[nid].sysdev.kobj));
	}
	NODEMASK_FREE(unlinked_nodes);
	return 0;
}

static int link_mem_sections(int nid)
{
	unsigned long start_pfn = NODE_DATA(nid)->node_start_pfn;
	unsigned long end_pfn = start_pfn + NODE_DATA(nid)->node_spanned_pages;
	unsigned long pfn;
	int err = 0;

	for (pfn = start_pfn; pfn < end_pfn; pfn += PAGES_PER_SECTION) {
		unsigned long section_nr = pfn_to_section_nr(pfn);
		struct mem_section *mem_sect;
		struct memory_block *mem_blk;
		int ret;

		if (!present_section_nr(section_nr))
			continue;
		mem_sect = __nr_to_section(section_nr);
		mem_blk = find_memory_block(mem_sect);
		ret = register_mem_sect_under_node(mem_blk, nid);
		if (!err)
			err = ret;

		/* discard ref obtained in find_memory_block() */
		kobject_put(&mem_blk->sysdev.kobj);
	}
	return err;
}

#ifdef CONFIG_HUGETLBFS
/*
 * Handle per node hstate attribute [un]registration on transistions
 * to/from memoryless state.
 */
static void node_hugetlb_work(struct work_struct *work)
{
	struct node *node = container_of(work, struct node, node_work);

	/*
	 * We only get here when a node transitions to/from memoryless state.
	 * We can detect which transition occurred by examining whether the
	 * node has memory now.  hugetlb_register_node() already check this
	 * so we try to register the attributes.  If that fails, then the
	 * node has transitioned to memoryless, try to unregister the
	 * attributes.
	 */
	if (!hugetlb_register_node(node))
		hugetlb_unregister_node(node);
}

static void init_node_hugetlb_work(int nid)
{
	INIT_WORK(&node_devices[nid].node_work, node_hugetlb_work);
}

static int node_memory_callback(struct notifier_block *self,
				unsigned long action, void *arg)
{
	struct memory_notify *mnb = arg;
	int nid = mnb->status_change_nid;

	switch (action) {
	case MEM_ONLINE:
	case MEM_OFFLINE:
		/*
		 * offload per node hstate [un]registration to a work thread
		 * when transitioning to/from memoryless state.
		 */
		if (nid != NUMA_NO_NODE)
			schedule_work(&node_devices[nid].node_work);
		break;

	case MEM_GOING_ONLINE:
	case MEM_GOING_OFFLINE:
	case MEM_CANCEL_ONLINE:
	case MEM_CANCEL_OFFLINE:
	default:
		break;
	}

	return NOTIFY_OK;
}
#endif	/* CONFIG_HUGETLBFS */
#else	/* !CONFIG_MEMORY_HOTPLUG_SPARSE */

static int link_mem_sections(int nid) { return 0; }
#endif	/* CONFIG_MEMORY_HOTPLUG_SPARSE */

#if !defined(CONFIG_MEMORY_HOTPLUG_SPARSE) || \
    !defined(CONFIG_HUGETLBFS)
static inline int node_memory_callback(struct notifier_block *self,
				unsigned long action, void *arg)
{
	return NOTIFY_OK;
}

static void init_node_hugetlb_work(int nid) { }

#endif

int register_one_node(int nid)
{
	int error = 0;
	int cpu;

	if (node_online(nid)) {
		int p_node = parent_node(nid);
		struct node *parent = NULL;

		if (p_node != nid)
			parent = &node_devices[p_node];

		error = register_node(&node_devices[nid], nid, parent);

		/* link cpu under this node */
		for_each_present_cpu(cpu) {
			if (cpu_to_node(cpu) == nid)
				register_cpu_under_node(cpu, nid);
		}

		/* link memory sections under this node */
		error = link_mem_sections(nid);

		/* initialize work queue for memory hot plug */
		init_node_hugetlb_work(nid);
	}

	return error;

}

void unregister_one_node(int nid)
{
	unregister_node(&node_devices[nid]);
}

/*
 * node states attributes
 */

static ssize_t print_nodes_state(enum node_states state, char *buf)
{
	int n;

	n = nodelist_scnprintf(buf, PAGE_SIZE, node_states[state]);
	if (n > 0 && PAGE_SIZE > n + 1) {
		*(buf + n++) = '\n';
		*(buf + n++) = '\0';
	}
	return n;
}

struct node_attr {
	struct sysdev_class_attribute attr;
	enum node_states state;
};

static ssize_t show_node_state(struct sysdev_class *class,
			       struct sysdev_class_attribute *attr, char *buf)
{
	struct node_attr *na = container_of(attr, struct node_attr, attr);
	return print_nodes_state(na->state, buf);
}

#define _NODE_ATTR(name, state) \
	{ _SYSDEV_CLASS_ATTR(name, 0444, show_node_state, NULL), state }

static struct node_attr node_state_attr[] = {
	_NODE_ATTR(possible, N_POSSIBLE),
	_NODE_ATTR(online, N_ONLINE),
	_NODE_ATTR(has_normal_memory, N_NORMAL_MEMORY),
	_NODE_ATTR(has_cpu, N_CPU),
#ifdef CONFIG_HIGHMEM
	_NODE_ATTR(has_high_memory, N_HIGH_MEMORY),
#endif
};

static struct sysdev_class_attribute *node_state_attrs[] = {
	&node_state_attr[0].attr,
	&node_state_attr[1].attr,
	&node_state_attr[2].attr,
	&node_state_attr[3].attr,
#ifdef CONFIG_HIGHMEM
	&node_state_attr[4].attr,
#endif
	NULL
};

#define NODE_CALLBACK_PRI	2	/* lower than SLAB */
static int __init register_node_type(void)
{
	int ret;

 	BUILD_BUG_ON(ARRAY_SIZE(node_state_attr) != NR_NODE_STATES);
 	BUILD_BUG_ON(ARRAY_SIZE(node_state_attrs)-1 != NR_NODE_STATES);

	ret = sysdev_class_register(&node_class);
	if (!ret) {
		hotplug_memory_notifier(node_memory_callback,
					NODE_CALLBACK_PRI);
	}

	/*
	 * Note:  we're not going to unregister the node class if we fail
	 * to register the node state class attribute files.
	 */
	return ret;
}
postcore_initcall(register_node_type);
Commit	Line	Data
1da177e4 LT	1	/*
	2	* drivers/base/node.c - basic Node class support
	3	*/
	4
	5	#include <linux/sysdev.h>
	6	#include <linux/module.h>
	7	#include <linux/init.h>
	8	#include <linux/mm.h>
c04fc586	9	#include <linux/memory.h>
1da177e4 LT	10	#include <linux/node.h>
	11	#include <linux/hugetlb.h>
	12	#include <linux/cpumask.h>
	13	#include <linux/topology.h>
	14	#include <linux/nodemask.h>
76b67ed9	15	#include <linux/cpu.h>
bde631a5	16	#include <linux/device.h>
af936a16	17	#include <linux/swap.h>
18e5b539	18	#include <linux/slab.h>
1da177e4	19
3701cde6 AK	20	static struct sysdev_class_attribute *node_state_attrs[];
3701cde6 AK	21
1da177e4	22	static struct sysdev_class node_class = {
af5ca3f4	23	.name = "node",
3701cde6	24	.attrs = node_state_attrs,
1da177e4 LT	25	};
	26
	27
39106dcf	28	static ssize_t node_read_cpumap(struct sys_device dev, int type, char buf)
1da177e4 LT	29	{
1da177e4 LT	30	struct node *node_dev = to_node(dev);
a70f7302	31	const struct cpumask *mask = cpumask_of_node(node_dev->sysdev.id);
1da177e4 LT	32	int len;
1da177e4 LT	33
39106dcf MT	34	/* 2008/04/07: buf currently PAGE_SIZE, need 9 chars per 32 bits. */
39106dcf MT	35	BUILD_BUG_ON((NR_CPUS/32 * 9) > (PAGE_SIZE-1));
1da177e4	36
39106dcf	37	len = type?
29c0177e RR	38	cpulist_scnprintf(buf, PAGE_SIZE-2, mask) :
29c0177e RR	39	cpumask_scnprintf(buf, PAGE_SIZE-2, mask);
c5f59f08 MT	40	buf[len++] = '\n';
c5f59f08 MT	41	buf[len] = '\0';
1da177e4 LT	42	return len;
	43	}
	44
4a0b2b4d AK	45	static inline ssize_t node_read_cpumask(struct sys_device *dev,
4a0b2b4d AK	46	struct sysdev_attribute attr, char buf)
39106dcf MT	47	{
	48	return node_read_cpumap(dev, 0, buf);
	49	}
4a0b2b4d AK	50	static inline ssize_t node_read_cpulist(struct sys_device *dev,
4a0b2b4d AK	51	struct sysdev_attribute attr, char buf)
39106dcf MT	52	{
	53	return node_read_cpumap(dev, 1, buf);
	54	}
	55
	56	static SYSDEV_ATTR(cpumap, S_IRUGO, node_read_cpumask, NULL);
	57	static SYSDEV_ATTR(cpulist, S_IRUGO, node_read_cpulist, NULL);
1da177e4 LT	58
1da177e4 LT	59	#define K(x) ((x) << (PAGE_SHIFT - 10))
4a0b2b4d AK	60	static ssize_t node_read_meminfo(struct sys_device * dev,
4a0b2b4d AK	61	struct sysdev_attribute attr, char buf)
1da177e4 LT	62	{
	63	int n;
	64	int nid = dev->id;
	65	struct sysinfo i;
1da177e4 LT	66
1da177e4 LT	67	si_meminfo_node(&i, nid);
c07e02db	68
1da177e4	69	n = sprintf(buf, "\n"
4f98a2fe RR	70	"Node %d MemTotal: %8lu kB\n"
	71	"Node %d MemFree: %8lu kB\n"
	72	"Node %d MemUsed: %8lu kB\n"
	73	"Node %d Active: %8lu kB\n"
	74	"Node %d Inactive: %8lu kB\n"
	75	"Node %d Active(anon): %8lu kB\n"
	76	"Node %d Inactive(anon): %8lu kB\n"
	77	"Node %d Active(file): %8lu kB\n"
	78	"Node %d Inactive(file): %8lu kB\n"
5344b7e6 NP	79	"Node %d Unevictable: %8lu kB\n"
5344b7e6 NP	80	"Node %d Mlocked: %8lu kB\n"
182e8e23	81	#ifdef CONFIG_HIGHMEM
4f98a2fe RR	82	"Node %d HighTotal: %8lu kB\n"
	83	"Node %d HighFree: %8lu kB\n"
	84	"Node %d LowTotal: %8lu kB\n"
	85	"Node %d LowFree: %8lu kB\n"
182e8e23	86	#endif
4f98a2fe RR	87	"Node %d Dirty: %8lu kB\n"
	88	"Node %d Writeback: %8lu kB\n"
	89	"Node %d FilePages: %8lu kB\n"
	90	"Node %d Mapped: %8lu kB\n"
	91	"Node %d AnonPages: %8lu kB\n"
4b02108a	92	"Node %d Shmem: %8lu kB\n"
c6a7f572	93	"Node %d KernelStack: %8lu kB\n"
4f98a2fe RR	94	"Node %d PageTables: %8lu kB\n"
	95	"Node %d NFS_Unstable: %8lu kB\n"
	96	"Node %d Bounce: %8lu kB\n"
	97	"Node %d WritebackTmp: %8lu kB\n"
	98	"Node %d Slab: %8lu kB\n"
	99	"Node %d SReclaimable: %8lu kB\n"
	100	"Node %d SUnreclaim: %8lu kB\n",
1da177e4 LT	101	nid, K(i.totalram),
	102	nid, K(i.freeram),
	103	nid, K(i.totalram - i.freeram),
4f98a2fe RR	104	nid, K(node_page_state(nid, NR_ACTIVE_ANON) +
	105	node_page_state(nid, NR_ACTIVE_FILE)),
	106	nid, K(node_page_state(nid, NR_INACTIVE_ANON) +
	107	node_page_state(nid, NR_INACTIVE_FILE)),
	108	nid, K(node_page_state(nid, NR_ACTIVE_ANON)),
	109	nid, K(node_page_state(nid, NR_INACTIVE_ANON)),
	110	nid, K(node_page_state(nid, NR_ACTIVE_FILE)),
	111	nid, K(node_page_state(nid, NR_INACTIVE_FILE)),
7b854121	112	nid, K(node_page_state(nid, NR_UNEVICTABLE)),
5344b7e6	113	nid, K(node_page_state(nid, NR_MLOCK)),
182e8e23	114	#ifdef CONFIG_HIGHMEM
1da177e4 LT	115	nid, K(i.totalhigh),
	116	nid, K(i.freehigh),
	117	nid, K(i.totalram - i.totalhigh),
c07e02db	118	nid, K(i.freeram - i.freehigh),
182e8e23	119	#endif
b1e7a8fd	120	nid, K(node_page_state(nid, NR_FILE_DIRTY)),
ce866b34	121	nid, K(node_page_state(nid, NR_WRITEBACK)),
347ce434	122	nid, K(node_page_state(nid, NR_FILE_PAGES)),
65ba55f5	123	nid, K(node_page_state(nid, NR_FILE_MAPPED)),
f3dbd344	124	nid, K(node_page_state(nid, NR_ANON_PAGES)),
4b02108a	125	nid, K(node_page_state(nid, NR_SHMEM)),
c6a7f572 KM	126	nid, node_page_state(nid, NR_KERNEL_STACK) *
c6a7f572 KM	127	THREAD_SIZE / 1024,
df849a15	128	nid, K(node_page_state(nid, NR_PAGETABLE)),
fd39fc85	129	nid, K(node_page_state(nid, NR_UNSTABLE_NFS)),
d2c5e30c	130	nid, K(node_page_state(nid, NR_BOUNCE)),
fc3ba692	131	nid, K(node_page_state(nid, NR_WRITEBACK_TEMP)),
972d1a7b CL	132	nid, K(node_page_state(nid, NR_SLAB_RECLAIMABLE) +
	133	node_page_state(nid, NR_SLAB_UNRECLAIMABLE)),
	134	nid, K(node_page_state(nid, NR_SLAB_RECLAIMABLE)),
	135	nid, K(node_page_state(nid, NR_SLAB_UNRECLAIMABLE)));
1da177e4 LT	136	n += hugetlb_report_node_meminfo(nid, buf + n);
	137	return n;
	138	}
	139
	140	#undef K
	141	static SYSDEV_ATTR(meminfo, S_IRUGO, node_read_meminfo, NULL);
	142
4a0b2b4d AK	143	static ssize_t node_read_numastat(struct sys_device * dev,
4a0b2b4d AK	144	struct sysdev_attribute attr, char buf)
1da177e4	145	{
1da177e4 LT	146	return sprintf(buf,
	147	"numa_hit %lu\n"
	148	"numa_miss %lu\n"
	149	"numa_foreign %lu\n"
	150	"interleave_hit %lu\n"
	151	"local_node %lu\n"
	152	"other_node %lu\n",
ca889e6c CL	153	node_page_state(dev->id, NUMA_HIT),
	154	node_page_state(dev->id, NUMA_MISS),
	155	node_page_state(dev->id, NUMA_FOREIGN),
	156	node_page_state(dev->id, NUMA_INTERLEAVE_HIT),
	157	node_page_state(dev->id, NUMA_LOCAL),
	158	node_page_state(dev->id, NUMA_OTHER));
1da177e4 LT	159	}
	160	static SYSDEV_ATTR(numastat, S_IRUGO, node_read_numastat, NULL);
	161
4a0b2b4d AK	162	static ssize_t node_read_distance(struct sys_device * dev,
4a0b2b4d AK	163	struct sysdev_attribute attr, char buf)
1da177e4 LT	164	{
	165	int nid = dev->id;
	166	int len = 0;
	167	int i;
	168
12ee3c0a DR	169	/*
	170	* buf is currently PAGE_SIZE in length and each node needs 4 chars
	171	* at the most (distance + space or newline).
	172	*/
	173	BUILD_BUG_ON(MAX_NUMNODES * 4 > PAGE_SIZE);
1da177e4 LT	174
	175	for_each_online_node(i)
	176	len += sprintf(buf + len, "%s%d", i ? " " : "", node_distance(nid, i));
	177
	178	len += sprintf(buf + len, "\n");
	179	return len;
	180	}
	181	static SYSDEV_ATTR(distance, S_IRUGO, node_read_distance, NULL);
	182
9a305230 LS	183	#ifdef CONFIG_HUGETLBFS
	184	/*
	185	* hugetlbfs per node attributes registration interface:
	186	* When/if hugetlb[fs] subsystem initializes [sometime after this module],
4faf8d95 LS	187	* it will register its per node attributes for all online nodes with
4faf8d95 LS	188	* memory. It will also call register_hugetlbfs_with_node(), below, to
9a305230 LS	189	* register its attribute registration functions with this node driver.
	190	* Once these hooks have been initialized, the node driver will call into
	191	* the hugetlb module to [un]register attributes for hot-plugged nodes.
	192	*/
	193	static node_registration_func_t __hugetlb_register_node;
	194	static node_registration_func_t __hugetlb_unregister_node;
	195
39da08cb	196	static inline bool hugetlb_register_node(struct node *node)
9a305230	197	{
4faf8d95	198	if (__hugetlb_register_node &&
39da08cb	199	node_state(node->sysdev.id, N_HIGH_MEMORY)) {
9a305230	200	__hugetlb_register_node(node);
39da08cb LS	201	return true;
	202	}
	203	return false;
9a305230 LS	204	}
	205
	206	static inline void hugetlb_unregister_node(struct node *node)
	207	{
	208	if (__hugetlb_unregister_node)
	209	__hugetlb_unregister_node(node);
	210	}
	211
	212	void register_hugetlbfs_with_node(node_registration_func_t doregister,
	213	node_registration_func_t unregister)
	214	{
	215	__hugetlb_register_node = doregister;
	216	__hugetlb_unregister_node = unregister;
	217	}
	218	#else
	219	static inline void hugetlb_register_node(struct node *node) {}
	220
	221	static inline void hugetlb_unregister_node(struct node *node) {}
	222	#endif
	223
1da177e4 LT	224
1da177e4 LT	225	/*
405ae7d3	226	* register_node - Setup a sysfs device for a node.
1da177e4 LT	227	* @num - Node number to use when creating the device.
	228	*
	229	* Initialize and register the node device.
	230	*/
4b45099b	231	int register_node(struct node node, int num, struct node parent)
1da177e4 LT	232	{
	233	int error;
	234
	235	node->sysdev.id = num;
	236	node->sysdev.cls = &node_class;
	237	error = sysdev_register(&node->sysdev);
	238
	239	if (!error){
	240	sysdev_create_file(&node->sysdev, &attr_cpumap);
39106dcf	241	sysdev_create_file(&node->sysdev, &attr_cpulist);
1da177e4 LT	242	sysdev_create_file(&node->sysdev, &attr_meminfo);
	243	sysdev_create_file(&node->sysdev, &attr_numastat);
	244	sysdev_create_file(&node->sysdev, &attr_distance);
af936a16 LS	245
af936a16 LS	246	scan_unevictable_register_node(node);
4faf8d95	247
9a305230	248	hugetlb_register_node(node);
1da177e4 LT	249	}
	250	return error;
	251	}
	252
4b45099b KT	253	/**
	254	* unregister_node - unregister a node device
	255	* @node: node going away
	256	*
	257	* Unregisters a node device @node. All the devices on the node must be
	258	* unregistered before calling this function.
	259	*/
	260	void unregister_node(struct node *node)
	261	{
	262	sysdev_remove_file(&node->sysdev, &attr_cpumap);
39106dcf	263	sysdev_remove_file(&node->sysdev, &attr_cpulist);
4b45099b KT	264	sysdev_remove_file(&node->sysdev, &attr_meminfo);
	265	sysdev_remove_file(&node->sysdev, &attr_numastat);
	266	sysdev_remove_file(&node->sysdev, &attr_distance);
	267
af936a16	268	scan_unevictable_unregister_node(node);
4faf8d95	269	hugetlb_unregister_node(node); /* no-op, if memoryless node */
af936a16	270
4b45099b KT	271	sysdev_unregister(&node->sysdev);
4b45099b KT	272	}
1da177e4	273
0fc44159 YG	274	struct node node_devices[MAX_NUMNODES];
0fc44159 YG	275
76b67ed9 KH	276	/*
	277	* register cpu under node
	278	*/
	279	int register_cpu_under_node(unsigned int cpu, unsigned int nid)
	280	{
1830794a	281	int ret;
f8246f31	282	struct sys_device *obj;
76b67ed9	283
f8246f31 AC	284	if (!node_online(nid))
	285	return 0;
	286
	287	obj = get_cpu_sysdev(cpu);
	288	if (!obj)
	289	return 0;
	290
1830794a	291	ret = sysfs_create_link(&node_devices[nid].sysdev.kobj,
f8246f31 AC	292	&obj->kobj,
f8246f31 AC	293	kobject_name(&obj->kobj));
1830794a AC	294	if (ret)
	295	return ret;
	296
	297	return sysfs_create_link(&obj->kobj,
	298	&node_devices[nid].sysdev.kobj,
	299	kobject_name(&node_devices[nid].sysdev.kobj));
76b67ed9 KH	300	}
	301
	302	int unregister_cpu_under_node(unsigned int cpu, unsigned int nid)
	303	{
b9d52dad AC	304	struct sys_device *obj;
	305
	306	if (!node_online(nid))
	307	return 0;
	308
	309	obj = get_cpu_sysdev(cpu);
	310	if (!obj)
	311	return 0;
	312
	313	sysfs_remove_link(&node_devices[nid].sysdev.kobj,
	314	kobject_name(&obj->kobj));
1830794a AC	315	sysfs_remove_link(&obj->kobj,
1830794a AC	316	kobject_name(&node_devices[nid].sysdev.kobj));
b9d52dad	317
76b67ed9 KH	318	return 0;
	319	}
	320
c04fc586 GH	321	#ifdef CONFIG_MEMORY_HOTPLUG_SPARSE
	322	#define page_initialized(page) (page->lru.next)
	323
	324	static int get_nid_for_pfn(unsigned long pfn)
	325	{
	326	struct page *page;
	327
	328	if (!pfn_valid_within(pfn))
	329	return -1;
	330	page = pfn_to_page(pfn);
	331	if (!page_initialized(page))
	332	return -1;
	333	return pfn_to_nid(pfn);
	334	}
	335
	336	/* register memory section under specified node if it spans that node */
	337	int register_mem_sect_under_node(struct memory_block *mem_blk, int nid)
	338	{
dee5d0d5	339	int ret;
c04fc586 GH	340	unsigned long pfn, sect_start_pfn, sect_end_pfn;
	341
	342	if (!mem_blk)
	343	return -EFAULT;
	344	if (!node_online(nid))
	345	return 0;
	346	sect_start_pfn = section_nr_to_pfn(mem_blk->phys_index);
	347	sect_end_pfn = sect_start_pfn + PAGES_PER_SECTION - 1;
	348	for (pfn = sect_start_pfn; pfn <= sect_end_pfn; pfn++) {
	349	int page_nid;
	350
	351	page_nid = get_nid_for_pfn(pfn);
	352	if (page_nid < 0)
	353	continue;
	354	if (page_nid != nid)
	355	continue;
dee5d0d5	356	ret = sysfs_create_link_nowarn(&node_devices[nid].sysdev.kobj,
c04fc586 GH	357	&mem_blk->sysdev.kobj,
c04fc586 GH	358	kobject_name(&mem_blk->sysdev.kobj));
dee5d0d5 AC	359	if (ret)
	360	return ret;
	361
	362	return sysfs_create_link_nowarn(&mem_blk->sysdev.kobj,
	363	&node_devices[nid].sysdev.kobj,
	364	kobject_name(&node_devices[nid].sysdev.kobj));
c04fc586 GH	365	}
	366	/* mem section does not span the specified node */
	367	return 0;
	368	}
	369
	370	/* unregister memory section under all nodes that it spans */
	371	int unregister_mem_sect_under_nodes(struct memory_block *mem_blk)
	372	{
9ae49fab	373	NODEMASK_ALLOC(nodemask_t, unlinked_nodes, GFP_KERNEL);
c04fc586 GH	374	unsigned long pfn, sect_start_pfn, sect_end_pfn;
c04fc586 GH	375
9ae49fab DR	376	if (!mem_blk) {
9ae49fab DR	377	NODEMASK_FREE(unlinked_nodes);
c04fc586	378	return -EFAULT;
9ae49fab DR	379	}
	380	if (!unlinked_nodes)
	381	return -ENOMEM;
	382	nodes_clear(*unlinked_nodes);
c04fc586 GH	383	sect_start_pfn = section_nr_to_pfn(mem_blk->phys_index);
	384	sect_end_pfn = sect_start_pfn + PAGES_PER_SECTION - 1;
	385	for (pfn = sect_start_pfn; pfn <= sect_end_pfn; pfn++) {
47504980	386	int nid;
c04fc586 GH	387
	388	nid = get_nid_for_pfn(pfn);
	389	if (nid < 0)
	390	continue;
	391	if (!node_online(nid))
	392	continue;
9ae49fab	393	if (node_test_and_set(nid, *unlinked_nodes))
c04fc586 GH	394	continue;
	395	sysfs_remove_link(&node_devices[nid].sysdev.kobj,
	396	kobject_name(&mem_blk->sysdev.kobj));
dee5d0d5 AC	397	sysfs_remove_link(&mem_blk->sysdev.kobj,
dee5d0d5 AC	398	kobject_name(&node_devices[nid].sysdev.kobj));
c04fc586	399	}
9ae49fab	400	NODEMASK_FREE(unlinked_nodes);
c04fc586 GH	401	return 0;
	402	}
	403
	404	static int link_mem_sections(int nid)
	405	{
	406	unsigned long start_pfn = NODE_DATA(nid)->node_start_pfn;
	407	unsigned long end_pfn = start_pfn + NODE_DATA(nid)->node_spanned_pages;
	408	unsigned long pfn;
	409	int err = 0;
	410
	411	for (pfn = start_pfn; pfn < end_pfn; pfn += PAGES_PER_SECTION) {
	412	unsigned long section_nr = pfn_to_section_nr(pfn);
	413	struct mem_section *mem_sect;
	414	struct memory_block *mem_blk;
	415	int ret;
	416
	417	if (!present_section_nr(section_nr))
	418	continue;
	419	mem_sect = __nr_to_section(section_nr);
	420	mem_blk = find_memory_block(mem_sect);
	421	ret = register_mem_sect_under_node(mem_blk, nid);
	422	if (!err)
	423	err = ret;
	424
	425	/* discard ref obtained in find_memory_block() */
	426	kobject_put(&mem_blk->sysdev.kobj);
	427	}
	428	return err;
	429	}
4faf8d95	430
39da08cb	431	#ifdef CONFIG_HUGETLBFS
4faf8d95 LS	432	/*
	433	* Handle per node hstate attribute [un]registration on transistions
	434	* to/from memoryless state.
	435	*/
39da08cb LS	436	static void node_hugetlb_work(struct work_struct *work)
	437	{
	438	struct node *node = container_of(work, struct node, node_work);
	439
	440	/*
	441	* We only get here when a node transitions to/from memoryless state.
	442	* We can detect which transition occurred by examining whether the
	443	* node has memory now. hugetlb_register_node() already check this
	444	* so we try to register the attributes. If that fails, then the
	445	* node has transitioned to memoryless, try to unregister the
	446	* attributes.
	447	*/
	448	if (!hugetlb_register_node(node))
	449	hugetlb_unregister_node(node);
	450	}
	451
	452	static void init_node_hugetlb_work(int nid)
	453	{
	454	INIT_WORK(&node_devices[nid].node_work, node_hugetlb_work);
	455	}
4faf8d95 LS	456
	457	static int node_memory_callback(struct notifier_block *self,
	458	unsigned long action, void *arg)
	459	{
	460	struct memory_notify *mnb = arg;
	461	int nid = mnb->status_change_nid;
	462
	463	switch (action) {
39da08cb LS	464	case MEM_ONLINE:
	465	case MEM_OFFLINE:
	466	/*
	467	* offload per node hstate [un]registration to a work thread
	468	* when transitioning to/from memoryless state.
	469	*/
4faf8d95	470	if (nid != NUMA_NO_NODE)
39da08cb	471	schedule_work(&node_devices[nid].node_work);
4faf8d95	472	break;
39da08cb	473
4faf8d95 LS	474	case MEM_GOING_ONLINE:
	475	case MEM_GOING_OFFLINE:
	476	case MEM_CANCEL_ONLINE:
	477	case MEM_CANCEL_OFFLINE:
	478	default:
	479	break;
	480	}
	481
	482	return NOTIFY_OK;
	483	}
39da08cb LS	484	#endif /* CONFIG_HUGETLBFS */
	485	#else /* !CONFIG_MEMORY_HOTPLUG_SPARSE */
	486
c04fc586	487	static int link_mem_sections(int nid) { return 0; }
39da08cb	488	#endif /* CONFIG_MEMORY_HOTPLUG_SPARSE */
4faf8d95	489
39da08cb LS	490	#if !defined(CONFIG_MEMORY_HOTPLUG_SPARSE) \|\| \
39da08cb LS	491	!defined(CONFIG_HUGETLBFS)
4faf8d95 LS	492	static inline int node_memory_callback(struct notifier_block *self,
	493	unsigned long action, void *arg)
	494	{
	495	return NOTIFY_OK;
	496	}
39da08cb LS	497
	498	static void init_node_hugetlb_work(int nid) { }
	499
	500	#endif
c04fc586	501
0fc44159 YG	502	int register_one_node(int nid)
	503	{
	504	int error = 0;
76b67ed9	505	int cpu;
0fc44159 YG	506
	507	if (node_online(nid)) {
	508	int p_node = parent_node(nid);
	509	struct node *parent = NULL;
	510
	511	if (p_node != nid)
	512	parent = &node_devices[p_node];
	513
	514	error = register_node(&node_devices[nid], nid, parent);
76b67ed9 KH	515
	516	/* link cpu under this node */
	517	for_each_present_cpu(cpu) {
	518	if (cpu_to_node(cpu) == nid)
	519	register_cpu_under_node(cpu, nid);
	520	}
c04fc586 GH	521
	522	/* link memory sections under this node */
	523	error = link_mem_sections(nid);
39da08cb LS	524
	525	/* initialize work queue for memory hot plug */
	526	init_node_hugetlb_work(nid);
0fc44159 YG	527	}
	528
	529	return error;
	530
	531	}
	532
	533	void unregister_one_node(int nid)
	534	{
	535	unregister_node(&node_devices[nid]);
	536	}
	537
bde631a5 LS	538	/*
	539	* node states attributes
	540	*/
	541
	542	static ssize_t print_nodes_state(enum node_states state, char *buf)
	543	{
	544	int n;
	545
	546	n = nodelist_scnprintf(buf, PAGE_SIZE, node_states[state]);
	547	if (n > 0 && PAGE_SIZE > n + 1) {
	548	*(buf + n++) = '\n';
	549	*(buf + n++) = '\0';
	550	}
	551	return n;
	552	}
	553
b15f562f AK	554	struct node_attr {
	555	struct sysdev_class_attribute attr;
	556	enum node_states state;
	557	};
bde631a5	558
b15f562f AK	559	static ssize_t show_node_state(struct sysdev_class *class,
b15f562f AK	560	struct sysdev_class_attribute attr, char buf)
bde631a5	561	{
b15f562f AK	562	struct node_attr *na = container_of(attr, struct node_attr, attr);
b15f562f AK	563	return print_nodes_state(na->state, buf);
bde631a5 LS	564	}
bde631a5 LS	565
b15f562f AK	566	#define _NODE_ATTR(name, state) \
b15f562f AK	567	{ _SYSDEV_CLASS_ATTR(name, 0444, show_node_state, NULL), state }
bde631a5	568
b15f562f AK	569	static struct node_attr node_state_attr[] = {
	570	_NODE_ATTR(possible, N_POSSIBLE),
	571	_NODE_ATTR(online, N_ONLINE),
	572	_NODE_ATTR(has_normal_memory, N_NORMAL_MEMORY),
	573	_NODE_ATTR(has_cpu, N_CPU),
bde631a5	574	#ifdef CONFIG_HIGHMEM
b15f562f	575	_NODE_ATTR(has_high_memory, N_HIGH_MEMORY),
bde631a5	576	#endif
bde631a5 LS	577	};
bde631a5 LS	578
3701cde6 AK	579	static struct sysdev_class_attribute *node_state_attrs[] = {
	580	&node_state_attr[0].attr,
	581	&node_state_attr[1].attr,
	582	&node_state_attr[2].attr,
	583	&node_state_attr[3].attr,
	584	#ifdef CONFIG_HIGHMEM
	585	&node_state_attr[4].attr,
	586	#endif
	587	NULL
	588	};
bde631a5	589
4faf8d95	590	#define NODE_CALLBACK_PRI 2 /* lower than SLAB */
4b45099b	591	static int __init register_node_type(void)
1da177e4	592	{
bde631a5 LS	593	int ret;
bde631a5 LS	594
3701cde6 AK	595	BUILD_BUG_ON(ARRAY_SIZE(node_state_attr) != NR_NODE_STATES);
	596	BUILD_BUG_ON(ARRAY_SIZE(node_state_attrs)-1 != NR_NODE_STATES);
	597
bde631a5	598	ret = sysdev_class_register(&node_class);
4faf8d95	599	if (!ret) {
4faf8d95 LS	600	hotplug_memory_notifier(node_memory_callback,
	601	NODE_CALLBACK_PRI);
	602	}
bde631a5 LS	603
	604	/*
	605	* Note: we're not going to unregister the node class if we fail
	606	* to register the node state class attribute files.
	607	*/
	608	return ret;
1da177e4 LT	609	}
1da177e4 LT	610	postcore_initcall(register_node_type);