[net-next-2.6.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/compaction.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,
		       "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",
		       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
	n += sprintf(buf + n,
		       "Node %d HighTotal:      %8lu kB\n"
		       "Node %d HighFree:       %8lu kB\n"
		       "Node %d LowTotal:       %8lu kB\n"
		       "Node %d LowFree:        %8lu kB\n",
		       nid, K(i.totalhigh),
		       nid, K(i.freehigh),
		       nid, K(i.totalram - i.totalhigh),
		       nid, K(i.freeram - i.freehigh));
#endif
	n += sprintf(buf + n,
		       "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(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);

		compaction_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>
1da177e4 LT	11	#include <linux/hugetlb.h>
ed4a6d7f	12	#include <linux/compaction.h>
1da177e4 LT	13	#include <linux/cpumask.h>
	14	#include <linux/topology.h>
	15	#include <linux/nodemask.h>
76b67ed9	16	#include <linux/cpu.h>
bde631a5	17	#include <linux/device.h>
af936a16	18	#include <linux/swap.h>
18e5b539	19	#include <linux/slab.h>
1da177e4	20
3701cde6 AK	21	static struct sysdev_class_attribute *node_state_attrs[];
3701cde6 AK	22
1da177e4	23	static struct sysdev_class node_class = {
af5ca3f4	24	.name = "node",
3701cde6	25	.attrs = node_state_attrs,
1da177e4 LT	26	};
	27
	28
39106dcf	29	static ssize_t node_read_cpumap(struct sys_device dev, int type, char buf)
1da177e4 LT	30	{
1da177e4 LT	31	struct node *node_dev = to_node(dev);
a70f7302	32	const struct cpumask *mask = cpumask_of_node(node_dev->sysdev.id);
1da177e4 LT	33	int len;
1da177e4 LT	34
39106dcf MT	35	/* 2008/04/07: buf currently PAGE_SIZE, need 9 chars per 32 bits. */
39106dcf MT	36	BUILD_BUG_ON((NR_CPUS/32 * 9) > (PAGE_SIZE-1));
1da177e4	37
39106dcf	38	len = type?
29c0177e RR	39	cpulist_scnprintf(buf, PAGE_SIZE-2, mask) :
29c0177e RR	40	cpumask_scnprintf(buf, PAGE_SIZE-2, mask);
c5f59f08 MT	41	buf[len++] = '\n';
c5f59f08 MT	42	buf[len] = '\0';
1da177e4 LT	43	return len;
	44	}
	45
4a0b2b4d AK	46	static inline ssize_t node_read_cpumask(struct sys_device *dev,
4a0b2b4d AK	47	struct sysdev_attribute attr, char buf)
39106dcf MT	48	{
	49	return node_read_cpumap(dev, 0, buf);
	50	}
4a0b2b4d AK	51	static inline ssize_t node_read_cpulist(struct sys_device *dev,
4a0b2b4d AK	52	struct sysdev_attribute attr, char buf)
39106dcf MT	53	{
	54	return node_read_cpumap(dev, 1, buf);
	55	}
	56
	57	static SYSDEV_ATTR(cpumap, S_IRUGO, node_read_cpumask, NULL);
	58	static SYSDEV_ATTR(cpulist, S_IRUGO, node_read_cpulist, NULL);
1da177e4 LT	59
1da177e4 LT	60	#define K(x) ((x) << (PAGE_SHIFT - 10))
4a0b2b4d AK	61	static ssize_t node_read_meminfo(struct sys_device * dev,
4a0b2b4d AK	62	struct sysdev_attribute attr, char buf)
1da177e4 LT	63	{
	64	int n;
	65	int nid = dev->id;
	66	struct sysinfo i;
1da177e4 LT	67
1da177e4 LT	68	si_meminfo_node(&i, nid);
7ee92255	69	n = sprintf(buf,
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	79	"Node %d Unevictable: %8lu kB\n"
7ee92255 KM	80	"Node %d Mlocked: %8lu kB\n",
	81	nid, K(i.totalram),
	82	nid, K(i.freeram),
	83	nid, K(i.totalram - i.freeram),
	84	nid, K(node_page_state(nid, NR_ACTIVE_ANON) +
	85	node_page_state(nid, NR_ACTIVE_FILE)),
	86	nid, K(node_page_state(nid, NR_INACTIVE_ANON) +
	87	node_page_state(nid, NR_INACTIVE_FILE)),
	88	nid, K(node_page_state(nid, NR_ACTIVE_ANON)),
	89	nid, K(node_page_state(nid, NR_INACTIVE_ANON)),
	90	nid, K(node_page_state(nid, NR_ACTIVE_FILE)),
	91	nid, K(node_page_state(nid, NR_INACTIVE_FILE)),
	92	nid, K(node_page_state(nid, NR_UNEVICTABLE)),
	93	nid, K(node_page_state(nid, NR_MLOCK)));
	94
182e8e23	95	#ifdef CONFIG_HIGHMEM
7ee92255	96	n += sprintf(buf + n,
4f98a2fe RR	97	"Node %d HighTotal: %8lu kB\n"
	98	"Node %d HighFree: %8lu kB\n"
	99	"Node %d LowTotal: %8lu kB\n"
7ee92255 KM	100	"Node %d LowFree: %8lu kB\n",
	101	nid, K(i.totalhigh),
	102	nid, K(i.freehigh),
	103	nid, K(i.totalram - i.totalhigh),
	104	nid, K(i.freeram - i.freehigh));
182e8e23	105	#endif
7ee92255	106	n += sprintf(buf + n,
4f98a2fe RR	107	"Node %d Dirty: %8lu kB\n"
	108	"Node %d Writeback: %8lu kB\n"
	109	"Node %d FilePages: %8lu kB\n"
	110	"Node %d Mapped: %8lu kB\n"
	111	"Node %d AnonPages: %8lu kB\n"
4b02108a	112	"Node %d Shmem: %8lu kB\n"
c6a7f572	113	"Node %d KernelStack: %8lu kB\n"
4f98a2fe RR	114	"Node %d PageTables: %8lu kB\n"
	115	"Node %d NFS_Unstable: %8lu kB\n"
	116	"Node %d Bounce: %8lu kB\n"
	117	"Node %d WritebackTmp: %8lu kB\n"
	118	"Node %d Slab: %8lu kB\n"
	119	"Node %d SReclaimable: %8lu kB\n"
	120	"Node %d SUnreclaim: %8lu kB\n",
b1e7a8fd	121	nid, K(node_page_state(nid, NR_FILE_DIRTY)),
ce866b34	122	nid, K(node_page_state(nid, NR_WRITEBACK)),
347ce434	123	nid, K(node_page_state(nid, NR_FILE_PAGES)),
65ba55f5	124	nid, K(node_page_state(nid, NR_FILE_MAPPED)),
f3dbd344	125	nid, K(node_page_state(nid, NR_ANON_PAGES)),
4b02108a	126	nid, K(node_page_state(nid, NR_SHMEM)),
c6a7f572 KM	127	nid, node_page_state(nid, NR_KERNEL_STACK) *
c6a7f572 KM	128	THREAD_SIZE / 1024,
df849a15	129	nid, K(node_page_state(nid, NR_PAGETABLE)),
fd39fc85	130	nid, K(node_page_state(nid, NR_UNSTABLE_NFS)),
d2c5e30c	131	nid, K(node_page_state(nid, NR_BOUNCE)),
fc3ba692	132	nid, K(node_page_state(nid, NR_WRITEBACK_TEMP)),
972d1a7b CL	133	nid, K(node_page_state(nid, NR_SLAB_RECLAIMABLE) +
	134	node_page_state(nid, NR_SLAB_UNRECLAIMABLE)),
	135	nid, K(node_page_state(nid, NR_SLAB_RECLAIMABLE)),
	136	nid, K(node_page_state(nid, NR_SLAB_UNRECLAIMABLE)));
1da177e4 LT	137	n += hugetlb_report_node_meminfo(nid, buf + n);
	138	return n;
	139	}
	140
	141	#undef K
	142	static SYSDEV_ATTR(meminfo, S_IRUGO, node_read_meminfo, NULL);
	143
4a0b2b4d AK	144	static ssize_t node_read_numastat(struct sys_device * dev,
4a0b2b4d AK	145	struct sysdev_attribute attr, char buf)
1da177e4	146	{
1da177e4 LT	147	return sprintf(buf,
	148	"numa_hit %lu\n"
	149	"numa_miss %lu\n"
	150	"numa_foreign %lu\n"
	151	"interleave_hit %lu\n"
	152	"local_node %lu\n"
	153	"other_node %lu\n",
ca889e6c CL	154	node_page_state(dev->id, NUMA_HIT),
	155	node_page_state(dev->id, NUMA_MISS),
	156	node_page_state(dev->id, NUMA_FOREIGN),
	157	node_page_state(dev->id, NUMA_INTERLEAVE_HIT),
	158	node_page_state(dev->id, NUMA_LOCAL),
	159	node_page_state(dev->id, NUMA_OTHER));
1da177e4 LT	160	}
	161	static SYSDEV_ATTR(numastat, S_IRUGO, node_read_numastat, NULL);
	162
4a0b2b4d AK	163	static ssize_t node_read_distance(struct sys_device * dev,
4a0b2b4d AK	164	struct sysdev_attribute attr, char buf)
1da177e4 LT	165	{
	166	int nid = dev->id;
	167	int len = 0;
	168	int i;
	169
12ee3c0a DR	170	/*
	171	* buf is currently PAGE_SIZE in length and each node needs 4 chars
	172	* at the most (distance + space or newline).
	173	*/
	174	BUILD_BUG_ON(MAX_NUMNODES * 4 > PAGE_SIZE);
1da177e4 LT	175
	176	for_each_online_node(i)
	177	len += sprintf(buf + len, "%s%d", i ? " " : "", node_distance(nid, i));
	178
	179	len += sprintf(buf + len, "\n");
	180	return len;
	181	}
	182	static SYSDEV_ATTR(distance, S_IRUGO, node_read_distance, NULL);
	183
9a305230 LS	184	#ifdef CONFIG_HUGETLBFS
	185	/*
	186	* hugetlbfs per node attributes registration interface:
	187	* When/if hugetlb[fs] subsystem initializes [sometime after this module],
4faf8d95 LS	188	* it will register its per node attributes for all online nodes with
4faf8d95 LS	189	* memory. It will also call register_hugetlbfs_with_node(), below, to
9a305230 LS	190	* register its attribute registration functions with this node driver.
	191	* Once these hooks have been initialized, the node driver will call into
	192	* the hugetlb module to [un]register attributes for hot-plugged nodes.
	193	*/
	194	static node_registration_func_t __hugetlb_register_node;
	195	static node_registration_func_t __hugetlb_unregister_node;
	196
39da08cb	197	static inline bool hugetlb_register_node(struct node *node)
9a305230	198	{
4faf8d95	199	if (__hugetlb_register_node &&
39da08cb	200	node_state(node->sysdev.id, N_HIGH_MEMORY)) {
9a305230	201	__hugetlb_register_node(node);
39da08cb LS	202	return true;
	203	}
	204	return false;
9a305230 LS	205	}
	206
	207	static inline void hugetlb_unregister_node(struct node *node)
	208	{
	209	if (__hugetlb_unregister_node)
	210	__hugetlb_unregister_node(node);
	211	}
	212
	213	void register_hugetlbfs_with_node(node_registration_func_t doregister,
	214	node_registration_func_t unregister)
	215	{
	216	__hugetlb_register_node = doregister;
	217	__hugetlb_unregister_node = unregister;
	218	}
	219	#else
	220	static inline void hugetlb_register_node(struct node *node) {}
	221
	222	static inline void hugetlb_unregister_node(struct node *node) {}
	223	#endif
	224
1da177e4 LT	225
1da177e4 LT	226	/*
405ae7d3	227	* register_node - Setup a sysfs device for a node.
1da177e4 LT	228	* @num - Node number to use when creating the device.
	229	*
	230	* Initialize and register the node device.
	231	*/
4b45099b	232	int register_node(struct node node, int num, struct node parent)
1da177e4 LT	233	{
	234	int error;
	235
	236	node->sysdev.id = num;
	237	node->sysdev.cls = &node_class;
	238	error = sysdev_register(&node->sysdev);
	239
	240	if (!error){
	241	sysdev_create_file(&node->sysdev, &attr_cpumap);
39106dcf	242	sysdev_create_file(&node->sysdev, &attr_cpulist);
1da177e4 LT	243	sysdev_create_file(&node->sysdev, &attr_meminfo);
	244	sysdev_create_file(&node->sysdev, &attr_numastat);
	245	sysdev_create_file(&node->sysdev, &attr_distance);
af936a16 LS	246
af936a16 LS	247	scan_unevictable_register_node(node);
4faf8d95	248
9a305230	249	hugetlb_register_node(node);
ed4a6d7f MG	250
ed4a6d7f MG	251	compaction_register_node(node);
1da177e4 LT	252	}
	253	return error;
	254	}
	255
4b45099b KT	256	/**
	257	* unregister_node - unregister a node device
	258	* @node: node going away
	259	*
	260	* Unregisters a node device @node. All the devices on the node must be
	261	* unregistered before calling this function.
	262	*/
	263	void unregister_node(struct node *node)
	264	{
	265	sysdev_remove_file(&node->sysdev, &attr_cpumap);
39106dcf	266	sysdev_remove_file(&node->sysdev, &attr_cpulist);
4b45099b KT	267	sysdev_remove_file(&node->sysdev, &attr_meminfo);
	268	sysdev_remove_file(&node->sysdev, &attr_numastat);
	269	sysdev_remove_file(&node->sysdev, &attr_distance);
	270
af936a16	271	scan_unevictable_unregister_node(node);
4faf8d95	272	hugetlb_unregister_node(node); /* no-op, if memoryless node */
af936a16	273
4b45099b KT	274	sysdev_unregister(&node->sysdev);
4b45099b KT	275	}
1da177e4	276
0fc44159 YG	277	struct node node_devices[MAX_NUMNODES];
0fc44159 YG	278
76b67ed9 KH	279	/*
	280	* register cpu under node
	281	*/
	282	int register_cpu_under_node(unsigned int cpu, unsigned int nid)
	283	{
1830794a	284	int ret;
f8246f31	285	struct sys_device *obj;
76b67ed9	286
f8246f31 AC	287	if (!node_online(nid))
	288	return 0;
	289
	290	obj = get_cpu_sysdev(cpu);
	291	if (!obj)
	292	return 0;
	293
1830794a	294	ret = sysfs_create_link(&node_devices[nid].sysdev.kobj,
f8246f31 AC	295	&obj->kobj,
f8246f31 AC	296	kobject_name(&obj->kobj));
1830794a AC	297	if (ret)
	298	return ret;
	299
	300	return sysfs_create_link(&obj->kobj,
	301	&node_devices[nid].sysdev.kobj,
	302	kobject_name(&node_devices[nid].sysdev.kobj));
76b67ed9 KH	303	}
	304
	305	int unregister_cpu_under_node(unsigned int cpu, unsigned int nid)
	306	{
b9d52dad AC	307	struct sys_device *obj;
	308
	309	if (!node_online(nid))
	310	return 0;
	311
	312	obj = get_cpu_sysdev(cpu);
	313	if (!obj)
	314	return 0;
	315
	316	sysfs_remove_link(&node_devices[nid].sysdev.kobj,
	317	kobject_name(&obj->kobj));
1830794a AC	318	sysfs_remove_link(&obj->kobj,
1830794a AC	319	kobject_name(&node_devices[nid].sysdev.kobj));
b9d52dad	320
76b67ed9 KH	321	return 0;
	322	}
	323
c04fc586 GH	324	#ifdef CONFIG_MEMORY_HOTPLUG_SPARSE
	325	#define page_initialized(page) (page->lru.next)
	326
	327	static int get_nid_for_pfn(unsigned long pfn)
	328	{
	329	struct page *page;
	330
	331	if (!pfn_valid_within(pfn))
	332	return -1;
	333	page = pfn_to_page(pfn);
	334	if (!page_initialized(page))
	335	return -1;
	336	return pfn_to_nid(pfn);
	337	}
	338
	339	/* register memory section under specified node if it spans that node */
	340	int register_mem_sect_under_node(struct memory_block *mem_blk, int nid)
	341	{
dee5d0d5	342	int ret;
c04fc586 GH	343	unsigned long pfn, sect_start_pfn, sect_end_pfn;
	344
	345	if (!mem_blk)
	346	return -EFAULT;
	347	if (!node_online(nid))
	348	return 0;
	349	sect_start_pfn = section_nr_to_pfn(mem_blk->phys_index);
	350	sect_end_pfn = sect_start_pfn + PAGES_PER_SECTION - 1;
	351	for (pfn = sect_start_pfn; pfn <= sect_end_pfn; pfn++) {
	352	int page_nid;
	353
	354	page_nid = get_nid_for_pfn(pfn);
	355	if (page_nid < 0)
	356	continue;
	357	if (page_nid != nid)
	358	continue;
dee5d0d5	359	ret = sysfs_create_link_nowarn(&node_devices[nid].sysdev.kobj,
c04fc586 GH	360	&mem_blk->sysdev.kobj,
c04fc586 GH	361	kobject_name(&mem_blk->sysdev.kobj));
dee5d0d5 AC	362	if (ret)
	363	return ret;
	364
	365	return sysfs_create_link_nowarn(&mem_blk->sysdev.kobj,
	366	&node_devices[nid].sysdev.kobj,
	367	kobject_name(&node_devices[nid].sysdev.kobj));
c04fc586 GH	368	}
	369	/* mem section does not span the specified node */
	370	return 0;
	371	}
	372
	373	/* unregister memory section under all nodes that it spans */
	374	int unregister_mem_sect_under_nodes(struct memory_block *mem_blk)
	375	{
9ae49fab	376	NODEMASK_ALLOC(nodemask_t, unlinked_nodes, GFP_KERNEL);
c04fc586 GH	377	unsigned long pfn, sect_start_pfn, sect_end_pfn;
c04fc586 GH	378
9ae49fab DR	379	if (!mem_blk) {
9ae49fab DR	380	NODEMASK_FREE(unlinked_nodes);
c04fc586	381	return -EFAULT;
9ae49fab DR	382	}
	383	if (!unlinked_nodes)
	384	return -ENOMEM;
	385	nodes_clear(*unlinked_nodes);
c04fc586 GH	386	sect_start_pfn = section_nr_to_pfn(mem_blk->phys_index);
	387	sect_end_pfn = sect_start_pfn + PAGES_PER_SECTION - 1;
	388	for (pfn = sect_start_pfn; pfn <= sect_end_pfn; pfn++) {
47504980	389	int nid;
c04fc586 GH	390
	391	nid = get_nid_for_pfn(pfn);
	392	if (nid < 0)
	393	continue;
	394	if (!node_online(nid))
	395	continue;
9ae49fab	396	if (node_test_and_set(nid, *unlinked_nodes))
c04fc586 GH	397	continue;
	398	sysfs_remove_link(&node_devices[nid].sysdev.kobj,
	399	kobject_name(&mem_blk->sysdev.kobj));
dee5d0d5 AC	400	sysfs_remove_link(&mem_blk->sysdev.kobj,
dee5d0d5 AC	401	kobject_name(&node_devices[nid].sysdev.kobj));
c04fc586	402	}
9ae49fab	403	NODEMASK_FREE(unlinked_nodes);
c04fc586 GH	404	return 0;
	405	}
	406
	407	static int link_mem_sections(int nid)
	408	{
	409	unsigned long start_pfn = NODE_DATA(nid)->node_start_pfn;
	410	unsigned long end_pfn = start_pfn + NODE_DATA(nid)->node_spanned_pages;
	411	unsigned long pfn;
	412	int err = 0;
	413
	414	for (pfn = start_pfn; pfn < end_pfn; pfn += PAGES_PER_SECTION) {
	415	unsigned long section_nr = pfn_to_section_nr(pfn);
	416	struct mem_section *mem_sect;
	417	struct memory_block *mem_blk;
	418	int ret;
	419
	420	if (!present_section_nr(section_nr))
	421	continue;
	422	mem_sect = __nr_to_section(section_nr);
	423	mem_blk = find_memory_block(mem_sect);
	424	ret = register_mem_sect_under_node(mem_blk, nid);
	425	if (!err)
	426	err = ret;
	427
	428	/* discard ref obtained in find_memory_block() */
	429	kobject_put(&mem_blk->sysdev.kobj);
	430	}
	431	return err;
	432	}
4faf8d95	433
39da08cb	434	#ifdef CONFIG_HUGETLBFS
4faf8d95 LS	435	/*
	436	* Handle per node hstate attribute [un]registration on transistions
	437	* to/from memoryless state.
	438	*/
39da08cb LS	439	static void node_hugetlb_work(struct work_struct *work)
	440	{
	441	struct node *node = container_of(work, struct node, node_work);
	442
	443	/*
	444	* We only get here when a node transitions to/from memoryless state.
	445	* We can detect which transition occurred by examining whether the
	446	* node has memory now. hugetlb_register_node() already check this
	447	* so we try to register the attributes. If that fails, then the
	448	* node has transitioned to memoryless, try to unregister the
	449	* attributes.
	450	*/
	451	if (!hugetlb_register_node(node))
	452	hugetlb_unregister_node(node);
	453	}
	454
	455	static void init_node_hugetlb_work(int nid)
	456	{
	457	INIT_WORK(&node_devices[nid].node_work, node_hugetlb_work);
	458	}
4faf8d95 LS	459
	460	static int node_memory_callback(struct notifier_block *self,
	461	unsigned long action, void *arg)
	462	{
	463	struct memory_notify *mnb = arg;
	464	int nid = mnb->status_change_nid;
	465
	466	switch (action) {
39da08cb LS	467	case MEM_ONLINE:
	468	case MEM_OFFLINE:
	469	/*
	470	* offload per node hstate [un]registration to a work thread
	471	* when transitioning to/from memoryless state.
	472	*/
4faf8d95	473	if (nid != NUMA_NO_NODE)
39da08cb	474	schedule_work(&node_devices[nid].node_work);
4faf8d95	475	break;
39da08cb	476
4faf8d95 LS	477	case MEM_GOING_ONLINE:
	478	case MEM_GOING_OFFLINE:
	479	case MEM_CANCEL_ONLINE:
	480	case MEM_CANCEL_OFFLINE:
	481	default:
	482	break;
	483	}
	484
	485	return NOTIFY_OK;
	486	}
39da08cb LS	487	#endif /* CONFIG_HUGETLBFS */
	488	#else /* !CONFIG_MEMORY_HOTPLUG_SPARSE */
	489
c04fc586	490	static int link_mem_sections(int nid) { return 0; }
39da08cb	491	#endif /* CONFIG_MEMORY_HOTPLUG_SPARSE */
4faf8d95	492
39da08cb LS	493	#if !defined(CONFIG_MEMORY_HOTPLUG_SPARSE) \|\| \
39da08cb LS	494	!defined(CONFIG_HUGETLBFS)
4faf8d95 LS	495	static inline int node_memory_callback(struct notifier_block *self,
	496	unsigned long action, void *arg)
	497	{
	498	return NOTIFY_OK;
	499	}
39da08cb LS	500
	501	static void init_node_hugetlb_work(int nid) { }
	502
	503	#endif
c04fc586	504
0fc44159 YG	505	int register_one_node(int nid)
	506	{
	507	int error = 0;
76b67ed9	508	int cpu;
0fc44159 YG	509
	510	if (node_online(nid)) {
	511	int p_node = parent_node(nid);
	512	struct node *parent = NULL;
	513
	514	if (p_node != nid)
	515	parent = &node_devices[p_node];
	516
	517	error = register_node(&node_devices[nid], nid, parent);
76b67ed9 KH	518
	519	/* link cpu under this node */
	520	for_each_present_cpu(cpu) {
	521	if (cpu_to_node(cpu) == nid)
	522	register_cpu_under_node(cpu, nid);
	523	}
c04fc586 GH	524
	525	/* link memory sections under this node */
	526	error = link_mem_sections(nid);
39da08cb LS	527
	528	/* initialize work queue for memory hot plug */
	529	init_node_hugetlb_work(nid);
0fc44159 YG	530	}
	531
	532	return error;
	533
	534	}
	535
	536	void unregister_one_node(int nid)
	537	{
	538	unregister_node(&node_devices[nid]);
	539	}
	540
bde631a5 LS	541	/*
	542	* node states attributes
	543	*/
	544
	545	static ssize_t print_nodes_state(enum node_states state, char *buf)
	546	{
	547	int n;
	548
	549	n = nodelist_scnprintf(buf, PAGE_SIZE, node_states[state]);
	550	if (n > 0 && PAGE_SIZE > n + 1) {
	551	*(buf + n++) = '\n';
	552	*(buf + n++) = '\0';
	553	}
	554	return n;
	555	}
	556
b15f562f AK	557	struct node_attr {
	558	struct sysdev_class_attribute attr;
	559	enum node_states state;
	560	};
bde631a5	561
b15f562f AK	562	static ssize_t show_node_state(struct sysdev_class *class,
b15f562f AK	563	struct sysdev_class_attribute attr, char buf)
bde631a5	564	{
b15f562f AK	565	struct node_attr *na = container_of(attr, struct node_attr, attr);
b15f562f AK	566	return print_nodes_state(na->state, buf);
bde631a5 LS	567	}
bde631a5 LS	568
b15f562f AK	569	#define _NODE_ATTR(name, state) \
b15f562f AK	570	{ _SYSDEV_CLASS_ATTR(name, 0444, show_node_state, NULL), state }
bde631a5	571
b15f562f AK	572	static struct node_attr node_state_attr[] = {
	573	_NODE_ATTR(possible, N_POSSIBLE),
	574	_NODE_ATTR(online, N_ONLINE),
	575	_NODE_ATTR(has_normal_memory, N_NORMAL_MEMORY),
	576	_NODE_ATTR(has_cpu, N_CPU),
bde631a5	577	#ifdef CONFIG_HIGHMEM
b15f562f	578	_NODE_ATTR(has_high_memory, N_HIGH_MEMORY),
bde631a5	579	#endif
bde631a5 LS	580	};
bde631a5 LS	581
3701cde6 AK	582	static struct sysdev_class_attribute *node_state_attrs[] = {
	583	&node_state_attr[0].attr,
	584	&node_state_attr[1].attr,
	585	&node_state_attr[2].attr,
	586	&node_state_attr[3].attr,
	587	#ifdef CONFIG_HIGHMEM
	588	&node_state_attr[4].attr,
	589	#endif
	590	NULL
	591	};
bde631a5	592
4faf8d95	593	#define NODE_CALLBACK_PRI 2 /* lower than SLAB */
4b45099b	594	static int __init register_node_type(void)
1da177e4	595	{
bde631a5 LS	596	int ret;
bde631a5 LS	597
3701cde6 AK	598	BUILD_BUG_ON(ARRAY_SIZE(node_state_attr) != NR_NODE_STATES);
	599	BUILD_BUG_ON(ARRAY_SIZE(node_state_attrs)-1 != NR_NODE_STATES);
	600
bde631a5	601	ret = sysdev_class_register(&node_class);
4faf8d95	602	if (!ret) {
4faf8d95 LS	603	hotplug_memory_notifier(node_memory_callback,
	604	NODE_CALLBACK_PRI);
	605	}
bde631a5 LS	606
	607	/*
	608	* Note: we're not going to unregister the node class if we fail
	609	* to register the node state class attribute files.
	610	*/
	611	return ret;
1da177e4 LT	612	}
1da177e4 LT	613	postcore_initcall(register_node_type);