[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/cpumask.h>
#include <linux/topology.h>
#include <linux/nodemask.h>
#include <linux/cpu.h>
#include <linux/device.h>
#include <linux/swap.h>

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


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 currently PAGE_SIZE, need ~4 chars per node */
	BUILD_BUG_ON(MAX_NUMNODES*4 > PAGE_SIZE/2);

	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;
}

static ssize_t print_nodes_possible(struct sysdev_class *class, char *buf)
{
	return print_nodes_state(N_POSSIBLE, buf);
}

static ssize_t print_nodes_online(struct sysdev_class *class, char *buf)
{
	return print_nodes_state(N_ONLINE, buf);
}

static ssize_t print_nodes_has_normal_memory(struct sysdev_class *class,
						char *buf)
{
	return print_nodes_state(N_NORMAL_MEMORY, buf);
}

static ssize_t print_nodes_has_cpu(struct sysdev_class *class, char *buf)
{
	return print_nodes_state(N_CPU, buf);
}

static SYSDEV_CLASS_ATTR(possible, 0444, print_nodes_possible, NULL);
static SYSDEV_CLASS_ATTR(online, 0444, print_nodes_online, NULL);
static SYSDEV_CLASS_ATTR(has_normal_memory, 0444, print_nodes_has_normal_memory,
									NULL);
static SYSDEV_CLASS_ATTR(has_cpu, 0444, print_nodes_has_cpu, NULL);

#ifdef CONFIG_HIGHMEM
static ssize_t print_nodes_has_high_memory(struct sysdev_class *class,
						 char *buf)
{
	return print_nodes_state(N_HIGH_MEMORY, buf);
}

static SYSDEV_CLASS_ATTR(has_high_memory, 0444, print_nodes_has_high_memory,
									 NULL);
#endif

struct sysdev_class_attribute *node_state_attr[] = {
	&attr_possible,
	&attr_online,
	&attr_has_normal_memory,
#ifdef CONFIG_HIGHMEM
	&attr_has_high_memory,
#endif
	&attr_has_cpu,
};

static int node_states_init(void)
{
	int i;
	int err = 0;

	for (i = 0;  i < NR_NODE_STATES; i++) {
		int ret;
		ret = sysdev_class_create_file(&node_class, node_state_attr[i]);
		if (!err)
			err = ret;
	}
	return err;
}

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

	ret = sysdev_class_register(&node_class);
	if (!ret) {
		ret = node_states_init();
		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>
1da177e4 LT	18
1da177e4 LT	19	static struct sysdev_class node_class = {
af5ca3f4	20	.name = "node",
1da177e4 LT	21	};
	22
	23
39106dcf	24	static ssize_t node_read_cpumap(struct sys_device dev, int type, char buf)
1da177e4 LT	25	{
1da177e4 LT	26	struct node *node_dev = to_node(dev);
a70f7302	27	const struct cpumask *mask = cpumask_of_node(node_dev->sysdev.id);
1da177e4 LT	28	int len;
1da177e4 LT	29
39106dcf MT	30	/* 2008/04/07: buf currently PAGE_SIZE, need 9 chars per 32 bits. */
39106dcf MT	31	BUILD_BUG_ON((NR_CPUS/32 * 9) > (PAGE_SIZE-1));
1da177e4	32
39106dcf	33	len = type?
29c0177e RR	34	cpulist_scnprintf(buf, PAGE_SIZE-2, mask) :
29c0177e RR	35	cpumask_scnprintf(buf, PAGE_SIZE-2, mask);
c5f59f08 MT	36	buf[len++] = '\n';
c5f59f08 MT	37	buf[len] = '\0';
1da177e4 LT	38	return len;
	39	}
	40
4a0b2b4d AK	41	static inline ssize_t node_read_cpumask(struct sys_device *dev,
4a0b2b4d AK	42	struct sysdev_attribute attr, char buf)
39106dcf MT	43	{
	44	return node_read_cpumap(dev, 0, buf);
	45	}
4a0b2b4d AK	46	static inline ssize_t node_read_cpulist(struct sys_device *dev,
4a0b2b4d AK	47	struct sysdev_attribute attr, char buf)
39106dcf MT	48	{
	49	return node_read_cpumap(dev, 1, buf);
	50	}
	51
	52	static SYSDEV_ATTR(cpumap, S_IRUGO, node_read_cpumask, NULL);
	53	static SYSDEV_ATTR(cpulist, S_IRUGO, node_read_cpulist, NULL);
1da177e4 LT	54
1da177e4 LT	55	#define K(x) ((x) << (PAGE_SHIFT - 10))
4a0b2b4d AK	56	static ssize_t node_read_meminfo(struct sys_device * dev,
4a0b2b4d AK	57	struct sysdev_attribute attr, char buf)
1da177e4 LT	58	{
	59	int n;
	60	int nid = dev->id;
	61	struct sysinfo i;
1da177e4 LT	62
1da177e4 LT	63	si_meminfo_node(&i, nid);
c07e02db	64
1da177e4	65	n = sprintf(buf, "\n"
4f98a2fe RR	66	"Node %d MemTotal: %8lu kB\n"
	67	"Node %d MemFree: %8lu kB\n"
	68	"Node %d MemUsed: %8lu kB\n"
	69	"Node %d Active: %8lu kB\n"
	70	"Node %d Inactive: %8lu kB\n"
	71	"Node %d Active(anon): %8lu kB\n"
	72	"Node %d Inactive(anon): %8lu kB\n"
	73	"Node %d Active(file): %8lu kB\n"
	74	"Node %d Inactive(file): %8lu kB\n"
5344b7e6 NP	75	"Node %d Unevictable: %8lu kB\n"
5344b7e6 NP	76	"Node %d Mlocked: %8lu kB\n"
182e8e23	77	#ifdef CONFIG_HIGHMEM
4f98a2fe RR	78	"Node %d HighTotal: %8lu kB\n"
	79	"Node %d HighFree: %8lu kB\n"
	80	"Node %d LowTotal: %8lu kB\n"
	81	"Node %d LowFree: %8lu kB\n"
182e8e23	82	#endif
4f98a2fe RR	83	"Node %d Dirty: %8lu kB\n"
	84	"Node %d Writeback: %8lu kB\n"
	85	"Node %d FilePages: %8lu kB\n"
	86	"Node %d Mapped: %8lu kB\n"
	87	"Node %d AnonPages: %8lu kB\n"
4b02108a	88	"Node %d Shmem: %8lu kB\n"
c6a7f572	89	"Node %d KernelStack: %8lu kB\n"
4f98a2fe RR	90	"Node %d PageTables: %8lu kB\n"
	91	"Node %d NFS_Unstable: %8lu kB\n"
	92	"Node %d Bounce: %8lu kB\n"
	93	"Node %d WritebackTmp: %8lu kB\n"
	94	"Node %d Slab: %8lu kB\n"
	95	"Node %d SReclaimable: %8lu kB\n"
	96	"Node %d SUnreclaim: %8lu kB\n",
1da177e4 LT	97	nid, K(i.totalram),
	98	nid, K(i.freeram),
	99	nid, K(i.totalram - i.freeram),
4f98a2fe RR	100	nid, K(node_page_state(nid, NR_ACTIVE_ANON) +
	101	node_page_state(nid, NR_ACTIVE_FILE)),
	102	nid, K(node_page_state(nid, NR_INACTIVE_ANON) +
	103	node_page_state(nid, NR_INACTIVE_FILE)),
	104	nid, K(node_page_state(nid, NR_ACTIVE_ANON)),
	105	nid, K(node_page_state(nid, NR_INACTIVE_ANON)),
	106	nid, K(node_page_state(nid, NR_ACTIVE_FILE)),
	107	nid, K(node_page_state(nid, NR_INACTIVE_FILE)),
7b854121	108	nid, K(node_page_state(nid, NR_UNEVICTABLE)),
5344b7e6	109	nid, K(node_page_state(nid, NR_MLOCK)),
182e8e23	110	#ifdef CONFIG_HIGHMEM
1da177e4 LT	111	nid, K(i.totalhigh),
	112	nid, K(i.freehigh),
	113	nid, K(i.totalram - i.totalhigh),
c07e02db	114	nid, K(i.freeram - i.freehigh),
182e8e23	115	#endif
b1e7a8fd	116	nid, K(node_page_state(nid, NR_FILE_DIRTY)),
ce866b34	117	nid, K(node_page_state(nid, NR_WRITEBACK)),
347ce434	118	nid, K(node_page_state(nid, NR_FILE_PAGES)),
65ba55f5	119	nid, K(node_page_state(nid, NR_FILE_MAPPED)),
f3dbd344	120	nid, K(node_page_state(nid, NR_ANON_PAGES)),
4b02108a	121	nid, K(node_page_state(nid, NR_SHMEM)),
c6a7f572 KM	122	nid, node_page_state(nid, NR_KERNEL_STACK) *
c6a7f572 KM	123	THREAD_SIZE / 1024,
df849a15	124	nid, K(node_page_state(nid, NR_PAGETABLE)),
fd39fc85	125	nid, K(node_page_state(nid, NR_UNSTABLE_NFS)),
d2c5e30c	126	nid, K(node_page_state(nid, NR_BOUNCE)),
fc3ba692	127	nid, K(node_page_state(nid, NR_WRITEBACK_TEMP)),
972d1a7b CL	128	nid, K(node_page_state(nid, NR_SLAB_RECLAIMABLE) +
	129	node_page_state(nid, NR_SLAB_UNRECLAIMABLE)),
	130	nid, K(node_page_state(nid, NR_SLAB_RECLAIMABLE)),
	131	nid, K(node_page_state(nid, NR_SLAB_UNRECLAIMABLE)));
1da177e4 LT	132	n += hugetlb_report_node_meminfo(nid, buf + n);
	133	return n;
	134	}
	135
	136	#undef K
	137	static SYSDEV_ATTR(meminfo, S_IRUGO, node_read_meminfo, NULL);
	138
4a0b2b4d AK	139	static ssize_t node_read_numastat(struct sys_device * dev,
4a0b2b4d AK	140	struct sysdev_attribute attr, char buf)
1da177e4	141	{
1da177e4 LT	142	return sprintf(buf,
	143	"numa_hit %lu\n"
	144	"numa_miss %lu\n"
	145	"numa_foreign %lu\n"
	146	"interleave_hit %lu\n"
	147	"local_node %lu\n"
	148	"other_node %lu\n",
ca889e6c CL	149	node_page_state(dev->id, NUMA_HIT),
	150	node_page_state(dev->id, NUMA_MISS),
	151	node_page_state(dev->id, NUMA_FOREIGN),
	152	node_page_state(dev->id, NUMA_INTERLEAVE_HIT),
	153	node_page_state(dev->id, NUMA_LOCAL),
	154	node_page_state(dev->id, NUMA_OTHER));
1da177e4 LT	155	}
	156	static SYSDEV_ATTR(numastat, S_IRUGO, node_read_numastat, NULL);
	157
4a0b2b4d AK	158	static ssize_t node_read_distance(struct sys_device * dev,
4a0b2b4d AK	159	struct sysdev_attribute attr, char buf)
1da177e4 LT	160	{
	161	int nid = dev->id;
	162	int len = 0;
	163	int i;
	164
	165	/* buf currently PAGE_SIZE, need ~4 chars per node */
	166	BUILD_BUG_ON(MAX_NUMNODES*4 > PAGE_SIZE/2);
	167
	168	for_each_online_node(i)
	169	len += sprintf(buf + len, "%s%d", i ? " " : "", node_distance(nid, i));
	170
	171	len += sprintf(buf + len, "\n");
	172	return len;
	173	}
	174	static SYSDEV_ATTR(distance, S_IRUGO, node_read_distance, NULL);
	175
9a305230 LS	176	#ifdef CONFIG_HUGETLBFS
	177	/*
	178	* hugetlbfs per node attributes registration interface:
	179	* When/if hugetlb[fs] subsystem initializes [sometime after this module],
4faf8d95 LS	180	* it will register its per node attributes for all online nodes with
4faf8d95 LS	181	* memory. It will also call register_hugetlbfs_with_node(), below, to
9a305230 LS	182	* register its attribute registration functions with this node driver.
	183	* Once these hooks have been initialized, the node driver will call into
	184	* the hugetlb module to [un]register attributes for hot-plugged nodes.
	185	*/
	186	static node_registration_func_t __hugetlb_register_node;
	187	static node_registration_func_t __hugetlb_unregister_node;
	188
39da08cb	189	static inline bool hugetlb_register_node(struct node *node)
9a305230	190	{
4faf8d95	191	if (__hugetlb_register_node &&
39da08cb	192	node_state(node->sysdev.id, N_HIGH_MEMORY)) {
9a305230	193	__hugetlb_register_node(node);
39da08cb LS	194	return true;
	195	}
	196	return false;
9a305230 LS	197	}
	198
	199	static inline void hugetlb_unregister_node(struct node *node)
	200	{
	201	if (__hugetlb_unregister_node)
	202	__hugetlb_unregister_node(node);
	203	}
	204
	205	void register_hugetlbfs_with_node(node_registration_func_t doregister,
	206	node_registration_func_t unregister)
	207	{
	208	__hugetlb_register_node = doregister;
	209	__hugetlb_unregister_node = unregister;
	210	}
	211	#else
	212	static inline void hugetlb_register_node(struct node *node) {}
	213
	214	static inline void hugetlb_unregister_node(struct node *node) {}
	215	#endif
	216
1da177e4 LT	217
1da177e4 LT	218	/*
405ae7d3	219	* register_node - Setup a sysfs device for a node.
1da177e4 LT	220	* @num - Node number to use when creating the device.
	221	*
	222	* Initialize and register the node device.
	223	*/
4b45099b	224	int register_node(struct node node, int num, struct node parent)
1da177e4 LT	225	{
	226	int error;
	227
	228	node->sysdev.id = num;
	229	node->sysdev.cls = &node_class;
	230	error = sysdev_register(&node->sysdev);
	231
	232	if (!error){
	233	sysdev_create_file(&node->sysdev, &attr_cpumap);
39106dcf	234	sysdev_create_file(&node->sysdev, &attr_cpulist);
1da177e4 LT	235	sysdev_create_file(&node->sysdev, &attr_meminfo);
	236	sysdev_create_file(&node->sysdev, &attr_numastat);
	237	sysdev_create_file(&node->sysdev, &attr_distance);
af936a16 LS	238
af936a16 LS	239	scan_unevictable_register_node(node);
4faf8d95	240
9a305230	241	hugetlb_register_node(node);
1da177e4 LT	242	}
	243	return error;
	244	}
	245
4b45099b KT	246	/**
	247	* unregister_node - unregister a node device
	248	* @node: node going away
	249	*
	250	* Unregisters a node device @node. All the devices on the node must be
	251	* unregistered before calling this function.
	252	*/
	253	void unregister_node(struct node *node)
	254	{
	255	sysdev_remove_file(&node->sysdev, &attr_cpumap);
39106dcf	256	sysdev_remove_file(&node->sysdev, &attr_cpulist);
4b45099b KT	257	sysdev_remove_file(&node->sysdev, &attr_meminfo);
	258	sysdev_remove_file(&node->sysdev, &attr_numastat);
	259	sysdev_remove_file(&node->sysdev, &attr_distance);
	260
af936a16	261	scan_unevictable_unregister_node(node);
4faf8d95	262	hugetlb_unregister_node(node); /* no-op, if memoryless node */
af936a16	263
4b45099b KT	264	sysdev_unregister(&node->sysdev);
4b45099b KT	265	}
1da177e4	266
0fc44159 YG	267	struct node node_devices[MAX_NUMNODES];
0fc44159 YG	268
76b67ed9 KH	269	/*
	270	* register cpu under node
	271	*/
	272	int register_cpu_under_node(unsigned int cpu, unsigned int nid)
	273	{
1830794a	274	int ret;
f8246f31	275	struct sys_device *obj;
76b67ed9	276
f8246f31 AC	277	if (!node_online(nid))
	278	return 0;
	279
	280	obj = get_cpu_sysdev(cpu);
	281	if (!obj)
	282	return 0;
	283
1830794a	284	ret = sysfs_create_link(&node_devices[nid].sysdev.kobj,
f8246f31 AC	285	&obj->kobj,
f8246f31 AC	286	kobject_name(&obj->kobj));
1830794a AC	287	if (ret)
	288	return ret;
	289
	290	return sysfs_create_link(&obj->kobj,
	291	&node_devices[nid].sysdev.kobj,
	292	kobject_name(&node_devices[nid].sysdev.kobj));
76b67ed9 KH	293	}
	294
	295	int unregister_cpu_under_node(unsigned int cpu, unsigned int nid)
	296	{
b9d52dad AC	297	struct sys_device *obj;
	298
	299	if (!node_online(nid))
	300	return 0;
	301
	302	obj = get_cpu_sysdev(cpu);
	303	if (!obj)
	304	return 0;
	305
	306	sysfs_remove_link(&node_devices[nid].sysdev.kobj,
	307	kobject_name(&obj->kobj));
1830794a AC	308	sysfs_remove_link(&obj->kobj,
1830794a AC	309	kobject_name(&node_devices[nid].sysdev.kobj));
b9d52dad	310
76b67ed9 KH	311	return 0;
	312	}
	313
c04fc586 GH	314	#ifdef CONFIG_MEMORY_HOTPLUG_SPARSE
	315	#define page_initialized(page) (page->lru.next)
	316
	317	static int get_nid_for_pfn(unsigned long pfn)
	318	{
	319	struct page *page;
	320
	321	if (!pfn_valid_within(pfn))
	322	return -1;
	323	page = pfn_to_page(pfn);
	324	if (!page_initialized(page))
	325	return -1;
	326	return pfn_to_nid(pfn);
	327	}
	328
	329	/* register memory section under specified node if it spans that node */
	330	int register_mem_sect_under_node(struct memory_block *mem_blk, int nid)
	331	{
dee5d0d5	332	int ret;
c04fc586 GH	333	unsigned long pfn, sect_start_pfn, sect_end_pfn;
	334
	335	if (!mem_blk)
	336	return -EFAULT;
	337	if (!node_online(nid))
	338	return 0;
	339	sect_start_pfn = section_nr_to_pfn(mem_blk->phys_index);
	340	sect_end_pfn = sect_start_pfn + PAGES_PER_SECTION - 1;
	341	for (pfn = sect_start_pfn; pfn <= sect_end_pfn; pfn++) {
	342	int page_nid;
	343
	344	page_nid = get_nid_for_pfn(pfn);
	345	if (page_nid < 0)
	346	continue;
	347	if (page_nid != nid)
	348	continue;
dee5d0d5	349	ret = sysfs_create_link_nowarn(&node_devices[nid].sysdev.kobj,
c04fc586 GH	350	&mem_blk->sysdev.kobj,
c04fc586 GH	351	kobject_name(&mem_blk->sysdev.kobj));
dee5d0d5 AC	352	if (ret)
	353	return ret;
	354
	355	return sysfs_create_link_nowarn(&mem_blk->sysdev.kobj,
	356	&node_devices[nid].sysdev.kobj,
	357	kobject_name(&node_devices[nid].sysdev.kobj));
c04fc586 GH	358	}
	359	/* mem section does not span the specified node */
	360	return 0;
	361	}
	362
	363	/* unregister memory section under all nodes that it spans */
	364	int unregister_mem_sect_under_nodes(struct memory_block *mem_blk)
	365	{
9ae49fab	366	NODEMASK_ALLOC(nodemask_t, unlinked_nodes, GFP_KERNEL);
c04fc586 GH	367	unsigned long pfn, sect_start_pfn, sect_end_pfn;
c04fc586 GH	368
9ae49fab DR	369	if (!mem_blk) {
9ae49fab DR	370	NODEMASK_FREE(unlinked_nodes);
c04fc586	371	return -EFAULT;
9ae49fab DR	372	}
	373	if (!unlinked_nodes)
	374	return -ENOMEM;
	375	nodes_clear(*unlinked_nodes);
c04fc586 GH	376	sect_start_pfn = section_nr_to_pfn(mem_blk->phys_index);
	377	sect_end_pfn = sect_start_pfn + PAGES_PER_SECTION - 1;
	378	for (pfn = sect_start_pfn; pfn <= sect_end_pfn; pfn++) {
47504980	379	int nid;
c04fc586 GH	380
	381	nid = get_nid_for_pfn(pfn);
	382	if (nid < 0)
	383	continue;
	384	if (!node_online(nid))
	385	continue;
9ae49fab	386	if (node_test_and_set(nid, *unlinked_nodes))
c04fc586 GH	387	continue;
	388	sysfs_remove_link(&node_devices[nid].sysdev.kobj,
	389	kobject_name(&mem_blk->sysdev.kobj));
dee5d0d5 AC	390	sysfs_remove_link(&mem_blk->sysdev.kobj,
dee5d0d5 AC	391	kobject_name(&node_devices[nid].sysdev.kobj));
c04fc586	392	}
9ae49fab	393	NODEMASK_FREE(unlinked_nodes);
c04fc586 GH	394	return 0;
	395	}
	396
	397	static int link_mem_sections(int nid)
	398	{
	399	unsigned long start_pfn = NODE_DATA(nid)->node_start_pfn;
	400	unsigned long end_pfn = start_pfn + NODE_DATA(nid)->node_spanned_pages;
	401	unsigned long pfn;
	402	int err = 0;
	403
	404	for (pfn = start_pfn; pfn < end_pfn; pfn += PAGES_PER_SECTION) {
	405	unsigned long section_nr = pfn_to_section_nr(pfn);
	406	struct mem_section *mem_sect;
	407	struct memory_block *mem_blk;
	408	int ret;
	409
	410	if (!present_section_nr(section_nr))
	411	continue;
	412	mem_sect = __nr_to_section(section_nr);
	413	mem_blk = find_memory_block(mem_sect);
	414	ret = register_mem_sect_under_node(mem_blk, nid);
	415	if (!err)
	416	err = ret;
	417
	418	/* discard ref obtained in find_memory_block() */
	419	kobject_put(&mem_blk->sysdev.kobj);
	420	}
	421	return err;
	422	}
4faf8d95	423
39da08cb	424	#ifdef CONFIG_HUGETLBFS
4faf8d95 LS	425	/*
	426	* Handle per node hstate attribute [un]registration on transistions
	427	* to/from memoryless state.
	428	*/
39da08cb LS	429	static void node_hugetlb_work(struct work_struct *work)
	430	{
	431	struct node *node = container_of(work, struct node, node_work);
	432
	433	/*
	434	* We only get here when a node transitions to/from memoryless state.
	435	* We can detect which transition occurred by examining whether the
	436	* node has memory now. hugetlb_register_node() already check this
	437	* so we try to register the attributes. If that fails, then the
	438	* node has transitioned to memoryless, try to unregister the
	439	* attributes.
	440	*/
	441	if (!hugetlb_register_node(node))
	442	hugetlb_unregister_node(node);
	443	}
	444
	445	static void init_node_hugetlb_work(int nid)
	446	{
	447	INIT_WORK(&node_devices[nid].node_work, node_hugetlb_work);
	448	}
4faf8d95 LS	449
	450	static int node_memory_callback(struct notifier_block *self,
	451	unsigned long action, void *arg)
	452	{
	453	struct memory_notify *mnb = arg;
	454	int nid = mnb->status_change_nid;
	455
	456	switch (action) {
39da08cb LS	457	case MEM_ONLINE:
	458	case MEM_OFFLINE:
	459	/*
	460	* offload per node hstate [un]registration to a work thread
	461	* when transitioning to/from memoryless state.
	462	*/
4faf8d95	463	if (nid != NUMA_NO_NODE)
39da08cb	464	schedule_work(&node_devices[nid].node_work);
4faf8d95	465	break;
39da08cb	466
4faf8d95 LS	467	case MEM_GOING_ONLINE:
	468	case MEM_GOING_OFFLINE:
	469	case MEM_CANCEL_ONLINE:
	470	case MEM_CANCEL_OFFLINE:
	471	default:
	472	break;
	473	}
	474
	475	return NOTIFY_OK;
	476	}
39da08cb LS	477	#endif /* CONFIG_HUGETLBFS */
	478	#else /* !CONFIG_MEMORY_HOTPLUG_SPARSE */
	479
c04fc586	480	static int link_mem_sections(int nid) { return 0; }
39da08cb	481	#endif /* CONFIG_MEMORY_HOTPLUG_SPARSE */
4faf8d95	482
39da08cb LS	483	#if !defined(CONFIG_MEMORY_HOTPLUG_SPARSE) \|\| \
39da08cb LS	484	!defined(CONFIG_HUGETLBFS)
4faf8d95 LS	485	static inline int node_memory_callback(struct notifier_block *self,
	486	unsigned long action, void *arg)
	487	{
	488	return NOTIFY_OK;
	489	}
39da08cb LS	490
	491	static void init_node_hugetlb_work(int nid) { }
	492
	493	#endif
c04fc586	494
0fc44159 YG	495	int register_one_node(int nid)
	496	{
	497	int error = 0;
76b67ed9	498	int cpu;
0fc44159 YG	499
	500	if (node_online(nid)) {
	501	int p_node = parent_node(nid);
	502	struct node *parent = NULL;
	503
	504	if (p_node != nid)
	505	parent = &node_devices[p_node];
	506
	507	error = register_node(&node_devices[nid], nid, parent);
76b67ed9 KH	508
	509	/* link cpu under this node */
	510	for_each_present_cpu(cpu) {
	511	if (cpu_to_node(cpu) == nid)
	512	register_cpu_under_node(cpu, nid);
	513	}
c04fc586 GH	514
	515	/* link memory sections under this node */
	516	error = link_mem_sections(nid);
39da08cb LS	517
	518	/* initialize work queue for memory hot plug */
	519	init_node_hugetlb_work(nid);
0fc44159 YG	520	}
	521
	522	return error;
	523
	524	}
	525
	526	void unregister_one_node(int nid)
	527	{
	528	unregister_node(&node_devices[nid]);
	529	}
	530
bde631a5 LS	531	/*
	532	* node states attributes
	533	*/
	534
	535	static ssize_t print_nodes_state(enum node_states state, char *buf)
	536	{
	537	int n;
	538
	539	n = nodelist_scnprintf(buf, PAGE_SIZE, node_states[state]);
	540	if (n > 0 && PAGE_SIZE > n + 1) {
	541	*(buf + n++) = '\n';
	542	*(buf + n++) = '\0';
	543	}
	544	return n;
	545	}
	546
	547	static ssize_t print_nodes_possible(struct sysdev_class class, char buf)
	548	{
	549	return print_nodes_state(N_POSSIBLE, buf);
	550	}
	551
	552	static ssize_t print_nodes_online(struct sysdev_class class, char buf)
	553	{
	554	return print_nodes_state(N_ONLINE, buf);
	555	}
	556
	557	static ssize_t print_nodes_has_normal_memory(struct sysdev_class *class,
	558	char *buf)
	559	{
	560	return print_nodes_state(N_NORMAL_MEMORY, buf);
	561	}
	562
	563	static ssize_t print_nodes_has_cpu(struct sysdev_class class, char buf)
	564	{
	565	return print_nodes_state(N_CPU, buf);
	566	}
	567
	568	static SYSDEV_CLASS_ATTR(possible, 0444, print_nodes_possible, NULL);
	569	static SYSDEV_CLASS_ATTR(online, 0444, print_nodes_online, NULL);
	570	static SYSDEV_CLASS_ATTR(has_normal_memory, 0444, print_nodes_has_normal_memory,
	571	NULL);
	572	static SYSDEV_CLASS_ATTR(has_cpu, 0444, print_nodes_has_cpu, NULL);
	573
	574	#ifdef CONFIG_HIGHMEM
	575	static ssize_t print_nodes_has_high_memory(struct sysdev_class *class,
	576	char *buf)
	577	{
	578	return print_nodes_state(N_HIGH_MEMORY, buf);
	579	}
	580
	581	static SYSDEV_CLASS_ATTR(has_high_memory, 0444, print_nodes_has_high_memory,
	582	NULL);
	583	#endif
	584
	585	struct sysdev_class_attribute *node_state_attr[] = {
	586	&attr_possible,
	587	&attr_online,
	588	&attr_has_normal_memory,
	589	#ifdef CONFIG_HIGHMEM
	590	&attr_has_high_memory,
	591	#endif
	592	&attr_has_cpu,
	593	};
	594
595	static int node_states_init(void)
596	{
597	int i;
598	int err = 0;
599
600	for (i = 0; i < NR_NODE_STATES; i++) {
601	int ret;
602	ret = sysdev_class_create_file(&node_class, node_state_attr[i]);
603	if (!err)
604	err = ret;
605	}
606	return err;
607	}
608
4faf8d95	609	#define NODE_CALLBACK_PRI 2 /* lower than SLAB */
4b45099b	610	static int __init register_node_type(void)
1da177e4	611	{
bde631a5 LS	612	int ret;
	613
	614	ret = sysdev_class_register(&node_class);
4faf8d95	615	if (!ret) {
bde631a5	616	ret = node_states_init();
4faf8d95 LS	617	hotplug_memory_notifier(node_memory_callback,
	618	NODE_CALLBACK_PRI);
	619	}
bde631a5 LS	620
	621	/*
	622	* Note: we're not going to unregister the node class if we fail
	623	* to register the node state class attribute files.
	624	*/
	625	return ret;
1da177e4 LT	626	}
1da177e4 LT	627	postcore_initcall(register_node_type);