[linux-2.6-block.git] / mm / swap.c

/*
 *  linux/mm/swap.c
 *
 *  Copyright (C) 1991, 1992, 1993, 1994  Linus Torvalds
 */

/*
 * This file contains the default values for the operation of the
 * Linux VM subsystem. Fine-tuning documentation can be found in
 * Documentation/sysctl/vm.txt.
 * Started 18.12.91
 * Swap aging added 23.2.95, Stephen Tweedie.
 * Buffermem limits added 12.3.98, Rik van Riel.
 */

#include <linux/mm.h>
#include <linux/sched.h>
#include <linux/kernel_stat.h>
#include <linux/swap.h>
#include <linux/mman.h>
#include <linux/pagemap.h>
#include <linux/pagevec.h>
#include <linux/init.h>
#include <linux/module.h>
#include <linux/mm_inline.h>
#include <linux/buffer_head.h>	/* for try_to_release_page() */
#include <linux/percpu_counter.h>
#include <linux/percpu.h>
#include <linux/cpu.h>
#include <linux/notifier.h>
#include <linux/backing-dev.h>
#include <linux/memcontrol.h>

/* How many pages do we try to swap or page in/out together? */
int page_cluster;

static DEFINE_PER_CPU(struct pagevec, lru_add_pvecs) = { 0, };
static DEFINE_PER_CPU(struct pagevec, lru_add_active_pvecs) = { 0, };
static DEFINE_PER_CPU(struct pagevec, lru_rotate_pvecs) = { 0, };

/*
 * This path almost never happens for VM activity - pages are normally
 * freed via pagevecs.  But it gets used by networking.
 */
static void __page_cache_release(struct page *page)
{
	if (PageLRU(page)) {
		unsigned long flags;
		struct zone *zone = page_zone(page);

		spin_lock_irqsave(&zone->lru_lock, flags);
		VM_BUG_ON(!PageLRU(page));
		__ClearPageLRU(page);
		del_page_from_lru(zone, page);
		spin_unlock_irqrestore(&zone->lru_lock, flags);
	}
	free_hot_page(page);
}

static void put_compound_page(struct page *page)
{
	page = compound_head(page);
	if (put_page_testzero(page)) {
		compound_page_dtor *dtor;

		dtor = get_compound_page_dtor(page);
		(*dtor)(page);
	}
}

void put_page(struct page *page)
{
	if (unlikely(PageCompound(page)))
		put_compound_page(page);
	else if (put_page_testzero(page))
		__page_cache_release(page);
}
EXPORT_SYMBOL(put_page);

/**
 * put_pages_list() - release a list of pages
 * @pages: list of pages threaded on page->lru
 *
 * Release a list of pages which are strung together on page.lru.  Currently
 * used by read_cache_pages() and related error recovery code.
 */
void put_pages_list(struct list_head *pages)
{
	while (!list_empty(pages)) {
		struct page *victim;

		victim = list_entry(pages->prev, struct page, lru);
		list_del(&victim->lru);
		page_cache_release(victim);
	}
}
EXPORT_SYMBOL(put_pages_list);

/*
 * pagevec_move_tail() must be called with IRQ disabled.
 * Otherwise this may cause nasty races.
 */
static void pagevec_move_tail(struct pagevec *pvec)
{
	int i;
	int pgmoved = 0;
	struct zone *zone = NULL;

	for (i = 0; i < pagevec_count(pvec); i++) {
		struct page *page = pvec->pages[i];
		struct zone *pagezone = page_zone(page);

		if (pagezone != zone) {
			if (zone)
				spin_unlock(&zone->lru_lock);
			zone = pagezone;
			spin_lock(&zone->lru_lock);
		}
		if (PageLRU(page) && !PageActive(page)) {
			list_move_tail(&page->lru, &zone->inactive_list);
			pgmoved++;
		}
	}
	if (zone)
		spin_unlock(&zone->lru_lock);
	__count_vm_events(PGROTATED, pgmoved);
	release_pages(pvec->pages, pvec->nr, pvec->cold);
	pagevec_reinit(pvec);
}

/*
 * Writeback is about to end against a page which has been marked for immediate
 * reclaim.  If it still appears to be reclaimable, move it to the tail of the
 * inactive list.
 *
 * Returns zero if it cleared PG_writeback.
 */
int rotate_reclaimable_page(struct page *page)
{
	struct pagevec *pvec;
	unsigned long flags;

	if (PageLocked(page))
		return 1;
	if (PageDirty(page))
		return 1;
	if (PageActive(page))
		return 1;
	if (!PageLRU(page))
		return 1;

	page_cache_get(page);
	local_irq_save(flags);
	pvec = &__get_cpu_var(lru_rotate_pvecs);
	if (!pagevec_add(pvec, page))
		pagevec_move_tail(pvec);
	local_irq_restore(flags);

	if (!test_clear_page_writeback(page))
		BUG();

	return 0;
}

/*
 * FIXME: speed this up?
 */
void activate_page(struct page *page)
{
	struct zone *zone = page_zone(page);

	spin_lock_irq(&zone->lru_lock);
	if (PageLRU(page) && !PageActive(page)) {
		del_page_from_inactive_list(zone, page);
		SetPageActive(page);
		add_page_to_active_list(zone, page);
		__count_vm_event(PGACTIVATE);
		mem_cgroup_move_lists(page, true);
	}
	spin_unlock_irq(&zone->lru_lock);
}

/*
 * Mark a page as having seen activity.
 *
 * inactive,unreferenced	->	inactive,referenced
 * inactive,referenced		->	active,unreferenced
 * active,unreferenced		->	active,referenced
 */
void mark_page_accessed(struct page *page)
{
	if (!PageActive(page) && PageReferenced(page) && PageLRU(page)) {
		activate_page(page);
		ClearPageReferenced(page);
	} else if (!PageReferenced(page)) {
		SetPageReferenced(page);
	}
}

EXPORT_SYMBOL(mark_page_accessed);

/**
 * lru_cache_add: add a page to the page lists
 * @page: the page to add
 */
void lru_cache_add(struct page *page)
{
	struct pagevec *pvec = &get_cpu_var(lru_add_pvecs);

	page_cache_get(page);
	if (!pagevec_add(pvec, page))
		__pagevec_lru_add(pvec);
	put_cpu_var(lru_add_pvecs);
}

void lru_cache_add_active(struct page *page)
{
	struct pagevec *pvec = &get_cpu_var(lru_add_active_pvecs);

	page_cache_get(page);
	if (!pagevec_add(pvec, page))
		__pagevec_lru_add_active(pvec);
	put_cpu_var(lru_add_active_pvecs);
}

/*
 * Drain pages out of the cpu's pagevecs.
 * Either "cpu" is the current CPU, and preemption has already been
 * disabled; or "cpu" is being hot-unplugged, and is already dead.
 */
static void drain_cpu_pagevecs(int cpu)
{
	struct pagevec *pvec;

	pvec = &per_cpu(lru_add_pvecs, cpu);
	if (pagevec_count(pvec))
		__pagevec_lru_add(pvec);

	pvec = &per_cpu(lru_add_active_pvecs, cpu);
	if (pagevec_count(pvec))
		__pagevec_lru_add_active(pvec);

	pvec = &per_cpu(lru_rotate_pvecs, cpu);
	if (pagevec_count(pvec)) {
		unsigned long flags;

		/* No harm done if a racing interrupt already did this */
		local_irq_save(flags);
		pagevec_move_tail(pvec);
		local_irq_restore(flags);
	}
}

void lru_add_drain(void)
{
	drain_cpu_pagevecs(get_cpu());
	put_cpu();
}

#ifdef CONFIG_NUMA
static void lru_add_drain_per_cpu(struct work_struct *dummy)
{
	lru_add_drain();
}

/*
 * Returns 0 for success
 */
int lru_add_drain_all(void)
{
	return schedule_on_each_cpu(lru_add_drain_per_cpu);
}

#else

/*
 * Returns 0 for success
 */
int lru_add_drain_all(void)
{
	lru_add_drain();
	return 0;
}
#endif

/*
 * Batched page_cache_release().  Decrement the reference count on all the
 * passed pages.  If it fell to zero then remove the page from the LRU and
 * free it.
 *
 * Avoid taking zone->lru_lock if possible, but if it is taken, retain it
 * for the remainder of the operation.
 *
 * The locking in this function is against shrink_cache(): we recheck the
 * page count inside the lock to see whether shrink_cache grabbed the page
 * via the LRU.  If it did, give up: shrink_cache will free it.
 */
void release_pages(struct page **pages, int nr, int cold)
{
	int i;
	struct pagevec pages_to_free;
	struct zone *zone = NULL;
	unsigned long uninitialized_var(flags);

	pagevec_init(&pages_to_free, cold);
	for (i = 0; i < nr; i++) {
		struct page *page = pages[i];

		if (unlikely(PageCompound(page))) {
			if (zone) {
				spin_unlock_irqrestore(&zone->lru_lock, flags);
				zone = NULL;
			}
			put_compound_page(page);
			continue;
		}

		if (!put_page_testzero(page))
			continue;

		if (PageLRU(page)) {
			struct zone *pagezone = page_zone(page);
			if (pagezone != zone) {
				if (zone)
					spin_unlock_irqrestore(&zone->lru_lock,
									flags);
				zone = pagezone;
				spin_lock_irqsave(&zone->lru_lock, flags);
			}
			VM_BUG_ON(!PageLRU(page));
			__ClearPageLRU(page);
			del_page_from_lru(zone, page);
		}

		if (!pagevec_add(&pages_to_free, page)) {
			if (zone) {
				spin_unlock_irqrestore(&zone->lru_lock, flags);
				zone = NULL;
			}
			__pagevec_free(&pages_to_free);
			pagevec_reinit(&pages_to_free);
  		}
	}
	if (zone)
		spin_unlock_irqrestore(&zone->lru_lock, flags);

	pagevec_free(&pages_to_free);
}

/*
 * The pages which we're about to release may be in the deferred lru-addition
 * queues.  That would prevent them from really being freed right now.  That's
 * OK from a correctness point of view but is inefficient - those pages may be
 * cache-warm and we want to give them back to the page allocator ASAP.
 *
 * So __pagevec_release() will drain those queues here.  __pagevec_lru_add()
 * and __pagevec_lru_add_active() call release_pages() directly to avoid
 * mutual recursion.
 */
void __pagevec_release(struct pagevec *pvec)
{
	lru_add_drain();
	release_pages(pvec->pages, pagevec_count(pvec), pvec->cold);
	pagevec_reinit(pvec);
}

EXPORT_SYMBOL(__pagevec_release);

/*
 * pagevec_release() for pages which are known to not be on the LRU
 *
 * This function reinitialises the caller's pagevec.
 */
void __pagevec_release_nonlru(struct pagevec *pvec)
{
	int i;
	struct pagevec pages_to_free;

	pagevec_init(&pages_to_free, pvec->cold);
	for (i = 0; i < pagevec_count(pvec); i++) {
		struct page *page = pvec->pages[i];

		VM_BUG_ON(PageLRU(page));
		if (put_page_testzero(page))
			pagevec_add(&pages_to_free, page);
	}
	pagevec_free(&pages_to_free);
	pagevec_reinit(pvec);
}

/*
 * Add the passed pages to the LRU, then drop the caller's refcount
 * on them.  Reinitialises the caller's pagevec.
 */
void __pagevec_lru_add(struct pagevec *pvec)
{
	int i;
	struct zone *zone = NULL;

	for (i = 0; i < pagevec_count(pvec); i++) {
		struct page *page = pvec->pages[i];
		struct zone *pagezone = page_zone(page);

		if (pagezone != zone) {
			if (zone)
				spin_unlock_irq(&zone->lru_lock);
			zone = pagezone;
			spin_lock_irq(&zone->lru_lock);
		}
		VM_BUG_ON(PageLRU(page));
		SetPageLRU(page);
		add_page_to_inactive_list(zone, page);
	}
	if (zone)
		spin_unlock_irq(&zone->lru_lock);
	release_pages(pvec->pages, pvec->nr, pvec->cold);
	pagevec_reinit(pvec);
}

EXPORT_SYMBOL(__pagevec_lru_add);

void __pagevec_lru_add_active(struct pagevec *pvec)
{
	int i;
	struct zone *zone = NULL;

	for (i = 0; i < pagevec_count(pvec); i++) {
		struct page *page = pvec->pages[i];
		struct zone *pagezone = page_zone(page);

		if (pagezone != zone) {
			if (zone)
				spin_unlock_irq(&zone->lru_lock);
			zone = pagezone;
			spin_lock_irq(&zone->lru_lock);
		}
		VM_BUG_ON(PageLRU(page));
		SetPageLRU(page);
		VM_BUG_ON(PageActive(page));
		SetPageActive(page);
		add_page_to_active_list(zone, page);
	}
	if (zone)
		spin_unlock_irq(&zone->lru_lock);
	release_pages(pvec->pages, pvec->nr, pvec->cold);
	pagevec_reinit(pvec);
}

/*
 * Try to drop buffers from the pages in a pagevec
 */
void pagevec_strip(struct pagevec *pvec)
{
	int i;

	for (i = 0; i < pagevec_count(pvec); i++) {
		struct page *page = pvec->pages[i];

		if (PagePrivate(page) && !TestSetPageLocked(page)) {
			if (PagePrivate(page))
				try_to_release_page(page, 0);
			unlock_page(page);
		}
	}
}

/**
 * pagevec_lookup - gang pagecache lookup
 * @pvec:	Where the resulting pages are placed
 * @mapping:	The address_space to search
 * @start:	The starting page index
 * @nr_pages:	The maximum number of pages
 *
 * pagevec_lookup() will search for and return a group of up to @nr_pages pages
 * in the mapping.  The pages are placed in @pvec.  pagevec_lookup() takes a
 * reference against the pages in @pvec.
 *
 * The search returns a group of mapping-contiguous pages with ascending
 * indexes.  There may be holes in the indices due to not-present pages.
 *
 * pagevec_lookup() returns the number of pages which were found.
 */
unsigned pagevec_lookup(struct pagevec *pvec, struct address_space *mapping,
		pgoff_t start, unsigned nr_pages)
{
	pvec->nr = find_get_pages(mapping, start, nr_pages, pvec->pages);
	return pagevec_count(pvec);
}

EXPORT_SYMBOL(pagevec_lookup);

unsigned pagevec_lookup_tag(struct pagevec *pvec, struct address_space *mapping,
		pgoff_t *index, int tag, unsigned nr_pages)
{
	pvec->nr = find_get_pages_tag(mapping, index, tag,
					nr_pages, pvec->pages);
	return pagevec_count(pvec);
}

EXPORT_SYMBOL(pagevec_lookup_tag);

#ifdef CONFIG_SMP
/*
 * We tolerate a little inaccuracy to avoid ping-ponging the counter between
 * CPUs
 */
#define ACCT_THRESHOLD	max(16, NR_CPUS * 2)

static DEFINE_PER_CPU(long, committed_space) = 0;

void vm_acct_memory(long pages)
{
	long *local;

	preempt_disable();
	local = &__get_cpu_var(committed_space);
	*local += pages;
	if (*local > ACCT_THRESHOLD || *local < -ACCT_THRESHOLD) {
		atomic_add(*local, &vm_committed_space);
		*local = 0;
	}
	preempt_enable();
}

#ifdef CONFIG_HOTPLUG_CPU

/* Drop the CPU's cached committed space back into the central pool. */
static int cpu_swap_callback(struct notifier_block *nfb,
			     unsigned long action,
			     void *hcpu)
{
	long *committed;

	committed = &per_cpu(committed_space, (long)hcpu);
	if (action == CPU_DEAD || action == CPU_DEAD_FROZEN) {
		atomic_add(*committed, &vm_committed_space);
		*committed = 0;
		drain_cpu_pagevecs((long)hcpu);
	}
	return NOTIFY_OK;
}
#endif /* CONFIG_HOTPLUG_CPU */
#endif /* CONFIG_SMP */

/*
 * Perform any setup for the swap system
 */
void __init swap_setup(void)
{
	unsigned long megs = num_physpages >> (20 - PAGE_SHIFT);

#ifdef CONFIG_SWAP
	bdi_init(swapper_space.backing_dev_info);
#endif

	/* Use a smaller cluster for small-memory machines */
	if (megs < 16)
		page_cluster = 2;
	else
		page_cluster = 3;
	/*
	 * Right now other parts of the system means that we
	 * _really_ don't want to cluster much more
	 */
#ifdef CONFIG_HOTPLUG_CPU
	hotcpu_notifier(cpu_swap_callback, 0);
#endif
}
Commit	Line	Data
1da177e4 LT	1	/*
	2	* linux/mm/swap.c
	3	*
	4	* Copyright (C) 1991, 1992, 1993, 1994 Linus Torvalds
	5	*/
	6
	7	/*
183ff22b	8	* This file contains the default values for the operation of the
1da177e4 LT	9	* Linux VM subsystem. Fine-tuning documentation can be found in
	10	* Documentation/sysctl/vm.txt.
	11	* Started 18.12.91
	12	* Swap aging added 23.2.95, Stephen Tweedie.
	13	* Buffermem limits added 12.3.98, Rik van Riel.
	14	*/
	15
	16	#include <linux/mm.h>
	17	#include <linux/sched.h>
	18	#include <linux/kernel_stat.h>
	19	#include <linux/swap.h>
	20	#include <linux/mman.h>
	21	#include <linux/pagemap.h>
	22	#include <linux/pagevec.h>
	23	#include <linux/init.h>
	24	#include <linux/module.h>
	25	#include <linux/mm_inline.h>
	26	#include <linux/buffer_head.h> /* for try_to_release_page() */
1da177e4 LT	27	#include <linux/percpu_counter.h>
	28	#include <linux/percpu.h>
	29	#include <linux/cpu.h>
	30	#include <linux/notifier.h>
e0bf68dd	31	#include <linux/backing-dev.h>
66e1707b	32	#include <linux/memcontrol.h>
1da177e4 LT	33
	34	/* How many pages do we try to swap or page in/out together? */
	35	int page_cluster;
	36
902aaed0 HH	37	static DEFINE_PER_CPU(struct pagevec, lru_add_pvecs) = { 0, };
	38	static DEFINE_PER_CPU(struct pagevec, lru_add_active_pvecs) = { 0, };
	39	static DEFINE_PER_CPU(struct pagevec, lru_rotate_pvecs) = { 0, };
	40
b221385b AB	41	/*
	42	* This path almost never happens for VM activity - pages are normally
	43	* freed via pagevecs. But it gets used by networking.
	44	*/
920c7a5d	45	static void __page_cache_release(struct page *page)
b221385b AB	46	{
	47	if (PageLRU(page)) {
	48	unsigned long flags;
	49	struct zone *zone = page_zone(page);
	50
	51	spin_lock_irqsave(&zone->lru_lock, flags);
	52	VM_BUG_ON(!PageLRU(page));
	53	__ClearPageLRU(page);
	54	del_page_from_lru(zone, page);
	55	spin_unlock_irqrestore(&zone->lru_lock, flags);
	56	}
	57	free_hot_page(page);
	58	}
	59
8519fb30	60	static void put_compound_page(struct page *page)
1da177e4	61	{
d85f3385	62	page = compound_head(page);
8519fb30	63	if (put_page_testzero(page)) {
33f2ef89	64	compound_page_dtor *dtor;
1da177e4	65
33f2ef89	66	dtor = get_compound_page_dtor(page);
8519fb30	67	(*dtor)(page);
1da177e4	68	}
8519fb30 NP	69	}
	70
	71	void put_page(struct page *page)
	72	{
	73	if (unlikely(PageCompound(page)))
	74	put_compound_page(page);
	75	else if (put_page_testzero(page))
1da177e4 LT	76	__page_cache_release(page);
	77	}
	78	EXPORT_SYMBOL(put_page);
1da177e4	79
1d7ea732	80	/**
7682486b RD	81	* put_pages_list() - release a list of pages
7682486b RD	82	* @pages: list of pages threaded on page->lru
1d7ea732 AZ	83	*
	84	* Release a list of pages which are strung together on page.lru. Currently
	85	* used by read_cache_pages() and related error recovery code.
1d7ea732 AZ	86	*/
	87	void put_pages_list(struct list_head *pages)
	88	{
	89	while (!list_empty(pages)) {
	90	struct page *victim;
	91
	92	victim = list_entry(pages->prev, struct page, lru);
	93	list_del(&victim->lru);
	94	page_cache_release(victim);
	95	}
	96	}
	97	EXPORT_SYMBOL(put_pages_list);
	98
902aaed0 HH	99	/*
	100	* pagevec_move_tail() must be called with IRQ disabled.
	101	* Otherwise this may cause nasty races.
	102	*/
	103	static void pagevec_move_tail(struct pagevec *pvec)
	104	{
	105	int i;
	106	int pgmoved = 0;
	107	struct zone *zone = NULL;
	108
	109	for (i = 0; i < pagevec_count(pvec); i++) {
	110	struct page *page = pvec->pages[i];
	111	struct zone *pagezone = page_zone(page);
	112
	113	if (pagezone != zone) {
	114	if (zone)
	115	spin_unlock(&zone->lru_lock);
	116	zone = pagezone;
	117	spin_lock(&zone->lru_lock);
	118	}
	119	if (PageLRU(page) && !PageActive(page)) {
	120	list_move_tail(&page->lru, &zone->inactive_list);
	121	pgmoved++;
	122	}
	123	}
	124	if (zone)
	125	spin_unlock(&zone->lru_lock);
	126	__count_vm_events(PGROTATED, pgmoved);
	127	release_pages(pvec->pages, pvec->nr, pvec->cold);
	128	pagevec_reinit(pvec);
	129	}
	130
1da177e4 LT	131	/*
	132	* Writeback is about to end against a page which has been marked for immediate
	133	* reclaim. If it still appears to be reclaimable, move it to the tail of the
902aaed0	134	* inactive list.
1da177e4 LT	135	*
	136	* Returns zero if it cleared PG_writeback.
	137	*/
	138	int rotate_reclaimable_page(struct page *page)
	139	{
902aaed0	140	struct pagevec *pvec;
1da177e4 LT	141	unsigned long flags;
	142
	143	if (PageLocked(page))
	144	return 1;
	145	if (PageDirty(page))
	146	return 1;
	147	if (PageActive(page))
	148	return 1;
	149	if (!PageLRU(page))
	150	return 1;
	151
902aaed0 HH	152	page_cache_get(page);
	153	local_irq_save(flags);
	154	pvec = &__get_cpu_var(lru_rotate_pvecs);
	155	if (!pagevec_add(pvec, page))
	156	pagevec_move_tail(pvec);
	157	local_irq_restore(flags);
	158
1da177e4 LT	159	if (!test_clear_page_writeback(page))
1da177e4 LT	160	BUG();
902aaed0	161
1da177e4 LT	162	return 0;
	163	}
	164
	165	/*
	166	* FIXME: speed this up?
	167	*/
920c7a5d	168	void activate_page(struct page *page)
1da177e4 LT	169	{
	170	struct zone *zone = page_zone(page);
	171
	172	spin_lock_irq(&zone->lru_lock);
	173	if (PageLRU(page) && !PageActive(page)) {
	174	del_page_from_inactive_list(zone, page);
	175	SetPageActive(page);
	176	add_page_to_active_list(zone, page);
f8891e5e	177	__count_vm_event(PGACTIVATE);
427d5416	178	mem_cgroup_move_lists(page, true);
1da177e4 LT	179	}
	180	spin_unlock_irq(&zone->lru_lock);
	181	}
	182
	183	/*
	184	* Mark a page as having seen activity.
	185	*
	186	* inactive,unreferenced -> inactive,referenced
	187	* inactive,referenced -> active,unreferenced
	188	* active,unreferenced -> active,referenced
	189	*/
920c7a5d	190	void mark_page_accessed(struct page *page)
1da177e4 LT	191	{
	192	if (!PageActive(page) && PageReferenced(page) && PageLRU(page)) {
	193	activate_page(page);
	194	ClearPageReferenced(page);
	195	} else if (!PageReferenced(page)) {
	196	SetPageReferenced(page);
	197	}
	198	}
	199
	200	EXPORT_SYMBOL(mark_page_accessed);
	201
	202	/**
	203	* lru_cache_add: add a page to the page lists
	204	* @page: the page to add
	205	*/
920c7a5d	206	void lru_cache_add(struct page *page)
1da177e4 LT	207	{
	208	struct pagevec *pvec = &get_cpu_var(lru_add_pvecs);
	209
	210	page_cache_get(page);
	211	if (!pagevec_add(pvec, page))
	212	__pagevec_lru_add(pvec);
	213	put_cpu_var(lru_add_pvecs);
	214	}
	215
920c7a5d	216	void lru_cache_add_active(struct page *page)
1da177e4 LT	217	{
	218	struct pagevec *pvec = &get_cpu_var(lru_add_active_pvecs);
	219
	220	page_cache_get(page);
	221	if (!pagevec_add(pvec, page))
	222	__pagevec_lru_add_active(pvec);
	223	put_cpu_var(lru_add_active_pvecs);
	224	}
	225
902aaed0 HH	226	/*
	227	* Drain pages out of the cpu's pagevecs.
	228	* Either "cpu" is the current CPU, and preemption has already been
	229	* disabled; or "cpu" is being hot-unplugged, and is already dead.
	230	*/
	231	static void drain_cpu_pagevecs(int cpu)
1da177e4	232	{
902aaed0	233	struct pagevec *pvec;
1da177e4	234
902aaed0	235	pvec = &per_cpu(lru_add_pvecs, cpu);
1da177e4 LT	236	if (pagevec_count(pvec))
1da177e4 LT	237	__pagevec_lru_add(pvec);
902aaed0	238
80bfed90	239	pvec = &per_cpu(lru_add_active_pvecs, cpu);
1da177e4 LT	240	if (pagevec_count(pvec))
1da177e4 LT	241	__pagevec_lru_add_active(pvec);
902aaed0 HH	242
	243	pvec = &per_cpu(lru_rotate_pvecs, cpu);
	244	if (pagevec_count(pvec)) {
	245	unsigned long flags;
	246
	247	/* No harm done if a racing interrupt already did this */
	248	local_irq_save(flags);
	249	pagevec_move_tail(pvec);
	250	local_irq_restore(flags);
	251	}
80bfed90 AM	252	}
	253
	254	void lru_add_drain(void)
	255	{
902aaed0	256	drain_cpu_pagevecs(get_cpu());
80bfed90	257	put_cpu();
1da177e4 LT	258	}
1da177e4 LT	259
053837fc	260	#ifdef CONFIG_NUMA
c4028958	261	static void lru_add_drain_per_cpu(struct work_struct *dummy)
053837fc NP	262	{
	263	lru_add_drain();
	264	}
	265
	266	/*
	267	* Returns 0 for success
	268	*/
	269	int lru_add_drain_all(void)
	270	{
c4028958	271	return schedule_on_each_cpu(lru_add_drain_per_cpu);
053837fc NP	272	}
	273
	274	#else
	275
	276	/*
	277	* Returns 0 for success
	278	*/
	279	int lru_add_drain_all(void)
	280	{
	281	lru_add_drain();
	282	return 0;
	283	}
	284	#endif
	285
1da177e4 LT	286	/*
	287	* Batched page_cache_release(). Decrement the reference count on all the
	288	* passed pages. If it fell to zero then remove the page from the LRU and
	289	* free it.
	290	*
	291	* Avoid taking zone->lru_lock if possible, but if it is taken, retain it
	292	* for the remainder of the operation.
	293	*
	294	* The locking in this function is against shrink_cache(): we recheck the
	295	* page count inside the lock to see whether shrink_cache grabbed the page
	296	* via the LRU. If it did, give up: shrink_cache will free it.
	297	*/
	298	void release_pages(struct page **pages, int nr, int cold)
	299	{
	300	int i;
	301	struct pagevec pages_to_free;
	302	struct zone *zone = NULL;
902aaed0	303	unsigned long uninitialized_var(flags);
1da177e4 LT	304
	305	pagevec_init(&pages_to_free, cold);
	306	for (i = 0; i < nr; i++) {
	307	struct page *page = pages[i];
1da177e4	308
8519fb30 NP	309	if (unlikely(PageCompound(page))) {
8519fb30 NP	310	if (zone) {
902aaed0	311	spin_unlock_irqrestore(&zone->lru_lock, flags);
8519fb30 NP	312	zone = NULL;
	313	}
	314	put_compound_page(page);
	315	continue;
	316	}
	317
b5810039	318	if (!put_page_testzero(page))
1da177e4 LT	319	continue;
1da177e4 LT	320
46453a6e NP	321	if (PageLRU(page)) {
	322	struct zone *pagezone = page_zone(page);
	323	if (pagezone != zone) {
	324	if (zone)
902aaed0 HH	325	spin_unlock_irqrestore(&zone->lru_lock,
902aaed0 HH	326	flags);
46453a6e	327	zone = pagezone;
902aaed0	328	spin_lock_irqsave(&zone->lru_lock, flags);
46453a6e	329	}
725d704e	330	VM_BUG_ON(!PageLRU(page));
67453911	331	__ClearPageLRU(page);
1da177e4	332	del_page_from_lru(zone, page);
46453a6e NP	333	}
	334
	335	if (!pagevec_add(&pages_to_free, page)) {
	336	if (zone) {
902aaed0	337	spin_unlock_irqrestore(&zone->lru_lock, flags);
46453a6e	338	zone = NULL;
1da177e4	339	}
46453a6e NP	340	__pagevec_free(&pages_to_free);
	341	pagevec_reinit(&pages_to_free);
	342	}
1da177e4 LT	343	}
1da177e4 LT	344	if (zone)
902aaed0	345	spin_unlock_irqrestore(&zone->lru_lock, flags);
1da177e4 LT	346
	347	pagevec_free(&pages_to_free);
	348	}
	349
	350	/*
	351	* The pages which we're about to release may be in the deferred lru-addition
	352	* queues. That would prevent them from really being freed right now. That's
	353	* OK from a correctness point of view but is inefficient - those pages may be
	354	* cache-warm and we want to give them back to the page allocator ASAP.
	355	*
	356	* So __pagevec_release() will drain those queues here. __pagevec_lru_add()
	357	* and __pagevec_lru_add_active() call release_pages() directly to avoid
	358	* mutual recursion.
	359	*/
	360	void __pagevec_release(struct pagevec *pvec)
	361	{
	362	lru_add_drain();
	363	release_pages(pvec->pages, pagevec_count(pvec), pvec->cold);
	364	pagevec_reinit(pvec);
	365	}
	366
7f285701 SF	367	EXPORT_SYMBOL(__pagevec_release);
7f285701 SF	368
1da177e4 LT	369	/*
	370	* pagevec_release() for pages which are known to not be on the LRU
	371	*
	372	* This function reinitialises the caller's pagevec.
	373	*/
	374	void __pagevec_release_nonlru(struct pagevec *pvec)
	375	{
	376	int i;
	377	struct pagevec pages_to_free;
	378
	379	pagevec_init(&pages_to_free, pvec->cold);
1da177e4 LT	380	for (i = 0; i < pagevec_count(pvec); i++) {
	381	struct page *page = pvec->pages[i];
	382
725d704e	383	VM_BUG_ON(PageLRU(page));
1da177e4 LT	384	if (put_page_testzero(page))
	385	pagevec_add(&pages_to_free, page);
	386	}
	387	pagevec_free(&pages_to_free);
	388	pagevec_reinit(pvec);
	389	}
	390
	391	/*
	392	* Add the passed pages to the LRU, then drop the caller's refcount
	393	* on them. Reinitialises the caller's pagevec.
	394	*/
	395	void __pagevec_lru_add(struct pagevec *pvec)
	396	{
	397	int i;
	398	struct zone *zone = NULL;
	399
	400	for (i = 0; i < pagevec_count(pvec); i++) {
	401	struct page *page = pvec->pages[i];
	402	struct zone *pagezone = page_zone(page);
	403
	404	if (pagezone != zone) {
	405	if (zone)
	406	spin_unlock_irq(&zone->lru_lock);
	407	zone = pagezone;
	408	spin_lock_irq(&zone->lru_lock);
	409	}
725d704e	410	VM_BUG_ON(PageLRU(page));
8d438f96	411	SetPageLRU(page);
1da177e4 LT	412	add_page_to_inactive_list(zone, page);
	413	}
	414	if (zone)
	415	spin_unlock_irq(&zone->lru_lock);
	416	release_pages(pvec->pages, pvec->nr, pvec->cold);
	417	pagevec_reinit(pvec);
	418	}
	419
	420	EXPORT_SYMBOL(__pagevec_lru_add);
	421
	422	void __pagevec_lru_add_active(struct pagevec *pvec)
	423	{
	424	int i;
	425	struct zone *zone = NULL;
	426
	427	for (i = 0; i < pagevec_count(pvec); i++) {
	428	struct page *page = pvec->pages[i];
	429	struct zone *pagezone = page_zone(page);
	430
	431	if (pagezone != zone) {
	432	if (zone)
	433	spin_unlock_irq(&zone->lru_lock);
	434	zone = pagezone;
	435	spin_lock_irq(&zone->lru_lock);
	436	}
725d704e	437	VM_BUG_ON(PageLRU(page));
8d438f96	438	SetPageLRU(page);
725d704e	439	VM_BUG_ON(PageActive(page));
4c84cacf	440	SetPageActive(page);
1da177e4 LT	441	add_page_to_active_list(zone, page);
	442	}
	443	if (zone)
	444	spin_unlock_irq(&zone->lru_lock);
	445	release_pages(pvec->pages, pvec->nr, pvec->cold);
	446	pagevec_reinit(pvec);
	447	}
	448
	449	/*
	450	* Try to drop buffers from the pages in a pagevec
	451	*/
	452	void pagevec_strip(struct pagevec *pvec)
	453	{
	454	int i;
	455
	456	for (i = 0; i < pagevec_count(pvec); i++) {
	457	struct page *page = pvec->pages[i];
	458
	459	if (PagePrivate(page) && !TestSetPageLocked(page)) {
5b40dc78 CL	460	if (PagePrivate(page))
5b40dc78 CL	461	try_to_release_page(page, 0);
1da177e4 LT	462	unlock_page(page);
	463	}
	464	}
	465	}
	466
	467	/**
	468	* pagevec_lookup - gang pagecache lookup
	469	* @pvec: Where the resulting pages are placed
	470	* @mapping: The address_space to search
	471	* @start: The starting page index
	472	* @nr_pages: The maximum number of pages
	473	*
	474	* pagevec_lookup() will search for and return a group of up to @nr_pages pages
	475	* in the mapping. The pages are placed in @pvec. pagevec_lookup() takes a
	476	* reference against the pages in @pvec.
	477	*
	478	* The search returns a group of mapping-contiguous pages with ascending
	479	* indexes. There may be holes in the indices due to not-present pages.
	480	*
	481	* pagevec_lookup() returns the number of pages which were found.
	482	*/
	483	unsigned pagevec_lookup(struct pagevec pvec, struct address_space mapping,
	484	pgoff_t start, unsigned nr_pages)
	485	{
	486	pvec->nr = find_get_pages(mapping, start, nr_pages, pvec->pages);
	487	return pagevec_count(pvec);
	488	}
	489
78539fdf CH	490	EXPORT_SYMBOL(pagevec_lookup);
78539fdf CH	491
1da177e4 LT	492	unsigned pagevec_lookup_tag(struct pagevec pvec, struct address_space mapping,
	493	pgoff_t *index, int tag, unsigned nr_pages)
	494	{
	495	pvec->nr = find_get_pages_tag(mapping, index, tag,
	496	nr_pages, pvec->pages);
	497	return pagevec_count(pvec);
	498	}
	499
7f285701	500	EXPORT_SYMBOL(pagevec_lookup_tag);
1da177e4 LT	501
	502	#ifdef CONFIG_SMP
	503	/*
	504	* We tolerate a little inaccuracy to avoid ping-ponging the counter between
	505	* CPUs
	506	*/
	507	#define ACCT_THRESHOLD max(16, NR_CPUS * 2)
	508
	509	static DEFINE_PER_CPU(long, committed_space) = 0;
	510
	511	void vm_acct_memory(long pages)
	512	{
	513	long *local;
	514
	515	preempt_disable();
	516	local = &__get_cpu_var(committed_space);
	517	*local += pages;
	518	if (local > ACCT_THRESHOLD \|\| local < -ACCT_THRESHOLD) {
	519	atomic_add(*local, &vm_committed_space);
	520	*local = 0;
	521	}
	522	preempt_enable();
	523	}
1da177e4 LT	524
1da177e4 LT	525	#ifdef CONFIG_HOTPLUG_CPU
1da177e4 LT	526
	527	/* Drop the CPU's cached committed space back into the central pool. */
	528	static int cpu_swap_callback(struct notifier_block *nfb,
	529	unsigned long action,
	530	void *hcpu)
	531	{
	532	long *committed;
	533
	534	committed = &per_cpu(committed_space, (long)hcpu);
8bb78442	535	if (action == CPU_DEAD \|\| action == CPU_DEAD_FROZEN) {
1da177e4 LT	536	atomic_add(*committed, &vm_committed_space);
1da177e4 LT	537	*committed = 0;
902aaed0	538	drain_cpu_pagevecs((long)hcpu);
1da177e4 LT	539	}
	540	return NOTIFY_OK;
	541	}
	542	#endif /* CONFIG_HOTPLUG_CPU */
	543	#endif /* CONFIG_SMP */
	544
1da177e4 LT	545	/*
	546	* Perform any setup for the swap system
	547	*/
	548	void __init swap_setup(void)
	549	{
	550	unsigned long megs = num_physpages >> (20 - PAGE_SHIFT);
	551
e0bf68dd PZ	552	#ifdef CONFIG_SWAP
	553	bdi_init(swapper_space.backing_dev_info);
	554	#endif
	555
1da177e4 LT	556	/* Use a smaller cluster for small-memory machines */
	557	if (megs < 16)
	558	page_cluster = 2;
	559	else
	560	page_cluster = 3;
	561	/*
	562	* Right now other parts of the system means that we
	563	* _really_ don't want to cluster much more
	564	*/
02316067	565	#ifdef CONFIG_HOTPLUG_CPU
1da177e4	566	hotcpu_notifier(cpu_swap_callback, 0);
02316067	567	#endif
1da177e4	568	}