[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
 *
 * Release a list of pages which are strung together on page.lru.  Currently
 * used by read_cache_pages() and related error recovery code.
 *
 * @pages: list of pages threaded on page->lru
 */
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 AZ	80	/**
	81	* put_pages_list(): release a list of pages
	82	*
	83	* Release a list of pages which are strung together on page.lru. Currently
	84	* used by read_cache_pages() and related error recovery code.
	85	*
	86	* @pages: list of pages threaded on page->lru
	87	*/
	88	void put_pages_list(struct list_head *pages)
	89	{
	90	while (!list_empty(pages)) {
	91	struct page *victim;
	92
	93	victim = list_entry(pages->prev, struct page, lru);
	94	list_del(&victim->lru);
	95	page_cache_release(victim);
	96	}
	97	}
	98	EXPORT_SYMBOL(put_pages_list);
	99
902aaed0 HH	100	/*
	101	* pagevec_move_tail() must be called with IRQ disabled.
	102	* Otherwise this may cause nasty races.
	103	*/
	104	static void pagevec_move_tail(struct pagevec *pvec)
	105	{
	106	int i;
	107	int pgmoved = 0;
	108	struct zone *zone = NULL;
	109
	110	for (i = 0; i < pagevec_count(pvec); i++) {
	111	struct page *page = pvec->pages[i];
	112	struct zone *pagezone = page_zone(page);
	113
	114	if (pagezone != zone) {
	115	if (zone)
	116	spin_unlock(&zone->lru_lock);
	117	zone = pagezone;
	118	spin_lock(&zone->lru_lock);
	119	}
	120	if (PageLRU(page) && !PageActive(page)) {
	121	list_move_tail(&page->lru, &zone->inactive_list);
	122	pgmoved++;
	123	}
	124	}
	125	if (zone)
	126	spin_unlock(&zone->lru_lock);
	127	__count_vm_events(PGROTATED, pgmoved);
	128	release_pages(pvec->pages, pvec->nr, pvec->cold);
	129	pagevec_reinit(pvec);
	130	}
	131
1da177e4 LT	132	/*
	133	* Writeback is about to end against a page which has been marked for immediate
	134	* reclaim. If it still appears to be reclaimable, move it to the tail of the
902aaed0	135	* inactive list.
1da177e4 LT	136	*
	137	* Returns zero if it cleared PG_writeback.
	138	*/
	139	int rotate_reclaimable_page(struct page *page)
	140	{
902aaed0	141	struct pagevec *pvec;
1da177e4 LT	142	unsigned long flags;
	143
	144	if (PageLocked(page))
	145	return 1;
	146	if (PageDirty(page))
	147	return 1;
	148	if (PageActive(page))
	149	return 1;
	150	if (!PageLRU(page))
	151	return 1;
	152
902aaed0 HH	153	page_cache_get(page);
	154	local_irq_save(flags);
	155	pvec = &__get_cpu_var(lru_rotate_pvecs);
	156	if (!pagevec_add(pvec, page))
	157	pagevec_move_tail(pvec);
	158	local_irq_restore(flags);
	159
1da177e4 LT	160	if (!test_clear_page_writeback(page))
1da177e4 LT	161	BUG();
902aaed0	162
1da177e4 LT	163	return 0;
	164	}
	165
	166	/*
	167	* FIXME: speed this up?
	168	*/
920c7a5d	169	void activate_page(struct page *page)
1da177e4 LT	170	{
	171	struct zone *zone = page_zone(page);
	172
	173	spin_lock_irq(&zone->lru_lock);
	174	if (PageLRU(page) && !PageActive(page)) {
	175	del_page_from_inactive_list(zone, page);
	176	SetPageActive(page);
	177	add_page_to_active_list(zone, page);
f8891e5e	178	__count_vm_event(PGACTIVATE);
427d5416	179	mem_cgroup_move_lists(page, true);
1da177e4 LT	180	}
	181	spin_unlock_irq(&zone->lru_lock);
	182	}
	183
	184	/*
	185	* Mark a page as having seen activity.
	186	*
	187	* inactive,unreferenced -> inactive,referenced
	188	* inactive,referenced -> active,unreferenced
	189	* active,unreferenced -> active,referenced
	190	*/
920c7a5d	191	void mark_page_accessed(struct page *page)
1da177e4 LT	192	{
	193	if (!PageActive(page) && PageReferenced(page) && PageLRU(page)) {
	194	activate_page(page);
	195	ClearPageReferenced(page);
	196	} else if (!PageReferenced(page)) {
	197	SetPageReferenced(page);
	198	}
	199	}
	200
	201	EXPORT_SYMBOL(mark_page_accessed);
	202
	203	/**
	204	* lru_cache_add: add a page to the page lists
	205	* @page: the page to add
	206	*/
920c7a5d	207	void lru_cache_add(struct page *page)
1da177e4 LT	208	{
	209	struct pagevec *pvec = &get_cpu_var(lru_add_pvecs);
	210
	211	page_cache_get(page);
	212	if (!pagevec_add(pvec, page))
	213	__pagevec_lru_add(pvec);
	214	put_cpu_var(lru_add_pvecs);
	215	}
	216
920c7a5d	217	void lru_cache_add_active(struct page *page)
1da177e4 LT	218	{
	219	struct pagevec *pvec = &get_cpu_var(lru_add_active_pvecs);
	220
	221	page_cache_get(page);
	222	if (!pagevec_add(pvec, page))
	223	__pagevec_lru_add_active(pvec);
	224	put_cpu_var(lru_add_active_pvecs);
	225	}
	226
902aaed0 HH	227	/*
	228	* Drain pages out of the cpu's pagevecs.
	229	* Either "cpu" is the current CPU, and preemption has already been
	230	* disabled; or "cpu" is being hot-unplugged, and is already dead.
	231	*/
	232	static void drain_cpu_pagevecs(int cpu)
1da177e4	233	{
902aaed0	234	struct pagevec *pvec;
1da177e4	235
902aaed0	236	pvec = &per_cpu(lru_add_pvecs, cpu);
1da177e4 LT	237	if (pagevec_count(pvec))
1da177e4 LT	238	__pagevec_lru_add(pvec);
902aaed0	239
80bfed90	240	pvec = &per_cpu(lru_add_active_pvecs, cpu);
1da177e4 LT	241	if (pagevec_count(pvec))
1da177e4 LT	242	__pagevec_lru_add_active(pvec);
902aaed0 HH	243
	244	pvec = &per_cpu(lru_rotate_pvecs, cpu);
	245	if (pagevec_count(pvec)) {
	246	unsigned long flags;
	247
	248	/* No harm done if a racing interrupt already did this */
	249	local_irq_save(flags);
	250	pagevec_move_tail(pvec);
	251	local_irq_restore(flags);
	252	}
80bfed90 AM	253	}
	254
	255	void lru_add_drain(void)
	256	{
902aaed0	257	drain_cpu_pagevecs(get_cpu());
80bfed90	258	put_cpu();
1da177e4 LT	259	}
1da177e4 LT	260
053837fc	261	#ifdef CONFIG_NUMA
c4028958	262	static void lru_add_drain_per_cpu(struct work_struct *dummy)
053837fc NP	263	{
	264	lru_add_drain();
	265	}
	266
	267	/*
	268	* Returns 0 for success
	269	*/
	270	int lru_add_drain_all(void)
	271	{
c4028958	272	return schedule_on_each_cpu(lru_add_drain_per_cpu);
053837fc NP	273	}
	274
	275	#else
	276
	277	/*
	278	* Returns 0 for success
	279	*/
	280	int lru_add_drain_all(void)
	281	{
	282	lru_add_drain();
	283	return 0;
	284	}
	285	#endif
	286
1da177e4 LT	287	/*
	288	* Batched page_cache_release(). Decrement the reference count on all the
	289	* passed pages. If it fell to zero then remove the page from the LRU and
	290	* free it.
	291	*
	292	* Avoid taking zone->lru_lock if possible, but if it is taken, retain it
	293	* for the remainder of the operation.
	294	*
	295	* The locking in this function is against shrink_cache(): we recheck the
	296	* page count inside the lock to see whether shrink_cache grabbed the page
	297	* via the LRU. If it did, give up: shrink_cache will free it.
	298	*/
	299	void release_pages(struct page **pages, int nr, int cold)
	300	{
	301	int i;
	302	struct pagevec pages_to_free;
	303	struct zone *zone = NULL;
902aaed0	304	unsigned long uninitialized_var(flags);
1da177e4 LT	305
	306	pagevec_init(&pages_to_free, cold);
	307	for (i = 0; i < nr; i++) {
	308	struct page *page = pages[i];
1da177e4	309
8519fb30 NP	310	if (unlikely(PageCompound(page))) {
8519fb30 NP	311	if (zone) {
902aaed0	312	spin_unlock_irqrestore(&zone->lru_lock, flags);
8519fb30 NP	313	zone = NULL;
	314	}
	315	put_compound_page(page);
	316	continue;
	317	}
	318
b5810039	319	if (!put_page_testzero(page))
1da177e4 LT	320	continue;
1da177e4 LT	321
46453a6e NP	322	if (PageLRU(page)) {
	323	struct zone *pagezone = page_zone(page);
	324	if (pagezone != zone) {
	325	if (zone)
902aaed0 HH	326	spin_unlock_irqrestore(&zone->lru_lock,
902aaed0 HH	327	flags);
46453a6e	328	zone = pagezone;
902aaed0	329	spin_lock_irqsave(&zone->lru_lock, flags);
46453a6e	330	}
725d704e	331	VM_BUG_ON(!PageLRU(page));
67453911	332	__ClearPageLRU(page);
1da177e4	333	del_page_from_lru(zone, page);
46453a6e NP	334	}
	335
	336	if (!pagevec_add(&pages_to_free, page)) {
	337	if (zone) {
902aaed0	338	spin_unlock_irqrestore(&zone->lru_lock, flags);
46453a6e	339	zone = NULL;
1da177e4	340	}
46453a6e NP	341	__pagevec_free(&pages_to_free);
	342	pagevec_reinit(&pages_to_free);
	343	}
1da177e4 LT	344	}
1da177e4 LT	345	if (zone)
902aaed0	346	spin_unlock_irqrestore(&zone->lru_lock, flags);
1da177e4 LT	347
	348	pagevec_free(&pages_to_free);
	349	}
	350
	351	/*
	352	* The pages which we're about to release may be in the deferred lru-addition
	353	* queues. That would prevent them from really being freed right now. That's
	354	* OK from a correctness point of view but is inefficient - those pages may be
	355	* cache-warm and we want to give them back to the page allocator ASAP.
	356	*
	357	* So __pagevec_release() will drain those queues here. __pagevec_lru_add()
	358	* and __pagevec_lru_add_active() call release_pages() directly to avoid
	359	* mutual recursion.
	360	*/
	361	void __pagevec_release(struct pagevec *pvec)
	362	{
	363	lru_add_drain();
	364	release_pages(pvec->pages, pagevec_count(pvec), pvec->cold);
	365	pagevec_reinit(pvec);
	366	}
	367
7f285701 SF	368	EXPORT_SYMBOL(__pagevec_release);
7f285701 SF	369
1da177e4 LT	370	/*
	371	* pagevec_release() for pages which are known to not be on the LRU
	372	*
	373	* This function reinitialises the caller's pagevec.
	374	*/
	375	void __pagevec_release_nonlru(struct pagevec *pvec)
	376	{
	377	int i;
	378	struct pagevec pages_to_free;
	379
	380	pagevec_init(&pages_to_free, pvec->cold);
1da177e4 LT	381	for (i = 0; i < pagevec_count(pvec); i++) {
	382	struct page *page = pvec->pages[i];
	383
725d704e	384	VM_BUG_ON(PageLRU(page));
1da177e4 LT	385	if (put_page_testzero(page))
	386	pagevec_add(&pages_to_free, page);
	387	}
	388	pagevec_free(&pages_to_free);
	389	pagevec_reinit(pvec);
	390	}
	391
	392	/*
	393	* Add the passed pages to the LRU, then drop the caller's refcount
	394	* on them. Reinitialises the caller's pagevec.
	395	*/
	396	void __pagevec_lru_add(struct pagevec *pvec)
	397	{
	398	int i;
	399	struct zone *zone = NULL;
	400
	401	for (i = 0; i < pagevec_count(pvec); i++) {
	402	struct page *page = pvec->pages[i];
	403	struct zone *pagezone = page_zone(page);
	404
	405	if (pagezone != zone) {
	406	if (zone)
	407	spin_unlock_irq(&zone->lru_lock);
	408	zone = pagezone;
	409	spin_lock_irq(&zone->lru_lock);
	410	}
725d704e	411	VM_BUG_ON(PageLRU(page));
8d438f96	412	SetPageLRU(page);
1da177e4 LT	413	add_page_to_inactive_list(zone, page);
	414	}
	415	if (zone)
	416	spin_unlock_irq(&zone->lru_lock);
	417	release_pages(pvec->pages, pvec->nr, pvec->cold);
	418	pagevec_reinit(pvec);
	419	}
	420
	421	EXPORT_SYMBOL(__pagevec_lru_add);
	422
	423	void __pagevec_lru_add_active(struct pagevec *pvec)
	424	{
	425	int i;
	426	struct zone *zone = NULL;
	427
	428	for (i = 0; i < pagevec_count(pvec); i++) {
	429	struct page *page = pvec->pages[i];
	430	struct zone *pagezone = page_zone(page);
	431
	432	if (pagezone != zone) {
	433	if (zone)
	434	spin_unlock_irq(&zone->lru_lock);
	435	zone = pagezone;
	436	spin_lock_irq(&zone->lru_lock);
	437	}
725d704e	438	VM_BUG_ON(PageLRU(page));
8d438f96	439	SetPageLRU(page);
725d704e	440	VM_BUG_ON(PageActive(page));
4c84cacf	441	SetPageActive(page);
1da177e4 LT	442	add_page_to_active_list(zone, page);
	443	}
	444	if (zone)
	445	spin_unlock_irq(&zone->lru_lock);
	446	release_pages(pvec->pages, pvec->nr, pvec->cold);
	447	pagevec_reinit(pvec);
	448	}
	449
	450	/*
	451	* Try to drop buffers from the pages in a pagevec
	452	*/
	453	void pagevec_strip(struct pagevec *pvec)
	454	{
	455	int i;
	456
	457	for (i = 0; i < pagevec_count(pvec); i++) {
	458	struct page *page = pvec->pages[i];
	459
	460	if (PagePrivate(page) && !TestSetPageLocked(page)) {
5b40dc78 CL	461	if (PagePrivate(page))
5b40dc78 CL	462	try_to_release_page(page, 0);
1da177e4 LT	463	unlock_page(page);
	464	}
	465	}
	466	}
	467
	468	/**
	469	* pagevec_lookup - gang pagecache lookup
	470	* @pvec: Where the resulting pages are placed
	471	* @mapping: The address_space to search
	472	* @start: The starting page index
	473	* @nr_pages: The maximum number of pages
	474	*
	475	* pagevec_lookup() will search for and return a group of up to @nr_pages pages
	476	* in the mapping. The pages are placed in @pvec. pagevec_lookup() takes a
	477	* reference against the pages in @pvec.
	478	*
	479	* The search returns a group of mapping-contiguous pages with ascending
	480	* indexes. There may be holes in the indices due to not-present pages.
	481	*
	482	* pagevec_lookup() returns the number of pages which were found.
	483	*/
	484	unsigned pagevec_lookup(struct pagevec pvec, struct address_space mapping,
	485	pgoff_t start, unsigned nr_pages)
	486	{
	487	pvec->nr = find_get_pages(mapping, start, nr_pages, pvec->pages);
	488	return pagevec_count(pvec);
	489	}
	490
78539fdf CH	491	EXPORT_SYMBOL(pagevec_lookup);
78539fdf CH	492
1da177e4 LT	493	unsigned pagevec_lookup_tag(struct pagevec pvec, struct address_space mapping,
	494	pgoff_t *index, int tag, unsigned nr_pages)
	495	{
	496	pvec->nr = find_get_pages_tag(mapping, index, tag,
	497	nr_pages, pvec->pages);
	498	return pagevec_count(pvec);
	499	}
	500
7f285701	501	EXPORT_SYMBOL(pagevec_lookup_tag);
1da177e4 LT	502
	503	#ifdef CONFIG_SMP
	504	/*
	505	* We tolerate a little inaccuracy to avoid ping-ponging the counter between
	506	* CPUs
	507	*/
	508	#define ACCT_THRESHOLD max(16, NR_CPUS * 2)
	509
	510	static DEFINE_PER_CPU(long, committed_space) = 0;
	511
	512	void vm_acct_memory(long pages)
	513	{
	514	long *local;
	515
	516	preempt_disable();
	517	local = &__get_cpu_var(committed_space);
	518	*local += pages;
	519	if (local > ACCT_THRESHOLD \|\| local < -ACCT_THRESHOLD) {
	520	atomic_add(*local, &vm_committed_space);
	521	*local = 0;
	522	}
	523	preempt_enable();
	524	}
1da177e4 LT	525
1da177e4 LT	526	#ifdef CONFIG_HOTPLUG_CPU
1da177e4 LT	527
	528	/* Drop the CPU's cached committed space back into the central pool. */
	529	static int cpu_swap_callback(struct notifier_block *nfb,
	530	unsigned long action,
	531	void *hcpu)
	532	{
	533	long *committed;
	534
	535	committed = &per_cpu(committed_space, (long)hcpu);
8bb78442	536	if (action == CPU_DEAD \|\| action == CPU_DEAD_FROZEN) {
1da177e4 LT	537	atomic_add(*committed, &vm_committed_space);
1da177e4 LT	538	*committed = 0;
902aaed0	539	drain_cpu_pagevecs((long)hcpu);
1da177e4 LT	540	}
	541	return NOTIFY_OK;
	542	}
	543	#endif /* CONFIG_HOTPLUG_CPU */
	544	#endif /* CONFIG_SMP */
	545
1da177e4 LT	546	/*
	547	* Perform any setup for the swap system
	548	*/
	549	void __init swap_setup(void)
	550	{
	551	unsigned long megs = num_physpages >> (20 - PAGE_SHIFT);
	552
e0bf68dd PZ	553	#ifdef CONFIG_SWAP
	554	bdi_init(swapper_space.backing_dev_info);
	555	#endif
	556
1da177e4 LT	557	/* Use a smaller cluster for small-memory machines */
	558	if (megs < 16)
	559	page_cluster = 2;
	560	else
	561	page_cluster = 3;
	562	/*
	563	* Right now other parts of the system means that we
	564	* _really_ don't want to cluster much more
	565	*/
02316067	566	#ifdef CONFIG_HOTPLUG_CPU
1da177e4	567	hotcpu_notifier(cpu_swap_callback, 0);
02316067	568	#endif
1da177e4	569	}