[linux.git] / mm / swap_state.c

/*
 *  linux/mm/swap_state.c
 *
 *  Copyright (C) 1991, 1992, 1993, 1994  Linus Torvalds
 *  Swap reorganised 29.12.95, Stephen Tweedie
 *
 *  Rewritten to use page cache, (C) 1998 Stephen Tweedie
 */
#include <linux/mm.h>
#include <linux/gfp.h>
#include <linux/kernel_stat.h>
#include <linux/swap.h>
#include <linux/swapops.h>
#include <linux/init.h>
#include <linux/pagemap.h>
#include <linux/backing-dev.h>
#include <linux/blkdev.h>
#include <linux/pagevec.h>
#include <linux/migrate.h>
#include <linux/page_cgroup.h>

#include <asm/pgtable.h>

/*
 * swapper_space is a fiction, retained to simplify the path through
 * vmscan's shrink_page_list.
 */
static const struct address_space_operations swap_aops = {
	.writepage	= swap_writepage,
	.set_page_dirty	= swap_set_page_dirty,
	.migratepage	= migrate_page,
};

static struct backing_dev_info swap_backing_dev_info = {
	.name		= "swap",
	.capabilities	= BDI_CAP_NO_ACCT_AND_WRITEBACK | BDI_CAP_SWAP_BACKED,
};

struct address_space swapper_spaces[MAX_SWAPFILES] = {
	[0 ... MAX_SWAPFILES - 1] = {
		.page_tree	= RADIX_TREE_INIT(GFP_ATOMIC|__GFP_NOWARN),
		.a_ops		= &swap_aops,
		.backing_dev_info = &swap_backing_dev_info,
	}
};

#define INC_CACHE_INFO(x)	do { swap_cache_info.x++; } while (0)

static struct {
	unsigned long add_total;
	unsigned long del_total;
	unsigned long find_success;
	unsigned long find_total;
} swap_cache_info;

unsigned long total_swapcache_pages(void)
{
	int i;
	unsigned long ret = 0;

	for (i = 0; i < MAX_SWAPFILES; i++)
		ret += swapper_spaces[i].nrpages;
	return ret;
}

void show_swap_cache_info(void)
{
	printk("%lu pages in swap cache\n", total_swapcache_pages());
	printk("Swap cache stats: add %lu, delete %lu, find %lu/%lu\n",
		swap_cache_info.add_total, swap_cache_info.del_total,
		swap_cache_info.find_success, swap_cache_info.find_total);
	printk("Free swap  = %ldkB\n",
		get_nr_swap_pages() << (PAGE_SHIFT - 10));
	printk("Total swap = %lukB\n", total_swap_pages << (PAGE_SHIFT - 10));
}

/*
 * __add_to_swap_cache resembles add_to_page_cache_locked on swapper_space,
 * but sets SwapCache flag and private instead of mapping and index.
 */
int __add_to_swap_cache(struct page *page, swp_entry_t entry)
{
	int error;
	struct address_space *address_space;

	VM_BUG_ON(!PageLocked(page));
	VM_BUG_ON(PageSwapCache(page));
	VM_BUG_ON(!PageSwapBacked(page));

	page_cache_get(page);
	SetPageSwapCache(page);
	set_page_private(page, entry.val);

	address_space = swap_address_space(entry);
	spin_lock_irq(&address_space->tree_lock);
	error = radix_tree_insert(&address_space->page_tree,
					entry.val, page);
	if (likely(!error)) {
		address_space->nrpages++;
		__inc_zone_page_state(page, NR_FILE_PAGES);
		INC_CACHE_INFO(add_total);
	}
	spin_unlock_irq(&address_space->tree_lock);

	if (unlikely(error)) {
		/*
		 * Only the context which have set SWAP_HAS_CACHE flag
		 * would call add_to_swap_cache().
		 * So add_to_swap_cache() doesn't returns -EEXIST.
		 */
		VM_BUG_ON(error == -EEXIST);
		set_page_private(page, 0UL);
		ClearPageSwapCache(page);
		page_cache_release(page);
	}

	return error;
}


int add_to_swap_cache(struct page *page, swp_entry_t entry, gfp_t gfp_mask)
{
	int error;

	error = radix_tree_maybe_preload(gfp_mask);
	if (!error) {
		error = __add_to_swap_cache(page, entry);
		radix_tree_preload_end();
	}
	return error;
}

/*
 * This must be called only on pages that have
 * been verified to be in the swap cache.
 */
void __delete_from_swap_cache(struct page *page)
{
	swp_entry_t entry;
	struct address_space *address_space;

	VM_BUG_ON(!PageLocked(page));
	VM_BUG_ON(!PageSwapCache(page));
	VM_BUG_ON(PageWriteback(page));

	entry.val = page_private(page);
	address_space = swap_address_space(entry);
	radix_tree_delete(&address_space->page_tree, page_private(page));
	set_page_private(page, 0);
	ClearPageSwapCache(page);
	address_space->nrpages--;
	__dec_zone_page_state(page, NR_FILE_PAGES);
	INC_CACHE_INFO(del_total);
}

/**
 * add_to_swap - allocate swap space for a page
 * @page: page we want to move to swap
 *
 * Allocate swap space for the page and add the page to the
 * swap cache.  Caller needs to hold the page lock. 
 */
int add_to_swap(struct page *page, struct list_head *list)
{
	swp_entry_t entry;
	int err;

	VM_BUG_ON(!PageLocked(page));
	VM_BUG_ON(!PageUptodate(page));

	entry = get_swap_page();
	if (!entry.val)
		return 0;

	if (unlikely(PageTransHuge(page)))
		if (unlikely(split_huge_page_to_list(page, list))) {
			swapcache_free(entry, NULL);
			return 0;
		}

	/*
	 * Radix-tree node allocations from PF_MEMALLOC contexts could
	 * completely exhaust the page allocator. __GFP_NOMEMALLOC
	 * stops emergency reserves from being allocated.
	 *
	 * TODO: this could cause a theoretical memory reclaim
	 * deadlock in the swap out path.
	 */
	/*
	 * Add it to the swap cache and mark it dirty
	 */
	err = add_to_swap_cache(page, entry,
			__GFP_HIGH|__GFP_NOMEMALLOC|__GFP_NOWARN);

	if (!err) {	/* Success */
		SetPageDirty(page);
		return 1;
	} else {	/* -ENOMEM radix-tree allocation failure */
		/*
		 * add_to_swap_cache() doesn't return -EEXIST, so we can safely
		 * clear SWAP_HAS_CACHE flag.
		 */
		swapcache_free(entry, NULL);
		return 0;
	}
}

/*
 * This must be called only on pages that have
 * been verified to be in the swap cache and locked.
 * It will never put the page into the free list,
 * the caller has a reference on the page.
 */
void delete_from_swap_cache(struct page *page)
{
	swp_entry_t entry;
	struct address_space *address_space;

	entry.val = page_private(page);

	address_space = swap_address_space(entry);
	spin_lock_irq(&address_space->tree_lock);
	__delete_from_swap_cache(page);
	spin_unlock_irq(&address_space->tree_lock);

	swapcache_free(entry, page);
	page_cache_release(page);
}

/* 
 * If we are the only user, then try to free up the swap cache. 
 * 
 * Its ok to check for PageSwapCache without the page lock
 * here because we are going to recheck again inside
 * try_to_free_swap() _with_ the lock.
 * 					- Marcelo
 */
static inline void free_swap_cache(struct page *page)
{
	if (PageSwapCache(page) && !page_mapped(page) && trylock_page(page)) {
		try_to_free_swap(page);
		unlock_page(page);
	}
}

/* 
 * Perform a free_page(), also freeing any swap cache associated with
 * this page if it is the last user of the page.
 */
void free_page_and_swap_cache(struct page *page)
{
	free_swap_cache(page);
	page_cache_release(page);
}

/*
 * Passed an array of pages, drop them all from swapcache and then release
 * them.  They are removed from the LRU and freed if this is their last use.
 */
void free_pages_and_swap_cache(struct page **pages, int nr)
{
	struct page **pagep = pages;

	lru_add_drain();
	while (nr) {
		int todo = min(nr, PAGEVEC_SIZE);
		int i;

		for (i = 0; i < todo; i++)
			free_swap_cache(pagep[i]);
		release_pages(pagep, todo, 0);
		pagep += todo;
		nr -= todo;
	}
}

/*
 * Lookup a swap entry in the swap cache. A found page will be returned
 * unlocked and with its refcount incremented - we rely on the kernel
 * lock getting page table operations atomic even if we drop the page
 * lock before returning.
 */
struct page * lookup_swap_cache(swp_entry_t entry)
{
	struct page *page;

	page = find_get_page(swap_address_space(entry), entry.val);

	if (page)
		INC_CACHE_INFO(find_success);

	INC_CACHE_INFO(find_total);
	return page;
}

/* 
 * Locate a page of swap in physical memory, reserving swap cache space
 * and reading the disk if it is not already cached.
 * A failure return means that either the page allocation failed or that
 * the swap entry is no longer in use.
 */
struct page *read_swap_cache_async(swp_entry_t entry, gfp_t gfp_mask,
			struct vm_area_struct *vma, unsigned long addr)
{
	struct page *found_page, *new_page = NULL;
	int err;

	do {
		/*
		 * First check the swap cache.  Since this is normally
		 * called after lookup_swap_cache() failed, re-calling
		 * that would confuse statistics.
		 */
		found_page = find_get_page(swap_address_space(entry),
					entry.val);
		if (found_page)
			break;

		/*
		 * Get a new page to read into from swap.
		 */
		if (!new_page) {
			new_page = alloc_page_vma(gfp_mask, vma, addr);
			if (!new_page)
				break;		/* Out of memory */
		}

		/*
		 * call radix_tree_preload() while we can wait.
		 */
		err = radix_tree_maybe_preload(gfp_mask & GFP_KERNEL);
		if (err)
			break;

		/*
		 * Swap entry may have been freed since our caller observed it.
		 */
		err = swapcache_prepare(entry);
		if (err == -EEXIST) {
			radix_tree_preload_end();
			/*
			 * We might race against get_swap_page() and stumble
			 * across a SWAP_HAS_CACHE swap_map entry whose page
			 * has not been brought into the swapcache yet, while
			 * the other end is scheduled away waiting on discard
			 * I/O completion at scan_swap_map().
			 *
			 * In order to avoid turning this transitory state
			 * into a permanent loop around this -EEXIST case
			 * if !CONFIG_PREEMPT and the I/O completion happens
			 * to be waiting on the CPU waitqueue where we are now
			 * busy looping, we just conditionally invoke the
			 * scheduler here, if there are some more important
			 * tasks to run.
			 */
			cond_resched();
			continue;
		}
		if (err) {		/* swp entry is obsolete ? */
			radix_tree_preload_end();
			break;
		}

		/* May fail (-ENOMEM) if radix-tree node allocation failed. */
		__set_page_locked(new_page);
		SetPageSwapBacked(new_page);
		err = __add_to_swap_cache(new_page, entry);
		if (likely(!err)) {
			radix_tree_preload_end();
			/*
			 * Initiate read into locked page and return.
			 */
			lru_cache_add_anon(new_page);
			swap_readpage(new_page);
			return new_page;
		}
		radix_tree_preload_end();
		ClearPageSwapBacked(new_page);
		__clear_page_locked(new_page);
		/*
		 * add_to_swap_cache() doesn't return -EEXIST, so we can safely
		 * clear SWAP_HAS_CACHE flag.
		 */
		swapcache_free(entry, NULL);
	} while (err != -ENOMEM);

	if (new_page)
		page_cache_release(new_page);
	return found_page;
}

/**
 * swapin_readahead - swap in pages in hope we need them soon
 * @entry: swap entry of this memory
 * @gfp_mask: memory allocation flags
 * @vma: user vma this address belongs to
 * @addr: target address for mempolicy
 *
 * Returns the struct page for entry and addr, after queueing swapin.
 *
 * Primitive swap readahead code. We simply read an aligned block of
 * (1 << page_cluster) entries in the swap area. This method is chosen
 * because it doesn't cost us any seek time.  We also make sure to queue
 * the 'original' request together with the readahead ones...
 *
 * This has been extended to use the NUMA policies from the mm triggering
 * the readahead.
 *
 * Caller must hold down_read on the vma->vm_mm if vma is not NULL.
 */
struct page *swapin_readahead(swp_entry_t entry, gfp_t gfp_mask,
			struct vm_area_struct *vma, unsigned long addr)
{
	struct page *page;
	unsigned long offset = swp_offset(entry);
	unsigned long start_offset, end_offset;
	unsigned long mask = (1UL << page_cluster) - 1;
	struct blk_plug plug;

	/* Read a page_cluster sized and aligned cluster around offset. */
	start_offset = offset & ~mask;
	end_offset = offset | mask;
	if (!start_offset)	/* First page is swap header. */
		start_offset++;

	blk_start_plug(&plug);
	for (offset = start_offset; offset <= end_offset ; offset++) {
		/* Ok, do the async read-ahead now */
		page = read_swap_cache_async(swp_entry(swp_type(entry), offset),
						gfp_mask, vma, addr);
		if (!page)
			continue;
		page_cache_release(page);
	}
	blk_finish_plug(&plug);

	lru_add_drain();	/* Push any new pages onto the LRU now */
	return read_swap_cache_async(entry, gfp_mask, vma, addr);
}
Commit	Line	Data
1da177e4 LT	1	/*
	2	* linux/mm/swap_state.c
	3	*
	4	* Copyright (C) 1991, 1992, 1993, 1994 Linus Torvalds
	5	* Swap reorganised 29.12.95, Stephen Tweedie
	6	*
	7	* Rewritten to use page cache, (C) 1998 Stephen Tweedie
	8	*/
1da177e4	9	#include <linux/mm.h>
5a0e3ad6	10	#include <linux/gfp.h>
1da177e4 LT	11	#include <linux/kernel_stat.h>
1da177e4 LT	12	#include <linux/swap.h>
46017e95	13	#include <linux/swapops.h>
1da177e4 LT	14	#include <linux/init.h>
1da177e4 LT	15	#include <linux/pagemap.h>
1da177e4	16	#include <linux/backing-dev.h>
3fb5c298	17	#include <linux/blkdev.h>
c484d410	18	#include <linux/pagevec.h>
b20a3503	19	#include <linux/migrate.h>
8c7c6e34	20	#include <linux/page_cgroup.h>
1da177e4 LT	21
	22	#include <asm/pgtable.h>
	23
	24	/*
	25	* swapper_space is a fiction, retained to simplify the path through
7eaceacc	26	* vmscan's shrink_page_list.
1da177e4	27	*/
f5e54d6e	28	static const struct address_space_operations swap_aops = {
1da177e4	29	.writepage = swap_writepage,
62c230bc	30	.set_page_dirty = swap_set_page_dirty,
e965f963	31	.migratepage = migrate_page,
1da177e4 LT	32	};
	33
	34	static struct backing_dev_info swap_backing_dev_info = {
d993831f	35	.name = "swap",
4f98a2fe	36	.capabilities = BDI_CAP_NO_ACCT_AND_WRITEBACK \| BDI_CAP_SWAP_BACKED,
1da177e4 LT	37	};
1da177e4 LT	38
33806f06 SL	39	struct address_space swapper_spaces[MAX_SWAPFILES] = {
	40	[0 ... MAX_SWAPFILES - 1] = {
	41	.page_tree = RADIX_TREE_INIT(GFP_ATOMIC\|__GFP_NOWARN),
	42	.a_ops = &swap_aops,
	43	.backing_dev_info = &swap_backing_dev_info,
	44	}
1da177e4	45	};
1da177e4 LT	46
	47	#define INC_CACHE_INFO(x) do { swap_cache_info.x++; } while (0)
	48
	49	static struct {
	50	unsigned long add_total;
	51	unsigned long del_total;
	52	unsigned long find_success;
	53	unsigned long find_total;
1da177e4 LT	54	} swap_cache_info;
1da177e4 LT	55
33806f06 SL	56	unsigned long total_swapcache_pages(void)
	57	{
	58	int i;
	59	unsigned long ret = 0;
	60
	61	for (i = 0; i < MAX_SWAPFILES; i++)
	62	ret += swapper_spaces[i].nrpages;
	63	return ret;
	64	}
	65
1da177e4 LT	66	void show_swap_cache_info(void)
1da177e4 LT	67	{
33806f06	68	printk("%lu pages in swap cache\n", total_swapcache_pages());
2c97b7fc	69	printk("Swap cache stats: add %lu, delete %lu, find %lu/%lu\n",
1da177e4	70	swap_cache_info.add_total, swap_cache_info.del_total,
bb63be0a	71	swap_cache_info.find_success, swap_cache_info.find_total);
ec8acf20 SL	72	printk("Free swap = %ldkB\n",
ec8acf20 SL	73	get_nr_swap_pages() << (PAGE_SHIFT - 10));
1da177e4 LT	74	printk("Total swap = %lukB\n", total_swap_pages << (PAGE_SHIFT - 10));
	75	}
	76
	77	/*
31a56396	78	* __add_to_swap_cache resembles add_to_page_cache_locked on swapper_space,
1da177e4 LT	79	* but sets SwapCache flag and private instead of mapping and index.
1da177e4 LT	80	*/
2f772e6c	81	int __add_to_swap_cache(struct page *page, swp_entry_t entry)
1da177e4 LT	82	{
1da177e4 LT	83	int error;
33806f06	84	struct address_space *address_space;
1da177e4	85
51726b12 HD	86	VM_BUG_ON(!PageLocked(page));
	87	VM_BUG_ON(PageSwapCache(page));
	88	VM_BUG_ON(!PageSwapBacked(page));
	89
31a56396 DN	90	page_cache_get(page);
	91	SetPageSwapCache(page);
	92	set_page_private(page, entry.val);
	93
33806f06 SL	94	address_space = swap_address_space(entry);
	95	spin_lock_irq(&address_space->tree_lock);
	96	error = radix_tree_insert(&address_space->page_tree,
	97	entry.val, page);
31a56396	98	if (likely(!error)) {
33806f06	99	address_space->nrpages++;
31a56396 DN	100	__inc_zone_page_state(page, NR_FILE_PAGES);
	101	INC_CACHE_INFO(add_total);
	102	}
33806f06	103	spin_unlock_irq(&address_space->tree_lock);
31a56396 DN	104
31a56396 DN	105	if (unlikely(error)) {
2ca4532a DN	106	/*
	107	* Only the context which have set SWAP_HAS_CACHE flag
	108	* would call add_to_swap_cache().
	109	* So add_to_swap_cache() doesn't returns -EEXIST.
	110	*/
	111	VM_BUG_ON(error == -EEXIST);
31a56396 DN	112	set_page_private(page, 0UL);
	113	ClearPageSwapCache(page);
	114	page_cache_release(page);
	115	}
	116
	117	return error;
	118	}
	119
	120
	121	int add_to_swap_cache(struct page *page, swp_entry_t entry, gfp_t gfp_mask)
	122	{
	123	int error;
	124
5e4c0d97	125	error = radix_tree_maybe_preload(gfp_mask);
35c754d7	126	if (!error) {
31a56396	127	error = __add_to_swap_cache(page, entry);
1da177e4	128	radix_tree_preload_end();
fa1de900	129	}
1da177e4 LT	130	return error;
	131	}
	132
1da177e4 LT	133	/*
	134	* This must be called only on pages that have
	135	* been verified to be in the swap cache.
	136	*/
	137	void __delete_from_swap_cache(struct page *page)
	138	{
33806f06 SL	139	swp_entry_t entry;
	140	struct address_space *address_space;
	141
51726b12 HD	142	VM_BUG_ON(!PageLocked(page));
	143	VM_BUG_ON(!PageSwapCache(page));
	144	VM_BUG_ON(PageWriteback(page));
1da177e4	145
33806f06 SL	146	entry.val = page_private(page);
	147	address_space = swap_address_space(entry);
	148	radix_tree_delete(&address_space->page_tree, page_private(page));
4c21e2f2	149	set_page_private(page, 0);
1da177e4	150	ClearPageSwapCache(page);
33806f06	151	address_space->nrpages--;
347ce434	152	__dec_zone_page_state(page, NR_FILE_PAGES);
1da177e4 LT	153	INC_CACHE_INFO(del_total);
	154	}
	155
	156	/**
	157	* add_to_swap - allocate swap space for a page
	158	* @page: page we want to move to swap
	159	*
	160	* Allocate swap space for the page and add the page to the
	161	* swap cache. Caller needs to hold the page lock.
	162	*/
5bc7b8ac	163	int add_to_swap(struct page page, struct list_head list)
1da177e4 LT	164	{
1da177e4 LT	165	swp_entry_t entry;
1da177e4 LT	166	int err;
1da177e4 LT	167
51726b12 HD	168	VM_BUG_ON(!PageLocked(page));
51726b12 HD	169	VM_BUG_ON(!PageUptodate(page));
1da177e4	170
2ca4532a DN	171	entry = get_swap_page();
	172	if (!entry.val)
	173	return 0;
	174
3f04f62f	175	if (unlikely(PageTransHuge(page)))
5bc7b8ac	176	if (unlikely(split_huge_page_to_list(page, list))) {
3f04f62f AA	177	swapcache_free(entry, NULL);
	178	return 0;
	179	}
	180
2ca4532a DN	181	/*
	182	* Radix-tree node allocations from PF_MEMALLOC contexts could
	183	* completely exhaust the page allocator. __GFP_NOMEMALLOC
	184	* stops emergency reserves from being allocated.
	185	*
	186	* TODO: this could cause a theoretical memory reclaim
	187	* deadlock in the swap out path.
	188	*/
	189	/*
	190	* Add it to the swap cache and mark it dirty
	191	*/
	192	err = add_to_swap_cache(page, entry,
	193	__GFP_HIGH\|__GFP_NOMEMALLOC\|__GFP_NOWARN);
	194
	195	if (!err) { /* Success */
	196	SetPageDirty(page);
	197	return 1;
	198	} else { /* -ENOMEM radix-tree allocation failure */
bd53b714	199	/*
2ca4532a DN	200	* add_to_swap_cache() doesn't return -EEXIST, so we can safely
2ca4532a DN	201	* clear SWAP_HAS_CACHE flag.
1da177e4	202	*/
2ca4532a DN	203	swapcache_free(entry, NULL);
2ca4532a DN	204	return 0;
1da177e4 LT	205	}
	206	}
	207
	208	/*
	209	* This must be called only on pages that have
	210	* been verified to be in the swap cache and locked.
	211	* It will never put the page into the free list,
	212	* the caller has a reference on the page.
	213	*/
	214	void delete_from_swap_cache(struct page *page)
	215	{
	216	swp_entry_t entry;
33806f06	217	struct address_space *address_space;
1da177e4	218
4c21e2f2	219	entry.val = page_private(page);
1da177e4	220
33806f06 SL	221	address_space = swap_address_space(entry);
33806f06 SL	222	spin_lock_irq(&address_space->tree_lock);
1da177e4	223	__delete_from_swap_cache(page);
33806f06	224	spin_unlock_irq(&address_space->tree_lock);
1da177e4	225
cb4b86ba	226	swapcache_free(entry, page);
1da177e4 LT	227	page_cache_release(page);
	228	}
	229
1da177e4 LT	230	/*
	231	* If we are the only user, then try to free up the swap cache.
	232	*
	233	* Its ok to check for PageSwapCache without the page lock
a2c43eed HD	234	* here because we are going to recheck again inside
a2c43eed HD	235	* try_to_free_swap() _with_ the lock.
1da177e4 LT	236	* - Marcelo
	237	*/
	238	static inline void free_swap_cache(struct page *page)
	239	{
a2c43eed HD	240	if (PageSwapCache(page) && !page_mapped(page) && trylock_page(page)) {
a2c43eed HD	241	try_to_free_swap(page);
1da177e4 LT	242	unlock_page(page);
	243	}
	244	}
	245
	246	/*
	247	* Perform a free_page(), also freeing any swap cache associated with
b8072f09	248	* this page if it is the last user of the page.
1da177e4 LT	249	*/
	250	void free_page_and_swap_cache(struct page *page)
	251	{
	252	free_swap_cache(page);
	253	page_cache_release(page);
	254	}
	255
	256	/*
	257	* Passed an array of pages, drop them all from swapcache and then release
	258	* them. They are removed from the LRU and freed if this is their last use.
	259	*/
	260	void free_pages_and_swap_cache(struct page **pages, int nr)
	261	{
1da177e4 LT	262	struct page **pagep = pages;
	263
	264	lru_add_drain();
	265	while (nr) {
c484d410	266	int todo = min(nr, PAGEVEC_SIZE);
1da177e4 LT	267	int i;
	268
	269	for (i = 0; i < todo; i++)
	270	free_swap_cache(pagep[i]);
	271	release_pages(pagep, todo, 0);
	272	pagep += todo;
	273	nr -= todo;
	274	}
	275	}
	276
	277	/*
	278	* Lookup a swap entry in the swap cache. A found page will be returned
	279	* unlocked and with its refcount incremented - we rely on the kernel
	280	* lock getting page table operations atomic even if we drop the page
	281	* lock before returning.
	282	*/
	283	struct page * lookup_swap_cache(swp_entry_t entry)
	284	{
	285	struct page *page;
	286
33806f06	287	page = find_get_page(swap_address_space(entry), entry.val);
1da177e4 LT	288
	289	if (page)
	290	INC_CACHE_INFO(find_success);
	291
	292	INC_CACHE_INFO(find_total);
	293	return page;
	294	}
	295
	296	/*
	297	* Locate a page of swap in physical memory, reserving swap cache space
	298	* and reading the disk if it is not already cached.
	299	* A failure return means that either the page allocation failed or that
	300	* the swap entry is no longer in use.
	301	*/
02098fea	302	struct page *read_swap_cache_async(swp_entry_t entry, gfp_t gfp_mask,
1da177e4 LT	303	struct vm_area_struct *vma, unsigned long addr)
	304	{
	305	struct page found_page, new_page = NULL;
	306	int err;
	307
	308	do {
	309	/*
	310	* First check the swap cache. Since this is normally
	311	* called after lookup_swap_cache() failed, re-calling
	312	* that would confuse statistics.
	313	*/
33806f06 SL	314	found_page = find_get_page(swap_address_space(entry),
33806f06 SL	315	entry.val);
1da177e4 LT	316	if (found_page)
	317	break;
	318
	319	/*
	320	* Get a new page to read into from swap.
	321	*/
	322	if (!new_page) {
02098fea	323	new_page = alloc_page_vma(gfp_mask, vma, addr);
1da177e4 LT	324	if (!new_page)
	325	break; /* Out of memory */
	326	}
	327
31a56396 DN	328	/*
	329	* call radix_tree_preload() while we can wait.
	330	*/
5e4c0d97	331	err = radix_tree_maybe_preload(gfp_mask & GFP_KERNEL);
31a56396 DN	332	if (err)
	333	break;
	334
f000944d HD	335	/*
	336	* Swap entry may have been freed since our caller observed it.
	337	*/
355cfa73	338	err = swapcache_prepare(entry);
cbab0e4e	339	if (err == -EEXIST) {
31a56396	340	radix_tree_preload_end();
cbab0e4e RA	341	/*
	342	* We might race against get_swap_page() and stumble
	343	* across a SWAP_HAS_CACHE swap_map entry whose page
	344	* has not been brought into the swapcache yet, while
	345	* the other end is scheduled away waiting on discard
	346	* I/O completion at scan_swap_map().
	347	*
	348	* In order to avoid turning this transitory state
	349	* into a permanent loop around this -EEXIST case
	350	* if !CONFIG_PREEMPT and the I/O completion happens
	351	* to be waiting on the CPU waitqueue where we are now
	352	* busy looping, we just conditionally invoke the
	353	* scheduler here, if there are some more important
	354	* tasks to run.
	355	*/
	356	cond_resched();
355cfa73	357	continue;
31a56396 DN	358	}
	359	if (err) { /* swp entry is obsolete ? */
	360	radix_tree_preload_end();
f000944d	361	break;
31a56396	362	}
f000944d	363
2ca4532a	364	/* May fail (-ENOMEM) if radix-tree node allocation failed. */
f45840b5	365	__set_page_locked(new_page);
b2e18538	366	SetPageSwapBacked(new_page);
31a56396	367	err = __add_to_swap_cache(new_page, entry);
529ae9aa	368	if (likely(!err)) {
31a56396	369	radix_tree_preload_end();
1da177e4 LT	370	/*
	371	* Initiate read into locked page and return.
	372	*/
c5fdae46	373	lru_cache_add_anon(new_page);
aca8bf32	374	swap_readpage(new_page);
1da177e4 LT	375	return new_page;
1da177e4 LT	376	}
31a56396	377	radix_tree_preload_end();
b2e18538	378	ClearPageSwapBacked(new_page);
f45840b5	379	__clear_page_locked(new_page);
2ca4532a DN	380	/*
	381	* add_to_swap_cache() doesn't return -EEXIST, so we can safely
	382	* clear SWAP_HAS_CACHE flag.
	383	*/
cb4b86ba	384	swapcache_free(entry, NULL);
f000944d	385	} while (err != -ENOMEM);
1da177e4 LT	386
	387	if (new_page)
	388	page_cache_release(new_page);
	389	return found_page;
	390	}
46017e95 HD	391
	392	/**
	393	* swapin_readahead - swap in pages in hope we need them soon
	394	* @entry: swap entry of this memory
7682486b	395	* @gfp_mask: memory allocation flags
46017e95 HD	396	* @vma: user vma this address belongs to
	397	* @addr: target address for mempolicy
	398	*
	399	* Returns the struct page for entry and addr, after queueing swapin.
	400	*
	401	* Primitive swap readahead code. We simply read an aligned block of
	402	* (1 << page_cluster) entries in the swap area. This method is chosen
	403	* because it doesn't cost us any seek time. We also make sure to queue
	404	* the 'original' request together with the readahead ones...
	405	*
	406	* This has been extended to use the NUMA policies from the mm triggering
	407	* the readahead.
	408	*
	409	* Caller must hold down_read on the vma->vm_mm if vma is not NULL.
	410	*/
02098fea	411	struct page *swapin_readahead(swp_entry_t entry, gfp_t gfp_mask,
46017e95 HD	412	struct vm_area_struct *vma, unsigned long addr)
46017e95 HD	413	{
46017e95	414	struct page *page;
67f96aa2 RR	415	unsigned long offset = swp_offset(entry);
	416	unsigned long start_offset, end_offset;
	417	unsigned long mask = (1UL << page_cluster) - 1;
3fb5c298	418	struct blk_plug plug;
46017e95	419
67f96aa2 RR	420	/* Read a page_cluster sized and aligned cluster around offset. */
	421	start_offset = offset & ~mask;
	422	end_offset = offset \| mask;
	423	if (!start_offset) /* First page is swap header. */
	424	start_offset++;
	425
3fb5c298	426	blk_start_plug(&plug);
67f96aa2	427	for (offset = start_offset; offset <= end_offset ; offset++) {
46017e95 HD	428	/* Ok, do the async read-ahead now */
46017e95 HD	429	page = read_swap_cache_async(swp_entry(swp_type(entry), offset),
02098fea	430	gfp_mask, vma, addr);
46017e95	431	if (!page)
67f96aa2	432	continue;
46017e95 HD	433	page_cache_release(page);
46017e95 HD	434	}
3fb5c298 CE	435	blk_finish_plug(&plug);
3fb5c298 CE	436
46017e95	437	lru_add_drain(); /* Push any new pages onto the LRU now */
02098fea	438	return read_swap_cache_async(entry, gfp_mask, vma, addr);
46017e95	439	}