[linux.git] / mm / mempool.c

// SPDX-License-Identifier: GPL-2.0
/*
 *  linux/mm/mempool.c
 *
 *  memory buffer pool support. Such pools are mostly used
 *  for guaranteed, deadlock-free memory allocations during
 *  extreme VM load.
 *
 *  started by Ingo Molnar, Copyright (C) 2001
 *  debugging by David Rientjes, Copyright (C) 2015
 */

#include <linux/mm.h>
#include <linux/slab.h>
#include <linux/highmem.h>
#include <linux/kasan.h>
#include <linux/kmemleak.h>
#include <linux/export.h>
#include <linux/mempool.h>
#include <linux/writeback.h>
#include "slab.h"

#if defined(CONFIG_DEBUG_SLAB) || defined(CONFIG_SLUB_DEBUG_ON)
static void poison_error(mempool_t *pool, void *element, size_t size,
			 size_t byte)
{
	const int nr = pool->curr_nr;
	const int start = max_t(int, byte - (BITS_PER_LONG / 8), 0);
	const int end = min_t(int, byte + (BITS_PER_LONG / 8), size);
	int i;

	pr_err("BUG: mempool element poison mismatch\n");
	pr_err("Mempool %p size %zu\n", pool, size);
	pr_err(" nr=%d @ %p: %s0x", nr, element, start > 0 ? "... " : "");
	for (i = start; i < end; i++)
		pr_cont("%x ", *(u8 *)(element + i));
	pr_cont("%s\n", end < size ? "..." : "");
	dump_stack();
}

static void __check_element(mempool_t *pool, void *element, size_t size)
{
	u8 *obj = element;
	size_t i;

	for (i = 0; i < size; i++) {
		u8 exp = (i < size - 1) ? POISON_FREE : POISON_END;

		if (obj[i] != exp) {
			poison_error(pool, element, size, i);
			return;
		}
	}
	memset(obj, POISON_INUSE, size);
}

static void check_element(mempool_t *pool, void *element)
{
	/* Mempools backed by slab allocator */
	if (pool->free == mempool_free_slab || pool->free == mempool_kfree) {
		__check_element(pool, element, ksize(element));
	} else if (pool->free == mempool_free_pages) {
		/* Mempools backed by page allocator */
		int order = (int)(long)pool->pool_data;
		void *addr = kmap_atomic((struct page *)element);

		__check_element(pool, addr, 1UL << (PAGE_SHIFT + order));
		kunmap_atomic(addr);
	}
}

static void __poison_element(void *element, size_t size)
{
	u8 *obj = element;

	memset(obj, POISON_FREE, size - 1);
	obj[size - 1] = POISON_END;
}

static void poison_element(mempool_t *pool, void *element)
{
	/* Mempools backed by slab allocator */
	if (pool->alloc == mempool_alloc_slab || pool->alloc == mempool_kmalloc) {
		__poison_element(element, ksize(element));
	} else if (pool->alloc == mempool_alloc_pages) {
		/* Mempools backed by page allocator */
		int order = (int)(long)pool->pool_data;
		void *addr = kmap_atomic((struct page *)element);

		__poison_element(addr, 1UL << (PAGE_SHIFT + order));
		kunmap_atomic(addr);
	}
}
#else /* CONFIG_DEBUG_SLAB || CONFIG_SLUB_DEBUG_ON */
static inline void check_element(mempool_t *pool, void *element)
{
}
static inline void poison_element(mempool_t *pool, void *element)
{
}
#endif /* CONFIG_DEBUG_SLAB || CONFIG_SLUB_DEBUG_ON */

static __always_inline void kasan_poison_element(mempool_t *pool, void *element)
{
	if (pool->alloc == mempool_alloc_slab || pool->alloc == mempool_kmalloc)
		kasan_slab_free_mempool(element);
	else if (pool->alloc == mempool_alloc_pages)
		kasan_poison_pages(element, (unsigned long)pool->pool_data,
				   false);
}

static void kasan_unpoison_element(mempool_t *pool, void *element)
{
	if (pool->alloc == mempool_alloc_slab || pool->alloc == mempool_kmalloc)
		kasan_unpoison_range(element, __ksize(element));
	else if (pool->alloc == mempool_alloc_pages)
		kasan_unpoison_pages(element, (unsigned long)pool->pool_data,
				     false);
}

static __always_inline void add_element(mempool_t *pool, void *element)
{
	BUG_ON(pool->curr_nr >= pool->min_nr);
	poison_element(pool, element);
	kasan_poison_element(pool, element);
	pool->elements[pool->curr_nr++] = element;
}

static void *remove_element(mempool_t *pool)
{
	void *element = pool->elements[--pool->curr_nr];

	BUG_ON(pool->curr_nr < 0);
	kasan_unpoison_element(pool, element);
	check_element(pool, element);
	return element;
}

/**
 * mempool_exit - exit a mempool initialized with mempool_init()
 * @pool:      pointer to the memory pool which was initialized with
 *             mempool_init().
 *
 * Free all reserved elements in @pool and @pool itself.  This function
 * only sleeps if the free_fn() function sleeps.
 *
 * May be called on a zeroed but uninitialized mempool (i.e. allocated with
 * kzalloc()).
 */
void mempool_exit(mempool_t *pool)
{
	while (pool->curr_nr) {
		void *element = remove_element(pool);
		pool->free(element, pool->pool_data);
	}
	kfree(pool->elements);
	pool->elements = NULL;
}
EXPORT_SYMBOL(mempool_exit);

/**
 * mempool_destroy - deallocate a memory pool
 * @pool:      pointer to the memory pool which was allocated via
 *             mempool_create().
 *
 * Free all reserved elements in @pool and @pool itself.  This function
 * only sleeps if the free_fn() function sleeps.
 */
void mempool_destroy(mempool_t *pool)
{
	if (unlikely(!pool))
		return;

	mempool_exit(pool);
	kfree(pool);
}
EXPORT_SYMBOL(mempool_destroy);

int mempool_init_node(mempool_t *pool, int min_nr, mempool_alloc_t *alloc_fn,
		      mempool_free_t *free_fn, void *pool_data,
		      gfp_t gfp_mask, int node_id)
{
	spin_lock_init(&pool->lock);
	pool->min_nr	= min_nr;
	pool->pool_data = pool_data;
	pool->alloc	= alloc_fn;
	pool->free	= free_fn;
	init_waitqueue_head(&pool->wait);

	pool->elements = kmalloc_array_node(min_nr, sizeof(void *),
					    gfp_mask, node_id);
	if (!pool->elements)
		return -ENOMEM;

	/*
	 * First pre-allocate the guaranteed number of buffers.
	 */
	while (pool->curr_nr < pool->min_nr) {
		void *element;

		element = pool->alloc(gfp_mask, pool->pool_data);
		if (unlikely(!element)) {
			mempool_exit(pool);
			return -ENOMEM;
		}
		add_element(pool, element);
	}

	return 0;
}
EXPORT_SYMBOL(mempool_init_node);

/**
 * mempool_init - initialize a memory pool
 * @pool:      pointer to the memory pool that should be initialized
 * @min_nr:    the minimum number of elements guaranteed to be
 *             allocated for this pool.
 * @alloc_fn:  user-defined element-allocation function.
 * @free_fn:   user-defined element-freeing function.
 * @pool_data: optional private data available to the user-defined functions.
 *
 * Like mempool_create(), but initializes the pool in (i.e. embedded in another
 * structure).
 *
 * Return: %0 on success, negative error code otherwise.
 */
int mempool_init(mempool_t *pool, int min_nr, mempool_alloc_t *alloc_fn,
		 mempool_free_t *free_fn, void *pool_data)
{
	return mempool_init_node(pool, min_nr, alloc_fn, free_fn,
				 pool_data, GFP_KERNEL, NUMA_NO_NODE);

}
EXPORT_SYMBOL(mempool_init);

/**
 * mempool_create - create a memory pool
 * @min_nr:    the minimum number of elements guaranteed to be
 *             allocated for this pool.
 * @alloc_fn:  user-defined element-allocation function.
 * @free_fn:   user-defined element-freeing function.
 * @pool_data: optional private data available to the user-defined functions.
 *
 * this function creates and allocates a guaranteed size, preallocated
 * memory pool. The pool can be used from the mempool_alloc() and mempool_free()
 * functions. This function might sleep. Both the alloc_fn() and the free_fn()
 * functions might sleep - as long as the mempool_alloc() function is not called
 * from IRQ contexts.
 *
 * Return: pointer to the created memory pool object or %NULL on error.
 */
mempool_t *mempool_create(int min_nr, mempool_alloc_t *alloc_fn,
				mempool_free_t *free_fn, void *pool_data)
{
	return mempool_create_node(min_nr, alloc_fn, free_fn, pool_data,
				   GFP_KERNEL, NUMA_NO_NODE);
}
EXPORT_SYMBOL(mempool_create);

mempool_t *mempool_create_node(int min_nr, mempool_alloc_t *alloc_fn,
			       mempool_free_t *free_fn, void *pool_data,
			       gfp_t gfp_mask, int node_id)
{
	mempool_t *pool;

	pool = kzalloc_node(sizeof(*pool), gfp_mask, node_id);
	if (!pool)
		return NULL;

	if (mempool_init_node(pool, min_nr, alloc_fn, free_fn, pool_data,
			      gfp_mask, node_id)) {
		kfree(pool);
		return NULL;
	}

	return pool;
}
EXPORT_SYMBOL(mempool_create_node);

/**
 * mempool_resize - resize an existing memory pool
 * @pool:       pointer to the memory pool which was allocated via
 *              mempool_create().
 * @new_min_nr: the new minimum number of elements guaranteed to be
 *              allocated for this pool.
 *
 * This function shrinks/grows the pool. In the case of growing,
 * it cannot be guaranteed that the pool will be grown to the new
 * size immediately, but new mempool_free() calls will refill it.
 * This function may sleep.
 *
 * Note, the caller must guarantee that no mempool_destroy is called
 * while this function is running. mempool_alloc() & mempool_free()
 * might be called (eg. from IRQ contexts) while this function executes.
 *
 * Return: %0 on success, negative error code otherwise.
 */
int mempool_resize(mempool_t *pool, int new_min_nr)
{
	void *element;
	void **new_elements;
	unsigned long flags;

	BUG_ON(new_min_nr <= 0);
	might_sleep();

	spin_lock_irqsave(&pool->lock, flags);
	if (new_min_nr <= pool->min_nr) {
		while (new_min_nr < pool->curr_nr) {
			element = remove_element(pool);
			spin_unlock_irqrestore(&pool->lock, flags);
			pool->free(element, pool->pool_data);
			spin_lock_irqsave(&pool->lock, flags);
		}
		pool->min_nr = new_min_nr;
		goto out_unlock;
	}
	spin_unlock_irqrestore(&pool->lock, flags);

	/* Grow the pool */
	new_elements = kmalloc_array(new_min_nr, sizeof(*new_elements),
				     GFP_KERNEL);
	if (!new_elements)
		return -ENOMEM;

	spin_lock_irqsave(&pool->lock, flags);
	if (unlikely(new_min_nr <= pool->min_nr)) {
		/* Raced, other resize will do our work */
		spin_unlock_irqrestore(&pool->lock, flags);
		kfree(new_elements);
		goto out;
	}
	memcpy(new_elements, pool->elements,
			pool->curr_nr * sizeof(*new_elements));
	kfree(pool->elements);
	pool->elements = new_elements;
	pool->min_nr = new_min_nr;

	while (pool->curr_nr < pool->min_nr) {
		spin_unlock_irqrestore(&pool->lock, flags);
		element = pool->alloc(GFP_KERNEL, pool->pool_data);
		if (!element)
			goto out;
		spin_lock_irqsave(&pool->lock, flags);
		if (pool->curr_nr < pool->min_nr) {
			add_element(pool, element);
		} else {
			spin_unlock_irqrestore(&pool->lock, flags);
			pool->free(element, pool->pool_data);	/* Raced */
			goto out;
		}
	}
out_unlock:
	spin_unlock_irqrestore(&pool->lock, flags);
out:
	return 0;
}
EXPORT_SYMBOL(mempool_resize);

/**
 * mempool_alloc - allocate an element from a specific memory pool
 * @pool:      pointer to the memory pool which was allocated via
 *             mempool_create().
 * @gfp_mask:  the usual allocation bitmask.
 *
 * this function only sleeps if the alloc_fn() function sleeps or
 * returns NULL. Note that due to preallocation, this function
 * *never* fails when called from process contexts. (it might
 * fail if called from an IRQ context.)
 * Note: using __GFP_ZERO is not supported.
 *
 * Return: pointer to the allocated element or %NULL on error.
 */
void *mempool_alloc(mempool_t *pool, gfp_t gfp_mask)
{
	void *element;
	unsigned long flags;
	wait_queue_entry_t wait;
	gfp_t gfp_temp;

	VM_WARN_ON_ONCE(gfp_mask & __GFP_ZERO);
	might_alloc(gfp_mask);

	gfp_mask |= __GFP_NOMEMALLOC;	/* don't allocate emergency reserves */
	gfp_mask |= __GFP_NORETRY;	/* don't loop in __alloc_pages */
	gfp_mask |= __GFP_NOWARN;	/* failures are OK */

	gfp_temp = gfp_mask & ~(__GFP_DIRECT_RECLAIM|__GFP_IO);

repeat_alloc:

	element = pool->alloc(gfp_temp, pool->pool_data);
	if (likely(element != NULL))
		return element;

	spin_lock_irqsave(&pool->lock, flags);
	if (likely(pool->curr_nr)) {
		element = remove_element(pool);
		spin_unlock_irqrestore(&pool->lock, flags);
		/* paired with rmb in mempool_free(), read comment there */
		smp_wmb();
		/*
		 * Update the allocation stack trace as this is more useful
		 * for debugging.
		 */
		kmemleak_update_trace(element);
		return element;
	}

	/*
	 * We use gfp mask w/o direct reclaim or IO for the first round.  If
	 * alloc failed with that and @pool was empty, retry immediately.
	 */
	if (gfp_temp != gfp_mask) {
		spin_unlock_irqrestore(&pool->lock, flags);
		gfp_temp = gfp_mask;
		goto repeat_alloc;
	}

	/* We must not sleep if !__GFP_DIRECT_RECLAIM */
	if (!(gfp_mask & __GFP_DIRECT_RECLAIM)) {
		spin_unlock_irqrestore(&pool->lock, flags);
		return NULL;
	}

	/* Let's wait for someone else to return an element to @pool */
	init_wait(&wait);
	prepare_to_wait(&pool->wait, &wait, TASK_UNINTERRUPTIBLE);

	spin_unlock_irqrestore(&pool->lock, flags);

	/*
	 * FIXME: this should be io_schedule().  The timeout is there as a
	 * workaround for some DM problems in 2.6.18.
	 */
	io_schedule_timeout(5*HZ);

	finish_wait(&pool->wait, &wait);
	goto repeat_alloc;
}
EXPORT_SYMBOL(mempool_alloc);

/**
 * mempool_free - return an element to the pool.
 * @element:   pool element pointer.
 * @pool:      pointer to the memory pool which was allocated via
 *             mempool_create().
 *
 * this function only sleeps if the free_fn() function sleeps.
 */
void mempool_free(void *element, mempool_t *pool)
{
	unsigned long flags;

	if (unlikely(element == NULL))
		return;

	/*
	 * Paired with the wmb in mempool_alloc().  The preceding read is
	 * for @element and the following @pool->curr_nr.  This ensures
	 * that the visible value of @pool->curr_nr is from after the
	 * allocation of @element.  This is necessary for fringe cases
	 * where @element was passed to this task without going through
	 * barriers.
	 *
	 * For example, assume @p is %NULL at the beginning and one task
	 * performs "p = mempool_alloc(...);" while another task is doing
	 * "while (!p) cpu_relax(); mempool_free(p, ...);".  This function
	 * may end up using curr_nr value which is from before allocation
	 * of @p without the following rmb.
	 */
	smp_rmb();

	/*
	 * For correctness, we need a test which is guaranteed to trigger
	 * if curr_nr + #allocated == min_nr.  Testing curr_nr < min_nr
	 * without locking achieves that and refilling as soon as possible
	 * is desirable.
	 *
	 * Because curr_nr visible here is always a value after the
	 * allocation of @element, any task which decremented curr_nr below
	 * min_nr is guaranteed to see curr_nr < min_nr unless curr_nr gets
	 * incremented to min_nr afterwards.  If curr_nr gets incremented
	 * to min_nr after the allocation of @element, the elements
	 * allocated after that are subject to the same guarantee.
	 *
	 * Waiters happen iff curr_nr is 0 and the above guarantee also
	 * ensures that there will be frees which return elements to the
	 * pool waking up the waiters.
	 */
	if (unlikely(READ_ONCE(pool->curr_nr) < pool->min_nr)) {
		spin_lock_irqsave(&pool->lock, flags);
		if (likely(pool->curr_nr < pool->min_nr)) {
			add_element(pool, element);
			spin_unlock_irqrestore(&pool->lock, flags);
			wake_up(&pool->wait);
			return;
		}
		spin_unlock_irqrestore(&pool->lock, flags);
	}
	pool->free(element, pool->pool_data);
}
EXPORT_SYMBOL(mempool_free);

/*
 * A commonly used alloc and free fn.
 */
void *mempool_alloc_slab(gfp_t gfp_mask, void *pool_data)
{
	struct kmem_cache *mem = pool_data;
	VM_BUG_ON(mem->ctor);
	return kmem_cache_alloc(mem, gfp_mask);
}
EXPORT_SYMBOL(mempool_alloc_slab);

void mempool_free_slab(void *element, void *pool_data)
{
	struct kmem_cache *mem = pool_data;
	kmem_cache_free(mem, element);
}
EXPORT_SYMBOL(mempool_free_slab);

/*
 * A commonly used alloc and free fn that kmalloc/kfrees the amount of memory
 * specified by pool_data
 */
void *mempool_kmalloc(gfp_t gfp_mask, void *pool_data)
{
	size_t size = (size_t)pool_data;
	return kmalloc(size, gfp_mask);
}
EXPORT_SYMBOL(mempool_kmalloc);

void mempool_kfree(void *element, void *pool_data)
{
	kfree(element);
}
EXPORT_SYMBOL(mempool_kfree);

/*
 * A simple mempool-backed page allocator that allocates pages
 * of the order specified by pool_data.
 */
void *mempool_alloc_pages(gfp_t gfp_mask, void *pool_data)
{
	int order = (int)(long)pool_data;
	return alloc_pages(gfp_mask, order);
}
EXPORT_SYMBOL(mempool_alloc_pages);

void mempool_free_pages(void *element, void *pool_data)
{
	int order = (int)(long)pool_data;
	__free_pages(element, order);
}
EXPORT_SYMBOL(mempool_free_pages);
Commit	Line	Data
b2441318	1	// SPDX-License-Identifier: GPL-2.0
1da177e4 LT	2	/*
	3	* linux/mm/mempool.c
	4	*
	5	* memory buffer pool support. Such pools are mostly used
	6	* for guaranteed, deadlock-free memory allocations during
	7	* extreme VM load.
	8	*
	9	* started by Ingo Molnar, Copyright (C) 2001
bdfedb76	10	* debugging by David Rientjes, Copyright (C) 2015
1da177e4 LT	11	*/
	12
	13	#include <linux/mm.h>
	14	#include <linux/slab.h>
bdfedb76	15	#include <linux/highmem.h>
92393615	16	#include <linux/kasan.h>
17411962	17	#include <linux/kmemleak.h>
b95f1b31	18	#include <linux/export.h>
1da177e4	19	#include <linux/mempool.h>
1da177e4	20	#include <linux/writeback.h>
e244c9e6	21	#include "slab.h"
1da177e4	22
bdfedb76 DR	23	#if defined(CONFIG_DEBUG_SLAB) \|\| defined(CONFIG_SLUB_DEBUG_ON)
	24	static void poison_error(mempool_t pool, void element, size_t size,
	25	size_t byte)
	26	{
	27	const int nr = pool->curr_nr;
	28	const int start = max_t(int, byte - (BITS_PER_LONG / 8), 0);
	29	const int end = min_t(int, byte + (BITS_PER_LONG / 8), size);
	30	int i;
	31
	32	pr_err("BUG: mempool element poison mismatch\n");
	33	pr_err("Mempool %p size %zu\n", pool, size);
	34	pr_err(" nr=%d @ %p: %s0x", nr, element, start > 0 ? "... " : "");
	35	for (i = start; i < end; i++)
	36	pr_cont("%x ", (u8 )(element + i));
	37	pr_cont("%s\n", end < size ? "..." : "");
	38	dump_stack();
	39	}
	40
	41	static void __check_element(mempool_t pool, void element, size_t size)
	42	{
	43	u8 *obj = element;
	44	size_t i;
	45
	46	for (i = 0; i < size; i++) {
	47	u8 exp = (i < size - 1) ? POISON_FREE : POISON_END;
	48
	49	if (obj[i] != exp) {
	50	poison_error(pool, element, size, i);
	51	return;
	52	}
	53	}
	54	memset(obj, POISON_INUSE, size);
	55	}
	56
	57	static void check_element(mempool_t pool, void element)
	58	{
	59	/* Mempools backed by slab allocator */
544941d7	60	if (pool->free == mempool_free_slab \|\| pool->free == mempool_kfree) {
bdfedb76	61	__check_element(pool, element, ksize(element));
544941d7 ML	62	} else if (pool->free == mempool_free_pages) {
544941d7 ML	63	/* Mempools backed by page allocator */
bdfedb76 DR	64	int order = (int)(long)pool->pool_data;
	65	void addr = kmap_atomic((struct page )element);
	66
	67	__check_element(pool, addr, 1UL << (PAGE_SHIFT + order));
	68	kunmap_atomic(addr);
	69	}
	70	}
	71
	72	static void __poison_element(void *element, size_t size)
	73	{
	74	u8 *obj = element;
	75
	76	memset(obj, POISON_FREE, size - 1);
	77	obj[size - 1] = POISON_END;
	78	}
	79
	80	static void poison_element(mempool_t pool, void element)
	81	{
	82	/* Mempools backed by slab allocator */
544941d7	83	if (pool->alloc == mempool_alloc_slab \|\| pool->alloc == mempool_kmalloc) {
bdfedb76	84	__poison_element(element, ksize(element));
544941d7 ML	85	} else if (pool->alloc == mempool_alloc_pages) {
544941d7 ML	86	/* Mempools backed by page allocator */
bdfedb76 DR	87	int order = (int)(long)pool->pool_data;
	88	void addr = kmap_atomic((struct page )element);
	89
	90	__poison_element(addr, 1UL << (PAGE_SHIFT + order));
	91	kunmap_atomic(addr);
	92	}
	93	}
	94	#else /* CONFIG_DEBUG_SLAB \|\| CONFIG_SLUB_DEBUG_ON */
	95	static inline void check_element(mempool_t pool, void element)
	96	{
	97	}
	98	static inline void poison_element(mempool_t pool, void element)
	99	{
	100	}
	101	#endif /* CONFIG_DEBUG_SLAB \|\| CONFIG_SLUB_DEBUG_ON */
	102
6860f634	103	static __always_inline void kasan_poison_element(mempool_t pool, void element)
92393615	104	{
9b75a867	105	if (pool->alloc == mempool_alloc_slab \|\| pool->alloc == mempool_kmalloc)
027b37b5	106	kasan_slab_free_mempool(element);
544941d7	107	else if (pool->alloc == mempool_alloc_pages)
7a3b8353 PC	108	kasan_poison_pages(element, (unsigned long)pool->pool_data,
7a3b8353 PC	109	false);
92393615 AR	110	}
92393615 AR	111
8cded866	112	static void kasan_unpoison_element(mempool_t pool, void element)
92393615	113	{
9b75a867	114	if (pool->alloc == mempool_alloc_slab \|\| pool->alloc == mempool_kmalloc)
bffe6907	115	kasan_unpoison_range(element, __ksize(element));
544941d7	116	else if (pool->alloc == mempool_alloc_pages)
7a3b8353 PC	117	kasan_unpoison_pages(element, (unsigned long)pool->pool_data,
7a3b8353 PC	118	false);
92393615 AR	119	}
92393615 AR	120
6860f634	121	static __always_inline void add_element(mempool_t pool, void element)
1da177e4 LT	122	{
1da177e4 LT	123	BUG_ON(pool->curr_nr >= pool->min_nr);
bdfedb76	124	poison_element(pool, element);
92393615	125	kasan_poison_element(pool, element);
1da177e4 LT	126	pool->elements[pool->curr_nr++] = element;
	127	}
	128
8cded866	129	static void remove_element(mempool_t pool)
1da177e4	130	{
bdfedb76 DR	131	void *element = pool->elements[--pool->curr_nr];
	132
	133	BUG_ON(pool->curr_nr < 0);
8cded866	134	kasan_unpoison_element(pool, element);
76401310	135	check_element(pool, element);
bdfedb76	136	return element;
1da177e4 LT	137	}
1da177e4 LT	138
c1a67fef KO	139	/**
	140	* mempool_exit - exit a mempool initialized with mempool_init()
	141	* @pool: pointer to the memory pool which was initialized with
	142	* mempool_init().
	143	*
	144	* Free all reserved elements in @pool and @pool itself. This function
	145	* only sleeps if the free_fn() function sleeps.
	146	*
	147	* May be called on a zeroed but uninitialized mempool (i.e. allocated with
	148	* kzalloc()).
	149	*/
	150	void mempool_exit(mempool_t *pool)
	151	{
	152	while (pool->curr_nr) {
8cded866	153	void *element = remove_element(pool);
c1a67fef KO	154	pool->free(element, pool->pool_data);
	155	}
	156	kfree(pool->elements);
	157	pool->elements = NULL;
	158	}
	159	EXPORT_SYMBOL(mempool_exit);
	160
0565d317 TH	161	/**
	162	* mempool_destroy - deallocate a memory pool
	163	* @pool: pointer to the memory pool which was allocated via
	164	* mempool_create().
	165	*
	166	* Free all reserved elements in @pool and @pool itself. This function
	167	* only sleeps if the free_fn() function sleeps.
	168	*/
	169	void mempool_destroy(mempool_t *pool)
1da177e4	170	{
4e3ca3e0 SS	171	if (unlikely(!pool))
	172	return;
	173
c1a67fef	174	mempool_exit(pool);
1da177e4 LT	175	kfree(pool);
1da177e4 LT	176	}
0565d317	177	EXPORT_SYMBOL(mempool_destroy);
1da177e4	178
c1a67fef KO	179	int mempool_init_node(mempool_t pool, int min_nr, mempool_alloc_t alloc_fn,
	180	mempool_free_t free_fn, void pool_data,
	181	gfp_t gfp_mask, int node_id)
	182	{
	183	spin_lock_init(&pool->lock);
	184	pool->min_nr = min_nr;
	185	pool->pool_data = pool_data;
	186	pool->alloc = alloc_fn;
	187	pool->free = free_fn;
	188	init_waitqueue_head(&pool->wait);
	189
	190	pool->elements = kmalloc_array_node(min_nr, sizeof(void *),
	191	gfp_mask, node_id);
	192	if (!pool->elements)
	193	return -ENOMEM;
	194
	195	/*
	196	* First pre-allocate the guaranteed number of buffers.
	197	*/
	198	while (pool->curr_nr < pool->min_nr) {
	199	void *element;
	200
	201	element = pool->alloc(gfp_mask, pool->pool_data);
	202	if (unlikely(!element)) {
	203	mempool_exit(pool);
	204	return -ENOMEM;
	205	}
	206	add_element(pool, element);
	207	}
	208
	209	return 0;
	210	}
	211	EXPORT_SYMBOL(mempool_init_node);
	212
	213	/**
	214	* mempool_init - initialize a memory pool
a3bf6ce3	215	* @pool: pointer to the memory pool that should be initialized
c1a67fef KO	216	* @min_nr: the minimum number of elements guaranteed to be
	217	* allocated for this pool.
	218	* @alloc_fn: user-defined element-allocation function.
	219	* @free_fn: user-defined element-freeing function.
	220	* @pool_data: optional private data available to the user-defined functions.
	221	*
	222	* Like mempool_create(), but initializes the pool in (i.e. embedded in another
	223	* structure).
a862f68a MR	224	*
a862f68a MR	225	* Return: %0 on success, negative error code otherwise.
c1a67fef KO	226	*/
	227	int mempool_init(mempool_t pool, int min_nr, mempool_alloc_t alloc_fn,
	228	mempool_free_t free_fn, void pool_data)
	229	{
	230	return mempool_init_node(pool, min_nr, alloc_fn, free_fn,
	231	pool_data, GFP_KERNEL, NUMA_NO_NODE);
	232
	233	}
	234	EXPORT_SYMBOL(mempool_init);
	235
1da177e4 LT	236	/**
	237	* mempool_create - create a memory pool
	238	* @min_nr: the minimum number of elements guaranteed to be
	239	* allocated for this pool.
	240	* @alloc_fn: user-defined element-allocation function.
	241	* @free_fn: user-defined element-freeing function.
	242	* @pool_data: optional private data available to the user-defined functions.
	243	*
	244	* this function creates and allocates a guaranteed size, preallocated
72fd4a35	245	* memory pool. The pool can be used from the mempool_alloc() and mempool_free()
1da177e4	246	* functions. This function might sleep. Both the alloc_fn() and the free_fn()
72fd4a35	247	* functions might sleep - as long as the mempool_alloc() function is not called
1da177e4	248	* from IRQ contexts.
a862f68a MR	249	*
a862f68a MR	250	* Return: pointer to the created memory pool object or %NULL on error.
1da177e4	251	*/
1946089a	252	mempool_t mempool_create(int min_nr, mempool_alloc_t alloc_fn,
1da177e4 LT	253	mempool_free_t free_fn, void pool_data)
1da177e4 LT	254	{
68d68ff6	255	return mempool_create_node(min_nr, alloc_fn, free_fn, pool_data,
a91a5ac6	256	GFP_KERNEL, NUMA_NO_NODE);
1946089a CL	257	}
1946089a CL	258	EXPORT_SYMBOL(mempool_create);
1da177e4	259
1946089a	260	mempool_t mempool_create_node(int min_nr, mempool_alloc_t alloc_fn,
a91a5ac6 TH	261	mempool_free_t free_fn, void pool_data,
a91a5ac6 TH	262	gfp_t gfp_mask, int node_id)
1946089a CL	263	{
1946089a CL	264	mempool_t *pool;
c1a67fef	265
7b5219db	266	pool = kzalloc_node(sizeof(*pool), gfp_mask, node_id);
1da177e4 LT	267	if (!pool)
1da177e4 LT	268	return NULL;
c1a67fef KO	269
	270	if (mempool_init_node(pool, min_nr, alloc_fn, free_fn, pool_data,
	271	gfp_mask, node_id)) {
1da177e4 LT	272	kfree(pool);
	273	return NULL;
	274	}
1da177e4	275
1da177e4 LT	276	return pool;
1da177e4 LT	277	}
1946089a	278	EXPORT_SYMBOL(mempool_create_node);
1da177e4 LT	279
	280	/**
	281	* mempool_resize - resize an existing memory pool
	282	* @pool: pointer to the memory pool which was allocated via
	283	* mempool_create().
	284	* @new_min_nr: the new minimum number of elements guaranteed to be
	285	* allocated for this pool.
1da177e4 LT	286	*
	287	* This function shrinks/grows the pool. In the case of growing,
	288	* it cannot be guaranteed that the pool will be grown to the new
	289	* size immediately, but new mempool_free() calls will refill it.
11d83360	290	* This function may sleep.
1da177e4 LT	291	*
	292	* Note, the caller must guarantee that no mempool_destroy is called
	293	* while this function is running. mempool_alloc() & mempool_free()
	294	* might be called (eg. from IRQ contexts) while this function executes.
a862f68a MR	295	*
a862f68a MR	296	* Return: %0 on success, negative error code otherwise.
1da177e4	297	*/
11d83360	298	int mempool_resize(mempool_t *pool, int new_min_nr)
1da177e4 LT	299	{
	300	void *element;
	301	void **new_elements;
	302	unsigned long flags;
	303
	304	BUG_ON(new_min_nr <= 0);
11d83360	305	might_sleep();
1da177e4 LT	306
	307	spin_lock_irqsave(&pool->lock, flags);
	308	if (new_min_nr <= pool->min_nr) {
	309	while (new_min_nr < pool->curr_nr) {
8cded866	310	element = remove_element(pool);
1da177e4 LT	311	spin_unlock_irqrestore(&pool->lock, flags);
	312	pool->free(element, pool->pool_data);
	313	spin_lock_irqsave(&pool->lock, flags);
	314	}
	315	pool->min_nr = new_min_nr;
	316	goto out_unlock;
	317	}
	318	spin_unlock_irqrestore(&pool->lock, flags);
	319
	320	/* Grow the pool */
11d83360 DR	321	new_elements = kmalloc_array(new_min_nr, sizeof(*new_elements),
11d83360 DR	322	GFP_KERNEL);
1da177e4 LT	323	if (!new_elements)
	324	return -ENOMEM;
	325
	326	spin_lock_irqsave(&pool->lock, flags);
	327	if (unlikely(new_min_nr <= pool->min_nr)) {
	328	/* Raced, other resize will do our work */
	329	spin_unlock_irqrestore(&pool->lock, flags);
	330	kfree(new_elements);
	331	goto out;
	332	}
	333	memcpy(new_elements, pool->elements,
	334	pool->curr_nr * sizeof(*new_elements));
	335	kfree(pool->elements);
	336	pool->elements = new_elements;
	337	pool->min_nr = new_min_nr;
	338
	339	while (pool->curr_nr < pool->min_nr) {
	340	spin_unlock_irqrestore(&pool->lock, flags);
11d83360	341	element = pool->alloc(GFP_KERNEL, pool->pool_data);
1da177e4 LT	342	if (!element)
	343	goto out;
	344	spin_lock_irqsave(&pool->lock, flags);
	345	if (pool->curr_nr < pool->min_nr) {
	346	add_element(pool, element);
	347	} else {
	348	spin_unlock_irqrestore(&pool->lock, flags);
	349	pool->free(element, pool->pool_data); /* Raced */
	350	goto out;
	351	}
	352	}
	353	out_unlock:
	354	spin_unlock_irqrestore(&pool->lock, flags);
	355	out:
	356	return 0;
	357	}
	358	EXPORT_SYMBOL(mempool_resize);
	359
1da177e4 LT	360	/**
	361	* mempool_alloc - allocate an element from a specific memory pool
	362	* @pool: pointer to the memory pool which was allocated via
	363	* mempool_create().
	364	* @gfp_mask: the usual allocation bitmask.
	365	*
72fd4a35	366	* this function only sleeps if the alloc_fn() function sleeps or
1da177e4 LT	367	* returns NULL. Note that due to preallocation, this function
	368	* never fails when called from process contexts. (it might
	369	* fail if called from an IRQ context.)
4e390b2b	370	* Note: using __GFP_ZERO is not supported.
a862f68a MR	371	*
a862f68a MR	372	* Return: pointer to the allocated element or %NULL on error.
1da177e4	373	*/
f9054c70	374	void mempool_alloc(mempool_t pool, gfp_t gfp_mask)
1da177e4 LT	375	{
	376	void *element;
	377	unsigned long flags;
ac6424b9	378	wait_queue_entry_t wait;
6daa0e28	379	gfp_t gfp_temp;
20a77776	380
8bf8fcb0	381	VM_WARN_ON_ONCE(gfp_mask & __GFP_ZERO);
21bfe8db	382	might_alloc(gfp_mask);
b84a35be	383
4e390b2b	384	gfp_mask \|= __GFP_NOMEMALLOC; /* don't allocate emergency reserves */
b84a35be NP	385	gfp_mask \|= __GFP_NORETRY; /* don't loop in __alloc_pages */
b84a35be NP	386	gfp_mask \|= __GFP_NOWARN; /* failures are OK */
1da177e4	387
d0164adc	388	gfp_temp = gfp_mask & ~(__GFP_DIRECT_RECLAIM\|__GFP_IO);
20a77776	389
1da177e4	390	repeat_alloc:
20a77776 NP	391
20a77776 NP	392	element = pool->alloc(gfp_temp, pool->pool_data);
1da177e4 LT	393	if (likely(element != NULL))
	394	return element;
	395
1da177e4 LT	396	spin_lock_irqsave(&pool->lock, flags);
1da177e4 LT	397	if (likely(pool->curr_nr)) {
8cded866	398	element = remove_element(pool);
1da177e4	399	spin_unlock_irqrestore(&pool->lock, flags);
5b990546 TH	400	/* paired with rmb in mempool_free(), read comment there */
5b990546 TH	401	smp_wmb();
17411962 CM	402	/*
	403	* Update the allocation stack trace as this is more useful
	404	* for debugging.
	405	*/
	406	kmemleak_update_trace(element);
1da177e4 LT	407	return element;
1da177e4 LT	408	}
1da177e4	409
1ebb7044	410	/*
d0164adc	411	* We use gfp mask w/o direct reclaim or IO for the first round. If
1ebb7044 TH	412	* alloc failed with that and @pool was empty, retry immediately.
1ebb7044 TH	413	*/
4e390b2b	414	if (gfp_temp != gfp_mask) {
1ebb7044 TH	415	spin_unlock_irqrestore(&pool->lock, flags);
	416	gfp_temp = gfp_mask;
	417	goto repeat_alloc;
	418	}
	419
d0164adc MG	420	/* We must not sleep if !__GFP_DIRECT_RECLAIM */
d0164adc MG	421	if (!(gfp_mask & __GFP_DIRECT_RECLAIM)) {
5b990546	422	spin_unlock_irqrestore(&pool->lock, flags);
1da177e4	423	return NULL;
5b990546	424	}
1da177e4	425
5b990546	426	/* Let's wait for someone else to return an element to @pool */
01890a4c	427	init_wait(&wait);
1da177e4	428	prepare_to_wait(&pool->wait, &wait, TASK_UNINTERRUPTIBLE);
1da177e4	429
5b990546 TH	430	spin_unlock_irqrestore(&pool->lock, flags);
	431
	432	/*
	433	* FIXME: this should be io_schedule(). The timeout is there as a
	434	* workaround for some DM problems in 2.6.18.
	435	*/
	436	io_schedule_timeout(5*HZ);
	437
	438	finish_wait(&pool->wait, &wait);
1da177e4 LT	439	goto repeat_alloc;
	440	}
	441	EXPORT_SYMBOL(mempool_alloc);
	442
	443	/**
	444	* mempool_free - return an element to the pool.
	445	* @element: pool element pointer.
	446	* @pool: pointer to the memory pool which was allocated via
	447	* mempool_create().
	448	*
	449	* this function only sleeps if the free_fn() function sleeps.
	450	*/
	451	void mempool_free(void element, mempool_t pool)
	452	{
	453	unsigned long flags;
	454
c80e7a82 RR	455	if (unlikely(element == NULL))
	456	return;
	457
5b990546 TH	458	/*
	459	* Paired with the wmb in mempool_alloc(). The preceding read is
	460	* for @element and the following @pool->curr_nr. This ensures
	461	* that the visible value of @pool->curr_nr is from after the
	462	* allocation of @element. This is necessary for fringe cases
	463	* where @element was passed to this task without going through
	464	* barriers.
	465	*
	466	* For example, assume @p is %NULL at the beginning and one task
	467	* performs "p = mempool_alloc(...);" while another task is doing
	468	* "while (!p) cpu_relax(); mempool_free(p, ...);". This function
	469	* may end up using curr_nr value which is from before allocation
	470	* of @p without the following rmb.
	471	*/
	472	smp_rmb();
	473
	474	/*
	475	* For correctness, we need a test which is guaranteed to trigger
	476	* if curr_nr + #allocated == min_nr. Testing curr_nr < min_nr
	477	* without locking achieves that and refilling as soon as possible
	478	* is desirable.
	479	*
	480	* Because curr_nr visible here is always a value after the
	481	* allocation of @element, any task which decremented curr_nr below
	482	* min_nr is guaranteed to see curr_nr < min_nr unless curr_nr gets
	483	* incremented to min_nr afterwards. If curr_nr gets incremented
	484	* to min_nr after the allocation of @element, the elements
	485	* allocated after that are subject to the same guarantee.
	486	*
	487	* Waiters happen iff curr_nr is 0 and the above guarantee also
	488	* ensures that there will be frees which return elements to the
	489	* pool waking up the waiters.
	490	*/
abe1de42	491	if (unlikely(READ_ONCE(pool->curr_nr) < pool->min_nr)) {
1da177e4	492	spin_lock_irqsave(&pool->lock, flags);
eb9a3c62	493	if (likely(pool->curr_nr < pool->min_nr)) {
1da177e4 LT	494	add_element(pool, element);
	495	spin_unlock_irqrestore(&pool->lock, flags);
	496	wake_up(&pool->wait);
	497	return;
	498	}
	499	spin_unlock_irqrestore(&pool->lock, flags);
	500	}
	501	pool->free(element, pool->pool_data);
	502	}
	503	EXPORT_SYMBOL(mempool_free);
	504
	505	/*
	506	* A commonly used alloc and free fn.
	507	*/
dd0fc66f	508	void mempool_alloc_slab(gfp_t gfp_mask, void pool_data)
1da177e4	509	{
fcc234f8	510	struct kmem_cache *mem = pool_data;
e244c9e6	511	VM_BUG_ON(mem->ctor);
1da177e4 LT	512	return kmem_cache_alloc(mem, gfp_mask);
	513	}
	514	EXPORT_SYMBOL(mempool_alloc_slab);
	515
	516	void mempool_free_slab(void element, void pool_data)
	517	{
fcc234f8	518	struct kmem_cache *mem = pool_data;
1da177e4 LT	519	kmem_cache_free(mem, element);
	520	}
	521	EXPORT_SYMBOL(mempool_free_slab);
6e0678f3	522
53184082 MD	523	/*
53184082 MD	524	* A commonly used alloc and free fn that kmalloc/kfrees the amount of memory
183ff22b	525	* specified by pool_data
53184082 MD	526	*/
	527	void mempool_kmalloc(gfp_t gfp_mask, void pool_data)
	528	{
5e2f89b5	529	size_t size = (size_t)pool_data;
53184082 MD	530	return kmalloc(size, gfp_mask);
	531	}
	532	EXPORT_SYMBOL(mempool_kmalloc);
	533
	534	void mempool_kfree(void element, void pool_data)
	535	{
	536	kfree(element);
	537	}
	538	EXPORT_SYMBOL(mempool_kfree);
	539
6e0678f3 MD	540	/*
	541	* A simple mempool-backed page allocator that allocates pages
	542	* of the order specified by pool_data.
	543	*/
	544	void mempool_alloc_pages(gfp_t gfp_mask, void pool_data)
	545	{
	546	int order = (int)(long)pool_data;
	547	return alloc_pages(gfp_mask, order);
	548	}
	549	EXPORT_SYMBOL(mempool_alloc_pages);
	550
	551	void mempool_free_pages(void element, void pool_data)
	552	{
	553	int order = (int)(long)pool_data;
	554	__free_pages(element, order);
	555	}
	556	EXPORT_SYMBOL(mempool_free_pages);