mempool.c 10.7 KB
Newer Older
Linus Torvalds's avatar
Linus Torvalds committed
1
2
3
4
5
6
7
8
9
10
11
12
/*
 *  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
 */

#include <linux/mm.h>
#include <linux/slab.h>
13
#include <linux/kmemleak.h>
14
#include <linux/export.h>
Linus Torvalds's avatar
Linus Torvalds committed
15
16
17
#include <linux/mempool.h>
#include <linux/blkdev.h>
#include <linux/writeback.h>
18
#include "slab.h"
Linus Torvalds's avatar
Linus Torvalds committed
19
20
21
22
23
24
25
26
27
28
29
30
31

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

static void *remove_element(mempool_t *pool)
{
	BUG_ON(pool->curr_nr <= 0);
	return pool->elements[--pool->curr_nr];
}

32
33
34
35
36
37
38
39
40
/**
 * 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)
Linus Torvalds's avatar
Linus Torvalds committed
41
42
43
44
45
46
47
48
{
	while (pool->curr_nr) {
		void *element = remove_element(pool);
		pool->free(element, pool->pool_data);
	}
	kfree(pool->elements);
	kfree(pool);
}
49
EXPORT_SYMBOL(mempool_destroy);
Linus Torvalds's avatar
Linus Torvalds committed
50
51
52
53
54
55
56
57
58
59

/**
 * 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
60
 * memory pool. The pool can be used from the mempool_alloc() and mempool_free()
Linus Torvalds's avatar
Linus Torvalds committed
61
 * functions. This function might sleep. Both the alloc_fn() and the free_fn()
62
 * functions might sleep - as long as the mempool_alloc() function is not called
Linus Torvalds's avatar
Linus Torvalds committed
63
64
 * from IRQ contexts.
 */
65
mempool_t *mempool_create(int min_nr, mempool_alloc_t *alloc_fn,
Linus Torvalds's avatar
Linus Torvalds committed
66
67
				mempool_free_t *free_fn, void *pool_data)
{
68
69
	return mempool_create_node(min_nr,alloc_fn,free_fn, pool_data,
				   GFP_KERNEL, NUMA_NO_NODE);
70
71
}
EXPORT_SYMBOL(mempool_create);
Linus Torvalds's avatar
Linus Torvalds committed
72

73
mempool_t *mempool_create_node(int min_nr, mempool_alloc_t *alloc_fn,
74
75
			       mempool_free_t *free_fn, void *pool_data,
			       gfp_t gfp_mask, int node_id)
76
77
{
	mempool_t *pool;
78
	pool = kzalloc_node(sizeof(*pool), gfp_mask, node_id);
Linus Torvalds's avatar
Linus Torvalds committed
79
80
	if (!pool)
		return NULL;
81
	pool->elements = kmalloc_node(min_nr * sizeof(void *),
82
				      gfp_mask, node_id);
Linus Torvalds's avatar
Linus Torvalds committed
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
	if (!pool->elements) {
		kfree(pool);
		return NULL;
	}
	spin_lock_init(&pool->lock);
	pool->min_nr = min_nr;
	pool->pool_data = pool_data;
	init_waitqueue_head(&pool->wait);
	pool->alloc = alloc_fn;
	pool->free = free_fn;

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

100
		element = pool->alloc(gfp_mask, pool->pool_data);
Linus Torvalds's avatar
Linus Torvalds committed
101
		if (unlikely(!element)) {
102
			mempool_destroy(pool);
Linus Torvalds's avatar
Linus Torvalds committed
103
104
105
106
107
108
			return NULL;
		}
		add_element(pool, element);
	}
	return pool;
}
109
EXPORT_SYMBOL(mempool_create_node);
Linus Torvalds's avatar
Linus Torvalds committed
110
111
112
113
114
115
116
117
118
119
120

/**
 * 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.
121
 * This function may sleep.
Linus Torvalds's avatar
Linus Torvalds committed
122
123
124
125
126
 *
 * 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.
 */
127
int mempool_resize(mempool_t *pool, int new_min_nr)
Linus Torvalds's avatar
Linus Torvalds committed
128
129
130
131
132
133
{
	void *element;
	void **new_elements;
	unsigned long flags;

	BUG_ON(new_min_nr <= 0);
134
	might_sleep();
Linus Torvalds's avatar
Linus Torvalds committed
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149

	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 */
150
151
	new_elements = kmalloc_array(new_min_nr, sizeof(*new_elements),
				     GFP_KERNEL);
Linus Torvalds's avatar
Linus Torvalds committed
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
	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);
170
		element = pool->alloc(GFP_KERNEL, pool->pool_data);
Linus Torvalds's avatar
Linus Torvalds committed
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
		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.
 *
195
 * this function only sleeps if the alloc_fn() function sleeps or
Linus Torvalds's avatar
Linus Torvalds committed
196
197
198
 * 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.)
199
 * Note: using __GFP_ZERO is not supported.
Linus Torvalds's avatar
Linus Torvalds committed
200
 */
201
void * mempool_alloc(mempool_t *pool, gfp_t gfp_mask)
Linus Torvalds's avatar
Linus Torvalds committed
202
203
204
{
	void *element;
	unsigned long flags;
205
	wait_queue_t wait;
Al Viro's avatar
Al Viro committed
206
	gfp_t gfp_temp;
Nick Piggin's avatar
Nick Piggin committed
207

208
	VM_WARN_ON_ONCE(gfp_mask & __GFP_ZERO);
Nick Piggin's avatar
Nick Piggin committed
209
	might_sleep_if(gfp_mask & __GFP_WAIT);
210
211
212
213

	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 */
Linus Torvalds's avatar
Linus Torvalds committed
214

Nick Piggin's avatar
Nick Piggin committed
215
216
	gfp_temp = gfp_mask & ~(__GFP_WAIT|__GFP_IO);

Linus Torvalds's avatar
Linus Torvalds committed
217
repeat_alloc:
Nick Piggin's avatar
Nick Piggin committed
218
219

	element = pool->alloc(gfp_temp, pool->pool_data);
Linus Torvalds's avatar
Linus Torvalds committed
220
221
222
223
224
225
226
	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);
227
228
		/* paired with rmb in mempool_free(), read comment there */
		smp_wmb();
229
230
231
232
233
		/*
		 * Update the allocation stack trace as this is more useful
		 * for debugging.
		 */
		kmemleak_update_trace(element);
Linus Torvalds's avatar
Linus Torvalds committed
234
235
236
		return element;
	}

237
238
239
240
241
242
243
244
245
246
247
	/*
	 * We use gfp mask w/o __GFP_WAIT 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_WAIT */
248
249
	if (!(gfp_mask & __GFP_WAIT)) {
		spin_unlock_irqrestore(&pool->lock, flags);
Linus Torvalds's avatar
Linus Torvalds committed
250
		return NULL;
251
	}
Linus Torvalds's avatar
Linus Torvalds committed
252

253
	/* Let's wait for someone else to return an element to @pool */
254
	init_wait(&wait);
Linus Torvalds's avatar
Linus Torvalds committed
255
256
	prepare_to_wait(&pool->wait, &wait, TASK_UNINTERRUPTIBLE);

257
258
259
260
261
262
263
264
265
	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);
Linus Torvalds's avatar
Linus Torvalds committed
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
	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;

Rusty Russell's avatar
Rusty Russell committed
282
283
284
	if (unlikely(element == NULL))
		return;

285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
	/*
	 * 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.
	 */
318
	if (unlikely(pool->curr_nr < pool->min_nr)) {
Linus Torvalds's avatar
Linus Torvalds committed
319
		spin_lock_irqsave(&pool->lock, flags);
320
		if (likely(pool->curr_nr < pool->min_nr)) {
Linus Torvalds's avatar
Linus Torvalds committed
321
322
323
324
325
326
327
328
329
330
331
332
333
334
			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.
 */
335
void *mempool_alloc_slab(gfp_t gfp_mask, void *pool_data)
Linus Torvalds's avatar
Linus Torvalds committed
336
{
337
	struct kmem_cache *mem = pool_data;
338
	VM_BUG_ON(mem->ctor);
Linus Torvalds's avatar
Linus Torvalds committed
339
340
341
342
343
344
	return kmem_cache_alloc(mem, gfp_mask);
}
EXPORT_SYMBOL(mempool_alloc_slab);

void mempool_free_slab(void *element, void *pool_data)
{
345
	struct kmem_cache *mem = pool_data;
Linus Torvalds's avatar
Linus Torvalds committed
346
347
348
	kmem_cache_free(mem, element);
}
EXPORT_SYMBOL(mempool_free_slab);
349

350
351
/*
 * A commonly used alloc and free fn that kmalloc/kfrees the amount of memory
Simon Arlott's avatar
Simon Arlott committed
352
 * specified by pool_data
353
354
355
 */
void *mempool_kmalloc(gfp_t gfp_mask, void *pool_data)
{
Figo.zhang's avatar
Figo.zhang committed
356
	size_t size = (size_t)pool_data;
357
358
359
360
361
362
363
364
365
366
	return kmalloc(size, gfp_mask);
}
EXPORT_SYMBOL(mempool_kmalloc);

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

367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
/*
 * 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);