mempool.c 13.2 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8
/*
 *  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
9
 *  debugging by David Rientjes, Copyright (C) 2015
L
Linus Torvalds 已提交
10 11 12 13
 */

#include <linux/mm.h>
#include <linux/slab.h>
14
#include <linux/highmem.h>
15
#include <linux/kmemleak.h>
16
#include <linux/export.h>
L
Linus Torvalds 已提交
17 18 19
#include <linux/mempool.h>
#include <linux/blkdev.h>
#include <linux/writeback.h>
20
#include "slab.h"
L
Linus Torvalds 已提交
21

22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103
#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));

	/* Mempools backed by page allocator */
	if (pool->free == mempool_free_pages) {
		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));

	/* Mempools backed by page allocator */
	if (pool->alloc == mempool_alloc_pages) {
		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 */

L
Linus Torvalds 已提交
104 105 106
static void add_element(mempool_t *pool, void *element)
{
	BUG_ON(pool->curr_nr >= pool->min_nr);
107
	poison_element(pool, element);
L
Linus Torvalds 已提交
108 109 110 111 112
	pool->elements[pool->curr_nr++] = element;
}

static void *remove_element(mempool_t *pool)
{
113 114 115 116 117
	void *element = pool->elements[--pool->curr_nr];

	BUG_ON(pool->curr_nr < 0);
	check_element(pool, element);
	return element;
L
Linus Torvalds 已提交
118 119
}

120 121 122 123 124 125 126 127 128
/**
 * 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)
L
Linus Torvalds 已提交
129 130 131 132 133 134 135 136
{
	while (pool->curr_nr) {
		void *element = remove_element(pool);
		pool->free(element, pool->pool_data);
	}
	kfree(pool->elements);
	kfree(pool);
}
137
EXPORT_SYMBOL(mempool_destroy);
L
Linus Torvalds 已提交
138 139 140 141 142 143 144 145 146 147

/**
 * 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
148
 * memory pool. The pool can be used from the mempool_alloc() and mempool_free()
L
Linus Torvalds 已提交
149
 * functions. This function might sleep. Both the alloc_fn() and the free_fn()
150
 * functions might sleep - as long as the mempool_alloc() function is not called
L
Linus Torvalds 已提交
151 152
 * from IRQ contexts.
 */
153
mempool_t *mempool_create(int min_nr, mempool_alloc_t *alloc_fn,
L
Linus Torvalds 已提交
154 155
				mempool_free_t *free_fn, void *pool_data)
{
156 157
	return mempool_create_node(min_nr,alloc_fn,free_fn, pool_data,
				   GFP_KERNEL, NUMA_NO_NODE);
158 159
}
EXPORT_SYMBOL(mempool_create);
L
Linus Torvalds 已提交
160

161
mempool_t *mempool_create_node(int min_nr, mempool_alloc_t *alloc_fn,
162 163
			       mempool_free_t *free_fn, void *pool_data,
			       gfp_t gfp_mask, int node_id)
164 165
{
	mempool_t *pool;
166
	pool = kzalloc_node(sizeof(*pool), gfp_mask, node_id);
L
Linus Torvalds 已提交
167 168
	if (!pool)
		return NULL;
169
	pool->elements = kmalloc_node(min_nr * sizeof(void *),
170
				      gfp_mask, node_id);
L
Linus Torvalds 已提交
171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187
	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;

188
		element = pool->alloc(gfp_mask, pool->pool_data);
L
Linus Torvalds 已提交
189
		if (unlikely(!element)) {
190
			mempool_destroy(pool);
L
Linus Torvalds 已提交
191 192 193 194 195 196
			return NULL;
		}
		add_element(pool, element);
	}
	return pool;
}
197
EXPORT_SYMBOL(mempool_create_node);
L
Linus Torvalds 已提交
198 199 200 201 202 203 204 205 206 207 208

/**
 * 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.
209
 * This function may sleep.
L
Linus Torvalds 已提交
210 211 212 213 214
 *
 * 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.
 */
215
int mempool_resize(mempool_t *pool, int new_min_nr)
L
Linus Torvalds 已提交
216 217 218 219 220 221
{
	void *element;
	void **new_elements;
	unsigned long flags;

	BUG_ON(new_min_nr <= 0);
222
	might_sleep();
L
Linus Torvalds 已提交
223 224 225 226 227 228 229 230 231 232 233 234 235 236 237

	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 */
238 239
	new_elements = kmalloc_array(new_min_nr, sizeof(*new_elements),
				     GFP_KERNEL);
L
Linus Torvalds 已提交
240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257
	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);
258
		element = pool->alloc(GFP_KERNEL, pool->pool_data);
L
Linus Torvalds 已提交
259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282
		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.
 *
283
 * this function only sleeps if the alloc_fn() function sleeps or
L
Linus Torvalds 已提交
284 285 286
 * 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.)
287
 * Note: using __GFP_ZERO is not supported.
L
Linus Torvalds 已提交
288
 */
A
Al Viro 已提交
289
void * mempool_alloc(mempool_t *pool, gfp_t gfp_mask)
L
Linus Torvalds 已提交
290 291 292
{
	void *element;
	unsigned long flags;
293
	wait_queue_t wait;
A
Al Viro 已提交
294
	gfp_t gfp_temp;
N
Nick Piggin 已提交
295

296
	VM_WARN_ON_ONCE(gfp_mask & __GFP_ZERO);
N
Nick Piggin 已提交
297
	might_sleep_if(gfp_mask & __GFP_WAIT);
298 299 300 301

	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 */
L
Linus Torvalds 已提交
302

N
Nick Piggin 已提交
303 304
	gfp_temp = gfp_mask & ~(__GFP_WAIT|__GFP_IO);

L
Linus Torvalds 已提交
305
repeat_alloc:
N
Nick Piggin 已提交
306 307

	element = pool->alloc(gfp_temp, pool->pool_data);
L
Linus Torvalds 已提交
308 309 310 311 312 313 314
	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);
315 316
		/* paired with rmb in mempool_free(), read comment there */
		smp_wmb();
317 318 319 320 321
		/*
		 * Update the allocation stack trace as this is more useful
		 * for debugging.
		 */
		kmemleak_update_trace(element);
L
Linus Torvalds 已提交
322 323 324
		return element;
	}

325 326 327 328 329 330 331 332 333 334 335
	/*
	 * 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 */
336 337
	if (!(gfp_mask & __GFP_WAIT)) {
		spin_unlock_irqrestore(&pool->lock, flags);
L
Linus Torvalds 已提交
338
		return NULL;
339
	}
L
Linus Torvalds 已提交
340

341
	/* Let's wait for someone else to return an element to @pool */
342
	init_wait(&wait);
L
Linus Torvalds 已提交
343 344
	prepare_to_wait(&pool->wait, &wait, TASK_UNINTERRUPTIBLE);

345 346 347 348 349 350 351 352 353
	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);
L
Linus Torvalds 已提交
354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369
	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;

R
Rusty Russell 已提交
370 371 372
	if (unlikely(element == NULL))
		return;

373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405
	/*
	 * 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.
	 */
406
	if (unlikely(pool->curr_nr < pool->min_nr)) {
L
Linus Torvalds 已提交
407
		spin_lock_irqsave(&pool->lock, flags);
408
		if (likely(pool->curr_nr < pool->min_nr)) {
L
Linus Torvalds 已提交
409 410 411 412 413 414 415 416 417 418 419 420 421 422
			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.
 */
A
Al Viro 已提交
423
void *mempool_alloc_slab(gfp_t gfp_mask, void *pool_data)
L
Linus Torvalds 已提交
424
{
425
	struct kmem_cache *mem = pool_data;
426
	VM_BUG_ON(mem->ctor);
L
Linus Torvalds 已提交
427 428 429 430 431 432
	return kmem_cache_alloc(mem, gfp_mask);
}
EXPORT_SYMBOL(mempool_alloc_slab);

void mempool_free_slab(void *element, void *pool_data)
{
433
	struct kmem_cache *mem = pool_data;
L
Linus Torvalds 已提交
434 435 436
	kmem_cache_free(mem, element);
}
EXPORT_SYMBOL(mempool_free_slab);
437

438 439
/*
 * A commonly used alloc and free fn that kmalloc/kfrees the amount of memory
S
Simon Arlott 已提交
440
 * specified by pool_data
441 442 443
 */
void *mempool_kmalloc(gfp_t gfp_mask, void *pool_data)
{
F
Figo.zhang 已提交
444
	size_t size = (size_t)pool_data;
445 446 447 448 449 450 451 452 453 454
	return kmalloc(size, gfp_mask);
}
EXPORT_SYMBOL(mempool_kmalloc);

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

455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471
/*
 * 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);