slob.c 17.2 KB
Newer Older
1 2 3 4 5
/*
 * SLOB Allocator: Simple List Of Blocks
 *
 * Matt Mackall <mpm@selenic.com> 12/30/03
 *
P
Paul Mundt 已提交
6 7
 * NUMA support by Paul Mundt, 2007.
 *
8 9 10 11
 * How SLOB works:
 *
 * The core of SLOB is a traditional K&R style heap allocator, with
 * support for returning aligned objects. The granularity of this
N
Nick Piggin 已提交
12 13
 * allocator is as little as 2 bytes, however typically most architectures
 * will require 4 bytes on 32-bit and 8 bytes on 64-bit.
N
Nick Piggin 已提交
14
 *
15 16 17 18 19 20 21 22 23 24 25
 * The slob heap is a set of linked list of pages from alloc_pages(),
 * and within each page, there is a singly-linked list of free blocks
 * (slob_t). The heap is grown on demand. To reduce fragmentation,
 * heap pages are segregated into three lists, with objects less than
 * 256 bytes, objects less than 1024 bytes, and all other objects.
 *
 * Allocation from heap involves first searching for a page with
 * sufficient free blocks (using a next-fit-like approach) followed by
 * a first-fit scan of the page. Deallocation inserts objects back
 * into the free list in address order, so this is effectively an
 * address-ordered first fit.
26 27
 *
 * Above this is an implementation of kmalloc/kfree. Blocks returned
N
Nick Piggin 已提交
28
 * from kmalloc are prepended with a 4-byte header with the kmalloc size.
29
 * If kmalloc is asked for objects of PAGE_SIZE or larger, it calls
P
Paul Mundt 已提交
30
 * alloc_pages() directly, allocating compound pages so the page order
N
Nick Piggin 已提交
31 32 33 34
 * does not have to be separately tracked, and also stores the exact
 * allocation size in page->private so that it can be used to accurately
 * provide ksize(). These objects are detected in kfree() because slob_page()
 * is false for them.
35 36
 *
 * SLAB is emulated on top of SLOB by simply calling constructors and
N
Nick Piggin 已提交
37 38 39 40
 * destructors for every SLAB allocation. Objects are returned with the
 * 4-byte alignment unless the SLAB_HWCACHE_ALIGN flag is set, in which
 * case the low-level allocator will fragment blocks to create the proper
 * alignment. Again, objects of page-size or greater are allocated by
P
Paul Mundt 已提交
41
 * calling alloc_pages(). As SLAB objects know their size, no separate
N
Nick Piggin 已提交
42
 * size bookkeeping is necessary and there is essentially no allocation
N
Nick Piggin 已提交
43 44
 * space overhead, and compound pages aren't needed for multi-page
 * allocations.
P
Paul Mundt 已提交
45 46 47 48
 *
 * NUMA support in SLOB is fairly simplistic, pushing most of the real
 * logic down to the page allocator, and simply doing the node accounting
 * on the upper levels. In the event that a node id is explicitly
49
 * provided, alloc_pages_exact_node() with the specified node id is used
P
Paul Mundt 已提交
50 51 52 53 54 55 56 57
 * instead. The common case (or when the node id isn't explicitly provided)
 * will default to the current node, as per numa_node_id().
 *
 * Node aware pages are still inserted in to the global freelist, and
 * these are scanned for by matching against the node id encoded in the
 * page flags. As a result, block allocations that can be satisfied from
 * the freelist will only be done so on pages residing on the same node,
 * in order to prevent random node placement.
58 59
 */

N
Nick Piggin 已提交
60
#include <linux/kernel.h>
61 62
#include <linux/slab.h>
#include <linux/mm.h>
N
Nick Piggin 已提交
63
#include <linux/swap.h> /* struct reclaim_state */
64 65 66
#include <linux/cache.h>
#include <linux/init.h>
#include <linux/module.h>
N
Nick Piggin 已提交
67
#include <linux/rcupdate.h>
N
Nick Piggin 已提交
68
#include <linux/list.h>
69
#include <linux/kmemtrace.h>
70
#include <linux/kmemleak.h>
N
Nick Piggin 已提交
71 72 73 74 75 76 77 78 79 80
#include <asm/atomic.h>

/*
 * slob_block has a field 'units', which indicates size of block if +ve,
 * or offset of next block if -ve (in SLOB_UNITs).
 *
 * Free blocks of size 1 unit simply contain the offset of the next block.
 * Those with larger size contain their size in the first SLOB_UNIT of
 * memory, and the offset of the next free block in the second SLOB_UNIT.
 */
N
Nick Piggin 已提交
81
#if PAGE_SIZE <= (32767 * 2)
N
Nick Piggin 已提交
82 83 84 85 86
typedef s16 slobidx_t;
#else
typedef s32 slobidx_t;
#endif

87
struct slob_block {
N
Nick Piggin 已提交
88
	slobidx_t units;
N
Nick Piggin 已提交
89
};
90 91
typedef struct slob_block slob_t;

N
Nick Piggin 已提交
92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122
/*
 * We use struct page fields to manage some slob allocation aspects,
 * however to avoid the horrible mess in include/linux/mm_types.h, we'll
 * just define our own struct page type variant here.
 */
struct slob_page {
	union {
		struct {
			unsigned long flags;	/* mandatory */
			atomic_t _count;	/* mandatory */
			slobidx_t units;	/* free units left in page */
			unsigned long pad[2];
			slob_t *free;		/* first free slob_t in page */
			struct list_head list;	/* linked list of free pages */
		};
		struct page page;
	};
};
static inline void struct_slob_page_wrong_size(void)
{ BUILD_BUG_ON(sizeof(struct slob_page) != sizeof(struct page)); }

/*
 * free_slob_page: call before a slob_page is returned to the page allocator.
 */
static inline void free_slob_page(struct slob_page *sp)
{
	reset_page_mapcount(&sp->page);
	sp->page.mapping = NULL;
}

/*
123
 * All partially free slob pages go on these lists.
N
Nick Piggin 已提交
124
 */
125 126 127 128 129
#define SLOB_BREAK1 256
#define SLOB_BREAK2 1024
static LIST_HEAD(free_slob_small);
static LIST_HEAD(free_slob_medium);
static LIST_HEAD(free_slob_large);
N
Nick Piggin 已提交
130 131

/*
A
Américo Wang 已提交
132
 * is_slob_page: True for all slob pages (false for bigblock pages)
N
Nick Piggin 已提交
133
 */
A
Américo Wang 已提交
134
static inline int is_slob_page(struct slob_page *sp)
N
Nick Piggin 已提交
135
{
136
	return PageSlab((struct page *)sp);
N
Nick Piggin 已提交
137 138 139 140
}

static inline void set_slob_page(struct slob_page *sp)
{
141
	__SetPageSlab((struct page *)sp);
N
Nick Piggin 已提交
142 143 144 145
}

static inline void clear_slob_page(struct slob_page *sp)
{
146
	__ClearPageSlab((struct page *)sp);
N
Nick Piggin 已提交
147 148
}

A
Américo Wang 已提交
149 150 151 152 153
static inline struct slob_page *slob_page(const void *addr)
{
	return (struct slob_page *)virt_to_page(addr);
}

N
Nick Piggin 已提交
154 155 156 157 158
/*
 * slob_page_free: true for pages on free_slob_pages list.
 */
static inline int slob_page_free(struct slob_page *sp)
{
159
	return PageSlobFree((struct page *)sp);
N
Nick Piggin 已提交
160 161
}

162
static void set_slob_page_free(struct slob_page *sp, struct list_head *list)
N
Nick Piggin 已提交
163
{
164
	list_add(&sp->list, list);
165
	__SetPageSlobFree((struct page *)sp);
N
Nick Piggin 已提交
166 167 168 169 170
}

static inline void clear_slob_page_free(struct slob_page *sp)
{
	list_del(&sp->list);
171
	__ClearPageSlobFree((struct page *)sp);
N
Nick Piggin 已提交
172 173
}

174 175 176 177
#define SLOB_UNIT sizeof(slob_t)
#define SLOB_UNITS(size) (((size) + SLOB_UNIT - 1)/SLOB_UNIT)
#define SLOB_ALIGN L1_CACHE_BYTES

N
Nick Piggin 已提交
178 179 180 181 182 183 184 185 186 187
/*
 * struct slob_rcu is inserted at the tail of allocated slob blocks, which
 * were created with a SLAB_DESTROY_BY_RCU slab. slob_rcu is used to free
 * the block using call_rcu.
 */
struct slob_rcu {
	struct rcu_head head;
	int size;
};

N
Nick Piggin 已提交
188 189 190
/*
 * slob_lock protects all slob allocator structures.
 */
191 192
static DEFINE_SPINLOCK(slob_lock);

N
Nick Piggin 已提交
193 194 195 196 197 198 199
/*
 * Encode the given size and next info into a free slob block s.
 */
static void set_slob(slob_t *s, slobidx_t size, slob_t *next)
{
	slob_t *base = (slob_t *)((unsigned long)s & PAGE_MASK);
	slobidx_t offset = next - base;
200

N
Nick Piggin 已提交
201 202 203 204 205 206
	if (size > 1) {
		s[0].units = size;
		s[1].units = offset;
	} else
		s[0].units = -offset;
}
207

N
Nick Piggin 已提交
208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240
/*
 * Return the size of a slob block.
 */
static slobidx_t slob_units(slob_t *s)
{
	if (s->units > 0)
		return s->units;
	return 1;
}

/*
 * Return the next free slob block pointer after this one.
 */
static slob_t *slob_next(slob_t *s)
{
	slob_t *base = (slob_t *)((unsigned long)s & PAGE_MASK);
	slobidx_t next;

	if (s[0].units < 0)
		next = -s[0].units;
	else
		next = s[1].units;
	return base+next;
}

/*
 * Returns true if s is the last free block in its page.
 */
static int slob_last(slob_t *s)
{
	return !((unsigned long)slob_next(s) & ~PAGE_MASK);
}

A
Américo Wang 已提交
241
static void *slob_new_pages(gfp_t gfp, int order, int node)
P
Paul Mundt 已提交
242 243 244 245 246
{
	void *page;

#ifdef CONFIG_NUMA
	if (node != -1)
247
		page = alloc_pages_exact_node(node, gfp, order);
P
Paul Mundt 已提交
248 249 250 251 252 253 254 255 256 257
	else
#endif
		page = alloc_pages(gfp, order);

	if (!page)
		return NULL;

	return page_address(page);
}

A
Américo Wang 已提交
258 259
static void slob_free_pages(void *b, int order)
{
N
Nick Piggin 已提交
260 261
	if (current->reclaim_state)
		current->reclaim_state->reclaimed_slab += 1 << order;
A
Américo Wang 已提交
262 263 264
	free_pages((unsigned long)b, order);
}

N
Nick Piggin 已提交
265 266 267 268
/*
 * Allocate a slob block within a given slob_page sp.
 */
static void *slob_page_alloc(struct slob_page *sp, size_t size, int align)
269
{
A
Américo Wang 已提交
270
	slob_t *prev, *cur, *aligned = NULL;
271 272
	int delta = 0, units = SLOB_UNITS(size);

N
Nick Piggin 已提交
273 274 275
	for (prev = NULL, cur = sp->free; ; prev = cur, cur = slob_next(cur)) {
		slobidx_t avail = slob_units(cur);

276 277 278 279
		if (align) {
			aligned = (slob_t *)ALIGN((unsigned long)cur, align);
			delta = aligned - cur;
		}
N
Nick Piggin 已提交
280 281 282
		if (avail >= units + delta) { /* room enough? */
			slob_t *next;

283
			if (delta) { /* need to fragment head to align? */
N
Nick Piggin 已提交
284 285 286
				next = slob_next(cur);
				set_slob(aligned, avail - delta, next);
				set_slob(cur, delta, aligned);
287 288
				prev = cur;
				cur = aligned;
N
Nick Piggin 已提交
289
				avail = slob_units(cur);
290 291
			}

N
Nick Piggin 已提交
292 293 294 295 296 297 298 299 300 301 302 303
			next = slob_next(cur);
			if (avail == units) { /* exact fit? unlink. */
				if (prev)
					set_slob(prev, slob_units(prev), next);
				else
					sp->free = next;
			} else { /* fragment */
				if (prev)
					set_slob(prev, slob_units(prev), cur + units);
				else
					sp->free = cur + units;
				set_slob(cur + units, avail - units, next);
304 305
			}

N
Nick Piggin 已提交
306 307 308
			sp->units -= units;
			if (!sp->units)
				clear_slob_page_free(sp);
309 310
			return cur;
		}
N
Nick Piggin 已提交
311 312 313 314
		if (slob_last(cur))
			return NULL;
	}
}
315

N
Nick Piggin 已提交
316 317 318
/*
 * slob_alloc: entry point into the slob allocator.
 */
P
Paul Mundt 已提交
319
static void *slob_alloc(size_t size, gfp_t gfp, int align, int node)
N
Nick Piggin 已提交
320 321
{
	struct slob_page *sp;
M
Matt Mackall 已提交
322
	struct list_head *prev;
323
	struct list_head *slob_list;
N
Nick Piggin 已提交
324 325
	slob_t *b = NULL;
	unsigned long flags;
326

327 328 329 330 331 332 333
	if (size < SLOB_BREAK1)
		slob_list = &free_slob_small;
	else if (size < SLOB_BREAK2)
		slob_list = &free_slob_medium;
	else
		slob_list = &free_slob_large;

N
Nick Piggin 已提交
334 335
	spin_lock_irqsave(&slob_lock, flags);
	/* Iterate through each partially free page, try to find room */
336
	list_for_each_entry(sp, slob_list, list) {
P
Paul Mundt 已提交
337 338 339 340 341 342 343 344
#ifdef CONFIG_NUMA
		/*
		 * If there's a node specification, search for a partial
		 * page with a matching node id in the freelist.
		 */
		if (node != -1 && page_to_nid(&sp->page) != node)
			continue;
#endif
M
Matt Mackall 已提交
345 346 347
		/* Enough room on this page? */
		if (sp->units < SLOB_UNITS(size))
			continue;
P
Paul Mundt 已提交
348

M
Matt Mackall 已提交
349 350 351 352 353 354 355 356 357
		/* Attempt to alloc */
		prev = sp->list.prev;
		b = slob_page_alloc(sp, size, align);
		if (!b)
			continue;

		/* Improve fragment distribution and reduce our average
		 * search time by starting our next search here. (see
		 * Knuth vol 1, sec 2.5, pg 449) */
358 359 360
		if (prev != slob_list->prev &&
				slob_list->next != prev->next)
			list_move_tail(slob_list, prev->next);
M
Matt Mackall 已提交
361
		break;
362
	}
N
Nick Piggin 已提交
363 364 365 366
	spin_unlock_irqrestore(&slob_lock, flags);

	/* Not enough space: must allocate a new page */
	if (!b) {
A
Américo Wang 已提交
367
		b = slob_new_pages(gfp & ~__GFP_ZERO, 0, node);
N
Nick Piggin 已提交
368
		if (!b)
A
Américo Wang 已提交
369 370
			return NULL;
		sp = slob_page(b);
N
Nick Piggin 已提交
371 372 373 374 375 376 377
		set_slob_page(sp);

		spin_lock_irqsave(&slob_lock, flags);
		sp->units = SLOB_UNITS(PAGE_SIZE);
		sp->free = b;
		INIT_LIST_HEAD(&sp->list);
		set_slob(b, SLOB_UNITS(PAGE_SIZE), b + SLOB_UNITS(PAGE_SIZE));
378
		set_slob_page_free(sp, slob_list);
N
Nick Piggin 已提交
379 380 381 382
		b = slob_page_alloc(sp, size, align);
		BUG_ON(!b);
		spin_unlock_irqrestore(&slob_lock, flags);
	}
383 384
	if (unlikely((gfp & __GFP_ZERO) && b))
		memset(b, 0, size);
N
Nick Piggin 已提交
385
	return b;
386 387
}

N
Nick Piggin 已提交
388 389 390
/*
 * slob_free: entry point into the slob allocator.
 */
391 392
static void slob_free(void *block, int size)
{
N
Nick Piggin 已提交
393 394 395
	struct slob_page *sp;
	slob_t *prev, *next, *b = (slob_t *)block;
	slobidx_t units;
396 397
	unsigned long flags;

398
	if (unlikely(ZERO_OR_NULL_PTR(block)))
399
		return;
N
Nick Piggin 已提交
400
	BUG_ON(!size);
401

A
Américo Wang 已提交
402
	sp = slob_page(block);
N
Nick Piggin 已提交
403
	units = SLOB_UNITS(size);
404 405 406

	spin_lock_irqsave(&slob_lock, flags);

N
Nick Piggin 已提交
407 408 409 410
	if (sp->units + units == SLOB_UNITS(PAGE_SIZE)) {
		/* Go directly to page allocator. Do not pass slob allocator */
		if (slob_page_free(sp))
			clear_slob_page_free(sp);
N
Nick Piggin 已提交
411
		spin_unlock_irqrestore(&slob_lock, flags);
N
Nick Piggin 已提交
412 413
		clear_slob_page(sp);
		free_slob_page(sp);
N
Nick Piggin 已提交
414
		slob_free_pages(b, 0);
N
Nick Piggin 已提交
415
		return;
N
Nick Piggin 已提交
416
	}
417

N
Nick Piggin 已提交
418 419 420 421 422 423 424
	if (!slob_page_free(sp)) {
		/* This slob page is about to become partially free. Easy! */
		sp->units = units;
		sp->free = b;
		set_slob(b, units,
			(void *)((unsigned long)(b +
					SLOB_UNITS(PAGE_SIZE)) & PAGE_MASK));
425
		set_slob_page_free(sp, &free_slob_small);
N
Nick Piggin 已提交
426 427 428 429 430 431 432 433
		goto out;
	}

	/*
	 * Otherwise the page is already partially free, so find reinsertion
	 * point.
	 */
	sp->units += units;
434

N
Nick Piggin 已提交
435
	if (b < sp->free) {
436 437 438 439
		if (b + units == sp->free) {
			units += slob_units(sp->free);
			sp->free = slob_next(sp->free);
		}
N
Nick Piggin 已提交
440 441 442 443 444 445 446 447 448
		set_slob(b, units, sp->free);
		sp->free = b;
	} else {
		prev = sp->free;
		next = slob_next(prev);
		while (b > next) {
			prev = next;
			next = slob_next(prev);
		}
449

N
Nick Piggin 已提交
450 451 452 453 454 455 456 457 458 459 460 461 462
		if (!slob_last(prev) && b + units == next) {
			units += slob_units(next);
			set_slob(b, units, slob_next(next));
		} else
			set_slob(b, units, next);

		if (prev + slob_units(prev) == b) {
			units = slob_units(b) + slob_units(prev);
			set_slob(prev, units, slob_next(b));
		} else
			set_slob(prev, slob_units(prev), b);
	}
out:
463 464 465
	spin_unlock_irqrestore(&slob_lock, flags);
}

N
Nick Piggin 已提交
466 467 468 469
/*
 * End of slob allocator proper. Begin kmem_cache_alloc and kmalloc frontend.
 */

N
Nick Piggin 已提交
470 471 472 473 474 475 476 477
#ifndef ARCH_KMALLOC_MINALIGN
#define ARCH_KMALLOC_MINALIGN __alignof__(unsigned long)
#endif

#ifndef ARCH_SLAB_MINALIGN
#define ARCH_SLAB_MINALIGN __alignof__(unsigned long)
#endif

P
Paul Mundt 已提交
478
void *__kmalloc_node(size_t size, gfp_t gfp, int node)
479
{
480
	unsigned int *m;
N
Nick Piggin 已提交
481
	int align = max(ARCH_KMALLOC_MINALIGN, ARCH_SLAB_MINALIGN);
E
Eduard - Gabriel Munteanu 已提交
482
	void *ret;
N
Nick Piggin 已提交
483

484
	lockdep_trace_alloc(gfp);
485

N
Nick Piggin 已提交
486
	if (size < PAGE_SIZE - align) {
487 488 489
		if (!size)
			return ZERO_SIZE_PTR;

P
Paul Mundt 已提交
490
		m = slob_alloc(size + align, gfp, align, node);
E
Eduard - Gabriel Munteanu 已提交
491

M
MinChan Kim 已提交
492 493 494
		if (!m)
			return NULL;
		*m = size;
E
Eduard - Gabriel Munteanu 已提交
495 496
		ret = (void *)m + align;

497 498
		trace_kmalloc_node(_RET_IP_, ret,
				   size, size + align, gfp, node);
N
Nick Piggin 已提交
499
	} else {
E
Eduard - Gabriel Munteanu 已提交
500
		unsigned int order = get_order(size);
N
Nick Piggin 已提交
501

A
Américo Wang 已提交
502
		ret = slob_new_pages(gfp | __GFP_COMP, get_order(size), node);
N
Nick Piggin 已提交
503 504 505 506 507
		if (ret) {
			struct page *page;
			page = virt_to_page(ret);
			page->private = size;
		}
E
Eduard - Gabriel Munteanu 已提交
508

509 510
		trace_kmalloc_node(_RET_IP_, ret,
				   size, PAGE_SIZE << order, gfp, node);
511
	}
E
Eduard - Gabriel Munteanu 已提交
512

513
	kmemleak_alloc(ret, size, 1, gfp);
E
Eduard - Gabriel Munteanu 已提交
514
	return ret;
515
}
P
Paul Mundt 已提交
516
EXPORT_SYMBOL(__kmalloc_node);
517 518 519

void kfree(const void *block)
{
N
Nick Piggin 已提交
520
	struct slob_page *sp;
521

522 523
	trace_kfree(_RET_IP_, block);

524
	if (unlikely(ZERO_OR_NULL_PTR(block)))
525
		return;
526
	kmemleak_free(block);
527

A
Américo Wang 已提交
528 529
	sp = slob_page(block);
	if (is_slob_page(sp)) {
N
Nick Piggin 已提交
530 531 532
		int align = max(ARCH_KMALLOC_MINALIGN, ARCH_SLAB_MINALIGN);
		unsigned int *m = (unsigned int *)(block - align);
		slob_free(m, *m + align);
N
Nick Piggin 已提交
533 534
	} else
		put_page(&sp->page);
535 536 537
}
EXPORT_SYMBOL(kfree);

N
Nick Piggin 已提交
538
/* can't use ksize for kmem_cache_alloc memory, only kmalloc */
P
Pekka Enberg 已提交
539
size_t ksize(const void *block)
540
{
N
Nick Piggin 已提交
541
	struct slob_page *sp;
542

543 544
	BUG_ON(!block);
	if (unlikely(block == ZERO_SIZE_PTR))
545 546
		return 0;

A
Américo Wang 已提交
547 548
	sp = slob_page(block);
	if (is_slob_page(sp)) {
549 550 551 552
		int align = max(ARCH_KMALLOC_MINALIGN, ARCH_SLAB_MINALIGN);
		unsigned int *m = (unsigned int *)(block - align);
		return SLOB_UNITS(*m) * SLOB_UNIT;
	} else
N
Nick Piggin 已提交
553
		return sp->page.private;
554
}
K
Kirill A. Shutemov 已提交
555
EXPORT_SYMBOL(ksize);
556 557 558

struct kmem_cache {
	unsigned int size, align;
N
Nick Piggin 已提交
559
	unsigned long flags;
560
	const char *name;
561
	void (*ctor)(void *);
562 563 564
};

struct kmem_cache *kmem_cache_create(const char *name, size_t size,
565
	size_t align, unsigned long flags, void (*ctor)(void *))
566 567 568
{
	struct kmem_cache *c;

569
	c = slob_alloc(sizeof(struct kmem_cache),
570
		GFP_KERNEL, ARCH_KMALLOC_MINALIGN, -1);
571 572 573 574

	if (c) {
		c->name = name;
		c->size = size;
N
Nick Piggin 已提交
575 576 577 578 579
		if (flags & SLAB_DESTROY_BY_RCU) {
			/* leave room for rcu footer at the end of object */
			c->size += sizeof(struct slob_rcu);
		}
		c->flags = flags;
580 581
		c->ctor = ctor;
		/* ignore alignment unless it's forced */
582
		c->align = (flags & SLAB_HWCACHE_ALIGN) ? SLOB_ALIGN : 0;
N
Nick Piggin 已提交
583 584
		if (c->align < ARCH_SLAB_MINALIGN)
			c->align = ARCH_SLAB_MINALIGN;
585 586
		if (c->align < align)
			c->align = align;
A
Akinobu Mita 已提交
587 588
	} else if (flags & SLAB_PANIC)
		panic("Cannot create slab cache %s\n", name);
589

590
	kmemleak_alloc(c, sizeof(struct kmem_cache), 1, GFP_KERNEL);
591 592 593 594
	return c;
}
EXPORT_SYMBOL(kmem_cache_create);

595
void kmem_cache_destroy(struct kmem_cache *c)
596
{
597
	kmemleak_free(c);
598 599
	if (c->flags & SLAB_DESTROY_BY_RCU)
		rcu_barrier();
600 601 602 603
	slob_free(c, sizeof(struct kmem_cache));
}
EXPORT_SYMBOL(kmem_cache_destroy);

P
Paul Mundt 已提交
604
void *kmem_cache_alloc_node(struct kmem_cache *c, gfp_t flags, int node)
605 606 607
{
	void *b;

E
Eduard - Gabriel Munteanu 已提交
608
	if (c->size < PAGE_SIZE) {
P
Paul Mundt 已提交
609
		b = slob_alloc(c->size, flags, c->align, node);
610 611 612
		trace_kmem_cache_alloc_node(_RET_IP_, b, c->size,
					    SLOB_UNITS(c->size) * SLOB_UNIT,
					    flags, node);
E
Eduard - Gabriel Munteanu 已提交
613
	} else {
A
Américo Wang 已提交
614
		b = slob_new_pages(flags, get_order(c->size), node);
615 616 617
		trace_kmem_cache_alloc_node(_RET_IP_, b, c->size,
					    PAGE_SIZE << get_order(c->size),
					    flags, node);
E
Eduard - Gabriel Munteanu 已提交
618
	}
619 620

	if (c->ctor)
621
		c->ctor(b);
622

623
	kmemleak_alloc_recursive(b, c->size, 1, c->flags, flags);
624 625
	return b;
}
P
Paul Mundt 已提交
626
EXPORT_SYMBOL(kmem_cache_alloc_node);
627

N
Nick Piggin 已提交
628
static void __kmem_cache_free(void *b, int size)
629
{
N
Nick Piggin 已提交
630 631
	if (size < PAGE_SIZE)
		slob_free(b, size);
632
	else
A
Américo Wang 已提交
633
		slob_free_pages(b, get_order(size));
N
Nick Piggin 已提交
634 635 636 637 638 639 640 641 642 643 644 645
}

static void kmem_rcu_free(struct rcu_head *head)
{
	struct slob_rcu *slob_rcu = (struct slob_rcu *)head;
	void *b = (void *)slob_rcu - (slob_rcu->size - sizeof(struct slob_rcu));

	__kmem_cache_free(b, slob_rcu->size);
}

void kmem_cache_free(struct kmem_cache *c, void *b)
{
646
	kmemleak_free_recursive(b, c->flags);
N
Nick Piggin 已提交
647 648 649 650 651 652 653 654 655
	if (unlikely(c->flags & SLAB_DESTROY_BY_RCU)) {
		struct slob_rcu *slob_rcu;
		slob_rcu = b + (c->size - sizeof(struct slob_rcu));
		INIT_RCU_HEAD(&slob_rcu->head);
		slob_rcu->size = c->size;
		call_rcu(&slob_rcu->head, kmem_rcu_free);
	} else {
		__kmem_cache_free(b, c->size);
	}
E
Eduard - Gabriel Munteanu 已提交
656

657
	trace_kmem_cache_free(_RET_IP_, b);
658 659 660 661 662 663 664 665 666 667 668 669 670 671 672
}
EXPORT_SYMBOL(kmem_cache_free);

unsigned int kmem_cache_size(struct kmem_cache *c)
{
	return c->size;
}
EXPORT_SYMBOL(kmem_cache_size);

const char *kmem_cache_name(struct kmem_cache *c)
{
	return c->name;
}
EXPORT_SYMBOL(kmem_cache_name);

673 674 675 676 677 678
int kmem_cache_shrink(struct kmem_cache *d)
{
	return 0;
}
EXPORT_SYMBOL(kmem_cache_shrink);

C
Christoph Lameter 已提交
679
int kmem_ptr_validate(struct kmem_cache *a, const void *b)
680 681 682 683
{
	return 0;
}

P
Paul Mundt 已提交
684 685 686 687 688 689 690
static unsigned int slob_ready __read_mostly;

int slab_is_available(void)
{
	return slob_ready;
}

691 692
void __init kmem_cache_init(void)
{
P
Paul Mundt 已提交
693
	slob_ready = 1;
694
}
695 696 697 698 699

void __init kmem_cache_init_late(void)
{
	/* Nothing to do */
}