slab.c 118 KB
Newer Older
Linus Torvalds's avatar
Linus Torvalds committed
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
/*
 * linux/mm/slab.c
 * Written by Mark Hemment, 1996/97.
 * (markhe@nextd.demon.co.uk)
 *
 * kmem_cache_destroy() + some cleanup - 1999 Andrea Arcangeli
 *
 * Major cleanup, different bufctl logic, per-cpu arrays
 *	(c) 2000 Manfred Spraul
 *
 * Cleanup, make the head arrays unconditional, preparation for NUMA
 * 	(c) 2002 Manfred Spraul
 *
 * An implementation of the Slab Allocator as described in outline in;
 *	UNIX Internals: The New Frontiers by Uresh Vahalia
 *	Pub: Prentice Hall	ISBN 0-13-101908-2
 * or with a little more detail in;
 *	The Slab Allocator: An Object-Caching Kernel Memory Allocator
 *	Jeff Bonwick (Sun Microsystems).
 *	Presented at: USENIX Summer 1994 Technical Conference
 *
 * The memory is organized in caches, one cache for each object type.
 * (e.g. inode_cache, dentry_cache, buffer_head, vm_area_struct)
 * Each cache consists out of many slabs (they are small (usually one
 * page long) and always contiguous), and each slab contains multiple
 * initialized objects.
 *
 * This means, that your constructor is used only for newly allocated
Simon Arlott's avatar
Simon Arlott committed
29
 * slabs and you must pass objects with the same initializations to
Linus Torvalds's avatar
Linus Torvalds committed
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
 * kmem_cache_free.
 *
 * Each cache can only support one memory type (GFP_DMA, GFP_HIGHMEM,
 * normal). If you need a special memory type, then must create a new
 * cache for that memory type.
 *
 * In order to reduce fragmentation, the slabs are sorted in 3 groups:
 *   full slabs with 0 free objects
 *   partial slabs
 *   empty slabs with no allocated objects
 *
 * If partial slabs exist, then new allocations come from these slabs,
 * otherwise from empty slabs or new slabs are allocated.
 *
 * kmem_cache_destroy() CAN CRASH if you try to allocate from the cache
 * during kmem_cache_destroy(). The caller must prevent concurrent allocs.
 *
 * Each cache has a short per-cpu head array, most allocs
 * and frees go into that array, and if that array overflows, then 1/2
 * of the entries in the array are given back into the global cache.
 * The head array is strictly LIFO and should improve the cache hit rates.
 * On SMP, it additionally reduces the spinlock operations.
 *
Andrew Morton's avatar
Andrew Morton committed
53
 * The c_cpuarray may not be read with enabled local interrupts -
Linus Torvalds's avatar
Linus Torvalds committed
54
55
56
57
 * it's changed with a smp_call_function().
 *
 * SMP synchronization:
 *  constructors and destructors are called without any locking.
58
 *  Several members in struct kmem_cache and struct slab never change, they
Linus Torvalds's avatar
Linus Torvalds committed
59
60
61
62
63
64
65
66
67
68
69
70
 *	are accessed without any locking.
 *  The per-cpu arrays are never accessed from the wrong cpu, no locking,
 *  	and local interrupts are disabled so slab code is preempt-safe.
 *  The non-constant members are protected with a per-cache irq spinlock.
 *
 * Many thanks to Mark Hemment, who wrote another per-cpu slab patch
 * in 2000 - many ideas in the current implementation are derived from
 * his patch.
 *
 * Further notes from the original documentation:
 *
 * 11 April '97.  Started multi-threading - markhe
Ingo Molnar's avatar
Ingo Molnar committed
71
 *	The global cache-chain is protected by the mutex 'cache_chain_mutex'.
Linus Torvalds's avatar
Linus Torvalds committed
72
73
74
75
76
77
 *	The sem is only needed when accessing/extending the cache-chain, which
 *	can never happen inside an interrupt (kmem_cache_create(),
 *	kmem_cache_shrink() and kmem_cache_reap()).
 *
 *	At present, each engine can be growing a cache.  This should be blocked.
 *
78
79
80
81
82
83
84
85
86
 * 15 March 2005. NUMA slab allocator.
 *	Shai Fultheim <shai@scalex86.org>.
 *	Shobhit Dayal <shobhit@calsoftinc.com>
 *	Alok N Kataria <alokk@calsoftinc.com>
 *	Christoph Lameter <christoph@lameter.com>
 *
 *	Modified the slab allocator to be node aware on NUMA systems.
 *	Each node has its own list of partial, free and full slabs.
 *	All object allocations for a node occur from node specific slab lists.
Linus Torvalds's avatar
Linus Torvalds committed
87
88
89
90
 */

#include	<linux/slab.h>
#include	<linux/mm.h>
91
#include	<linux/poison.h>
Linus Torvalds's avatar
Linus Torvalds committed
92
93
94
95
96
#include	<linux/swap.h>
#include	<linux/cache.h>
#include	<linux/interrupt.h>
#include	<linux/init.h>
#include	<linux/compiler.h>
97
#include	<linux/cpuset.h>
98
#include	<linux/proc_fs.h>
Linus Torvalds's avatar
Linus Torvalds committed
99
100
101
102
103
104
105
#include	<linux/seq_file.h>
#include	<linux/notifier.h>
#include	<linux/kallsyms.h>
#include	<linux/cpu.h>
#include	<linux/sysctl.h>
#include	<linux/module.h>
#include	<linux/rcupdate.h>
106
#include	<linux/string.h>
107
#include	<linux/uaccess.h>
108
#include	<linux/nodemask.h>
109
#include	<linux/kmemleak.h>
110
#include	<linux/mempolicy.h>
Ingo Molnar's avatar
Ingo Molnar committed
111
#include	<linux/mutex.h>
112
#include	<linux/fault-inject.h>
Ingo Molnar's avatar
Ingo Molnar committed
113
#include	<linux/rtmutex.h>
114
#include	<linux/reciprocal_div.h>
115
#include	<linux/debugobjects.h>
Pekka Enberg's avatar
Pekka Enberg committed
116
#include	<linux/kmemcheck.h>
117
#include	<linux/memory.h>
Linus Torvalds's avatar
Linus Torvalds committed
118
119
120
121
122
123

#include	<asm/cacheflush.h>
#include	<asm/tlbflush.h>
#include	<asm/page.h>

/*
124
 * DEBUG	- 1 for kmem_cache_create() to honour; SLAB_RED_ZONE & SLAB_POISON.
Linus Torvalds's avatar
Linus Torvalds committed
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
 *		  0 for faster, smaller code (especially in the critical paths).
 *
 * STATS	- 1 to collect stats for /proc/slabinfo.
 *		  0 for faster, smaller code (especially in the critical paths).
 *
 * FORCED_DEBUG	- 1 enables SLAB_RED_ZONE and SLAB_POISON (if possible)
 */

#ifdef CONFIG_DEBUG_SLAB
#define	DEBUG		1
#define	STATS		1
#define	FORCED_DEBUG	1
#else
#define	DEBUG		0
#define	STATS		0
#define	FORCED_DEBUG	0
#endif

/* Shouldn't this be in a header file somewhere? */
#define	BYTES_PER_WORD		sizeof(void *)
David Woodhouse's avatar
David Woodhouse committed
145
#define	REDZONE_ALIGN		max(BYTES_PER_WORD, __alignof__(unsigned long long))
Linus Torvalds's avatar
Linus Torvalds committed
146
147
148
149
150
151
152

#ifndef ARCH_KMALLOC_FLAGS
#define ARCH_KMALLOC_FLAGS SLAB_HWCACHE_ALIGN
#endif

/* Legal flag mask for kmem_cache_create(). */
#if DEBUG
153
# define CREATE_MASK	(SLAB_RED_ZONE | \
Linus Torvalds's avatar
Linus Torvalds committed
154
			 SLAB_POISON | SLAB_HWCACHE_ALIGN | \
155
			 SLAB_CACHE_DMA | \
156
			 SLAB_STORE_USER | \
Linus Torvalds's avatar
Linus Torvalds committed
157
			 SLAB_RECLAIM_ACCOUNT | SLAB_PANIC | \
158
			 SLAB_DESTROY_BY_RCU | SLAB_MEM_SPREAD | \
Pekka Enberg's avatar
Pekka Enberg committed
159
			 SLAB_DEBUG_OBJECTS | SLAB_NOLEAKTRACE | SLAB_NOTRACK)
Linus Torvalds's avatar
Linus Torvalds committed
160
#else
161
# define CREATE_MASK	(SLAB_HWCACHE_ALIGN | \
162
			 SLAB_CACHE_DMA | \
Linus Torvalds's avatar
Linus Torvalds committed
163
			 SLAB_RECLAIM_ACCOUNT | SLAB_PANIC | \
164
			 SLAB_DESTROY_BY_RCU | SLAB_MEM_SPREAD | \
Pekka Enberg's avatar
Pekka Enberg committed
165
			 SLAB_DEBUG_OBJECTS | SLAB_NOLEAKTRACE | SLAB_NOTRACK)
Linus Torvalds's avatar
Linus Torvalds committed
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
#endif

/*
 * kmem_bufctl_t:
 *
 * Bufctl's are used for linking objs within a slab
 * linked offsets.
 *
 * This implementation relies on "struct page" for locating the cache &
 * slab an object belongs to.
 * This allows the bufctl structure to be small (one int), but limits
 * the number of objects a slab (not a cache) can contain when off-slab
 * bufctls are used. The limit is the size of the largest general cache
 * that does not use off-slab slabs.
 * For 32bit archs with 4 kB pages, is this 56.
 * This is not serious, as it is only for large objects, when it is unwise
 * to have too many per slab.
 * Note: This limit can be raised by introducing a general cache whose size
 * is less than 512 (PAGE_SIZE<<3), but greater than 256.
 */

187
typedef unsigned int kmem_bufctl_t;
Linus Torvalds's avatar
Linus Torvalds committed
188
189
#define BUFCTL_END	(((kmem_bufctl_t)(~0U))-0)
#define BUFCTL_FREE	(((kmem_bufctl_t)(~0U))-1)
190
191
#define	BUFCTL_ACTIVE	(((kmem_bufctl_t)(~0U))-2)
#define	SLAB_LIMIT	(((kmem_bufctl_t)(~0U))-3)
Linus Torvalds's avatar
Linus Torvalds committed
192
193
194
195
196
197
198
199
200

/*
 * struct slab
 *
 * Manages the objs in a slab. Placed either at the beginning of mem allocated
 * for a slab, or allocated from an general cache.
 * Slabs are chained into three list: fully used, partial, fully free slabs.
 */
struct slab {
201
202
203
204
205
206
	struct list_head list;
	unsigned long colouroff;
	void *s_mem;		/* including colour offset */
	unsigned int inuse;	/* num of objs active in slab */
	kmem_bufctl_t free;
	unsigned short nodeid;
Linus Torvalds's avatar
Linus Torvalds committed
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
};

/*
 * struct slab_rcu
 *
 * slab_destroy on a SLAB_DESTROY_BY_RCU cache uses this structure to
 * arrange for kmem_freepages to be called via RCU.  This is useful if
 * we need to approach a kernel structure obliquely, from its address
 * obtained without the usual locking.  We can lock the structure to
 * stabilize it and check it's still at the given address, only if we
 * can be sure that the memory has not been meanwhile reused for some
 * other kind of object (which our subsystem's lock might corrupt).
 *
 * rcu_read_lock before reading the address, then rcu_read_unlock after
 * taking the spinlock within the structure expected at that address.
 *
 * We assume struct slab_rcu can overlay struct slab when destroying.
 */
struct slab_rcu {
226
	struct rcu_head head;
227
	struct kmem_cache *cachep;
228
	void *addr;
Linus Torvalds's avatar
Linus Torvalds committed
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
};

/*
 * struct array_cache
 *
 * Purpose:
 * - LIFO ordering, to hand out cache-warm objects from _alloc
 * - reduce the number of linked list operations
 * - reduce spinlock operations
 *
 * The limit is stored in the per-cpu structure to reduce the data cache
 * footprint.
 *
 */
struct array_cache {
	unsigned int avail;
	unsigned int limit;
	unsigned int batchcount;
	unsigned int touched;
248
	spinlock_t lock;
249
	void *entry[];	/*
Andrew Morton's avatar
Andrew Morton committed
250
251
252
253
			 * Must have this definition in here for the proper
			 * alignment of array_cache. Also simplifies accessing
			 * the entries.
			 */
Linus Torvalds's avatar
Linus Torvalds committed
254
255
};

Andrew Morton's avatar
Andrew Morton committed
256
257
258
/*
 * bootstrap: The caches do not work without cpuarrays anymore, but the
 * cpuarrays are allocated from the generic caches...
Linus Torvalds's avatar
Linus Torvalds committed
259
260
261
262
 */
#define BOOT_CPUCACHE_ENTRIES	1
struct arraycache_init {
	struct array_cache cache;
263
	void *entries[BOOT_CPUCACHE_ENTRIES];
Linus Torvalds's avatar
Linus Torvalds committed
264
265
266
};

/*
267
 * The slab lists for all objects.
Linus Torvalds's avatar
Linus Torvalds committed
268
269
 */
struct kmem_list3 {
270
271
272
273
274
	struct list_head slabs_partial;	/* partial list first, better asm code */
	struct list_head slabs_full;
	struct list_head slabs_free;
	unsigned long free_objects;
	unsigned int free_limit;
275
	unsigned int colour_next;	/* Per-node cache coloring */
276
277
278
	spinlock_t list_lock;
	struct array_cache *shared;	/* shared per node */
	struct array_cache **alien;	/* on other nodes */
279
280
	unsigned long next_reap;	/* updated without locking */
	int free_touched;		/* updated without locking */
Linus Torvalds's avatar
Linus Torvalds committed
281
282
};

283
284
285
/*
 * Need this for bootstrapping a per node allocator.
 */
286
#define NUM_INIT_LISTS (3 * MAX_NUMNODES)
287
static struct kmem_list3 __initdata initkmem_list3[NUM_INIT_LISTS];
288
#define	CACHE_CACHE 0
289
290
#define	SIZE_AC MAX_NUMNODES
#define	SIZE_L3 (2 * MAX_NUMNODES)
291

292
293
294
295
static int drain_freelist(struct kmem_cache *cache,
			struct kmem_list3 *l3, int tofree);
static void free_block(struct kmem_cache *cachep, void **objpp, int len,
			int node);
296
static int enable_cpucache(struct kmem_cache *cachep, gfp_t gfp);
297
static void cache_reap(struct work_struct *unused);
298

299
/*
Andrew Morton's avatar
Andrew Morton committed
300
301
 * This function must be completely optimized away if a constant is passed to
 * it.  Mostly the same as what is in linux/slab.h except it returns an index.
302
 */
303
static __always_inline int index_of(const size_t size)
304
{
305
306
	extern void __bad_size(void);

307
308
309
310
311
312
313
314
	if (__builtin_constant_p(size)) {
		int i = 0;

#define CACHE(x) \
	if (size <=x) \
		return i; \
	else \
		i++;
315
#include <linux/kmalloc_sizes.h>
316
#undef CACHE
317
		__bad_size();
318
	} else
319
		__bad_size();
320
321
322
	return 0;
}

323
324
static int slab_early_init = 1;

325
326
#define INDEX_AC index_of(sizeof(struct arraycache_init))
#define INDEX_L3 index_of(sizeof(struct kmem_list3))
Linus Torvalds's avatar
Linus Torvalds committed
327

Pekka Enberg's avatar
Pekka Enberg committed
328
static void kmem_list3_init(struct kmem_list3 *parent)
329
330
331
332
333
334
{
	INIT_LIST_HEAD(&parent->slabs_full);
	INIT_LIST_HEAD(&parent->slabs_partial);
	INIT_LIST_HEAD(&parent->slabs_free);
	parent->shared = NULL;
	parent->alien = NULL;
335
	parent->colour_next = 0;
336
337
338
339
340
	spin_lock_init(&parent->list_lock);
	parent->free_objects = 0;
	parent->free_touched = 0;
}

Andrew Morton's avatar
Andrew Morton committed
341
342
343
344
#define MAKE_LIST(cachep, listp, slab, nodeid)				\
	do {								\
		INIT_LIST_HEAD(listp);					\
		list_splice(&(cachep->nodelists[nodeid]->slab), listp);	\
345
346
	} while (0)

Andrew Morton's avatar
Andrew Morton committed
347
348
#define	MAKE_ALL_LISTS(cachep, ptr, nodeid)				\
	do {								\
349
350
351
352
	MAKE_LIST((cachep), (&(ptr)->slabs_full), slabs_full, nodeid);	\
	MAKE_LIST((cachep), (&(ptr)->slabs_partial), slabs_partial, nodeid); \
	MAKE_LIST((cachep), (&(ptr)->slabs_free), slabs_free, nodeid);	\
	} while (0)
Linus Torvalds's avatar
Linus Torvalds committed
353
354
355
356
357

#define CFLGS_OFF_SLAB		(0x80000000UL)
#define	OFF_SLAB(x)	((x)->flags & CFLGS_OFF_SLAB)

#define BATCHREFILL_LIMIT	16
Andrew Morton's avatar
Andrew Morton committed
358
359
360
/*
 * Optimization question: fewer reaps means less probability for unnessary
 * cpucache drain/refill cycles.
Linus Torvalds's avatar
Linus Torvalds committed
361
 *
Adrian Bunk's avatar
Adrian Bunk committed
362
 * OTOH the cpuarrays can contain lots of objects,
Linus Torvalds's avatar
Linus Torvalds committed
363
364
365
366
367
368
369
370
371
372
 * which could lock up otherwise freeable slabs.
 */
#define REAPTIMEOUT_CPUC	(2*HZ)
#define REAPTIMEOUT_LIST3	(4*HZ)

#if STATS
#define	STATS_INC_ACTIVE(x)	((x)->num_active++)
#define	STATS_DEC_ACTIVE(x)	((x)->num_active--)
#define	STATS_INC_ALLOCED(x)	((x)->num_allocations++)
#define	STATS_INC_GROWN(x)	((x)->grown++)
373
#define	STATS_ADD_REAPED(x,y)	((x)->reaped += (y))
Andrew Morton's avatar
Andrew Morton committed
374
375
376
377
378
#define	STATS_SET_HIGH(x)						\
	do {								\
		if ((x)->num_active > (x)->high_mark)			\
			(x)->high_mark = (x)->num_active;		\
	} while (0)
Linus Torvalds's avatar
Linus Torvalds committed
379
380
#define	STATS_INC_ERR(x)	((x)->errors++)
#define	STATS_INC_NODEALLOCS(x)	((x)->node_allocs++)
381
#define	STATS_INC_NODEFREES(x)	((x)->node_frees++)
382
#define STATS_INC_ACOVERFLOW(x)   ((x)->node_overflow++)
Andrew Morton's avatar
Andrew Morton committed
383
384
385
386
387
#define	STATS_SET_FREEABLE(x, i)					\
	do {								\
		if ((x)->max_freeable < i)				\
			(x)->max_freeable = i;				\
	} while (0)
Linus Torvalds's avatar
Linus Torvalds committed
388
389
390
391
392
393
394
395
396
#define STATS_INC_ALLOCHIT(x)	atomic_inc(&(x)->allochit)
#define STATS_INC_ALLOCMISS(x)	atomic_inc(&(x)->allocmiss)
#define STATS_INC_FREEHIT(x)	atomic_inc(&(x)->freehit)
#define STATS_INC_FREEMISS(x)	atomic_inc(&(x)->freemiss)
#else
#define	STATS_INC_ACTIVE(x)	do { } while (0)
#define	STATS_DEC_ACTIVE(x)	do { } while (0)
#define	STATS_INC_ALLOCED(x)	do { } while (0)
#define	STATS_INC_GROWN(x)	do { } while (0)
397
#define	STATS_ADD_REAPED(x,y)	do { (void)(y); } while (0)
Linus Torvalds's avatar
Linus Torvalds committed
398
399
400
#define	STATS_SET_HIGH(x)	do { } while (0)
#define	STATS_INC_ERR(x)	do { } while (0)
#define	STATS_INC_NODEALLOCS(x)	do { } while (0)
401
#define	STATS_INC_NODEFREES(x)	do { } while (0)
402
#define STATS_INC_ACOVERFLOW(x)   do { } while (0)
Andrew Morton's avatar
Andrew Morton committed
403
#define	STATS_SET_FREEABLE(x, i) do { } while (0)
Linus Torvalds's avatar
Linus Torvalds committed
404
405
406
407
408
409
410
411
#define STATS_INC_ALLOCHIT(x)	do { } while (0)
#define STATS_INC_ALLOCMISS(x)	do { } while (0)
#define STATS_INC_FREEHIT(x)	do { } while (0)
#define STATS_INC_FREEMISS(x)	do { } while (0)
#endif

#if DEBUG

Andrew Morton's avatar
Andrew Morton committed
412
413
/*
 * memory layout of objects:
Linus Torvalds's avatar
Linus Torvalds committed
414
 * 0		: objp
415
 * 0 .. cachep->obj_offset - BYTES_PER_WORD - 1: padding. This ensures that
Linus Torvalds's avatar
Linus Torvalds committed
416
417
 * 		the end of an object is aligned with the end of the real
 * 		allocation. Catches writes behind the end of the allocation.
418
 * cachep->obj_offset - BYTES_PER_WORD .. cachep->obj_offset - 1:
Linus Torvalds's avatar
Linus Torvalds committed
419
 * 		redzone word.
420
421
 * cachep->obj_offset: The real object.
 * cachep->buffer_size - 2* BYTES_PER_WORD: redzone word [BYTES_PER_WORD long]
Andrew Morton's avatar
Andrew Morton committed
422
423
 * cachep->buffer_size - 1* BYTES_PER_WORD: last caller address
 *					[BYTES_PER_WORD long]
Linus Torvalds's avatar
Linus Torvalds committed
424
 */
425
static int obj_offset(struct kmem_cache *cachep)
Linus Torvalds's avatar
Linus Torvalds committed
426
{
427
	return cachep->obj_offset;
Linus Torvalds's avatar
Linus Torvalds committed
428
429
}

430
static int obj_size(struct kmem_cache *cachep)
Linus Torvalds's avatar
Linus Torvalds committed
431
{
432
	return cachep->obj_size;
Linus Torvalds's avatar
Linus Torvalds committed
433
434
}

435
static unsigned long long *dbg_redzone1(struct kmem_cache *cachep, void *objp)
Linus Torvalds's avatar
Linus Torvalds committed
436
437
{
	BUG_ON(!(cachep->flags & SLAB_RED_ZONE));
438
439
	return (unsigned long long*) (objp + obj_offset(cachep) -
				      sizeof(unsigned long long));
Linus Torvalds's avatar
Linus Torvalds committed
440
441
}

442
static unsigned long long *dbg_redzone2(struct kmem_cache *cachep, void *objp)
Linus Torvalds's avatar
Linus Torvalds committed
443
444
445
{
	BUG_ON(!(cachep->flags & SLAB_RED_ZONE));
	if (cachep->flags & SLAB_STORE_USER)
446
447
		return (unsigned long long *)(objp + cachep->buffer_size -
					      sizeof(unsigned long long) -
David Woodhouse's avatar
David Woodhouse committed
448
					      REDZONE_ALIGN);
449
450
	return (unsigned long long *) (objp + cachep->buffer_size -
				       sizeof(unsigned long long));
Linus Torvalds's avatar
Linus Torvalds committed
451
452
}

453
static void **dbg_userword(struct kmem_cache *cachep, void *objp)
Linus Torvalds's avatar
Linus Torvalds committed
454
455
{
	BUG_ON(!(cachep->flags & SLAB_STORE_USER));
456
	return (void **)(objp + cachep->buffer_size - BYTES_PER_WORD);
Linus Torvalds's avatar
Linus Torvalds committed
457
458
459
460
}

#else

461
462
#define obj_offset(x)			0
#define obj_size(cachep)		(cachep->buffer_size)
463
464
#define dbg_redzone1(cachep, objp)	({BUG(); (unsigned long long *)NULL;})
#define dbg_redzone2(cachep, objp)	({BUG(); (unsigned long long *)NULL;})
Linus Torvalds's avatar
Linus Torvalds committed
465
466
467
468
#define dbg_userword(cachep, objp)	({BUG(); (void **)NULL;})

#endif

469
#ifdef CONFIG_TRACING
470
471
472
473
474
475
476
size_t slab_buffer_size(struct kmem_cache *cachep)
{
	return cachep->buffer_size;
}
EXPORT_SYMBOL(slab_buffer_size);
#endif

Linus Torvalds's avatar
Linus Torvalds committed
477
478
479
480
481
482
483
/*
 * Do not go above this order unless 0 objects fit into the slab.
 */
#define	BREAK_GFP_ORDER_HI	1
#define	BREAK_GFP_ORDER_LO	0
static int slab_break_gfp_order = BREAK_GFP_ORDER_LO;

Andrew Morton's avatar
Andrew Morton committed
484
485
486
487
/*
 * Functions for storing/retrieving the cachep and or slab from the page
 * allocator.  These are used to find the slab an obj belongs to.  With kfree(),
 * these are used to find the cache which an obj belongs to.
Linus Torvalds's avatar
Linus Torvalds committed
488
 */
489
490
491
492
493
494
495
static inline void page_set_cache(struct page *page, struct kmem_cache *cache)
{
	page->lru.next = (struct list_head *)cache;
}

static inline struct kmem_cache *page_get_cache(struct page *page)
{
496
	page = compound_head(page);
497
	BUG_ON(!PageSlab(page));
498
499
500
501
502
503
504
505
506
507
	return (struct kmem_cache *)page->lru.next;
}

static inline void page_set_slab(struct page *page, struct slab *slab)
{
	page->lru.prev = (struct list_head *)slab;
}

static inline struct slab *page_get_slab(struct page *page)
{
508
	BUG_ON(!PageSlab(page));
509
510
	return (struct slab *)page->lru.prev;
}
Linus Torvalds's avatar
Linus Torvalds committed
511

512
513
static inline struct kmem_cache *virt_to_cache(const void *obj)
{
514
	struct page *page = virt_to_head_page(obj);
515
516
517
518
519
	return page_get_cache(page);
}

static inline struct slab *virt_to_slab(const void *obj)
{
520
	struct page *page = virt_to_head_page(obj);
521
522
523
	return page_get_slab(page);
}

524
525
526
527
528
529
static inline void *index_to_obj(struct kmem_cache *cache, struct slab *slab,
				 unsigned int idx)
{
	return slab->s_mem + cache->buffer_size * idx;
}

530
531
532
533
534
535
536
537
/*
 * We want to avoid an expensive divide : (offset / cache->buffer_size)
 *   Using the fact that buffer_size is a constant for a particular cache,
 *   we can replace (offset / cache->buffer_size) by
 *   reciprocal_divide(offset, cache->reciprocal_buffer_size)
 */
static inline unsigned int obj_to_index(const struct kmem_cache *cache,
					const struct slab *slab, void *obj)
538
{
539
540
	u32 offset = (obj - slab->s_mem);
	return reciprocal_divide(offset, cache->reciprocal_buffer_size);
541
542
}

Andrew Morton's avatar
Andrew Morton committed
543
544
545
/*
 * These are the default caches for kmalloc. Custom caches can have other sizes.
 */
Linus Torvalds's avatar
Linus Torvalds committed
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
struct cache_sizes malloc_sizes[] = {
#define CACHE(x) { .cs_size = (x) },
#include <linux/kmalloc_sizes.h>
	CACHE(ULONG_MAX)
#undef CACHE
};
EXPORT_SYMBOL(malloc_sizes);

/* Must match cache_sizes above. Out of line to keep cache footprint low. */
struct cache_names {
	char *name;
	char *name_dma;
};

static struct cache_names __initdata cache_names[] = {
#define CACHE(x) { .name = "size-" #x, .name_dma = "size-" #x "(DMA)" },
#include <linux/kmalloc_sizes.h>
563
	{NULL,}
Linus Torvalds's avatar
Linus Torvalds committed
564
565
566
567
#undef CACHE
};

static struct arraycache_init initarray_cache __initdata =
568
    { {0, BOOT_CPUCACHE_ENTRIES, 1, 0} };
Linus Torvalds's avatar
Linus Torvalds committed
569
static struct arraycache_init initarray_generic =
570
    { {0, BOOT_CPUCACHE_ENTRIES, 1, 0} };
Linus Torvalds's avatar
Linus Torvalds committed
571
572

/* internal cache of cache description objs */
573
static struct kmem_cache cache_cache = {
574
575
576
	.batchcount = 1,
	.limit = BOOT_CPUCACHE_ENTRIES,
	.shared = 1,
577
	.buffer_size = sizeof(struct kmem_cache),
578
	.name = "kmem_cache",
Linus Torvalds's avatar
Linus Torvalds committed
579
580
};

581
582
#define BAD_ALIEN_MAGIC 0x01020304ul

583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
/*
 * chicken and egg problem: delay the per-cpu array allocation
 * until the general caches are up.
 */
static enum {
	NONE,
	PARTIAL_AC,
	PARTIAL_L3,
	EARLY,
	FULL
} g_cpucache_up;

/*
 * used by boot code to determine if it can use slab based allocator
 */
int slab_is_available(void)
{
	return g_cpucache_up >= EARLY;
}

603
604
605
606
607
608
609
610
#ifdef CONFIG_LOCKDEP

/*
 * Slab sometimes uses the kmalloc slabs to store the slab headers
 * for other slabs "off slab".
 * The locking for this is tricky in that it nests within the locks
 * of all other slabs in a few places; to deal with this special
 * locking we put on-slab caches into a separate lock-class.
611
612
613
614
 *
 * We set lock class for alien array caches which are up during init.
 * The lock annotation will be lost if all cpus of a node goes down and
 * then comes back up during hotplug
615
 */
616
617
618
static struct lock_class_key on_slab_l3_key;
static struct lock_class_key on_slab_alc_key;

619
static void init_node_lock_keys(int q)
620
{
621
622
	struct cache_sizes *s = malloc_sizes;

623
624
625
626
627
628
629
630
631
632
	if (g_cpucache_up != FULL)
		return;

	for (s = malloc_sizes; s->cs_size != ULONG_MAX; s++) {
		struct array_cache **alc;
		struct kmem_list3 *l3;
		int r;

		l3 = s->cs_cachep->nodelists[q];
		if (!l3 || OFF_SLAB(s->cs_cachep))
633
			continue;
634
635
636
637
638
639
640
641
642
643
		lockdep_set_class(&l3->list_lock, &on_slab_l3_key);
		alc = l3->alien;
		/*
		 * FIXME: This check for BAD_ALIEN_MAGIC
		 * should go away when common slab code is taught to
		 * work even without alien caches.
		 * Currently, non NUMA code returns BAD_ALIEN_MAGIC
		 * for alloc_alien_cache,
		 */
		if (!alc || (unsigned long)alc == BAD_ALIEN_MAGIC)
644
			continue;
645
646
647
648
		for_each_node(r) {
			if (alc[r])
				lockdep_set_class(&alc[r]->lock,
					&on_slab_alc_key);
649
		}
650
651
	}
}
652
653
654
655
656
657
658
659

static inline void init_lock_keys(void)
{
	int node;

	for_each_node(node)
		init_node_lock_keys(node);
}
660
#else
661
662
663
664
static void init_node_lock_keys(int q)
{
}

665
static inline void init_lock_keys(void)
666
667
668
669
{
}
#endif

670
/*
671
 * Guard access to the cache-chain.
672
 */
Ingo Molnar's avatar
Ingo Molnar committed
673
static DEFINE_MUTEX(cache_chain_mutex);
Linus Torvalds's avatar
Linus Torvalds committed
674
675
static struct list_head cache_chain;

676
static DEFINE_PER_CPU(struct delayed_work, slab_reap_work);
Linus Torvalds's avatar
Linus Torvalds committed
677

678
static inline struct array_cache *cpu_cache_get(struct kmem_cache *cachep)
Linus Torvalds's avatar
Linus Torvalds committed
679
680
681
682
{
	return cachep->array[smp_processor_id()];
}

Andrew Morton's avatar
Andrew Morton committed
683
684
static inline struct kmem_cache *__find_general_cachep(size_t size,
							gfp_t gfpflags)
Linus Torvalds's avatar
Linus Torvalds committed
685
686
687
688
689
{
	struct cache_sizes *csizep = malloc_sizes;

#if DEBUG
	/* This happens if someone tries to call
690
691
692
	 * kmem_cache_create(), or __kmalloc(), before
	 * the generic caches are initialized.
	 */
693
	BUG_ON(malloc_sizes[INDEX_AC].cs_cachep == NULL);
Linus Torvalds's avatar
Linus Torvalds committed
694
#endif
695
696
697
	if (!size)
		return ZERO_SIZE_PTR;

Linus Torvalds's avatar
Linus Torvalds committed
698
699
700
701
	while (size > csizep->cs_size)
		csizep++;

	/*
702
	 * Really subtle: The last entry with cs->cs_size==ULONG_MAX
Linus Torvalds's avatar
Linus Torvalds committed
703
704
705
	 * has cs_{dma,}cachep==NULL. Thus no special case
	 * for large kmalloc calls required.
	 */
706
#ifdef CONFIG_ZONE_DMA
Linus Torvalds's avatar
Linus Torvalds committed
707
708
	if (unlikely(gfpflags & GFP_DMA))
		return csizep->cs_dmacachep;
709
#endif
Linus Torvalds's avatar
Linus Torvalds committed
710
711
712
	return csizep->cs_cachep;
}

713
static struct kmem_cache *kmem_find_general_cachep(size_t size, gfp_t gfpflags)
714
715
716
717
{
	return __find_general_cachep(size, gfpflags);
}

718
static size_t slab_mgmt_size(size_t nr_objs, size_t align)
Linus Torvalds's avatar
Linus Torvalds committed
719
{
720
721
	return ALIGN(sizeof(struct slab)+nr_objs*sizeof(kmem_bufctl_t), align);
}
Linus Torvalds's avatar
Linus Torvalds committed
722

Andrew Morton's avatar
Andrew Morton committed
723
724
725
/*
 * Calculate the number of objects and left-over bytes for a given buffer size.
 */
726
727
728
729
730
731
732
static void cache_estimate(unsigned long gfporder, size_t buffer_size,
			   size_t align, int flags, size_t *left_over,
			   unsigned int *num)
{
	int nr_objs;
	size_t mgmt_size;
	size_t slab_size = PAGE_SIZE << gfporder;
Linus Torvalds's avatar
Linus Torvalds committed
733

734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
	/*
	 * The slab management structure can be either off the slab or
	 * on it. For the latter case, the memory allocated for a
	 * slab is used for:
	 *
	 * - The struct slab
	 * - One kmem_bufctl_t for each object
	 * - Padding to respect alignment of @align
	 * - @buffer_size bytes for each object
	 *
	 * If the slab management structure is off the slab, then the
	 * alignment will already be calculated into the size. Because
	 * the slabs are all pages aligned, the objects will be at the
	 * correct alignment when allocated.
	 */
	if (flags & CFLGS_OFF_SLAB) {
		mgmt_size = 0;
		nr_objs = slab_size / buffer_size;

		if (nr_objs > SLAB_LIMIT)
			nr_objs = SLAB_LIMIT;
	} else {
		/*
		 * Ignore padding for the initial guess. The padding
		 * is at most @align-1 bytes, and @buffer_size is at
		 * least @align. In the worst case, this result will
		 * be one greater than the number of objects that fit
		 * into the memory allocation when taking the padding
		 * into account.
		 */
		nr_objs = (slab_size - sizeof(struct slab)) /
			  (buffer_size + sizeof(kmem_bufctl_t));

		/*
		 * This calculated number will be either the right
		 * amount, or one greater than what we want.
		 */
		if (slab_mgmt_size(nr_objs, align) + nr_objs*buffer_size
		       > slab_size)
			nr_objs--;

		if (nr_objs > SLAB_LIMIT)
			nr_objs = SLAB_LIMIT;

		mgmt_size = slab_mgmt_size(nr_objs, align);
	}
	*num = nr_objs;
	*left_over = slab_size - nr_objs*buffer_size - mgmt_size;
Linus Torvalds's avatar
Linus Torvalds committed
782
783
}

784
#define slab_error(cachep, msg) __slab_error(__func__, cachep, msg)
Linus Torvalds's avatar
Linus Torvalds committed
785

Andrew Morton's avatar
Andrew Morton committed
786
787
static void __slab_error(const char *function, struct kmem_cache *cachep,
			char *msg)
Linus Torvalds's avatar
Linus Torvalds committed
788
789
{
	printk(KERN_ERR "slab error in %s(): cache `%s': %s\n",
790
	       function, cachep->name, msg);
Linus Torvalds's avatar
Linus Torvalds committed
791
792
793
	dump_stack();
}

794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
/*
 * By default on NUMA we use alien caches to stage the freeing of
 * objects allocated from other nodes. This causes massive memory
 * inefficiencies when using fake NUMA setup to split memory into a
 * large number of small nodes, so it can be disabled on the command
 * line
  */

static int use_alien_caches __read_mostly = 1;
static int __init noaliencache_setup(char *s)
{
	use_alien_caches = 0;
	return 1;
}
__setup("noaliencache", noaliencache_setup);

810
811
812
813
814
815
816
#ifdef CONFIG_NUMA
/*
 * Special reaping functions for NUMA systems called from cache_reap().
 * These take care of doing round robin flushing of alien caches (containing
 * objects freed on different nodes from which they were allocated) and the
 * flushing of remote pcps by calling drain_node_pages.
 */
817
static DEFINE_PER_CPU(unsigned long, slab_reap_node);
818
819
820
821
822

static void init_reap_node(int cpu)
{
	int node;

823
	node = next_node(cpu_to_mem(cpu), node_online_map);
824
	if (node == MAX_NUMNODES)
825
		node = first_node(node_online_map);
826

827
	per_cpu(slab_reap_node, cpu) = node;
828
829
830
831
}

static void next_reap_node(void)
{
832
	int node = __this_cpu_read(slab_reap_node);
833
834
835
836

	node = next_node(node, node_online_map);
	if (unlikely(node >= MAX_NUMNODES))
		node = first_node(node_online_map);
837
	__this_cpu_write(slab_reap_node, node);
838
839
840
841
842
843
844
}

#else
#define init_reap_node(cpu) do { } while (0)
#define next_reap_node(void) do { } while (0)
#endif

Linus Torvalds's avatar
Linus Torvalds committed
845
846
847
848
849
850
851
/*
 * Initiate the reap timer running on the target CPU.  We run at around 1 to 2Hz
 * via the workqueue/eventd.
 * Add the CPU number into the expiration time to minimize the possibility of
 * the CPUs getting into lockstep and contending for the global cache chain
 * lock.
 */
852
static void __cpuinit start_cpu_timer(int cpu)
Linus Torvalds's avatar
Linus Torvalds committed
853
{
854
	struct delayed_work *reap_work = &per_cpu(slab_reap_work, cpu);
Linus Torvalds's avatar
Linus Torvalds committed
855
856
857
858
859
860

	/*
	 * When this gets called from do_initcalls via cpucache_init(),
	 * init_workqueues() has already run, so keventd will be setup
	 * at that time.
	 */
861
	if (keventd_up() && reap_work->work.func == NULL) {
862
		init_reap_node(cpu);
863
		INIT_DELAYED_WORK_DEFERRABLE(reap_work, cache_reap);
864
865
		schedule_delayed_work_on(cpu, reap_work,
					__round_jiffies_relative(HZ, cpu));
Linus Torvalds's avatar
Linus Torvalds committed
866
867
868
	}
}

869
static struct array_cache *alloc_arraycache(int node, int entries,
870
					    int batchcount, gfp_t gfp)
Linus Torvalds's avatar
Linus Torvalds committed
871
{
872
	int memsize = sizeof(void *) * entries + sizeof(struct array_cache);
Linus Torvalds's avatar
Linus Torvalds committed
873
874
	struct array_cache *nc = NULL;

875
	nc = kmalloc_node(memsize, gfp, node);
876
877
878
879
880
881
882
883
	/*
	 * The array_cache structures contain pointers to free object.
	 * However, when such objects are allocated or transfered to another
	 * cache the pointers are not cleared and they could be counted as
	 * valid references during a kmemleak scan. Therefore, kmemleak must
	 * not scan such objects.
	 */
	kmemleak_no_scan(nc);
Linus Torvalds's avatar
Linus Torvalds committed
884
885
886
887
888
	if (nc) {
		nc->avail = 0;
		nc->limit = entries;
		nc->batchcount = batchcount;
		nc->touched = 0;
889
		spin_lock_init(&nc->lock);
Linus Torvalds's avatar
Linus Torvalds committed
890
891
892
893
	}
	return nc;
}

894
895
896
897
898
899
900
901
902
903
/*
 * Transfer objects in one arraycache to another.
 * Locking must be handled by the caller.
 *
 * Return the number of entries transferred.
 */
static int transfer_objects(struct array_cache *to,
		struct array_cache *from, unsigned int max)
{
	/* Figure out how many entries to transfer */
904
	int nr = min3(from->avail, max, to->limit - to->avail);
905
906
907
908
909
910
911
912
913
914
915
916

	if (!nr)
		return 0;

	memcpy(to->entry + to->avail, from->entry + from->avail -nr,
			sizeof(void *) *nr);

	from->avail -= nr;
	to->avail += nr;
	return nr;
}

917
918
919
920
921
#ifndef CONFIG_NUMA

#define drain_alien_cache(cachep, alien) do { } while (0)
#define reap_alien(cachep, l3) do { } while (0)

922
static inline struct array_cache **alloc_alien_cache(int node, int limit, gfp_t gfp)
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
{
	return (struct array_cache **)BAD_ALIEN_MAGIC;
}

static inline void free_alien_cache(struct array_cache **ac_ptr)
{
}

static inline int cache_free_alien(struct kmem_cache *cachep, void *objp)
{
	return 0;
}

static inline void *alternate_node_alloc(struct kmem_cache *cachep,
		gfp_t flags)
{
	return NULL;
}

942
static inline void *____cache_alloc_node(struct kmem_cache *cachep,
943
944
945
946
947
948
949
		 gfp_t flags, int nodeid)
{
	return NULL;
}

#else	/* CONFIG_NUMA */

950
static void *____cache_alloc_node(struct kmem_cache *, gfp_t, int);
951
static void *alternate_node_alloc(struct kmem_cache *, gfp_t);
952

953
static struct array_cache **alloc_alien_cache(int node, int limit, gfp_t gfp)
954
955
{
	struct array_cache **ac_ptr;
956
	int memsize = sizeof(void *) * nr_node_ids;
957
958
959
960
	int i;

	if (limit > 1)
		limit = 12;
961
	ac_ptr = kzalloc_node(memsize, gfp, node);
962
963
	if (ac_ptr) {
		for_each_node(i) {
964
			if (i == node || !node_online(i))
965
				continue;
966
			ac_ptr[i] = alloc_arraycache(node, limit, 0xbaadf00d, gfp);
967
			if (!ac_ptr[i]) {
968
				for (i--; i >= 0; i--)
969
970
971
972
973
974
975
976
977
					kfree(ac_ptr[i]);
				kfree(ac_ptr);
				return NULL;
			}
		}
	}
	return ac_ptr;
}

Pekka Enberg's avatar
Pekka Enberg committed
978
static void free_alien_cache(struct array_cache **ac_ptr)
979
980
981
982
983
984
{
	int i;

	if (!ac_ptr)
		return;
	for_each_node(i)
985
	    kfree(ac_ptr[i]);
986
987
988
	kfree(ac_ptr);
}

989
static void __drain_alien_cache(struct kmem_cache *cachep,
Pekka Enberg's avatar
Pekka Enberg committed
990
				struct array_cache *ac, int node)
991
992
993
994
995
{
	struct kmem_list3 *rl3 = cachep->nodelists[node];

	if (ac->avail) {
		spin_lock(&rl3->list_lock);
996
997
998
999
1000
		/*
		 * Stuff objects into the remote nodes shared array first.
		 * That way we could avoid the overhead of putting the objects
		 * into the free lists and getting them back later.
		 */