slab.c 117 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
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
/*
 * 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
 * slabs and you must pass objects with the same intializations to
 * 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>
Linus Torvalds's avatar
Linus Torvalds committed
98
99
100
101
102
103
104
#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>
105
#include	<linux/string.h>
106
#include	<linux/uaccess.h>
107
#include	<linux/nodemask.h>
108
#include	<linux/mempolicy.h>
Ingo Molnar's avatar
Ingo Molnar committed
109
#include	<linux/mutex.h>
110
#include	<linux/fault-inject.h>
Ingo Molnar's avatar
Ingo Molnar committed
111
#include	<linux/rtmutex.h>
112
#include	<linux/reciprocal_div.h>
Linus Torvalds's avatar
Linus Torvalds committed
113
114
115
116
117
118

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

/*
119
 * DEBUG	- 1 for kmem_cache_create() to honour; SLAB_RED_ZONE & SLAB_POISON.
Linus Torvalds's avatar
Linus Torvalds committed
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
 *		  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 *)

#ifndef cache_line_size
#define cache_line_size()	L1_CACHE_BYTES
#endif

#ifndef ARCH_KMALLOC_MINALIGN
/*
 * Enforce a minimum alignment for the kmalloc caches.
 * Usually, the kmalloc caches are cache_line_size() aligned, except when
 * DEBUG and FORCED_DEBUG are enabled, then they are BYTES_PER_WORD aligned.
 * Some archs want to perform DMA into kmalloc caches and need a guaranteed
 * alignment larger than BYTES_PER_WORD. ARCH_KMALLOC_MINALIGN allows that.
 * Note that this flag disables some debug features.
 */
#define ARCH_KMALLOC_MINALIGN 0
#endif

#ifndef ARCH_SLAB_MINALIGN
/*
 * Enforce a minimum alignment for all caches.
 * Intended for archs that get misalignment faults even for BYTES_PER_WORD
 * aligned buffers. Includes ARCH_KMALLOC_MINALIGN.
 * If possible: Do not enable this flag for CONFIG_DEBUG_SLAB, it disables
 * some debug features.
 */
#define ARCH_SLAB_MINALIGN 0
#endif

#ifndef ARCH_KMALLOC_FLAGS
#define ARCH_KMALLOC_FLAGS SLAB_HWCACHE_ALIGN
#endif

/* Legal flag mask for kmem_cache_create(). */
#if DEBUG
174
# define CREATE_MASK	(SLAB_RED_ZONE | \
Linus Torvalds's avatar
Linus Torvalds committed
175
			 SLAB_POISON | SLAB_HWCACHE_ALIGN | \
176
			 SLAB_CACHE_DMA | \
177
			 SLAB_STORE_USER | \
Linus Torvalds's avatar
Linus Torvalds committed
178
			 SLAB_RECLAIM_ACCOUNT | SLAB_PANIC | \
179
			 SLAB_DESTROY_BY_RCU | SLAB_MEM_SPREAD)
Linus Torvalds's avatar
Linus Torvalds committed
180
#else
181
# define CREATE_MASK	(SLAB_HWCACHE_ALIGN | \
182
			 SLAB_CACHE_DMA | \
Linus Torvalds's avatar
Linus Torvalds committed
183
			 SLAB_RECLAIM_ACCOUNT | SLAB_PANIC | \
184
			 SLAB_DESTROY_BY_RCU | SLAB_MEM_SPREAD)
Linus Torvalds's avatar
Linus Torvalds committed
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
#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.
 */

206
typedef unsigned int kmem_bufctl_t;
Linus Torvalds's avatar
Linus Torvalds committed
207
208
#define BUFCTL_END	(((kmem_bufctl_t)(~0U))-0)
#define BUFCTL_FREE	(((kmem_bufctl_t)(~0U))-1)
209
210
#define	BUFCTL_ACTIVE	(((kmem_bufctl_t)(~0U))-2)
#define	SLAB_LIMIT	(((kmem_bufctl_t)(~0U))-3)
Linus Torvalds's avatar
Linus Torvalds committed
211
212
213
214
215
216
217
218
219

/*
 * 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 {
220
221
222
223
224
225
	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
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
};

/*
 * 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 {
245
	struct rcu_head head;
246
	struct kmem_cache *cachep;
247
	void *addr;
Linus Torvalds's avatar
Linus Torvalds committed
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
};

/*
 * 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;
267
	spinlock_t lock;
Andrew Morton's avatar
Andrew Morton committed
268
269
270
271
272
273
	void *entry[0];	/*
			 * Must have this definition in here for the proper
			 * alignment of array_cache. Also simplifies accessing
			 * the entries.
			 * [0] is for gcc 2.95. It should really be [].
			 */
Linus Torvalds's avatar
Linus Torvalds committed
274
275
};

Andrew Morton's avatar
Andrew Morton committed
276
277
278
/*
 * 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
279
280
281
282
 */
#define BOOT_CPUCACHE_ENTRIES	1
struct arraycache_init {
	struct array_cache cache;
283
	void *entries[BOOT_CPUCACHE_ENTRIES];
Linus Torvalds's avatar
Linus Torvalds committed
284
285
286
};

/*
287
 * The slab lists for all objects.
Linus Torvalds's avatar
Linus Torvalds committed
288
289
 */
struct kmem_list3 {
290
291
292
293
294
	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;
295
	unsigned int colour_next;	/* Per-node cache coloring */
296
297
298
	spinlock_t list_lock;
	struct array_cache *shared;	/* shared per node */
	struct array_cache **alien;	/* on other nodes */
299
300
	unsigned long next_reap;	/* updated without locking */
	int free_touched;		/* updated without locking */
Linus Torvalds's avatar
Linus Torvalds committed
301
302
};

303
304
305
306
307
308
309
310
311
/*
 * Need this for bootstrapping a per node allocator.
 */
#define NUM_INIT_LISTS (2 * MAX_NUMNODES + 1)
struct kmem_list3 __initdata initkmem_list3[NUM_INIT_LISTS];
#define	CACHE_CACHE 0
#define	SIZE_AC 1
#define	SIZE_L3 (1 + MAX_NUMNODES)

312
313
314
315
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);
316
static int enable_cpucache(struct kmem_cache *cachep);
317
static void cache_reap(struct work_struct *unused);
318

319
/*
Andrew Morton's avatar
Andrew Morton committed
320
321
 * 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.
322
 */
323
static __always_inline int index_of(const size_t size)
324
{
325
326
	extern void __bad_size(void);

327
328
329
330
331
332
333
334
335
336
	if (__builtin_constant_p(size)) {
		int i = 0;

#define CACHE(x) \
	if (size <=x) \
		return i; \
	else \
		i++;
#include "linux/kmalloc_sizes.h"
#undef CACHE
337
		__bad_size();
338
	} else
339
		__bad_size();
340
341
342
	return 0;
}

343
344
static int slab_early_init = 1;

345
346
#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
347

Pekka Enberg's avatar
Pekka Enberg committed
348
static void kmem_list3_init(struct kmem_list3 *parent)
349
350
351
352
353
354
{
	INIT_LIST_HEAD(&parent->slabs_full);
	INIT_LIST_HEAD(&parent->slabs_partial);
	INIT_LIST_HEAD(&parent->slabs_free);
	parent->shared = NULL;
	parent->alien = NULL;
355
	parent->colour_next = 0;
356
357
358
359
360
	spin_lock_init(&parent->list_lock);
	parent->free_objects = 0;
	parent->free_touched = 0;
}

Andrew Morton's avatar
Andrew Morton committed
361
362
363
364
#define MAKE_LIST(cachep, listp, slab, nodeid)				\
	do {								\
		INIT_LIST_HEAD(listp);					\
		list_splice(&(cachep->nodelists[nodeid]->slab), listp);	\
365
366
	} while (0)

Andrew Morton's avatar
Andrew Morton committed
367
368
#define	MAKE_ALL_LISTS(cachep, ptr, nodeid)				\
	do {								\
369
370
371
372
	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
373
374

/*
375
 * struct kmem_cache
Linus Torvalds's avatar
Linus Torvalds committed
376
377
378
 *
 * manages a cache.
 */
379

380
struct kmem_cache {
Linus Torvalds's avatar
Linus Torvalds committed
381
/* 1) per-cpu data, touched during every alloc/free */
382
	struct array_cache *array[NR_CPUS];
383
/* 2) Cache tunables. Protected by cache_chain_mutex */
384
385
386
	unsigned int batchcount;
	unsigned int limit;
	unsigned int shared;
387

388
	unsigned int buffer_size;
389
	u32 reciprocal_buffer_size;
390
391
/* 3) touched by every alloc & free from the backend */

Andrew Morton's avatar
Andrew Morton committed
392
393
	unsigned int flags;		/* constant flags */
	unsigned int num;		/* # of objs per slab */
Linus Torvalds's avatar
Linus Torvalds committed
394

395
/* 4) cache_grow/shrink */
Linus Torvalds's avatar
Linus Torvalds committed
396
	/* order of pgs per slab (2^n) */
397
	unsigned int gfporder;
Linus Torvalds's avatar
Linus Torvalds committed
398
399

	/* force GFP flags, e.g. GFP_DMA */
400
	gfp_t gfpflags;
Linus Torvalds's avatar
Linus Torvalds committed
401

Andrew Morton's avatar
Andrew Morton committed
402
	size_t colour;			/* cache colouring range */
403
	unsigned int colour_off;	/* colour offset */
404
	struct kmem_cache *slabp_cache;
405
	unsigned int slab_size;
Andrew Morton's avatar
Andrew Morton committed
406
	unsigned int dflags;		/* dynamic flags */
Linus Torvalds's avatar
Linus Torvalds committed
407
408

	/* constructor func */
409
	void (*ctor) (void *, struct kmem_cache *, unsigned long);
Linus Torvalds's avatar
Linus Torvalds committed
410
411

	/* de-constructor func */
412
	void (*dtor) (void *, struct kmem_cache *, unsigned long);
Linus Torvalds's avatar
Linus Torvalds committed
413

414
/* 5) cache creation/removal */
415
416
	const char *name;
	struct list_head next;
Linus Torvalds's avatar
Linus Torvalds committed
417

418
/* 6) statistics */
Linus Torvalds's avatar
Linus Torvalds committed
419
#if STATS
420
421
422
423
424
425
426
427
428
	unsigned long num_active;
	unsigned long num_allocations;
	unsigned long high_mark;
	unsigned long grown;
	unsigned long reaped;
	unsigned long errors;
	unsigned long max_freeable;
	unsigned long node_allocs;
	unsigned long node_frees;
429
	unsigned long node_overflow;
430
431
432
433
	atomic_t allochit;
	atomic_t allocmiss;
	atomic_t freehit;
	atomic_t freemiss;
Linus Torvalds's avatar
Linus Torvalds committed
434
435
#endif
#if DEBUG
436
437
438
439
440
441
442
443
	/*
	 * If debugging is enabled, then the allocator can add additional
	 * fields and/or padding to every object. buffer_size contains the total
	 * object size including these internal fields, the following two
	 * variables contain the offset to the user object and its size.
	 */
	int obj_offset;
	int obj_size;
Linus Torvalds's avatar
Linus Torvalds committed
444
#endif
Eric Dumazet's avatar
Eric Dumazet committed
445
446
447
448
449
450
451
452
453
454
455
	/*
	 * We put nodelists[] at the end of kmem_cache, because we want to size
	 * this array to nr_node_ids slots instead of MAX_NUMNODES
	 * (see kmem_cache_init())
	 * We still use [MAX_NUMNODES] and not [1] or [0] because cache_cache
	 * is statically defined, so we reserve the max number of nodes.
	 */
	struct kmem_list3 *nodelists[MAX_NUMNODES];
	/*
	 * Do not add fields after nodelists[]
	 */
Linus Torvalds's avatar
Linus Torvalds committed
456
457
458
459
460
461
};

#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
462
463
464
/*
 * Optimization question: fewer reaps means less probability for unnessary
 * cpucache drain/refill cycles.
Linus Torvalds's avatar
Linus Torvalds committed
465
 *
Adrian Bunk's avatar
Adrian Bunk committed
466
 * OTOH the cpuarrays can contain lots of objects,
Linus Torvalds's avatar
Linus Torvalds committed
467
468
469
470
471
472
473
474
475
476
 * 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++)
477
#define	STATS_ADD_REAPED(x,y)	((x)->reaped += (y))
Andrew Morton's avatar
Andrew Morton committed
478
479
480
481
482
#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
483
484
#define	STATS_INC_ERR(x)	((x)->errors++)
#define	STATS_INC_NODEALLOCS(x)	((x)->node_allocs++)
485
#define	STATS_INC_NODEFREES(x)	((x)->node_frees++)
486
#define STATS_INC_ACOVERFLOW(x)   ((x)->node_overflow++)
Andrew Morton's avatar
Andrew Morton committed
487
488
489
490
491
#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
492
493
494
495
496
497
498
499
500
#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)
501
#define	STATS_ADD_REAPED(x,y)	do { } while (0)
Linus Torvalds's avatar
Linus Torvalds committed
502
503
504
#define	STATS_SET_HIGH(x)	do { } while (0)
#define	STATS_INC_ERR(x)	do { } while (0)
#define	STATS_INC_NODEALLOCS(x)	do { } while (0)
505
#define	STATS_INC_NODEFREES(x)	do { } while (0)
506
#define STATS_INC_ACOVERFLOW(x)   do { } while (0)
Andrew Morton's avatar
Andrew Morton committed
507
#define	STATS_SET_FREEABLE(x, i) do { } while (0)
Linus Torvalds's avatar
Linus Torvalds committed
508
509
510
511
512
513
514
515
#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
516
517
/*
 * memory layout of objects:
Linus Torvalds's avatar
Linus Torvalds committed
518
 * 0		: objp
519
 * 0 .. cachep->obj_offset - BYTES_PER_WORD - 1: padding. This ensures that
Linus Torvalds's avatar
Linus Torvalds committed
520
521
 * 		the end of an object is aligned with the end of the real
 * 		allocation. Catches writes behind the end of the allocation.
522
 * cachep->obj_offset - BYTES_PER_WORD .. cachep->obj_offset - 1:
Linus Torvalds's avatar
Linus Torvalds committed
523
 * 		redzone word.
524
525
 * 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
526
527
 * cachep->buffer_size - 1* BYTES_PER_WORD: last caller address
 *					[BYTES_PER_WORD long]
Linus Torvalds's avatar
Linus Torvalds committed
528
 */
529
static int obj_offset(struct kmem_cache *cachep)
Linus Torvalds's avatar
Linus Torvalds committed
530
{
531
	return cachep->obj_offset;
Linus Torvalds's avatar
Linus Torvalds committed
532
533
}

534
static int obj_size(struct kmem_cache *cachep)
Linus Torvalds's avatar
Linus Torvalds committed
535
{
536
	return cachep->obj_size;
Linus Torvalds's avatar
Linus Torvalds committed
537
538
}

539
static unsigned long *dbg_redzone1(struct kmem_cache *cachep, void *objp)
Linus Torvalds's avatar
Linus Torvalds committed
540
541
{
	BUG_ON(!(cachep->flags & SLAB_RED_ZONE));
542
	return (unsigned long*) (objp+obj_offset(cachep)-BYTES_PER_WORD);
Linus Torvalds's avatar
Linus Torvalds committed
543
544
}

545
static unsigned long *dbg_redzone2(struct kmem_cache *cachep, void *objp)
Linus Torvalds's avatar
Linus Torvalds committed
546
547
548
{
	BUG_ON(!(cachep->flags & SLAB_RED_ZONE));
	if (cachep->flags & SLAB_STORE_USER)
549
		return (unsigned long *)(objp + cachep->buffer_size -
550
					 2 * BYTES_PER_WORD);
551
	return (unsigned long *)(objp + cachep->buffer_size - BYTES_PER_WORD);
Linus Torvalds's avatar
Linus Torvalds committed
552
553
}

554
static void **dbg_userword(struct kmem_cache *cachep, void *objp)
Linus Torvalds's avatar
Linus Torvalds committed
555
556
{
	BUG_ON(!(cachep->flags & SLAB_STORE_USER));
557
	return (void **)(objp + cachep->buffer_size - BYTES_PER_WORD);
Linus Torvalds's avatar
Linus Torvalds committed
558
559
560
561
}

#else

562
563
#define obj_offset(x)			0
#define obj_size(cachep)		(cachep->buffer_size)
Linus Torvalds's avatar
Linus Torvalds committed
564
565
566
567
568
569
570
#define dbg_redzone1(cachep, objp)	({BUG(); (unsigned long *)NULL;})
#define dbg_redzone2(cachep, objp)	({BUG(); (unsigned long *)NULL;})
#define dbg_userword(cachep, objp)	({BUG(); (void **)NULL;})

#endif

/*
Andrew Morton's avatar
Andrew Morton committed
571
572
 * Maximum size of an obj (in 2^order pages) and absolute limit for the gfp
 * order.
Linus Torvalds's avatar
Linus Torvalds committed
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
 */
#if defined(CONFIG_LARGE_ALLOCS)
#define	MAX_OBJ_ORDER	13	/* up to 32Mb */
#define	MAX_GFP_ORDER	13	/* up to 32Mb */
#elif defined(CONFIG_MMU)
#define	MAX_OBJ_ORDER	5	/* 32 pages */
#define	MAX_GFP_ORDER	5	/* 32 pages */
#else
#define	MAX_OBJ_ORDER	8	/* up to 1Mb */
#define	MAX_GFP_ORDER	8	/* up to 1Mb */
#endif

/*
 * 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
592
593
594
595
/*
 * 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
596
 */
597
598
599
600
601
602
603
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)
{
604
	page = compound_head(page);
605
	BUG_ON(!PageSlab(page));
606
607
608
609
610
611
612
613
614
615
	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)
{
616
	BUG_ON(!PageSlab(page));
617
618
	return (struct slab *)page->lru.prev;
}
Linus Torvalds's avatar
Linus Torvalds committed
619

620
621
static inline struct kmem_cache *virt_to_cache(const void *obj)
{
622
	struct page *page = virt_to_head_page(obj);
623
624
625
626
627
	return page_get_cache(page);
}

static inline struct slab *virt_to_slab(const void *obj)
{
628
	struct page *page = virt_to_head_page(obj);
629
630
631
	return page_get_slab(page);
}

632
633
634
635
636
637
static inline void *index_to_obj(struct kmem_cache *cache, struct slab *slab,
				 unsigned int idx)
{
	return slab->s_mem + cache->buffer_size * idx;
}

638
639
640
641
642
643
644
645
/*
 * 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)
646
{
647
648
	u32 offset = (obj - slab->s_mem);
	return reciprocal_divide(offset, cache->reciprocal_buffer_size);
649
650
}

Andrew Morton's avatar
Andrew Morton committed
651
652
653
/*
 * These are the default caches for kmalloc. Custom caches can have other sizes.
 */
Linus Torvalds's avatar
Linus Torvalds committed
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
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>
671
	{NULL,}
Linus Torvalds's avatar
Linus Torvalds committed
672
673
674
675
#undef CACHE
};

static struct arraycache_init initarray_cache __initdata =
676
    { {0, BOOT_CPUCACHE_ENTRIES, 1, 0} };
Linus Torvalds's avatar
Linus Torvalds committed
677
static struct arraycache_init initarray_generic =
678
    { {0, BOOT_CPUCACHE_ENTRIES, 1, 0} };
Linus Torvalds's avatar
Linus Torvalds committed
679
680

/* internal cache of cache description objs */
681
static struct kmem_cache cache_cache = {
682
683
684
	.batchcount = 1,
	.limit = BOOT_CPUCACHE_ENTRIES,
	.shared = 1,
685
	.buffer_size = sizeof(struct kmem_cache),
686
	.name = "kmem_cache",
Linus Torvalds's avatar
Linus Torvalds committed
687
688
};

689
690
#define BAD_ALIEN_MAGIC 0x01020304ul

691
692
693
694
695
696
697
698
#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.
699
700
701
702
 *
 * 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
703
 */
704
705
706
707
static struct lock_class_key on_slab_l3_key;
static struct lock_class_key on_slab_alc_key;

static inline void init_lock_keys(void)
708
709
710

{
	int q;
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
	struct cache_sizes *s = malloc_sizes;

	while (s->cs_size != ULONG_MAX) {
		for_each_node(q) {
			struct array_cache **alc;
			int r;
			struct kmem_list3 *l3 = s->cs_cachep->nodelists[q];
			if (!l3 || OFF_SLAB(s->cs_cachep))
				continue;
			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)
				continue;
			for_each_node(r) {
				if (alc[r])
					lockdep_set_class(&alc[r]->lock,
					     &on_slab_alc_key);
			}
		}
		s++;
738
739
740
	}
}
#else
741
static inline void init_lock_keys(void)
742
743
744
745
{
}
#endif

746
747
748
749
/*
 * 1. Guard access to the cache-chain.
 * 2. Protect sanity of cpu_online_map against cpu hotplug events
 */
Ingo Molnar's avatar
Ingo Molnar committed
750
static DEFINE_MUTEX(cache_chain_mutex);
Linus Torvalds's avatar
Linus Torvalds committed
751
752
753
754
755
756
757
758
static struct list_head cache_chain;

/*
 * chicken and egg problem: delay the per-cpu array allocation
 * until the general caches are up.
 */
static enum {
	NONE,
759
760
	PARTIAL_AC,
	PARTIAL_L3,
Linus Torvalds's avatar
Linus Torvalds committed
761
762
763
	FULL
} g_cpucache_up;

764
765
766
767
768
769
770
771
/*
 * used by boot code to determine if it can use slab based allocator
 */
int slab_is_available(void)
{
	return g_cpucache_up == FULL;
}

772
static DEFINE_PER_CPU(struct delayed_work, reap_work);
Linus Torvalds's avatar
Linus Torvalds committed
773

774
static inline struct array_cache *cpu_cache_get(struct kmem_cache *cachep)
Linus Torvalds's avatar
Linus Torvalds committed
775
776
777
778
{
	return cachep->array[smp_processor_id()];
}

Andrew Morton's avatar
Andrew Morton committed
779
780
static inline struct kmem_cache *__find_general_cachep(size_t size,
							gfp_t gfpflags)
Linus Torvalds's avatar
Linus Torvalds committed
781
782
783
784
785
{
	struct cache_sizes *csizep = malloc_sizes;

#if DEBUG
	/* This happens if someone tries to call
786
787
788
	 * kmem_cache_create(), or __kmalloc(), before
	 * the generic caches are initialized.
	 */
789
	BUG_ON(malloc_sizes[INDEX_AC].cs_cachep == NULL);
Linus Torvalds's avatar
Linus Torvalds committed
790
791
792
793
794
#endif
	while (size > csizep->cs_size)
		csizep++;

	/*
795
	 * Really subtle: The last entry with cs->cs_size==ULONG_MAX
Linus Torvalds's avatar
Linus Torvalds committed
796
797
798
	 * has cs_{dma,}cachep==NULL. Thus no special case
	 * for large kmalloc calls required.
	 */
799
#ifdef CONFIG_ZONE_DMA
Linus Torvalds's avatar
Linus Torvalds committed
800
801
	if (unlikely(gfpflags & GFP_DMA))
		return csizep->cs_dmacachep;
802
#endif
Linus Torvalds's avatar
Linus Torvalds committed
803
804
805
	return csizep->cs_cachep;
}

806
static struct kmem_cache *kmem_find_general_cachep(size_t size, gfp_t gfpflags)
807
808
809
810
{
	return __find_general_cachep(size, gfpflags);
}

811
static size_t slab_mgmt_size(size_t nr_objs, size_t align)
Linus Torvalds's avatar
Linus Torvalds committed
812
{
813
814
	return ALIGN(sizeof(struct slab)+nr_objs*sizeof(kmem_bufctl_t), align);
}
Linus Torvalds's avatar
Linus Torvalds committed
815

Andrew Morton's avatar
Andrew Morton committed
816
817
818
/*
 * Calculate the number of objects and left-over bytes for a given buffer size.
 */
819
820
821
822
823
824
825
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
826

827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
	/*
	 * 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
875
876
877
878
}

#define slab_error(cachep, msg) __slab_error(__FUNCTION__, cachep, msg)

Andrew Morton's avatar
Andrew Morton committed
879
880
static void __slab_error(const char *function, struct kmem_cache *cachep,
			char *msg)
Linus Torvalds's avatar
Linus Torvalds committed
881
882
{
	printk(KERN_ERR "slab error in %s(): cache `%s': %s\n",
883
	       function, cachep->name, msg);
Linus Torvalds's avatar
Linus Torvalds committed
884
885
886
	dump_stack();
}

887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
/*
 * 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);

903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
#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.
 */
static DEFINE_PER_CPU(unsigned long, reap_node);

static void init_reap_node(int cpu)
{
	int node;

	node = next_node(cpu_to_node(cpu), node_online_map);
	if (node == MAX_NUMNODES)
918
		node = first_node(node_online_map);
919

920
	per_cpu(reap_node, cpu) = node;
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
}

static void next_reap_node(void)
{
	int node = __get_cpu_var(reap_node);

	/*
	 * Also drain per cpu pages on remote zones
	 */
	if (node != numa_node_id())
		drain_node_pages(node);

	node = next_node(node, node_online_map);
	if (unlikely(node >= MAX_NUMNODES))
		node = first_node(node_online_map);
	__get_cpu_var(reap_node) = node;
}

#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
944
945
946
947
948
949
950
951
952
/*
 * 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.
 */
static void __devinit start_cpu_timer(int cpu)
{
953
	struct delayed_work *reap_work = &per_cpu(reap_work, cpu);
Linus Torvalds's avatar
Linus Torvalds committed
954
955
956
957
958
959

	/*
	 * When this gets called from do_initcalls via cpucache_init(),
	 * init_workqueues() has already run, so keventd will be setup
	 * at that time.
	 */
960
	if (keventd_up() && reap_work->work.func == NULL) {
961
		init_reap_node(cpu);
962
		INIT_DELAYED_WORK(reap_work, cache_reap);
963
964
		schedule_delayed_work_on(cpu, reap_work,
					__round_jiffies_relative(HZ, cpu));
Linus Torvalds's avatar
Linus Torvalds committed
965
966
967
	}
}

968
static struct array_cache *alloc_arraycache(int node, int entries,
969
					    int batchcount)
Linus Torvalds's avatar
Linus Torvalds committed
970
{
971
	int memsize = sizeof(void *) * entries + sizeof(struct array_cache);
Linus Torvalds's avatar
Linus Torvalds committed
972
973
	struct array_cache *nc = NULL;

974
	nc = kmalloc_node(memsize, GFP_KERNEL, node);
Linus Torvalds's avatar
Linus Torvalds committed
975
976
977
978
979
	if (nc) {
		nc->avail = 0;
		nc->limit = entries;
		nc->batchcount = batchcount;
		nc->touched = 0;
980
		spin_lock_init(&nc->lock);
Linus Torvalds's avatar
Linus Torvalds committed
981
982
983
984
	}
	return nc;
}

985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
/*
 * 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 */
	int nr = min(min(from->avail, max), to->limit - to->avail);

	if (!nr)
		return 0;

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

	from->avail -= nr;
	to->avail += nr;
	to->touched = 1;
	return nr;
}

1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
#ifndef CONFIG_NUMA

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

static inline struct array_cache **alloc_alien_cache(int node, int limit)
{
	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;
}

1034
static inline void *____cache_alloc_node(struct kmem_cache *cachep,
1035
1036
1037
1038
1039
1040
1041
		 gfp_t flags, int nodeid)
{
	return NULL;
}

#else	/* CONFIG_NUMA */

1042
static void *____cache_alloc_node(struct kmem_cache *, gfp_t, int);
1043
static void *alternate_node_alloc(struct kmem_cache *, gfp_t);
1044

Pekka Enberg's avatar
Pekka Enberg committed
1045
static struct array_cache **alloc_alien_cache(int node, int limit)
1046
1047
{
	struct array_cache **ac_ptr;
1048
	int memsize = sizeof(void *) * nr_node_ids;
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
	int i;

	if (limit > 1)
		limit = 12;
	ac_ptr = kmalloc_node(memsize, GFP_KERNEL, node);
	if (ac_ptr) {
		for_each_node(i) {
			if (i == node || !node_online(i)) {
				ac_ptr[i] = NULL;
				continue;
			}
			ac_ptr[i] = alloc_arraycache(node, limit, 0xbaadf00d);
			if (!ac_ptr[i]) {
1062
				for (i--; i <= 0; i--)
1063
1064
1065
1066
1067
1068
1069
1070
1071
					kfree(ac_ptr[i]);
				kfree(ac_ptr);
				return NULL;
			}
		}
	}
	return ac_ptr;
}

Pekka Enberg's avatar
Pekka Enberg committed
1072
static void free_alien_cache(struct array_cache **ac_ptr)
1073
1074
1075
1076
1077
1078
{
	int i;

	if (!ac_ptr)
		return;
	for_each_node(i)
1079
	    kfree(ac_ptr[i]);
1080
1081
1082
	kfree(ac_ptr);
}

1083
static void __drain_alien_cache(struct kmem_cache *cachep,
Pekka Enberg's avatar
Pekka Enberg committed
1084
				struct array_cache *ac, int node)
1085
1086
1087
1088
1089
{
	struct kmem_list3 *rl3 = cachep->nodelists[node];

	if (ac->avail) {
		spin_lock(&rl3->list_lock);
1090
1091
1092
1093
1094
		/*
		 * 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.
		 */
1095
1096
		if (rl3->shared)
			transfer_objects(rl3->shared, ac, ac->limit);
1097

1098
		free_block(cachep, ac->entry, ac->avail, node);
1099
1100
1101
1102
1103
		ac->avail = 0;
		spin_unlock(&rl3->list_lock);
	}
}

1104
1105
1106
1107
1108
1109
1110
1111
1112
/*
 * Called from cache_reap() to regularly drain alien caches round robin.
 */
static void reap_alien(struct kmem_cache *cachep, struct kmem_list3 *l3)
{
	int node = __get_cpu_var(reap_node);

	if (l3->alien) {
		struct array_cache *ac = l3->alien[node];
1113
1114

		if (ac && ac->avail && spin_trylock_irq(&ac->lock)) {
1115
1116
1117
1118
1119
1120
			__drain_alien_cache(cachep, ac, node);
			spin_unlock_irq(&ac->lock);
		}
	}
}

Andrew Morton's avatar
Andrew Morton committed
1121
1122
static void drain_alien_cache(struct kmem_cache *cachep,
				struct array_cache **alien)
1123
{
1124
	int i = 0;
1125
1126
1127
1128
	struct array_cache *ac;
	unsigned long flags;

	for_each_online_node(i) {
1129
		ac = alien[i];
1130
1131
1132
1133
1134
1135
1136
		if (ac) {
			spin_lock_irqsave(&ac->lock, flags);
			__drain_alien_cache(cachep, ac, i);
			spin_unlock_irqrestore(&ac->lock, flags);
		}
	}
}
1137

1138
static inline int cache_free_alien(struct kmem_cache *cachep, void *objp)
1139
1140
1141
1142
1143
{
	struct slab *slabp = virt_to_slab(objp);
	int nodeid = slabp->nodeid;
	struct kmem_list3 *l3;
	struct array_cache *alien = NULL;
1144
1145
1146
	int node;

	node = numa_node_id();
1147
1148
1149
1150
1151

	/*
	 * Make sure we are not freeing a object from another node to the array
	 * cache on this cpu.
	 */
1152
	if (likely(slabp->nodeid == node))
1153
1154
		return 0;

1155
	l3 = cachep->nodelists[node];
1156
1157
1158
	STATS_INC_NODEFREES(cachep);
	if (l3->alien && l3->alien[nodeid]) {
		alien = l3->alien[nodeid];
1159
		spin_lock(&alien->lock);
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
		if (unlikely(alien->avail == alien->limit)) {
			STATS_INC_ACOVERFLOW(cachep);
			__drain_alien_cache(cachep, alien, nodeid);
		}
		alien->entry[alien->avail++] = objp;
		spin_unlock(&alien->lock);
	} else {
		spin_lock(&(cachep->nodelists[nodeid])->list_lock);
		free_block(cachep, &objp, 1, nodeid);
		spin_unlock(&(cachep->nodelists[nodeid])->list_lock);
	}
	return 1;
}
1173
1174
#endif

1175
static int __cpuinit cpuup_callback(struct notifier_block *nfb,
1176
				    unsigned long action, void *hcpu)
Linus Torvalds's avatar
Linus Torvalds committed
1177
1178
{
	long cpu = (long)hcpu;
1179
	struct kmem_cache *cachep;
1180
1181
1182
	struct kmem_list3 *l3 = NULL;
	int node = cpu_to_node(cpu);
	int memsize = sizeof(struct kmem_list3);
Linus Torvalds's avatar
Linus Torvalds committed
1183
1184
1185

	switch (action) {
	case CPU_UP_PREPARE:
Ingo Molnar's avatar
Ingo Molnar committed
1186
		mutex_lock(&cache_chain_mutex);
Andrew Morton's avatar
Andrew Morton committed
1187
1188
		/*
		 * We need to do this right in the beginning since
1189
1190
1191
1192
1193
		 * alloc_arraycache's are going to use this list.
		 * kmalloc_node allows us to add the slab to the right
		 * kmem_list3 and not this cpu's kmem_list3
		 */

Linus Torvalds's avatar
Linus Torvalds committed
1194
		list_for_each_entry(cachep, &cache_chain, next) {
Andrew Morton's avatar
Andrew Morton committed
1195
1196
			/*
			 * Set up the size64 kmemlist for cpu before we can
1197
1198
1199
1200
			 * begin anything. Make sure some other cpu on this
			 * node has not already allocated this
			 */
			if (!cachep->nodelists[node]) {
Andrew Morton's avatar
Andrew Morton committed
1201
1202
				l3 = kmalloc_node(memsize, GFP_KERNEL, node);
				if (!l3)
1203
1204
1205
					goto bad;
				kmem_list3_init(l3);
				l3->next_reap = jiffies + REAPTIMEOUT_LIST3 +
1206
				    ((unsigned long)cachep) % REAPTIMEOUT_LIST3;
1207

1208
1209
1210
1211
1212
				/*
				 * The l3s don't come and go as CPUs come and
				 * go.  cache_chain_mutex is sufficient
				 * protection here.
				 */
1213
1214
				cachep->nodelists[node] = l3;
			}
Linus Torvalds's avatar
Linus Torvalds committed
1215

1216
1217
			spin_lock_irq(&cachep->nodelists[node]->list_lock);
			cachep->nodelists[node]->free_limit =
Andrew Morton's avatar
Andrew Morton committed
1218
1219
				(1 + nr_cpus_node(node)) *
				cachep->batchcount + cachep->num;
1220
1221
1222
			spin_unlock_irq(&cachep->nodelists[node]->list_lock);
		}

Andrew Morton's avatar
Andrew Morton committed
1223
1224
1225
1226
		/*
		 * Now we can go ahead with allocating the shared arrays and
		 * array caches
		 */
1227
		list_for_each_entry(cachep, &cache_chain, next) {
1228
			struct array_cache *nc;
1229
			struct array_cache *shared = NULL;
1230
			struct array_cache **alien = NULL;
1231

1232
			nc = alloc_arraycache(node, cachep->limit,
1233
						cachep->batchcount);
Linus Torvalds's avatar
Linus Torvalds committed
1234
1235
			if (!nc)
				goto bad;
1236
1237
			if (cachep->shared) {
				shared = alloc_arraycache(node,
1238
1239
					cachep->shared * cachep->batchcount,
					0xbaadf00d);
1240
1241
1242
				if (!shared)
					goto bad;
			}
1243
1244
1245
1246
1247
			if (use_alien_caches) {
                                alien = alloc_alien_cache(node, cachep->limit);
                                if (!alien)
                                        goto bad;
                        }
Linus Torvalds's avatar
Linus Torvalds committed
1248
			cachep->array[cpu] = nc;
1249
1250
1251
			l3 = cachep->nodelists[node];
			BUG_ON(!l3);

1252
1253
1254
1255
1256
1257
1258
1259
			spin_lock_irq(&l3->list_lock);
			if (!l3->shared) {
				/*
				 * We are serialised from CPU_DEAD or
				 * CPU_UP_CANCELLED by the cpucontrol lock
				 */
				l3->shared = shared;
				shared = NULL;
1260
			}
1261
1262
1263
1264
1265
1266
1267
1268
1269
#ifdef CONFIG_NUMA
			if (!l3->alien) {
				l3->alien = alien;
				alien = NULL;
			}
#endif
			spin_unlock_irq(&l3->list_lock);
			kfree(shared);
			free_alien_cache(alien);
Linus Torvalds's avatar
Linus Torvalds committed
1270
1271
1272
		}
		break;
	case CPU_ONLINE:
1273
		mutex_unlock(&cache_chain_mutex);
Linus Torvalds's avatar
Linus Torvalds committed
1274
1275
1276
		start_cpu_timer(cpu);
		break;
#ifdef CONFIG_HOTPLUG_CPU
1277
1278
1279
1280
1281
1282
	case CPU_DOWN_PREPARE:
		mutex_lock(&cache_chain_mutex);
		break;
	case CPU_DOWN_FAILED:
		mutex_unlock(&cache_chain_mutex);
		break;
Linus Torvalds's avatar
Linus Torvalds committed
1283
	case CPU_DEAD:
1284
1285
1286
1287
1288
1289
1290
1291
		/*
		 * Even if all the cpus of a node are down, we don't free the
		 * kmem_list3 of any cache. This to avoid a race between
		 * cpu_down, and a kmalloc allocation from another cpu for
		 * memory from the node of the cpu going down.  The list3
		 * structure is usually allocated from kmem_cache_create() and
		 * gets destroyed at kmem_cache_destroy().
		 */
Linus Torvalds's avatar
Linus Torvalds committed
1292
		/* fall thru */
1293
#endif
Linus Torvalds's avatar
Linus Torvalds committed
1294
1295
1296
	case CPU_UP_CANCELED:
		list_for_each_entry(cachep, &cache_chain, next) {
			struct array_cache *nc;
1297
1298
			struct array_cache *shared;
			struct array_cache **alien;
1299
			cpumask_t mask;
Linus Torvalds's avatar
Linus Torvalds committed
1300

1301
			mask = node_to_cpumask(node);
Linus Torvalds's avatar
Linus Torvalds committed
1302
1303
1304
			/* cpu is dead; no one can alloc from it. */
			nc = cachep->array[cpu];
			cachep->array[cpu] = NULL;
1305
1306
1307
			l3 = cachep->nodelists[node];

			if (!l3)
1308
				goto free_array_cache;
1309

1310
			spin_lock_irq(&l3->list_lock);
1311
1312
1313
1314

			/* Free limit for this kmem_list3 */
			l3->free_limit -= cachep->batchcount;
			if (nc)
1315
				free_block(cachep, nc->entry, nc->avail, node);
1316
1317

			if (!cpus_empty(mask)) {
1318
				spin_unlock_irq(&l3->list_lock);
1319
				goto free_array_cache;
1320
			}
1321

1322
1323
			shared = l3->shared;
			if (shared) {
1324
1325
				free_block(cachep, shared->entry,
					   shared->avail, node);
1326
1327
1328
				l3->shared = NULL;
			}

1329
1330
1331
1332
1333
1334
1335
1336
1337
			alien = l3->alien;
			l3->alien = NULL;

			spin_unlock_irq(&l3->list_lock);

			kfree(shared);
			if (alien) {
				drain_alien_cache(cachep, alien);
				free_alien_cache(alien);
1338
			}
1339
free_array_cache:
Linus Torvalds's avatar
Linus Torvalds committed
1340
1341
			kfree(nc);
		}
1342
1343
1344
1345
1346
1347
1348
1349
1350
		/*
		 * In the previous loop, all the objects were freed to
		 * the respective cache's slabs,  now we can go ahead and
		 * shrink each nodelist to its limit.
		 */
		list_for_each_entry(cachep, &cache_chain, next) {
			l3 = cachep->nodelists[node];
			if (!l3)
				continue;