slab.c 114 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
71
 *	The global cache-chain is protected by the mutex 'slab_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>
118
#include	<linux/prefetch.h>
Linus Torvalds's avatar
Linus Torvalds committed
119

120
121
#include	<net/sock.h>

Linus Torvalds's avatar
Linus Torvalds committed
122
123
124
125
#include	<asm/cacheflush.h>
#include	<asm/tlbflush.h>
#include	<asm/page.h>

126
127
#include <trace/events/kmem.h>

128
129
#include	"internal.h"

130
131
#include	"slab.h"

Linus Torvalds's avatar
Linus Torvalds committed
132
/*
133
 * DEBUG	- 1 for kmem_cache_create() to honour; SLAB_RED_ZONE & SLAB_POISON.
Linus Torvalds's avatar
Linus Torvalds committed
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
 *		  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
154
#define	REDZONE_ALIGN		max(BYTES_PER_WORD, __alignof__(unsigned long long))
Linus Torvalds's avatar
Linus Torvalds committed
155
156
157
158
159

#ifndef ARCH_KMALLOC_FLAGS
#define ARCH_KMALLOC_FLAGS SLAB_HWCACHE_ALIGN
#endif

160
161
162
163
164
165
/*
 * true if a page was allocated from pfmemalloc reserves for network-based
 * swap
 */
static bool pfmemalloc_active __read_mostly;

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
/*
 * 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.
 */

185
typedef unsigned int kmem_bufctl_t;
Linus Torvalds's avatar
Linus Torvalds committed
186
187
#define BUFCTL_END	(((kmem_bufctl_t)(~0U))-0)
#define BUFCTL_FREE	(((kmem_bufctl_t)(~0U))-1)
188
189
#define	BUFCTL_ACTIVE	(((kmem_bufctl_t)(~0U))-2)
#define	SLAB_LIMIT	(((kmem_bufctl_t)(~0U))-3)
Linus Torvalds's avatar
Linus Torvalds committed
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205

/*
 * 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.
 */
struct slab_rcu {
206
	struct rcu_head head;
207
	struct kmem_cache *cachep;
208
	struct page *page;
Linus Torvalds's avatar
Linus Torvalds committed
209
210
};

211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
/*
 * 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 {
	union {
		struct {
			struct list_head list;
			void *s_mem;		/* including colour offset */
			unsigned int inuse;	/* num of objs active in slab */
			kmem_bufctl_t free;
			unsigned short nodeid;
		};
		struct slab_rcu __slab_cover_slab_rcu;
	};
};

Linus Torvalds's avatar
Linus Torvalds committed
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
			 * Must have this definition in here for the proper
			 * alignment of array_cache. Also simplifies accessing
			 * the entries.
253
254
255
256
			 *
			 * Entries should not be directly dereferenced as
			 * entries belonging to slabs marked pfmemalloc will
			 * have the lower bits set SLAB_OBJ_PFMEMALLOC
Andrew Morton's avatar
Andrew Morton committed
257
			 */
Linus Torvalds's avatar
Linus Torvalds committed
258
259
};

260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
#define SLAB_OBJ_PFMEMALLOC	1
static inline bool is_obj_pfmemalloc(void *objp)
{
	return (unsigned long)objp & SLAB_OBJ_PFMEMALLOC;
}

static inline void set_obj_pfmemalloc(void **objp)
{
	*objp = (void *)((unsigned long)*objp | SLAB_OBJ_PFMEMALLOC);
	return;
}

static inline void clear_obj_pfmemalloc(void **objp)
{
	*objp = (void *)((unsigned long)*objp & ~SLAB_OBJ_PFMEMALLOC);
}

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

287
288
289
/*
 * Need this for bootstrapping a per node allocator.
 */
290
#define NUM_INIT_LISTS (3 * MAX_NUMNODES)
291
static struct kmem_cache_node __initdata init_kmem_cache_node[NUM_INIT_LISTS];
292
#define	CACHE_CACHE 0
293
#define	SIZE_AC MAX_NUMNODES
294
#define	SIZE_NODE (2 * MAX_NUMNODES)
295

296
static int drain_freelist(struct kmem_cache *cache,
297
			struct kmem_cache_node *n, int tofree);
298
299
static void free_block(struct kmem_cache *cachep, void **objpp, int len,
			int node);
300
static int enable_cpucache(struct kmem_cache *cachep, gfp_t gfp);
301
static void cache_reap(struct work_struct *unused);
302

303
304
static int slab_early_init = 1;

305
#define INDEX_AC kmalloc_index(sizeof(struct arraycache_init))
306
#define INDEX_NODE kmalloc_index(sizeof(struct kmem_cache_node))
Linus Torvalds's avatar
Linus Torvalds committed
307

308
static void kmem_cache_node_init(struct kmem_cache_node *parent)
309
310
311
312
313
314
{
	INIT_LIST_HEAD(&parent->slabs_full);
	INIT_LIST_HEAD(&parent->slabs_partial);
	INIT_LIST_HEAD(&parent->slabs_free);
	parent->shared = NULL;
	parent->alien = NULL;
315
	parent->colour_next = 0;
316
317
318
319
320
	spin_lock_init(&parent->list_lock);
	parent->free_objects = 0;
	parent->free_touched = 0;
}

Andrew Morton's avatar
Andrew Morton committed
321
322
323
#define MAKE_LIST(cachep, listp, slab, nodeid)				\
	do {								\
		INIT_LIST_HEAD(listp);					\
324
		list_splice(&(cachep->node[nodeid]->slab), listp);	\
325
326
	} while (0)

Andrew Morton's avatar
Andrew Morton committed
327
328
#define	MAKE_ALL_LISTS(cachep, ptr, nodeid)				\
	do {								\
329
330
331
332
	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
333
334
335
336
337

#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
338
339
340
/*
 * Optimization question: fewer reaps means less probability for unnessary
 * cpucache drain/refill cycles.
Linus Torvalds's avatar
Linus Torvalds committed
341
 *
Adrian Bunk's avatar
Adrian Bunk committed
342
 * OTOH the cpuarrays can contain lots of objects,
Linus Torvalds's avatar
Linus Torvalds committed
343
344
345
346
347
348
349
350
351
352
 * 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++)
353
#define	STATS_ADD_REAPED(x,y)	((x)->reaped += (y))
Andrew Morton's avatar
Andrew Morton committed
354
355
356
357
358
#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
359
360
#define	STATS_INC_ERR(x)	((x)->errors++)
#define	STATS_INC_NODEALLOCS(x)	((x)->node_allocs++)
361
#define	STATS_INC_NODEFREES(x)	((x)->node_frees++)
362
#define STATS_INC_ACOVERFLOW(x)   ((x)->node_overflow++)
Andrew Morton's avatar
Andrew Morton committed
363
364
365
366
367
#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
368
369
370
371
372
373
374
375
376
#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)
377
#define	STATS_ADD_REAPED(x,y)	do { (void)(y); } while (0)
Linus Torvalds's avatar
Linus Torvalds committed
378
379
380
#define	STATS_SET_HIGH(x)	do { } while (0)
#define	STATS_INC_ERR(x)	do { } while (0)
#define	STATS_INC_NODEALLOCS(x)	do { } while (0)
381
#define	STATS_INC_NODEFREES(x)	do { } while (0)
382
#define STATS_INC_ACOVERFLOW(x)   do { } while (0)
Andrew Morton's avatar
Andrew Morton committed
383
#define	STATS_SET_FREEABLE(x, i) do { } while (0)
Linus Torvalds's avatar
Linus Torvalds committed
384
385
386
387
388
389
390
391
#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
392
393
/*
 * memory layout of objects:
Linus Torvalds's avatar
Linus Torvalds committed
394
 * 0		: objp
395
 * 0 .. cachep->obj_offset - BYTES_PER_WORD - 1: padding. This ensures that
Linus Torvalds's avatar
Linus Torvalds committed
396
397
 * 		the end of an object is aligned with the end of the real
 * 		allocation. Catches writes behind the end of the allocation.
398
 * cachep->obj_offset - BYTES_PER_WORD .. cachep->obj_offset - 1:
Linus Torvalds's avatar
Linus Torvalds committed
399
 * 		redzone word.
400
 * cachep->obj_offset: The real object.
401
402
 * cachep->size - 2* BYTES_PER_WORD: redzone word [BYTES_PER_WORD long]
 * cachep->size - 1* BYTES_PER_WORD: last caller address
Andrew Morton's avatar
Andrew Morton committed
403
 *					[BYTES_PER_WORD long]
Linus Torvalds's avatar
Linus Torvalds committed
404
 */
405
static int obj_offset(struct kmem_cache *cachep)
Linus Torvalds's avatar
Linus Torvalds committed
406
{
407
	return cachep->obj_offset;
Linus Torvalds's avatar
Linus Torvalds committed
408
409
}

410
static unsigned long long *dbg_redzone1(struct kmem_cache *cachep, void *objp)
Linus Torvalds's avatar
Linus Torvalds committed
411
412
{
	BUG_ON(!(cachep->flags & SLAB_RED_ZONE));
413
414
	return (unsigned long long*) (objp + obj_offset(cachep) -
				      sizeof(unsigned long long));
Linus Torvalds's avatar
Linus Torvalds committed
415
416
}

417
static unsigned long long *dbg_redzone2(struct kmem_cache *cachep, void *objp)
Linus Torvalds's avatar
Linus Torvalds committed
418
419
420
{
	BUG_ON(!(cachep->flags & SLAB_RED_ZONE));
	if (cachep->flags & SLAB_STORE_USER)
421
		return (unsigned long long *)(objp + cachep->size -
422
					      sizeof(unsigned long long) -
David Woodhouse's avatar
David Woodhouse committed
423
					      REDZONE_ALIGN);
424
	return (unsigned long long *) (objp + cachep->size -
425
				       sizeof(unsigned long long));
Linus Torvalds's avatar
Linus Torvalds committed
426
427
}

428
static void **dbg_userword(struct kmem_cache *cachep, void *objp)
Linus Torvalds's avatar
Linus Torvalds committed
429
430
{
	BUG_ON(!(cachep->flags & SLAB_STORE_USER));
431
	return (void **)(objp + cachep->size - BYTES_PER_WORD);
Linus Torvalds's avatar
Linus Torvalds committed
432
433
434
435
}

#else

436
#define obj_offset(x)			0
437
438
#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
439
440
441
442
443
#define dbg_userword(cachep, objp)	({BUG(); (void **)NULL;})

#endif

/*
444
445
 * Do not go above this order unless 0 objects fit into the slab or
 * overridden on the command line.
Linus Torvalds's avatar
Linus Torvalds committed
446
 */
447
448
449
#define	SLAB_MAX_ORDER_HI	1
#define	SLAB_MAX_ORDER_LO	0
static int slab_max_order = SLAB_MAX_ORDER_LO;
450
static bool slab_max_order_set __initdata;
Linus Torvalds's avatar
Linus Torvalds committed
451

452
453
static inline struct kmem_cache *virt_to_cache(const void *obj)
{
454
	struct page *page = virt_to_head_page(obj);
455
	return page->slab_cache;
456
457
458
459
}

static inline struct slab *virt_to_slab(const void *obj)
{
460
	struct page *page = virt_to_head_page(obj);
461
462
463

	VM_BUG_ON(!PageSlab(page));
	return page->slab_page;
464
465
}

466
467
468
static inline void *index_to_obj(struct kmem_cache *cache, struct slab *slab,
				 unsigned int idx)
{
469
	return slab->s_mem + cache->size * idx;
470
471
}

472
/*
473
474
475
 * We want to avoid an expensive divide : (offset / cache->size)
 *   Using the fact that size is a constant for a particular cache,
 *   we can replace (offset / cache->size) by
476
477
478
479
 *   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)
480
{
481
482
	u32 offset = (obj - slab->s_mem);
	return reciprocal_divide(offset, cache->reciprocal_buffer_size);
483
484
}

Linus Torvalds's avatar
Linus Torvalds committed
485
static struct arraycache_init initarray_generic =
486
    { {0, BOOT_CPUCACHE_ENTRIES, 1, 0} };
Linus Torvalds's avatar
Linus Torvalds committed
487
488

/* internal cache of cache description objs */
489
static struct kmem_cache kmem_cache_boot = {
490
491
492
	.batchcount = 1,
	.limit = BOOT_CPUCACHE_ENTRIES,
	.shared = 1,
493
	.size = sizeof(struct kmem_cache),
494
	.name = "kmem_cache",
Linus Torvalds's avatar
Linus Torvalds committed
495
496
};

497
498
#define BAD_ALIEN_MAGIC 0x01020304ul

499
500
501
502
503
504
505
506
#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.
507
508
509
510
 *
 * 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
511
 */
512
513
514
static struct lock_class_key on_slab_l3_key;
static struct lock_class_key on_slab_alc_key;

515
516
517
518
519
520
521
522
static struct lock_class_key debugobj_l3_key;
static struct lock_class_key debugobj_alc_key;

static void slab_set_lock_classes(struct kmem_cache *cachep,
		struct lock_class_key *l3_key, struct lock_class_key *alc_key,
		int q)
{
	struct array_cache **alc;
523
	struct kmem_cache_node *n;
524
525
	int r;

526
527
	n = cachep->node[q];
	if (!n)
528
529
		return;

530
531
	lockdep_set_class(&n->list_lock, l3_key);
	alc = n->alien;
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
	/*
	 * 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)
		return;
	for_each_node(r) {
		if (alc[r])
			lockdep_set_class(&alc[r]->lock, alc_key);
	}
}

static void slab_set_debugobj_lock_classes_node(struct kmem_cache *cachep, int node)
{
	slab_set_lock_classes(cachep, &debugobj_l3_key, &debugobj_alc_key, node);
}

static void slab_set_debugobj_lock_classes(struct kmem_cache *cachep)
{
	int node;

	for_each_online_node(node)
		slab_set_debugobj_lock_classes_node(cachep, node);
}

560
static void init_node_lock_keys(int q)
561
{
562
	int i;
563

564
	if (slab_state < UP)
565
566
		return;

Christoph Lameter's avatar
Christoph Lameter committed
567
	for (i = 1; i <= KMALLOC_SHIFT_HIGH; i++) {
568
		struct kmem_cache_node *n;
569
570
571
572
		struct kmem_cache *cache = kmalloc_caches[i];

		if (!cache)
			continue;
573

574
575
		n = cache->node[q];
		if (!n || OFF_SLAB(cache))
576
			continue;
577

578
		slab_set_lock_classes(cache, &on_slab_l3_key,
579
				&on_slab_alc_key, q);
580
581
	}
}
582

583
584
static void on_slab_lock_classes_node(struct kmem_cache *cachep, int q)
{
585
	if (!cachep->node[q])
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
		return;

	slab_set_lock_classes(cachep, &on_slab_l3_key,
			&on_slab_alc_key, q);
}

static inline void on_slab_lock_classes(struct kmem_cache *cachep)
{
	int node;

	VM_BUG_ON(OFF_SLAB(cachep));
	for_each_node(node)
		on_slab_lock_classes_node(cachep, node);
}

601
602
603
604
605
606
607
static inline void init_lock_keys(void)
{
	int node;

	for_each_node(node)
		init_node_lock_keys(node);
}
608
#else
609
610
611
612
static void init_node_lock_keys(int q)
{
}

613
static inline void init_lock_keys(void)
614
615
{
}
616

617
618
619
620
621
622
623
624
static inline void on_slab_lock_classes(struct kmem_cache *cachep)
{
}

static inline void on_slab_lock_classes_node(struct kmem_cache *cachep, int node)
{
}

625
626
627
628
629
630
631
static void slab_set_debugobj_lock_classes_node(struct kmem_cache *cachep, int node)
{
}

static void slab_set_debugobj_lock_classes(struct kmem_cache *cachep)
{
}
632
633
#endif

634
static DEFINE_PER_CPU(struct delayed_work, slab_reap_work);
Linus Torvalds's avatar
Linus Torvalds committed
635

636
static inline struct array_cache *cpu_cache_get(struct kmem_cache *cachep)
Linus Torvalds's avatar
Linus Torvalds committed
637
638
639
640
{
	return cachep->array[smp_processor_id()];
}

641
static size_t slab_mgmt_size(size_t nr_objs, size_t align)
Linus Torvalds's avatar
Linus Torvalds committed
642
{
643
644
	return ALIGN(sizeof(struct slab)+nr_objs*sizeof(kmem_bufctl_t), align);
}
Linus Torvalds's avatar
Linus Torvalds committed
645

Andrew Morton's avatar
Andrew Morton committed
646
647
648
/*
 * Calculate the number of objects and left-over bytes for a given buffer size.
 */
649
650
651
652
653
654
655
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
656

657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
	/*
	 * 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
705
706
}

707
#if DEBUG
708
#define slab_error(cachep, msg) __slab_error(__func__, cachep, msg)
Linus Torvalds's avatar
Linus Torvalds committed
709

Andrew Morton's avatar
Andrew Morton committed
710
711
static void __slab_error(const char *function, struct kmem_cache *cachep,
			char *msg)
Linus Torvalds's avatar
Linus Torvalds committed
712
713
{
	printk(KERN_ERR "slab error in %s(): cache `%s': %s\n",
714
	       function, cachep->name, msg);
Linus Torvalds's avatar
Linus Torvalds committed
715
	dump_stack();
716
	add_taint(TAINT_BAD_PAGE, LOCKDEP_NOW_UNRELIABLE);
Linus Torvalds's avatar
Linus Torvalds committed
717
}
718
#endif
Linus Torvalds's avatar
Linus Torvalds committed
719

720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
/*
 * 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);

736
737
738
739
740
741
742
743
744
745
746
static int __init slab_max_order_setup(char *str)
{
	get_option(&str, &slab_max_order);
	slab_max_order = slab_max_order < 0 ? 0 :
				min(slab_max_order, MAX_ORDER - 1);
	slab_max_order_set = true;

	return 1;
}
__setup("slab_max_order=", slab_max_order_setup);

747
748
749
750
751
752
753
#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.
 */
754
static DEFINE_PER_CPU(unsigned long, slab_reap_node);
755
756
757
758
759

static void init_reap_node(int cpu)
{
	int node;

760
	node = next_node(cpu_to_mem(cpu), node_online_map);
761
	if (node == MAX_NUMNODES)
762
		node = first_node(node_online_map);
763

764
	per_cpu(slab_reap_node, cpu) = node;
765
766
767
768
}

static void next_reap_node(void)
{
769
	int node = __this_cpu_read(slab_reap_node);
770
771
772
773

	node = next_node(node, node_online_map);
	if (unlikely(node >= MAX_NUMNODES))
		node = first_node(node_online_map);
774
	__this_cpu_write(slab_reap_node, node);
775
776
777
778
779
780
781
}

#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
782
783
784
785
786
787
788
/*
 * 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.
 */
789
static void start_cpu_timer(int cpu)
Linus Torvalds's avatar
Linus Torvalds committed
790
{
791
	struct delayed_work *reap_work = &per_cpu(slab_reap_work, cpu);
Linus Torvalds's avatar
Linus Torvalds committed
792
793
794
795
796
797

	/*
	 * When this gets called from do_initcalls via cpucache_init(),
	 * init_workqueues() has already run, so keventd will be setup
	 * at that time.
	 */
798
	if (keventd_up() && reap_work->work.func == NULL) {
799
		init_reap_node(cpu);
800
		INIT_DEFERRABLE_WORK(reap_work, cache_reap);
801
802
		schedule_delayed_work_on(cpu, reap_work,
					__round_jiffies_relative(HZ, cpu));
Linus Torvalds's avatar
Linus Torvalds committed
803
804
805
	}
}

806
static struct array_cache *alloc_arraycache(int node, int entries,
807
					    int batchcount, gfp_t gfp)
Linus Torvalds's avatar
Linus Torvalds committed
808
{
809
	int memsize = sizeof(void *) * entries + sizeof(struct array_cache);
Linus Torvalds's avatar
Linus Torvalds committed
810
811
	struct array_cache *nc = NULL;

812
	nc = kmalloc_node(memsize, gfp, node);
813
814
	/*
	 * The array_cache structures contain pointers to free object.
Lucas De Marchi's avatar
Lucas De Marchi committed
815
	 * However, when such objects are allocated or transferred to another
816
817
818
819
820
	 * 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
821
822
823
824
825
	if (nc) {
		nc->avail = 0;
		nc->limit = entries;
		nc->batchcount = batchcount;
		nc->touched = 0;
826
		spin_lock_init(&nc->lock);
Linus Torvalds's avatar
Linus Torvalds committed
827
828
829
830
	}
	return nc;
}

831
832
833
834
835
836
837
838
839
840
841
static inline bool is_slab_pfmemalloc(struct slab *slabp)
{
	struct page *page = virt_to_page(slabp->s_mem);

	return PageSlabPfmemalloc(page);
}

/* Clears pfmemalloc_active if no slabs have pfmalloc set */
static void recheck_pfmemalloc_active(struct kmem_cache *cachep,
						struct array_cache *ac)
{
842
	struct kmem_cache_node *n = cachep->node[numa_mem_id()];
843
844
845
846
847
848
	struct slab *slabp;
	unsigned long flags;

	if (!pfmemalloc_active)
		return;

849
850
	spin_lock_irqsave(&n->list_lock, flags);
	list_for_each_entry(slabp, &n->slabs_full, list)
851
852
853
		if (is_slab_pfmemalloc(slabp))
			goto out;

854
	list_for_each_entry(slabp, &n->slabs_partial, list)
855
856
857
		if (is_slab_pfmemalloc(slabp))
			goto out;

858
	list_for_each_entry(slabp, &n->slabs_free, list)
859
860
861
862
863
		if (is_slab_pfmemalloc(slabp))
			goto out;

	pfmemalloc_active = false;
out:
864
	spin_unlock_irqrestore(&n->list_lock, flags);
865
866
}

867
static void *__ac_get_obj(struct kmem_cache *cachep, struct array_cache *ac,
868
869
870
871
872
873
874
						gfp_t flags, bool force_refill)
{
	int i;
	void *objp = ac->entry[--ac->avail];

	/* Ensure the caller is allowed to use objects from PFMEMALLOC slab */
	if (unlikely(is_obj_pfmemalloc(objp))) {
875
		struct kmem_cache_node *n;
876
877
878
879
880
881
882

		if (gfp_pfmemalloc_allowed(flags)) {
			clear_obj_pfmemalloc(&objp);
			return objp;
		}

		/* The caller cannot use PFMEMALLOC objects, find another one */
883
		for (i = 0; i < ac->avail; i++) {
884
885
886
887
888
889
890
891
892
893
894
895
896
			/* If a !PFMEMALLOC object is found, swap them */
			if (!is_obj_pfmemalloc(ac->entry[i])) {
				objp = ac->entry[i];
				ac->entry[i] = ac->entry[ac->avail];
				ac->entry[ac->avail] = objp;
				return objp;
			}
		}

		/*
		 * If there are empty slabs on the slabs_free list and we are
		 * being forced to refill the cache, mark this one !pfmemalloc.
		 */
897
898
		n = cachep->node[numa_mem_id()];
		if (!list_empty(&n->slabs_free) && force_refill) {
899
			struct slab *slabp = virt_to_slab(objp);
900
			ClearPageSlabPfmemalloc(virt_to_head_page(slabp->s_mem));
901
902
903
904
905
906
907
908
909
910
911
912
913
			clear_obj_pfmemalloc(&objp);
			recheck_pfmemalloc_active(cachep, ac);
			return objp;
		}

		/* No !PFMEMALLOC objects available */
		ac->avail++;
		objp = NULL;
	}

	return objp;
}

914
915
916
917
918
919
920
921
922
923
924
925
926
927
static inline void *ac_get_obj(struct kmem_cache *cachep,
			struct array_cache *ac, gfp_t flags, bool force_refill)
{
	void *objp;

	if (unlikely(sk_memalloc_socks()))
		objp = __ac_get_obj(cachep, ac, flags, force_refill);
	else
		objp = ac->entry[--ac->avail];

	return objp;
}

static void *__ac_put_obj(struct kmem_cache *cachep, struct array_cache *ac,
928
929
930
931
								void *objp)
{
	if (unlikely(pfmemalloc_active)) {
		/* Some pfmemalloc slabs exist, check if this is one */
Joonsoo Kim's avatar
Joonsoo Kim committed
932
933
		struct slab *slabp = virt_to_slab(objp);
		struct page *page = virt_to_head_page(slabp->s_mem);
934
935
936
937
		if (PageSlabPfmemalloc(page))
			set_obj_pfmemalloc(&objp);
	}

938
939
940
941
942
943
944
945
946
	return objp;
}

static inline void ac_put_obj(struct kmem_cache *cachep, struct array_cache *ac,
								void *objp)
{
	if (unlikely(sk_memalloc_socks()))
		objp = __ac_put_obj(cachep, ac, objp);

947
948
949
	ac->entry[ac->avail++] = objp;
}

950
951
952
953
954
955
956
957
958
959
/*
 * 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 */
960
	int nr = min3(from->avail, max, to->limit - to->avail);
961
962
963
964
965
966
967
968
969
970
971
972

	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;
}

973
974
975
#ifndef CONFIG_NUMA

#define drain_alien_cache(cachep, alien) do { } while (0)
976
#define reap_alien(cachep, n) do { } while (0)
977

978
static inline struct array_cache **alloc_alien_cache(int node, int limit, gfp_t gfp)
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
{
	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;
}

998
static inline void *____cache_alloc_node(struct kmem_cache *cachep,
999
1000
		 gfp_t flags, int nodeid)
{