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
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
	void *addr;
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
231
/*
 * 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;
			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;
		};
		struct slab_rcu __slab_cover_slab_rcu;
	};
};

Linus Torvalds's avatar
Linus Torvalds committed
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
/*
 * 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;
249
	spinlock_t lock;
250
	void *entry[];	/*
Andrew Morton's avatar
Andrew Morton committed
251
252
253
			 * Must have this definition in here for the proper
			 * alignment of array_cache. Also simplifies accessing
			 * the entries.
254
255
256
257
			 *
			 * 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
258
			 */
Linus Torvalds's avatar
Linus Torvalds committed
259
260
};

261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
#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
278
279
280
/*
 * 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
281
282
283
284
 */
#define BOOT_CPUCACHE_ENTRIES	1
struct arraycache_init {
	struct array_cache cache;
285
	void *entries[BOOT_CPUCACHE_ENTRIES];
Linus Torvalds's avatar
Linus Torvalds committed
286
287
288
};

/*
289
 * The slab lists for all objects.
Linus Torvalds's avatar
Linus Torvalds committed
290
291
 */
struct kmem_list3 {
292
293
294
295
296
	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;
297
	unsigned int colour_next;	/* Per-node cache coloring */
298
299
300
	spinlock_t list_lock;
	struct array_cache *shared;	/* shared per node */
	struct array_cache **alien;	/* on other nodes */
301
302
	unsigned long next_reap;	/* updated without locking */
	int free_touched;		/* updated without locking */
Linus Torvalds's avatar
Linus Torvalds committed
303
304
};

305
306
307
/*
 * Need this for bootstrapping a per node allocator.
 */
308
#define NUM_INIT_LISTS (3 * MAX_NUMNODES)
309
static struct kmem_list3 __initdata initkmem_list3[NUM_INIT_LISTS];
310
#define	CACHE_CACHE 0
311
312
#define	SIZE_AC MAX_NUMNODES
#define	SIZE_L3 (2 * MAX_NUMNODES)
313

314
315
316
317
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);
318
static int enable_cpucache(struct kmem_cache *cachep, gfp_t gfp);
319
static void cache_reap(struct work_struct *unused);
320

321
/*
Andrew Morton's avatar
Andrew Morton committed
322
323
 * 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.
324
 */
325
static __always_inline int index_of(const size_t size)
326
{
327
328
	extern void __bad_size(void);

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++;
337
#include <linux/kmalloc_sizes.h>
338
#undef CACHE
339
		__bad_size();
340
	} else
341
		__bad_size();
342
343
344
	return 0;
}

345
346
static int slab_early_init = 1;

347
348
#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
349

Pekka Enberg's avatar
Pekka Enberg committed
350
static void kmem_list3_init(struct kmem_list3 *parent)
351
352
353
354
355
356
{
	INIT_LIST_HEAD(&parent->slabs_full);
	INIT_LIST_HEAD(&parent->slabs_partial);
	INIT_LIST_HEAD(&parent->slabs_free);
	parent->shared = NULL;
	parent->alien = NULL;
357
	parent->colour_next = 0;
358
359
360
361
362
	spin_lock_init(&parent->list_lock);
	parent->free_objects = 0;
	parent->free_touched = 0;
}

Andrew Morton's avatar
Andrew Morton committed
363
364
365
366
#define MAKE_LIST(cachep, listp, slab, nodeid)				\
	do {								\
		INIT_LIST_HEAD(listp);					\
		list_splice(&(cachep->nodelists[nodeid]->slab), listp);	\
367
368
	} while (0)

Andrew Morton's avatar
Andrew Morton committed
369
370
#define	MAKE_ALL_LISTS(cachep, ptr, nodeid)				\
	do {								\
371
372
373
374
	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
375
376
377
378
379

#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
380
381
382
/*
 * Optimization question: fewer reaps means less probability for unnessary
 * cpucache drain/refill cycles.
Linus Torvalds's avatar
Linus Torvalds committed
383
 *
Adrian Bunk's avatar
Adrian Bunk committed
384
 * OTOH the cpuarrays can contain lots of objects,
Linus Torvalds's avatar
Linus Torvalds committed
385
386
387
388
389
390
391
392
393
394
 * 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++)
395
#define	STATS_ADD_REAPED(x,y)	((x)->reaped += (y))
Andrew Morton's avatar
Andrew Morton committed
396
397
398
399
400
#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
401
402
#define	STATS_INC_ERR(x)	((x)->errors++)
#define	STATS_INC_NODEALLOCS(x)	((x)->node_allocs++)
403
#define	STATS_INC_NODEFREES(x)	((x)->node_frees++)
404
#define STATS_INC_ACOVERFLOW(x)   ((x)->node_overflow++)
Andrew Morton's avatar
Andrew Morton committed
405
406
407
408
409
#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
410
411
412
413
414
415
416
417
418
#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)
419
#define	STATS_ADD_REAPED(x,y)	do { (void)(y); } while (0)
Linus Torvalds's avatar
Linus Torvalds committed
420
421
422
#define	STATS_SET_HIGH(x)	do { } while (0)
#define	STATS_INC_ERR(x)	do { } while (0)
#define	STATS_INC_NODEALLOCS(x)	do { } while (0)
423
#define	STATS_INC_NODEFREES(x)	do { } while (0)
424
#define STATS_INC_ACOVERFLOW(x)   do { } while (0)
Andrew Morton's avatar
Andrew Morton committed
425
#define	STATS_SET_FREEABLE(x, i) do { } while (0)
Linus Torvalds's avatar
Linus Torvalds committed
426
427
428
429
430
431
432
433
#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
434
435
/*
 * memory layout of objects:
Linus Torvalds's avatar
Linus Torvalds committed
436
 * 0		: objp
437
 * 0 .. cachep->obj_offset - BYTES_PER_WORD - 1: padding. This ensures that
Linus Torvalds's avatar
Linus Torvalds committed
438
439
 * 		the end of an object is aligned with the end of the real
 * 		allocation. Catches writes behind the end of the allocation.
440
 * cachep->obj_offset - BYTES_PER_WORD .. cachep->obj_offset - 1:
Linus Torvalds's avatar
Linus Torvalds committed
441
 * 		redzone word.
442
 * cachep->obj_offset: The real object.
443
444
 * 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
445
 *					[BYTES_PER_WORD long]
Linus Torvalds's avatar
Linus Torvalds committed
446
 */
447
static int obj_offset(struct kmem_cache *cachep)
Linus Torvalds's avatar
Linus Torvalds committed
448
{
449
	return cachep->obj_offset;
Linus Torvalds's avatar
Linus Torvalds committed
450
451
}

452
static unsigned long long *dbg_redzone1(struct kmem_cache *cachep, void *objp)
Linus Torvalds's avatar
Linus Torvalds committed
453
454
{
	BUG_ON(!(cachep->flags & SLAB_RED_ZONE));
455
456
	return (unsigned long long*) (objp + obj_offset(cachep) -
				      sizeof(unsigned long long));
Linus Torvalds's avatar
Linus Torvalds committed
457
458
}

459
static unsigned long long *dbg_redzone2(struct kmem_cache *cachep, void *objp)
Linus Torvalds's avatar
Linus Torvalds committed
460
461
462
{
	BUG_ON(!(cachep->flags & SLAB_RED_ZONE));
	if (cachep->flags & SLAB_STORE_USER)
463
		return (unsigned long long *)(objp + cachep->size -
464
					      sizeof(unsigned long long) -
David Woodhouse's avatar
David Woodhouse committed
465
					      REDZONE_ALIGN);
466
	return (unsigned long long *) (objp + cachep->size -
467
				       sizeof(unsigned long long));
Linus Torvalds's avatar
Linus Torvalds committed
468
469
}

470
static void **dbg_userword(struct kmem_cache *cachep, void *objp)
Linus Torvalds's avatar
Linus Torvalds committed
471
472
{
	BUG_ON(!(cachep->flags & SLAB_STORE_USER));
473
	return (void **)(objp + cachep->size - BYTES_PER_WORD);
Linus Torvalds's avatar
Linus Torvalds committed
474
475
476
477
}

#else

478
#define obj_offset(x)			0
479
480
#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
481
482
483
484
485
#define dbg_userword(cachep, objp)	({BUG(); (void **)NULL;})

#endif

/*
486
487
 * 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
488
 */
489
490
491
#define	SLAB_MAX_ORDER_HI	1
#define	SLAB_MAX_ORDER_LO	0
static int slab_max_order = SLAB_MAX_ORDER_LO;
492
static bool slab_max_order_set __initdata;
Linus Torvalds's avatar
Linus Torvalds committed
493

494
495
static inline struct kmem_cache *virt_to_cache(const void *obj)
{
496
	struct page *page = virt_to_head_page(obj);
497
	return page->slab_cache;
498
499
500
501
}

static inline struct slab *virt_to_slab(const void *obj)
{
502
	struct page *page = virt_to_head_page(obj);
503
504
505

	VM_BUG_ON(!PageSlab(page));
	return page->slab_page;
506
507
}

508
509
510
static inline void *index_to_obj(struct kmem_cache *cache, struct slab *slab,
				 unsigned int idx)
{
511
	return slab->s_mem + cache->size * idx;
512
513
}

514
/*
515
516
517
 * 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
518
519
520
521
 *   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)
522
{
523
524
	u32 offset = (obj - slab->s_mem);
	return reciprocal_divide(offset, cache->reciprocal_buffer_size);
525
526
}

Andrew Morton's avatar
Andrew Morton committed
527
528
529
/*
 * These are the default caches for kmalloc. Custom caches can have other sizes.
 */
Linus Torvalds's avatar
Linus Torvalds committed
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
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>
547
	{NULL,}
Linus Torvalds's avatar
Linus Torvalds committed
548
549
550
551
#undef CACHE
};

static struct arraycache_init initarray_generic =
552
    { {0, BOOT_CPUCACHE_ENTRIES, 1, 0} };
Linus Torvalds's avatar
Linus Torvalds committed
553
554

/* internal cache of cache description objs */
555
static struct kmem_cache kmem_cache_boot = {
556
557
558
	.batchcount = 1,
	.limit = BOOT_CPUCACHE_ENTRIES,
	.shared = 1,
559
	.size = sizeof(struct kmem_cache),
560
	.name = "kmem_cache",
Linus Torvalds's avatar
Linus Torvalds committed
561
562
};

563
564
#define BAD_ALIEN_MAGIC 0x01020304ul

565
566
567
568
569
570
571
572
#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.
573
574
575
576
 *
 * 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
577
 */
578
579
580
static struct lock_class_key on_slab_l3_key;
static struct lock_class_key on_slab_alc_key;

581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
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;
	struct kmem_list3 *l3;
	int r;

	l3 = cachep->nodelists[q];
	if (!l3)
		return;

	lockdep_set_class(&l3->list_lock, 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)
		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);
}

626
static void init_node_lock_keys(int q)
627
{
628
629
	struct cache_sizes *s = malloc_sizes;

630
	if (slab_state < UP)
631
632
633
634
635
636
637
		return;

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

		l3 = s->cs_cachep->nodelists[q];
		if (!l3 || OFF_SLAB(s->cs_cachep))
638
			continue;
639
640
641

		slab_set_lock_classes(s->cs_cachep, &on_slab_l3_key,
				&on_slab_alc_key, q);
642
643
	}
}
644

645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
static void on_slab_lock_classes_node(struct kmem_cache *cachep, int q)
{
	struct kmem_list3 *l3;
	l3 = cachep->nodelists[q];
	if (!l3)
		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);
}

665
666
667
668
669
670
671
static inline void init_lock_keys(void)
{
	int node;

	for_each_node(node)
		init_node_lock_keys(node);
}
672
#else
673
674
675
676
static void init_node_lock_keys(int q)
{
}

677
static inline void init_lock_keys(void)
678
679
{
}
680

681
682
683
684
685
686
687
688
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)
{
}

689
690
691
692
693
694
695
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)
{
}
696
697
#endif

698
static DEFINE_PER_CPU(struct delayed_work, slab_reap_work);
Linus Torvalds's avatar
Linus Torvalds committed
699

700
static inline struct array_cache *cpu_cache_get(struct kmem_cache *cachep)
Linus Torvalds's avatar
Linus Torvalds committed
701
702
703
704
{
	return cachep->array[smp_processor_id()];
}

Andrew Morton's avatar
Andrew Morton committed
705
706
static inline struct kmem_cache *__find_general_cachep(size_t size,
							gfp_t gfpflags)
Linus Torvalds's avatar
Linus Torvalds committed
707
708
709
710
711
{
	struct cache_sizes *csizep = malloc_sizes;

#if DEBUG
	/* This happens if someone tries to call
712
713
714
	 * kmem_cache_create(), or __kmalloc(), before
	 * the generic caches are initialized.
	 */
715
	BUG_ON(malloc_sizes[INDEX_AC].cs_cachep == NULL);
Linus Torvalds's avatar
Linus Torvalds committed
716
#endif
717
718
719
	if (!size)
		return ZERO_SIZE_PTR;

Linus Torvalds's avatar
Linus Torvalds committed
720
721
722
723
	while (size > csizep->cs_size)
		csizep++;

	/*
724
	 * Really subtle: The last entry with cs->cs_size==ULONG_MAX
Linus Torvalds's avatar
Linus Torvalds committed
725
726
727
	 * has cs_{dma,}cachep==NULL. Thus no special case
	 * for large kmalloc calls required.
	 */
728
#ifdef CONFIG_ZONE_DMA
Linus Torvalds's avatar
Linus Torvalds committed
729
730
	if (unlikely(gfpflags & GFP_DMA))
		return csizep->cs_dmacachep;
731
#endif
Linus Torvalds's avatar
Linus Torvalds committed
732
733
734
	return csizep->cs_cachep;
}

735
static struct kmem_cache *kmem_find_general_cachep(size_t size, gfp_t gfpflags)
736
737
738
739
{
	return __find_general_cachep(size, gfpflags);
}

740
static size_t slab_mgmt_size(size_t nr_objs, size_t align)
Linus Torvalds's avatar
Linus Torvalds committed
741
{
742
743
	return ALIGN(sizeof(struct slab)+nr_objs*sizeof(kmem_bufctl_t), align);
}
Linus Torvalds's avatar
Linus Torvalds committed
744

Andrew Morton's avatar
Andrew Morton committed
745
746
747
/*
 * Calculate the number of objects and left-over bytes for a given buffer size.
 */
748
749
750
751
752
753
754
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
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
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
	/*
	 * 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
804
805
}

806
#if DEBUG
807
#define slab_error(cachep, msg) __slab_error(__func__, cachep, msg)
Linus Torvalds's avatar
Linus Torvalds committed
808

Andrew Morton's avatar
Andrew Morton committed
809
810
static void __slab_error(const char *function, struct kmem_cache *cachep,
			char *msg)
Linus Torvalds's avatar
Linus Torvalds committed
811
812
{
	printk(KERN_ERR "slab error in %s(): cache `%s': %s\n",
813
	       function, cachep->name, msg);
Linus Torvalds's avatar
Linus Torvalds committed
814
	dump_stack();
815
	add_taint(TAINT_BAD_PAGE);
Linus Torvalds's avatar
Linus Torvalds committed
816
}
817
#endif
Linus Torvalds's avatar
Linus Torvalds committed
818

819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
/*
 * 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);

835
836
837
838
839
840
841
842
843
844
845
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);

846
847
848
849
850
851
852
#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.
 */
853
static DEFINE_PER_CPU(unsigned long, slab_reap_node);
854
855
856
857
858

static void init_reap_node(int cpu)
{
	int node;

859
	node = next_node(cpu_to_mem(cpu), node_online_map);
860
	if (node == MAX_NUMNODES)
861
		node = first_node(node_online_map);
862

863
	per_cpu(slab_reap_node, cpu) = node;
864
865
866
867
}

static void next_reap_node(void)
{
868
	int node = __this_cpu_read(slab_reap_node);
869
870
871
872

	node = next_node(node, node_online_map);
	if (unlikely(node >= MAX_NUMNODES))
		node = first_node(node_online_map);
873
	__this_cpu_write(slab_reap_node, node);
874
875
876
877
878
879
880
}

#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
881
882
883
884
885
886
887
/*
 * 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.
 */
888
static void __cpuinit start_cpu_timer(int cpu)
Linus Torvalds's avatar
Linus Torvalds committed
889
{
890
	struct delayed_work *reap_work = &per_cpu(slab_reap_work, cpu);
Linus Torvalds's avatar
Linus Torvalds committed
891
892
893
894
895
896

	/*
	 * When this gets called from do_initcalls via cpucache_init(),
	 * init_workqueues() has already run, so keventd will be setup
	 * at that time.
	 */
897
	if (keventd_up() && reap_work->work.func == NULL) {
898
		init_reap_node(cpu);
899
		INIT_DEFERRABLE_WORK(reap_work, cache_reap);
900
901
		schedule_delayed_work_on(cpu, reap_work,
					__round_jiffies_relative(HZ, cpu));
Linus Torvalds's avatar
Linus Torvalds committed
902
903
904
	}
}

905
static struct array_cache *alloc_arraycache(int node, int entries,
906
					    int batchcount, gfp_t gfp)
Linus Torvalds's avatar
Linus Torvalds committed
907
{
908
	int memsize = sizeof(void *) * entries + sizeof(struct array_cache);
Linus Torvalds's avatar
Linus Torvalds committed
909
910
	struct array_cache *nc = NULL;

911
	nc = kmalloc_node(memsize, gfp, node);
912
913
	/*
	 * The array_cache structures contain pointers to free object.
Lucas De Marchi's avatar
Lucas De Marchi committed
914
	 * However, when such objects are allocated or transferred to another
915
916
917
918
919
	 * 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
920
921
922
923
924
	if (nc) {
		nc->avail = 0;
		nc->limit = entries;
		nc->batchcount = batchcount;
		nc->touched = 0;
925
		spin_lock_init(&nc->lock);
Linus Torvalds's avatar
Linus Torvalds committed
926
927
928
929
	}
	return nc;
}

930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
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)
{
	struct kmem_list3 *l3 = cachep->nodelists[numa_mem_id()];
	struct slab *slabp;
	unsigned long flags;

	if (!pfmemalloc_active)
		return;

	spin_lock_irqsave(&l3->list_lock, flags);
	list_for_each_entry(slabp, &l3->slabs_full, list)
		if (is_slab_pfmemalloc(slabp))
			goto out;

	list_for_each_entry(slabp, &l3->slabs_partial, list)
		if (is_slab_pfmemalloc(slabp))
			goto out;

	list_for_each_entry(slabp, &l3->slabs_free, list)
		if (is_slab_pfmemalloc(slabp))
			goto out;

	pfmemalloc_active = false;
out:
	spin_unlock_irqrestore(&l3->list_lock, flags);
}

966
static void *__ac_get_obj(struct kmem_cache *cachep, struct array_cache *ac,
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
						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))) {
		struct kmem_list3 *l3;

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

		/* The caller cannot use PFMEMALLOC objects, find another one */
982
		for (i = 0; i < ac->avail; i++) {
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
			/* 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.
		 */
		l3 = cachep->nodelists[numa_mem_id()];
		if (!list_empty(&l3->slabs_free) && force_refill) {
			struct slab *slabp = virt_to_slab(objp);
999
			ClearPageSlabPfmemalloc(virt_to_head_page(slabp->s_mem));
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
			clear_obj_pfmemalloc(&objp);
			recheck_pfmemalloc_active(cachep, ac);
			return objp;
		}

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

	return objp;
}

1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
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,
1027
1028
1029
1030
								void *objp)
{
	if (unlikely(pfmemalloc_active)) {
		/* Some pfmemalloc slabs exist, check if this is one */
1031
		struct page *page = virt_to_head_page(objp);
1032
1033
1034
1035
		if (PageSlabPfmemalloc(page))
			set_obj_pfmemalloc(&objp);
	}

1036
1037
1038
1039
1040
1041
1042
1043
1044
	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);

1045
1046
1047
	ac->entry[ac->avail++] = objp;
}

1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
/*
 * 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 */
1058
	int nr = min3(from->avail, max, to->limit - to->avail);
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070

	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;
}

1071
1072
1073
1074
1075
#ifndef CONFIG_NUMA

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

1076
static inline struct array_cache **alloc_alien_cache(int node, int limit, gfp_t gfp)
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
{
	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;
}

1096
static inline void *____cache_alloc_node(struct kmem_cache *cachep,
1097
1098
1099
1100
1101
1102
1103
		 gfp_t flags, int nodeid)
{
	return NULL;
}

#else	/* CONFIG_NUMA */

1104
static void *____cache_alloc_node(struct kmem_cache *, gfp_t, int);
1105
static void *alternate_node_alloc(struct kmem_cache *, gfp_t);
1106

1107
static struct array_cache **alloc_alien_cache(int node, int limit, gfp_t gfp)
1108
1109
{
	struct array_cache **ac_ptr;
1110
	int memsize = sizeof(void *) * nr_node_ids;
1111
1112
1113
1114
	int i;

	if (limit > 1)
		limit = 12;
1115
	ac_ptr = kzalloc_node(memsize, gfp, node);
1116
1117
	if (ac_ptr) {
		for_each_node(i) {
1118
			if (i == node || !node_online(i))
1119
				continue;
1120
			ac_ptr[i] = alloc_arraycache(node, limit, 0xbaadf00d, gfp);
1121
			if (!ac_ptr[i]) {
1122
				for (i--; i >= 0; i--)
1123
1124
1125
1126
1127
1128
1129
1130
1131
					kfree(ac_ptr[i]);
				kfree(ac_ptr);
				return NULL;
			}
		}
	}
	return ac_ptr;
}

Pekka Enberg's avatar
Pekka Enberg committed
1132
static void free_alien_cache(struct array_cache **ac_ptr)
1133
1134
1135
1136
1137
1138
{
	int i;

	if (!ac_ptr)
		return;
	for_each_node(i)
1139
	    kfree(ac_ptr[i]);
1140
1141
1142
	kfree(ac_ptr);
}

1143
static void __drain_alien_cache(struct kmem_cache *cachep,
Pekka Enberg's avatar
Pekka Enberg committed
1144
				struct array_cache *ac, int node)
1145
1146
1147
1148
1149
{
	struct kmem_list3 *rl3 = cachep->nodelists[node];

	if (ac->avail) {
		spin_lock(&rl3->list_lock);
1150
1151
1152
1153
1154
		/*
		 * 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.
		 */
1155
1156
		if (rl3->shared)
			transfer_objects(rl3->shared, ac, ac->limit);
1157

1158
		free_block(cachep, ac->entry, ac->avail, node);
1159
1160
1161
1162
1163
		ac->avail = 0;
		spin_unlock(&rl3->list_lock);
	}
}

1164
1165
1166
1167
1168
/*
 * Called from cache_reap() to regularly drain alien caches round robin.
 */
static void reap_alien(struct kmem_cache *cachep, struct kmem_list3 *l3)
{
1169
	int node = __this_cpu_read(slab_reap_node);
1170
1171
1172

	if (l3->alien) {
		struct array_cache *ac = l3->alien[node];
1173
1174

		if (ac && ac->avail && spin_trylock_irq(&ac->lock)) {
1175
1176
1177
1178
1179
1180
			__drain_alien_cache(cachep, ac, node);
			spin_unlock_irq(&ac->lock);
		}
	}
}

Andrew Morton's avatar
Andrew Morton committed
1181
1182
static void drain_alien_cache(struct kmem_cache *cachep,
				struct array_cache **alien)
1183
{
1184
	int i = 0;
1185
1186
1187
1188
	struct array_cache *ac;
	unsigned long flags;

	for_each_online_node(i) {
1189
		ac = alien[i];
1190
1191
1192
1193
1194
1195
1196
		if (ac) {
			spin_lock_irqsave(&ac->lock, flags);
			__drain_alien_cache(cachep, ac, i);
			spin_unlock_irqrestore(&ac->lock, flags);
		}
	}
}
1197

1198
static inline int cache_free_alien(struct kmem_cache *cachep, void *objp)
1199
1200
1201
1202
1203
{
	struct slab *slabp = virt_to_slab(objp);
	int nodeid = slabp->nodeid;
	struct kmem_list3 *l3;
	struct array_cache *alien = NULL;
1204
1205
	int node;

1206
	node = numa_mem_id();
1207
1208
1209
1210
1211

	/*
	 * Make sure we are not freeing a object from another node to the array
	 * cache on this cpu.
	 */
1212
	if (likely(slabp->nodeid == node))
1213
1214
		return 0;

1215
	l3 = cachep->nodelists[node];
1216
1217
1218
	STATS_INC_NODEFREES(cachep);
	if (l3->alien && l3->alien[nodeid]) {
		alien = l3->alien[nodeid];
1219
		spin_lock(&alien->lock);
1220
1221
1222
1223
		if (unlikely(alien->avail == alien->limit)) {
			STATS_INC_ACOVERFLOW(cachep);
			__drain_alien_cache(cachep, alien, nodeid);
		}
1224
		ac_put_obj(cachep, alien, objp);
1225
1226
1227
1228
1229
1230
1231
1232
		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;
}
1233
1234
#endif

1235
1236
1237
1238
1239
1240
1241
/*
 * Allocates and initializes nodelists for a node on each slab cache, used for
 * either memory or cpu hotplug.  If memory is being hot-added, the kmem_list3
 * will be allocated off-node since memory is not yet online for the new node.
 * When hotplugging memory or a cpu, existing nodelists are not replaced if
 * already in use.
 *
1242
 * Must hold slab_mutex.
1243
1244
1245
1246
1247
1248
1249
 */
static int init_cache_nodelists_node(int node)
{
	struct kmem_cache *cachep;
	struct kmem_list3 *l3;
	const int memsize = sizeof(struct kmem_list3);

1250
	list_for_each_entry(cachep, &slab_caches, list) {
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
		/*
		 * Set up the size64 kmemlist for cpu before we can
		 * begin anything. Make sure some other cpu on this
		 * node has not already allocated this
		 */
		if (!cachep->nodelists[node]) {
			l3 = kmalloc_node(memsize, GFP_KERNEL, node);
			if (!l3)
				return -ENOMEM;
			kmem_list3_init(l3);
			l3->next_reap = jiffies + REAPTIMEOUT_LIST3 +
			    ((unsigned long)cachep) % REAPTIMEOUT_LIST3;

			/*
			 * The l3s don't come and go as CPUs come and
1266
			 * go.  slab_mutex is sufficient
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
			 * protection here.
			 */
			cachep->nodelists[node] = l3;
		}

		spin_lock_irq(&cachep->nodelists[node]->list_lock);
		cachep->nodelists[node]->free_limit =
			(1 + nr_cpus_node(node)) *
			cachep->batchcount + cachep->num;
		spin_unlock_irq(&cachep->nodelists[node]->list_lock);
	}
	return 0;
}

1281
1282
1283
1284
static void __cpuinit cpuup_canceled(long cpu)
{
	struct kmem_cache *cachep;
	struct kmem_list3 *l3 = NULL;
1285
	int node = cpu_to_mem(cpu);
1286
	const struct cpumask *mask = cpumask_of_node(node);
1287

1288
	list_for_each_entry(cachep, &slab_caches, list) {
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
		struct array_cache *nc;
		struct array_cache *shared;
		struct array_cache **alien;

		/* cpu is dead; no one can alloc from it. */
		nc = cachep->array[cpu];
		cachep->array[cpu] = NULL;
		l3 = cachep->nodelists[node];

		if (!l3)
			goto free_array_cache;

		spin_lock_irq(&l3->list_lock);

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

1308
		if (!cpumask_empty(mask)) {
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
			spin_unlock_irq(&l3->list_lock);
			goto free_array_cache;
		}

		shared = l3->shared;
		if (shared) {
			free_block(cachep, shared->entry,
				   shared->avail, node);
			l3->shared = NULL;
		}

		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);
		}
free_array_cache:
		kfree(nc);
	}
	/*
	 * 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.
	 */
1338
	list_for_each_entry(cachep, &slab_caches, list) {
1339
1340
1341
1342
1343
1344
1345
1346
		l3 = cachep->nodelists[node];
		if (!l3)
			continue;
		drain_freelist(cachep, l3, l3->free_objects);
	}
}

static int __cpuinit cpuup_prepare(long cpu)
Linus Torvalds's avatar
Linus Torvalds committed
1347
{
1348
	struct kmem_cache *cachep;
1349
	struct kmem_list3 *l3 = NULL;
1350
	int node = cpu_to_mem(cpu);
1351
	int err;
Linus Torvalds's avatar
Linus Torvalds committed
1352

1353
1354
1355
1356
1357
1358
	/*
	 * We need to do this right in the beginning since
	 * 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
	 */
1359
1360
1361
	err = init_cache_nodelists_node(node);
	if (err < 0)
		goto bad;
1362
1363
1364
1365
1366

	/*
	 * Now we can go ahead with allocating the shared arrays and
	 * array caches
	 */
1367
	list_for_each_entry(cachep, &slab_caches, list) {
1368