slub.c 105 KB
Newer Older
Christoph Lameter's avatar
Christoph Lameter committed
1
2
3
4
5
6
7
/*
 * SLUB: A slab allocator that limits cache line use instead of queuing
 * objects in per cpu and per node lists.
 *
 * The allocator synchronizes using per slab locks and only
 * uses a centralized lock to manage a pool of partial slabs.
 *
Christoph Lameter's avatar
Christoph Lameter committed
8
 * (C) 2007 SGI, Christoph Lameter
Christoph Lameter's avatar
Christoph Lameter committed
9
10
11
12
13
14
15
16
17
18
19
20
21
 */

#include <linux/mm.h>
#include <linux/module.h>
#include <linux/bit_spinlock.h>
#include <linux/interrupt.h>
#include <linux/bitops.h>
#include <linux/slab.h>
#include <linux/seq_file.h>
#include <linux/cpu.h>
#include <linux/cpuset.h>
#include <linux/mempolicy.h>
#include <linux/ctype.h>
22
#include <linux/debugobjects.h>
Christoph Lameter's avatar
Christoph Lameter committed
23
#include <linux/kallsyms.h>
24
#include <linux/memory.h>
Roman Zippel's avatar
Roman Zippel committed
25
#include <linux/math64.h>
Christoph Lameter's avatar
Christoph Lameter committed
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
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71

/*
 * Lock order:
 *   1. slab_lock(page)
 *   2. slab->list_lock
 *
 *   The slab_lock protects operations on the object of a particular
 *   slab and its metadata in the page struct. If the slab lock
 *   has been taken then no allocations nor frees can be performed
 *   on the objects in the slab nor can the slab be added or removed
 *   from the partial or full lists since this would mean modifying
 *   the page_struct of the slab.
 *
 *   The list_lock protects the partial and full list on each node and
 *   the partial slab counter. If taken then no new slabs may be added or
 *   removed from the lists nor make the number of partial slabs be modified.
 *   (Note that the total number of slabs is an atomic value that may be
 *   modified without taking the list lock).
 *
 *   The list_lock is a centralized lock and thus we avoid taking it as
 *   much as possible. As long as SLUB does not have to handle partial
 *   slabs, operations can continue without any centralized lock. F.e.
 *   allocating a long series of objects that fill up slabs does not require
 *   the list lock.
 *
 *   The lock order is sometimes inverted when we are trying to get a slab
 *   off a list. We take the list_lock and then look for a page on the list
 *   to use. While we do that objects in the slabs may be freed. We can
 *   only operate on the slab if we have also taken the slab_lock. So we use
 *   a slab_trylock() on the slab. If trylock was successful then no frees
 *   can occur anymore and we can use the slab for allocations etc. If the
 *   slab_trylock() does not succeed then frees are in progress in the slab and
 *   we must stay away from it for a while since we may cause a bouncing
 *   cacheline if we try to acquire the lock. So go onto the next slab.
 *   If all pages are busy then we may allocate a new slab instead of reusing
 *   a partial slab. A new slab has noone operating on it and thus there is
 *   no danger of cacheline contention.
 *
 *   Interrupts are disabled during allocation and deallocation in order to
 *   make the slab allocator safe to use in the context of an irq. In addition
 *   interrupts are disabled to ensure that the processor does not change
 *   while handling per_cpu slabs, due to kernel preemption.
 *
 * SLUB assigns one slab for allocation to each processor.
 * Allocations only occur from these slabs called cpu slabs.
 *
Christoph Lameter's avatar
Christoph Lameter committed
72
73
 * Slabs with free elements are kept on a partial list and during regular
 * operations no list for full slabs is used. If an object in a full slab is
Christoph Lameter's avatar
Christoph Lameter committed
74
 * freed then the slab will show up again on the partial lists.
Christoph Lameter's avatar
Christoph Lameter committed
75
76
 * We track full slabs for debugging purposes though because otherwise we
 * cannot scan all objects.
Christoph Lameter's avatar
Christoph Lameter committed
77
78
79
80
81
82
83
 *
 * Slabs are freed when they become empty. Teardown and setup is
 * minimal so we rely on the page allocators per cpu caches for
 * fast frees and allocs.
 *
 * Overloading of page flags that are otherwise used for LRU management.
 *
84
85
86
87
88
89
90
91
92
93
94
95
 * PageActive 		The slab is frozen and exempt from list processing.
 * 			This means that the slab is dedicated to a purpose
 * 			such as satisfying allocations for a specific
 * 			processor. Objects may be freed in the slab while
 * 			it is frozen but slab_free will then skip the usual
 * 			list operations. It is up to the processor holding
 * 			the slab to integrate the slab into the slab lists
 * 			when the slab is no longer needed.
 *
 * 			One use of this flag is to mark slabs that are
 * 			used for allocations. Then such a slab becomes a cpu
 * 			slab. The cpu slab may be equipped with an additional
96
 * 			freelist that allows lockless access to
97
98
 * 			free objects in addition to the regular freelist
 * 			that requires the slab lock.
Christoph Lameter's avatar
Christoph Lameter committed
99
100
101
 *
 * PageError		Slab requires special handling due to debug
 * 			options set. This moves	slab handling out of
102
 * 			the fast path and disables lockless freelists.
Christoph Lameter's avatar
Christoph Lameter committed
103
104
 */

105
106
107
108
109
110
111
112
#define FROZEN (1 << PG_active)

#ifdef CONFIG_SLUB_DEBUG
#define SLABDEBUG (1 << PG_error)
#else
#define SLABDEBUG 0
#endif

113
114
static inline int SlabFrozen(struct page *page)
{
115
	return page->flags & FROZEN;
116
117
118
119
}

static inline void SetSlabFrozen(struct page *page)
{
120
	page->flags |= FROZEN;
121
122
123
124
}

static inline void ClearSlabFrozen(struct page *page)
{
125
	page->flags &= ~FROZEN;
126
127
}

128
129
static inline int SlabDebug(struct page *page)
{
130
	return page->flags & SLABDEBUG;
131
132
133
134
}

static inline void SetSlabDebug(struct page *page)
{
135
	page->flags |= SLABDEBUG;
136
137
138
139
}

static inline void ClearSlabDebug(struct page *page)
{
140
	page->flags &= ~SLABDEBUG;
141
142
}

Christoph Lameter's avatar
Christoph Lameter committed
143
144
145
146
147
148
149
150
151
152
153
/*
 * Issues still to be resolved:
 *
 * - Support PAGE_ALLOC_DEBUG. Should be easy to do.
 *
 * - Variable sizing of the per node arrays
 */

/* Enable to test recovery from slab corruption on boot */
#undef SLUB_RESILIENCY_TEST

154
155
156
157
/*
 * Mininum number of partial slabs. These will be left on the partial
 * lists even if they are empty. kmem_cache_shrink may reclaim them.
 */
158
#define MIN_PARTIAL 5
Christoph Lameter's avatar
Christoph Lameter committed
159

160
161
162
163
164
165
166
/*
 * Maximum number of desirable partial slabs.
 * The existence of more partial slabs makes kmem_cache_shrink
 * sort the partial list by the number of objects in the.
 */
#define MAX_PARTIAL 10

Christoph Lameter's avatar
Christoph Lameter committed
167
168
#define DEBUG_DEFAULT_FLAGS (SLAB_DEBUG_FREE | SLAB_RED_ZONE | \
				SLAB_POISON | SLAB_STORE_USER)
Christoph Lameter's avatar
Christoph Lameter committed
169

Christoph Lameter's avatar
Christoph Lameter committed
170
171
172
173
174
175
176
177
178
179
/*
 * Set of flags that will prevent slab merging
 */
#define SLUB_NEVER_MERGE (SLAB_RED_ZONE | SLAB_POISON | SLAB_STORE_USER | \
		SLAB_TRACE | SLAB_DESTROY_BY_RCU)

#define SLUB_MERGE_SAME (SLAB_DEBUG_FREE | SLAB_RECLAIM_ACCOUNT | \
		SLAB_CACHE_DMA)

#ifndef ARCH_KMALLOC_MINALIGN
180
#define ARCH_KMALLOC_MINALIGN __alignof__(unsigned long long)
Christoph Lameter's avatar
Christoph Lameter committed
181
182
183
#endif

#ifndef ARCH_SLAB_MINALIGN
184
#define ARCH_SLAB_MINALIGN __alignof__(unsigned long long)
Christoph Lameter's avatar
Christoph Lameter committed
185
186
187
#endif

/* Internal SLUB flags */
188
189
#define __OBJECT_POISON		0x80000000 /* Poison object */
#define __SYSFS_ADD_DEFERRED	0x40000000 /* Not yet visible via sysfs */
Christoph Lameter's avatar
Christoph Lameter committed
190
191
192
193
194
195
196
197
198
199

static int kmem_size = sizeof(struct kmem_cache);

#ifdef CONFIG_SMP
static struct notifier_block slab_notifier;
#endif

static enum {
	DOWN,		/* No slab functionality available */
	PARTIAL,	/* kmem_cache_open() works but kmalloc does not */
Christoph Lameter's avatar
Christoph Lameter committed
200
	UP,		/* Everything works but does not show up in sysfs */
Christoph Lameter's avatar
Christoph Lameter committed
201
202
203
204
205
	SYSFS		/* Sysfs up */
} slab_state = DOWN;

/* A list of all slab caches on the system */
static DECLARE_RWSEM(slub_lock);
Adrian Bunk's avatar
Adrian Bunk committed
206
static LIST_HEAD(slab_caches);
Christoph Lameter's avatar
Christoph Lameter committed
207

208
209
210
211
212
213
214
215
216
217
218
219
/*
 * Tracking user of a slab.
 */
struct track {
	void *addr;		/* Called from address */
	int cpu;		/* Was running on cpu */
	int pid;		/* Pid context */
	unsigned long when;	/* When did the operation occur */
};

enum track_item { TRACK_ALLOC, TRACK_FREE };

220
#ifdef CONFIG_SLUB_DEBUG
Christoph Lameter's avatar
Christoph Lameter committed
221
222
223
static int sysfs_slab_add(struct kmem_cache *);
static int sysfs_slab_alias(struct kmem_cache *, const char *);
static void sysfs_slab_remove(struct kmem_cache *);
224

Christoph Lameter's avatar
Christoph Lameter committed
225
#else
226
227
228
static inline int sysfs_slab_add(struct kmem_cache *s) { return 0; }
static inline int sysfs_slab_alias(struct kmem_cache *s, const char *p)
							{ return 0; }
229
230
231
232
static inline void sysfs_slab_remove(struct kmem_cache *s)
{
	kfree(s);
}
233

Christoph Lameter's avatar
Christoph Lameter committed
234
235
#endif

236
237
238
239
240
241
242
static inline void stat(struct kmem_cache_cpu *c, enum stat_item si)
{
#ifdef CONFIG_SLUB_STATS
	c->stat[si]++;
#endif
}

Christoph Lameter's avatar
Christoph Lameter committed
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
/********************************************************************
 * 			Core slab cache functions
 *******************************************************************/

int slab_is_available(void)
{
	return slab_state >= UP;
}

static inline struct kmem_cache_node *get_node(struct kmem_cache *s, int node)
{
#ifdef CONFIG_NUMA
	return s->node[node];
#else
	return &s->local_node;
#endif
}

261
262
static inline struct kmem_cache_cpu *get_cpu_slab(struct kmem_cache *s, int cpu)
{
263
264
265
266
267
#ifdef CONFIG_SMP
	return s->cpu_slab[cpu];
#else
	return &s->cpu_slab;
#endif
268
269
}

Christoph Lameter's avatar
Christoph Lameter committed
270
/* Verify that a pointer has an address that is valid within a slab page */
271
272
273
274
275
static inline int check_valid_pointer(struct kmem_cache *s,
				struct page *page, const void *object)
{
	void *base;

276
	if (!object)
277
278
		return 1;

279
	base = page_address(page);
280
	if (object < base || object >= base + page->objects * s->size ||
281
282
283
284
285
286
287
		(object - base) % s->size) {
		return 0;
	}

	return 1;
}

288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
/*
 * Slow version of get and set free pointer.
 *
 * This version requires touching the cache lines of kmem_cache which
 * we avoid to do in the fast alloc free paths. There we obtain the offset
 * from the page struct.
 */
static inline void *get_freepointer(struct kmem_cache *s, void *object)
{
	return *(void **)(object + s->offset);
}

static inline void set_freepointer(struct kmem_cache *s, void *object, void *fp)
{
	*(void **)(object + s->offset) = fp;
}

/* Loop over all objects in a slab */
306
307
#define for_each_object(__p, __s, __addr, __objects) \
	for (__p = (__addr); __p < (__addr) + (__objects) * (__s)->size;\
308
309
310
311
			__p += (__s)->size)

/* Scan freelist */
#define for_each_free_object(__p, __s, __free) \
312
	for (__p = (__free); __p; __p = get_freepointer((__s), __p))
313
314
315
316
317
318
319

/* Determine object index from a given position */
static inline int slab_index(void *p, struct kmem_cache *s, void *addr)
{
	return (p - addr) / s->size;
}

320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
static inline struct kmem_cache_order_objects oo_make(int order,
						unsigned long size)
{
	struct kmem_cache_order_objects x = {
		(order << 16) + (PAGE_SIZE << order) / size
	};

	return x;
}

static inline int oo_order(struct kmem_cache_order_objects x)
{
	return x.x >> 16;
}

static inline int oo_objects(struct kmem_cache_order_objects x)
{
	return x.x & ((1 << 16) - 1);
}

340
341
342
343
#ifdef CONFIG_SLUB_DEBUG
/*
 * Debug settings:
 */
344
345
346
#ifdef CONFIG_SLUB_DEBUG_ON
static int slub_debug = DEBUG_DEFAULT_FLAGS;
#else
347
static int slub_debug;
348
#endif
349
350
351

static char *slub_debug_slabs;

Christoph Lameter's avatar
Christoph Lameter committed
352
353
354
355
356
357
358
359
360
361
362
363
364
/*
 * Object debugging
 */
static void print_section(char *text, u8 *addr, unsigned int length)
{
	int i, offset;
	int newline = 1;
	char ascii[17];

	ascii[16] = 0;

	for (i = 0; i < length; i++) {
		if (newline) {
365
			printk(KERN_ERR "%8s 0x%p: ", text, addr + i);
Christoph Lameter's avatar
Christoph Lameter committed
366
367
			newline = 0;
		}
368
		printk(KERN_CONT " %02x", addr[i]);
Christoph Lameter's avatar
Christoph Lameter committed
369
370
371
		offset = i % 16;
		ascii[offset] = isgraph(addr[i]) ? addr[i] : '.';
		if (offset == 15) {
372
			printk(KERN_CONT " %s\n", ascii);
Christoph Lameter's avatar
Christoph Lameter committed
373
374
375
376
377
378
			newline = 1;
		}
	}
	if (!newline) {
		i %= 16;
		while (i < 16) {
379
			printk(KERN_CONT "   ");
Christoph Lameter's avatar
Christoph Lameter committed
380
381
382
			ascii[i] = ' ';
			i++;
		}
383
		printk(KERN_CONT " %s\n", ascii);
Christoph Lameter's avatar
Christoph Lameter committed
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
	}
}

static struct track *get_track(struct kmem_cache *s, void *object,
	enum track_item alloc)
{
	struct track *p;

	if (s->offset)
		p = object + s->offset + sizeof(void *);
	else
		p = object + s->inuse;

	return p + alloc;
}

static void set_track(struct kmem_cache *s, void *object,
				enum track_item alloc, void *addr)
{
	struct track *p;

	if (s->offset)
		p = object + s->offset + sizeof(void *);
	else
		p = object + s->inuse;

	p += alloc;
	if (addr) {
		p->addr = addr;
		p->cpu = smp_processor_id();
		p->pid = current ? current->pid : -1;
		p->when = jiffies;
	} else
		memset(p, 0, sizeof(struct track));
}

static void init_tracking(struct kmem_cache *s, void *object)
{
422
423
424
425
426
	if (!(s->flags & SLAB_STORE_USER))
		return;

	set_track(s, object, TRACK_FREE, NULL);
	set_track(s, object, TRACK_ALLOC, NULL);
Christoph Lameter's avatar
Christoph Lameter committed
427
428
429
430
431
432
433
}

static void print_track(const char *s, struct track *t)
{
	if (!t->addr)
		return;

434
435
	printk(KERN_ERR "INFO: %s in %pS age=%lu cpu=%u pid=%d\n",
		s, t->addr, jiffies - t->when, t->cpu, t->pid);
436
437
438
439
440
441
442
443
444
445
446
447
448
}

static void print_tracking(struct kmem_cache *s, void *object)
{
	if (!(s->flags & SLAB_STORE_USER))
		return;

	print_track("Allocated", get_track(s, object, TRACK_ALLOC));
	print_track("Freed", get_track(s, object, TRACK_FREE));
}

static void print_page_info(struct page *page)
{
449
450
	printk(KERN_ERR "INFO: Slab 0x%p objects=%u used=%u fp=0x%p flags=0x%04lx\n",
		page, page->objects, page->inuse, page->freelist, page->flags);
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466

}

static void slab_bug(struct kmem_cache *s, char *fmt, ...)
{
	va_list args;
	char buf[100];

	va_start(args, fmt);
	vsnprintf(buf, sizeof(buf), fmt, args);
	va_end(args);
	printk(KERN_ERR "========================================"
			"=====================================\n");
	printk(KERN_ERR "BUG %s: %s\n", s->name, buf);
	printk(KERN_ERR "----------------------------------------"
			"-------------------------------------\n\n");
Christoph Lameter's avatar
Christoph Lameter committed
467
468
}

469
470
471
472
473
474
475
476
477
478
479
480
static void slab_fix(struct kmem_cache *s, char *fmt, ...)
{
	va_list args;
	char buf[100];

	va_start(args, fmt);
	vsnprintf(buf, sizeof(buf), fmt, args);
	va_end(args);
	printk(KERN_ERR "FIX %s: %s\n", s->name, buf);
}

static void print_trailer(struct kmem_cache *s, struct page *page, u8 *p)
Christoph Lameter's avatar
Christoph Lameter committed
481
482
{
	unsigned int off;	/* Offset of last byte */
483
	u8 *addr = page_address(page);
484
485
486
487
488
489
490
491
492
493
494
495

	print_tracking(s, p);

	print_page_info(page);

	printk(KERN_ERR "INFO: Object 0x%p @offset=%tu fp=0x%p\n\n",
			p, p - addr, get_freepointer(s, p));

	if (p > addr + 16)
		print_section("Bytes b4", p - 16, 16);

	print_section("Object", p, min(s->objsize, 128));
Christoph Lameter's avatar
Christoph Lameter committed
496
497
498
499
500
501
502
503
504
505

	if (s->flags & SLAB_RED_ZONE)
		print_section("Redzone", p + s->objsize,
			s->inuse - s->objsize);

	if (s->offset)
		off = s->offset + sizeof(void *);
	else
		off = s->inuse;

506
	if (s->flags & SLAB_STORE_USER)
Christoph Lameter's avatar
Christoph Lameter committed
507
508
509
510
		off += 2 * sizeof(struct track);

	if (off != s->size)
		/* Beginning of the filler is the free pointer */
511
512
513
		print_section("Padding", p + off, s->size - off);

	dump_stack();
Christoph Lameter's avatar
Christoph Lameter committed
514
515
516
517
518
}

static void object_err(struct kmem_cache *s, struct page *page,
			u8 *object, char *reason)
{
519
	slab_bug(s, "%s", reason);
520
	print_trailer(s, page, object);
Christoph Lameter's avatar
Christoph Lameter committed
521
522
}

523
static void slab_err(struct kmem_cache *s, struct page *page, char *fmt, ...)
Christoph Lameter's avatar
Christoph Lameter committed
524
525
526
527
{
	va_list args;
	char buf[100];

528
529
	va_start(args, fmt);
	vsnprintf(buf, sizeof(buf), fmt, args);
Christoph Lameter's avatar
Christoph Lameter committed
530
	va_end(args);
531
	slab_bug(s, "%s", buf);
532
	print_page_info(page);
Christoph Lameter's avatar
Christoph Lameter committed
533
534
535
536
537
538
539
540
541
	dump_stack();
}

static void init_object(struct kmem_cache *s, void *object, int active)
{
	u8 *p = object;

	if (s->flags & __OBJECT_POISON) {
		memset(p, POISON_FREE, s->objsize - 1);
542
		p[s->objsize - 1] = POISON_END;
Christoph Lameter's avatar
Christoph Lameter committed
543
544
545
546
547
548
549
550
	}

	if (s->flags & SLAB_RED_ZONE)
		memset(p + s->objsize,
			active ? SLUB_RED_ACTIVE : SLUB_RED_INACTIVE,
			s->inuse - s->objsize);
}

551
static u8 *check_bytes(u8 *start, unsigned int value, unsigned int bytes)
Christoph Lameter's avatar
Christoph Lameter committed
552
553
554
{
	while (bytes) {
		if (*start != (u8)value)
555
			return start;
Christoph Lameter's avatar
Christoph Lameter committed
556
557
558
		start++;
		bytes--;
	}
559
560
561
562
563
564
565
566
567
568
569
570
	return NULL;
}

static void restore_bytes(struct kmem_cache *s, char *message, u8 data,
						void *from, void *to)
{
	slab_fix(s, "Restoring 0x%p-0x%p=0x%x\n", from, to - 1, data);
	memset(from, data, to - from);
}

static int check_bytes_and_report(struct kmem_cache *s, struct page *page,
			u8 *object, char *what,
571
			u8 *start, unsigned int value, unsigned int bytes)
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
{
	u8 *fault;
	u8 *end;

	fault = check_bytes(start, value, bytes);
	if (!fault)
		return 1;

	end = start + bytes;
	while (end > fault && end[-1] == value)
		end--;

	slab_bug(s, "%s overwritten", what);
	printk(KERN_ERR "INFO: 0x%p-0x%p. First byte 0x%x instead of 0x%x\n",
					fault, end - 1, fault[0], value);
	print_trailer(s, page, object);

	restore_bytes(s, what, value, fault, end);
	return 0;
Christoph Lameter's avatar
Christoph Lameter committed
591
592
593
594
595
596
597
598
599
}

/*
 * Object layout:
 *
 * object address
 * 	Bytes of the object to be managed.
 * 	If the freepointer may overlay the object then the free
 * 	pointer is the first word of the object.
Christoph Lameter's avatar
Christoph Lameter committed
600
 *
Christoph Lameter's avatar
Christoph Lameter committed
601
602
603
604
605
 * 	Poisoning uses 0x6b (POISON_FREE) and the last byte is
 * 	0xa5 (POISON_END)
 *
 * object + s->objsize
 * 	Padding to reach word boundary. This is also used for Redzoning.
Christoph Lameter's avatar
Christoph Lameter committed
606
607
608
 * 	Padding is extended by another word if Redzoning is enabled and
 * 	objsize == inuse.
 *
Christoph Lameter's avatar
Christoph Lameter committed
609
610
611
612
 * 	We fill with 0xbb (RED_INACTIVE) for inactive objects and with
 * 	0xcc (RED_ACTIVE) for objects in use.
 *
 * object + s->inuse
Christoph Lameter's avatar
Christoph Lameter committed
613
614
 * 	Meta data starts here.
 *
Christoph Lameter's avatar
Christoph Lameter committed
615
616
 * 	A. Free pointer (if we cannot overwrite object on free)
 * 	B. Tracking data for SLAB_STORE_USER
Christoph Lameter's avatar
Christoph Lameter committed
617
 * 	C. Padding to reach required alignment boundary or at mininum
Christoph Lameter's avatar
Christoph Lameter committed
618
 * 		one word if debugging is on to be able to detect writes
Christoph Lameter's avatar
Christoph Lameter committed
619
620
621
 * 		before the word boundary.
 *
 *	Padding is done using 0x5a (POISON_INUSE)
Christoph Lameter's avatar
Christoph Lameter committed
622
623
 *
 * object + s->size
Christoph Lameter's avatar
Christoph Lameter committed
624
 * 	Nothing is used beyond s->size.
Christoph Lameter's avatar
Christoph Lameter committed
625
 *
Christoph Lameter's avatar
Christoph Lameter committed
626
627
 * If slabcaches are merged then the objsize and inuse boundaries are mostly
 * ignored. And therefore no slab options that rely on these boundaries
Christoph Lameter's avatar
Christoph Lameter committed
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
 * may be used with merged slabcaches.
 */

static int check_pad_bytes(struct kmem_cache *s, struct page *page, u8 *p)
{
	unsigned long off = s->inuse;	/* The end of info */

	if (s->offset)
		/* Freepointer is placed after the object. */
		off += sizeof(void *);

	if (s->flags & SLAB_STORE_USER)
		/* We also have user information there */
		off += 2 * sizeof(struct track);

	if (s->size == off)
		return 1;

646
647
	return check_bytes_and_report(s, page, p, "Object padding",
				p + off, POISON_INUSE, s->size - off);
Christoph Lameter's avatar
Christoph Lameter committed
648
649
}

650
/* Check the pad bytes at the end of a slab page */
Christoph Lameter's avatar
Christoph Lameter committed
651
652
static int slab_pad_check(struct kmem_cache *s, struct page *page)
{
653
654
655
656
657
	u8 *start;
	u8 *fault;
	u8 *end;
	int length;
	int remainder;
Christoph Lameter's avatar
Christoph Lameter committed
658
659
660
661

	if (!(s->flags & SLAB_POISON))
		return 1;

662
	start = page_address(page);
663
	length = (PAGE_SIZE << compound_order(page));
664
665
	end = start + length;
	remainder = length % s->size;
Christoph Lameter's avatar
Christoph Lameter committed
666
667
668
	if (!remainder)
		return 1;

669
	fault = check_bytes(end - remainder, POISON_INUSE, remainder);
670
671
672
673
674
675
	if (!fault)
		return 1;
	while (end > fault && end[-1] == POISON_INUSE)
		end--;

	slab_err(s, page, "Padding overwritten. 0x%p-0x%p", fault, end - 1);
676
	print_section("Padding", end - remainder, remainder);
677
678
679

	restore_bytes(s, "slab padding", POISON_INUSE, start, end);
	return 0;
Christoph Lameter's avatar
Christoph Lameter committed
680
681
682
683
684
685
686
687
688
689
690
691
}

static int check_object(struct kmem_cache *s, struct page *page,
					void *object, int active)
{
	u8 *p = object;
	u8 *endobject = object + s->objsize;

	if (s->flags & SLAB_RED_ZONE) {
		unsigned int red =
			active ? SLUB_RED_ACTIVE : SLUB_RED_INACTIVE;

692
693
		if (!check_bytes_and_report(s, page, object, "Redzone",
			endobject, red, s->inuse - s->objsize))
Christoph Lameter's avatar
Christoph Lameter committed
694
695
			return 0;
	} else {
Ingo Molnar's avatar
Ingo Molnar committed
696
697
698
699
		if ((s->flags & SLAB_POISON) && s->objsize < s->inuse) {
			check_bytes_and_report(s, page, p, "Alignment padding",
				endobject, POISON_INUSE, s->inuse - s->objsize);
		}
Christoph Lameter's avatar
Christoph Lameter committed
700
701
702
703
	}

	if (s->flags & SLAB_POISON) {
		if (!active && (s->flags & __OBJECT_POISON) &&
704
705
706
			(!check_bytes_and_report(s, page, p, "Poison", p,
					POISON_FREE, s->objsize - 1) ||
			 !check_bytes_and_report(s, page, p, "Poison",
707
				p + s->objsize - 1, POISON_END, 1)))
Christoph Lameter's avatar
Christoph Lameter committed
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
			return 0;
		/*
		 * check_pad_bytes cleans up on its own.
		 */
		check_pad_bytes(s, page, p);
	}

	if (!s->offset && active)
		/*
		 * Object and freepointer overlap. Cannot check
		 * freepointer while object is allocated.
		 */
		return 1;

	/* Check free pointer validity */
	if (!check_valid_pointer(s, page, get_freepointer(s, p))) {
		object_err(s, page, p, "Freepointer corrupt");
		/*
		 * No choice but to zap it and thus loose the remainder
		 * of the free objects in this slab. May cause
Christoph Lameter's avatar
Christoph Lameter committed
728
		 * another error because the object count is now wrong.
Christoph Lameter's avatar
Christoph Lameter committed
729
		 */
730
		set_freepointer(s, p, NULL);
Christoph Lameter's avatar
Christoph Lameter committed
731
732
733
734
735
736
737
		return 0;
	}
	return 1;
}

static int check_slab(struct kmem_cache *s, struct page *page)
{
738
739
	int maxobj;

Christoph Lameter's avatar
Christoph Lameter committed
740
741
742
	VM_BUG_ON(!irqs_disabled());

	if (!PageSlab(page)) {
743
		slab_err(s, page, "Not a valid slab page");
Christoph Lameter's avatar
Christoph Lameter committed
744
745
		return 0;
	}
746
747
748
749
750
751
752
753

	maxobj = (PAGE_SIZE << compound_order(page)) / s->size;
	if (page->objects > maxobj) {
		slab_err(s, page, "objects %u > max %u",
			s->name, page->objects, maxobj);
		return 0;
	}
	if (page->inuse > page->objects) {
754
		slab_err(s, page, "inuse %u > max %u",
755
			s->name, page->inuse, page->objects);
Christoph Lameter's avatar
Christoph Lameter committed
756
757
758
759
760
761
762
763
		return 0;
	}
	/* Slab_pad_check fixes things up after itself */
	slab_pad_check(s, page);
	return 1;
}

/*
Christoph Lameter's avatar
Christoph Lameter committed
764
765
 * Determine if a certain object on a page is on the freelist. Must hold the
 * slab lock to guarantee that the chains are in a consistent state.
Christoph Lameter's avatar
Christoph Lameter committed
766
767
768
769
770
771
 */
static int on_freelist(struct kmem_cache *s, struct page *page, void *search)
{
	int nr = 0;
	void *fp = page->freelist;
	void *object = NULL;
772
	unsigned long max_objects;
Christoph Lameter's avatar
Christoph Lameter committed
773

774
	while (fp && nr <= page->objects) {
Christoph Lameter's avatar
Christoph Lameter committed
775
776
777
778
779
780
		if (fp == search)
			return 1;
		if (!check_valid_pointer(s, page, fp)) {
			if (object) {
				object_err(s, page, object,
					"Freechain corrupt");
781
				set_freepointer(s, object, NULL);
Christoph Lameter's avatar
Christoph Lameter committed
782
783
				break;
			} else {
784
				slab_err(s, page, "Freepointer corrupt");
785
				page->freelist = NULL;
786
				page->inuse = page->objects;
787
				slab_fix(s, "Freelist cleared");
Christoph Lameter's avatar
Christoph Lameter committed
788
789
790
791
792
793
794
795
796
				return 0;
			}
			break;
		}
		object = fp;
		fp = get_freepointer(s, object);
		nr++;
	}

797
798
799
800
801
802
803
804
805
806
	max_objects = (PAGE_SIZE << compound_order(page)) / s->size;
	if (max_objects > 65535)
		max_objects = 65535;

	if (page->objects != max_objects) {
		slab_err(s, page, "Wrong number of objects. Found %d but "
			"should be %d", page->objects, max_objects);
		page->objects = max_objects;
		slab_fix(s, "Number of objects adjusted.");
	}
807
	if (page->inuse != page->objects - nr) {
808
		slab_err(s, page, "Wrong object count. Counter is %d but "
809
810
			"counted were %d", page->inuse, page->objects - nr);
		page->inuse = page->objects - nr;
811
		slab_fix(s, "Object count adjusted.");
Christoph Lameter's avatar
Christoph Lameter committed
812
813
814
815
	}
	return search == NULL;
}

816
817
static void trace(struct kmem_cache *s, struct page *page, void *object,
								int alloc)
Christoph Lameter's avatar
Christoph Lameter committed
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
{
	if (s->flags & SLAB_TRACE) {
		printk(KERN_INFO "TRACE %s %s 0x%p inuse=%d fp=0x%p\n",
			s->name,
			alloc ? "alloc" : "free",
			object, page->inuse,
			page->freelist);

		if (!alloc)
			print_section("Object", (void *)object, s->objsize);

		dump_stack();
	}
}

833
/*
Christoph Lameter's avatar
Christoph Lameter committed
834
 * Tracking of fully allocated slabs for debugging purposes.
835
 */
Christoph Lameter's avatar
Christoph Lameter committed
836
static void add_full(struct kmem_cache_node *n, struct page *page)
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
{
	spin_lock(&n->list_lock);
	list_add(&page->lru, &n->full);
	spin_unlock(&n->list_lock);
}

static void remove_full(struct kmem_cache *s, struct page *page)
{
	struct kmem_cache_node *n;

	if (!(s->flags & SLAB_STORE_USER))
		return;

	n = get_node(s, page_to_nid(page));

	spin_lock(&n->list_lock);
	list_del(&page->lru);
	spin_unlock(&n->list_lock);
}

857
858
859
860
861
862
863
864
/* Tracking of the number of slabs for debugging purposes */
static inline unsigned long slabs_node(struct kmem_cache *s, int node)
{
	struct kmem_cache_node *n = get_node(s, node);

	return atomic_long_read(&n->nr_slabs);
}

865
static inline void inc_slabs_node(struct kmem_cache *s, int node, int objects)
866
867
868
869
870
871
872
873
874
{
	struct kmem_cache_node *n = get_node(s, node);

	/*
	 * May be called early in order to allocate a slab for the
	 * kmem_cache_node structure. Solve the chicken-egg
	 * dilemma by deferring the increment of the count during
	 * bootstrap (see early_kmem_cache_node_alloc).
	 */
875
	if (!NUMA_BUILD || n) {
876
		atomic_long_inc(&n->nr_slabs);
877
878
		atomic_long_add(objects, &n->total_objects);
	}
879
}
880
static inline void dec_slabs_node(struct kmem_cache *s, int node, int objects)
881
882
883
884
{
	struct kmem_cache_node *n = get_node(s, node);

	atomic_long_dec(&n->nr_slabs);
885
	atomic_long_sub(objects, &n->total_objects);
886
887
888
}

/* Object debug checks for alloc/free paths */
Christoph Lameter's avatar
Christoph Lameter committed
889
890
891
892
893
894
895
896
897
898
899
900
static void setup_object_debug(struct kmem_cache *s, struct page *page,
								void *object)
{
	if (!(s->flags & (SLAB_STORE_USER|SLAB_RED_ZONE|__OBJECT_POISON)))
		return;

	init_object(s, object, 0);
	init_tracking(s, object);
}

static int alloc_debug_processing(struct kmem_cache *s, struct page *page,
						void *object, void *addr)
Christoph Lameter's avatar
Christoph Lameter committed
901
902
903
904
{
	if (!check_slab(s, page))
		goto bad;

905
	if (!on_freelist(s, page, object)) {
906
		object_err(s, page, object, "Object already allocated");
907
		goto bad;
Christoph Lameter's avatar
Christoph Lameter committed
908
909
910
911
	}

	if (!check_valid_pointer(s, page, object)) {
		object_err(s, page, object, "Freelist Pointer check fails");
912
		goto bad;
Christoph Lameter's avatar
Christoph Lameter committed
913
914
	}

915
	if (!check_object(s, page, object, 0))
Christoph Lameter's avatar
Christoph Lameter committed
916
917
		goto bad;

Christoph Lameter's avatar
Christoph Lameter committed
918
919
920
921
922
	/* Success perform special debug activities for allocs */
	if (s->flags & SLAB_STORE_USER)
		set_track(s, object, TRACK_ALLOC, addr);
	trace(s, page, object, 1);
	init_object(s, object, 1);
Christoph Lameter's avatar
Christoph Lameter committed
923
	return 1;
Christoph Lameter's avatar
Christoph Lameter committed
924

Christoph Lameter's avatar
Christoph Lameter committed
925
926
927
928
929
bad:
	if (PageSlab(page)) {
		/*
		 * If this is a slab page then lets do the best we can
		 * to avoid issues in the future. Marking all objects
Christoph Lameter's avatar
Christoph Lameter committed
930
		 * as used avoids touching the remaining objects.
Christoph Lameter's avatar
Christoph Lameter committed
931
		 */
932
		slab_fix(s, "Marking all objects used");
933
		page->inuse = page->objects;
934
		page->freelist = NULL;
Christoph Lameter's avatar
Christoph Lameter committed
935
936
937
938
	}
	return 0;
}

Christoph Lameter's avatar
Christoph Lameter committed
939
940
static int free_debug_processing(struct kmem_cache *s, struct page *page,
						void *object, void *addr)
Christoph Lameter's avatar
Christoph Lameter committed
941
942
943
944
945
{
	if (!check_slab(s, page))
		goto fail;

	if (!check_valid_pointer(s, page, object)) {
946
		slab_err(s, page, "Invalid object pointer 0x%p", object);
Christoph Lameter's avatar
Christoph Lameter committed
947
948
949
950
		goto fail;
	}

	if (on_freelist(s, page, object)) {
951
		object_err(s, page, object, "Object already free");
Christoph Lameter's avatar
Christoph Lameter committed
952
953
954
955
956
957
958
		goto fail;
	}

	if (!check_object(s, page, object, 1))
		return 0;

	if (unlikely(s != page->slab)) {
Ingo Molnar's avatar
Ingo Molnar committed
959
		if (!PageSlab(page)) {
960
961
			slab_err(s, page, "Attempt to free object(0x%p) "
				"outside of slab", object);
Ingo Molnar's avatar
Ingo Molnar committed
962
		} else if (!page->slab) {
Christoph Lameter's avatar
Christoph Lameter committed
963
			printk(KERN_ERR
964
				"SLUB <none>: no slab for object 0x%p.\n",
Christoph Lameter's avatar
Christoph Lameter committed
965
						object);
966
			dump_stack();
967
		} else
968
969
			object_err(s, page, object,
					"page slab pointer corrupt.");
Christoph Lameter's avatar
Christoph Lameter committed
970
971
		goto fail;
	}
Christoph Lameter's avatar
Christoph Lameter committed
972
973

	/* Special debug activities for freeing objects */
974
	if (!SlabFrozen(page) && !page->freelist)
Christoph Lameter's avatar
Christoph Lameter committed
975
976
977
978
979
		remove_full(s, page);
	if (s->flags & SLAB_STORE_USER)
		set_track(s, object, TRACK_FREE, addr);
	trace(s, page, object, 0);
	init_object(s, object, 0);
Christoph Lameter's avatar
Christoph Lameter committed
980
	return 1;
Christoph Lameter's avatar
Christoph Lameter committed
981

Christoph Lameter's avatar
Christoph Lameter committed
982
fail:
983
	slab_fix(s, "Object at 0x%p not freed", object);
Christoph Lameter's avatar
Christoph Lameter committed
984
985
986
	return 0;
}

987
988
static int __init setup_slub_debug(char *str)
{
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
	slub_debug = DEBUG_DEFAULT_FLAGS;
	if (*str++ != '=' || !*str)
		/*
		 * No options specified. Switch on full debugging.
		 */
		goto out;

	if (*str == ',')
		/*
		 * No options but restriction on slabs. This means full
		 * debugging for slabs matching a pattern.
		 */
		goto check_slabs;

	slub_debug = 0;
	if (*str == '-')
		/*
		 * Switch off all debugging measures.
		 */
		goto out;

	/*
	 * Determine which debug features should be switched on
	 */
1013
	for (; *str && *str != ','; str++) {
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
		switch (tolower(*str)) {
		case 'f':
			slub_debug |= SLAB_DEBUG_FREE;
			break;
		case 'z':
			slub_debug |= SLAB_RED_ZONE;
			break;
		case 'p':
			slub_debug |= SLAB_POISON;
			break;
		case 'u':
			slub_debug |= SLAB_STORE_USER;
			break;
		case 't':
			slub_debug |= SLAB_TRACE;
			break;
		default:
			printk(KERN_ERR "slub_debug option '%c' "
1032
				"unknown. skipped\n", *str);
1033
		}
1034
1035
	}

1036
check_slabs:
1037
1038
	if (*str == ',')
		slub_debug_slabs = str + 1;
1039
out:
1040
1041
1042
1043
1044
	return 1;
}

__setup("slub_debug", setup_slub_debug);

1045
1046
static unsigned long kmem_cache_flags(unsigned long objsize,
	unsigned long flags, const char *name,
1047
	void (*ctor)(struct kmem_cache *, void *))
1048
1049
{
	/*
1050
	 * Enable debugging if selected on the kernel commandline.
1051
	 */
1052
1053
1054
	if (slub_debug && (!slub_debug_slabs ||
	    strncmp(slub_debug_slabs, name, strlen(slub_debug_slabs)) == 0))
			flags |= slub_debug;
1055
1056

	return flags;
1057
1058
}
#else
Christoph Lameter's avatar
Christoph Lameter committed
1059
1060
static inline void setup_object_debug(struct kmem_cache *s,
			struct page *page, void *object) {}
1061

Christoph Lameter's avatar
Christoph Lameter committed
1062
1063
static inline int alloc_debug_processing(struct kmem_cache *s,
	struct page *page, void *object, void *addr) { return 0; }
1064

Christoph Lameter's avatar
Christoph Lameter committed
1065
1066
static inline int free_debug_processing(struct kmem_cache *s,
	struct page *page, void *object, void *addr) { return 0; }
1067
1068
1069
1070
1071

static inline int slab_pad_check(struct kmem_cache *s, struct page *page)
			{ return 1; }
static inline int check_object(struct kmem_cache *s, struct page *page,
			void *object, int active) { return 1; }
Christoph Lameter's avatar
Christoph Lameter committed
1072
static inline void add_full(struct kmem_cache_node *n, struct page *page) {}
1073
1074
static inline unsigned long kmem_cache_flags(unsigned long objsize,
	unsigned long flags, const char *name,
1075
	void (*ctor)(struct kmem_cache *, void *))
1076
1077
1078
{
	return flags;
}
1079
#define slub_debug 0
1080
1081
1082

static inline unsigned long slabs_node(struct kmem_cache *s, int node)
							{ return 0; }
1083
1084
1085
1086
static inline void inc_slabs_node(struct kmem_cache *s, int node,
							int objects) {}
static inline void dec_slabs_node(struct kmem_cache *s, int node,
							int objects) {}
1087
#endif
1088

Christoph Lameter's avatar
Christoph Lameter committed
1089
1090
1091
/*
 * Slab allocation and freeing
 */
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
static inline struct page *alloc_slab_page(gfp_t flags, int node,
					struct kmem_cache_order_objects oo)
{
	int order = oo_order(oo);

	if (node == -1)
		return alloc_pages(flags, order);
	else
		return alloc_pages_node(node, flags, order);
}

Christoph Lameter's avatar
Christoph Lameter committed
1103
1104
static struct page *allocate_slab(struct kmem_cache *s, gfp_t flags, int node)
{
1105
	struct page *page;
1106
	struct kmem_cache_order_objects oo = s->oo;
Christoph Lameter's avatar
Christoph Lameter committed
1107

1108
	flags |= s->allocflags;
1109

1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
	page = alloc_slab_page(flags | __GFP_NOWARN | __GFP_NORETRY, node,
									oo);
	if (unlikely(!page)) {
		oo = s->min;
		/*
		 * Allocation may have failed due to fragmentation.
		 * Try a lower order alloc if possible
		 */
		page = alloc_slab_page(flags, node, oo);
		if (!page)
			return NULL;
Christoph Lameter's avatar
Christoph Lameter committed
1121

1122
1123
		stat(get_cpu_slab(s, raw_smp_processor_id()), ORDER_FALLBACK);
	}
1124
	page->objects = oo_objects(oo);
Christoph Lameter's avatar
Christoph Lameter committed
1125
1126
1127
	mod_zone_page_state(page_zone(page),
		(s->flags & SLAB_RECLAIM_ACCOUNT) ?
		NR_SLAB_RECLAIMABLE : NR_SLAB_UNRECLAIMABLE,
1128
		1 << oo_order(oo));
Christoph Lameter's avatar
Christoph Lameter committed
1129
1130
1131
1132
1133
1134
1135

	return page;
}

static void setup_object(struct kmem_cache *s, struct page *page,
				void *object)
{
Christoph Lameter's avatar
Christoph Lameter committed
1136
	setup_object_debug(s, page, object);
1137
	if (unlikely(s->ctor))
1138
		s->ctor(s, object);
Christoph Lameter's avatar
Christoph Lameter committed
1139
1140
1141
1142
1143
1144
1145
1146
1147
}

static struct page *new_slab(struct kmem_cache *s, gfp_t flags, int node)
{
	struct page *page;
	void *start;
	void *last;
	void *p;

Christoph Lameter's avatar
Christoph Lameter committed
1148
	BUG_ON(flags & GFP_SLAB_BUG_MASK);
Christoph Lameter's avatar
Christoph Lameter committed
1149

Christoph Lameter's avatar
Christoph Lameter committed
1150
1151
	page = allocate_slab(s,
		flags & (GFP_RECLAIM_MASK | GFP_CONSTRAINT_MASK), node);
Christoph Lameter's avatar
Christoph Lameter committed
1152
1153
1154
	if (!page)
		goto out;

1155
	inc_slabs_node(s, page_to_nid(page), page->objects);
Christoph Lameter's avatar
Christoph Lameter committed
1156
1157
1158
1159
	page->slab = s;
	page->flags |= 1 << PG_slab;
	if (s->flags & (SLAB_DEBUG_FREE | SLAB_RED_ZONE | SLAB_POISON |
			SLAB_STORE_USER | SLAB_TRACE))
1160
		SetSlabDebug(page);
Christoph Lameter's avatar
Christoph Lameter committed
1161
1162
1163
1164

	start = page_address(page);

	if (unlikely(s->flags & SLAB_POISON))
1165
		memset(start, POISON_INUSE, PAGE_SIZE << compound_order(page));
Christoph Lameter's avatar
Christoph Lameter committed
1166
1167

	last = start;
1168
	for_each_object(p, s, start, page->objects) {
Christoph Lameter's avatar
Christoph Lameter committed
1169
1170
1171
1172
1173
		setup_object(s, page, last);
		set_freepointer(s, last, p);
		last = p;
	}
	setup_object(s, page, last);
1174
	set_freepointer(s, last, NULL);
Christoph Lameter's avatar
Christoph Lameter committed
1175
1176
1177
1178
1179
1180
1181
1182
1183

	page->freelist = start;
	page->inuse = 0;
out:
	return page;
}

static void __free_slab(struct kmem_cache *s, struct page *page)
{
1184
1185
	int order = compound_order(page);
	int pages = 1 << order;
Christoph Lameter's avatar
Christoph Lameter committed
1186

1187
	if (unlikely(SlabDebug(page))) {
Christoph Lameter's avatar
Christoph Lameter committed
1188
1189
1190
		void *p;

		slab_pad_check(s, page);
1191
1192
		for_each_object(p, s, page_address(page),
						page->objects)
Christoph Lameter's avatar
Christoph Lameter committed
1193
			check_object(s, page, p, 0);
1194
		ClearSlabDebug(page);
Christoph Lameter's avatar
Christoph Lameter committed
1195
1196
1197
1198
1199
	}

	mod_zone_page_state(page_zone(page),
		(s->flags & SLAB_RECLAIM_ACCOUNT) ?
		NR_SLAB_RECLAIMABLE : NR_SLAB_UNRECLAIMABLE,
1200
		-pages);
Christoph Lameter's avatar
Christoph Lameter committed
1201

1202
1203
	__ClearPageSlab(page);
	reset_page_mapcount(page);
1204
	__free_pages(page, order);
Christoph Lameter's avatar
Christoph Lameter committed
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
}

static void rcu_free_slab(struct rcu_head *h)
{
	struct page *page;

	page = container_of((struct list_head *)h, struct page, lru);
	__free_slab(page->slab, page);
}

static void free_slab(struct kmem_cache *s, struct page *page)
{
	if (unlikely(s->flags & SLAB_DESTROY_BY_RCU)) {
		/*
		 * RCU free overloads the RCU head over the LRU
		 */
		struct rcu_head *head = (void *)&page->lru;

		call_rcu(head, rcu_free_slab);
	} else
		__free_slab(s, page);
}

static void discard_slab(struct kmem_cache *s, struct page *page)
{
1230
	dec_slabs_node(s, page_to_nid(page), page->objects);
Christoph Lameter's avatar
Christoph Lameter committed
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
	free_slab(s, page);
}

/*
 * Per slab locking using the pagelock
 */
static __always_inline void slab_lock(struct page *page)
{
	bit_spin_lock(PG_locked, &page->flags);
}

static __always_inline void slab_unlock(struct page *page)
{
Nick Piggin's avatar
Nick Piggin committed
1244
	__bit_spin_unlock(PG_locked, &page->flags);
Christoph Lameter's avatar
Christoph Lameter committed
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
}

static __always_inline int slab_trylock(struct page *page)
{
	int rc = 1;

	rc = bit_spin_trylock(PG_locked, &page->flags);
	return rc;
}

/*
 * Management of partially allocated slabs
 */
1258
1259
static void add_partial(struct kmem_cache_node *n,
				struct page *page, int tail)
Christoph Lameter's avatar
Christoph Lameter committed
1260
{
Christoph Lameter's avatar
Christoph Lameter committed
1261
1262
	spin_lock(&n->list_lock);
	n->nr_partial++;
1263
1264
1265
1266
	if (tail)
		list_add_tail(&page->lru, &n->partial);
	else
		list_add(&page->lru, &n->partial);
Christoph Lameter's avatar
Christoph Lameter committed
1267
1268
1269
	spin_unlock(&n->list_lock);
}

1270
static void remove_partial(struct kmem_cache *s, struct page *page)
Christoph Lameter's avatar
Christoph Lameter committed
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
{
	struct kmem_cache_node *n = get_node(s, page_to_nid(page));

	spin_lock(&n->list_lock);
	list_del(&page->lru);
	n->nr_partial--;
	spin_unlock(&n->list_lock);
}

/*
Christoph Lameter's avatar
Christoph Lameter committed
1281
 * Lock slab and remove from the partial list.
Christoph Lameter's avatar
Christoph Lameter committed
1282
 *
Christoph Lameter's avatar
Christoph Lameter committed
1283
 * Must hold list_lock.
Christoph Lameter's avatar
Christoph Lameter committed
1284
 */
1285
1286
static inline int lock_and_freeze_slab(struct kmem_cache_node *n,
							struct page *page)
Christoph Lameter's avatar
Christoph Lameter committed
1287
1288
1289
1290
{
	if (slab_trylock(page)) {
		list_del(&page->lru);
		n->nr_partial--;
1291
		SetSlabFrozen(page);
Christoph Lameter's avatar
Christoph Lameter committed
1292
1293
1294
1295
1296
1297
		return 1;
	}
	return 0;
}

/*
Christoph Lameter's avatar
Christoph Lameter committed
1298
 * Try to allocate a partial slab from a specific node.
Christoph Lameter's avatar
Christoph Lameter committed
1299
1300
1301
1302
1303
1304
1305
1306
 */
static struct page *get_partial_node(struct kmem_cache_node *n)
{
	struct page *page;

	/*
	 * Racy check. If we mistakenly see no partial slabs then we
	 * just allocate an empty slab. If we mistakenly try to get a
Christoph Lameter's avatar
Christoph Lameter committed
1307
1308
	 * partial slab and there is none available then get_partials()
	 * will return NULL.
Christoph Lameter's avatar
Christoph Lameter committed
1309
1310
1311
1312
1313
1314
	 */
	if (!n || !n->nr_partial)
		return NULL;

	spin_lock(&n->list_lock);
	list_for_each_entry(page, &n->partial, lru)
1315
		if (lock_and_freeze_slab(n, page))
Christoph Lameter's avatar
Christoph Lameter committed
1316
1317
1318
1319
1320
1321
1322
1323
			goto out;
	page = NULL;
out:
	spin_unlock(&n->list_lock);
	return page;
}

/*
Christoph Lameter's avatar
Christoph Lameter committed
1324
 * Get a page from somewhere. Search in increasing NUMA distances.
Christoph Lameter's avatar
Christoph Lameter committed
1325
1326
1327
1328
1329
 */
static struct page *get_any_partial(struct kmem_cache *s, gfp_t flags)
{
#ifdef CONFIG_NUMA
	struct zonelist *zonelist;
1330
	struct zoneref *z;
1331
1332
	struct zone *zone;
	enum zone_type high_zoneidx = gfp_zone(flags);
Christoph Lameter's avatar
Christoph Lameter committed
1333
1334
1335
	struct page *page;

	/*
Christoph Lameter's avatar
Christoph Lameter committed
1336
1337
1338
1339
	 * The defrag ratio allows a configuration of the tradeoffs between
	 * inter node defragmentation and node local allocations. A lower
	 * defrag_ratio increases the tendency to do local allocations
	 * instead of attempting to obtain partial slabs from other nodes.
Christoph Lameter's avatar
Christoph Lameter committed
1340
	 *
Christoph Lameter's avatar
Christoph Lameter committed
1341
1342
1343
1344
	 * If the defrag_ratio is set to 0 then kmalloc() always
	 * returns node local objects. If the ratio is higher then kmalloc()
	 * may return off node objects because partial slabs are obtained
	 * from other nodes and filled up.
Christoph Lameter's avatar
Christoph Lameter committed
1345
	 *
Christoph Lameter's avatar
Christoph Lameter committed
1346
	 * If /sys/kernel/slab/xx/defrag_ratio is set to 100 (which makes
Christoph Lameter's avatar
Christoph Lameter committed
1347
1348
1349
1350
1351
	 * defrag_ratio = 1000) then every (well almost) allocation will
	 * first attempt to defrag slab caches on other nodes. This means
	 * scanning over all nodes to look for partial slabs which may be
	 * expensive if we do it every time we are trying to find a slab
	 * with available objects.
Christoph Lameter's avatar
Christoph Lameter committed
1352
	 */
1353
1354
	if (!s->remote_node_defrag_ratio ||
			get_cycles() % 1024 > s->remote_node_defrag_ratio)
Christoph Lameter's avatar
Christoph Lameter committed
1355
1356
		return NULL;

1357
	zonelist = node_zonelist(slab_node(current->mempolicy), flags);
1358
	for_each_zone_zonelist(zone, z, zonelist, high_zoneidx) {
Christoph Lameter's avatar
Christoph Lameter committed
1359
1360
		struct kmem_cache_node *n;

1361
		n = get_node(s, zone_to_nid(zone));
Christoph Lameter's avatar
Christoph Lameter committed
1362

1363
		if (n && cpuset_zone_allowed_hardwall(zone, flags) &&
Christoph Lameter's avatar
Christoph Lameter committed
1364