slub.c 87.3 KB
Newer Older
Christoph Lameter's avatar
Christoph Lameter committed
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
/*
 * 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.
 *
 * (C) 2007 SGI, Christoph Lameter <clameter@sgi.com>
 */

#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>
#include <linux/kallsyms.h>

/*
 * 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
69
70
 * 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
71
 * freed then the slab will show up again on the partial lists.
Christoph Lameter's avatar
Christoph Lameter committed
72
73
 * We track full slabs for debugging purposes though because otherwise we
 * cannot scan all objects.
Christoph Lameter's avatar
Christoph Lameter committed
74
75
76
77
78
79
80
 *
 * 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.
 *
81
82
83
84
85
86
87
88
89
90
91
92
 * 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
93
94
95
 * 			lockless_freelist that allows lockless access to
 * 			free objects in addition to the regular freelist
 * 			that requires the slab lock.
Christoph Lameter's avatar
Christoph Lameter committed
96
97
98
 *
 * PageError		Slab requires special handling due to debug
 * 			options set. This moves	slab handling out of
99
 * 			the fast path and disables lockless freelists.
Christoph Lameter's avatar
Christoph Lameter committed
100
101
 */

102
103
104
105
106
107
108
109
#define FROZEN (1 << PG_active)

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

110
111
static inline int SlabFrozen(struct page *page)
{
112
	return page->flags & FROZEN;
113
114
115
116
}

static inline void SetSlabFrozen(struct page *page)
{
117
	page->flags |= FROZEN;
118
119
120
121
}

static inline void ClearSlabFrozen(struct page *page)
{
122
	page->flags &= ~FROZEN;
123
124
}

125
126
static inline int SlabDebug(struct page *page)
{
127
	return page->flags & SLABDEBUG;
128
129
130
131
}

static inline void SetSlabDebug(struct page *page)
{
132
	page->flags |= SLABDEBUG;
133
134
135
136
}

static inline void ClearSlabDebug(struct page *page)
{
137
	page->flags &= ~SLABDEBUG;
138
139
}

Christoph Lameter's avatar
Christoph Lameter committed
140
141
142
143
144
/*
 * Issues still to be resolved:
 *
 * - The per cpu array is updated for each new slab and and is a remote
 *   cacheline for most nodes. This could become a bouncing cacheline given
Christoph Lameter's avatar
Christoph Lameter committed
145
146
 *   enough frequent updates. There are 16 pointers in a cacheline, so at
 *   max 16 cpus could compete for the cacheline which may be okay.
Christoph Lameter's avatar
Christoph Lameter committed
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
 *
 * - 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

#if PAGE_SHIFT <= 12

/*
 * Small page size. Make sure that we do not fragment memory
 */
#define DEFAULT_MAX_ORDER 1
#define DEFAULT_MIN_OBJECTS 4

#else

/*
 * Large page machines are customarily able to handle larger
 * page orders.
 */
#define DEFAULT_MAX_ORDER 2
#define DEFAULT_MIN_OBJECTS 8

#endif

175
176
177
178
/*
 * Mininum number of partial slabs. These will be left on the partial
 * lists even if they are empty. kmem_cache_shrink may reclaim them.
 */
Christoph Lameter's avatar
Christoph Lameter committed
179
180
#define MIN_PARTIAL 2

181
182
183
184
185
186
187
/*
 * 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
188
189
#define DEBUG_DEFAULT_FLAGS (SLAB_DEBUG_FREE | SLAB_RED_ZONE | \
				SLAB_POISON | SLAB_STORE_USER)
Christoph Lameter's avatar
Christoph Lameter committed
190

Christoph Lameter's avatar
Christoph Lameter committed
191
192
193
194
195
196
197
198
199
200
/*
 * 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
201
#define ARCH_KMALLOC_MINALIGN __alignof__(unsigned long long)
Christoph Lameter's avatar
Christoph Lameter committed
202
203
204
#endif

#ifndef ARCH_SLAB_MINALIGN
205
#define ARCH_SLAB_MINALIGN __alignof__(unsigned long long)
Christoph Lameter's avatar
Christoph Lameter committed
206
207
208
209
210
#endif

/* Internal SLUB flags */
#define __OBJECT_POISON 0x80000000	/* Poison object */

211
212
213
214
215
/* Not all arches define cache_line_size */
#ifndef cache_line_size
#define cache_line_size()	L1_CACHE_BYTES
#endif

Christoph Lameter's avatar
Christoph Lameter committed
216
217
218
219
220
221
222
223
224
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
225
	UP,		/* Everything works but does not show up in sysfs */
Christoph Lameter's avatar
Christoph Lameter committed
226
227
228
229
230
231
232
	SYSFS		/* Sysfs up */
} slab_state = DOWN;

/* A list of all slab caches on the system */
static DECLARE_RWSEM(slub_lock);
LIST_HEAD(slab_caches);

233
234
235
236
237
238
239
240
241
242
243
244
/*
 * 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 };

245
#if defined(CONFIG_SYSFS) && defined(CONFIG_SLUB_DEBUG)
Christoph Lameter's avatar
Christoph Lameter committed
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
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 *);
#else
static int sysfs_slab_add(struct kmem_cache *s) { return 0; }
static int sysfs_slab_alias(struct kmem_cache *s, const char *p) { return 0; }
static void sysfs_slab_remove(struct kmem_cache *s) {}
#endif

/********************************************************************
 * 			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
}

273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
static inline int check_valid_pointer(struct kmem_cache *s,
				struct page *page, const void *object)
{
	void *base;

	if (!object)
		return 1;

	base = page_address(page);
	if (object < base || object >= base + s->objects * s->size ||
		(object - base) % s->size) {
		return 0;
	}

	return 1;
}

290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
/*
 * 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 */
#define for_each_object(__p, __s, __addr) \
	for (__p = (__addr); __p < (__addr) + (__s)->objects * (__s)->size;\
			__p += (__s)->size)

/* Scan freelist */
#define for_each_free_object(__p, __s, __free) \
	for (__p = (__free); __p; __p = get_freepointer((__s), __p))

/* 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;
}

322
323
324
325
#ifdef CONFIG_SLUB_DEBUG
/*
 * Debug settings:
 */
326
327
328
#ifdef CONFIG_SLUB_DEBUG_ON
static int slub_debug = DEBUG_DEFAULT_FLAGS;
#else
329
static int slub_debug;
330
#endif
331
332
333

static char *slub_debug_slabs;

Christoph Lameter's avatar
Christoph Lameter committed
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
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
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
/*
 * 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) {
			printk(KERN_ERR "%10s 0x%p: ", text, addr + i);
			newline = 0;
		}
		printk(" %02x", addr[i]);
		offset = i % 16;
		ascii[offset] = isgraph(addr[i]) ? addr[i] : '.';
		if (offset == 15) {
			printk(" %s\n",ascii);
			newline = 1;
		}
	}
	if (!newline) {
		i %= 16;
		while (i < 16) {
			printk("   ");
			ascii[i] = ' ';
			i++;
		}
		printk(" %s\n", ascii);
	}
}

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)
{
	if (s->flags & SLAB_STORE_USER) {
		set_track(s, object, TRACK_FREE, NULL);
		set_track(s, object, TRACK_ALLOC, NULL);
	}
}

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

	printk(KERN_ERR "%s: ", s);
	__print_symbol("%s", (unsigned long)t->addr);
	printk(" jiffies_ago=%lu cpu=%u pid=%d\n", jiffies - t->when, t->cpu, t->pid);
}

static void print_trailer(struct kmem_cache *s, u8 *p)
{
	unsigned int off;	/* Offset of last byte */

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

	printk(KERN_ERR "FreePointer 0x%p -> 0x%p\n",
			p + s->offset,
			get_freepointer(s, p));

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

	if (s->flags & SLAB_STORE_USER) {
		print_track("Last alloc", get_track(s, p, TRACK_ALLOC));
		print_track("Last free ", get_track(s, p, TRACK_FREE));
		off += 2 * sizeof(struct track);
	}

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

static void object_err(struct kmem_cache *s, struct page *page,
			u8 *object, char *reason)
{
	u8 *addr = page_address(page);

	printk(KERN_ERR "*** SLUB %s: %s@0x%p slab 0x%p\n",
			s->name, reason, object, page);
	printk(KERN_ERR "    offset=%tu flags=0x%04lx inuse=%u freelist=0x%p\n",
		object - addr, page->flags, page->inuse, page->freelist);
	if (object > addr + 16)
		print_section("Bytes b4", object - 16, 16);
	print_section("Object", object, min(s->objsize, 128));
	print_trailer(s, object);
	dump_stack();
}

static void slab_err(struct kmem_cache *s, struct page *page, char *reason, ...)
{
	va_list args;
	char buf[100];

	va_start(args, reason);
	vsnprintf(buf, sizeof(buf), reason, args);
	va_end(args);
	printk(KERN_ERR "*** SLUB %s: %s in slab @0x%p\n", s->name, buf,
		page);
	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);
		p[s->objsize -1] = POISON_END;
	}

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

static int check_bytes(u8 *start, unsigned int value, unsigned int bytes)
{
	while (bytes) {
		if (*start != (u8)value)
			return 0;
		start++;
		bytes--;
	}
	return 1;
}

/*
 * 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
510
 *
Christoph Lameter's avatar
Christoph Lameter committed
511
512
513
514
515
 * 	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
516
517
518
 * 	Padding is extended by another word if Redzoning is enabled and
 * 	objsize == inuse.
 *
Christoph Lameter's avatar
Christoph Lameter committed
519
520
521
522
 * 	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
523
524
 * 	Meta data starts here.
 *
Christoph Lameter's avatar
Christoph Lameter committed
525
526
 * 	A. Free pointer (if we cannot overwrite object on free)
 * 	B. Tracking data for SLAB_STORE_USER
Christoph Lameter's avatar
Christoph Lameter committed
527
528
529
530
531
 * 	C. Padding to reach required alignment boundary or at mininum
 * 		one word if debuggin is on to be able to detect writes
 * 		before the word boundary.
 *
 *	Padding is done using 0x5a (POISON_INUSE)
Christoph Lameter's avatar
Christoph Lameter committed
532
533
 *
 * object + s->size
Christoph Lameter's avatar
Christoph Lameter committed
534
 * 	Nothing is used beyond s->size.
Christoph Lameter's avatar
Christoph Lameter committed
535
 *
Christoph Lameter's avatar
Christoph Lameter committed
536
537
 * 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
538
539
540
541
542
543
 * may be used with merged slabcaches.
 */

static void restore_bytes(struct kmem_cache *s, char *message, u8 data,
						void *from, void *to)
{
544
	printk(KERN_ERR "@@@ SLUB %s: Restoring %s (0x%x) from 0x%p-0x%p\n",
Christoph Lameter's avatar
Christoph Lameter committed
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
		s->name, message, data, from, to - 1);
	memset(from, data, to - from);
}

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;

	if (check_bytes(p + off, POISON_INUSE, s->size - off))
		return 1;

	object_err(s, page, p, "Object padding check fails");

	/*
	 * Restore padding
	 */
	restore_bytes(s, "object padding", POISON_INUSE, p + off, p + s->size);
	return 0;
}

static int slab_pad_check(struct kmem_cache *s, struct page *page)
{
	u8 *p;
	int length, remainder;

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

	p = page_address(page);
	length = s->objects * s->size;
	remainder = (PAGE_SIZE << s->order) - length;
	if (!remainder)
		return 1;

	if (!check_bytes(p + length, POISON_INUSE, remainder)) {
591
		slab_err(s, page, "Padding check failed");
Christoph Lameter's avatar
Christoph Lameter committed
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
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
		restore_bytes(s, "slab padding", POISON_INUSE, p + length,
			p + length + remainder);
		return 0;
	}
	return 1;
}

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;

		if (!check_bytes(endobject, red, s->inuse - s->objsize)) {
			object_err(s, page, object,
			active ? "Redzone Active" : "Redzone Inactive");
			restore_bytes(s, "redzone", red,
				endobject, object + s->inuse);
			return 0;
		}
	} else {
		if ((s->flags & SLAB_POISON) && s->objsize < s->inuse &&
			!check_bytes(endobject, POISON_INUSE,
					s->inuse - s->objsize)) {
		object_err(s, page, p, "Alignment padding check fails");
		/*
		 * Fix it so that there will not be another report.
		 *
		 * Hmmm... We may be corrupting an object that now expects
		 * to be longer than allowed.
		 */
		restore_bytes(s, "alignment padding", POISON_INUSE,
			endobject, object + s->inuse);
		}
	}

	if (s->flags & SLAB_POISON) {
		if (!active && (s->flags & __OBJECT_POISON) &&
			(!check_bytes(p, POISON_FREE, s->objsize - 1) ||
				p[s->objsize - 1] != POISON_END)) {

			object_err(s, page, p, "Poison check failed");
			restore_bytes(s, "Poison", POISON_FREE,
						p, p + s->objsize -1);
			restore_bytes(s, "Poison", POISON_END,
					p + s->objsize - 1, p + s->objsize);
			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
663
		 * another error because the object count is now wrong.
Christoph Lameter's avatar
Christoph Lameter committed
664
665
666
667
668
669
670
671
672
673
674
675
		 */
		set_freepointer(s, p, NULL);
		return 0;
	}
	return 1;
}

static int check_slab(struct kmem_cache *s, struct page *page)
{
	VM_BUG_ON(!irqs_disabled());

	if (!PageSlab(page)) {
676
677
		slab_err(s, page, "Not a valid slab page flags=%lx "
			"mapping=0x%p count=%d", page->flags, page->mapping,
Christoph Lameter's avatar
Christoph Lameter committed
678
679
680
681
			page_count(page));
		return 0;
	}
	if (page->offset * sizeof(void *) != s->offset) {
682
683
		slab_err(s, page, "Corrupted offset %lu flags=0x%lx "
			"mapping=0x%p count=%d",
Christoph Lameter's avatar
Christoph Lameter committed
684
685
686
687
688
689
690
			(unsigned long)(page->offset * sizeof(void *)),
			page->flags,
			page->mapping,
			page_count(page));
		return 0;
	}
	if (page->inuse > s->objects) {
691
692
693
		slab_err(s, page, "inuse %u > max %u @0x%p flags=%lx "
			"mapping=0x%p count=%d",
			s->name, page->inuse, s->objects, page->flags,
Christoph Lameter's avatar
Christoph Lameter committed
694
695
696
697
698
699
700
701
702
			page->mapping, page_count(page));
		return 0;
	}
	/* Slab_pad_check fixes things up after itself */
	slab_pad_check(s, page);
	return 1;
}

/*
Christoph Lameter's avatar
Christoph Lameter committed
703
704
 * 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
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
 */
static int on_freelist(struct kmem_cache *s, struct page *page, void *search)
{
	int nr = 0;
	void *fp = page->freelist;
	void *object = NULL;

	while (fp && nr <= s->objects) {
		if (fp == search)
			return 1;
		if (!check_valid_pointer(s, page, fp)) {
			if (object) {
				object_err(s, page, object,
					"Freechain corrupt");
				set_freepointer(s, object, NULL);
				break;
			} else {
722
723
				slab_err(s, page, "Freepointer 0x%p corrupt",
									fp);
Christoph Lameter's avatar
Christoph Lameter committed
724
725
				page->freelist = NULL;
				page->inuse = s->objects;
726
727
728
				printk(KERN_ERR "@@@ SLUB %s: Freelist "
					"cleared. Slab 0x%p\n",
					s->name, page);
Christoph Lameter's avatar
Christoph Lameter committed
729
730
731
732
733
734
735
736
737
738
				return 0;
			}
			break;
		}
		object = fp;
		fp = get_freepointer(s, object);
		nr++;
	}

	if (page->inuse != s->objects - nr) {
739
740
741
		slab_err(s, page, "Wrong object count. Counter is %d but "
			"counted were %d", s, page, page->inuse,
							s->objects - nr);
Christoph Lameter's avatar
Christoph Lameter committed
742
		page->inuse = s->objects - nr;
743
744
		printk(KERN_ERR "@@@ SLUB %s: Object count adjusted. "
			"Slab @0x%p\n", s->name, page);
Christoph Lameter's avatar
Christoph Lameter committed
745
746
747
748
	}
	return search == NULL;
}

Christoph Lameter's avatar
Christoph Lameter committed
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
static void trace(struct kmem_cache *s, struct page *page, void *object, int alloc)
{
	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();
	}
}

765
/*
Christoph Lameter's avatar
Christoph Lameter committed
766
 * Tracking of fully allocated slabs for debugging purposes.
767
 */
Christoph Lameter's avatar
Christoph Lameter committed
768
static void add_full(struct kmem_cache_node *n, struct page *page)
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
{
	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);
}

Christoph Lameter's avatar
Christoph Lameter committed
789
790
791
792
793
794
795
796
797
798
799
800
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
801
802
803
804
805
{
	if (!check_slab(s, page))
		goto bad;

	if (object && !on_freelist(s, page, object)) {
806
807
		slab_err(s, page, "Object 0x%p already allocated", object);
		goto bad;
Christoph Lameter's avatar
Christoph Lameter committed
808
809
810
811
	}

	if (!check_valid_pointer(s, page, object)) {
		object_err(s, page, object, "Freelist Pointer check fails");
812
		goto bad;
Christoph Lameter's avatar
Christoph Lameter committed
813
814
	}

Christoph Lameter's avatar
Christoph Lameter committed
815
	if (object && !check_object(s, page, object, 0))
Christoph Lameter's avatar
Christoph Lameter committed
816
817
		goto bad;

Christoph Lameter's avatar
Christoph Lameter committed
818
819
820
821
822
	/* 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
823
	return 1;
Christoph Lameter's avatar
Christoph Lameter committed
824

Christoph Lameter's avatar
Christoph Lameter committed
825
826
827
828
829
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
830
		 * as used avoids touching the remaining objects.
Christoph Lameter's avatar
Christoph Lameter committed
831
832
833
834
835
836
837
838
839
840
841
		 */
		printk(KERN_ERR "@@@ SLUB: %s slab 0x%p. Marking all objects used.\n",
			s->name, page);
		page->inuse = s->objects;
		page->freelist = NULL;
		/* Fix up fields that may be corrupted */
		page->offset = s->offset / sizeof(void *);
	}
	return 0;
}

Christoph Lameter's avatar
Christoph Lameter committed
842
843
static int free_debug_processing(struct kmem_cache *s, struct page *page,
						void *object, void *addr)
Christoph Lameter's avatar
Christoph Lameter committed
844
845
846
847
848
{
	if (!check_slab(s, page))
		goto fail;

	if (!check_valid_pointer(s, page, object)) {
849
		slab_err(s, page, "Invalid object pointer 0x%p", object);
Christoph Lameter's avatar
Christoph Lameter committed
850
851
852
853
		goto fail;
	}

	if (on_freelist(s, page, object)) {
854
		slab_err(s, page, "Object 0x%p already free", object);
Christoph Lameter's avatar
Christoph Lameter committed
855
856
857
858
859
860
861
862
		goto fail;
	}

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

	if (unlikely(s != page->slab)) {
		if (!PageSlab(page))
863
864
			slab_err(s, page, "Attempt to free object(0x%p) "
				"outside of slab", object);
Christoph Lameter's avatar
Christoph Lameter committed
865
		else
866
		if (!page->slab) {
Christoph Lameter's avatar
Christoph Lameter committed
867
			printk(KERN_ERR
868
				"SLUB <none>: no slab for object 0x%p.\n",
Christoph Lameter's avatar
Christoph Lameter committed
869
						object);
870
871
			dump_stack();
		}
Christoph Lameter's avatar
Christoph Lameter committed
872
		else
873
874
			slab_err(s, page, "object at 0x%p belongs "
				"to slab %s", object, page->slab->name);
Christoph Lameter's avatar
Christoph Lameter committed
875
876
		goto fail;
	}
Christoph Lameter's avatar
Christoph Lameter committed
877
878
879
880
881
882
883
884

	/* Special debug activities for freeing objects */
	if (!SlabFrozen(page) && !page->freelist)
		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
885
	return 1;
Christoph Lameter's avatar
Christoph Lameter committed
886

Christoph Lameter's avatar
Christoph Lameter committed
887
888
889
890
891
892
fail:
	printk(KERN_ERR "@@@ SLUB: %s slab 0x%p object at 0x%p not freed.\n",
		s->name, page, object);
	return 0;
}

893
894
static int __init setup_slub_debug(char *str)
{
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
	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
	 */
	for ( ;*str && *str != ','; str++) {
		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' "
				"unknown. skipped\n",*str);
		}
940
941
	}

942
check_slabs:
943
944
	if (*str == ',')
		slub_debug_slabs = str + 1;
945
out:
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
	return 1;
}

__setup("slub_debug", setup_slub_debug);

static void kmem_cache_open_debug_check(struct kmem_cache *s)
{
	/*
	 * The page->offset field is only 16 bit wide. This is an offset
	 * in units of words from the beginning of an object. If the slab
	 * size is bigger then we cannot move the free pointer behind the
	 * object anymore.
	 *
	 * On 32 bit platforms the limit is 256k. On 64bit platforms
	 * the limit is 512k.
	 *
962
	 * Debugging or ctor may create a need to move the free
963
964
	 * pointer. Fail if this happens.
	 */
965
	if (s->objsize >= 65535 * sizeof(void *)) {
966
967
		BUG_ON(s->flags & (SLAB_RED_ZONE | SLAB_POISON |
				SLAB_STORE_USER | SLAB_DESTROY_BY_RCU));
968
		BUG_ON(s->ctor);
969
970
971
972
973
974
975
976
977
978
979
	}
	else
		/*
		 * Enable debugging if selected on the kernel commandline.
		 */
		if (slub_debug && (!slub_debug_slabs ||
		    strncmp(slub_debug_slabs, s->name,
		    	strlen(slub_debug_slabs)) == 0))
				s->flags |= slub_debug;
}
#else
Christoph Lameter's avatar
Christoph Lameter committed
980
981
static inline void setup_object_debug(struct kmem_cache *s,
			struct page *page, void *object) {}
982

Christoph Lameter's avatar
Christoph Lameter committed
983
984
static inline int alloc_debug_processing(struct kmem_cache *s,
	struct page *page, void *object, void *addr) { return 0; }
985

Christoph Lameter's avatar
Christoph Lameter committed
986
987
static inline int free_debug_processing(struct kmem_cache *s,
	struct page *page, void *object, void *addr) { return 0; }
988
989
990
991
992

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
993
static inline void add_full(struct kmem_cache_node *n, struct page *page) {}
994
995
996
static inline void kmem_cache_open_debug_check(struct kmem_cache *s) {}
#define slub_debug 0
#endif
Christoph Lameter's avatar
Christoph Lameter committed
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
/*
 * Slab allocation and freeing
 */
static struct page *allocate_slab(struct kmem_cache *s, gfp_t flags, int node)
{
	struct page * page;
	int pages = 1 << s->order;

	if (s->order)
		flags |= __GFP_COMP;

	if (s->flags & SLAB_CACHE_DMA)
		flags |= SLUB_DMA;

	if (node == -1)
		page = alloc_pages(flags, s->order);
	else
		page = alloc_pages_node(node, flags, s->order);

	if (!page)
		return NULL;

	mod_zone_page_state(page_zone(page),
		(s->flags & SLAB_RECLAIM_ACCOUNT) ?
		NR_SLAB_RECLAIMABLE : NR_SLAB_UNRECLAIMABLE,
		pages);

	return page;
}

static void setup_object(struct kmem_cache *s, struct page *page,
				void *object)
{
Christoph Lameter's avatar
Christoph Lameter committed
1030
	setup_object_debug(s, page, object);
1031
	if (unlikely(s->ctor))
1032
		s->ctor(object, s, 0);
Christoph Lameter's avatar
Christoph Lameter committed
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
}

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

	BUG_ON(flags & ~(GFP_DMA | GFP_LEVEL_MASK));

	if (flags & __GFP_WAIT)
		local_irq_enable();

	page = allocate_slab(s, flags & GFP_LEVEL_MASK, node);
	if (!page)
		goto out;

	n = get_node(s, page_to_nid(page));
	if (n)
		atomic_long_inc(&n->nr_slabs);
	page->offset = s->offset / sizeof(void *);
	page->slab = s;
	page->flags |= 1 << PG_slab;
	if (s->flags & (SLAB_DEBUG_FREE | SLAB_RED_ZONE | SLAB_POISON |
			SLAB_STORE_USER | SLAB_TRACE))
1061
		SetSlabDebug(page);
Christoph Lameter's avatar
Christoph Lameter committed
1062
1063
1064
1065
1066
1067
1068
1069

	start = page_address(page);
	end = start + s->objects * s->size;

	if (unlikely(s->flags & SLAB_POISON))
		memset(start, POISON_INUSE, PAGE_SIZE << s->order);

	last = start;
1070
	for_each_object(p, s, start) {
Christoph Lameter's avatar
Christoph Lameter committed
1071
1072
1073
1074
1075
1076
1077
1078
		setup_object(s, page, last);
		set_freepointer(s, last, p);
		last = p;
	}
	setup_object(s, page, last);
	set_freepointer(s, last, NULL);

	page->freelist = start;
1079
	page->lockless_freelist = NULL;
Christoph Lameter's avatar
Christoph Lameter committed
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
	page->inuse = 0;
out:
	if (flags & __GFP_WAIT)
		local_irq_disable();
	return page;
}

static void __free_slab(struct kmem_cache *s, struct page *page)
{
	int pages = 1 << s->order;

1091
	if (unlikely(SlabDebug(page))) {
Christoph Lameter's avatar
Christoph Lameter committed
1092
1093
1094
		void *p;

		slab_pad_check(s, page);
1095
		for_each_object(p, s, page_address(page))
Christoph Lameter's avatar
Christoph Lameter committed
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
			check_object(s, page, p, 0);
	}

	mod_zone_page_state(page_zone(page),
		(s->flags & SLAB_RECLAIM_ACCOUNT) ?
		NR_SLAB_RECLAIMABLE : NR_SLAB_UNRECLAIMABLE,
		- pages);

	page->mapping = NULL;
	__free_pages(page, s->order);
}

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)
{
	struct kmem_cache_node *n = get_node(s, page_to_nid(page));

	atomic_long_dec(&n->nr_slabs);
	reset_page_mapcount(page);
1135
1136
	ClearSlabDebug(page);
	__ClearPageSlab(page);
Christoph Lameter's avatar
Christoph Lameter committed
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
	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)
{
	bit_spin_unlock(PG_locked, &page->flags);
}

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
 */
Christoph Lameter's avatar
Christoph Lameter committed
1164
static void add_partial_tail(struct kmem_cache_node *n, struct page *page)
Christoph Lameter's avatar
Christoph Lameter committed
1165
{
Christoph Lameter's avatar
Christoph Lameter committed
1166
1167
1168
1169
1170
	spin_lock(&n->list_lock);
	n->nr_partial++;
	list_add_tail(&page->lru, &n->partial);
	spin_unlock(&n->list_lock);
}
Christoph Lameter's avatar
Christoph Lameter committed
1171

Christoph Lameter's avatar
Christoph Lameter committed
1172
1173
static void add_partial(struct kmem_cache_node *n, struct page *page)
{
Christoph Lameter's avatar
Christoph Lameter committed
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
	spin_lock(&n->list_lock);
	n->nr_partial++;
	list_add(&page->lru, &n->partial);
	spin_unlock(&n->list_lock);
}

static void remove_partial(struct kmem_cache *s,
						struct page *page)
{
	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
1192
 * Lock slab and remove from the partial list.
Christoph Lameter's avatar
Christoph Lameter committed
1193
 *
Christoph Lameter's avatar
Christoph Lameter committed
1194
 * Must hold list_lock.
Christoph Lameter's avatar
Christoph Lameter committed
1195
 */
1196
static inline int lock_and_freeze_slab(struct kmem_cache_node *n, struct page *page)
Christoph Lameter's avatar
Christoph Lameter committed
1197
1198
1199
1200
{
	if (slab_trylock(page)) {
		list_del(&page->lru);
		n->nr_partial--;
1201
		SetSlabFrozen(page);
Christoph Lameter's avatar
Christoph Lameter committed
1202
1203
1204
1205
1206
1207
		return 1;
	}
	return 0;
}

/*
Christoph Lameter's avatar
Christoph Lameter committed
1208
 * Try to allocate a partial slab from a specific node.
Christoph Lameter's avatar
Christoph Lameter committed
1209
1210
1211
1212
1213
1214
1215
1216
 */
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
1217
1218
	 * partial slab and there is none available then get_partials()
	 * will return NULL.
Christoph Lameter's avatar
Christoph Lameter committed
1219
1220
1221
1222
1223
1224
	 */
	if (!n || !n->nr_partial)
		return NULL;

	spin_lock(&n->list_lock);
	list_for_each_entry(page, &n->partial, lru)
1225
		if (lock_and_freeze_slab(n, page))
Christoph Lameter's avatar
Christoph Lameter committed
1226
1227
1228
1229
1230
1231
1232
1233
			goto out;
	page = NULL;
out:
	spin_unlock(&n->list_lock);
	return page;
}

/*
Christoph Lameter's avatar
Christoph Lameter committed
1234
 * Get a page from somewhere. Search in increasing NUMA distances.
Christoph Lameter's avatar
Christoph Lameter committed
1235
1236
1237
1238
1239
1240
1241
1242
1243
 */
static struct page *get_any_partial(struct kmem_cache *s, gfp_t flags)
{
#ifdef CONFIG_NUMA
	struct zonelist *zonelist;
	struct zone **z;
	struct page *page;

	/*
Christoph Lameter's avatar
Christoph Lameter committed
1244
1245
1246
1247
	 * 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
1248
	 *
Christoph Lameter's avatar
Christoph Lameter committed
1249
1250
1251
1252
	 * 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
1253
1254
	 *
	 * If /sys/slab/xx/defrag_ratio is set to 100 (which makes
Christoph Lameter's avatar
Christoph Lameter committed
1255
1256
1257
1258
1259
	 * 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
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
	 */
	if (!s->defrag_ratio || get_cycles() % 1024 > s->defrag_ratio)
		return NULL;

	zonelist = &NODE_DATA(slab_node(current->mempolicy))
					->node_zonelists[gfp_zone(flags)];
	for (z = zonelist->zones; *z; z++) {
		struct kmem_cache_node *n;

		n = get_node(s, zone_to_nid(*z));

		if (n && cpuset_zone_allowed_hardwall(*z, flags) &&
Christoph Lameter's avatar
Christoph Lameter committed
1272
				n->nr_partial > MIN_PARTIAL) {
Christoph Lameter's avatar
Christoph Lameter committed
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
			page = get_partial_node(n);
			if (page)
				return page;
		}
	}
#endif
	return NULL;
}

/*
 * Get a partial page, lock it and return it.
 */
static struct page *get_partial(struct kmem_cache *s, gfp_t flags, int node)
{
	struct page *page;
	int searchnode = (node == -1) ? numa_node_id() : node;

	page = get_partial_node(get_node(s, searchnode));
	if (page || (flags & __GFP_THISNODE))
		return page;

	return get_any_partial(s, flags);
}

/*
 * Move a page back to the lists.
 *
 * Must be called with the slab lock held.
 *
 * On exit the slab lock will have been dropped.
 */
1304
static void unfreeze_slab(struct kmem_cache *s, struct page *page)
Christoph Lameter's avatar
Christoph Lameter committed
1305
{
Christoph Lameter's avatar
Christoph Lameter committed
1306
1307
	struct kmem_cache_node *n = get_node(s, page_to_nid(page));

1308
	ClearSlabFrozen(page);
Christoph Lameter's avatar
Christoph Lameter committed
1309
	if (page->inuse) {
Christoph Lameter's avatar
Christoph Lameter committed
1310

Christoph Lameter's avatar
Christoph Lameter committed
1311
		if (page->freelist)
Christoph Lameter's avatar
Christoph Lameter committed
1312
			add_partial(n, page);
1313
		else if (SlabDebug(page) && (s->flags & SLAB_STORE_USER))
Christoph Lameter's avatar
Christoph Lameter committed
1314
			add_full(n, page);
Christoph Lameter's avatar
Christoph Lameter committed
1315
		slab_unlock(page);
Christoph Lameter's avatar
Christoph Lameter committed
1316

Christoph Lameter's avatar
Christoph Lameter committed
1317
	} else {
Christoph Lameter's avatar
Christoph Lameter committed
1318
1319
		if (n->nr_partial < MIN_PARTIAL) {
			/*
Christoph Lameter's avatar
Christoph Lameter committed
1320
1321
1322
1323
1324
1325
			 * Adding an empty slab to the partial slabs in order
			 * to avoid page allocator overhead. This slab needs
			 * to come after the other slabs with objects in
			 * order to fill them up. That way the size of the
			 * partial list stays small. kmem_cache_shrink can
			 * reclaim empty slabs from the partial list.
Christoph Lameter's avatar
Christoph Lameter committed
1326
1327
1328
1329
1330
1331
1332
			 */
			add_partial_tail(n, page);
			slab_unlock(page);
		} else {
			slab_unlock(page);
			discard_slab(s, page);
		}
Christoph Lameter's avatar
Christoph Lameter committed
1333
1334
1335
1336
1337
1338
1339
1340
	}
}

/*
 * Remove the cpu slab
 */
static void deactivate_slab(struct kmem_cache *s, struct page *page, int cpu)
{
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
	/*
	 * Merge cpu freelist into freelist. Typically we get here
	 * because both freelists are empty. So this is unlikely
	 * to occur.
	 */
	while (unlikely(page->lockless_freelist)) {
		void **object;

		/* Retrieve object from cpu_freelist */
		object = page->lockless_freelist;
		page->lockless_freelist = page->lockless_freelist[page->offset];

		/* And put onto the regular freelist */
		object[page->offset] = page->freelist;
		page->freelist = object;
		page->inuse--;
	}
Christoph Lameter's avatar
Christoph Lameter committed
1358
	s->cpu_slab[cpu] = NULL;
1359
	unfreeze_slab(s, page);
Christoph Lameter's avatar
Christoph Lameter committed
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
}

static void flush_slab(struct kmem_cache *s, struct page *page, int cpu)
{
	slab_lock(page);
	deactivate_slab(s, page, cpu);
}

/*
 * Flush cpu slab.
 * Called from IPI handler with interrupts disabled.
 */
static void __flush_cpu_slab(struct kmem_cache *s, int cpu)
{
	struct page *page = s->cpu_slab[cpu];

	if (likely(page))
		flush_slab(s, page, cpu);
}

static void flush_cpu_slab(void *d)
{
	struct kmem_cache *s = d;
	int cpu = smp_processor_id();

	__flush_cpu_slab(s, cpu);
}

static void flush_all(struct kmem_cache *s)
{
#ifdef CONFIG_SMP
	on_each_cpu(flush_cpu_slab, s, 1, 1);
#else
	unsigned long flags;

	local_irq_save(flags);
	flush_cpu_slab(s);
	local_irq_restore(flags);
#endif
}

/*
1402
1403
1404
1405
 * Slow path. The lockless freelist is empty or we need to perform
 * debugging duties.
 *
 * Interrupts are disabled.
Christoph Lameter's avatar
Christoph Lameter committed
1406
 *
1407
1408
1409
 * Processing is still very fast if new objects have been freed to the
 * regular freelist. In that case we simply take over the regular freelist
 * as the lockless freelist and zap the regular freelist.
Christoph Lameter's avatar
Christoph Lameter committed
1410
 *
1411
1412
1413
 * If that is not working then we fall back to the partial lists. We take the
 * first element of the freelist as the object to allocate now and move the
 * rest of the freelist to the lockless freelist.
Christoph Lameter's avatar
Christoph Lameter committed
1414
 *
1415
1416
 * And if we were unable to get a new slab from the partial slab lists then
 * we need to allocate a new slab. This is slowest path since we may sleep.
Christoph Lameter's avatar
Christoph Lameter committed
1417
 */
1418
1419
static void *__slab_alloc(struct kmem_cache *s,
		gfp_t gfpflags, int node, void *addr, struct page *page)
Christoph Lameter's avatar
Christoph Lameter committed
1420
1421
{
	void **object;
1422
	int cpu = smp_processor_id();
Christoph Lameter's avatar
Christoph Lameter committed
1423
1424
1425
1426
1427
1428
1429

	if (!page)
		goto new_slab;

	slab_lock(page);
	if (unlikely(node != -1 && page_to_nid(page) != node))
		goto another_slab;
1430
load_freelist:
Christoph Lameter's avatar
Christoph Lameter committed
1431
1432
1433
	object = page->freelist;
	if (unlikely(!object))
		goto another_slab;
1434
	if (unlikely(SlabDebug(page)))
Christoph Lameter's avatar
Christoph Lameter committed
1435
1436
		goto debug;

1437
1438
1439
1440
	object = page->freelist;
	page->lockless_freelist = object[page->offset];
	page->inuse = s->objects;
	page->freelist = NULL;
Christoph Lameter's avatar
Christoph Lameter committed
1441
1442
1443
1444
1445
1446
1447
1448
	slab_unlock(page);
	return object;

another_slab:
	deactivate_slab(s, page, cpu);

new_slab:
	page = get_partial(s, gfpflags, node);
1449
	if (page) {
Christoph Lameter's avatar
Christoph Lameter committed
1450
		s->cpu_slab[cpu] = page;
1451
		goto load_freelist;
Christoph Lameter's avatar
Christoph Lameter committed
1452
1453
1454
1455
1456
1457
1458
	}

	page = new_slab(s, gfpflags, node);
	if (page) {
		cpu = smp_processor_id();
		if (s->cpu_slab[cpu]) {
			/*
Christoph Lameter's avatar
Christoph Lameter committed
1459
1460
1461
1462
1463
			 * Someone else populated the cpu_slab while we
			 * enabled interrupts, or we have gotten scheduled
			 * on another cpu. The page may not be on the
			 * requested node even if __GFP_THISNODE was
			 * specified. So we need to recheck.
Christoph Lameter's avatar
Christoph Lameter committed
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
			 */
			if (node == -1 ||
				page_to_nid(s->cpu_slab[cpu]) == node) {
				/*
				 * Current cpuslab is acceptable and we
				 * want the current one since its cache hot
				 */
				discard_slab(s, page);
				page = s->cpu_slab[cpu];
				slab_lock(page);
1474
				goto load_freelist;
Christoph Lameter's avatar
Christoph Lameter committed
1475
			}
Christoph Lameter's avatar
Christoph Lameter committed
1476
			/* New slab does not fit our expectations */
Christoph Lameter's avatar
Christoph Lameter committed
1477
1478
1479
			flush_slab(s, s->cpu_slab[cpu], cpu);
		}
		slab_lock(page);
1480
1481
1482
		SetSlabFrozen(page);
		s->cpu_slab[cpu] = page;
		goto load_freelist;
Christoph Lameter's avatar
Christoph Lameter committed
1483
1484
1485
	}
	return NULL;
debug:
1486
	object = page->freelist;
Christoph Lameter's avatar
Christoph Lameter committed
1487
	if (!alloc_debug_processing(s, page, object, addr))
Christoph Lameter's avatar
Christoph Lameter committed
1488
		goto another_slab;
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525

	page->inuse++;
	page->freelist = object[page->offset];
	slab_unlock(page);
	return object;
}

/*
 * Inlined fastpath so that allocation functions (kmalloc, kmem_cache_alloc)
 * have the fastpath folded into their functions. So no function call
 * overhead for requests that can be satisfied on the fastpath.
 *
 * The fastpath works by first checking if the lockless freelist can be used.
 * If not then __slab_alloc is called for slow processing.
 *
 * Otherwise we can simply pick the next object from the lockless free list.
 */
static void __always_inline *slab_alloc(struct kmem_cache *s,
				gfp_t gfpflags, int node, void *addr)
{
	struct page *page;
	void **object;
	unsigned long flags;

	local_irq_save(flags);
	page = s->cpu_slab[smp_processor_id()];
	if (unlikely(!page || !page->lockless_freelist ||
			(node != -1 && page_to_nid(page) != node)))

		object = __slab_alloc(s, gfpflags, node, addr, page);

	else {
		object = page->lockless_freelist;
		page->lockless_freelist = object[page->offset];
	}
	local_irq_restore(flags);
	return object;
Christoph Lameter's avatar
Christoph Lameter committed
1526
1527
1528
1529
}

void *kmem_cache_alloc(struct kmem_cache *s, gfp_t gfpflags)
{
Christoph Lameter's avatar
Christoph Lameter committed
1530
	return slab_alloc(s, gfpflags, -1, __builtin_return_address(0));
Christoph Lameter's avatar
Christoph Lameter committed
1531
1532
1533
1534
1535
1536
}
EXPORT_SYMBOL(kmem_cache_alloc);

#ifdef CONFIG_NUMA
void *kmem_cache_alloc_node(struct kmem_cache *s, gfp_t gfpflags, int node)
{
Christoph Lameter's avatar
Christoph Lameter committed
1537
	return slab_alloc(s, gfpflags, node, __builtin_return_address(0));
Christoph Lameter's avatar
Christoph Lameter committed
1538
1539
1540
1541
1542
}
EXPORT_SYMBOL(kmem_cache_alloc_node);
#endif

/*
1543
1544
 * Slow patch handling. This may still be called frequently since objects
 * have a longer lifetime than the cpu slabs in most processing loads.
Christoph Lameter's avatar
Christoph Lameter committed
1545
 *
1546
1547
1548
 * So we still attempt to reduce cache line usage. Just take the slab
 * lock and free the item. If there is no additional partial page
 * handling required then we can return immediately.
Christoph Lameter's avatar
Christoph Lameter committed
1549
 */
1550
static void __slab_free(struct kmem_cache *s, struct page *page,
Christoph Lameter's avatar
Christoph Lameter committed
1551
					void *x, void *addr)
Christoph Lameter's avatar
Christoph Lameter committed
1552
1553
1554
1555
1556
1557
{
	void *prior;
	void **object = (void *)x;

	slab_lock(page);

1558
	if (unlikely(SlabDebug(page)))
Christoph Lameter's avatar
Christoph Lameter committed
1559
1560
1561
1562
1563
1564
		goto debug;
checks_ok:
	prior = object[page->offset] = page->freelist;
	page->freelist = object;
	page->inuse--;

1565
	if (unlikely(SlabFrozen(page)))
Christoph Lameter's avatar
Christoph Lameter committed
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
		goto out_unlock;

	if (unlikely(!page->inuse))
		goto slab_empty;

	/*
	 * Objects left in the slab. If it
	 * was not on the partial list before
	 * then add it.
	 */
	if (unlikely(!prior))
Christoph Lameter's avatar
Christoph Lameter committed
1577
		add_partial(get_node(s, page_to_nid(page)), page);
Christoph Lameter's avatar
Christoph Lameter committed
1578
1579
1580
1581
1582
1583
1584
1585

out_unlock:
	slab_unlock(page);
	return;

slab_empty:
	if (prior)
		/*
Christoph Lameter's avatar
Christoph Lameter committed
1586
		 * Slab still on the partial list.
Christoph Lameter's avatar
Christoph Lameter committed
1587
1588
1589
1590
1591
1592
1593
1594
		 */
		remove_partial(s, page);

	slab_unlock(page);
	discard_slab(s, page);
	return;

debug:
Christoph Lameter's avatar
Christoph Lameter committed
1595
	if (!free_debug_processing(s, page, x, addr))
Christoph Lameter's avatar
Christoph Lameter committed
1596
1597
		goto out_unlock;
	goto checks_ok;
Christoph Lameter's avatar
Christoph Lameter committed
1598
1599
}

1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
/*
 * Fastpath with forced inlining to produce a kfree and kmem_cache_free that
 * can perform fastpath freeing without additional function calls.
 *
 * The fastpath is only possible if we are freeing to the current cpu slab
 * of this processor. This typically the case if we have just allocated
 * the item before.
 *
 * If fastpath is not possible then fall back to __slab_free where we deal
 * with all sorts of special processing.
 */
static void __always_inline slab_free(struct kmem_cache *s,
			struct page *page, void *x, void *addr)
{
	void **object = (void *)x;
	unsigned long flags;

	local_irq_save(flags);
	if (likely(page == s->cpu_slab[smp_processor_id()] &&
						!SlabDebug(page))) {
		object[page->offset] = page->lockless_freelist;
		page->lockless_freelist = object;
	} else
		__slab_free(s, page, x, addr);

	local_irq_restore(flags);
}

Christoph Lameter's avatar
Christoph Lameter committed
1628
1629
void kmem_cache_free(struct kmem_cache *s, void *x)
{
Christoph Lameter's avatar
Christoph Lameter committed
1630
	struct page *page;
Christoph Lameter's avatar
Christoph Lameter committed
1631

1632
	page = virt_to_head_page(x);
Christoph Lameter's avatar
Christoph Lameter committed
1633

Christoph Lameter's avatar
Christoph Lameter committed
1634
	slab_free(s, page, x, __builtin_return_address(0));
Christoph Lameter's avatar
Christoph Lameter committed
1635
1636
1637
1638
1639
1640
}
EXPORT_SYMBOL(kmem_cache_free);

/* Figure out on which slab object the object resides */
static struct page *get_object_page(const void *x)
{
1641
	struct page *page = virt_to_head_page(x);
Christoph Lameter's avatar
Christoph Lameter committed
1642
1643
1644
1645
1646
1647
1648
1649

	if (!PageSlab(page))
		return NULL;

	return page;
}

/*
Christoph Lameter's avatar
Christoph Lameter committed
1650
1651
1652
1653
 * Object placement in a slab is made very easy because we always start at
 * offset 0. If we tune the size of the object to the alignment then we can
 * get the required alignment by putting one properly sized object after
 * another.
Christoph Lameter's avatar
Christoph Lameter committed
1654
1655
1656
1657
 *
 * Notice that the allocation order determines the sizes of the per cpu
 * caches. Each processor has always one slab available for allocations.
 * Increasing the allocation order reduces the number of times that slabs
Christoph Lameter's avatar
Christoph Lameter committed
1658
 * must be moved on and off the partial lists and is therefore a factor in
Christoph Lameter's avatar
Christoph Lameter committed
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
 * locking overhead.
 */

/*
 * Mininum / Maximum order of slab pages. This influences locking overhead
 * and slab fragmentation. A higher order reduces the number of partial slabs
 * and increases the number of allocations possible without having to
 * take the list_lock.
 */
static int slub_min_order;
static int slub_max_order = DEFAULT_MAX_ORDER;
static int slub_min_objects = DEFAULT_MIN_OBJECTS;

/*
 * Merge control. If this is set then no merging of slab caches will occur.
Christoph Lameter's avatar
Christoph Lameter committed
1674
 * (Could be removed. This was introduced to pacify the merge skeptics.)
Christoph Lameter's avatar
Christoph Lameter committed
1675
1676
1677
1678
1679
1680
 */
static int slub_nomerge;

/*
 * Calculate the order of allocation given an slab object size.
 *
Christoph Lameter's avatar
Christoph Lameter committed
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
 * The order of allocation has significant impact on performance and other
 * system components. Generally order 0 allocations should be preferred since
 * order 0 does not cause fragmentation in the page allocator. Larger objects
 * be problematic to put into order 0 slabs because there may be too much
 * unused space left. We go to a higher order if more than 1/8th of the slab
 * would be wasted.
 *
 * In order to reach satisfactory performance we must ensure that a minimum
 * number of objects is in one slab. Otherwise we may generate too much
 * activity on the partial lists which requires taking the list_lock. This is
 * less a concern for large slabs though which are rarely used.
Christoph Lameter's avatar
Christoph Lameter committed
1692
 *
Christoph Lameter's avatar
Christoph Lameter committed
1693
1694
1695
1696
 * slub_max_order specifies the order where we begin to stop considering the
 * number of objects in a slab as critical. If we reach slub_max_order then
 * we try to keep the page order as low as possible. So we accept more waste
 * of space in favor of a small page order.
Christoph Lameter's avatar
Christoph Lameter committed
1697
 *
Christoph Lameter's avatar
Christoph Lameter committed
1698
1699
1700
1701
 * Higher order allocations also allow the placement of more objects in a
 * slab and thereby reduce object handling overhead. If the user has
 * requested a higher mininum order then we start with that one instead of
 * the smallest order which will fit the object.
Christoph Lameter's avatar
Christoph Lameter committed
1702
 */
1703
1704
static inline int slab_order(int size, int min_objects,
				int max_order, int fract_leftover)
Christoph Lameter's avatar
Christoph Lameter committed
1705
1706
1707
1708
{
	int order;
	int rem;

1709
1710
1711
	for (order = max(slub_min_order,
				fls(min_objects * size - 1) - PAGE_SHIFT);
			order <= max_order; order++) {
Christoph Lameter's avatar
Christoph Lameter committed
1712

1713
		unsigned long slab_size = PAGE_SIZE << order;
Christoph Lameter's avatar
Christoph Lameter committed
1714

1715
		if (slab_size < min_objects * size)
Christoph Lameter's avatar
Christoph Lameter committed
1716
1717
1718
1719
			continue;

		rem = slab_size % size;

1720
		if (rem <= slab_size / fract_leftover)
Christoph Lameter's avatar
Christoph Lameter committed
1721
1722
1723
			break;

	}
Christoph Lameter's avatar
Christoph Lameter committed
1724

Christoph Lameter's avatar
Christoph Lameter committed
1725
1726
1727
	return order;
}

1728
1729
1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766