slub.c 102 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
/*
 * 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>
23
#include <linux/memory.h>
Christoph Lameter's avatar
Christoph Lameter committed
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
69

/*
 * 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
70
71
 * 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
72
 * freed then the slab will show up again on the partial lists.
Christoph Lameter's avatar
Christoph Lameter committed
73
74
 * We track full slabs for debugging purposes though because otherwise we
 * cannot scan all objects.
Christoph Lameter's avatar
Christoph Lameter committed
75
76
77
78
79
80
81
 *
 * 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.
 *
82
83
84
85
86
87
88
89
90
91
92
93
 * 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
94
 * 			freelist that allows lockless access to
95
96
 * 			free objects in addition to the regular freelist
 * 			that requires the slab lock.
Christoph Lameter's avatar
Christoph Lameter committed
97
98
99
 *
 * PageError		Slab requires special handling due to debug
 * 			options set. This moves	slab handling out of
100
 * 			the fast path and disables lockless freelists.
Christoph Lameter's avatar
Christoph Lameter committed
101
102
 */

103
104
105
106
107
108
109
110
#define FROZEN (1 << PG_active)

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

111
112
static inline int SlabFrozen(struct page *page)
{
113
	return page->flags & FROZEN;
114
115
116
117
}

static inline void SetSlabFrozen(struct page *page)
{
118
	page->flags |= FROZEN;
119
120
121
122
}

static inline void ClearSlabFrozen(struct page *page)
{
123
	page->flags &= ~FROZEN;
124
125
}

126
127
static inline int SlabDebug(struct page *page)
{
128
	return page->flags & SLABDEBUG;
129
130
131
132
}

static inline void SetSlabDebug(struct page *page)
{
133
	page->flags |= SLABDEBUG;
134
135
136
137
}

static inline void ClearSlabDebug(struct page *page)
{
138
	page->flags &= ~SLABDEBUG;
139
140
}

Christoph Lameter's avatar
Christoph Lameter committed
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
/*
 * 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

#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

171
172
173
174
/*
 * Mininum number of partial slabs. These will be left on the partial
 * lists even if they are empty. kmem_cache_shrink may reclaim them.
 */
175
#define MIN_PARTIAL 5
Christoph Lameter's avatar
Christoph Lameter committed
176

177
178
179
180
181
182
183
/*
 * 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
184
185
#define DEBUG_DEFAULT_FLAGS (SLAB_DEBUG_FREE | SLAB_RED_ZONE | \
				SLAB_POISON | SLAB_STORE_USER)
Christoph Lameter's avatar
Christoph Lameter committed
186

Christoph Lameter's avatar
Christoph Lameter committed
187
188
189
190
191
192
193
194
195
196
/*
 * 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
197
#define ARCH_KMALLOC_MINALIGN __alignof__(unsigned long long)
Christoph Lameter's avatar
Christoph Lameter committed
198
199
200
#endif

#ifndef ARCH_SLAB_MINALIGN
201
#define ARCH_SLAB_MINALIGN __alignof__(unsigned long long)
Christoph Lameter's avatar
Christoph Lameter committed
202
203
204
#endif

/* Internal SLUB flags */
205
206
#define __OBJECT_POISON		0x80000000 /* Poison object */
#define __SYSFS_ADD_DEFERRED	0x40000000 /* Not yet visible via sysfs */
207
208
#define __KMALLOC_CACHE		0x20000000 /* objects freed using kfree */
#define __PAGE_ALLOC_FALLBACK	0x10000000 /* Allow fallback to page alloc */
Christoph Lameter's avatar
Christoph Lameter committed
209

210
211
212
213
214
/* 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
215
216
217
218
219
220
221
222
223
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
224
	UP,		/* Everything works but does not show up in sysfs */
Christoph Lameter's avatar
Christoph Lameter committed
225
226
227
228
229
	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
230
static LIST_HEAD(slab_caches);
Christoph Lameter's avatar
Christoph Lameter committed
231

232
233
234
235
236
237
238
239
240
241
242
243
/*
 * 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 };

244
#if defined(CONFIG_SYSFS) && defined(CONFIG_SLUB_DEBUG)
Christoph Lameter's avatar
Christoph Lameter committed
245
246
247
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 *);
248

Christoph Lameter's avatar
Christoph Lameter committed
249
#else
250
251
252
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; }
253
254
255
256
static inline void sysfs_slab_remove(struct kmem_cache *s)
{
	kfree(s);
}
257

Christoph Lameter's avatar
Christoph Lameter committed
258
259
#endif

260
261
262
263
264
265
266
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
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
/********************************************************************
 * 			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
}

285
286
static inline struct kmem_cache_cpu *get_cpu_slab(struct kmem_cache *s, int cpu)
{
287
288
289
290
291
#ifdef CONFIG_SMP
	return s->cpu_slab[cpu];
#else
	return &s->cpu_slab;
#endif
292
293
}

294
295
296
297
298
299
300
301
302
303
304
305
/*
 * The end pointer in a slab is special. It points to the first object in the
 * slab but has bit 0 set to mark it.
 *
 * Note that SLUB relies on page_mapping returning NULL for pages with bit 0
 * in the mapping set.
 */
static inline int is_end(void *addr)
{
	return (unsigned long)addr & PAGE_MAPPING_ANON;
}

306
static void *slab_address(struct page *page)
307
308
309
310
{
	return page->end - PAGE_MAPPING_ANON;
}

311
312
313
314
315
static inline int check_valid_pointer(struct kmem_cache *s,
				struct page *page, const void *object)
{
	void *base;

316
	if (object == page->end)
317
318
		return 1;

319
	base = slab_address(page);
320
321
322
323
324
325
326
327
	if (object < base || object >= base + s->objects * s->size ||
		(object - base) % s->size) {
		return 0;
	}

	return 1;
}

328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
/*
 * 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) \
352
353
	for (__p = (__free); (__p) != page->end; __p = get_freepointer((__s),\
		__p))
354
355
356
357
358
359
360

/* 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;
}

361
362
363
364
#ifdef CONFIG_SLUB_DEBUG
/*
 * Debug settings:
 */
365
366
367
#ifdef CONFIG_SLUB_DEBUG_ON
static int slub_debug = DEBUG_DEFAULT_FLAGS;
#else
368
static int slub_debug;
369
#endif
370
371
372

static char *slub_debug_slabs;

Christoph Lameter's avatar
Christoph Lameter committed
373
374
375
376
377
378
379
380
381
382
383
384
385
/*
 * 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) {
386
			printk(KERN_ERR "%8s 0x%p: ", text, addr + i);
Christoph Lameter's avatar
Christoph Lameter committed
387
388
			newline = 0;
		}
389
		printk(KERN_CONT " %02x", addr[i]);
Christoph Lameter's avatar
Christoph Lameter committed
390
391
392
		offset = i % 16;
		ascii[offset] = isgraph(addr[i]) ? addr[i] : '.';
		if (offset == 15) {
393
			printk(KERN_CONT " %s\n", ascii);
Christoph Lameter's avatar
Christoph Lameter committed
394
395
396
397
398
399
			newline = 1;
		}
	}
	if (!newline) {
		i %= 16;
		while (i < 16) {
400
			printk(KERN_CONT "   ");
Christoph Lameter's avatar
Christoph Lameter committed
401
402
403
			ascii[i] = ' ';
			i++;
		}
404
		printk(KERN_CONT " %s\n", ascii);
Christoph Lameter's avatar
Christoph Lameter committed
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
	}
}

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)
{
443
444
445
446
447
	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
448
449
450
451
452
453
454
}

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

455
	printk(KERN_ERR "INFO: %s in ", s);
Christoph Lameter's avatar
Christoph Lameter committed
456
	__print_symbol("%s", (unsigned long)t->addr);
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
	printk(" age=%lu cpu=%u pid=%d\n", jiffies - t->when, t->cpu, t->pid);
}

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)
{
	printk(KERN_ERR "INFO: Slab 0x%p used=%u fp=0x%p flags=0x%04lx\n",
		page, page->inuse, page->freelist, page->flags);

}

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
489
490
}

491
492
493
494
495
496
497
498
499
500
501
502
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
503
504
{
	unsigned int off;	/* Offset of last byte */
505
	u8 *addr = slab_address(page);
506
507
508
509
510
511
512
513
514
515
516
517

	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
518
519
520
521
522
523
524
525
526
527

	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;

528
	if (s->flags & SLAB_STORE_USER)
Christoph Lameter's avatar
Christoph Lameter committed
529
530
531
532
		off += 2 * sizeof(struct track);

	if (off != s->size)
		/* Beginning of the filler is the free pointer */
533
534
535
		print_section("Padding", p + off, s->size - off);

	dump_stack();
Christoph Lameter's avatar
Christoph Lameter committed
536
537
538
539
540
}

static void object_err(struct kmem_cache *s, struct page *page,
			u8 *object, char *reason)
{
541
542
	slab_bug(s, reason);
	print_trailer(s, page, object);
Christoph Lameter's avatar
Christoph Lameter committed
543
544
}

545
static void slab_err(struct kmem_cache *s, struct page *page, char *fmt, ...)
Christoph Lameter's avatar
Christoph Lameter committed
546
547
548
549
{
	va_list args;
	char buf[100];

550
551
	va_start(args, fmt);
	vsnprintf(buf, sizeof(buf), fmt, args);
Christoph Lameter's avatar
Christoph Lameter committed
552
	va_end(args);
553
554
	slab_bug(s, fmt);
	print_page_info(page);
Christoph Lameter's avatar
Christoph Lameter committed
555
556
557
558
559
560
561
562
563
	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);
564
		p[s->objsize - 1] = POISON_END;
Christoph Lameter's avatar
Christoph Lameter committed
565
566
567
568
569
570
571
572
	}

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

573
static u8 *check_bytes(u8 *start, unsigned int value, unsigned int bytes)
Christoph Lameter's avatar
Christoph Lameter committed
574
575
576
{
	while (bytes) {
		if (*start != (u8)value)
577
			return start;
Christoph Lameter's avatar
Christoph Lameter committed
578
579
580
		start++;
		bytes--;
	}
581
582
583
584
585
586
587
588
589
590
591
592
	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,
593
			u8 *start, unsigned int value, unsigned int bytes)
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
{
	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
613
614
615
616
617
618
619
620
621
}

/*
 * 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
622
 *
Christoph Lameter's avatar
Christoph Lameter committed
623
624
625
626
627
 * 	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
628
629
630
 * 	Padding is extended by another word if Redzoning is enabled and
 * 	objsize == inuse.
 *
Christoph Lameter's avatar
Christoph Lameter committed
631
632
633
634
 * 	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
635
636
 * 	Meta data starts here.
 *
Christoph Lameter's avatar
Christoph Lameter committed
637
638
 * 	A. Free pointer (if we cannot overwrite object on free)
 * 	B. Tracking data for SLAB_STORE_USER
Christoph Lameter's avatar
Christoph Lameter committed
639
640
641
642
643
 * 	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
644
645
 *
 * object + s->size
Christoph Lameter's avatar
Christoph Lameter committed
646
 * 	Nothing is used beyond s->size.
Christoph Lameter's avatar
Christoph Lameter committed
647
 *
Christoph Lameter's avatar
Christoph Lameter committed
648
649
 * 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
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
 * 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;

668
669
	return check_bytes_and_report(s, page, p, "Object padding",
				p + off, POISON_INUSE, s->size - off);
Christoph Lameter's avatar
Christoph Lameter committed
670
671
672
673
}

static int slab_pad_check(struct kmem_cache *s, struct page *page)
{
674
675
676
677
678
	u8 *start;
	u8 *fault;
	u8 *end;
	int length;
	int remainder;
Christoph Lameter's avatar
Christoph Lameter committed
679
680
681
682

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

683
	start = slab_address(page);
684
	end = start + (PAGE_SIZE << s->order);
Christoph Lameter's avatar
Christoph Lameter committed
685
	length = s->objects * s->size;
686
	remainder = end - (start + length);
Christoph Lameter's avatar
Christoph Lameter committed
687
688
689
	if (!remainder)
		return 1;

690
691
692
693
694
695
696
697
698
699
700
	fault = check_bytes(start + length, POISON_INUSE, remainder);
	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);
	print_section("Padding", start, length);

	restore_bytes(s, "slab padding", POISON_INUSE, start, end);
	return 0;
Christoph Lameter's avatar
Christoph Lameter committed
701
702
703
704
705
706
707
708
709
710
711
712
}

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;

713
714
		if (!check_bytes_and_report(s, page, object, "Redzone",
			endobject, red, s->inuse - s->objsize))
Christoph Lameter's avatar
Christoph Lameter committed
715
716
			return 0;
	} else {
Ingo Molnar's avatar
Ingo Molnar committed
717
718
719
720
		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
721
722
723
724
	}

	if (s->flags & SLAB_POISON) {
		if (!active && (s->flags & __OBJECT_POISON) &&
725
726
727
			(!check_bytes_and_report(s, page, p, "Poison", p,
					POISON_FREE, s->objsize - 1) ||
			 !check_bytes_and_report(s, page, p, "Poison",
728
				p + s->objsize - 1, POISON_END, 1)))
Christoph Lameter's avatar
Christoph Lameter committed
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
			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
749
		 * another error because the object count is now wrong.
Christoph Lameter's avatar
Christoph Lameter committed
750
		 */
751
		set_freepointer(s, p, page->end);
Christoph Lameter's avatar
Christoph Lameter committed
752
753
754
755
756
757
758
759
760
761
		return 0;
	}
	return 1;
}

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

	if (!PageSlab(page)) {
762
		slab_err(s, page, "Not a valid slab page");
Christoph Lameter's avatar
Christoph Lameter committed
763
764
765
		return 0;
	}
	if (page->inuse > s->objects) {
766
767
		slab_err(s, page, "inuse %u > max %u",
			s->name, page->inuse, s->objects);
Christoph Lameter's avatar
Christoph Lameter committed
768
769
770
771
772
773
774
775
		return 0;
	}
	/* Slab_pad_check fixes things up after itself */
	slab_pad_check(s, page);
	return 1;
}

/*
Christoph Lameter's avatar
Christoph Lameter committed
776
777
 * 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
778
779
780
781
782
783
784
 */
static int on_freelist(struct kmem_cache *s, struct page *page, void *search)
{
	int nr = 0;
	void *fp = page->freelist;
	void *object = NULL;

785
	while (fp != page->end && nr <= s->objects) {
Christoph Lameter's avatar
Christoph Lameter committed
786
787
788
789
790
791
		if (fp == search)
			return 1;
		if (!check_valid_pointer(s, page, fp)) {
			if (object) {
				object_err(s, page, object,
					"Freechain corrupt");
792
				set_freepointer(s, object, page->end);
Christoph Lameter's avatar
Christoph Lameter committed
793
794
				break;
			} else {
795
				slab_err(s, page, "Freepointer corrupt");
796
				page->freelist = page->end;
Christoph Lameter's avatar
Christoph Lameter committed
797
				page->inuse = s->objects;
798
				slab_fix(s, "Freelist cleared");
Christoph Lameter's avatar
Christoph Lameter committed
799
800
801
802
803
804
805
806
807
808
				return 0;
			}
			break;
		}
		object = fp;
		fp = get_freepointer(s, object);
		nr++;
	}

	if (page->inuse != s->objects - nr) {
809
		slab_err(s, page, "Wrong object count. Counter is %d but "
810
			"counted were %d", page->inuse, s->objects - nr);
Christoph Lameter's avatar
Christoph Lameter committed
811
		page->inuse = s->objects - nr;
812
		slab_fix(s, "Object count adjusted.");
Christoph Lameter's avatar
Christoph Lameter committed
813
814
815
816
	}
	return search == NULL;
}

Christoph Lameter's avatar
Christoph Lameter committed
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
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();
	}
}

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);
}

Christoph Lameter's avatar
Christoph Lameter committed
857
858
859
860
861
862
863
864
865
866
867
868
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
869
870
871
872
873
{
	if (!check_slab(s, page))
		goto bad;

	if (object && !on_freelist(s, page, object)) {
874
		object_err(s, page, object, "Object already allocated");
875
		goto bad;
Christoph Lameter's avatar
Christoph Lameter committed
876
877
878
879
	}

	if (!check_valid_pointer(s, page, object)) {
		object_err(s, page, object, "Freelist Pointer check fails");
880
		goto bad;
Christoph Lameter's avatar
Christoph Lameter committed
881
882
	}

Christoph Lameter's avatar
Christoph Lameter committed
883
	if (object && !check_object(s, page, object, 0))
Christoph Lameter's avatar
Christoph Lameter committed
884
885
		goto bad;

Christoph Lameter's avatar
Christoph Lameter committed
886
887
888
889
890
	/* 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
891
	return 1;
Christoph Lameter's avatar
Christoph Lameter committed
892

Christoph Lameter's avatar
Christoph Lameter committed
893
894
895
896
897
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
898
		 * as used avoids touching the remaining objects.
Christoph Lameter's avatar
Christoph Lameter committed
899
		 */
900
		slab_fix(s, "Marking all objects used");
Christoph Lameter's avatar
Christoph Lameter committed
901
		page->inuse = s->objects;
902
		page->freelist = page->end;
Christoph Lameter's avatar
Christoph Lameter committed
903
904
905
906
	}
	return 0;
}

Christoph Lameter's avatar
Christoph Lameter committed
907
908
static int free_debug_processing(struct kmem_cache *s, struct page *page,
						void *object, void *addr)
Christoph Lameter's avatar
Christoph Lameter committed
909
910
911
912
913
{
	if (!check_slab(s, page))
		goto fail;

	if (!check_valid_pointer(s, page, object)) {
914
		slab_err(s, page, "Invalid object pointer 0x%p", object);
Christoph Lameter's avatar
Christoph Lameter committed
915
916
917
918
		goto fail;
	}

	if (on_freelist(s, page, object)) {
919
		object_err(s, page, object, "Object already free");
Christoph Lameter's avatar
Christoph Lameter committed
920
921
922
923
924
925
926
		goto fail;
	}

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

	if (unlikely(s != page->slab)) {
Ingo Molnar's avatar
Ingo Molnar committed
927
		if (!PageSlab(page)) {
928
929
			slab_err(s, page, "Attempt to free object(0x%p) "
				"outside of slab", object);
Ingo Molnar's avatar
Ingo Molnar committed
930
		} else if (!page->slab) {
Christoph Lameter's avatar
Christoph Lameter committed
931
			printk(KERN_ERR
932
				"SLUB <none>: no slab for object 0x%p.\n",
Christoph Lameter's avatar
Christoph Lameter committed
933
						object);
934
			dump_stack();
935
		} else
936
937
			object_err(s, page, object,
					"page slab pointer corrupt.");
Christoph Lameter's avatar
Christoph Lameter committed
938
939
		goto fail;
	}
Christoph Lameter's avatar
Christoph Lameter committed
940
941

	/* Special debug activities for freeing objects */
942
	if (!SlabFrozen(page) && page->freelist == page->end)
Christoph Lameter's avatar
Christoph Lameter committed
943
944
945
946
947
		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
948
	return 1;
Christoph Lameter's avatar
Christoph Lameter committed
949

Christoph Lameter's avatar
Christoph Lameter committed
950
fail:
951
	slab_fix(s, "Object at 0x%p not freed", object);
Christoph Lameter's avatar
Christoph Lameter committed
952
953
954
	return 0;
}

955
956
static int __init setup_slub_debug(char *str)
{
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
	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
	 */
981
	for (; *str && *str != ','; str++) {
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
		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' "
1000
				"unknown. skipped\n", *str);
1001
		}
1002
1003
	}

1004
check_slabs:
1005
1006
	if (*str == ',')
		slub_debug_slabs = str + 1;
1007
out:
1008
1009
1010
1011
1012
	return 1;
}

__setup("slub_debug", setup_slub_debug);

1013
1014
static unsigned long kmem_cache_flags(unsigned long objsize,
	unsigned long flags, const char *name,
1015
	void (*ctor)(struct kmem_cache *, void *))
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
{
	/*
	 * 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.
	 *
1026
	 * Debugging or ctor may create a need to move the free
1027
1028
	 * pointer. Fail if this happens.
	 */
1029
1030
	if (objsize >= 65535 * sizeof(void *)) {
		BUG_ON(flags & (SLAB_RED_ZONE | SLAB_POISON |
1031
				SLAB_STORE_USER | SLAB_DESTROY_BY_RCU));
1032
1033
		BUG_ON(ctor);
	} else {
1034
1035
1036
1037
		/*
		 * Enable debugging if selected on the kernel commandline.
		 */
		if (slub_debug && (!slub_debug_slabs ||
1038
		    strncmp(slub_debug_slabs, name,
Ingo Molnar's avatar
Ingo Molnar committed
1039
			strlen(slub_debug_slabs)) == 0))
1040
1041
1042
1043
				flags |= slub_debug;
	}

	return flags;
1044
1045
}
#else
Christoph Lameter's avatar
Christoph Lameter committed
1046
1047
static inline void setup_object_debug(struct kmem_cache *s,
			struct page *page, void *object) {}
1048

Christoph Lameter's avatar
Christoph Lameter committed
1049
1050
static inline int alloc_debug_processing(struct kmem_cache *s,
	struct page *page, void *object, void *addr) { return 0; }
1051

Christoph Lameter's avatar
Christoph Lameter committed
1052
1053
static inline int free_debug_processing(struct kmem_cache *s,
	struct page *page, void *object, void *addr) { return 0; }
1054
1055
1056
1057
1058

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
1059
static inline void add_full(struct kmem_cache_node *n, struct page *page) {}
1060
1061
static inline unsigned long kmem_cache_flags(unsigned long objsize,
	unsigned long flags, const char *name,
1062
	void (*ctor)(struct kmem_cache *, void *))
1063
1064
1065
{
	return flags;
}
1066
1067
#define slub_debug 0
#endif
Christoph Lameter's avatar
Christoph Lameter committed
1068
1069
1070
1071
1072
/*
 * Slab allocation and freeing
 */
static struct page *allocate_slab(struct kmem_cache *s, gfp_t flags, int node)
{
1073
	struct page *page;
Christoph Lameter's avatar
Christoph Lameter committed
1074
1075
	int pages = 1 << s->order;

1076
	flags |= s->allocflags;
1077

Christoph Lameter's avatar
Christoph Lameter committed
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
	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
1097
	setup_object_debug(s, page, object);
1098
	if (unlikely(s->ctor))
1099
		s->ctor(s, object);
Christoph Lameter's avatar
Christoph Lameter committed
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
}

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 *last;
	void *p;

Christoph Lameter's avatar
Christoph Lameter committed
1110
	BUG_ON(flags & GFP_SLAB_BUG_MASK);
Christoph Lameter's avatar
Christoph Lameter committed
1111

Christoph Lameter's avatar
Christoph Lameter committed
1112
1113
	page = allocate_slab(s,
		flags & (GFP_RECLAIM_MASK | GFP_CONSTRAINT_MASK), node);
Christoph Lameter's avatar
Christoph Lameter committed
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
	if (!page)
		goto out;

	n = get_node(s, page_to_nid(page));
	if (n)
		atomic_long_inc(&n->nr_slabs);
	page->slab = s;
	page->flags |= 1 << PG_slab;
	if (s->flags & (SLAB_DEBUG_FREE | SLAB_RED_ZONE | SLAB_POISON |
			SLAB_STORE_USER | SLAB_TRACE))
1124
		SetSlabDebug(page);
Christoph Lameter's avatar
Christoph Lameter committed
1125
1126

	start = page_address(page);
1127
	page->end = start + 1;
Christoph Lameter's avatar
Christoph Lameter committed
1128
1129
1130
1131
1132

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

	last = start;
1133
	for_each_object(p, s, start) {
Christoph Lameter's avatar
Christoph Lameter committed
1134
1135
1136
1137
1138
		setup_object(s, page, last);
		set_freepointer(s, last, p);
		last = p;
	}
	setup_object(s, page, last);
1139
	set_freepointer(s, last, page->end);
Christoph Lameter's avatar
Christoph Lameter committed
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150

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

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

1151
	if (unlikely(SlabDebug(page))) {
Christoph Lameter's avatar
Christoph Lameter committed
1152
1153
1154
		void *p;

		slab_pad_check(s, page);
1155
		for_each_object(p, s, slab_address(page))
Christoph Lameter's avatar
Christoph Lameter committed
1156
			check_object(s, page, p, 0);
1157
		ClearSlabDebug(page);
Christoph Lameter's avatar
Christoph Lameter committed
1158
1159
1160
1161
1162
	}

	mod_zone_page_state(page_zone(page),
		(s->flags & SLAB_RECLAIM_ACCOUNT) ?
		NR_SLAB_RECLAIMABLE : NR_SLAB_UNRECLAIMABLE,
1163
		-pages);
Christoph Lameter's avatar
Christoph Lameter committed
1164

1165
	page->mapping = NULL;
Christoph Lameter's avatar
Christoph Lameter committed
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
	__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);
1196
	__ClearPageSlab(page);
Christoph Lameter's avatar
Christoph Lameter committed
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
	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
1210
	__bit_spin_unlock(PG_locked, &page->flags);
Christoph Lameter's avatar
Christoph Lameter committed
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
}

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
 */
1224
1225
static void add_partial(struct kmem_cache_node *n,
				struct page *page, int tail)
Christoph Lameter's avatar
Christoph Lameter committed
1226
{
Christoph Lameter's avatar
Christoph Lameter committed
1227
1228
	spin_lock(&n->list_lock);
	n->nr_partial++;
1229
1230
1231
1232
	if (tail)
		list_add_tail(&page->lru, &n->partial);
	else
		list_add(&page->lru, &n->partial);
Christoph Lameter's avatar
Christoph Lameter committed
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
	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
1248
 * Lock slab and remove from the partial list.
Christoph Lameter's avatar
Christoph Lameter committed
1249
 *
Christoph Lameter's avatar
Christoph Lameter committed
1250
 * Must hold list_lock.
Christoph Lameter's avatar
Christoph Lameter committed
1251
 */
1252
static inline int lock_and_freeze_slab(struct kmem_cache_node *n, struct page *page)
Christoph Lameter's avatar
Christoph Lameter committed
1253
1254
1255
1256
{
	if (slab_trylock(page)) {
		list_del(&page->lru);
		n->nr_partial--;
1257
		SetSlabFrozen(page);
Christoph Lameter's avatar
Christoph Lameter committed
1258
1259
1260
1261
1262
1263
		return 1;
	}
	return 0;
}

/*
Christoph Lameter's avatar
Christoph Lameter committed
1264
 * Try to allocate a partial slab from a specific node.
Christoph Lameter's avatar
Christoph Lameter committed
1265
1266
1267
1268
1269
1270
1271
1272
 */
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
1273
1274
	 * partial slab and there is none available then get_partials()
	 * will return NULL.
Christoph Lameter's avatar
Christoph Lameter committed
1275
1276
1277
1278
1279
1280
	 */
	if (!n || !n->nr_partial)
		return NULL;

	spin_lock(&n->list_lock);
	list_for_each_entry(page, &n->partial, lru)
1281
		if (lock_and_freeze_slab(n, page))
Christoph Lameter's avatar
Christoph Lameter committed
1282
1283
1284
1285
1286
1287
1288
1289
			goto out;
	page = NULL;
out:
	spin_unlock(&n->list_lock);
	return page;
}

/*
Christoph Lameter's avatar
Christoph Lameter committed
1290
 * Get a page from somewhere. Search in increasing NUMA distances.
Christoph Lameter's avatar
Christoph Lameter committed
1291
1292
1293
1294
1295
1296
1297
1298
1299
 */
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
1300
1301
1302
1303
	 * 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
1304
	 *
Christoph Lameter's avatar
Christoph Lameter committed
1305
1306
1307
1308
	 * 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
1309
1310
	 *
	 * If /sys/slab/xx/defrag_ratio is set to 100 (which makes
Christoph Lameter's avatar
Christoph Lameter committed
1311
1312
1313
1314
1315
	 * 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
1316
	 */
1317
1318
	if (!s->remote_node_defrag_ratio ||
			get_cycles() % 1024 > s->remote_node_defrag_ratio)
Christoph Lameter's avatar
Christoph Lameter committed
1319
1320
		return NULL;

Ingo Molnar's avatar
Ingo Molnar committed
1321
1322
	zonelist = &NODE_DATA(
		slab_node(current->mempolicy))->node_zonelists[gfp_zone(flags)];
Christoph Lameter's avatar
Christoph Lameter committed
1323
1324
1325
1326
1327
1328
	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
1329
				n->nr_partial > MIN_PARTIAL) {
Christoph Lameter's avatar
Christoph Lameter committed
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
			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.
 */
1361
static void unfreeze_slab(struct kmem_cache *s, struct page *page, int tail)
Christoph Lameter's avatar
Christoph Lameter committed
1362
{
Christoph Lameter's avatar
Christoph Lameter committed
1363
	struct kmem_cache_node *n = get_node(s, page_to_nid(page));
1364
	struct kmem_cache_cpu *c = get_cpu_slab(s, smp_processor_id());
Christoph Lameter's avatar
Christoph Lameter committed
1365

1366
	ClearSlabFrozen(page);
Christoph Lameter's avatar
Christoph Lameter committed
1367
	if (page->inuse) {
Christoph Lameter's avatar
Christoph Lameter committed
1368

1369
		if (page->freelist != page->end) {
1370
			add_partial(n, page, tail);
1371
1372
1373
1374
1375
1376
			stat(c, tail ? DEACTIVATE_TO_TAIL : DEACTIVATE_TO_HEAD);
		} else {
			stat(c, DEACTIVATE_FULL);
			if (SlabDebug(page) && (s->flags & SLAB_STORE_USER))
				add_full(n, page);
		}
Christoph Lameter's avatar
Christoph Lameter committed
1377
1378
		slab_unlock(page);
	} else {
1379
		stat(c, DEACTIVATE_EMPTY);
Christoph Lameter's avatar
Christoph Lameter committed
1380
1381
		if (n->nr_partial < MIN_PARTIAL) {
			/*
Christoph Lameter's avatar
Christoph Lameter committed
1382
1383
1384
1385
1386
1387
			 * 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
1388
			 */
1389
			add_partial(n, page, 1);
Christoph Lameter's avatar
Christoph Lameter committed
1390
1391
1392
			slab_unlock(page);
		} else {
			slab_unlock(page);
1393
			stat(get_cpu_slab(s, raw_smp_processor_id()), FREE_SLAB);
Christoph Lameter's avatar
Christoph Lameter committed
1394
1395
			discard_slab(s, page);
		}
Christoph Lameter's avatar
Christoph Lameter committed
1396
1397
1398
1399
1400
1401
	}
}

/*
 * Remove the cpu slab
 */
1402
static void deactivate_slab(struct kmem_cache *s, struct kmem_cache_cpu *c)
Christoph Lameter's avatar
Christoph Lameter committed
1403
{
1404
	struct page *page = c->page;
1405
	int tail = 1;
1406
1407
1408

	if (c->freelist)
		stat(c, DEACTIVATE_REMOTE_FREES);
1409
1410
1411
1412
	/*
	 * Merge cpu freelist into freelist. Typically we get here
	 * because both freelists are empty. So this is unlikely
	 * to occur.
1413
1414
1415
1416
	 *
	 * We need to use _is_end here because deactivate slab may
	 * be called for a debug slab. Then c->freelist may contain
	 * a dummy pointer.
1417
	 */
1418
	while (unlikely(!is_end(c->freelist))) {
1419
1420
		void **object;

1421
1422
		tail = 0;	/* Hot objects. Put the slab first */

1423
		/* Retrieve object from cpu_freelist */
1424
		object = c->freelist;
1425
		c->freelist = c->freelist[c->offset];
1426
1427

		/* And put onto the regular freelist */
1428
		object[c->offset] = page->freelist;
1429
1430
1431
		page->freelist = object;
		page->inuse--;
	}
1432
	c->page = NULL;
1433
	unfreeze_slab(s, page, tail);
Christoph Lameter's avatar
Christoph Lameter committed
1434
1435
}

1436
static inline void flush_slab(struct kmem_cache *s, struct