slub.c 105 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
/*
 * 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

152
153
154
155
/*
 * Mininum number of partial slabs. These will be left on the partial
 * lists even if they are empty. kmem_cache_shrink may reclaim them.
 */
156
#define MIN_PARTIAL 5
Christoph Lameter's avatar
Christoph Lameter committed
157

158
159
160
161
162
163
164
/*
 * 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
165
166
#define DEBUG_DEFAULT_FLAGS (SLAB_DEBUG_FREE | SLAB_RED_ZONE | \
				SLAB_POISON | SLAB_STORE_USER)
Christoph Lameter's avatar
Christoph Lameter committed
167

Christoph Lameter's avatar
Christoph Lameter committed
168
169
170
171
172
173
174
175
176
177
/*
 * 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
178
#define ARCH_KMALLOC_MINALIGN __alignof__(unsigned long long)
Christoph Lameter's avatar
Christoph Lameter committed
179
180
181
#endif

#ifndef ARCH_SLAB_MINALIGN
182
#define ARCH_SLAB_MINALIGN __alignof__(unsigned long long)
Christoph Lameter's avatar
Christoph Lameter committed
183
184
185
#endif

/* Internal SLUB flags */
186
187
#define __OBJECT_POISON		0x80000000 /* Poison object */
#define __SYSFS_ADD_DEFERRED	0x40000000 /* Not yet visible via sysfs */
Christoph Lameter's avatar
Christoph Lameter committed
188
189
190
191
192
193
194
195
196
197

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
198
	UP,		/* Everything works but does not show up in sysfs */
Christoph Lameter's avatar
Christoph Lameter committed
199
200
201
202
203
	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
204
static LIST_HEAD(slab_caches);
Christoph Lameter's avatar
Christoph Lameter committed
205

206
207
208
209
210
211
212
213
214
215
216
217
/*
 * 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 };

218
#if defined(CONFIG_SYSFS) && defined(CONFIG_SLUB_DEBUG)
Christoph Lameter's avatar
Christoph Lameter committed
219
220
221
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 *);
222

Christoph Lameter's avatar
Christoph Lameter committed
223
#else
224
225
226
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; }
227
228
229
230
static inline void sysfs_slab_remove(struct kmem_cache *s)
{
	kfree(s);
}
231

Christoph Lameter's avatar
Christoph Lameter committed
232
233
#endif

234
235
236
237
238
239
240
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
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
/********************************************************************
 * 			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
}

259
260
static inline struct kmem_cache_cpu *get_cpu_slab(struct kmem_cache *s, int cpu)
{
261
262
263
264
265
#ifdef CONFIG_SMP
	return s->cpu_slab[cpu];
#else
	return &s->cpu_slab;
#endif
266
267
}

Christoph Lameter's avatar
Christoph Lameter committed
268
/* Verify that a pointer has an address that is valid within a slab page */
269
270
271
272
273
static inline int check_valid_pointer(struct kmem_cache *s,
				struct page *page, const void *object)
{
	void *base;

274
	if (!object)
275
276
		return 1;

277
	base = page_address(page);
278
	if (object < base || object >= base + page->objects * s->size ||
279
280
281
282
283
284
285
		(object - base) % s->size) {
		return 0;
	}

	return 1;
}

286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
/*
 * 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 */
304
305
#define for_each_object(__p, __s, __addr, __objects) \
	for (__p = (__addr); __p < (__addr) + (__objects) * (__s)->size;\
306
307
308
309
			__p += (__s)->size)

/* Scan freelist */
#define for_each_free_object(__p, __s, __free) \
310
	for (__p = (__free); __p; __p = get_freepointer((__s), __p))
311
312
313
314
315
316
317

/* 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;
}

318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
static inline struct kmem_cache_order_objects oo_make(int order,
						unsigned long size)
{
	struct kmem_cache_order_objects x = {
		(order << 16) + (PAGE_SIZE << order) / size
	};

	return x;
}

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

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

338
339
340
341
#ifdef CONFIG_SLUB_DEBUG
/*
 * Debug settings:
 */
342
343
344
#ifdef CONFIG_SLUB_DEBUG_ON
static int slub_debug = DEBUG_DEFAULT_FLAGS;
#else
345
static int slub_debug;
346
#endif
347
348
349

static char *slub_debug_slabs;

Christoph Lameter's avatar
Christoph Lameter committed
350
351
352
353
354
355
356
357
358
359
360
361
362
/*
 * 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) {
363
			printk(KERN_ERR "%8s 0x%p: ", text, addr + i);
Christoph Lameter's avatar
Christoph Lameter committed
364
365
			newline = 0;
		}
366
		printk(KERN_CONT " %02x", addr[i]);
Christoph Lameter's avatar
Christoph Lameter committed
367
368
369
		offset = i % 16;
		ascii[offset] = isgraph(addr[i]) ? addr[i] : '.';
		if (offset == 15) {
370
			printk(KERN_CONT " %s\n", ascii);
Christoph Lameter's avatar
Christoph Lameter committed
371
372
373
374
375
376
			newline = 1;
		}
	}
	if (!newline) {
		i %= 16;
		while (i < 16) {
377
			printk(KERN_CONT "   ");
Christoph Lameter's avatar
Christoph Lameter committed
378
379
380
			ascii[i] = ' ';
			i++;
		}
381
		printk(KERN_CONT " %s\n", ascii);
Christoph Lameter's avatar
Christoph Lameter committed
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
	}
}

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)
{
420
421
422
423
424
	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
425
426
427
428
429
430
431
}

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

432
	printk(KERN_ERR "INFO: %s in ", s);
Christoph Lameter's avatar
Christoph Lameter committed
433
	__print_symbol("%s", (unsigned long)t->addr);
434
435
436
437
438
439
440
441
442
443
444
445
446
447
	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)
{
448
449
	printk(KERN_ERR "INFO: Slab 0x%p objects=%u used=%u fp=0x%p flags=0x%04lx\n",
		page, page->objects, page->inuse, page->freelist, page->flags);
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465

}

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
466
467
}

468
469
470
471
472
473
474
475
476
477
478
479
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
480
481
{
	unsigned int off;	/* Offset of last byte */
482
	u8 *addr = page_address(page);
483
484
485
486
487
488
489
490
491
492
493
494

	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
495
496
497
498
499
500
501
502
503
504

	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;

505
	if (s->flags & SLAB_STORE_USER)
Christoph Lameter's avatar
Christoph Lameter committed
506
507
508
509
		off += 2 * sizeof(struct track);

	if (off != s->size)
		/* Beginning of the filler is the free pointer */
510
511
512
		print_section("Padding", p + off, s->size - off);

	dump_stack();
Christoph Lameter's avatar
Christoph Lameter committed
513
514
515
516
517
}

static void object_err(struct kmem_cache *s, struct page *page,
			u8 *object, char *reason)
{
518
	slab_bug(s, "%s", reason);
519
	print_trailer(s, page, object);
Christoph Lameter's avatar
Christoph Lameter committed
520
521
}

522
static void slab_err(struct kmem_cache *s, struct page *page, char *fmt, ...)
Christoph Lameter's avatar
Christoph Lameter committed
523
524
525
526
{
	va_list args;
	char buf[100];

527
528
	va_start(args, fmt);
	vsnprintf(buf, sizeof(buf), fmt, args);
Christoph Lameter's avatar
Christoph Lameter committed
529
	va_end(args);
530
	slab_bug(s, "%s", buf);
531
	print_page_info(page);
Christoph Lameter's avatar
Christoph Lameter committed
532
533
534
535
536
537
538
539
540
	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);
541
		p[s->objsize - 1] = POISON_END;
Christoph Lameter's avatar
Christoph Lameter committed
542
543
544
545
546
547
548
549
	}

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

550
static u8 *check_bytes(u8 *start, unsigned int value, unsigned int bytes)
Christoph Lameter's avatar
Christoph Lameter committed
551
552
553
{
	while (bytes) {
		if (*start != (u8)value)
554
			return start;
Christoph Lameter's avatar
Christoph Lameter committed
555
556
557
		start++;
		bytes--;
	}
558
559
560
561
562
563
564
565
566
567
568
569
	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,
570
			u8 *start, unsigned int value, unsigned int bytes)
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
{
	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
590
591
592
593
594
595
596
597
598
}

/*
 * 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
599
 *
Christoph Lameter's avatar
Christoph Lameter committed
600
601
602
603
604
 * 	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
605
606
607
 * 	Padding is extended by another word if Redzoning is enabled and
 * 	objsize == inuse.
 *
Christoph Lameter's avatar
Christoph Lameter committed
608
609
610
611
 * 	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
612
613
 * 	Meta data starts here.
 *
Christoph Lameter's avatar
Christoph Lameter committed
614
615
 * 	A. Free pointer (if we cannot overwrite object on free)
 * 	B. Tracking data for SLAB_STORE_USER
Christoph Lameter's avatar
Christoph Lameter committed
616
 * 	C. Padding to reach required alignment boundary or at mininum
Christoph Lameter's avatar
Christoph Lameter committed
617
 * 		one word if debugging is on to be able to detect writes
Christoph Lameter's avatar
Christoph Lameter committed
618
619
620
 * 		before the word boundary.
 *
 *	Padding is done using 0x5a (POISON_INUSE)
Christoph Lameter's avatar
Christoph Lameter committed
621
622
 *
 * object + s->size
Christoph Lameter's avatar
Christoph Lameter committed
623
 * 	Nothing is used beyond s->size.
Christoph Lameter's avatar
Christoph Lameter committed
624
 *
Christoph Lameter's avatar
Christoph Lameter committed
625
626
 * 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
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
 * 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;

645
646
	return check_bytes_and_report(s, page, p, "Object padding",
				p + off, POISON_INUSE, s->size - off);
Christoph Lameter's avatar
Christoph Lameter committed
647
648
}

649
/* Check the pad bytes at the end of a slab page */
Christoph Lameter's avatar
Christoph Lameter committed
650
651
static int slab_pad_check(struct kmem_cache *s, struct page *page)
{
652
653
654
655
656
	u8 *start;
	u8 *fault;
	u8 *end;
	int length;
	int remainder;
Christoph Lameter's avatar
Christoph Lameter committed
657
658
659
660

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

661
	start = page_address(page);
662
	length = (PAGE_SIZE << compound_order(page));
663
664
	end = start + length;
	remainder = length % s->size;
Christoph Lameter's avatar
Christoph Lameter committed
665
666
667
	if (!remainder)
		return 1;

668
	fault = check_bytes(end - remainder, POISON_INUSE, remainder);
669
670
671
672
673
674
	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);
675
	print_section("Padding", end - remainder, remainder);
676
677
678

	restore_bytes(s, "slab padding", POISON_INUSE, start, end);
	return 0;
Christoph Lameter's avatar
Christoph Lameter committed
679
680
681
682
683
684
685
686
687
688
689
690
}

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;

691
692
		if (!check_bytes_and_report(s, page, object, "Redzone",
			endobject, red, s->inuse - s->objsize))
Christoph Lameter's avatar
Christoph Lameter committed
693
694
			return 0;
	} else {
Ingo Molnar's avatar
Ingo Molnar committed
695
696
697
698
		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
699
700
701
702
	}

	if (s->flags & SLAB_POISON) {
		if (!active && (s->flags & __OBJECT_POISON) &&
703
704
705
			(!check_bytes_and_report(s, page, p, "Poison", p,
					POISON_FREE, s->objsize - 1) ||
			 !check_bytes_and_report(s, page, p, "Poison",
706
				p + s->objsize - 1, POISON_END, 1)))
Christoph Lameter's avatar
Christoph Lameter committed
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
			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
727
		 * another error because the object count is now wrong.
Christoph Lameter's avatar
Christoph Lameter committed
728
		 */
729
		set_freepointer(s, p, NULL);
Christoph Lameter's avatar
Christoph Lameter committed
730
731
732
733
734
735
736
		return 0;
	}
	return 1;
}

static int check_slab(struct kmem_cache *s, struct page *page)
{
737
738
	int maxobj;

Christoph Lameter's avatar
Christoph Lameter committed
739
740
741
	VM_BUG_ON(!irqs_disabled());

	if (!PageSlab(page)) {
742
		slab_err(s, page, "Not a valid slab page");
Christoph Lameter's avatar
Christoph Lameter committed
743
744
		return 0;
	}
745
746
747
748
749
750
751
752

	maxobj = (PAGE_SIZE << compound_order(page)) / s->size;
	if (page->objects > maxobj) {
		slab_err(s, page, "objects %u > max %u",
			s->name, page->objects, maxobj);
		return 0;
	}
	if (page->inuse > page->objects) {
753
		slab_err(s, page, "inuse %u > max %u",
754
			s->name, page->inuse, page->objects);
Christoph Lameter's avatar
Christoph Lameter committed
755
756
757
758
759
760
761
762
		return 0;
	}
	/* Slab_pad_check fixes things up after itself */
	slab_pad_check(s, page);
	return 1;
}

/*
Christoph Lameter's avatar
Christoph Lameter committed
763
764
 * 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
765
766
767
768
769
770
 */
static int on_freelist(struct kmem_cache *s, struct page *page, void *search)
{
	int nr = 0;
	void *fp = page->freelist;
	void *object = NULL;
771
	unsigned long max_objects;
Christoph Lameter's avatar
Christoph Lameter committed
772

773
	while (fp && nr <= page->objects) {
Christoph Lameter's avatar
Christoph Lameter committed
774
775
776
777
778
779
		if (fp == search)
			return 1;
		if (!check_valid_pointer(s, page, fp)) {
			if (object) {
				object_err(s, page, object,
					"Freechain corrupt");
780
				set_freepointer(s, object, NULL);
Christoph Lameter's avatar
Christoph Lameter committed
781
782
				break;
			} else {
783
				slab_err(s, page, "Freepointer corrupt");
784
				page->freelist = NULL;
785
				page->inuse = page->objects;
786
				slab_fix(s, "Freelist cleared");
Christoph Lameter's avatar
Christoph Lameter committed
787
788
789
790
791
792
793
794
795
				return 0;
			}
			break;
		}
		object = fp;
		fp = get_freepointer(s, object);
		nr++;
	}

796
797
798
799
800
801
802
803
804
805
	max_objects = (PAGE_SIZE << compound_order(page)) / s->size;
	if (max_objects > 65535)
		max_objects = 65535;

	if (page->objects != max_objects) {
		slab_err(s, page, "Wrong number of objects. Found %d but "
			"should be %d", page->objects, max_objects);
		page->objects = max_objects;
		slab_fix(s, "Number of objects adjusted.");
	}
806
	if (page->inuse != page->objects - nr) {
807
		slab_err(s, page, "Wrong object count. Counter is %d but "
808
809
			"counted were %d", page->inuse, page->objects - nr);
		page->inuse = page->objects - nr;
810
		slab_fix(s, "Object count adjusted.");
Christoph Lameter's avatar
Christoph Lameter committed
811
812
813
814
	}
	return search == NULL;
}

Christoph Lameter's avatar
Christoph Lameter committed
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
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();
	}
}

831
/*
Christoph Lameter's avatar
Christoph Lameter committed
832
 * Tracking of fully allocated slabs for debugging purposes.
833
 */
Christoph Lameter's avatar
Christoph Lameter committed
834
static void add_full(struct kmem_cache_node *n, struct page *page)
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
{
	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);
}

855
856
857
858
859
860
861
862
/* Tracking of the number of slabs for debugging purposes */
static inline unsigned long slabs_node(struct kmem_cache *s, int node)
{
	struct kmem_cache_node *n = get_node(s, node);

	return atomic_long_read(&n->nr_slabs);
}

863
static inline void inc_slabs_node(struct kmem_cache *s, int node, int objects)
864
865
866
867
868
869
870
871
872
{
	struct kmem_cache_node *n = get_node(s, node);

	/*
	 * May be called early in order to allocate a slab for the
	 * kmem_cache_node structure. Solve the chicken-egg
	 * dilemma by deferring the increment of the count during
	 * bootstrap (see early_kmem_cache_node_alloc).
	 */
873
	if (!NUMA_BUILD || n) {
874
		atomic_long_inc(&n->nr_slabs);
875
876
		atomic_long_add(objects, &n->total_objects);
	}
877
}
878
static inline void dec_slabs_node(struct kmem_cache *s, int node, int objects)
879
880
881
882
{
	struct kmem_cache_node *n = get_node(s, node);

	atomic_long_dec(&n->nr_slabs);
883
	atomic_long_sub(objects, &n->total_objects);
884
885
886
}

/* Object debug checks for alloc/free paths */
Christoph Lameter's avatar
Christoph Lameter committed
887
888
889
890
891
892
893
894
895
896
897
898
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
899
900
901
902
{
	if (!check_slab(s, page))
		goto bad;

903
	if (!on_freelist(s, page, object)) {
904
		object_err(s, page, object, "Object already allocated");
905
		goto bad;
Christoph Lameter's avatar
Christoph Lameter committed
906
907
908
909
	}

	if (!check_valid_pointer(s, page, object)) {
		object_err(s, page, object, "Freelist Pointer check fails");
910
		goto bad;
Christoph Lameter's avatar
Christoph Lameter committed
911
912
	}

913
	if (!check_object(s, page, object, 0))
Christoph Lameter's avatar
Christoph Lameter committed
914
915
		goto bad;

Christoph Lameter's avatar
Christoph Lameter committed
916
917
918
919
920
	/* 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
921
	return 1;
Christoph Lameter's avatar
Christoph Lameter committed
922

Christoph Lameter's avatar
Christoph Lameter committed
923
924
925
926
927
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
928
		 * as used avoids touching the remaining objects.
Christoph Lameter's avatar
Christoph Lameter committed
929
		 */
930
		slab_fix(s, "Marking all objects used");
931
		page->inuse = page->objects;
932
		page->freelist = NULL;
Christoph Lameter's avatar
Christoph Lameter committed
933
934
935
936
	}
	return 0;
}

Christoph Lameter's avatar
Christoph Lameter committed
937
938
static int free_debug_processing(struct kmem_cache *s, struct page *page,
						void *object, void *addr)
Christoph Lameter's avatar
Christoph Lameter committed
939
940
941
942
943
{
	if (!check_slab(s, page))
		goto fail;

	if (!check_valid_pointer(s, page, object)) {
944
		slab_err(s, page, "Invalid object pointer 0x%p", object);
Christoph Lameter's avatar
Christoph Lameter committed
945
946
947
948
		goto fail;
	}

	if (on_freelist(s, page, object)) {
949
		object_err(s, page, object, "Object already free");
Christoph Lameter's avatar
Christoph Lameter committed
950
951
952
953
954
955
956
		goto fail;
	}

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

	if (unlikely(s != page->slab)) {
Ingo Molnar's avatar
Ingo Molnar committed
957
		if (!PageSlab(page)) {
958
959
			slab_err(s, page, "Attempt to free object(0x%p) "
				"outside of slab", object);
Ingo Molnar's avatar
Ingo Molnar committed
960
		} else if (!page->slab) {
Christoph Lameter's avatar
Christoph Lameter committed
961
			printk(KERN_ERR
962
				"SLUB <none>: no slab for object 0x%p.\n",
Christoph Lameter's avatar
Christoph Lameter committed
963
						object);
964
			dump_stack();
965
		} else
966
967
			object_err(s, page, object,
					"page slab pointer corrupt.");
Christoph Lameter's avatar
Christoph Lameter committed
968
969
		goto fail;
	}
Christoph Lameter's avatar
Christoph Lameter committed
970
971

	/* Special debug activities for freeing objects */
972
	if (!SlabFrozen(page) && !page->freelist)
Christoph Lameter's avatar
Christoph Lameter committed
973
974
975
976
977
		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
978
	return 1;
Christoph Lameter's avatar
Christoph Lameter committed
979

Christoph Lameter's avatar
Christoph Lameter committed
980
fail:
981
	slab_fix(s, "Object at 0x%p not freed", object);
Christoph Lameter's avatar
Christoph Lameter committed
982
983
984
	return 0;
}

985
986
static int __init setup_slub_debug(char *str)
{
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
	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
	 */
1011
	for (; *str && *str != ','; str++) {
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
		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' "
1030
				"unknown. skipped\n", *str);
1031
		}
1032
1033
	}

1034
check_slabs:
1035
1036
	if (*str == ',')
		slub_debug_slabs = str + 1;
1037
out:
1038
1039
1040
1041
1042
	return 1;
}

__setup("slub_debug", setup_slub_debug);

1043
1044
static unsigned long kmem_cache_flags(unsigned long objsize,
	unsigned long flags, const char *name,
1045
	void (*ctor)(struct kmem_cache *, void *))
1046
1047
{
	/*
1048
	 * Enable debugging if selected on the kernel commandline.
1049
	 */
1050
1051
1052
	if (slub_debug && (!slub_debug_slabs ||
	    strncmp(slub_debug_slabs, name, strlen(slub_debug_slabs)) == 0))
			flags |= slub_debug;
1053
1054

	return flags;
1055
1056
}
#else
Christoph Lameter's avatar
Christoph Lameter committed
1057
1058
static inline void setup_object_debug(struct kmem_cache *s,
			struct page *page, void *object) {}
1059

Christoph Lameter's avatar
Christoph Lameter committed
1060
1061
static inline int alloc_debug_processing(struct kmem_cache *s,
	struct page *page, void *object, void *addr) { return 0; }
1062

Christoph Lameter's avatar
Christoph Lameter committed
1063
1064
static inline int free_debug_processing(struct kmem_cache *s,
	struct page *page, void *object, void *addr) { return 0; }
1065
1066
1067
1068
1069

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
1070
static inline void add_full(struct kmem_cache_node *n, struct page *page) {}
1071
1072
static inline unsigned long kmem_cache_flags(unsigned long objsize,
	unsigned long flags, const char *name,
1073
	void (*ctor)(struct kmem_cache *, void *))
1074
1075
1076
{
	return flags;
}
1077
#define slub_debug 0
1078
1079
1080

static inline unsigned long slabs_node(struct kmem_cache *s, int node)
							{ return 0; }
1081
1082
1083
1084
static inline void inc_slabs_node(struct kmem_cache *s, int node,
							int objects) {}
static inline void dec_slabs_node(struct kmem_cache *s, int node,
							int objects) {}
1085
#endif
1086

Christoph Lameter's avatar
Christoph Lameter committed
1087
1088
1089
/*
 * Slab allocation and freeing
 */
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
static inline struct page *alloc_slab_page(gfp_t flags, int node,
					struct kmem_cache_order_objects oo)
{
	int order = oo_order(oo);

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

Christoph Lameter's avatar
Christoph Lameter committed
1101
1102
static struct page *allocate_slab(struct kmem_cache *s, gfp_t flags, int node)
{
1103
	struct page *page;
1104
	struct kmem_cache_order_objects oo = s->oo;
Christoph Lameter's avatar
Christoph Lameter committed
1105

1106
	flags |= s->allocflags;
1107

1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
	page = alloc_slab_page(flags | __GFP_NOWARN | __GFP_NORETRY, node,
									oo);
	if (unlikely(!page)) {
		oo = s->min;
		/*
		 * Allocation may have failed due to fragmentation.
		 * Try a lower order alloc if possible
		 */
		page = alloc_slab_page(flags, node, oo);
		if (!page)
			return NULL;
Christoph Lameter's avatar
Christoph Lameter committed
1119

1120
1121
		stat(get_cpu_slab(s, raw_smp_processor_id()), ORDER_FALLBACK);
	}
1122
	page->objects = oo_objects(oo);
Christoph Lameter's avatar
Christoph Lameter committed
1123
1124
1125
	mod_zone_page_state(page_zone(page),
		(s->flags & SLAB_RECLAIM_ACCOUNT) ?
		NR_SLAB_RECLAIMABLE : NR_SLAB_UNRECLAIMABLE,
1126
		1 << oo_order(oo));
Christoph Lameter's avatar
Christoph Lameter committed
1127
1128
1129
1130
1131
1132
1133

	return page;
}

static void setup_object(struct kmem_cache *s, struct page *page,
				void *object)
{
Christoph Lameter's avatar
Christoph Lameter committed
1134
	setup_object_debug(s, page, object);
1135
	if (unlikely(s->ctor))
1136
		s->ctor(s, object);
Christoph Lameter's avatar
Christoph Lameter committed
1137
1138
1139
1140
1141
1142
1143
1144
1145
}

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

Christoph Lameter's avatar
Christoph Lameter committed
1146
	BUG_ON(flags & GFP_SLAB_BUG_MASK);
Christoph Lameter's avatar
Christoph Lameter committed
1147

Christoph Lameter's avatar
Christoph Lameter committed
1148
1149
	page = allocate_slab(s,
		flags & (GFP_RECLAIM_MASK | GFP_CONSTRAINT_MASK), node);
Christoph Lameter's avatar
Christoph Lameter committed
1150
1151
1152
	if (!page)
		goto out;

1153
	inc_slabs_node(s, page_to_nid(page), page->objects);
Christoph Lameter's avatar
Christoph Lameter committed
1154
1155
1156
1157
	page->slab = s;
	page->flags |= 1 << PG_slab;
	if (s->flags & (SLAB_DEBUG_FREE | SLAB_RED_ZONE | SLAB_POISON |
			SLAB_STORE_USER | SLAB_TRACE))
1158
		SetSlabDebug(page);
Christoph Lameter's avatar
Christoph Lameter committed
1159
1160
1161
1162

	start = page_address(page);

	if (unlikely(s->flags & SLAB_POISON))
1163
		memset(start, POISON_INUSE, PAGE_SIZE << compound_order(page));
Christoph Lameter's avatar
Christoph Lameter committed
1164
1165

	last = start;
1166
	for_each_object(p, s, start, page->objects) {
Christoph Lameter's avatar
Christoph Lameter committed
1167
1168
1169
1170
1171
		setup_object(s, page, last);
		set_freepointer(s, last, p);
		last = p;
	}
	setup_object(s, page, last);
1172
	set_freepointer(s, last, NULL);
Christoph Lameter's avatar
Christoph Lameter committed
1173
1174
1175
1176
1177
1178
1179
1180
1181

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

static void __free_slab(struct kmem_cache *s, struct page *page)
{
1182
1183
	int order = compound_order(page);
	int pages = 1 << order;
Christoph Lameter's avatar
Christoph Lameter committed
1184

1185
	if (unlikely(SlabDebug(page))) {
Christoph Lameter's avatar
Christoph Lameter committed
1186
1187
1188
		void *p;

		slab_pad_check(s, page);
1189
1190
		for_each_object(p, s, page_address(page),
						page->objects)
Christoph Lameter's avatar
Christoph Lameter committed
1191
			check_object(s, page, p, 0);
1192
		ClearSlabDebug(page);
Christoph Lameter's avatar
Christoph Lameter committed
1193
1194
1195
1196
1197
	}

	mod_zone_page_state(page_zone(page),
		(s->flags & SLAB_RECLAIM_ACCOUNT) ?
		NR_SLAB_RECLAIMABLE : NR_SLAB_UNRECLAIMABLE,
1198
		-pages);
Christoph Lameter's avatar
Christoph Lameter committed
1199

1200
1201
	__ClearPageSlab(page);
	reset_page_mapcount(page);
1202
	__free_pages(page, order);
Christoph Lameter's avatar
Christoph Lameter committed
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
}

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)
{
1228
	dec_slabs_node(s, page_to_nid(page), page->objects);
Christoph Lameter's avatar
Christoph Lameter committed
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
	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
1242
	__bit_spin_unlock(PG_locked, &page->flags);
Christoph Lameter's avatar
Christoph Lameter committed
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
}

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
 */
1256
1257
static void add_partial(struct kmem_cache_node *n,
				struct page *page, int tail)
Christoph Lameter's avatar
Christoph Lameter committed
1258
{
Christoph Lameter's avatar
Christoph Lameter committed
1259
1260
	spin_lock(&n->list_lock);
	n->nr_partial++;
1261
1262
1263
1264
	if (tail)
		list_add_tail(&page->lru, &n->partial);
	else
		list_add(&page->lru, &n->partial);
Christoph Lameter's avatar
Christoph Lameter committed
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
	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
1280
 * Lock slab and remove from the partial list.
Christoph Lameter's avatar
Christoph Lameter committed
1281
 *
Christoph Lameter's avatar
Christoph Lameter committed
1282
 * Must hold list_lock.
Christoph Lameter's avatar
Christoph Lameter committed
1283
 */
1284
static inline int lock_and_freeze_slab(struct kmem_cache_node *n, struct page *page)
Christoph Lameter's avatar
Christoph Lameter committed
1285
1286
1287
1288
{
	if (slab_trylock(page)) {
		list_del(&page->lru);
		n->nr_partial--;
1289
		SetSlabFrozen(page);
Christoph Lameter's avatar
Christoph Lameter committed
1290
1291
1292
1293
1294
1295
		return 1;
	}
	return 0;
}

/*
Christoph Lameter's avatar
Christoph Lameter committed
1296
 * Try to allocate a partial slab from a specific node.
Christoph Lameter's avatar
Christoph Lameter committed
1297
1298
1299
1300
1301
1302
1303
1304
 */
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
1305
1306
	 * partial slab and there is none available then get_partials()
	 * will return NULL.
Christoph Lameter's avatar
Christoph Lameter committed
1307
1308
1309
1310
1311
1312
	 */
	if (!n || !n->nr_partial)
		return NULL;

	spin_lock(&n->list_lock);
	list_for_each_entry(page, &n->partial, lru)
1313
		if (lock_and_freeze_slab(n, page))
Christoph Lameter's avatar
Christoph Lameter committed
1314
1315
1316
1317
1318
1319
1320
1321
			goto out;
	page = NULL;
out:
	spin_unlock(&n->list_lock);
	return page;
}

/*
Christoph Lameter's avatar
Christoph Lameter committed
1322
 * Get a page from somewhere. Search in increasing NUMA distances.
Christoph Lameter's avatar
Christoph Lameter committed
1323
1324
1325
1326
1327
 */
static struct page *get_any_partial(struct kmem_cache *s, gfp_t flags)
{
#ifdef CONFIG_NUMA
	struct zonelist *zonelist;
1328
	struct zoneref *z;
1329
1330
	struct zone *zone;
	enum zone_type high_zoneidx = gfp_zone(flags);
Christoph Lameter's avatar
Christoph Lameter committed
1331
1332
1333
	struct page *page;

	/*
Christoph Lameter's avatar
Christoph Lameter committed
1334
1335
1336
1337
	 * 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
1338
	 *
Christoph Lameter's avatar
Christoph Lameter committed
1339
1340
1341
1342
	 * 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
1343
	 *
Christoph Lameter's avatar
Christoph Lameter committed
1344
	 * If /sys/kernel/slab/xx/defrag_ratio is set to 100 (which makes
Christoph Lameter's avatar
Christoph Lameter committed
1345
1346
1347
1348
1349
	 * 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
1350
	 */
1351
1352
	if (!s->remote_node_defrag_ratio ||
			get_cycles() % 1024 > s->remote_node_defrag_ratio)
Christoph Lameter's avatar
Christoph Lameter committed
1353
1354
		return NULL;

1355
	zonelist = node_zonelist(slab_node(current->mempolicy), flags);
1356
	for_each_zone_zonelist(zone, z, zonelist, high_zoneidx) {
Christoph Lameter's avatar
Christoph Lameter committed
1357
1358
		struct kmem_cache_node *n;

1359
		n = get_node(s, zone_to_nid(zone));
Christoph Lameter's avatar
Christoph Lameter committed
1360

1361
		if (n && cpuset_zone_allowed_hardwall(zone, flags) &&
Christoph Lameter's avatar
Christoph Lameter committed
1362
				n->nr_partial > MIN_PARTIAL) {
Christoph Lameter's avatar
Christoph Lameter committed
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
			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;