slub.c 106 KB
Newer Older
Christoph Lameter's avatar
Christoph Lameter committed
1
2
3
4
5
6
7
/*
 * SLUB: A slab allocator that limits cache line use instead of queuing
 * objects in per cpu and per node lists.
 *
 * The allocator synchronizes using per slab locks and only
 * uses a centralized lock to manage a pool of partial slabs.
 *
Christoph Lameter's avatar
Christoph Lameter committed
8
 * (C) 2007 SGI, Christoph Lameter
Christoph Lameter's avatar
Christoph Lameter committed
9
10
11
12
13
14
15
16
 */

#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>
17
#include <linux/proc_fs.h>
Christoph Lameter's avatar
Christoph Lameter committed
18
19
20
21
22
#include <linux/seq_file.h>
#include <linux/cpu.h>
#include <linux/cpuset.h>
#include <linux/mempolicy.h>
#include <linux/ctype.h>
23
#include <linux/debugobjects.h>
Christoph Lameter's avatar
Christoph Lameter committed
24
#include <linux/kallsyms.h>
25
#include <linux/memory.h>
Roman Zippel's avatar
Roman Zippel committed
26
#include <linux/math64.h>
Akinobu Mita's avatar
Akinobu Mita committed
27
#include <linux/fault-inject.h>
Christoph Lameter's avatar
Christoph Lameter committed
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73

/*
 * 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
74
75
 * 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
76
 * freed then the slab will show up again on the partial lists.
Christoph Lameter's avatar
Christoph Lameter committed
77
78
 * We track full slabs for debugging purposes though because otherwise we
 * cannot scan all objects.
Christoph Lameter's avatar
Christoph Lameter committed
79
80
81
82
83
84
85
 *
 * 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.
 *
86
87
88
89
90
91
92
93
94
95
96
97
 * 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
98
 * 			freelist that allows lockless access to
99
100
 * 			free objects in addition to the regular freelist
 * 			that requires the slab lock.
Christoph Lameter's avatar
Christoph Lameter committed
101
102
103
 *
 * PageError		Slab requires special handling due to debug
 * 			options set. This moves	slab handling out of
104
 * 			the fast path and disables lockless freelists.
Christoph Lameter's avatar
Christoph Lameter committed
105
106
 */

107
#ifdef CONFIG_SLUB_DEBUG
108
#define SLABDEBUG 1
109
110
111
112
#else
#define SLABDEBUG 0
#endif

Christoph Lameter's avatar
Christoph Lameter committed
113
114
115
116
117
118
119
120
121
122
123
/*
 * 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

124
125
126
127
/*
 * Mininum number of partial slabs. These will be left on the partial
 * lists even if they are empty. kmem_cache_shrink may reclaim them.
 */
128
#define MIN_PARTIAL 5
Christoph Lameter's avatar
Christoph Lameter committed
129

130
131
132
133
134
135
136
/*
 * 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
137
138
#define DEBUG_DEFAULT_FLAGS (SLAB_DEBUG_FREE | SLAB_RED_ZONE | \
				SLAB_POISON | SLAB_STORE_USER)
Christoph Lameter's avatar
Christoph Lameter committed
139

Christoph Lameter's avatar
Christoph Lameter committed
140
141
142
143
144
145
146
147
148
149
/*
 * 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
150
#define ARCH_KMALLOC_MINALIGN __alignof__(unsigned long long)
Christoph Lameter's avatar
Christoph Lameter committed
151
152
153
#endif

#ifndef ARCH_SLAB_MINALIGN
154
#define ARCH_SLAB_MINALIGN __alignof__(unsigned long long)
Christoph Lameter's avatar
Christoph Lameter committed
155
156
#endif

157
158
159
160
#define OO_SHIFT	16
#define OO_MASK		((1 << OO_SHIFT) - 1)
#define MAX_OBJS_PER_PAGE	65535 /* since page.objects is u16 */

Christoph Lameter's avatar
Christoph Lameter committed
161
/* Internal SLUB flags */
162
163
#define __OBJECT_POISON		0x80000000 /* Poison object */
#define __SYSFS_ADD_DEFERRED	0x40000000 /* Not yet visible via sysfs */
Christoph Lameter's avatar
Christoph Lameter committed
164
165
166
167
168
169
170
171
172
173

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
174
	UP,		/* Everything works but does not show up in sysfs */
Christoph Lameter's avatar
Christoph Lameter committed
175
176
177
178
179
	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
180
static LIST_HEAD(slab_caches);
Christoph Lameter's avatar
Christoph Lameter committed
181

182
183
184
185
/*
 * Tracking user of a slab.
 */
struct track {
186
	unsigned long addr;	/* Called from address */
187
188
189
190
191
192
193
	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 };

194
#ifdef CONFIG_SLUB_DEBUG
Christoph Lameter's avatar
Christoph Lameter committed
195
196
197
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 *);
198

Christoph Lameter's avatar
Christoph Lameter committed
199
#else
200
201
202
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; }
203
204
205
206
static inline void sysfs_slab_remove(struct kmem_cache *s)
{
	kfree(s);
}
207

Christoph Lameter's avatar
Christoph Lameter committed
208
209
#endif

210
211
212
213
214
215
216
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
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
/********************************************************************
 * 			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
}

235
236
static inline struct kmem_cache_cpu *get_cpu_slab(struct kmem_cache *s, int cpu)
{
237
238
239
240
241
#ifdef CONFIG_SMP
	return s->cpu_slab[cpu];
#else
	return &s->cpu_slab;
#endif
242
243
}

Christoph Lameter's avatar
Christoph Lameter committed
244
/* Verify that a pointer has an address that is valid within a slab page */
245
246
247
248
249
static inline int check_valid_pointer(struct kmem_cache *s,
				struct page *page, const void *object)
{
	void *base;

250
	if (!object)
251
252
		return 1;

253
	base = page_address(page);
254
	if (object < base || object >= base + page->objects * s->size ||
255
256
257
258
259
260
261
		(object - base) % s->size) {
		return 0;
	}

	return 1;
}

262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
/*
 * 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 */
280
281
#define for_each_object(__p, __s, __addr, __objects) \
	for (__p = (__addr); __p < (__addr) + (__objects) * (__s)->size;\
282
283
284
285
			__p += (__s)->size)

/* Scan freelist */
#define for_each_free_object(__p, __s, __free) \
286
	for (__p = (__free); __p; __p = get_freepointer((__s), __p))
287
288
289
290
291
292
293

/* 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;
}

294
295
296
297
static inline struct kmem_cache_order_objects oo_make(int order,
						unsigned long size)
{
	struct kmem_cache_order_objects x = {
298
		(order << OO_SHIFT) + (PAGE_SIZE << order) / size
299
300
301
302
303
304
305
	};

	return x;
}

static inline int oo_order(struct kmem_cache_order_objects x)
{
306
	return x.x >> OO_SHIFT;
307
308
309
310
}

static inline int oo_objects(struct kmem_cache_order_objects x)
{
311
	return x.x & OO_MASK;
312
313
}

314
315
316
317
#ifdef CONFIG_SLUB_DEBUG
/*
 * Debug settings:
 */
318
319
320
#ifdef CONFIG_SLUB_DEBUG_ON
static int slub_debug = DEBUG_DEFAULT_FLAGS;
#else
321
static int slub_debug;
322
#endif
323
324
325

static char *slub_debug_slabs;

Christoph Lameter's avatar
Christoph Lameter committed
326
327
328
329
330
331
332
333
334
335
336
337
338
/*
 * 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) {
339
			printk(KERN_ERR "%8s 0x%p: ", text, addr + i);
Christoph Lameter's avatar
Christoph Lameter committed
340
341
			newline = 0;
		}
342
		printk(KERN_CONT " %02x", addr[i]);
Christoph Lameter's avatar
Christoph Lameter committed
343
344
345
		offset = i % 16;
		ascii[offset] = isgraph(addr[i]) ? addr[i] : '.';
		if (offset == 15) {
346
			printk(KERN_CONT " %s\n", ascii);
Christoph Lameter's avatar
Christoph Lameter committed
347
348
349
350
351
352
			newline = 1;
		}
	}
	if (!newline) {
		i %= 16;
		while (i < 16) {
353
			printk(KERN_CONT "   ");
Christoph Lameter's avatar
Christoph Lameter committed
354
355
356
			ascii[i] = ' ';
			i++;
		}
357
		printk(KERN_CONT " %s\n", ascii);
Christoph Lameter's avatar
Christoph Lameter committed
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
	}
}

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,
375
			enum track_item alloc, unsigned long addr)
Christoph Lameter's avatar
Christoph Lameter committed
376
377
378
379
380
381
382
383
384
385
386
387
{
	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();
388
		p->pid = current->pid;
Christoph Lameter's avatar
Christoph Lameter committed
389
390
391
392
393
394
395
		p->when = jiffies;
	} else
		memset(p, 0, sizeof(struct track));
}

static void init_tracking(struct kmem_cache *s, void *object)
{
396
397
398
	if (!(s->flags & SLAB_STORE_USER))
		return;

399
400
	set_track(s, object, TRACK_FREE, 0UL);
	set_track(s, object, TRACK_ALLOC, 0UL);
Christoph Lameter's avatar
Christoph Lameter committed
401
402
403
404
405
406
407
}

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

408
	printk(KERN_ERR "INFO: %s in %pS age=%lu cpu=%u pid=%d\n",
409
		s, (void *)t->addr, jiffies - t->when, t->cpu, t->pid);
410
411
412
413
414
415
416
417
418
419
420
421
422
}

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)
{
423
424
	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);
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440

}

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
441
442
}

443
444
445
446
447
448
449
450
451
452
453
454
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
455
456
{
	unsigned int off;	/* Offset of last byte */
457
	u8 *addr = page_address(page);
458
459
460
461
462
463
464
465
466
467
468

	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);

469
	print_section("Object", p, min_t(unsigned long, s->objsize, PAGE_SIZE));
Christoph Lameter's avatar
Christoph Lameter committed
470
471
472
473
474
475
476
477
478
479

	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;

480
	if (s->flags & SLAB_STORE_USER)
Christoph Lameter's avatar
Christoph Lameter committed
481
482
483
484
		off += 2 * sizeof(struct track);

	if (off != s->size)
		/* Beginning of the filler is the free pointer */
485
486
487
		print_section("Padding", p + off, s->size - off);

	dump_stack();
Christoph Lameter's avatar
Christoph Lameter committed
488
489
490
491
492
}

static void object_err(struct kmem_cache *s, struct page *page,
			u8 *object, char *reason)
{
493
	slab_bug(s, "%s", reason);
494
	print_trailer(s, page, object);
Christoph Lameter's avatar
Christoph Lameter committed
495
496
}

497
static void slab_err(struct kmem_cache *s, struct page *page, char *fmt, ...)
Christoph Lameter's avatar
Christoph Lameter committed
498
499
500
501
{
	va_list args;
	char buf[100];

502
503
	va_start(args, fmt);
	vsnprintf(buf, sizeof(buf), fmt, args);
Christoph Lameter's avatar
Christoph Lameter committed
504
	va_end(args);
505
	slab_bug(s, "%s", buf);
506
	print_page_info(page);
Christoph Lameter's avatar
Christoph Lameter committed
507
508
509
510
511
512
513
514
515
	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);
516
		p[s->objsize - 1] = POISON_END;
Christoph Lameter's avatar
Christoph Lameter committed
517
518
519
520
521
522
523
524
	}

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

525
static u8 *check_bytes(u8 *start, unsigned int value, unsigned int bytes)
Christoph Lameter's avatar
Christoph Lameter committed
526
527
528
{
	while (bytes) {
		if (*start != (u8)value)
529
			return start;
Christoph Lameter's avatar
Christoph Lameter committed
530
531
532
		start++;
		bytes--;
	}
533
534
535
536
537
538
539
540
541
542
543
544
	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,
545
			u8 *start, unsigned int value, unsigned int bytes)
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
{
	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
565
566
567
568
569
570
571
572
573
}

/*
 * 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
574
 *
Christoph Lameter's avatar
Christoph Lameter committed
575
576
577
578
579
 * 	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
580
581
582
 * 	Padding is extended by another word if Redzoning is enabled and
 * 	objsize == inuse.
 *
Christoph Lameter's avatar
Christoph Lameter committed
583
584
585
586
 * 	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
587
588
 * 	Meta data starts here.
 *
Christoph Lameter's avatar
Christoph Lameter committed
589
590
 * 	A. Free pointer (if we cannot overwrite object on free)
 * 	B. Tracking data for SLAB_STORE_USER
Christoph Lameter's avatar
Christoph Lameter committed
591
 * 	C. Padding to reach required alignment boundary or at mininum
Christoph Lameter's avatar
Christoph Lameter committed
592
 * 		one word if debugging is on to be able to detect writes
Christoph Lameter's avatar
Christoph Lameter committed
593
594
595
 * 		before the word boundary.
 *
 *	Padding is done using 0x5a (POISON_INUSE)
Christoph Lameter's avatar
Christoph Lameter committed
596
597
 *
 * object + s->size
Christoph Lameter's avatar
Christoph Lameter committed
598
 * 	Nothing is used beyond s->size.
Christoph Lameter's avatar
Christoph Lameter committed
599
 *
Christoph Lameter's avatar
Christoph Lameter committed
600
601
 * 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
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
 * 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;

620
621
	return check_bytes_and_report(s, page, p, "Object padding",
				p + off, POISON_INUSE, s->size - off);
Christoph Lameter's avatar
Christoph Lameter committed
622
623
}

624
/* Check the pad bytes at the end of a slab page */
Christoph Lameter's avatar
Christoph Lameter committed
625
626
static int slab_pad_check(struct kmem_cache *s, struct page *page)
{
627
628
629
630
631
	u8 *start;
	u8 *fault;
	u8 *end;
	int length;
	int remainder;
Christoph Lameter's avatar
Christoph Lameter committed
632
633
634
635

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

636
	start = page_address(page);
637
	length = (PAGE_SIZE << compound_order(page));
638
639
	end = start + length;
	remainder = length % s->size;
Christoph Lameter's avatar
Christoph Lameter committed
640
641
642
	if (!remainder)
		return 1;

643
	fault = check_bytes(end - remainder, POISON_INUSE, remainder);
644
645
646
647
648
649
	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);
650
	print_section("Padding", end - remainder, remainder);
651
652
653

	restore_bytes(s, "slab padding", POISON_INUSE, start, end);
	return 0;
Christoph Lameter's avatar
Christoph Lameter committed
654
655
656
657
658
659
660
661
662
663
664
665
}

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;

666
667
		if (!check_bytes_and_report(s, page, object, "Redzone",
			endobject, red, s->inuse - s->objsize))
Christoph Lameter's avatar
Christoph Lameter committed
668
669
			return 0;
	} else {
Ingo Molnar's avatar
Ingo Molnar committed
670
671
672
673
		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
674
675
676
677
	}

	if (s->flags & SLAB_POISON) {
		if (!active && (s->flags & __OBJECT_POISON) &&
678
679
680
			(!check_bytes_and_report(s, page, p, "Poison", p,
					POISON_FREE, s->objsize - 1) ||
			 !check_bytes_and_report(s, page, p, "Poison",
681
				p + s->objsize - 1, POISON_END, 1)))
Christoph Lameter's avatar
Christoph Lameter committed
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
			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");
		/*
Nick Andrew's avatar
Nick Andrew committed
700
		 * No choice but to zap it and thus lose the remainder
Christoph Lameter's avatar
Christoph Lameter committed
701
		 * of the free objects in this slab. May cause
Christoph Lameter's avatar
Christoph Lameter committed
702
		 * another error because the object count is now wrong.
Christoph Lameter's avatar
Christoph Lameter committed
703
		 */
704
		set_freepointer(s, p, NULL);
Christoph Lameter's avatar
Christoph Lameter committed
705
706
707
708
709
710
711
		return 0;
	}
	return 1;
}

static int check_slab(struct kmem_cache *s, struct page *page)
{
712
713
	int maxobj;

Christoph Lameter's avatar
Christoph Lameter committed
714
715
716
	VM_BUG_ON(!irqs_disabled());

	if (!PageSlab(page)) {
717
		slab_err(s, page, "Not a valid slab page");
Christoph Lameter's avatar
Christoph Lameter committed
718
719
		return 0;
	}
720
721
722
723
724
725
726
727

	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) {
728
		slab_err(s, page, "inuse %u > max %u",
729
			s->name, page->inuse, page->objects);
Christoph Lameter's avatar
Christoph Lameter committed
730
731
732
733
734
735
736
737
		return 0;
	}
	/* Slab_pad_check fixes things up after itself */
	slab_pad_check(s, page);
	return 1;
}

/*
Christoph Lameter's avatar
Christoph Lameter committed
738
739
 * 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
740
741
742
743
744
745
 */
static int on_freelist(struct kmem_cache *s, struct page *page, void *search)
{
	int nr = 0;
	void *fp = page->freelist;
	void *object = NULL;
746
	unsigned long max_objects;
Christoph Lameter's avatar
Christoph Lameter committed
747

748
	while (fp && nr <= page->objects) {
Christoph Lameter's avatar
Christoph Lameter committed
749
750
751
752
753
754
		if (fp == search)
			return 1;
		if (!check_valid_pointer(s, page, fp)) {
			if (object) {
				object_err(s, page, object,
					"Freechain corrupt");
755
				set_freepointer(s, object, NULL);
Christoph Lameter's avatar
Christoph Lameter committed
756
757
				break;
			} else {
758
				slab_err(s, page, "Freepointer corrupt");
759
				page->freelist = NULL;
760
				page->inuse = page->objects;
761
				slab_fix(s, "Freelist cleared");
Christoph Lameter's avatar
Christoph Lameter committed
762
763
764
765
766
767
768
769
770
				return 0;
			}
			break;
		}
		object = fp;
		fp = get_freepointer(s, object);
		nr++;
	}

771
	max_objects = (PAGE_SIZE << compound_order(page)) / s->size;
772
773
	if (max_objects > MAX_OBJS_PER_PAGE)
		max_objects = MAX_OBJS_PER_PAGE;
774
775
776
777
778
779
780

	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.");
	}
781
	if (page->inuse != page->objects - nr) {
782
		slab_err(s, page, "Wrong object count. Counter is %d but "
783
784
			"counted were %d", page->inuse, page->objects - nr);
		page->inuse = page->objects - nr;
785
		slab_fix(s, "Object count adjusted.");
Christoph Lameter's avatar
Christoph Lameter committed
786
787
788
789
	}
	return search == NULL;
}

790
791
static void trace(struct kmem_cache *s, struct page *page, void *object,
								int alloc)
Christoph Lameter's avatar
Christoph Lameter committed
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
{
	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();
	}
}

807
/*
Christoph Lameter's avatar
Christoph Lameter committed
808
 * Tracking of fully allocated slabs for debugging purposes.
809
 */
Christoph Lameter's avatar
Christoph Lameter committed
810
static void add_full(struct kmem_cache_node *n, struct page *page)
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
{
	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);
}

831
832
833
834
835
836
837
838
/* 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);
}

839
static inline void inc_slabs_node(struct kmem_cache *s, int node, int objects)
840
841
842
843
844
845
846
847
848
{
	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).
	 */
849
	if (!NUMA_BUILD || n) {
850
		atomic_long_inc(&n->nr_slabs);
851
852
		atomic_long_add(objects, &n->total_objects);
	}
853
}
854
static inline void dec_slabs_node(struct kmem_cache *s, int node, int objects)
855
856
857
858
{
	struct kmem_cache_node *n = get_node(s, node);

	atomic_long_dec(&n->nr_slabs);
859
	atomic_long_sub(objects, &n->total_objects);
860
861
862
}

/* Object debug checks for alloc/free paths */
Christoph Lameter's avatar
Christoph Lameter committed
863
864
865
866
867
868
869
870
871
872
873
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,
874
					void *object, unsigned long addr)
Christoph Lameter's avatar
Christoph Lameter committed
875
876
877
878
{
	if (!check_slab(s, page))
		goto bad;

879
	if (!on_freelist(s, page, object)) {
880
		object_err(s, page, object, "Object already allocated");
881
		goto bad;
Christoph Lameter's avatar
Christoph Lameter committed
882
883
884
885
	}

	if (!check_valid_pointer(s, page, object)) {
		object_err(s, page, object, "Freelist Pointer check fails");
886
		goto bad;
Christoph Lameter's avatar
Christoph Lameter committed
887
888
	}

889
	if (!check_object(s, page, object, 0))
Christoph Lameter's avatar
Christoph Lameter committed
890
891
		goto bad;

Christoph Lameter's avatar
Christoph Lameter committed
892
893
894
895
896
	/* 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
897
	return 1;
Christoph Lameter's avatar
Christoph Lameter committed
898

Christoph Lameter's avatar
Christoph Lameter committed
899
900
901
902
903
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
904
		 * as used avoids touching the remaining objects.
Christoph Lameter's avatar
Christoph Lameter committed
905
		 */
906
		slab_fix(s, "Marking all objects used");
907
		page->inuse = page->objects;
908
		page->freelist = NULL;
Christoph Lameter's avatar
Christoph Lameter committed
909
910
911
912
	}
	return 0;
}

Christoph Lameter's avatar
Christoph Lameter committed
913
static int free_debug_processing(struct kmem_cache *s, struct page *page,
914
					void *object, unsigned long addr)
Christoph Lameter's avatar
Christoph Lameter committed
915
916
917
918
919
{
	if (!check_slab(s, page))
		goto fail;

	if (!check_valid_pointer(s, page, object)) {
920
		slab_err(s, page, "Invalid object pointer 0x%p", object);
Christoph Lameter's avatar
Christoph Lameter committed
921
922
923
924
		goto fail;
	}

	if (on_freelist(s, page, object)) {
925
		object_err(s, page, object, "Object already free");
Christoph Lameter's avatar
Christoph Lameter committed
926
927
928
929
930
931
932
		goto fail;
	}

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

	if (unlikely(s != page->slab)) {
Ingo Molnar's avatar
Ingo Molnar committed
933
		if (!PageSlab(page)) {
934
935
			slab_err(s, page, "Attempt to free object(0x%p) "
				"outside of slab", object);
Ingo Molnar's avatar
Ingo Molnar committed
936
		} else if (!page->slab) {
Christoph Lameter's avatar
Christoph Lameter committed
937
			printk(KERN_ERR
938
				"SLUB <none>: no slab for object 0x%p.\n",
Christoph Lameter's avatar
Christoph Lameter committed
939
						object);
940
			dump_stack();
941
		} else
942
943
			object_err(s, page, object,
					"page slab pointer corrupt.");
Christoph Lameter's avatar
Christoph Lameter committed
944
945
		goto fail;
	}
Christoph Lameter's avatar
Christoph Lameter committed
946
947

	/* Special debug activities for freeing objects */
948
	if (!PageSlubFrozen(page) && !page->freelist)
Christoph Lameter's avatar
Christoph Lameter committed
949
950
951
952
953
		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
954
	return 1;
Christoph Lameter's avatar
Christoph Lameter committed
955

Christoph Lameter's avatar
Christoph Lameter committed
956
fail:
957
	slab_fix(s, "Object at 0x%p not freed", object);
Christoph Lameter's avatar
Christoph Lameter committed
958
959
960
	return 0;
}

961
962
static int __init setup_slub_debug(char *str)
{
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
	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
	 */
987
	for (; *str && *str != ','; str++) {
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
		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' "
1006
				"unknown. skipped\n", *str);
1007
		}
1008
1009
	}

1010
check_slabs:
1011
1012
	if (*str == ',')
		slub_debug_slabs = str + 1;
1013
out:
1014
1015
1016
1017
1018
	return 1;
}

__setup("slub_debug", setup_slub_debug);

1019
1020
static unsigned long kmem_cache_flags(unsigned long objsize,
	unsigned long flags, const char *name,
1021
	void (*ctor)(void *))
1022
1023
{
	/*
1024
	 * Enable debugging if selected on the kernel commandline.
1025
	 */
1026
1027
1028
	if (slub_debug && (!slub_debug_slabs ||
	    strncmp(slub_debug_slabs, name, strlen(slub_debug_slabs)) == 0))
			flags |= slub_debug;
1029
1030

	return flags;
1031
1032
}
#else
Christoph Lameter's avatar
Christoph Lameter committed
1033
1034
static inline void setup_object_debug(struct kmem_cache *s,
			struct page *page, void *object) {}
1035

Christoph Lameter's avatar
Christoph Lameter committed
1036
static inline int alloc_debug_processing(struct kmem_cache *s,
1037
	struct page *page, void *object, unsigned long addr) { return 0; }
1038

Christoph Lameter's avatar
Christoph Lameter committed
1039
static inline int free_debug_processing(struct kmem_cache *s,
1040
	struct page *page, void *object, unsigned long addr) { return 0; }
1041
1042
1043
1044
1045

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
1046
static inline void add_full(struct kmem_cache_node *n, struct page *page) {}
1047
1048
static inline unsigned long kmem_cache_flags(unsigned long objsize,
	unsigned long flags, const char *name,
1049
	void (*ctor)(void *))
1050
1051
1052
{
	return flags;
}
1053
#define slub_debug 0
1054
1055
1056

static inline unsigned long slabs_node(struct kmem_cache *s, int node)
							{ return 0; }
1057
1058
1059
1060
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) {}
1061
#endif
1062

Christoph Lameter's avatar
Christoph Lameter committed
1063
1064
1065
/*
 * Slab allocation and freeing
 */
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
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
1077
1078
static struct page *allocate_slab(struct kmem_cache *s, gfp_t flags, int node)
{
1079
	struct page *page;
1080
	struct kmem_cache_order_objects oo = s->oo;
Christoph Lameter's avatar
Christoph Lameter committed
1081

1082
	flags |= s->allocflags;
1083

1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
	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
1095

1096
1097
		stat(get_cpu_slab(s, raw_smp_processor_id()), ORDER_FALLBACK);
	}
1098
	page->objects = oo_objects(oo);
Christoph Lameter's avatar
Christoph Lameter committed
1099
1100
1101
	mod_zone_page_state(page_zone(page),
		(s->flags & SLAB_RECLAIM_ACCOUNT) ?
		NR_SLAB_RECLAIMABLE : NR_SLAB_UNRECLAIMABLE,
1102
		1 << oo_order(oo));
Christoph Lameter's avatar
Christoph Lameter committed
1103
1104
1105
1106
1107
1108
1109

	return page;
}

static void setup_object(struct kmem_cache *s, struct page *page,
				void *object)
{
Christoph Lameter's avatar
Christoph Lameter committed
1110
	setup_object_debug(s, page, object);
1111
	if (unlikely(s->ctor))
1112
		s->ctor(object);
Christoph Lameter's avatar
Christoph Lameter committed
1113
1114
1115
1116
1117
1118
1119
1120
1121
}

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
1122
	BUG_ON(flags & GFP_SLAB_BUG_MASK);
Christoph Lameter's avatar
Christoph Lameter committed
1123

Christoph Lameter's avatar
Christoph Lameter committed
1124
1125
	page = allocate_slab(s,
		flags & (GFP_RECLAIM_MASK | GFP_CONSTRAINT_MASK), node);
Christoph Lameter's avatar
Christoph Lameter committed
1126
1127
1128
	if (!page)
		goto out;

1129
	inc_slabs_node(s, page_to_nid(page), page->objects);
Christoph Lameter's avatar
Christoph Lameter committed
1130
1131
1132
1133
	page->slab = s;
	page->flags |= 1 << PG_slab;
	if (s->flags & (SLAB_DEBUG_FREE | SLAB_RED_ZONE | SLAB_POISON |
			SLAB_STORE_USER | SLAB_TRACE))
1134
		__SetPageSlubDebug(page);
Christoph Lameter's avatar
Christoph Lameter committed
1135
1136
1137
1138

	start = page_address(page);

	if (unlikely(s->flags & SLAB_POISON))
1139
		memset(start, POISON_INUSE, PAGE_SIZE << compound_order(page));
Christoph Lameter's avatar
Christoph Lameter committed
1140
1141

	last = start;
1142
	for_each_object(p, s, start, page->objects) {
Christoph Lameter's avatar
Christoph Lameter committed
1143
1144
1145
1146
1147
		setup_object(s, page, last);
		set_freepointer(s, last, p);
		last = p;
	}
	setup_object(s, page, last);
1148
	set_freepointer(s, last, NULL);
Christoph Lameter's avatar
Christoph Lameter committed
1149
1150
1151
1152
1153
1154
1155
1156
1157

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

static void __free_slab(struct kmem_cache *s, struct page *page)
{
1158
1159
	int order = compound_order(page);
	int pages = 1 << order;
Christoph Lameter's avatar
Christoph Lameter committed
1160

1161
	if (unlikely(SLABDEBUG && PageSlubDebug(page))) {
Christoph Lameter's avatar
Christoph Lameter committed
1162
1163
1164
		void *p;

		slab_pad_check(s, page);
1165
1166
		for_each_object(p, s, page_address(page),
						page->objects)
Christoph Lameter's avatar
Christoph Lameter committed
1167
			check_object(s, page, p, 0);
1168
		__ClearPageSlubDebug(page);
Christoph Lameter's avatar
Christoph Lameter committed
1169
1170
1171
1172
1173
	}

	mod_zone_page_state(page_zone(page),
		(s->flags & SLAB_RECLAIM_ACCOUNT) ?
		NR_SLAB_RECLAIMABLE : NR_SLAB_UNRECLAIMABLE,
1174
		-pages);
Christoph Lameter's avatar
Christoph Lameter committed
1175

1176
1177
	__ClearPageSlab(page);
	reset_page_mapcount(page);
1178
	__free_pages(page, order);
Christoph Lameter's avatar
Christoph Lameter committed
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
}

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)
{
1204
	dec_slabs_node(s, page_to_nid(page), page->objects);
Christoph Lameter's avatar
Christoph Lameter committed
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
	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
1218
	__bit_spin_unlock(PG_locked, &page->flags);
Christoph Lameter's avatar
Christoph Lameter committed
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
}

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
 */
1232
1233
static void add_partial(struct kmem_cache_node *n,
				struct page *page, int tail)
Christoph Lameter's avatar
Christoph Lameter committed
1234
{
Christoph Lameter's avatar
Christoph Lameter committed
1235
1236
	spin_lock(&n->list_lock);
	n->nr_partial++;
1237
1238
1239
1240
	if (tail)
		list_add_tail(&page->lru, &n->partial);
	else
		list_add(&page->lru, &n->partial);
Christoph Lameter's avatar
Christoph Lameter committed
1241
1242
1243
	spin_unlock(&n->list_lock);
}

1244
static void remove_partial(struct kmem_cache *s, struct page *page)
Christoph Lameter's avatar
Christoph Lameter committed
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
{
	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
1255
 * Lock slab and remove from the partial list.
Christoph Lameter's avatar
Christoph Lameter committed
1256
 *
Christoph Lameter's avatar
Christoph Lameter committed
1257
 * Must hold list_lock.
Christoph Lameter's avatar
Christoph Lameter committed
1258
 */
1259
1260
static inline int lock_and_freeze_slab(struct kmem_cache_node *n,
							struct page *page)
Christoph Lameter's avatar
Christoph Lameter committed
1261
1262
1263
1264
{
	if (slab_trylock(page)) {
		list_del(&page->lru);
		n->nr_partial--;
1265
		__SetPageSlubFrozen(page);
Christoph Lameter's avatar
Christoph Lameter committed
1266
1267
1268
1269
1270
1271
		return 1;
	}
	return 0;
}

/*
Christoph Lameter's avatar
Christoph Lameter committed
1272
 * Try to allocate a partial slab from a specific node.
Christoph Lameter's avatar
Christoph Lameter committed
1273
1274
1275
1276
1277
1278
1279
1280
 */
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
1281
1282
	 * partial slab and there is none available then get_partials()
	 * will return NULL.
Christoph Lameter's avatar
Christoph Lameter committed
1283
1284
1285
1286
1287
1288
	 */
	if (!n || !n->nr_partial)
		return NULL;

	spin_lock(&n->list_lock);
	list_for_each_entry(page, &n->partial, lru)
1289
		if (lock_and_freeze_slab(n, page))
Christoph Lameter's avatar
Christoph Lameter committed
1290
1291
1292
1293
1294
1295
1296
1297
			goto out;
	page = NULL;
out:
	spin_unlock(&n->list_lock);
	return page;
}

/*
Christoph Lameter's avatar
Christoph Lameter committed
1298
 * Get a page from somewhere. Search in increasing NUMA distances.
Christoph Lameter's avatar
Christoph Lameter committed
1299
1300
1301
1302
1303
 */
static struct page *get_any_partial(struct kmem_cache *s, gfp_t flags)
{
#ifdef CONFIG_NUMA
	struct zonelist *zonelist;
1304
	struct zoneref *z;
1305
1306
	struct zone *zone;
	enum zone_type high_zoneidx = gfp_zone(flags);
Christoph Lameter's avatar
Christoph Lameter committed
1307
1308
1309
	struct page *page;

	/*
Christoph Lameter's avatar
Christoph Lameter committed
1310
1311
1312
1313
	 * 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
1314
	 *
Christoph Lameter's avatar
Christoph Lameter committed
1315
1316
1317
1318
	 * 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
1319
	 *
Christoph Lameter's avatar
Christoph Lameter committed
1320
	 * If /sys/kernel/slab/xx/defrag_ratio is set to 100 (which makes
Christoph Lameter's avatar
Christoph Lameter committed
1321
1322
1323
1324
1325
	 * 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
1326
	 */
1327
1328
	if (!s->remote_node_defrag_ratio ||
			get_cycles() % 1024 > s->remote_node_defrag_ratio)