slub.c 108 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
#include <linux/seq_file.h>
19
#include <trace/kmemtrace.h>
Christoph Lameter's avatar
Christoph Lameter committed
20
21
22
23
#include <linux/cpu.h>
#include <linux/cpuset.h>
#include <linux/mempolicy.h>
#include <linux/ctype.h>
24
#include <linux/debugobjects.h>
Christoph Lameter's avatar
Christoph Lameter committed
25
#include <linux/kallsyms.h>
26
#include <linux/memory.h>
Roman Zippel's avatar
Roman Zippel committed
27
#include <linux/math64.h>
Akinobu Mita's avatar
Akinobu Mita committed
28
#include <linux/fault-inject.h>
Christoph Lameter's avatar
Christoph Lameter committed
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
74

/*
 * 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
75
76
 * 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
77
 * freed then the slab will show up again on the partial lists.
Christoph Lameter's avatar
Christoph Lameter committed
78
79
 * We track full slabs for debugging purposes though because otherwise we
 * cannot scan all objects.
Christoph Lameter's avatar
Christoph Lameter committed
80
81
82
83
84
85
86
 *
 * 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.
 *
87
88
89
90
91
92
93
94
95
96
97
98
 * 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
99
 * 			freelist that allows lockless access to
100
101
 * 			free objects in addition to the regular freelist
 * 			that requires the slab lock.
Christoph Lameter's avatar
Christoph Lameter committed
102
103
104
 *
 * PageError		Slab requires special handling due to debug
 * 			options set. This moves	slab handling out of
105
 * 			the fast path and disables lockless freelists.
Christoph Lameter's avatar
Christoph Lameter committed
106
107
 */

108
#ifdef CONFIG_SLUB_DEBUG
109
#define SLABDEBUG 1
110
111
112
113
#else
#define SLABDEBUG 0
#endif

Christoph Lameter's avatar
Christoph Lameter committed
114
115
116
117
118
119
120
121
122
123
124
/*
 * 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

125
126
127
128
/*
 * Mininum number of partial slabs. These will be left on the partial
 * lists even if they are empty. kmem_cache_shrink may reclaim them.
 */
129
#define MIN_PARTIAL 5
Christoph Lameter's avatar
Christoph Lameter committed
130

131
132
133
134
135
136
137
/*
 * 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
138
139
#define DEBUG_DEFAULT_FLAGS (SLAB_DEBUG_FREE | SLAB_RED_ZONE | \
				SLAB_POISON | SLAB_STORE_USER)
Christoph Lameter's avatar
Christoph Lameter committed
140

Christoph Lameter's avatar
Christoph Lameter committed
141
142
143
144
145
146
147
148
149
150
/*
 * 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
151
#define ARCH_KMALLOC_MINALIGN __alignof__(unsigned long long)
Christoph Lameter's avatar
Christoph Lameter committed
152
153
154
#endif

#ifndef ARCH_SLAB_MINALIGN
155
#define ARCH_SLAB_MINALIGN __alignof__(unsigned long long)
Christoph Lameter's avatar
Christoph Lameter committed
156
157
#endif

158
159
160
161
#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
162
/* Internal SLUB flags */
163
164
#define __OBJECT_POISON		0x80000000 /* Poison object */
#define __SYSFS_ADD_DEFERRED	0x40000000 /* Not yet visible via sysfs */
Christoph Lameter's avatar
Christoph Lameter committed
165
166
167
168
169
170
171
172
173
174

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
175
	UP,		/* Everything works but does not show up in sysfs */
Christoph Lameter's avatar
Christoph Lameter committed
176
177
178
179
180
	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
181
static LIST_HEAD(slab_caches);
Christoph Lameter's avatar
Christoph Lameter committed
182

183
184
185
186
/*
 * Tracking user of a slab.
 */
struct track {
187
	unsigned long addr;	/* Called from address */
188
189
190
191
192
193
194
	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 };

195
#ifdef CONFIG_SLUB_DEBUG
Christoph Lameter's avatar
Christoph Lameter committed
196
197
198
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 *);
199

Christoph Lameter's avatar
Christoph Lameter committed
200
#else
201
202
203
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; }
204
205
206
207
static inline void sysfs_slab_remove(struct kmem_cache *s)
{
	kfree(s);
}
208

Christoph Lameter's avatar
Christoph Lameter committed
209
210
#endif

211
212
213
214
215
216
217
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
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
/********************************************************************
 * 			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
}

236
237
static inline struct kmem_cache_cpu *get_cpu_slab(struct kmem_cache *s, int cpu)
{
238
239
240
241
242
#ifdef CONFIG_SMP
	return s->cpu_slab[cpu];
#else
	return &s->cpu_slab;
#endif
243
244
}

Christoph Lameter's avatar
Christoph Lameter committed
245
/* Verify that a pointer has an address that is valid within a slab page */
246
247
248
249
250
static inline int check_valid_pointer(struct kmem_cache *s,
				struct page *page, const void *object)
{
	void *base;

251
	if (!object)
252
253
		return 1;

254
	base = page_address(page);
255
	if (object < base || object >= base + page->objects * s->size ||
256
257
258
259
260
261
262
		(object - base) % s->size) {
		return 0;
	}

	return 1;
}

263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
/*
 * 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 */
281
282
#define for_each_object(__p, __s, __addr, __objects) \
	for (__p = (__addr); __p < (__addr) + (__objects) * (__s)->size;\
283
284
285
286
			__p += (__s)->size)

/* Scan freelist */
#define for_each_free_object(__p, __s, __free) \
287
	for (__p = (__free); __p; __p = get_freepointer((__s), __p))
288
289
290
291
292
293
294

/* 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;
}

295
296
297
298
static inline struct kmem_cache_order_objects oo_make(int order,
						unsigned long size)
{
	struct kmem_cache_order_objects x = {
299
		(order << OO_SHIFT) + (PAGE_SIZE << order) / size
300
301
302
303
304
305
306
	};

	return x;
}

static inline int oo_order(struct kmem_cache_order_objects x)
{
307
	return x.x >> OO_SHIFT;
308
309
310
311
}

static inline int oo_objects(struct kmem_cache_order_objects x)
{
312
	return x.x & OO_MASK;
313
314
}

315
316
317
318
#ifdef CONFIG_SLUB_DEBUG
/*
 * Debug settings:
 */
319
320
321
#ifdef CONFIG_SLUB_DEBUG_ON
static int slub_debug = DEBUG_DEFAULT_FLAGS;
#else
322
static int slub_debug;
323
#endif
324
325
326

static char *slub_debug_slabs;

Christoph Lameter's avatar
Christoph Lameter committed
327
328
329
330
331
332
333
334
335
336
337
338
339
/*
 * 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) {
340
			printk(KERN_ERR "%8s 0x%p: ", text, addr + i);
Christoph Lameter's avatar
Christoph Lameter committed
341
342
			newline = 0;
		}
343
		printk(KERN_CONT " %02x", addr[i]);
Christoph Lameter's avatar
Christoph Lameter committed
344
345
346
		offset = i % 16;
		ascii[offset] = isgraph(addr[i]) ? addr[i] : '.';
		if (offset == 15) {
347
			printk(KERN_CONT " %s\n", ascii);
Christoph Lameter's avatar
Christoph Lameter committed
348
349
350
351
352
353
			newline = 1;
		}
	}
	if (!newline) {
		i %= 16;
		while (i < 16) {
354
			printk(KERN_CONT "   ");
Christoph Lameter's avatar
Christoph Lameter committed
355
356
357
			ascii[i] = ' ';
			i++;
		}
358
		printk(KERN_CONT " %s\n", ascii);
Christoph Lameter's avatar
Christoph Lameter committed
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
	}
}

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,
376
			enum track_item alloc, unsigned long addr)
Christoph Lameter's avatar
Christoph Lameter committed
377
{
Akinobu Mita's avatar
Akinobu Mita committed
378
	struct track *p = get_track(s, object, alloc);
Christoph Lameter's avatar
Christoph Lameter committed
379
380
381
382

	if (addr) {
		p->addr = addr;
		p->cpu = smp_processor_id();
383
		p->pid = current->pid;
Christoph Lameter's avatar
Christoph Lameter committed
384
385
386
387
388
389
390
		p->when = jiffies;
	} else
		memset(p, 0, sizeof(struct track));
}

static void init_tracking(struct kmem_cache *s, void *object)
{
391
392
393
	if (!(s->flags & SLAB_STORE_USER))
		return;

394
395
	set_track(s, object, TRACK_FREE, 0UL);
	set_track(s, object, TRACK_ALLOC, 0UL);
Christoph Lameter's avatar
Christoph Lameter committed
396
397
398
399
400
401
402
}

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

403
	printk(KERN_ERR "INFO: %s in %pS age=%lu cpu=%u pid=%d\n",
404
		s, (void *)t->addr, jiffies - t->when, t->cpu, t->pid);
405
406
407
408
409
410
411
412
413
414
415
416
417
}

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)
{
418
419
	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);
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435

}

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
436
437
}

438
439
440
441
442
443
444
445
446
447
448
449
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
450
451
{
	unsigned int off;	/* Offset of last byte */
452
	u8 *addr = page_address(page);
453
454
455
456
457
458
459
460
461
462
463

	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);

464
	print_section("Object", p, min_t(unsigned long, s->objsize, PAGE_SIZE));
Christoph Lameter's avatar
Christoph Lameter committed
465
466
467
468
469
470
471
472
473
474

	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;

475
	if (s->flags & SLAB_STORE_USER)
Christoph Lameter's avatar
Christoph Lameter committed
476
477
478
479
		off += 2 * sizeof(struct track);

	if (off != s->size)
		/* Beginning of the filler is the free pointer */
480
481
482
		print_section("Padding", p + off, s->size - off);

	dump_stack();
Christoph Lameter's avatar
Christoph Lameter committed
483
484
485
486
487
}

static void object_err(struct kmem_cache *s, struct page *page,
			u8 *object, char *reason)
{
488
	slab_bug(s, "%s", reason);
489
	print_trailer(s, page, object);
Christoph Lameter's avatar
Christoph Lameter committed
490
491
}

492
static void slab_err(struct kmem_cache *s, struct page *page, char *fmt, ...)
Christoph Lameter's avatar
Christoph Lameter committed
493
494
495
496
{
	va_list args;
	char buf[100];

497
498
	va_start(args, fmt);
	vsnprintf(buf, sizeof(buf), fmt, args);
Christoph Lameter's avatar
Christoph Lameter committed
499
	va_end(args);
500
	slab_bug(s, "%s", buf);
501
	print_page_info(page);
Christoph Lameter's avatar
Christoph Lameter committed
502
503
504
505
506
507
508
509
510
	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);
511
		p[s->objsize - 1] = POISON_END;
Christoph Lameter's avatar
Christoph Lameter committed
512
513
514
515
516
517
518
519
	}

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

520
static u8 *check_bytes(u8 *start, unsigned int value, unsigned int bytes)
Christoph Lameter's avatar
Christoph Lameter committed
521
522
523
{
	while (bytes) {
		if (*start != (u8)value)
524
			return start;
Christoph Lameter's avatar
Christoph Lameter committed
525
526
527
		start++;
		bytes--;
	}
528
529
530
531
532
533
534
535
536
537
538
539
	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,
540
			u8 *start, unsigned int value, unsigned int bytes)
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
{
	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
560
561
562
563
564
565
566
567
568
}

/*
 * 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
569
 *
Christoph Lameter's avatar
Christoph Lameter committed
570
571
572
573
574
 * 	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
575
576
577
 * 	Padding is extended by another word if Redzoning is enabled and
 * 	objsize == inuse.
 *
Christoph Lameter's avatar
Christoph Lameter committed
578
579
580
581
 * 	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
582
583
 * 	Meta data starts here.
 *
Christoph Lameter's avatar
Christoph Lameter committed
584
585
 * 	A. Free pointer (if we cannot overwrite object on free)
 * 	B. Tracking data for SLAB_STORE_USER
Christoph Lameter's avatar
Christoph Lameter committed
586
 * 	C. Padding to reach required alignment boundary or at mininum
Christoph Lameter's avatar
Christoph Lameter committed
587
 * 		one word if debugging is on to be able to detect writes
Christoph Lameter's avatar
Christoph Lameter committed
588
589
590
 * 		before the word boundary.
 *
 *	Padding is done using 0x5a (POISON_INUSE)
Christoph Lameter's avatar
Christoph Lameter committed
591
592
 *
 * object + s->size
Christoph Lameter's avatar
Christoph Lameter committed
593
 * 	Nothing is used beyond s->size.
Christoph Lameter's avatar
Christoph Lameter committed
594
 *
Christoph Lameter's avatar
Christoph Lameter committed
595
596
 * 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
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
 * 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;

615
616
	return check_bytes_and_report(s, page, p, "Object padding",
				p + off, POISON_INUSE, s->size - off);
Christoph Lameter's avatar
Christoph Lameter committed
617
618
}

619
/* Check the pad bytes at the end of a slab page */
Christoph Lameter's avatar
Christoph Lameter committed
620
621
static int slab_pad_check(struct kmem_cache *s, struct page *page)
{
622
623
624
625
626
	u8 *start;
	u8 *fault;
	u8 *end;
	int length;
	int remainder;
Christoph Lameter's avatar
Christoph Lameter committed
627
628
629
630

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

631
	start = page_address(page);
632
	length = (PAGE_SIZE << compound_order(page));
633
634
	end = start + length;
	remainder = length % s->size;
Christoph Lameter's avatar
Christoph Lameter committed
635
636
637
	if (!remainder)
		return 1;

638
	fault = check_bytes(end - remainder, POISON_INUSE, remainder);
639
640
641
642
643
644
	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);
645
	print_section("Padding", end - remainder, remainder);
646
647
648

	restore_bytes(s, "slab padding", POISON_INUSE, start, end);
	return 0;
Christoph Lameter's avatar
Christoph Lameter committed
649
650
651
652
653
654
655
656
657
658
659
660
}

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;

661
662
		if (!check_bytes_and_report(s, page, object, "Redzone",
			endobject, red, s->inuse - s->objsize))
Christoph Lameter's avatar
Christoph Lameter committed
663
664
			return 0;
	} else {
Ingo Molnar's avatar
Ingo Molnar committed
665
666
667
668
		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
669
670
671
672
	}

	if (s->flags & SLAB_POISON) {
		if (!active && (s->flags & __OBJECT_POISON) &&
673
674
675
			(!check_bytes_and_report(s, page, p, "Poison", p,
					POISON_FREE, s->objsize - 1) ||
			 !check_bytes_and_report(s, page, p, "Poison",
676
				p + s->objsize - 1, POISON_END, 1)))
Christoph Lameter's avatar
Christoph Lameter committed
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
			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
695
		 * No choice but to zap it and thus lose the remainder
Christoph Lameter's avatar
Christoph Lameter committed
696
		 * of the free objects in this slab. May cause
Christoph Lameter's avatar
Christoph Lameter committed
697
		 * another error because the object count is now wrong.
Christoph Lameter's avatar
Christoph Lameter committed
698
		 */
699
		set_freepointer(s, p, NULL);
Christoph Lameter's avatar
Christoph Lameter committed
700
701
702
703
704
705
706
		return 0;
	}
	return 1;
}

static int check_slab(struct kmem_cache *s, struct page *page)
{
707
708
	int maxobj;

Christoph Lameter's avatar
Christoph Lameter committed
709
710
711
	VM_BUG_ON(!irqs_disabled());

	if (!PageSlab(page)) {
712
		slab_err(s, page, "Not a valid slab page");
Christoph Lameter's avatar
Christoph Lameter committed
713
714
		return 0;
	}
715
716
717
718
719
720
721
722

	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) {
723
		slab_err(s, page, "inuse %u > max %u",
724
			s->name, page->inuse, page->objects);
Christoph Lameter's avatar
Christoph Lameter committed
725
726
727
728
729
730
731
732
		return 0;
	}
	/* Slab_pad_check fixes things up after itself */
	slab_pad_check(s, page);
	return 1;
}

/*
Christoph Lameter's avatar
Christoph Lameter committed
733
734
 * 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
735
736
737
738
739
740
 */
static int on_freelist(struct kmem_cache *s, struct page *page, void *search)
{
	int nr = 0;
	void *fp = page->freelist;
	void *object = NULL;
741
	unsigned long max_objects;
Christoph Lameter's avatar
Christoph Lameter committed
742

743
	while (fp && nr <= page->objects) {
Christoph Lameter's avatar
Christoph Lameter committed
744
745
746
747
748
749
		if (fp == search)
			return 1;
		if (!check_valid_pointer(s, page, fp)) {
			if (object) {
				object_err(s, page, object,
					"Freechain corrupt");
750
				set_freepointer(s, object, NULL);
Christoph Lameter's avatar
Christoph Lameter committed
751
752
				break;
			} else {
753
				slab_err(s, page, "Freepointer corrupt");
754
				page->freelist = NULL;
755
				page->inuse = page->objects;
756
				slab_fix(s, "Freelist cleared");
Christoph Lameter's avatar
Christoph Lameter committed
757
758
759
760
761
762
763
764
765
				return 0;
			}
			break;
		}
		object = fp;
		fp = get_freepointer(s, object);
		nr++;
	}

766
	max_objects = (PAGE_SIZE << compound_order(page)) / s->size;
767
768
	if (max_objects > MAX_OBJS_PER_PAGE)
		max_objects = MAX_OBJS_PER_PAGE;
769
770
771
772
773
774
775

	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.");
	}
776
	if (page->inuse != page->objects - nr) {
777
		slab_err(s, page, "Wrong object count. Counter is %d but "
778
779
			"counted were %d", page->inuse, page->objects - nr);
		page->inuse = page->objects - nr;
780
		slab_fix(s, "Object count adjusted.");
Christoph Lameter's avatar
Christoph Lameter committed
781
782
783
784
	}
	return search == NULL;
}

785
786
static void trace(struct kmem_cache *s, struct page *page, void *object,
								int alloc)
Christoph Lameter's avatar
Christoph Lameter committed
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
{
	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();
	}
}

802
/*
Christoph Lameter's avatar
Christoph Lameter committed
803
 * Tracking of fully allocated slabs for debugging purposes.
804
 */
Christoph Lameter's avatar
Christoph Lameter committed
805
static void add_full(struct kmem_cache_node *n, struct page *page)
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
{
	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);
}

826
827
828
829
830
831
832
833
/* 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);
}

834
static inline void inc_slabs_node(struct kmem_cache *s, int node, int objects)
835
836
837
838
839
840
841
842
843
{
	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).
	 */
844
	if (!NUMA_BUILD || n) {
845
		atomic_long_inc(&n->nr_slabs);
846
847
		atomic_long_add(objects, &n->total_objects);
	}
848
}
849
static inline void dec_slabs_node(struct kmem_cache *s, int node, int objects)
850
851
852
853
{
	struct kmem_cache_node *n = get_node(s, node);

	atomic_long_dec(&n->nr_slabs);
854
	atomic_long_sub(objects, &n->total_objects);
855
856
857
}

/* Object debug checks for alloc/free paths */
Christoph Lameter's avatar
Christoph Lameter committed
858
859
860
861
862
863
864
865
866
867
868
static void setup_object_debug(struct kmem_cache *s, struct page *page,
								void *object)
{
	if (!(s->flags & (SLAB_STORE_USER|SLAB_RED_ZONE|__OBJECT_POISON)))
		return;

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

static int alloc_debug_processing(struct kmem_cache *s, struct page *page,
869
					void *object, unsigned long addr)
Christoph Lameter's avatar
Christoph Lameter committed
870
871
872
873
{
	if (!check_slab(s, page))
		goto bad;

874
	if (!on_freelist(s, page, object)) {
875
		object_err(s, page, object, "Object already allocated");
876
		goto bad;
Christoph Lameter's avatar
Christoph Lameter committed
877
878
879
880
	}

	if (!check_valid_pointer(s, page, object)) {
		object_err(s, page, object, "Freelist Pointer check fails");
881
		goto bad;
Christoph Lameter's avatar
Christoph Lameter committed
882
883
	}

884
	if (!check_object(s, page, object, 0))
Christoph Lameter's avatar
Christoph Lameter committed
885
886
		goto bad;

Christoph Lameter's avatar
Christoph Lameter committed
887
888
889
890
891
	/* 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
892
	return 1;
Christoph Lameter's avatar
Christoph Lameter committed
893

Christoph Lameter's avatar
Christoph Lameter committed
894
895
896
897
898
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
899
		 * as used avoids touching the remaining objects.
Christoph Lameter's avatar
Christoph Lameter committed
900
		 */
901
		slab_fix(s, "Marking all objects used");
902
		page->inuse = page->objects;
903
		page->freelist = NULL;
Christoph Lameter's avatar
Christoph Lameter committed
904
905
906
907
	}
	return 0;
}

Christoph Lameter's avatar
Christoph Lameter committed
908
static int free_debug_processing(struct kmem_cache *s, struct page *page,
909
					void *object, unsigned long addr)
Christoph Lameter's avatar
Christoph Lameter committed
910
911
912
913
914
{
	if (!check_slab(s, page))
		goto fail;

	if (!check_valid_pointer(s, page, object)) {
915
		slab_err(s, page, "Invalid object pointer 0x%p", object);
Christoph Lameter's avatar
Christoph Lameter committed
916
917
918
919
		goto fail;
	}

	if (on_freelist(s, page, object)) {
920
		object_err(s, page, object, "Object already free");
Christoph Lameter's avatar
Christoph Lameter committed
921
922
923
924
925
926
927
		goto fail;
	}

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

	if (unlikely(s != page->slab)) {
Ingo Molnar's avatar
Ingo Molnar committed
928
		if (!PageSlab(page)) {
929
930
			slab_err(s, page, "Attempt to free object(0x%p) "
				"outside of slab", object);
Ingo Molnar's avatar
Ingo Molnar committed
931
		} else if (!page->slab) {
Christoph Lameter's avatar
Christoph Lameter committed
932
			printk(KERN_ERR
933
				"SLUB <none>: no slab for object 0x%p.\n",
Christoph Lameter's avatar
Christoph Lameter committed
934
						object);
935
			dump_stack();
936
		} else
937
938
			object_err(s, page, object,
					"page slab pointer corrupt.");
Christoph Lameter's avatar
Christoph Lameter committed
939
940
		goto fail;
	}
Christoph Lameter's avatar
Christoph Lameter committed
941
942

	/* Special debug activities for freeing objects */
943
	if (!PageSlubFrozen(page) && !page->freelist)
Christoph Lameter's avatar
Christoph Lameter committed
944
945
946
947
948
		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
949
	return 1;
Christoph Lameter's avatar
Christoph Lameter committed
950

Christoph Lameter's avatar
Christoph Lameter committed
951
fail:
952
	slab_fix(s, "Object at 0x%p not freed", object);
Christoph Lameter's avatar
Christoph Lameter committed
953
954
955
	return 0;
}

956
957
static int __init setup_slub_debug(char *str)
{
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
	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
	 */
982
	for (; *str && *str != ','; str++) {
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
		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' "
1001
				"unknown. skipped\n", *str);
1002
		}
1003
1004
	}

1005
check_slabs:
1006
1007
	if (*str == ',')
		slub_debug_slabs = str + 1;
1008
out:
1009
1010
1011
1012
1013
	return 1;
}

__setup("slub_debug", setup_slub_debug);

1014
1015
static unsigned long kmem_cache_flags(unsigned long objsize,
	unsigned long flags, const char *name,
1016
	void (*ctor)(void *))
1017
1018
{
	/*
1019
	 * Enable debugging if selected on the kernel commandline.
1020
	 */
1021
1022
1023
	if (slub_debug && (!slub_debug_slabs ||
	    strncmp(slub_debug_slabs, name, strlen(slub_debug_slabs)) == 0))
			flags |= slub_debug;
1024
1025

	return flags;
1026
1027
}
#else
Christoph Lameter's avatar
Christoph Lameter committed
1028
1029
static inline void setup_object_debug(struct kmem_cache *s,
			struct page *page, void *object) {}
1030

Christoph Lameter's avatar
Christoph Lameter committed
1031
static inline int alloc_debug_processing(struct kmem_cache *s,
1032
	struct page *page, void *object, unsigned long addr) { return 0; }
1033

Christoph Lameter's avatar
Christoph Lameter committed
1034
static inline int free_debug_processing(struct kmem_cache *s,
1035
	struct page *page, void *object, unsigned long addr) { return 0; }
1036
1037
1038
1039
1040

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
1041
static inline void add_full(struct kmem_cache_node *n, struct page *page) {}
1042
1043
static inline unsigned long kmem_cache_flags(unsigned long objsize,
	unsigned long flags, const char *name,
1044
	void (*ctor)(void *))
1045
1046
1047
{
	return flags;
}
1048
#define slub_debug 0
1049
1050
1051

static inline unsigned long slabs_node(struct kmem_cache *s, int node)
							{ return 0; }
1052
1053
1054
1055
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) {}
1056
#endif
1057

Christoph Lameter's avatar
Christoph Lameter committed
1058
1059
1060
/*
 * Slab allocation and freeing
 */
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
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
1072
1073
static struct page *allocate_slab(struct kmem_cache *s, gfp_t flags, int node)
{
1074
	struct page *page;
1075
	struct kmem_cache_order_objects oo = s->oo;
Christoph Lameter's avatar
Christoph Lameter committed
1076

1077
	flags |= s->allocflags;
1078

1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
	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
1090

1091
1092
		stat(get_cpu_slab(s, raw_smp_processor_id()), ORDER_FALLBACK);
	}
1093
	page->objects = oo_objects(oo);
Christoph Lameter's avatar
Christoph Lameter committed
1094
1095
1096
	mod_zone_page_state(page_zone(page),
		(s->flags & SLAB_RECLAIM_ACCOUNT) ?
		NR_SLAB_RECLAIMABLE : NR_SLAB_UNRECLAIMABLE,
1097
		1 << oo_order(oo));
Christoph Lameter's avatar
Christoph Lameter committed
1098
1099
1100
1101
1102
1103
1104

	return page;
}

static void setup_object(struct kmem_cache *s, struct page *page,
				void *object)
{
Christoph Lameter's avatar
Christoph Lameter committed
1105
	setup_object_debug(s, page, object);
1106
	if (unlikely(s->ctor))
1107
		s->ctor(object);
Christoph Lameter's avatar
Christoph Lameter committed
1108
1109
1110
1111
1112
1113
1114
1115
1116
}

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
1117
	BUG_ON(flags & GFP_SLAB_BUG_MASK);
Christoph Lameter's avatar
Christoph Lameter committed
1118

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

1124
	inc_slabs_node(s, page_to_nid(page), page->objects);
Christoph Lameter's avatar
Christoph Lameter committed
1125
1126
1127
1128
	page->slab = s;
	page->flags |= 1 << PG_slab;
	if (s->flags & (SLAB_DEBUG_FREE | SLAB_RED_ZONE | SLAB_POISON |
			SLAB_STORE_USER | SLAB_TRACE))
1129
		__SetPageSlubDebug(page);
Christoph Lameter's avatar
Christoph Lameter committed
1130
1131
1132
1133

	start = page_address(page);

	if (unlikely(s->flags & SLAB_POISON))
1134
		memset(start, POISON_INUSE, PAGE_SIZE << compound_order(page));
Christoph Lameter's avatar
Christoph Lameter committed
1135
1136

	last = start;
1137
	for_each_object(p, s, start, page->objects) {
Christoph Lameter's avatar
Christoph Lameter committed
1138
1139
1140
1141
1142
		setup_object(s, page, last);
		set_freepointer(s, last, p);
		last = p;
	}
	setup_object(s, page, last);
1143
	set_freepointer(s, last, NULL);
Christoph Lameter's avatar
Christoph Lameter committed
1144
1145
1146
1147
1148
1149
1150
1151
1152

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

static void __free_slab(struct kmem_cache *s, struct page *page)
{
1153
1154
	int order = compound_order(page);
	int pages = 1 << order;
Christoph Lameter's avatar
Christoph Lameter committed
1155

1156
	if (unlikely(SLABDEBUG && PageSlubDebug(page))) {
Christoph Lameter's avatar
Christoph Lameter committed
1157
1158
1159
		void *p;

		slab_pad_check(s, page);
1160
1161
		for_each_object(p, s, page_address(page),
						page->objects)
Christoph Lameter's avatar
Christoph Lameter committed
1162
			check_object(s, page, p, 0);
1163
		__ClearPageSlubDebug(page);
Christoph Lameter's avatar
Christoph Lameter committed
1164
1165
1166
1167
1168
	}

	mod_zone_page_state(page_zone(page),
		(s->flags & SLAB_RECLAIM_ACCOUNT) ?
		NR_SLAB_RECLAIMABLE : NR_SLAB_UNRECLAIMABLE,
1169
		-pages);
Christoph Lameter's avatar
Christoph Lameter committed
1170

1171
1172
	__ClearPageSlab(page);
	reset_page_mapcount(page);
1173
	__free_pages(page, order);
Christoph Lameter's avatar
Christoph Lameter committed
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
}

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)
{
1199
	dec_slabs_node(s, page_to_nid(page), page->objects);
Christoph Lameter's avatar
Christoph Lameter committed
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
	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
1213
	__bit_spin_unlock(PG_locked, &page->flags);
Christoph Lameter's avatar
Christoph Lameter committed
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
}

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
 */
1227
1228
static void add_partial(struct kmem_cache_node *n,
				struct page *page, int tail)
Christoph Lameter's avatar
Christoph Lameter committed
1229
{
Christoph Lameter's avatar
Christoph Lameter committed
1230
1231
	spin_lock(&n->list_lock);
	n->nr_partial++;
1232
1233
1234
1235
	if (tail)
		list_add_tail(&page->lru, &n->partial);
	else
		list_add(&page->lru, &n->partial);
Christoph Lameter's avatar
Christoph Lameter committed
1236
1237
1238
	spin_unlock(&n->list_lock);
}

1239
static void remove_partial(struct kmem_cache *s, struct page *page)
Christoph Lameter's avatar
Christoph Lameter committed
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
{
	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
1250
 * Lock slab and remove from the partial list.
Christoph Lameter's avatar
Christoph Lameter committed
1251
 *
Christoph Lameter's avatar
Christoph Lameter committed
1252
 * Must hold list_lock.
Christoph Lameter's avatar
Christoph Lameter committed
1253
 */
1254
1255
static inline int lock_and_freeze_slab(struct kmem_cache_node *n,
							struct page *page)
Christoph Lameter's avatar
Christoph Lameter committed
1256
1257
1258
1259
{
	if (slab_trylock(page)) {
		list_del(&page->lru);
		n->nr_partial--;
1260
		__SetPageSlubFrozen(page);
Christoph Lameter's avatar
Christoph Lameter committed
1261
1262
1263
1264
1265
1266
		return 1;
	}
	return 0;
}

/*
Christoph Lameter's avatar
Christoph Lameter committed
1267
 * Try to allocate a partial slab from a specific node.
Christoph Lameter's avatar
Christoph Lameter committed
1268
1269
1270
1271
1272
1273
1274
1275
 */
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
1276
1277
	 * partial slab and there is none available then get_partials()
	 * will return NULL.
Christoph Lameter's avatar
Christoph Lameter committed
1278
1279
1280
1281
1282
1283
	 */
	if (!n || !n->nr_partial)
		return NULL;

	spin_lock(&n->list_lock);
	list_for_each_entry(page, &n->partial, lru)
1284
		if (lock_and_freeze_slab(n, page))
Christoph Lameter's avatar
Christoph Lameter committed
1285
1286
1287
1288
1289
1290
1291
1292
			goto out;
	page = NULL;
out:
	spin_unlock(&n->list_lock);
	return page;
}

/*
Christoph Lameter's avatar
Christoph Lameter committed
1293
 * Get a page from somewhere. Search in increasing NUMA distances.
Christoph Lameter's avatar
Christoph Lameter committed
1294
1295
1296
1297
1298
 */
static struct page *get_any_partial(struct kmem_cache *s, gfp_t flags)
{
#ifdef CONFIG_NUMA
	struct zonelist *zonelist;
1299
	struct zoneref *z;
1300
1301
	struct zone *zone;
	enum zone_type high_zoneidx = gfp_zone(flags);
Christoph Lameter's avatar
Christoph Lameter committed
1302
1303
1304
	struct page *page;

	/*
Christoph Lameter's avatar
Christoph Lameter committed
1305
1306
1307
1308
	 * 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
1309
	 *
Christoph Lameter's avatar
Christoph Lameter committed
1310
1311
1312
1313
	 * 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
1314
	 *
Christoph Lameter's avatar
Christoph Lameter committed
1315
	 * If /sys/kernel/slab/xx/defrag_ratio is set to 100 (which makes
Christoph Lameter's avatar
Christoph Lameter committed
1316
1317
1318
1319
1320
	 * 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.