slub.c 111 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
 */

#include <linux/mm.h>
Nick Piggin's avatar
Nick Piggin committed
12
#include <linux/swap.h> /* struct reclaim_state */
Christoph Lameter's avatar
Christoph Lameter committed
13
14
15
16
17
#include <linux/module.h>
#include <linux/bit_spinlock.h>
#include <linux/interrupt.h>
#include <linux/bitops.h>
#include <linux/slab.h>
18
#include <linux/proc_fs.h>
Christoph Lameter's avatar
Christoph Lameter committed
19
#include <linux/seq_file.h>
20
#include <linux/kmemtrace.h>
Vegard Nossum's avatar
Vegard Nossum committed
21
#include <linux/kmemcheck.h>
Christoph Lameter's avatar
Christoph Lameter committed
22
23
#include <linux/cpu.h>
#include <linux/cpuset.h>
24
#include <linux/kmemleak.h>
Christoph Lameter's avatar
Christoph Lameter committed
25
26
#include <linux/mempolicy.h>
#include <linux/ctype.h>
27
#include <linux/debugobjects.h>
Christoph Lameter's avatar
Christoph Lameter committed
28
#include <linux/kallsyms.h>
29
#include <linux/memory.h>
Roman Zippel's avatar
Roman Zippel committed
30
#include <linux/math64.h>
Akinobu Mita's avatar
Akinobu Mita committed
31
#include <linux/fault-inject.h>
Christoph Lameter's avatar
Christoph Lameter committed
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
75
76
77

/*
 * 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
78
79
 * 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
80
 * freed then the slab will show up again on the partial lists.
Christoph Lameter's avatar
Christoph Lameter committed
81
82
 * We track full slabs for debugging purposes though because otherwise we
 * cannot scan all objects.
Christoph Lameter's avatar
Christoph Lameter committed
83
84
85
86
87
88
89
 *
 * 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.
 *
90
91
92
93
94
95
96
97
98
99
100
101
 * 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
102
 * 			freelist that allows lockless access to
103
104
 * 			free objects in addition to the regular freelist
 * 			that requires the slab lock.
Christoph Lameter's avatar
Christoph Lameter committed
105
106
107
 *
 * PageError		Slab requires special handling due to debug
 * 			options set. This moves	slab handling out of
108
 * 			the fast path and disables lockless freelists.
Christoph Lameter's avatar
Christoph Lameter committed
109
110
 */

111
#ifdef CONFIG_SLUB_DEBUG
112
#define SLABDEBUG 1
113
114
115
116
#else
#define SLABDEBUG 0
#endif

Christoph Lameter's avatar
Christoph Lameter committed
117
118
119
120
121
122
123
124
125
126
127
/*
 * 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

128
129
130
131
/*
 * Mininum number of partial slabs. These will be left on the partial
 * lists even if they are empty. kmem_cache_shrink may reclaim them.
 */
132
#define MIN_PARTIAL 5
Christoph Lameter's avatar
Christoph Lameter committed
133

134
135
136
137
138
139
140
/*
 * 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
141
142
#define DEBUG_DEFAULT_FLAGS (SLAB_DEBUG_FREE | SLAB_RED_ZONE | \
				SLAB_POISON | SLAB_STORE_USER)
Christoph Lameter's avatar
Christoph Lameter committed
143

Christoph Lameter's avatar
Christoph Lameter committed
144
145
146
147
/*
 * Set of flags that will prevent slab merging
 */
#define SLUB_NEVER_MERGE (SLAB_RED_ZONE | SLAB_POISON | SLAB_STORE_USER | \
148
		SLAB_TRACE | SLAB_DESTROY_BY_RCU | SLAB_NOLEAKTRACE)
Christoph Lameter's avatar
Christoph Lameter committed
149
150

#define SLUB_MERGE_SAME (SLAB_DEBUG_FREE | SLAB_RECLAIM_ACCOUNT | \
Vegard Nossum's avatar
Vegard Nossum committed
151
		SLAB_CACHE_DMA | SLAB_NOTRACK)
Christoph Lameter's avatar
Christoph Lameter committed
152
153

#ifndef ARCH_KMALLOC_MINALIGN
154
#define ARCH_KMALLOC_MINALIGN __alignof__(unsigned long long)
Christoph Lameter's avatar
Christoph Lameter committed
155
156
157
#endif

#ifndef ARCH_SLAB_MINALIGN
158
#define ARCH_SLAB_MINALIGN __alignof__(unsigned long long)
Christoph Lameter's avatar
Christoph Lameter committed
159
160
#endif

161
162
163
164
#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
165
/* Internal SLUB flags */
166
167
#define __OBJECT_POISON		0x80000000 /* Poison object */
#define __SYSFS_ADD_DEFERRED	0x40000000 /* Not yet visible via sysfs */
Christoph Lameter's avatar
Christoph Lameter committed
168
169
170
171
172
173
174
175
176
177

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
178
	UP,		/* Everything works but does not show up in sysfs */
Christoph Lameter's avatar
Christoph Lameter committed
179
180
181
182
183
	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
184
static LIST_HEAD(slab_caches);
Christoph Lameter's avatar
Christoph Lameter committed
185

186
187
188
189
/*
 * Tracking user of a slab.
 */
struct track {
190
	unsigned long addr;	/* Called from address */
191
192
193
194
195
196
197
	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 };

198
#ifdef CONFIG_SLUB_DEBUG
Christoph Lameter's avatar
Christoph Lameter committed
199
200
201
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 *);
202

Christoph Lameter's avatar
Christoph Lameter committed
203
#else
204
205
206
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; }
207
208
209
210
static inline void sysfs_slab_remove(struct kmem_cache *s)
{
	kfree(s);
}
211

Christoph Lameter's avatar
Christoph Lameter committed
212
213
#endif

214
215
216
217
218
219
220
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
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
/********************************************************************
 * 			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
}

239
240
static inline struct kmem_cache_cpu *get_cpu_slab(struct kmem_cache *s, int cpu)
{
241
242
243
244
245
#ifdef CONFIG_SMP
	return s->cpu_slab[cpu];
#else
	return &s->cpu_slab;
#endif
246
247
}

Christoph Lameter's avatar
Christoph Lameter committed
248
/* Verify that a pointer has an address that is valid within a slab page */
249
250
251
252
253
static inline int check_valid_pointer(struct kmem_cache *s,
				struct page *page, const void *object)
{
	void *base;

254
	if (!object)
255
256
		return 1;

257
	base = page_address(page);
258
	if (object < base || object >= base + page->objects * s->size ||
259
260
261
262
263
264
265
		(object - base) % s->size) {
		return 0;
	}

	return 1;
}

266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
/*
 * 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 */
284
285
#define for_each_object(__p, __s, __addr, __objects) \
	for (__p = (__addr); __p < (__addr) + (__objects) * (__s)->size;\
286
287
288
289
			__p += (__s)->size)

/* Scan freelist */
#define for_each_free_object(__p, __s, __free) \
290
	for (__p = (__free); __p; __p = get_freepointer((__s), __p))
291
292
293
294
295
296
297

/* 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;
}

298
299
300
301
static inline struct kmem_cache_order_objects oo_make(int order,
						unsigned long size)
{
	struct kmem_cache_order_objects x = {
302
		(order << OO_SHIFT) + (PAGE_SIZE << order) / size
303
304
305
306
307
308
309
	};

	return x;
}

static inline int oo_order(struct kmem_cache_order_objects x)
{
310
	return x.x >> OO_SHIFT;
311
312
313
314
}

static inline int oo_objects(struct kmem_cache_order_objects x)
{
315
	return x.x & OO_MASK;
316
317
}

318
319
320
321
#ifdef CONFIG_SLUB_DEBUG
/*
 * Debug settings:
 */
322
323
324
#ifdef CONFIG_SLUB_DEBUG_ON
static int slub_debug = DEBUG_DEFAULT_FLAGS;
#else
325
static int slub_debug;
326
#endif
327
328
329

static char *slub_debug_slabs;

Christoph Lameter's avatar
Christoph Lameter committed
330
331
332
333
334
335
336
337
338
339
340
341
342
/*
 * 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) {
343
			printk(KERN_ERR "%8s 0x%p: ", text, addr + i);
Christoph Lameter's avatar
Christoph Lameter committed
344
345
			newline = 0;
		}
346
		printk(KERN_CONT " %02x", addr[i]);
Christoph Lameter's avatar
Christoph Lameter committed
347
348
349
		offset = i % 16;
		ascii[offset] = isgraph(addr[i]) ? addr[i] : '.';
		if (offset == 15) {
350
			printk(KERN_CONT " %s\n", ascii);
Christoph Lameter's avatar
Christoph Lameter committed
351
352
353
354
355
356
			newline = 1;
		}
	}
	if (!newline) {
		i %= 16;
		while (i < 16) {
357
			printk(KERN_CONT "   ");
Christoph Lameter's avatar
Christoph Lameter committed
358
359
360
			ascii[i] = ' ';
			i++;
		}
361
		printk(KERN_CONT " %s\n", ascii);
Christoph Lameter's avatar
Christoph Lameter committed
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
	}
}

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,
379
			enum track_item alloc, unsigned long addr)
Christoph Lameter's avatar
Christoph Lameter committed
380
{
Akinobu Mita's avatar
Akinobu Mita committed
381
	struct track *p = get_track(s, object, alloc);
Christoph Lameter's avatar
Christoph Lameter committed
382
383
384
385

	if (addr) {
		p->addr = addr;
		p->cpu = smp_processor_id();
386
		p->pid = current->pid;
Christoph Lameter's avatar
Christoph Lameter committed
387
388
389
390
391
392
393
		p->when = jiffies;
	} else
		memset(p, 0, sizeof(struct track));
}

static void init_tracking(struct kmem_cache *s, void *object)
{
394
395
396
	if (!(s->flags & SLAB_STORE_USER))
		return;

397
398
	set_track(s, object, TRACK_FREE, 0UL);
	set_track(s, object, TRACK_ALLOC, 0UL);
Christoph Lameter's avatar
Christoph Lameter committed
399
400
401
402
403
404
405
}

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

406
	printk(KERN_ERR "INFO: %s in %pS age=%lu cpu=%u pid=%d\n",
407
		s, (void *)t->addr, jiffies - t->when, t->cpu, t->pid);
408
409
410
411
412
413
414
415
416
417
418
419
420
}

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)
{
421
422
	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);
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438

}

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
439
440
}

441
442
443
444
445
446
447
448
449
450
451
452
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
453
454
{
	unsigned int off;	/* Offset of last byte */
455
	u8 *addr = page_address(page);
456
457
458
459
460
461
462
463
464
465
466

	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);

467
	print_section("Object", p, min_t(unsigned long, s->objsize, PAGE_SIZE));
Christoph Lameter's avatar
Christoph Lameter committed
468
469
470
471
472
473
474
475
476
477

	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;

478
	if (s->flags & SLAB_STORE_USER)
Christoph Lameter's avatar
Christoph Lameter committed
479
480
481
482
		off += 2 * sizeof(struct track);

	if (off != s->size)
		/* Beginning of the filler is the free pointer */
483
484
485
		print_section("Padding", p + off, s->size - off);

	dump_stack();
Christoph Lameter's avatar
Christoph Lameter committed
486
487
488
489
490
}

static void object_err(struct kmem_cache *s, struct page *page,
			u8 *object, char *reason)
{
491
	slab_bug(s, "%s", reason);
492
	print_trailer(s, page, object);
Christoph Lameter's avatar
Christoph Lameter committed
493
494
}

495
static void slab_err(struct kmem_cache *s, struct page *page, char *fmt, ...)
Christoph Lameter's avatar
Christoph Lameter committed
496
497
498
499
{
	va_list args;
	char buf[100];

500
501
	va_start(args, fmt);
	vsnprintf(buf, sizeof(buf), fmt, args);
Christoph Lameter's avatar
Christoph Lameter committed
502
	va_end(args);
503
	slab_bug(s, "%s", buf);
504
	print_page_info(page);
Christoph Lameter's avatar
Christoph Lameter committed
505
506
507
508
509
510
511
512
513
	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);
514
		p[s->objsize - 1] = POISON_END;
Christoph Lameter's avatar
Christoph Lameter committed
515
516
517
518
519
520
521
522
	}

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

523
static u8 *check_bytes(u8 *start, unsigned int value, unsigned int bytes)
Christoph Lameter's avatar
Christoph Lameter committed
524
525
526
{
	while (bytes) {
		if (*start != (u8)value)
527
			return start;
Christoph Lameter's avatar
Christoph Lameter committed
528
529
530
		start++;
		bytes--;
	}
531
532
533
534
535
536
537
538
539
540
541
542
	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,
543
			u8 *start, unsigned int value, unsigned int bytes)
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
{
	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
563
564
565
566
567
568
569
570
571
}

/*
 * 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
572
 *
Christoph Lameter's avatar
Christoph Lameter committed
573
574
575
576
577
 * 	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
578
579
580
 * 	Padding is extended by another word if Redzoning is enabled and
 * 	objsize == inuse.
 *
Christoph Lameter's avatar
Christoph Lameter committed
581
582
583
584
 * 	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
585
586
 * 	Meta data starts here.
 *
Christoph Lameter's avatar
Christoph Lameter committed
587
588
 * 	A. Free pointer (if we cannot overwrite object on free)
 * 	B. Tracking data for SLAB_STORE_USER
Christoph Lameter's avatar
Christoph Lameter committed
589
 * 	C. Padding to reach required alignment boundary or at mininum
Christoph Lameter's avatar
Christoph Lameter committed
590
 * 		one word if debugging is on to be able to detect writes
Christoph Lameter's avatar
Christoph Lameter committed
591
592
593
 * 		before the word boundary.
 *
 *	Padding is done using 0x5a (POISON_INUSE)
Christoph Lameter's avatar
Christoph Lameter committed
594
595
 *
 * object + s->size
Christoph Lameter's avatar
Christoph Lameter committed
596
 * 	Nothing is used beyond s->size.
Christoph Lameter's avatar
Christoph Lameter committed
597
 *
Christoph Lameter's avatar
Christoph Lameter committed
598
599
 * 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
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
 * 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;

618
619
	return check_bytes_and_report(s, page, p, "Object padding",
				p + off, POISON_INUSE, s->size - off);
Christoph Lameter's avatar
Christoph Lameter committed
620
621
}

622
/* Check the pad bytes at the end of a slab page */
Christoph Lameter's avatar
Christoph Lameter committed
623
624
static int slab_pad_check(struct kmem_cache *s, struct page *page)
{
625
626
627
628
629
	u8 *start;
	u8 *fault;
	u8 *end;
	int length;
	int remainder;
Christoph Lameter's avatar
Christoph Lameter committed
630
631
632
633

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

634
	start = page_address(page);
635
	length = (PAGE_SIZE << compound_order(page));
636
637
	end = start + length;
	remainder = length % s->size;
Christoph Lameter's avatar
Christoph Lameter committed
638
639
640
	if (!remainder)
		return 1;

641
	fault = check_bytes(end - remainder, POISON_INUSE, remainder);
642
643
644
645
646
647
	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);
648
	print_section("Padding", end - remainder, remainder);
649
650
651

	restore_bytes(s, "slab padding", POISON_INUSE, start, end);
	return 0;
Christoph Lameter's avatar
Christoph Lameter committed
652
653
654
655
656
657
658
659
660
661
662
663
}

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;

664
665
		if (!check_bytes_and_report(s, page, object, "Redzone",
			endobject, red, s->inuse - s->objsize))
Christoph Lameter's avatar
Christoph Lameter committed
666
667
			return 0;
	} else {
Ingo Molnar's avatar
Ingo Molnar committed
668
669
670
671
		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
672
673
674
675
	}

	if (s->flags & SLAB_POISON) {
		if (!active && (s->flags & __OBJECT_POISON) &&
676
677
678
			(!check_bytes_and_report(s, page, p, "Poison", p,
					POISON_FREE, s->objsize - 1) ||
			 !check_bytes_and_report(s, page, p, "Poison",
679
				p + s->objsize - 1, POISON_END, 1)))
Christoph Lameter's avatar
Christoph Lameter committed
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
			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
698
		 * No choice but to zap it and thus lose the remainder
Christoph Lameter's avatar
Christoph Lameter committed
699
		 * of the free objects in this slab. May cause
Christoph Lameter's avatar
Christoph Lameter committed
700
		 * another error because the object count is now wrong.
Christoph Lameter's avatar
Christoph Lameter committed
701
		 */
702
		set_freepointer(s, p, NULL);
Christoph Lameter's avatar
Christoph Lameter committed
703
704
705
706
707
708
709
		return 0;
	}
	return 1;
}

static int check_slab(struct kmem_cache *s, struct page *page)
{
710
711
	int maxobj;

Christoph Lameter's avatar
Christoph Lameter committed
712
713
714
	VM_BUG_ON(!irqs_disabled());

	if (!PageSlab(page)) {
715
		slab_err(s, page, "Not a valid slab page");
Christoph Lameter's avatar
Christoph Lameter committed
716
717
		return 0;
	}
718
719
720
721
722
723
724
725

	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) {
726
		slab_err(s, page, "inuse %u > max %u",
727
			s->name, page->inuse, page->objects);
Christoph Lameter's avatar
Christoph Lameter committed
728
729
730
731
732
733
734
735
		return 0;
	}
	/* Slab_pad_check fixes things up after itself */
	slab_pad_check(s, page);
	return 1;
}

/*
Christoph Lameter's avatar
Christoph Lameter committed
736
737
 * 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
738
739
740
741
742
743
 */
static int on_freelist(struct kmem_cache *s, struct page *page, void *search)
{
	int nr = 0;
	void *fp = page->freelist;
	void *object = NULL;
744
	unsigned long max_objects;
Christoph Lameter's avatar
Christoph Lameter committed
745

746
	while (fp && nr <= page->objects) {
Christoph Lameter's avatar
Christoph Lameter committed
747
748
749
750
751
752
		if (fp == search)
			return 1;
		if (!check_valid_pointer(s, page, fp)) {
			if (object) {
				object_err(s, page, object,
					"Freechain corrupt");
753
				set_freepointer(s, object, NULL);
Christoph Lameter's avatar
Christoph Lameter committed
754
755
				break;
			} else {
756
				slab_err(s, page, "Freepointer corrupt");
757
				page->freelist = NULL;
758
				page->inuse = page->objects;
759
				slab_fix(s, "Freelist cleared");
Christoph Lameter's avatar
Christoph Lameter committed
760
761
762
763
764
765
766
767
768
				return 0;
			}
			break;
		}
		object = fp;
		fp = get_freepointer(s, object);
		nr++;
	}

769
	max_objects = (PAGE_SIZE << compound_order(page)) / s->size;
770
771
	if (max_objects > MAX_OBJS_PER_PAGE)
		max_objects = MAX_OBJS_PER_PAGE;
772
773
774
775
776
777
778

	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.");
	}
779
	if (page->inuse != page->objects - nr) {
780
		slab_err(s, page, "Wrong object count. Counter is %d but "
781
782
			"counted were %d", page->inuse, page->objects - nr);
		page->inuse = page->objects - nr;
783
		slab_fix(s, "Object count adjusted.");
Christoph Lameter's avatar
Christoph Lameter committed
784
785
786
787
	}
	return search == NULL;
}

788
789
static void trace(struct kmem_cache *s, struct page *page, void *object,
								int alloc)
Christoph Lameter's avatar
Christoph Lameter committed
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
{
	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();
	}
}

805
/*
Christoph Lameter's avatar
Christoph Lameter committed
806
 * Tracking of fully allocated slabs for debugging purposes.
807
 */
Christoph Lameter's avatar
Christoph Lameter committed
808
static void add_full(struct kmem_cache_node *n, struct page *page)
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
{
	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);
}

829
830
831
832
833
834
835
836
/* 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);
}

837
838
839
840
841
static inline unsigned long node_nr_slabs(struct kmem_cache_node *n)
{
	return atomic_long_read(&n->nr_slabs);
}

842
static inline void inc_slabs_node(struct kmem_cache *s, int node, int objects)
843
844
845
846
847
848
849
850
851
{
	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).
	 */
852
	if (!NUMA_BUILD || n) {
853
		atomic_long_inc(&n->nr_slabs);
854
855
		atomic_long_add(objects, &n->total_objects);
	}
856
}
857
static inline void dec_slabs_node(struct kmem_cache *s, int node, int objects)
858
859
860
861
{
	struct kmem_cache_node *n = get_node(s, node);

	atomic_long_dec(&n->nr_slabs);
862
	atomic_long_sub(objects, &n->total_objects);
863
864
865
}

/* Object debug checks for alloc/free paths */
Christoph Lameter's avatar
Christoph Lameter committed
866
867
868
869
870
871
872
873
874
875
876
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,
877
					void *object, unsigned long addr)
Christoph Lameter's avatar
Christoph Lameter committed
878
879
880
881
{
	if (!check_slab(s, page))
		goto bad;

882
	if (!on_freelist(s, page, object)) {
883
		object_err(s, page, object, "Object already allocated");
884
		goto bad;
Christoph Lameter's avatar
Christoph Lameter committed
885
886
887
888
	}

	if (!check_valid_pointer(s, page, object)) {
		object_err(s, page, object, "Freelist Pointer check fails");
889
		goto bad;
Christoph Lameter's avatar
Christoph Lameter committed
890
891
	}

892
	if (!check_object(s, page, object, 0))
Christoph Lameter's avatar
Christoph Lameter committed
893
894
		goto bad;

Christoph Lameter's avatar
Christoph Lameter committed
895
896
897
898
899
	/* 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
900
	return 1;
Christoph Lameter's avatar
Christoph Lameter committed
901

Christoph Lameter's avatar
Christoph Lameter committed
902
903
904
905
906
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
907
		 * as used avoids touching the remaining objects.
Christoph Lameter's avatar
Christoph Lameter committed
908
		 */
909
		slab_fix(s, "Marking all objects used");
910
		page->inuse = page->objects;
911
		page->freelist = NULL;
Christoph Lameter's avatar
Christoph Lameter committed
912
913
914
915
	}
	return 0;
}

Christoph Lameter's avatar
Christoph Lameter committed
916
static int free_debug_processing(struct kmem_cache *s, struct page *page,
917
					void *object, unsigned long addr)
Christoph Lameter's avatar
Christoph Lameter committed
918
919
920
921
922
{
	if (!check_slab(s, page))
		goto fail;

	if (!check_valid_pointer(s, page, object)) {
923
		slab_err(s, page, "Invalid object pointer 0x%p", object);
Christoph Lameter's avatar
Christoph Lameter committed
924
925
926
927
		goto fail;
	}

	if (on_freelist(s, page, object)) {
928
		object_err(s, page, object, "Object already free");
Christoph Lameter's avatar
Christoph Lameter committed
929
930
931
932
933
934
935
		goto fail;
	}

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

	if (unlikely(s != page->slab)) {
Ingo Molnar's avatar
Ingo Molnar committed
936
		if (!PageSlab(page)) {
937
938
			slab_err(s, page, "Attempt to free object(0x%p) "
				"outside of slab", object);
Ingo Molnar's avatar
Ingo Molnar committed
939
		} else if (!page->slab) {
Christoph Lameter's avatar
Christoph Lameter committed
940
			printk(KERN_ERR
941
				"SLUB <none>: no slab for object 0x%p.\n",
Christoph Lameter's avatar
Christoph Lameter committed
942
						object);
943
			dump_stack();
944
		} else
945
946
			object_err(s, page, object,
					"page slab pointer corrupt.");
Christoph Lameter's avatar
Christoph Lameter committed
947
948
		goto fail;
	}
Christoph Lameter's avatar
Christoph Lameter committed
949
950

	/* Special debug activities for freeing objects */
951
	if (!PageSlubFrozen(page) && !page->freelist)
Christoph Lameter's avatar
Christoph Lameter committed
952
953
954
955
956
		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
957
	return 1;
Christoph Lameter's avatar
Christoph Lameter committed
958

Christoph Lameter's avatar
Christoph Lameter committed
959
fail:
960
	slab_fix(s, "Object at 0x%p not freed", object);
Christoph Lameter's avatar
Christoph Lameter committed
961
962
963
	return 0;
}

964
965
static int __init setup_slub_debug(char *str)
{
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
	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
	 */
990
	for (; *str && *str != ','; str++) {
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
		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' "
1009
				"unknown. skipped\n", *str);
1010
		}
1011
1012
	}

1013
check_slabs:
1014
1015
	if (*str == ',')
		slub_debug_slabs = str + 1;
1016
out:
1017
1018
1019
1020
1021
	return 1;
}

__setup("slub_debug", setup_slub_debug);

1022
1023
static unsigned long kmem_cache_flags(unsigned long objsize,
	unsigned long flags, const char *name,
1024
	void (*ctor)(void *))
1025
1026
{
	/*
1027
	 * Enable debugging if selected on the kernel commandline.
1028
	 */
1029
1030
1031
	if (slub_debug && (!slub_debug_slabs ||
	    strncmp(slub_debug_slabs, name, strlen(slub_debug_slabs)) == 0))
			flags |= slub_debug;
1032
1033

	return flags;
1034
1035
}
#else
Christoph Lameter's avatar
Christoph Lameter committed
1036
1037
static inline void setup_object_debug(struct kmem_cache *s,
			struct page *page, void *object) {}
1038

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

Christoph Lameter's avatar
Christoph Lameter committed
1042
static inline int free_debug_processing(struct kmem_cache *s,
1043
	struct page *page, void *object, unsigned long addr) { return 0; }
1044
1045
1046
1047
1048

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
1049
static inline void add_full(struct kmem_cache_node *n, struct page *page) {}
1050
1051
static inline unsigned long kmem_cache_flags(unsigned long objsize,
	unsigned long flags, const char *name,
1052
	void (*ctor)(void *))
1053
1054
1055
{
	return flags;
}
1056
#define slub_debug 0
1057
1058
1059

static inline unsigned long slabs_node(struct kmem_cache *s, int node)
							{ return 0; }
1060
1061
static inline unsigned long node_nr_slabs(struct kmem_cache_node *n)
							{ return 0; }
1062
1063
1064
1065
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) {}
1066
#endif
1067

Christoph Lameter's avatar
Christoph Lameter committed
1068
1069
1070
/*
 * Slab allocation and freeing
 */
1071
1072
1073
1074
1075
static inline struct page *alloc_slab_page(gfp_t flags, int node,
					struct kmem_cache_order_objects oo)
{
	int order = oo_order(oo);

1076
1077
	flags |= __GFP_NOTRACK;

1078
1079
1080
1081
1082
1083
	if (node == -1)
		return alloc_pages(flags, order);
	else
		return alloc_pages_node(node, flags, order);
}

Christoph Lameter's avatar
Christoph Lameter committed
1084
1085
static struct page *allocate_slab(struct kmem_cache *s, gfp_t flags, int node)
{
1086
	struct page *page;
1087
	struct kmem_cache_order_objects oo = s->oo;
Christoph Lameter's avatar
Christoph Lameter committed
1088

1089
	flags |= s->allocflags;
1090

1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
	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
1102

1103
1104
		stat(get_cpu_slab(s, raw_smp_processor_id()), ORDER_FALLBACK);
	}
Vegard Nossum's avatar
Vegard Nossum committed
1105
1106
1107
1108

	if (kmemcheck_enabled
		&& !(s->flags & (SLAB_NOTRACK | DEBUG_DEFAULT_FLAGS)))
	{
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
		int pages = 1 << oo_order(oo);

		kmemcheck_alloc_shadow(page, oo_order(oo), flags, node);

		/*
		 * Objects from caches that have a constructor don't get
		 * cleared when they're allocated, so we need to do it here.
		 */
		if (s->ctor)
			kmemcheck_mark_uninitialized_pages(page, pages);
		else
			kmemcheck_mark_unallocated_pages(page, pages);
Vegard Nossum's avatar
Vegard Nossum committed
1121
1122
	}

1123
	page->objects = oo_objects(oo);
Christoph Lameter's avatar
Christoph Lameter committed
1124
1125
1126
	mod_zone_page_state(page_zone(page),
		(s->flags & SLAB_RECLAIM_ACCOUNT) ?
		NR_SLAB_RECLAIMABLE : NR_SLAB_UNRECLAIMABLE,
1127
		1 << oo_order(oo));
Christoph Lameter's avatar
Christoph Lameter committed
1128
1129
1130
1131
1132
1133
1134

	return page;
}

static void setup_object(struct kmem_cache *s, struct page *page,
				void *object)
{
Christoph Lameter's avatar
Christoph Lameter committed
1135
	setup_object_debug(s, page, object);
1136
	if (unlikely(s->ctor))
1137
		s->ctor(object);
Christoph Lameter's avatar
Christoph Lameter committed
1138
1139
1140
1141
1142
1143
1144
1145
1146
}

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
1147
	BUG_ON(flags & GFP_SLAB_BUG_MASK);
Christoph Lameter's avatar
Christoph Lameter committed
1148

Christoph Lameter's avatar
Christoph Lameter committed
1149
1150
	page = allocate_slab(s,
		flags & (GFP_RECLAIM_MASK | GFP_CONSTRAINT_MASK), node);
Christoph Lameter's avatar
Christoph Lameter committed
1151
1152
1153
	if (!page)
		goto out;

1154
	inc_slabs_node(s, page_to_nid(page), page->objects);
Christoph Lameter's avatar
Christoph Lameter committed
1155
1156
1157
1158
	page->slab = s;
	page->flags |= 1 << PG_slab;
	if (s->flags & (SLAB_DEBUG_FREE | SLAB_RED_ZONE | SLAB_POISON |
			SLAB_STORE_USER | SLAB_TRACE))
1159
		__SetPageSlubDebug(page);
Christoph Lameter's avatar
Christoph Lameter committed
1160
1161
1162
1163

	start = page_address(page);

	if (unlikely(s->flags & SLAB_POISON))
1164
		memset(start, POISON_INUSE, PAGE_SIZE << compound_order(page));
Christoph Lameter's avatar
Christoph Lameter committed
1165
1166

	last = start;
1167
	for_each_object(p, s, start, page->objects) {
Christoph Lameter's avatar
Christoph Lameter committed
1168
1169
1170
1171
1172
		setup_object(s, page, last);
		set_freepointer(s, last, p);
		last = p;
	}
	setup_object(s, page, last);
1173
	set_freepointer(s, last, NULL);
Christoph Lameter's avatar
Christoph Lameter committed
1174
1175
1176
1177
1178
1179
1180
1181
1182

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

static void __free_slab(struct kmem_cache *s, struct page *page)
{
1183
1184
	int order = compound_order(page);
	int pages = 1 << order;
Christoph Lameter's avatar
Christoph Lameter committed
1185

1186
	if (unlikely(SLABDEBUG && PageSlubDebug(page))) {
Christoph Lameter's avatar
Christoph Lameter committed
1187
1188
1189
		void *p;

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

1196
	kmemcheck_free_shadow(page, compound_order(page));
Vegard Nossum's avatar
Vegard Nossum committed
1197

Christoph Lameter's avatar
Christoph Lameter committed
1198
1199
1200
	mod_zone_page_state(page_zone(page),
		(s->flags & SLAB_RECLAIM_ACCOUNT) ?
		NR_SLAB_RECLAIMABLE : NR_SLAB_UNRECLAIMABLE,
1201
		-pages);
Christoph Lameter's avatar
Christoph Lameter committed
1202

1203
1204
	__ClearPageSlab(page);
	reset_page_mapcount(page);
Nick Piggin's avatar
Nick Piggin committed
1205
1206
	if (current->reclaim_state)
		current->reclaim_state->reclaimed_slab += pages;
1207
	__free_pages(page, order);
Christoph Lameter's avatar
Christoph Lameter committed
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
}

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)
{
1233
	dec_slabs_node(s, page_to_nid(page), page->objects);
Christoph Lameter's avatar
Christoph Lameter committed
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
	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
1247
	__bit_spin_unlock(PG_locked, &page->flags);
Christoph Lameter's avatar
Christoph Lameter committed
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
}

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
 */
1261
1262
static void add_partial(struct kmem_cache_node *n,
				struct page *page, int tail)
Christoph Lameter's avatar
Christoph Lameter committed
1263
{
Christoph Lameter's avatar
Christoph Lameter committed
1264
1265
	spin_lock(&n->list_lock);
	n->nr_partial++;
1266
1267
1268
1269
	if (tail)
		list_add_tail(&page->lru, &n->partial);
	else
		list_add(&page->lru, &n->partial);
Christoph Lameter's avatar
Christoph Lameter committed
1270
1271
1272
	spin_unlock(&n->list_lock);
}

1273
static void remove_partial(struct kmem_cache *s, struct page *page)
Christoph Lameter's avatar
Christoph Lameter committed
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
{
	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
1284
 * Lock slab and remove from the partial list.
Christoph Lameter's avatar
Christoph Lameter committed
1285
 *
Christoph Lameter's avatar
Christoph Lameter committed
1286
 * Must hold list_lock.
Christoph Lameter's avatar
Christoph Lameter committed
1287
 */
1288
1289
static inline int