slub.c 119 KB
Newer Older
Christoph Lameter's avatar
Christoph Lameter committed
1
2
3
4
/*
 * SLUB: A slab allocator that limits cache line use instead of queuing
 * objects in per cpu and per node lists.
 *
5
6
 * The allocator synchronizes using per slab locks or atomic operatios
 * and only uses a centralized lock to manage a pool of partial slabs.
Christoph Lameter's avatar
Christoph Lameter committed
7
 *
Christoph Lameter's avatar
Christoph Lameter committed
8
 * (C) 2007 SGI, Christoph Lameter
9
 * (C) 2011 Linux Foundation, Christoph Lameter
Christoph Lameter's avatar
Christoph Lameter committed
10
11
12
 */

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

32
33
#include <trace/events/kmem.h>

Christoph Lameter's avatar
Christoph Lameter committed
34
35
/*
 * Lock order:
36
37
38
 *   1. slub_lock (Global Semaphore)
 *   2. node->list_lock
 *   3. slab_lock(page) (Only on some arches and for debugging)
Christoph Lameter's avatar
Christoph Lameter committed
39
 *
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
 *   slub_lock
 *
 *   The role of the slub_lock is to protect the list of all the slabs
 *   and to synchronize major metadata changes to slab cache structures.
 *
 *   The slab_lock is only used for debugging and on arches that do not
 *   have the ability to do a cmpxchg_double. It only protects the second
 *   double word in the page struct. Meaning
 *	A. page->freelist	-> List of object free in a page
 *	B. page->counters	-> Counters of objects
 *	C. page->frozen		-> frozen state
 *
 *   If a slab is frozen then it is exempt from list management. It is not
 *   on any list. The processor that froze the slab is the one who can
 *   perform list operations on the page. Other processors may put objects
 *   onto the freelist but the processor that froze the slab is the only
 *   one that can retrieve the objects from the page's freelist.
Christoph Lameter's avatar
Christoph Lameter committed
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
 *
 *   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.
 *   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
77
78
 * 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
79
 * freed then the slab will show up again on the partial lists.
Christoph Lameter's avatar
Christoph Lameter committed
80
81
 * We track full slabs for debugging purposes though because otherwise we
 * cannot scan all objects.
Christoph Lameter's avatar
Christoph Lameter committed
82
83
84
85
86
87
88
 *
 * 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.
 *
89
90
91
92
93
94
95
96
97
98
99
100
 * 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
101
 * 			freelist that allows lockless access to
102
103
 * 			free objects in addition to the regular freelist
 * 			that requires the slab lock.
Christoph Lameter's avatar
Christoph Lameter committed
104
105
106
 *
 * PageError		Slab requires special handling due to debug
 * 			options set. This moves	slab handling out of
107
 * 			the fast path and disables lockless freelists.
Christoph Lameter's avatar
Christoph Lameter committed
108
109
 */

110
111
112
113
114
#define SLAB_DEBUG_FLAGS (SLAB_RED_ZONE | SLAB_POISON | SLAB_STORE_USER | \
		SLAB_TRACE | SLAB_DEBUG_FREE)

static inline int kmem_cache_debug(struct kmem_cache *s)
{
115
#ifdef CONFIG_SLUB_DEBUG
116
	return unlikely(s->flags & SLAB_DEBUG_FLAGS);
117
#else
118
	return 0;
119
#endif
120
}
121

Christoph Lameter's avatar
Christoph Lameter committed
122
123
124
125
126
127
128
129
130
131
132
/*
 * 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

133
134
135
/* Enable to log cmpxchg failures */
#undef SLUB_DEBUG_CMPXCHG

136
137
138
139
/*
 * Mininum number of partial slabs. These will be left on the partial
 * lists even if they are empty. kmem_cache_shrink may reclaim them.
 */
140
#define MIN_PARTIAL 5
Christoph Lameter's avatar
Christoph Lameter committed
141

142
143
144
145
146
147
148
/*
 * 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
149
150
#define DEBUG_DEFAULT_FLAGS (SLAB_DEBUG_FREE | SLAB_RED_ZONE | \
				SLAB_POISON | SLAB_STORE_USER)
Christoph Lameter's avatar
Christoph Lameter committed
151

152
/*
153
154
155
 * Debugging flags that require metadata to be stored in the slab.  These get
 * disabled when slub_debug=O is used and a cache's min order increases with
 * metadata.
156
 */
157
#define DEBUG_METADATA_FLAGS (SLAB_RED_ZONE | SLAB_POISON | SLAB_STORE_USER)
158

Christoph Lameter's avatar
Christoph Lameter committed
159
160
161
162
/*
 * Set of flags that will prevent slab merging
 */
#define SLUB_NEVER_MERGE (SLAB_RED_ZONE | SLAB_POISON | SLAB_STORE_USER | \
163
164
		SLAB_TRACE | SLAB_DESTROY_BY_RCU | SLAB_NOLEAKTRACE | \
		SLAB_FAILSLAB)
Christoph Lameter's avatar
Christoph Lameter committed
165
166

#define SLUB_MERGE_SAME (SLAB_DEBUG_FREE | SLAB_RECLAIM_ACCOUNT | \
Vegard Nossum's avatar
Vegard Nossum committed
167
		SLAB_CACHE_DMA | SLAB_NOTRACK)
Christoph Lameter's avatar
Christoph Lameter committed
168

169
170
#define OO_SHIFT	16
#define OO_MASK		((1 << OO_SHIFT) - 1)
171
#define MAX_OBJS_PER_PAGE	32767 /* since page.objects is u15 */
172

Christoph Lameter's avatar
Christoph Lameter committed
173
/* Internal SLUB flags */
Christoph Lameter's avatar
Christoph Lameter committed
174
#define __OBJECT_POISON		0x80000000UL /* Poison object */
175
#define __CMPXCHG_DOUBLE	0x40000000UL /* Use cmpxchg_double */
Christoph Lameter's avatar
Christoph Lameter committed
176
177
178
179
180
181
182
183
184

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 */
185
	PARTIAL,	/* Kmem_cache_node works */
Christoph Lameter's avatar
Christoph Lameter committed
186
	UP,		/* Everything works but does not show up in sysfs */
Christoph Lameter's avatar
Christoph Lameter committed
187
188
189
190
191
	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
192
static LIST_HEAD(slab_caches);
Christoph Lameter's avatar
Christoph Lameter committed
193

194
195
196
197
/*
 * Tracking user of a slab.
 */
struct track {
198
	unsigned long addr;	/* Called from address */
199
200
201
202
203
204
205
	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 };

206
#ifdef CONFIG_SYSFS
Christoph Lameter's avatar
Christoph Lameter committed
207
208
209
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 *);
210

Christoph Lameter's avatar
Christoph Lameter committed
211
#else
212
213
214
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; }
215
216
static inline void sysfs_slab_remove(struct kmem_cache *s)
{
217
	kfree(s->name);
218
219
	kfree(s);
}
220

Christoph Lameter's avatar
Christoph Lameter committed
221
222
#endif

223
static inline void stat(const struct kmem_cache *s, enum stat_item si)
224
225
{
#ifdef CONFIG_SLUB_STATS
226
	__this_cpu_inc(s->cpu_slab->stat[si]);
227
228
229
#endif
}

Christoph Lameter's avatar
Christoph Lameter committed
230
231
232
233
234
235
236
237
238
239
240
241
242
243
/********************************************************************
 * 			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)
{
	return s->node[node];
}

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

250
	if (!object)
251
252
		return 1;

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

	return 1;
}

262
263
264
265
266
static inline void *get_freepointer(struct kmem_cache *s, void *object)
{
	return *(void **)(object + s->offset);
}

267
268
269
270
271
272
273
274
275
276
277
278
static inline void *get_freepointer_safe(struct kmem_cache *s, void *object)
{
	void *p;

#ifdef CONFIG_DEBUG_PAGEALLOC
	probe_kernel_read(&p, (void **)(object + s->offset), sizeof(p));
#else
	p = get_freepointer(s, object);
#endif
	return p;
}

279
280
281
282
283
284
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 */
285
286
#define for_each_object(__p, __s, __addr, __objects) \
	for (__p = (__addr); __p < (__addr) + (__objects) * (__s)->size;\
287
288
289
290
291
292
293
294
			__p += (__s)->size)

/* 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
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
static inline size_t slab_ksize(const struct kmem_cache *s)
{
#ifdef CONFIG_SLUB_DEBUG
	/*
	 * Debugging requires use of the padding between object
	 * and whatever may come after it.
	 */
	if (s->flags & (SLAB_RED_ZONE | SLAB_POISON))
		return s->objsize;

#endif
	/*
	 * If we have the need to store the freelist pointer
	 * back there or track user information then we can
	 * only use the space before that information.
	 */
	if (s->flags & (SLAB_DESTROY_BY_RCU | SLAB_STORE_USER))
		return s->inuse;
	/*
	 * Else we can use all the padding etc for the allocation
	 */
	return s->size;
}

319
320
321
322
323
static inline int order_objects(int order, unsigned long size, int reserved)
{
	return ((PAGE_SIZE << order) - reserved) / size;
}

324
static inline struct kmem_cache_order_objects oo_make(int order,
325
		unsigned long size, int reserved)
326
327
{
	struct kmem_cache_order_objects x = {
328
		(order << OO_SHIFT) + order_objects(order, size, reserved)
329
330
331
332
333
334
335
	};

	return x;
}

static inline int oo_order(struct kmem_cache_order_objects x)
{
336
	return x.x >> OO_SHIFT;
337
338
339
340
}

static inline int oo_objects(struct kmem_cache_order_objects x)
{
341
	return x.x & OO_MASK;
342
343
}

344
345
346
347
348
349
350
351
352
353
354
355
356
/*
 * 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)
{
	__bit_spin_unlock(PG_locked, &page->flags);
}

357
358
359
360
361
362
363
364
365
366
367
368
369
370
static inline bool cmpxchg_double_slab(struct kmem_cache *s, struct page *page,
		void *freelist_old, unsigned long counters_old,
		void *freelist_new, unsigned long counters_new,
		const char *n)
{
#ifdef CONFIG_CMPXCHG_DOUBLE
	if (s->flags & __CMPXCHG_DOUBLE) {
		if (cmpxchg_double(&page->freelist,
			freelist_old, counters_old,
			freelist_new, counters_new))
		return 1;
	} else
#endif
	{
371
		slab_lock(page);
372
373
374
		if (page->freelist == freelist_old && page->counters == counters_old) {
			page->freelist = freelist_new;
			page->counters = counters_new;
375
			slab_unlock(page);
376
377
			return 1;
		}
378
		slab_unlock(page);
379
380
381
382
383
384
385
386
387
388
389
390
	}

	cpu_relax();
	stat(s, CMPXCHG_DOUBLE_FAIL);

#ifdef SLUB_DEBUG_CMPXCHG
	printk(KERN_INFO "%s %s: cmpxchg double redo ", n, s->name);
#endif

	return 0;
}

391
#ifdef CONFIG_SLUB_DEBUG
392
393
394
/*
 * Determine a map of object in use on a page.
 *
395
 * Node listlock must be held to guarantee that the page does
396
397
398
399
400
401
402
403
404
405
406
 * not vanish from under us.
 */
static void get_map(struct kmem_cache *s, struct page *page, unsigned long *map)
{
	void *p;
	void *addr = page_address(page);

	for (p = page->freelist; p; p = get_freepointer(s, p))
		set_bit(slab_index(p, s, addr), map);
}

407
408
409
/*
 * Debug settings:
 */
410
411
412
#ifdef CONFIG_SLUB_DEBUG_ON
static int slub_debug = DEBUG_DEFAULT_FLAGS;
#else
413
static int slub_debug;
414
#endif
415
416

static char *slub_debug_slabs;
417
static int disable_higher_order_debug;
418

Christoph Lameter's avatar
Christoph Lameter committed
419
420
421
422
423
424
425
426
427
428
429
430
431
/*
 * 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) {
432
			printk(KERN_ERR "%8s 0x%p: ", text, addr + i);
Christoph Lameter's avatar
Christoph Lameter committed
433
434
			newline = 0;
		}
435
		printk(KERN_CONT " %02x", addr[i]);
Christoph Lameter's avatar
Christoph Lameter committed
436
437
438
		offset = i % 16;
		ascii[offset] = isgraph(addr[i]) ? addr[i] : '.';
		if (offset == 15) {
439
			printk(KERN_CONT " %s\n", ascii);
Christoph Lameter's avatar
Christoph Lameter committed
440
441
442
443
444
445
			newline = 1;
		}
	}
	if (!newline) {
		i %= 16;
		while (i < 16) {
446
			printk(KERN_CONT "   ");
Christoph Lameter's avatar
Christoph Lameter committed
447
448
449
			ascii[i] = ' ';
			i++;
		}
450
		printk(KERN_CONT " %s\n", ascii);
Christoph Lameter's avatar
Christoph Lameter committed
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
	}
}

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,
468
			enum track_item alloc, unsigned long addr)
Christoph Lameter's avatar
Christoph Lameter committed
469
{
Akinobu Mita's avatar
Akinobu Mita committed
470
	struct track *p = get_track(s, object, alloc);
Christoph Lameter's avatar
Christoph Lameter committed
471
472
473
474

	if (addr) {
		p->addr = addr;
		p->cpu = smp_processor_id();
475
		p->pid = current->pid;
Christoph Lameter's avatar
Christoph Lameter committed
476
477
478
479
480
481
482
		p->when = jiffies;
	} else
		memset(p, 0, sizeof(struct track));
}

static void init_tracking(struct kmem_cache *s, void *object)
{
483
484
485
	if (!(s->flags & SLAB_STORE_USER))
		return;

486
487
	set_track(s, object, TRACK_FREE, 0UL);
	set_track(s, object, TRACK_ALLOC, 0UL);
Christoph Lameter's avatar
Christoph Lameter committed
488
489
490
491
492
493
494
}

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

495
	printk(KERN_ERR "INFO: %s in %pS age=%lu cpu=%u pid=%d\n",
496
		s, (void *)t->addr, jiffies - t->when, t->cpu, t->pid);
497
498
499
500
501
502
503
504
505
506
507
508
509
}

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)
{
510
511
	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);
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527

}

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
528
529
}

530
531
532
533
534
535
536
537
538
539
540
541
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
542
543
{
	unsigned int off;	/* Offset of last byte */
544
	u8 *addr = page_address(page);
545
546
547
548
549
550
551
552
553
554
555

	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);

556
	print_section("Object", p, min_t(unsigned long, s->objsize, PAGE_SIZE));
Christoph Lameter's avatar
Christoph Lameter committed
557
558
559
560
561
562
563
564
565
566

	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;

567
	if (s->flags & SLAB_STORE_USER)
Christoph Lameter's avatar
Christoph Lameter committed
568
569
570
571
		off += 2 * sizeof(struct track);

	if (off != s->size)
		/* Beginning of the filler is the free pointer */
572
573
574
		print_section("Padding", p + off, s->size - off);

	dump_stack();
Christoph Lameter's avatar
Christoph Lameter committed
575
576
577
578
579
}

static void object_err(struct kmem_cache *s, struct page *page,
			u8 *object, char *reason)
{
580
	slab_bug(s, "%s", reason);
581
	print_trailer(s, page, object);
Christoph Lameter's avatar
Christoph Lameter committed
582
583
}

584
static void slab_err(struct kmem_cache *s, struct page *page, char *fmt, ...)
Christoph Lameter's avatar
Christoph Lameter committed
585
586
587
588
{
	va_list args;
	char buf[100];

589
590
	va_start(args, fmt);
	vsnprintf(buf, sizeof(buf), fmt, args);
Christoph Lameter's avatar
Christoph Lameter committed
591
	va_end(args);
592
	slab_bug(s, "%s", buf);
593
	print_page_info(page);
Christoph Lameter's avatar
Christoph Lameter committed
594
595
596
	dump_stack();
}

597
static void init_object(struct kmem_cache *s, void *object, u8 val)
Christoph Lameter's avatar
Christoph Lameter committed
598
599
600
601
602
{
	u8 *p = object;

	if (s->flags & __OBJECT_POISON) {
		memset(p, POISON_FREE, s->objsize - 1);
603
		p[s->objsize - 1] = POISON_END;
Christoph Lameter's avatar
Christoph Lameter committed
604
605
606
	}

	if (s->flags & SLAB_RED_ZONE)
607
		memset(p + s->objsize, val, s->inuse - s->objsize);
Christoph Lameter's avatar
Christoph Lameter committed
608
609
}

610
static u8 *check_bytes(u8 *start, unsigned int value, unsigned int bytes)
Christoph Lameter's avatar
Christoph Lameter committed
611
612
613
{
	while (bytes) {
		if (*start != (u8)value)
614
			return start;
Christoph Lameter's avatar
Christoph Lameter committed
615
616
617
		start++;
		bytes--;
	}
618
619
620
621
622
623
624
625
626
627
628
629
	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,
630
			u8 *start, unsigned int value, unsigned int bytes)
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
{
	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
650
651
652
653
654
655
656
657
658
}

/*
 * 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
659
 *
Christoph Lameter's avatar
Christoph Lameter committed
660
661
662
663
664
 * 	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
665
666
667
 * 	Padding is extended by another word if Redzoning is enabled and
 * 	objsize == inuse.
 *
Christoph Lameter's avatar
Christoph Lameter committed
668
669
670
671
 * 	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
672
673
 * 	Meta data starts here.
 *
Christoph Lameter's avatar
Christoph Lameter committed
674
675
 * 	A. Free pointer (if we cannot overwrite object on free)
 * 	B. Tracking data for SLAB_STORE_USER
Christoph Lameter's avatar
Christoph Lameter committed
676
 * 	C. Padding to reach required alignment boundary or at mininum
Christoph Lameter's avatar
Christoph Lameter committed
677
 * 		one word if debugging is on to be able to detect writes
Christoph Lameter's avatar
Christoph Lameter committed
678
679
680
 * 		before the word boundary.
 *
 *	Padding is done using 0x5a (POISON_INUSE)
Christoph Lameter's avatar
Christoph Lameter committed
681
682
 *
 * object + s->size
Christoph Lameter's avatar
Christoph Lameter committed
683
 * 	Nothing is used beyond s->size.
Christoph Lameter's avatar
Christoph Lameter committed
684
 *
Christoph Lameter's avatar
Christoph Lameter committed
685
686
 * 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
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
 * 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;

705
706
	return check_bytes_and_report(s, page, p, "Object padding",
				p + off, POISON_INUSE, s->size - off);
Christoph Lameter's avatar
Christoph Lameter committed
707
708
}

709
/* Check the pad bytes at the end of a slab page */
Christoph Lameter's avatar
Christoph Lameter committed
710
711
static int slab_pad_check(struct kmem_cache *s, struct page *page)
{
712
713
714
715
716
	u8 *start;
	u8 *fault;
	u8 *end;
	int length;
	int remainder;
Christoph Lameter's avatar
Christoph Lameter committed
717
718
719
720

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

721
	start = page_address(page);
722
	length = (PAGE_SIZE << compound_order(page)) - s->reserved;
723
724
	end = start + length;
	remainder = length % s->size;
Christoph Lameter's avatar
Christoph Lameter committed
725
726
727
	if (!remainder)
		return 1;

728
	fault = check_bytes(end - remainder, POISON_INUSE, remainder);
729
730
731
732
733
734
	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);
735
	print_section("Padding", end - remainder, remainder);
736

Eric Dumazet's avatar
Eric Dumazet committed
737
	restore_bytes(s, "slab padding", POISON_INUSE, end - remainder, end);
738
	return 0;
Christoph Lameter's avatar
Christoph Lameter committed
739
740
741
}

static int check_object(struct kmem_cache *s, struct page *page,
742
					void *object, u8 val)
Christoph Lameter's avatar
Christoph Lameter committed
743
744
745
746
747
{
	u8 *p = object;
	u8 *endobject = object + s->objsize;

	if (s->flags & SLAB_RED_ZONE) {
748
		if (!check_bytes_and_report(s, page, object, "Redzone",
749
			endobject, val, s->inuse - s->objsize))
Christoph Lameter's avatar
Christoph Lameter committed
750
751
			return 0;
	} else {
Ingo Molnar's avatar
Ingo Molnar committed
752
753
754
755
		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
756
757
758
	}

	if (s->flags & SLAB_POISON) {
759
		if (val != SLUB_RED_ACTIVE && (s->flags & __OBJECT_POISON) &&
760
761
762
			(!check_bytes_and_report(s, page, p, "Poison", p,
					POISON_FREE, s->objsize - 1) ||
			 !check_bytes_and_report(s, page, p, "Poison",
763
				p + s->objsize - 1, POISON_END, 1)))
Christoph Lameter's avatar
Christoph Lameter committed
764
765
766
767
768
769
770
			return 0;
		/*
		 * check_pad_bytes cleans up on its own.
		 */
		check_pad_bytes(s, page, p);
	}

771
	if (!s->offset && val == SLUB_RED_ACTIVE)
Christoph Lameter's avatar
Christoph Lameter committed
772
773
774
775
776
777
778
779
780
781
		/*
		 * 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
782
		 * No choice but to zap it and thus lose the remainder
Christoph Lameter's avatar
Christoph Lameter committed
783
		 * of the free objects in this slab. May cause
Christoph Lameter's avatar
Christoph Lameter committed
784
		 * another error because the object count is now wrong.
Christoph Lameter's avatar
Christoph Lameter committed
785
		 */
786
		set_freepointer(s, p, NULL);
Christoph Lameter's avatar
Christoph Lameter committed
787
788
789
790
791
792
793
		return 0;
	}
	return 1;
}

static int check_slab(struct kmem_cache *s, struct page *page)
{
794
795
	int maxobj;

Christoph Lameter's avatar
Christoph Lameter committed
796
797
798
	VM_BUG_ON(!irqs_disabled());

	if (!PageSlab(page)) {
799
		slab_err(s, page, "Not a valid slab page");
Christoph Lameter's avatar
Christoph Lameter committed
800
801
		return 0;
	}
802

803
	maxobj = order_objects(compound_order(page), s->size, s->reserved);
804
805
806
807
808
809
	if (page->objects > maxobj) {
		slab_err(s, page, "objects %u > max %u",
			s->name, page->objects, maxobj);
		return 0;
	}
	if (page->inuse > page->objects) {
810
		slab_err(s, page, "inuse %u > max %u",
811
			s->name, page->inuse, page->objects);
Christoph Lameter's avatar
Christoph Lameter committed
812
813
814
815
816
817
818
819
		return 0;
	}
	/* Slab_pad_check fixes things up after itself */
	slab_pad_check(s, page);
	return 1;
}

/*
Christoph Lameter's avatar
Christoph Lameter committed
820
821
 * 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
822
823
824
825
 */
static int on_freelist(struct kmem_cache *s, struct page *page, void *search)
{
	int nr = 0;
826
	void *fp;
Christoph Lameter's avatar
Christoph Lameter committed
827
	void *object = NULL;
828
	unsigned long max_objects;
Christoph Lameter's avatar
Christoph Lameter committed
829

830
	fp = page->freelist;
831
	while (fp && nr <= page->objects) {
Christoph Lameter's avatar
Christoph Lameter committed
832
833
834
835
836
837
		if (fp == search)
			return 1;
		if (!check_valid_pointer(s, page, fp)) {
			if (object) {
				object_err(s, page, object,
					"Freechain corrupt");
838
				set_freepointer(s, object, NULL);
Christoph Lameter's avatar
Christoph Lameter committed
839
840
				break;
			} else {
841
				slab_err(s, page, "Freepointer corrupt");
842
				page->freelist = NULL;
843
				page->inuse = page->objects;
844
				slab_fix(s, "Freelist cleared");
Christoph Lameter's avatar
Christoph Lameter committed
845
846
847
848
849
850
851
852
853
				return 0;
			}
			break;
		}
		object = fp;
		fp = get_freepointer(s, object);
		nr++;
	}

854
	max_objects = order_objects(compound_order(page), s->size, s->reserved);
855
856
	if (max_objects > MAX_OBJS_PER_PAGE)
		max_objects = MAX_OBJS_PER_PAGE;
857
858
859
860
861
862
863

	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.");
	}
864
	if (page->inuse != page->objects - nr) {
865
		slab_err(s, page, "Wrong object count. Counter is %d but "
866
867
			"counted were %d", page->inuse, page->objects - nr);
		page->inuse = page->objects - nr;
868
		slab_fix(s, "Object count adjusted.");
Christoph Lameter's avatar
Christoph Lameter committed
869
870
871
872
	}
	return search == NULL;
}

873
874
static void trace(struct kmem_cache *s, struct page *page, void *object,
								int alloc)
Christoph Lameter's avatar
Christoph Lameter committed
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
{
	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();
	}
}

890
891
892
893
894
895
/*
 * Hooks for other subsystems that check memory allocations. In a typical
 * production configuration these hooks all should produce no code at all.
 */
static inline int slab_pre_alloc_hook(struct kmem_cache *s, gfp_t flags)
{
896
	flags &= gfp_allowed_mask;
897
898
899
900
901
902
903
904
	lockdep_trace_alloc(flags);
	might_sleep_if(flags & __GFP_WAIT);

	return should_failslab(s->objsize, flags, s->flags);
}

static inline void slab_post_alloc_hook(struct kmem_cache *s, gfp_t flags, void *object)
{
905
	flags &= gfp_allowed_mask;
906
	kmemcheck_slab_alloc(s, flags, object, slab_ksize(s));
907
908
909
910
911
912
913
	kmemleak_alloc_recursive(object, s->objsize, 1, s->flags, flags);
}

static inline void slab_free_hook(struct kmem_cache *s, void *x)
{
	kmemleak_free_recursive(x, s->flags);

914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
	/*
	 * Trouble is that we may no longer disable interupts in the fast path
	 * So in order to make the debug calls that expect irqs to be
	 * disabled we need to disable interrupts temporarily.
	 */
#if defined(CONFIG_KMEMCHECK) || defined(CONFIG_LOCKDEP)
	{
		unsigned long flags;

		local_irq_save(flags);
		kmemcheck_slab_free(s, x, s->objsize);
		debug_check_no_locks_freed(x, s->objsize);
		local_irq_restore(flags);
	}
#endif
929
930
	if (!(s->flags & SLAB_DEBUG_OBJECTS))
		debug_check_no_obj_freed(x, s->objsize);
931
932
}

933
/*
Christoph Lameter's avatar
Christoph Lameter committed
934
 * Tracking of fully allocated slabs for debugging purposes.
935
936
 *
 * list_lock must be held.
937
 */
938
939
static void add_full(struct kmem_cache *s,
	struct kmem_cache_node *n, struct page *page)
940
{
941
942
943
	if (!(s->flags & SLAB_STORE_USER))
		return;

944
945
946
	list_add(&page->lru, &n->full);
}

947
948
949
/*
 * list_lock must be held.
 */
950
951
952
953
954
955
956
957
static void remove_full(struct kmem_cache *s, struct page *page)
{
	if (!(s->flags & SLAB_STORE_USER))
		return;

	list_del(&page->lru);
}

958
959
960
961
962
963
964
965
/* 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);
}

966
967
968
969
970
static inline unsigned long node_nr_slabs(struct kmem_cache_node *n)
{
	return atomic_long_read(&n->nr_slabs);
}

971
static inline void inc_slabs_node(struct kmem_cache *s, int node, int objects)
972
973
974
975
976
977
978
979
980
{
	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).
	 */
981
	if (n) {
982
		atomic_long_inc(&n->nr_slabs);
983
984
		atomic_long_add(objects, &n->total_objects);
	}
985
}
986
static inline void dec_slabs_node(struct kmem_cache *s, int node, int objects)
987
988
989
990
{
	struct kmem_cache_node *n = get_node(s, node);

	atomic_long_dec(&n->nr_slabs);
991
	atomic_long_sub(objects, &n->total_objects);
992
993
994
}

/* Object debug checks for alloc/free paths */
Christoph Lameter's avatar
Christoph Lameter committed
995
996
997
998
999
1000
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;