slub.c 129 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 "slab.h"
20
#include <linux/proc_fs.h>
Christoph Lameter's avatar
Christoph Lameter committed
21
#include <linux/seq_file.h>
Vegard Nossum's avatar
Vegard Nossum committed
22
#include <linux/kmemcheck.h>
Christoph Lameter's avatar
Christoph Lameter committed
23
24
25
26
#include <linux/cpu.h>
#include <linux/cpuset.h>
#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>
32
#include <linux/stacktrace.h>
33
#include <linux/prefetch.h>
Christoph Lameter's avatar
Christoph Lameter committed
34

35
36
#include <trace/events/kmem.h>

Christoph Lameter's avatar
Christoph Lameter committed
37
38
/*
 * Lock order:
39
 *   1. slab_mutex (Global Mutex)
40
41
 *   2. node->list_lock
 *   3. slab_lock(page) (Only on some arches and for debugging)
Christoph Lameter's avatar
Christoph Lameter committed
42
 *
43
 *   slab_mutex
44
 *
45
 *   The role of the slab_mutex is to protect the list of all the slabs
46
47
48
49
50
51
52
53
54
55
56
57
58
59
 *   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
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
 *
 *   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
80
81
 * 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
82
 * freed then the slab will show up again on the partial lists.
Christoph Lameter's avatar
Christoph Lameter committed
83
84
 * We track full slabs for debugging purposes though because otherwise we
 * cannot scan all objects.
Christoph Lameter's avatar
Christoph Lameter committed
85
86
87
88
89
90
91
 *
 * 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.
 *
92
93
94
95
96
97
98
99
100
101
102
103
 * 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
104
 * 			freelist that allows lockless access to
105
106
 * 			free objects in addition to the regular freelist
 * 			that requires the slab lock.
Christoph Lameter's avatar
Christoph Lameter committed
107
108
109
 *
 * PageError		Slab requires special handling due to debug
 * 			options set. This moves	slab handling out of
110
 * 			the fast path and disables lockless freelists.
Christoph Lameter's avatar
Christoph Lameter committed
111
112
 */

113
114
115
116
117
#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)
{
118
#ifdef CONFIG_SLUB_DEBUG
119
	return unlikely(s->flags & SLAB_DEBUG_FLAGS);
120
#else
121
	return 0;
122
#endif
123
}
124

Christoph Lameter's avatar
Christoph Lameter committed
125
126
127
128
129
130
131
132
133
134
135
/*
 * 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

136
137
138
/* Enable to log cmpxchg failures */
#undef SLUB_DEBUG_CMPXCHG

139
140
141
142
/*
 * Mininum number of partial slabs. These will be left on the partial
 * lists even if they are empty. kmem_cache_shrink may reclaim them.
 */
143
#define MIN_PARTIAL 5
Christoph Lameter's avatar
Christoph Lameter committed
144

145
146
147
148
149
150
151
/*
 * 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
152
153
#define DEBUG_DEFAULT_FLAGS (SLAB_DEBUG_FREE | SLAB_RED_ZONE | \
				SLAB_POISON | SLAB_STORE_USER)
Christoph Lameter's avatar
Christoph Lameter committed
154

155
/*
156
157
158
 * 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.
159
 */
160
#define DEBUG_METADATA_FLAGS (SLAB_RED_ZONE | SLAB_POISON | SLAB_STORE_USER)
161

Christoph Lameter's avatar
Christoph Lameter committed
162
163
164
165
/*
 * Set of flags that will prevent slab merging
 */
#define SLUB_NEVER_MERGE (SLAB_RED_ZONE | SLAB_POISON | SLAB_STORE_USER | \
166
167
		SLAB_TRACE | SLAB_DESTROY_BY_RCU | SLAB_NOLEAKTRACE | \
		SLAB_FAILSLAB)
Christoph Lameter's avatar
Christoph Lameter committed
168
169

#define SLUB_MERGE_SAME (SLAB_DEBUG_FREE | SLAB_RECLAIM_ACCOUNT | \
Vegard Nossum's avatar
Vegard Nossum committed
170
		SLAB_CACHE_DMA | SLAB_NOTRACK)
Christoph Lameter's avatar
Christoph Lameter committed
171

172
173
#define OO_SHIFT	16
#define OO_MASK		((1 << OO_SHIFT) - 1)
174
#define MAX_OBJS_PER_PAGE	32767 /* since page.objects is u15 */
175

Christoph Lameter's avatar
Christoph Lameter committed
176
/* Internal SLUB flags */
Christoph Lameter's avatar
Christoph Lameter committed
177
#define __OBJECT_POISON		0x80000000UL /* Poison object */
178
#define __CMPXCHG_DOUBLE	0x40000000UL /* Use cmpxchg_double */
Christoph Lameter's avatar
Christoph Lameter committed
179
180
181
182
183
184
185

static int kmem_size = sizeof(struct kmem_cache);

#ifdef CONFIG_SMP
static struct notifier_block slab_notifier;
#endif

186
187
188
/*
 * Tracking user of a slab.
 */
189
#define TRACK_ADDRS_COUNT 16
190
struct track {
191
	unsigned long addr;	/* Called from address */
192
193
194
#ifdef CONFIG_STACKTRACE
	unsigned long addrs[TRACK_ADDRS_COUNT];	/* Called from address */
#endif
195
196
197
198
199
200
201
	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 };

202
#ifdef CONFIG_SYSFS
Christoph Lameter's avatar
Christoph Lameter committed
203
204
205
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 *);
206

Christoph Lameter's avatar
Christoph Lameter committed
207
#else
208
209
210
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; }
211
212
static inline void sysfs_slab_remove(struct kmem_cache *s)
{
213
	kfree(s->name);
214
215
	kfree(s);
}
216

Christoph Lameter's avatar
Christoph Lameter committed
217
218
#endif

219
static inline void stat(const struct kmem_cache *s, enum stat_item si)
220
221
{
#ifdef CONFIG_SLUB_STATS
222
	__this_cpu_inc(s->cpu_slab->stat[si]);
223
224
225
#endif
}

Christoph Lameter's avatar
Christoph Lameter committed
226
227
228
229
230
231
232
233
234
/********************************************************************
 * 			Core slab cache functions
 *******************************************************************/

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
235
/* Verify that a pointer has an address that is valid within a slab page */
236
237
238
239
240
static inline int check_valid_pointer(struct kmem_cache *s,
				struct page *page, const void *object)
{
	void *base;

241
	if (!object)
242
243
		return 1;

244
	base = page_address(page);
245
	if (object < base || object >= base + page->objects * s->size ||
246
247
248
249
250
251
252
		(object - base) % s->size) {
		return 0;
	}

	return 1;
}

253
254
255
256
257
static inline void *get_freepointer(struct kmem_cache *s, void *object)
{
	return *(void **)(object + s->offset);
}

258
259
260
261
262
static void prefetch_freepointer(const struct kmem_cache *s, void *object)
{
	prefetch(object + s->offset);
}

263
264
265
266
267
268
269
270
271
272
273
274
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;
}

275
276
277
278
279
280
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
287
288
289
290
			__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;
}

291
292
293
294
295
296
297
298
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))
299
		return s->object_size;
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314

#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;
}

315
316
317
318
319
static inline int order_objects(int order, unsigned long size, int reserved)
{
	return ((PAGE_SIZE << order) - reserved) / size;
}

320
static inline struct kmem_cache_order_objects oo_make(int order,
321
		unsigned long size, int reserved)
322
323
{
	struct kmem_cache_order_objects x = {
324
		(order << OO_SHIFT) + order_objects(order, size, reserved)
325
326
327
328
329
330
331
	};

	return x;
}

static inline int oo_order(struct kmem_cache_order_objects x)
{
332
	return x.x >> OO_SHIFT;
333
334
335
336
}

static inline int oo_objects(struct kmem_cache_order_objects x)
{
337
	return x.x & OO_MASK;
338
339
}

340
341
342
343
344
345
346
347
348
349
350
351
352
/*
 * 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);
}

353
354
355
356
357
358
359
/* Interrupts must be disabled (for the fallback code to work right) */
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)
{
	VM_BUG_ON(!irqs_disabled());
360
361
#if defined(CONFIG_HAVE_CMPXCHG_DOUBLE) && \
    defined(CONFIG_HAVE_ALIGNED_STRUCT_PAGE)
362
	if (s->flags & __CMPXCHG_DOUBLE) {
363
		if (cmpxchg_double(&page->freelist, &page->counters,
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
			freelist_old, counters_old,
			freelist_new, counters_new))
		return 1;
	} else
#endif
	{
		slab_lock(page);
		if (page->freelist == freelist_old && page->counters == counters_old) {
			page->freelist = freelist_new;
			page->counters = counters_new;
			slab_unlock(page);
			return 1;
		}
		slab_unlock(page);
	}

	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;
}

390
391
392
393
394
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)
{
395
396
#if defined(CONFIG_HAVE_CMPXCHG_DOUBLE) && \
    defined(CONFIG_HAVE_ALIGNED_STRUCT_PAGE)
397
	if (s->flags & __CMPXCHG_DOUBLE) {
398
		if (cmpxchg_double(&page->freelist, &page->counters,
399
400
401
402
403
404
			freelist_old, counters_old,
			freelist_new, counters_new))
		return 1;
	} else
#endif
	{
405
406
407
		unsigned long flags;

		local_irq_save(flags);
408
		slab_lock(page);
409
410
411
		if (page->freelist == freelist_old && page->counters == counters_old) {
			page->freelist = freelist_new;
			page->counters = counters_new;
412
			slab_unlock(page);
413
			local_irq_restore(flags);
414
415
			return 1;
		}
416
		slab_unlock(page);
417
		local_irq_restore(flags);
418
419
420
421
422
423
424
425
426
427
428
429
	}

	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;
}

430
#ifdef CONFIG_SLUB_DEBUG
431
432
433
/*
 * Determine a map of object in use on a page.
 *
434
 * Node listlock must be held to guarantee that the page does
435
436
437
438
439
440
441
442
443
444
445
 * 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);
}

446
447
448
/*
 * Debug settings:
 */
449
450
451
#ifdef CONFIG_SLUB_DEBUG_ON
static int slub_debug = DEBUG_DEFAULT_FLAGS;
#else
452
static int slub_debug;
453
#endif
454
455

static char *slub_debug_slabs;
456
static int disable_higher_order_debug;
457

Christoph Lameter's avatar
Christoph Lameter committed
458
459
460
461
462
/*
 * Object debugging
 */
static void print_section(char *text, u8 *addr, unsigned int length)
{
463
464
	print_hex_dump(KERN_ERR, text, DUMP_PREFIX_ADDRESS, 16, 1, addr,
			length, 1);
Christoph Lameter's avatar
Christoph Lameter committed
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
}

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,
481
			enum track_item alloc, unsigned long addr)
Christoph Lameter's avatar
Christoph Lameter committed
482
{
Akinobu Mita's avatar
Akinobu Mita committed
483
	struct track *p = get_track(s, object, alloc);
Christoph Lameter's avatar
Christoph Lameter committed
484
485

	if (addr) {
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
#ifdef CONFIG_STACKTRACE
		struct stack_trace trace;
		int i;

		trace.nr_entries = 0;
		trace.max_entries = TRACK_ADDRS_COUNT;
		trace.entries = p->addrs;
		trace.skip = 3;
		save_stack_trace(&trace);

		/* See rant in lockdep.c */
		if (trace.nr_entries != 0 &&
		    trace.entries[trace.nr_entries - 1] == ULONG_MAX)
			trace.nr_entries--;

		for (i = trace.nr_entries; i < TRACK_ADDRS_COUNT; i++)
			p->addrs[i] = 0;
#endif
Christoph Lameter's avatar
Christoph Lameter committed
504
505
		p->addr = addr;
		p->cpu = smp_processor_id();
506
		p->pid = current->pid;
Christoph Lameter's avatar
Christoph Lameter committed
507
508
509
510
511
512
513
		p->when = jiffies;
	} else
		memset(p, 0, sizeof(struct track));
}

static void init_tracking(struct kmem_cache *s, void *object)
{
514
515
516
	if (!(s->flags & SLAB_STORE_USER))
		return;

517
518
	set_track(s, object, TRACK_FREE, 0UL);
	set_track(s, object, TRACK_ALLOC, 0UL);
Christoph Lameter's avatar
Christoph Lameter committed
519
520
521
522
523
524
525
}

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

526
	printk(KERN_ERR "INFO: %s in %pS age=%lu cpu=%u pid=%d\n",
527
		s, (void *)t->addr, jiffies - t->when, t->cpu, t->pid);
528
529
530
531
532
533
534
535
536
537
#ifdef CONFIG_STACKTRACE
	{
		int i;
		for (i = 0; i < TRACK_ADDRS_COUNT; i++)
			if (t->addrs[i])
				printk(KERN_ERR "\t%pS\n", (void *)t->addrs[i]);
			else
				break;
	}
#endif
538
539
540
541
542
543
544
545
546
547
548
549
550
}

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)
{
551
552
	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);
553
554
555
556
557
558
559
560
561
562
563
564
565

}

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");
566
	printk(KERN_ERR "BUG %s (%s): %s\n", s->name, print_tainted(), buf);
567
568
	printk(KERN_ERR "----------------------------------------"
			"-------------------------------------\n\n");
Christoph Lameter's avatar
Christoph Lameter committed
569
570
}

571
572
573
574
575
576
577
578
579
580
581
582
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
583
584
{
	unsigned int off;	/* Offset of last byte */
585
	u8 *addr = page_address(page);
586
587
588
589
590
591
592
593
594

	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)
595
		print_section("Bytes b4 ", p - 16, 16);
Christoph Lameter's avatar
Christoph Lameter committed
596

597
	print_section("Object ", p, min_t(unsigned long, s->object_size,
598
				PAGE_SIZE));
Christoph Lameter's avatar
Christoph Lameter committed
599
	if (s->flags & SLAB_RED_ZONE)
600
601
		print_section("Redzone ", p + s->object_size,
			s->inuse - s->object_size);
Christoph Lameter's avatar
Christoph Lameter committed
602
603
604
605
606
607

	if (s->offset)
		off = s->offset + sizeof(void *);
	else
		off = s->inuse;

608
	if (s->flags & SLAB_STORE_USER)
Christoph Lameter's avatar
Christoph Lameter committed
609
610
611
612
		off += 2 * sizeof(struct track);

	if (off != s->size)
		/* Beginning of the filler is the free pointer */
613
		print_section("Padding ", p + off, s->size - off);
614
615

	dump_stack();
Christoph Lameter's avatar
Christoph Lameter committed
616
617
618
619
620
}

static void object_err(struct kmem_cache *s, struct page *page,
			u8 *object, char *reason)
{
621
	slab_bug(s, "%s", reason);
622
	print_trailer(s, page, object);
Christoph Lameter's avatar
Christoph Lameter committed
623
624
}

625
static void slab_err(struct kmem_cache *s, struct page *page, char *fmt, ...)
Christoph Lameter's avatar
Christoph Lameter committed
626
627
628
629
{
	va_list args;
	char buf[100];

630
631
	va_start(args, fmt);
	vsnprintf(buf, sizeof(buf), fmt, args);
Christoph Lameter's avatar
Christoph Lameter committed
632
	va_end(args);
633
	slab_bug(s, "%s", buf);
634
	print_page_info(page);
Christoph Lameter's avatar
Christoph Lameter committed
635
636
637
	dump_stack();
}

638
static void init_object(struct kmem_cache *s, void *object, u8 val)
Christoph Lameter's avatar
Christoph Lameter committed
639
640
641
642
{
	u8 *p = object;

	if (s->flags & __OBJECT_POISON) {
643
644
		memset(p, POISON_FREE, s->object_size - 1);
		p[s->object_size - 1] = POISON_END;
Christoph Lameter's avatar
Christoph Lameter committed
645
646
647
	}

	if (s->flags & SLAB_RED_ZONE)
648
		memset(p + s->object_size, val, s->inuse - s->object_size);
Christoph Lameter's avatar
Christoph Lameter committed
649
650
}

651
652
653
654
655
656
657
658
659
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,
660
			u8 *start, unsigned int value, unsigned int bytes)
661
662
663
664
{
	u8 *fault;
	u8 *end;

665
	fault = memchr_inv(start, value, bytes);
666
667
668
669
670
671
672
673
674
675
676
677
678
679
	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
680
681
682
683
684
685
686
687
688
}

/*
 * 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
689
 *
Christoph Lameter's avatar
Christoph Lameter committed
690
691
692
 * 	Poisoning uses 0x6b (POISON_FREE) and the last byte is
 * 	0xa5 (POISON_END)
 *
693
 * object + s->object_size
Christoph Lameter's avatar
Christoph Lameter committed
694
 * 	Padding to reach word boundary. This is also used for Redzoning.
Christoph Lameter's avatar
Christoph Lameter committed
695
 * 	Padding is extended by another word if Redzoning is enabled and
696
 * 	object_size == inuse.
Christoph Lameter's avatar
Christoph Lameter committed
697
 *
Christoph Lameter's avatar
Christoph Lameter committed
698
699
700
701
 * 	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
702
703
 * 	Meta data starts here.
 *
Christoph Lameter's avatar
Christoph Lameter committed
704
705
 * 	A. Free pointer (if we cannot overwrite object on free)
 * 	B. Tracking data for SLAB_STORE_USER
Christoph Lameter's avatar
Christoph Lameter committed
706
 * 	C. Padding to reach required alignment boundary or at mininum
Christoph Lameter's avatar
Christoph Lameter committed
707
 * 		one word if debugging is on to be able to detect writes
Christoph Lameter's avatar
Christoph Lameter committed
708
709
710
 * 		before the word boundary.
 *
 *	Padding is done using 0x5a (POISON_INUSE)
Christoph Lameter's avatar
Christoph Lameter committed
711
712
 *
 * object + s->size
Christoph Lameter's avatar
Christoph Lameter committed
713
 * 	Nothing is used beyond s->size.
Christoph Lameter's avatar
Christoph Lameter committed
714
 *
715
 * If slabcaches are merged then the object_size and inuse boundaries are mostly
Christoph Lameter's avatar
Christoph Lameter committed
716
 * ignored. And therefore no slab options that rely on these boundaries
Christoph Lameter's avatar
Christoph Lameter committed
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
 * 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;

735
736
	return check_bytes_and_report(s, page, p, "Object padding",
				p + off, POISON_INUSE, s->size - off);
Christoph Lameter's avatar
Christoph Lameter committed
737
738
}

739
/* Check the pad bytes at the end of a slab page */
Christoph Lameter's avatar
Christoph Lameter committed
740
741
static int slab_pad_check(struct kmem_cache *s, struct page *page)
{
742
743
744
745
746
	u8 *start;
	u8 *fault;
	u8 *end;
	int length;
	int remainder;
Christoph Lameter's avatar
Christoph Lameter committed
747
748
749
750

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

751
	start = page_address(page);
752
	length = (PAGE_SIZE << compound_order(page)) - s->reserved;
753
754
	end = start + length;
	remainder = length % s->size;
Christoph Lameter's avatar
Christoph Lameter committed
755
756
757
	if (!remainder)
		return 1;

758
	fault = memchr_inv(end - remainder, POISON_INUSE, remainder);
759
760
761
762
763
764
	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);
765
	print_section("Padding ", end - remainder, remainder);
766

Eric Dumazet's avatar
Eric Dumazet committed
767
	restore_bytes(s, "slab padding", POISON_INUSE, end - remainder, end);
768
	return 0;
Christoph Lameter's avatar
Christoph Lameter committed
769
770
771
}

static int check_object(struct kmem_cache *s, struct page *page,
772
					void *object, u8 val)
Christoph Lameter's avatar
Christoph Lameter committed
773
774
{
	u8 *p = object;
775
	u8 *endobject = object + s->object_size;
Christoph Lameter's avatar
Christoph Lameter committed
776
777

	if (s->flags & SLAB_RED_ZONE) {
778
		if (!check_bytes_and_report(s, page, object, "Redzone",
779
			endobject, val, s->inuse - s->object_size))
Christoph Lameter's avatar
Christoph Lameter committed
780
781
			return 0;
	} else {
782
		if ((s->flags & SLAB_POISON) && s->object_size < s->inuse) {
Ingo Molnar's avatar
Ingo Molnar committed
783
			check_bytes_and_report(s, page, p, "Alignment padding",
784
				endobject, POISON_INUSE, s->inuse - s->object_size);
Ingo Molnar's avatar
Ingo Molnar committed
785
		}
Christoph Lameter's avatar
Christoph Lameter committed
786
787
788
	}

	if (s->flags & SLAB_POISON) {
789
		if (val != SLUB_RED_ACTIVE && (s->flags & __OBJECT_POISON) &&
790
			(!check_bytes_and_report(s, page, p, "Poison", p,
791
					POISON_FREE, s->object_size - 1) ||
792
			 !check_bytes_and_report(s, page, p, "Poison",
793
				p + s->object_size - 1, POISON_END, 1)))
Christoph Lameter's avatar
Christoph Lameter committed
794
795
796
797
798
799
800
			return 0;
		/*
		 * check_pad_bytes cleans up on its own.
		 */
		check_pad_bytes(s, page, p);
	}

801
	if (!s->offset && val == SLUB_RED_ACTIVE)
Christoph Lameter's avatar
Christoph Lameter committed
802
803
804
805
806
807
808
809
810
811
		/*
		 * 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
812
		 * No choice but to zap it and thus lose the remainder
Christoph Lameter's avatar
Christoph Lameter committed
813
		 * of the free objects in this slab. May cause
Christoph Lameter's avatar
Christoph Lameter committed
814
		 * another error because the object count is now wrong.
Christoph Lameter's avatar
Christoph Lameter committed
815
		 */
816
		set_freepointer(s, p, NULL);
Christoph Lameter's avatar
Christoph Lameter committed
817
818
819
820
821
822
823
		return 0;
	}
	return 1;
}

static int check_slab(struct kmem_cache *s, struct page *page)
{
824
825
	int maxobj;

Christoph Lameter's avatar
Christoph Lameter committed
826
827
828
	VM_BUG_ON(!irqs_disabled());

	if (!PageSlab(page)) {
829
		slab_err(s, page, "Not a valid slab page");
Christoph Lameter's avatar
Christoph Lameter committed
830
831
		return 0;
	}
832

833
	maxobj = order_objects(compound_order(page), s->size, s->reserved);
834
835
836
837
838
839
	if (page->objects > maxobj) {
		slab_err(s, page, "objects %u > max %u",
			s->name, page->objects, maxobj);
		return 0;
	}
	if (page->inuse > page->objects) {
840
		slab_err(s, page, "inuse %u > max %u",
841
			s->name, page->inuse, page->objects);
Christoph Lameter's avatar
Christoph Lameter committed
842
843
844
845
846
847
848
849
		return 0;
	}
	/* Slab_pad_check fixes things up after itself */
	slab_pad_check(s, page);
	return 1;
}

/*
Christoph Lameter's avatar
Christoph Lameter committed
850
851
 * 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
852
853
854
855
 */
static int on_freelist(struct kmem_cache *s, struct page *page, void *search)
{
	int nr = 0;
856
	void *fp;
Christoph Lameter's avatar
Christoph Lameter committed
857
	void *object = NULL;
858
	unsigned long max_objects;
Christoph Lameter's avatar
Christoph Lameter committed
859

860
	fp = page->freelist;
861
	while (fp && nr <= page->objects) {
Christoph Lameter's avatar
Christoph Lameter committed
862
863
864
865
866
867
		if (fp == search)
			return 1;
		if (!check_valid_pointer(s, page, fp)) {
			if (object) {
				object_err(s, page, object,
					"Freechain corrupt");
868
				set_freepointer(s, object, NULL);
Christoph Lameter's avatar
Christoph Lameter committed
869
870
				break;
			} else {
871
				slab_err(s, page, "Freepointer corrupt");
872
				page->freelist = NULL;
873
				page->inuse = page->objects;
874
				slab_fix(s, "Freelist cleared");
Christoph Lameter's avatar
Christoph Lameter committed
875
876
877
878
879
880
881
882
883
				return 0;
			}
			break;
		}
		object = fp;
		fp = get_freepointer(s, object);
		nr++;
	}

884
	max_objects = order_objects(compound_order(page), s->size, s->reserved);
885
886
	if (max_objects > MAX_OBJS_PER_PAGE)
		max_objects = MAX_OBJS_PER_PAGE;
887
888
889
890
891
892
893

	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.");
	}
894
	if (page->inuse != page->objects - nr) {
895
		slab_err(s, page, "Wrong object count. Counter is %d but "
896
897
			"counted were %d", page->inuse, page->objects - nr);
		page->inuse = page->objects - nr;
898
		slab_fix(s, "Object count adjusted.");
Christoph Lameter's avatar
Christoph Lameter committed
899
900
901
902
	}
	return search == NULL;
}

903
904
static void trace(struct kmem_cache *s, struct page *page, void *object,
								int alloc)
Christoph Lameter's avatar
Christoph Lameter committed
905
906
907
908
909
910
911
912
913
{
	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)
914
			print_section("Object ", (void *)object, s->object_size);
Christoph Lameter's avatar
Christoph Lameter committed
915
916
917
918
919

		dump_stack();
	}
}

920
921
922
923
924
925
/*
 * 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)
{
926
	flags &= gfp_allowed_mask;
927
928
929
	lockdep_trace_alloc(flags);
	might_sleep_if(flags & __GFP_WAIT);

930
	return should_failslab(s->object_size, flags, s->flags);
931
932
933
934
}

static inline void slab_post_alloc_hook(struct kmem_cache *s, gfp_t flags, void *object)
{
935
	flags &= gfp_allowed_mask;
936
	kmemcheck_slab_alloc(s, flags, object, slab_ksize(s));
937
	kmemleak_alloc_recursive(object, s->object_size, 1, s->flags, flags);
938
939
940
941
942
943
}

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

944
945
946
947
948
949
950
951
952
953
	/*
	 * 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);
954
955
		kmemcheck_slab_free(s, x, s->object_size);
		debug_check_no_locks_freed(x, s->object_size);
956
957
958
		local_irq_restore(flags);
	}
#endif
959
	if (!(s->flags & SLAB_DEBUG_OBJECTS))
960
		debug_check_no_obj_freed(x, s->object_size);
961
962
}

963
/*
Christoph Lameter's avatar
Christoph Lameter committed
964
 * Tracking of fully allocated slabs for debugging purposes.
965
966
 *
 * list_lock must be held.
967
 */
968
969
static void add_full(struct kmem_cache *s,
	struct kmem_cache_node *n, struct page *page)
970
{
971
972
973
	if (!(s->flags & SLAB_STORE_USER))
		return;

974
975
976
	list_add(&page->lru, &n->full);
}

977
978
979
/*
 * list_lock must be held.
 */
980
981
982
983
984
985
986
987
static void remove_full(struct kmem_cache *s, struct page *page)
{
	if (!(s->flags & SLAB_STORE_USER))
		return;

	list_del(&page->lru);
}

988
989
990
991
992
993
994
995
/* 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);
}

996
997
998
999
1000
static inline unsigned long node_nr_slabs(struct kmem_cache_node *n)
{
	return atomic_long_read(&n->nr_slabs);
}

1001
static inline void inc_slabs_node(struct kmem_cache *s, int node, int objects)
1002
1003
1004
1005
1006
1007
1008
1009
1010
{
	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).
	 */
1011
	if (n) {
1012
		atomic_long_inc(&n->nr_slabs);
1013
1014
		atomic_long_add(objects, &n->total_objects);
	}
1015
}
1016
static inline void dec_slabs_node(struct kmem_cache *s, int node, int objects)
1017
1018
1019
1020
{
	struct kmem_cache_node *n = get_node(s, node);

	atomic_long_dec(&n->nr_slabs);
1021
	atomic_long_sub(objects, &n->total_objects);
1022
1023
1024
}

/* Object debug checks for alloc/free paths */
Christoph Lameter's avatar
Christoph Lameter committed
1025
1026
1027
1028
1029
1030
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;

1031
	init_object(s, object, SLUB_RED_INACTIVE);
Christoph Lameter's avatar
Christoph Lameter committed
1032
1033
1034
	init_tracking(s, object);
}

1035
static noinline int alloc_debug_processing(struct kmem_cache *s, struct page *page,
1036
					void *object, unsigned long addr)
Christoph Lameter's avatar
Christoph Lameter committed
1037
1038
1039
1040
1041
1042
{
	if (!check_slab(s, page))
		goto bad;

	if (!check_valid_pointer(s, page, object)) {
		object_err(s, page, object, "Freelist Pointer check fails");
1043
		goto bad;
Christoph Lameter's avatar
Christoph Lameter committed
1044
1045
	}

1046
	if (!check_object(s, page, object, SLUB_RED_INACTIVE))
Christoph Lameter's avatar
Christoph Lameter committed
1047
1048
		goto bad;

Christoph Lameter's avatar
Christoph Lameter committed
1049
1050
1051
1052
	/* 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);
1053
	init_object(s, object, SLUB_RED_ACTIVE);
Christoph Lameter's avatar
Christoph Lameter committed
1054
	return 1;
Christoph Lameter's avatar
Christoph Lameter committed
1055

Christoph Lameter's avatar
Christoph Lameter committed
1056
1057
1058
1059
1060
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
1061
		 * as used avoids touching the remaining objects.
Christoph Lameter's avatar
Christoph Lameter committed
1062
		 */
1063
		slab_fix(s, "Marking all objects used");
1064
		page->inuse = page->objects;
1065
		page->freelist = NULL;
Christoph Lameter's avatar
Christoph Lameter committed
1066
1067
1068
1069
	}
	return 0;
}

1070
1071
static noinline int free_debug_processing(struct kmem_cache *s,
		 struct page *page, void *object, unsigned long addr)
Christoph Lameter's avatar
Christoph Lameter committed
1072
{
1073
1074
1075
1076
	unsigned long flags;
	int rc = 0;

	local_irq_save(flags);
1077
1078
	slab_lock(page);

Christoph Lameter's avatar
Christoph Lameter committed
1079
1080
1081
1082
	if (!check_slab(s, page))
		goto fail;

	if (!check_valid_pointer(s, page, object)) {
1083
		slab_err(s, page, "Invalid object pointer 0x%p", object);
Christoph Lameter's avatar
Christoph Lameter committed
1084
1085
1086
1087
		goto fail;
	}

	if (on_freelist(s, page, object)) {
1088
		object_err(s, page, object, "Object already free");
Christoph Lameter's avatar
Christoph Lameter committed
1089
1090
1091
		goto fail;
	}

1092
	if (!check_object(s, page, object, SLUB_RED_ACTIVE))
1093
		goto out;
Christoph Lameter's avatar
Christoph Lameter committed
1094
1095

	if (unlikely(s != page->slab)) {
Ingo Molnar's avatar
Ingo Molnar committed
1096
		if (!PageSlab(page)) {
1097
1098
			slab_err(s, page, "Attempt to free object(0x%p) "
				"outside of slab", object);
Ingo Molnar's avatar
Ingo Molnar committed
1099
		} else if (!page->slab) {
Christoph Lameter's avatar
Christoph Lameter committed
1100
			printk(KERN_ERR
1101
				"SLUB <none>: no slab for object 0x%p.\n",
Christoph Lameter's avatar
Christoph Lameter committed
1102
						object);
1103
			dump_stack();
1104
		} else
1105
1106
			object_err(s, page, object,
					"page slab pointer corrupt.");
Christoph Lameter's avatar
Christoph Lameter committed
1107
1108
		goto fail;
	}
Christoph Lameter's avatar
Christoph Lameter committed
1109
1110
1111
1112

	if (s->flags & SLAB_STORE_USER)
		set_track(s, object, TRACK_FREE, addr);
	trace(s, page, object, 0);
1113
	init_object(s, object, SLUB_RED_INACTIVE);
1114
1115
	rc = 1;
out:
1116
	slab_unlock(page);
1117
1118
	local_irq_restore(flags);
	return rc;
Christoph Lameter's avatar
Christoph Lameter committed
1119

Christoph Lameter's avatar
Christoph Lameter committed
1120
fail:
1121
	slab_fix(s, "Object at 0x%p not freed", object);
1122
	goto out;
Christoph Lameter's avatar
Christoph Lameter committed
1123
1124
}

1125
1126
static int __init setup_slub_debug(char *str)
{
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
	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;

1141
1142
1143
1144
1145
1146
1147
1148
1149
	if (tolower(*str) == 'o') {
		/*
		 * Avoid enabling debugging on caches if its minimum order
		 * would increase as a result.
		 */
		disable_higher_order_debug = 1;
		goto out;
	}

1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
	slub_debug = 0;
	if (*str == '-')
		/*
		 * Switch off all debugging measures.
		 */
		goto out;

	/*
	 * Determine which debug features should be switched on
	 */
1160
	for (; *str && *str != ','; str++) {
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
		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;
1177
1178
1179
		case 'a':
			slub_debug |= SLAB_FAILSLAB;
			break;
1180
1181
		default:
			printk(KERN_ERR "slub_debug option '%c' "
1182
				"unknown. skipped\n", *str);
1183
		}
1184
1185
	}

1186
check_slabs:
1187
1188
	if (*str == ',')
		slub_debug_slabs = str + 1;
1189
out:
1190
1191
1192
1193
1194