slub.c 126 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>
34
#include <linux/memcontrol.h>
Christoph Lameter's avatar
Christoph Lameter committed
35

36
37
#include <trace/events/kmem.h>

38
39
#include "internal.h"

Christoph Lameter's avatar
Christoph Lameter committed
40
41
/*
 * Lock order:
42
 *   1. slab_mutex (Global Mutex)
43
44
 *   2. node->list_lock
 *   3. slab_lock(page) (Only on some arches and for debugging)
Christoph Lameter's avatar
Christoph Lameter committed
45
 *
46
 *   slab_mutex
47
 *
48
 *   The role of the slab_mutex is to protect the list of all the slabs
49
50
51
52
53
54
55
56
57
58
59
60
61
62
 *   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
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
 *
 *   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
83
84
 * 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
85
 * freed then the slab will show up again on the partial lists.
Christoph Lameter's avatar
Christoph Lameter committed
86
87
 * We track full slabs for debugging purposes though because otherwise we
 * cannot scan all objects.
Christoph Lameter's avatar
Christoph Lameter committed
88
89
90
91
92
93
94
 *
 * 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.
 *
95
96
97
98
99
100
101
102
103
104
105
106
 * 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
107
 * 			freelist that allows lockless access to
108
109
 * 			free objects in addition to the regular freelist
 * 			that requires the slab lock.
Christoph Lameter's avatar
Christoph Lameter committed
110
111
112
 *
 * PageError		Slab requires special handling due to debug
 * 			options set. This moves	slab handling out of
113
 * 			the fast path and disables lockless freelists.
Christoph Lameter's avatar
Christoph Lameter committed
114
115
 */

116
117
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

#ifdef CONFIG_SMP
static struct notifier_block slab_notifier;
#endif

184
185
186
/*
 * Tracking user of a slab.
 */
187
#define TRACK_ADDRS_COUNT 16
188
struct track {
189
	unsigned long addr;	/* Called from address */
190
191
192
#ifdef CONFIG_STACKTRACE
	unsigned long addrs[TRACK_ADDRS_COUNT];	/* Called from address */
#endif
193
194
195
196
197
198
199
	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 };

200
#ifdef CONFIG_SYSFS
Christoph Lameter's avatar
Christoph Lameter committed
201
202
203
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 *);
204
static void memcg_propagate_slab_attrs(struct kmem_cache *s);
Christoph Lameter's avatar
Christoph Lameter committed
205
#else
206
207
208
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; }
209
static inline void sysfs_slab_remove(struct kmem_cache *s) { }
210

211
static inline void memcg_propagate_slab_attrs(struct kmem_cache *s) { }
Christoph Lameter's avatar
Christoph Lameter committed
212
213
#endif

214
static inline void stat(const struct kmem_cache *s, enum stat_item si)
215
216
{
#ifdef CONFIG_SLUB_STATS
217
	__this_cpu_inc(s->cpu_slab->stat[si]);
218
219
220
#endif
}

Christoph Lameter's avatar
Christoph Lameter committed
221
222
223
224
225
226
227
228
229
/********************************************************************
 * 			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
230
/* Verify that a pointer has an address that is valid within a slab page */
231
232
233
234
235
static inline int check_valid_pointer(struct kmem_cache *s,
				struct page *page, const void *object)
{
	void *base;

236
	if (!object)
237
238
		return 1;

239
	base = page_address(page);
240
	if (object < base || object >= base + page->objects * s->size ||
241
242
243
244
245
246
247
		(object - base) % s->size) {
		return 0;
	}

	return 1;
}

248
249
250
251
252
static inline void *get_freepointer(struct kmem_cache *s, void *object)
{
	return *(void **)(object + s->offset);
}

253
254
255
256
257
static void prefetch_freepointer(const struct kmem_cache *s, void *object)
{
	prefetch(object + s->offset);
}

258
259
260
261
262
263
264
265
266
267
268
269
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;
}

270
271
272
273
274
275
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 */
276
277
#define for_each_object(__p, __s, __addr, __objects) \
	for (__p = (__addr); __p < (__addr) + (__objects) * (__s)->size;\
278
279
280
281
282
283
284
285
			__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;
}

286
287
288
289
290
291
292
293
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))
294
		return s->object_size;
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309

#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;
}

310
311
312
313
314
static inline int order_objects(int order, unsigned long size, int reserved)
{
	return ((PAGE_SIZE << order) - reserved) / size;
}

315
static inline struct kmem_cache_order_objects oo_make(int order,
316
		unsigned long size, int reserved)
317
318
{
	struct kmem_cache_order_objects x = {
319
		(order << OO_SHIFT) + order_objects(order, size, reserved)
320
321
322
323
324
325
326
	};

	return x;
}

static inline int oo_order(struct kmem_cache_order_objects x)
{
327
	return x.x >> OO_SHIFT;
328
329
330
331
}

static inline int oo_objects(struct kmem_cache_order_objects x)
{
332
	return x.x & OO_MASK;
333
334
}

335
336
337
338
339
340
341
342
343
344
345
346
347
/*
 * 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);
}

348
349
350
351
352
353
354
/* 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());
355
356
#if defined(CONFIG_HAVE_CMPXCHG_DOUBLE) && \
    defined(CONFIG_HAVE_ALIGNED_STRUCT_PAGE)
357
	if (s->flags & __CMPXCHG_DOUBLE) {
358
		if (cmpxchg_double(&page->freelist, &page->counters,
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
			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;
}

385
386
387
388
389
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)
{
390
391
#if defined(CONFIG_HAVE_CMPXCHG_DOUBLE) && \
    defined(CONFIG_HAVE_ALIGNED_STRUCT_PAGE)
392
	if (s->flags & __CMPXCHG_DOUBLE) {
393
		if (cmpxchg_double(&page->freelist, &page->counters,
394
395
396
397
398
399
			freelist_old, counters_old,
			freelist_new, counters_new))
		return 1;
	} else
#endif
	{
400
401
402
		unsigned long flags;

		local_irq_save(flags);
403
		slab_lock(page);
404
405
406
		if (page->freelist == freelist_old && page->counters == counters_old) {
			page->freelist = freelist_new;
			page->counters = counters_new;
407
			slab_unlock(page);
408
			local_irq_restore(flags);
409
410
			return 1;
		}
411
		slab_unlock(page);
412
		local_irq_restore(flags);
413
414
415
416
417
418
419
420
421
422
423
424
	}

	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;
}

425
#ifdef CONFIG_SLUB_DEBUG
426
427
428
/*
 * Determine a map of object in use on a page.
 *
429
 * Node listlock must be held to guarantee that the page does
430
431
432
433
434
435
436
437
438
439
440
 * 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);
}

441
442
443
/*
 * Debug settings:
 */
444
445
446
#ifdef CONFIG_SLUB_DEBUG_ON
static int slub_debug = DEBUG_DEFAULT_FLAGS;
#else
447
static int slub_debug;
448
#endif
449
450

static char *slub_debug_slabs;
451
static int disable_higher_order_debug;
452

Christoph Lameter's avatar
Christoph Lameter committed
453
454
455
456
457
/*
 * Object debugging
 */
static void print_section(char *text, u8 *addr, unsigned int length)
{
458
459
	print_hex_dump(KERN_ERR, text, DUMP_PREFIX_ADDRESS, 16, 1, addr,
			length, 1);
Christoph Lameter's avatar
Christoph Lameter committed
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
}

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,
476
			enum track_item alloc, unsigned long addr)
Christoph Lameter's avatar
Christoph Lameter committed
477
{
Akinobu Mita's avatar
Akinobu Mita committed
478
	struct track *p = get_track(s, object, alloc);
Christoph Lameter's avatar
Christoph Lameter committed
479
480

	if (addr) {
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
#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
499
500
		p->addr = addr;
		p->cpu = smp_processor_id();
501
		p->pid = current->pid;
Christoph Lameter's avatar
Christoph Lameter committed
502
503
504
505
506
507
508
		p->when = jiffies;
	} else
		memset(p, 0, sizeof(struct track));
}

static void init_tracking(struct kmem_cache *s, void *object)
{
509
510
511
	if (!(s->flags & SLAB_STORE_USER))
		return;

512
513
	set_track(s, object, TRACK_FREE, 0UL);
	set_track(s, object, TRACK_ALLOC, 0UL);
Christoph Lameter's avatar
Christoph Lameter committed
514
515
516
517
518
519
520
}

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

521
	printk(KERN_ERR "INFO: %s in %pS age=%lu cpu=%u pid=%d\n",
522
		s, (void *)t->addr, jiffies - t->when, t->cpu, t->pid);
523
524
525
526
527
528
529
530
531
532
#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
533
534
535
536
537
538
539
540
541
542
543
544
545
}

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)
{
546
547
	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);
548
549
550
551
552
553
554
555
556
557
558
559
560

}

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");
561
	printk(KERN_ERR "BUG %s (%s): %s\n", s->name, print_tainted(), buf);
562
563
	printk(KERN_ERR "----------------------------------------"
			"-------------------------------------\n\n");
564
565

	add_taint(TAINT_BAD_PAGE);
Christoph Lameter's avatar
Christoph Lameter committed
566
567
}

568
569
570
571
572
573
574
575
576
577
578
579
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
580
581
{
	unsigned int off;	/* Offset of last byte */
582
	u8 *addr = page_address(page);
583
584
585
586
587
588
589
590
591

	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)
592
		print_section("Bytes b4 ", p - 16, 16);
Christoph Lameter's avatar
Christoph Lameter committed
593

594
	print_section("Object ", p, min_t(unsigned long, s->object_size,
595
				PAGE_SIZE));
Christoph Lameter's avatar
Christoph Lameter committed
596
	if (s->flags & SLAB_RED_ZONE)
597
598
		print_section("Redzone ", p + s->object_size,
			s->inuse - s->object_size);
Christoph Lameter's avatar
Christoph Lameter committed
599
600
601
602
603
604

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

605
	if (s->flags & SLAB_STORE_USER)
Christoph Lameter's avatar
Christoph Lameter committed
606
607
608
609
		off += 2 * sizeof(struct track);

	if (off != s->size)
		/* Beginning of the filler is the free pointer */
610
		print_section("Padding ", p + off, s->size - off);
611
612

	dump_stack();
Christoph Lameter's avatar
Christoph Lameter committed
613
614
615
616
617
}

static void object_err(struct kmem_cache *s, struct page *page,
			u8 *object, char *reason)
{
618
	slab_bug(s, "%s", reason);
619
	print_trailer(s, page, object);
Christoph Lameter's avatar
Christoph Lameter committed
620
621
}

622
static void slab_err(struct kmem_cache *s, struct page *page, const char *fmt, ...)
Christoph Lameter's avatar
Christoph Lameter committed
623
624
625
626
{
	va_list args;
	char buf[100];

627
628
	va_start(args, fmt);
	vsnprintf(buf, sizeof(buf), fmt, args);
Christoph Lameter's avatar
Christoph Lameter committed
629
	va_end(args);
630
	slab_bug(s, "%s", buf);
631
	print_page_info(page);
Christoph Lameter's avatar
Christoph Lameter committed
632
633
634
	dump_stack();
}

635
static void init_object(struct kmem_cache *s, void *object, u8 val)
Christoph Lameter's avatar
Christoph Lameter committed
636
637
638
639
{
	u8 *p = object;

	if (s->flags & __OBJECT_POISON) {
640
641
		memset(p, POISON_FREE, s->object_size - 1);
		p[s->object_size - 1] = POISON_END;
Christoph Lameter's avatar
Christoph Lameter committed
642
643
644
	}

	if (s->flags & SLAB_RED_ZONE)
645
		memset(p + s->object_size, val, s->inuse - s->object_size);
Christoph Lameter's avatar
Christoph Lameter committed
646
647
}

648
649
650
651
652
653
654
655
656
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,
657
			u8 *start, unsigned int value, unsigned int bytes)
658
659
660
661
{
	u8 *fault;
	u8 *end;

662
	fault = memchr_inv(start, value, bytes);
663
664
665
666
667
668
669
670
671
672
673
674
675
676
	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
677
678
679
680
681
682
683
684
685
}

/*
 * 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
686
 *
Christoph Lameter's avatar
Christoph Lameter committed
687
688
689
 * 	Poisoning uses 0x6b (POISON_FREE) and the last byte is
 * 	0xa5 (POISON_END)
 *
690
 * object + s->object_size
Christoph Lameter's avatar
Christoph Lameter committed
691
 * 	Padding to reach word boundary. This is also used for Redzoning.
Christoph Lameter's avatar
Christoph Lameter committed
692
 * 	Padding is extended by another word if Redzoning is enabled and
693
 * 	object_size == inuse.
Christoph Lameter's avatar
Christoph Lameter committed
694
 *
Christoph Lameter's avatar
Christoph Lameter committed
695
696
697
698
 * 	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
699
700
 * 	Meta data starts here.
 *
Christoph Lameter's avatar
Christoph Lameter committed
701
702
 * 	A. Free pointer (if we cannot overwrite object on free)
 * 	B. Tracking data for SLAB_STORE_USER
Christoph Lameter's avatar
Christoph Lameter committed
703
 * 	C. Padding to reach required alignment boundary or at mininum
Christoph Lameter's avatar
Christoph Lameter committed
704
 * 		one word if debugging is on to be able to detect writes
Christoph Lameter's avatar
Christoph Lameter committed
705
706
707
 * 		before the word boundary.
 *
 *	Padding is done using 0x5a (POISON_INUSE)
Christoph Lameter's avatar
Christoph Lameter committed
708
709
 *
 * object + s->size
Christoph Lameter's avatar
Christoph Lameter committed
710
 * 	Nothing is used beyond s->size.
Christoph Lameter's avatar
Christoph Lameter committed
711
 *
712
 * If slabcaches are merged then the object_size and inuse boundaries are mostly
Christoph Lameter's avatar
Christoph Lameter committed
713
 * ignored. And therefore no slab options that rely on these boundaries
Christoph Lameter's avatar
Christoph Lameter committed
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
 * 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;

732
733
	return check_bytes_and_report(s, page, p, "Object padding",
				p + off, POISON_INUSE, s->size - off);
Christoph Lameter's avatar
Christoph Lameter committed
734
735
}

736
/* Check the pad bytes at the end of a slab page */
Christoph Lameter's avatar
Christoph Lameter committed
737
738
static int slab_pad_check(struct kmem_cache *s, struct page *page)
{
739
740
741
742
743
	u8 *start;
	u8 *fault;
	u8 *end;
	int length;
	int remainder;
Christoph Lameter's avatar
Christoph Lameter committed
744
745
746
747

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

748
	start = page_address(page);
749
	length = (PAGE_SIZE << compound_order(page)) - s->reserved;
750
751
	end = start + length;
	remainder = length % s->size;
Christoph Lameter's avatar
Christoph Lameter committed
752
753
754
	if (!remainder)
		return 1;

755
	fault = memchr_inv(end - remainder, POISON_INUSE, remainder);
756
757
758
759
760
761
	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);
762
	print_section("Padding ", end - remainder, remainder);
763

Eric Dumazet's avatar
Eric Dumazet committed
764
	restore_bytes(s, "slab padding", POISON_INUSE, end - remainder, end);
765
	return 0;
Christoph Lameter's avatar
Christoph Lameter committed
766
767
768
}

static int check_object(struct kmem_cache *s, struct page *page,
769
					void *object, u8 val)
Christoph Lameter's avatar
Christoph Lameter committed
770
771
{
	u8 *p = object;
772
	u8 *endobject = object + s->object_size;
Christoph Lameter's avatar
Christoph Lameter committed
773
774

	if (s->flags & SLAB_RED_ZONE) {
775
		if (!check_bytes_and_report(s, page, object, "Redzone",
776
			endobject, val, s->inuse - s->object_size))
Christoph Lameter's avatar
Christoph Lameter committed
777
778
			return 0;
	} else {
779
		if ((s->flags & SLAB_POISON) && s->object_size < s->inuse) {
Ingo Molnar's avatar
Ingo Molnar committed
780
			check_bytes_and_report(s, page, p, "Alignment padding",
781
				endobject, POISON_INUSE, s->inuse - s->object_size);
Ingo Molnar's avatar
Ingo Molnar committed
782
		}
Christoph Lameter's avatar
Christoph Lameter committed
783
784
785
	}

	if (s->flags & SLAB_POISON) {
786
		if (val != SLUB_RED_ACTIVE && (s->flags & __OBJECT_POISON) &&
787
			(!check_bytes_and_report(s, page, p, "Poison", p,
788
					POISON_FREE, s->object_size - 1) ||
789
			 !check_bytes_and_report(s, page, p, "Poison",
790
				p + s->object_size - 1, POISON_END, 1)))
Christoph Lameter's avatar
Christoph Lameter committed
791
792
793
794
795
796
797
			return 0;
		/*
		 * check_pad_bytes cleans up on its own.
		 */
		check_pad_bytes(s, page, p);
	}

798
	if (!s->offset && val == SLUB_RED_ACTIVE)
Christoph Lameter's avatar
Christoph Lameter committed
799
800
801
802
803
804
805
806
807
808
		/*
		 * 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
809
		 * No choice but to zap it and thus lose the remainder
Christoph Lameter's avatar
Christoph Lameter committed
810
		 * of the free objects in this slab. May cause
Christoph Lameter's avatar
Christoph Lameter committed
811
		 * another error because the object count is now wrong.
Christoph Lameter's avatar
Christoph Lameter committed
812
		 */
813
		set_freepointer(s, p, NULL);
Christoph Lameter's avatar
Christoph Lameter committed
814
815
816
817
818
819
820
		return 0;
	}
	return 1;
}

static int check_slab(struct kmem_cache *s, struct page *page)
{
821
822
	int maxobj;

Christoph Lameter's avatar
Christoph Lameter committed
823
824
825
	VM_BUG_ON(!irqs_disabled());

	if (!PageSlab(page)) {
826
		slab_err(s, page, "Not a valid slab page");
Christoph Lameter's avatar
Christoph Lameter committed
827
828
		return 0;
	}
829

830
	maxobj = order_objects(compound_order(page), s->size, s->reserved);
831
832
833
834
835
836
	if (page->objects > maxobj) {
		slab_err(s, page, "objects %u > max %u",
			s->name, page->objects, maxobj);
		return 0;
	}
	if (page->inuse > page->objects) {
837
		slab_err(s, page, "inuse %u > max %u",
838
			s->name, page->inuse, page->objects);
Christoph Lameter's avatar
Christoph Lameter committed
839
840
841
842
843
844
845
846
		return 0;
	}
	/* Slab_pad_check fixes things up after itself */
	slab_pad_check(s, page);
	return 1;
}

/*
Christoph Lameter's avatar
Christoph Lameter committed
847
848
 * 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
849
850
851
852
 */
static int on_freelist(struct kmem_cache *s, struct page *page, void *search)
{
	int nr = 0;
853
	void *fp;
Christoph Lameter's avatar
Christoph Lameter committed
854
	void *object = NULL;
855
	unsigned long max_objects;
Christoph Lameter's avatar
Christoph Lameter committed
856

857
	fp = page->freelist;
858
	while (fp && nr <= page->objects) {
Christoph Lameter's avatar
Christoph Lameter committed
859
860
861
862
863
864
		if (fp == search)
			return 1;
		if (!check_valid_pointer(s, page, fp)) {
			if (object) {
				object_err(s, page, object,
					"Freechain corrupt");
865
				set_freepointer(s, object, NULL);
Christoph Lameter's avatar
Christoph Lameter committed
866
867
				break;
			} else {
868
				slab_err(s, page, "Freepointer corrupt");
869
				page->freelist = NULL;
870
				page->inuse = page->objects;
871
				slab_fix(s, "Freelist cleared");
Christoph Lameter's avatar
Christoph Lameter committed
872
873
874
875
876
877
878
879
880
				return 0;
			}
			break;
		}
		object = fp;
		fp = get_freepointer(s, object);
		nr++;
	}

881
	max_objects = order_objects(compound_order(page), s->size, s->reserved);
882
883
	if (max_objects > MAX_OBJS_PER_PAGE)
		max_objects = MAX_OBJS_PER_PAGE;
884
885
886
887
888
889
890

	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.");
	}
891
	if (page->inuse != page->objects - nr) {
892
		slab_err(s, page, "Wrong object count. Counter is %d but "
893
894
			"counted were %d", page->inuse, page->objects - nr);
		page->inuse = page->objects - nr;
895
		slab_fix(s, "Object count adjusted.");
Christoph Lameter's avatar
Christoph Lameter committed
896
897
898
899
	}
	return search == NULL;
}

900
901
static void trace(struct kmem_cache *s, struct page *page, void *object,
								int alloc)
Christoph Lameter's avatar
Christoph Lameter committed
902
903
904
905
906
907
908
909
910
{
	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)
911
			print_section("Object ", (void *)object, s->object_size);
Christoph Lameter's avatar
Christoph Lameter committed
912
913
914
915
916

		dump_stack();
	}
}

917
918
919
920
921
922
/*
 * 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)
{
923
	flags &= gfp_allowed_mask;
924
925
926
	lockdep_trace_alloc(flags);
	might_sleep_if(flags & __GFP_WAIT);

927
	return should_failslab(s->object_size, flags, s->flags);
928
929
930
931
}

static inline void slab_post_alloc_hook(struct kmem_cache *s, gfp_t flags, void *object)
{
932
	flags &= gfp_allowed_mask;
933
	kmemcheck_slab_alloc(s, flags, object, slab_ksize(s));
934
	kmemleak_alloc_recursive(object, s->object_size, 1, s->flags, flags);
935
936
937
938
939
940
}

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

941
942
943
944
945
946
947
948
949
950
	/*
	 * 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);
951
952
		kmemcheck_slab_free(s, x, s->object_size);
		debug_check_no_locks_freed(x, s->object_size);
953
954
955
		local_irq_restore(flags);
	}
#endif
956
	if (!(s->flags & SLAB_DEBUG_OBJECTS))
957
		debug_check_no_obj_freed(x, s->object_size);
958
959
}

960
/*
Christoph Lameter's avatar
Christoph Lameter committed
961
 * Tracking of fully allocated slabs for debugging purposes.
962
963
 *
 * list_lock must be held.
964
 */
965
966
static void add_full(struct kmem_cache *s,
	struct kmem_cache_node *n, struct page *page)
967
{
968
969
970
	if (!(s->flags & SLAB_STORE_USER))
		return;

971
972
973
	list_add(&page->lru, &n->full);
}

974
975
976
/*
 * list_lock must be held.
 */
977
978
979
980
981
982
983
984
static void remove_full(struct kmem_cache *s, struct page *page)
{
	if (!(s->flags & SLAB_STORE_USER))
		return;

	list_del(&page->lru);
}

985
986
987
988
989
990
991
992
/* 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);
}

993
994
995
996
997
static inline unsigned long node_nr_slabs(struct kmem_cache_node *n)
{
	return atomic_long_read(&n->nr_slabs);
}

998
static inline void inc_slabs_node(struct kmem_cache *s, int node, int objects)
999
1000
1001
1002
1003
1004
1005
1006
1007
{
	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).
	 */
1008
	if (likely(n)) {
1009
		atomic_long_inc(&n->nr_slabs);
1010
1011
		atomic_long_add(objects, &n->total_objects);
	}
1012
}
1013
static inline void dec_slabs_node(struct kmem_cache *s, int node, int objects)
1014
1015
1016
1017
{
	struct kmem_cache_node *n = get_node(s, node);

	atomic_long_dec(&n->nr_slabs);
1018
	atomic_long_sub(objects, &n->total_objects);
1019
1020
1021
}

/* Object debug checks for alloc/free paths */
Christoph Lameter's avatar
Christoph Lameter committed
1022
1023
1024
1025
1026
1027
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;

1028
	init_object(s, object, SLUB_RED_INACTIVE);
Christoph Lameter's avatar
Christoph Lameter committed
1029
1030
1031
	init_tracking(s, object);
}

1032
static noinline int alloc_debug_processing(struct kmem_cache *s, struct page *page,
1033
					void *object, unsigned long addr)
Christoph Lameter's avatar
Christoph Lameter committed
1034
1035
1036
1037
1038
1039
{
	if (!check_slab(s, page))
		goto bad;

	if (!check_valid_pointer(s, page, object)) {
		object_err(s, page, object, "Freelist Pointer check fails");
1040
		goto bad;
Christoph Lameter's avatar
Christoph Lameter committed
1041
1042
	}

1043
	if (!check_object(s, page, object, SLUB_RED_INACTIVE))
Christoph Lameter's avatar
Christoph Lameter committed
1044
1045
		goto bad;

Christoph Lameter's avatar
Christoph Lameter committed
1046
1047
1048
1049
	/* 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);
1050
	init_object(s, object, SLUB_RED_ACTIVE);
Christoph Lameter's avatar
Christoph Lameter committed
1051
	return 1;
Christoph Lameter's avatar
Christoph Lameter committed
1052

Christoph Lameter's avatar
Christoph Lameter committed
1053
1054
1055
1056
1057
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
1058
		 * as used avoids touching the remaining objects.
Christoph Lameter's avatar
Christoph Lameter committed
1059
		 */
1060
		slab_fix(s, "Marking all objects used");
1061
		page->inuse = page->objects;
1062
		page->freelist = NULL;
Christoph Lameter's avatar
Christoph Lameter committed
1063
1064
1065
1066
	}
	return 0;
}

1067
1068
1069
static noinline struct kmem_cache_node *free_debug_processing(
	struct kmem_cache *s, struct page *page, void *object,
	unsigned long addr, unsigned long *flags)
Christoph Lameter's avatar
Christoph Lameter committed
1070
{
1071
	struct kmem_cache_node *n = get_node(s, page_to_nid(page));
1072

1073
	spin_lock_irqsave(&n->list_lock, *flags);
1074
1075
	slab_lock(page);

Christoph Lameter's avatar
Christoph Lameter committed
1076
1077
1078
1079
	if (!check_slab(s, page))
		goto fail;

	if (!check_valid_pointer(s, page, object)) {
1080
		slab_err(s, page, "Invalid object pointer 0x%p", object);
Christoph Lameter's avatar
Christoph Lameter committed
1081
1082
1083
1084
		goto fail;
	}

	if (on_freelist(s, page, object)) {
1085
		object_err(s, page, object, "Object already free");
Christoph Lameter's avatar
Christoph Lameter committed
1086
1087
1088
		goto fail;
	}

1089
	if (!check_object(s, page, object, SLUB_RED_ACTIVE))
1090
		goto out;
Christoph Lameter's avatar
Christoph Lameter committed
1091

1092
	if (unlikely(s != page->slab_cache)) {
Ingo Molnar's avatar
Ingo Molnar committed
1093
		if (!PageSlab(page)) {
1094
1095
			slab_err(s, page, "Attempt to free object(0x%p) "
				"outside of slab", object);
1096
		} else if (!page->slab_cache) {
Christoph Lameter's avatar
Christoph Lameter committed
1097
			printk(KERN_ERR
1098
				"SLUB <none>: no slab for object 0x%p.\n",
Christoph Lameter's avatar
Christoph Lameter committed
1099
						object);
1100
			dump_stack();
1101
		} else
1102
1103
			object_err(s, page, object,
					"page slab pointer corrupt.");
Christoph Lameter's avatar
Christoph Lameter committed
1104
1105
		goto fail;
	}
Christoph Lameter's avatar
Christoph Lameter committed
1106
1107
1108
1109

	if (s->flags & SLAB_STORE_USER)
		set_track(s, object, TRACK_FREE, addr);
	trace(s, page, object, 0);
1110
	init_object(s, object, SLUB_RED_INACTIVE);
1111
out:
1112
	slab_unlock(page);
1113
1114
1115
1116
1117
	/*
	 * Keep node_lock to preserve integrity
	 * until the object is actually freed
	 */
	return n;
Christoph Lameter's avatar
Christoph Lameter committed
1118

Christoph Lameter's avatar
Christoph Lameter committed
1119
fail:
1120
1121
	slab_unlock(page);
	spin_unlock_irqrestore(&n->list_lock, *flags);
1122
	slab_fix(s, "Object at 0x%p not freed", object);
1123
	return NULL;
Christoph Lameter's avatar
Christoph Lameter committed
1124
1125
}

1126
1127
static int __init setup_slub_debug(char *str)
{
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
	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;

1142
1143
1144
1145
1146
1147
1148
1149
1150
	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;
	}

1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
	slub_debug = 0;
	if (*str == '-')
		/*
		 * Switch off all debugging measures.
		 */
		goto out;

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