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>
Christoph Lameter's avatar
Christoph Lameter committed
34

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

37
38
#include "internal.h"

Christoph Lameter's avatar
Christoph Lameter committed
39
40
/*
 * Lock order:
41
 *   1. slab_mutex (Global Mutex)
42
43
 *   2. node->list_lock
 *   3. slab_lock(page) (Only on some arches and for debugging)
Christoph Lameter's avatar
Christoph Lameter committed
44
 *
45
 *   slab_mutex
46
 *
47
 *   The role of the slab_mutex is to protect the list of all the slabs
48
49
50
51
52
53
54
55
56
57
58
59
60
61
 *   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
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
 *
 *   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
82
83
 * 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
84
 * freed then the slab will show up again on the partial lists.
Christoph Lameter's avatar
Christoph Lameter committed
85
86
 * We track full slabs for debugging purposes though because otherwise we
 * cannot scan all objects.
Christoph Lameter's avatar
Christoph Lameter committed
87
88
89
90
91
92
93
 *
 * 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.
 *
94
95
96
97
98
99
100
101
102
103
104
105
 * 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
106
 * 			freelist that allows lockless access to
107
108
 * 			free objects in addition to the regular freelist
 * 			that requires the slab lock.
Christoph Lameter's avatar
Christoph Lameter committed
109
110
111
 *
 * PageError		Slab requires special handling due to debug
 * 			options set. This moves	slab handling out of
112
 * 			the fast path and disables lockless freelists.
Christoph Lameter's avatar
Christoph Lameter committed
113
114
 */

115
116
static inline int kmem_cache_debug(struct kmem_cache *s)
{
117
#ifdef CONFIG_SLUB_DEBUG
118
	return unlikely(s->flags & SLAB_DEBUG_FLAGS);
119
#else
120
	return 0;
121
#endif
122
}
123

Christoph Lameter's avatar
Christoph Lameter committed
124
125
126
127
128
129
130
131
132
133
134
/*
 * 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

135
136
137
/* Enable to log cmpxchg failures */
#undef SLUB_DEBUG_CMPXCHG

138
139
140
141
/*
 * Mininum number of partial slabs. These will be left on the partial
 * lists even if they are empty. kmem_cache_shrink may reclaim them.
 */
142
#define MIN_PARTIAL 5
Christoph Lameter's avatar
Christoph Lameter committed
143

144
145
146
147
148
149
150
/*
 * 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
151
152
#define DEBUG_DEFAULT_FLAGS (SLAB_DEBUG_FREE | SLAB_RED_ZONE | \
				SLAB_POISON | SLAB_STORE_USER)
Christoph Lameter's avatar
Christoph Lameter committed
153

154
/*
155
156
157
 * 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.
158
 */
159
#define DEBUG_METADATA_FLAGS (SLAB_RED_ZONE | SLAB_POISON | SLAB_STORE_USER)
160

Christoph Lameter's avatar
Christoph Lameter committed
161
162
163
164
/*
 * Set of flags that will prevent slab merging
 */
#define SLUB_NEVER_MERGE (SLAB_RED_ZONE | SLAB_POISON | SLAB_STORE_USER | \
165
166
		SLAB_TRACE | SLAB_DESTROY_BY_RCU | SLAB_NOLEAKTRACE | \
		SLAB_FAILSLAB)
Christoph Lameter's avatar
Christoph Lameter committed
167
168

#define SLUB_MERGE_SAME (SLAB_DEBUG_FREE | SLAB_RECLAIM_ACCOUNT | \
Vegard Nossum's avatar
Vegard Nossum committed
169
		SLAB_CACHE_DMA | SLAB_NOTRACK)
Christoph Lameter's avatar
Christoph Lameter committed
170

171
172
#define OO_SHIFT	16
#define OO_MASK		((1 << OO_SHIFT) - 1)
173
#define MAX_OBJS_PER_PAGE	32767 /* since page.objects is u15 */
174

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

#ifdef CONFIG_SMP
static struct notifier_block slab_notifier;
#endif

183
184
185
/*
 * Tracking user of a slab.
 */
186
#define TRACK_ADDRS_COUNT 16
187
struct track {
188
	unsigned long addr;	/* Called from address */
189
190
191
#ifdef CONFIG_STACKTRACE
	unsigned long addrs[TRACK_ADDRS_COUNT];	/* Called from address */
#endif
192
193
194
195
196
197
198
	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 };

199
#ifdef CONFIG_SYSFS
Christoph Lameter's avatar
Christoph Lameter committed
200
201
202
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 *);
203

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

Christoph Lameter's avatar
Christoph Lameter committed
210
211
#endif

212
static inline void stat(const struct kmem_cache *s, enum stat_item si)
213
214
{
#ifdef CONFIG_SLUB_STATS
215
	__this_cpu_inc(s->cpu_slab->stat[si]);
216
217
218
#endif
}

Christoph Lameter's avatar
Christoph Lameter committed
219
220
221
222
223
224
225
226
227
/********************************************************************
 * 			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
228
/* Verify that a pointer has an address that is valid within a slab page */
229
230
231
232
233
static inline int check_valid_pointer(struct kmem_cache *s,
				struct page *page, const void *object)
{
	void *base;

234
	if (!object)
235
236
		return 1;

237
	base = page_address(page);
238
	if (object < base || object >= base + page->objects * s->size ||
239
240
241
242
243
244
245
		(object - base) % s->size) {
		return 0;
	}

	return 1;
}

246
247
248
249
250
static inline void *get_freepointer(struct kmem_cache *s, void *object)
{
	return *(void **)(object + s->offset);
}

251
252
253
254
255
static void prefetch_freepointer(const struct kmem_cache *s, void *object)
{
	prefetch(object + s->offset);
}

256
257
258
259
260
261
262
263
264
265
266
267
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;
}

268
269
270
271
272
273
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 */
274
275
#define for_each_object(__p, __s, __addr, __objects) \
	for (__p = (__addr); __p < (__addr) + (__objects) * (__s)->size;\
276
277
278
279
280
281
282
283
			__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;
}

284
285
286
287
288
289
290
291
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))
292
		return s->object_size;
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307

#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;
}

308
309
310
311
312
static inline int order_objects(int order, unsigned long size, int reserved)
{
	return ((PAGE_SIZE << order) - reserved) / size;
}

313
static inline struct kmem_cache_order_objects oo_make(int order,
314
		unsigned long size, int reserved)
315
316
{
	struct kmem_cache_order_objects x = {
317
		(order << OO_SHIFT) + order_objects(order, size, reserved)
318
319
320
321
322
323
324
	};

	return x;
}

static inline int oo_order(struct kmem_cache_order_objects x)
{
325
	return x.x >> OO_SHIFT;
326
327
328
329
}

static inline int oo_objects(struct kmem_cache_order_objects x)
{
330
	return x.x & OO_MASK;
331
332
}

333
334
335
336
337
338
339
340
341
342
343
344
345
/*
 * 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);
}

346
347
348
349
350
351
352
/* 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());
353
354
#if defined(CONFIG_HAVE_CMPXCHG_DOUBLE) && \
    defined(CONFIG_HAVE_ALIGNED_STRUCT_PAGE)
355
	if (s->flags & __CMPXCHG_DOUBLE) {
356
		if (cmpxchg_double(&page->freelist, &page->counters,
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
			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;
}

383
384
385
386
387
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)
{
388
389
#if defined(CONFIG_HAVE_CMPXCHG_DOUBLE) && \
    defined(CONFIG_HAVE_ALIGNED_STRUCT_PAGE)
390
	if (s->flags & __CMPXCHG_DOUBLE) {
391
		if (cmpxchg_double(&page->freelist, &page->counters,
392
393
394
395
396
397
			freelist_old, counters_old,
			freelist_new, counters_new))
		return 1;
	} else
#endif
	{
398
399
400
		unsigned long flags;

		local_irq_save(flags);
401
		slab_lock(page);
402
403
404
		if (page->freelist == freelist_old && page->counters == counters_old) {
			page->freelist = freelist_new;
			page->counters = counters_new;
405
			slab_unlock(page);
406
			local_irq_restore(flags);
407
408
			return 1;
		}
409
		slab_unlock(page);
410
		local_irq_restore(flags);
411
412
413
414
415
416
417
418
419
420
421
422
	}

	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;
}

423
#ifdef CONFIG_SLUB_DEBUG
424
425
426
/*
 * Determine a map of object in use on a page.
 *
427
 * Node listlock must be held to guarantee that the page does
428
429
430
431
432
433
434
435
436
437
438
 * 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);
}

439
440
441
/*
 * Debug settings:
 */
442
443
444
#ifdef CONFIG_SLUB_DEBUG_ON
static int slub_debug = DEBUG_DEFAULT_FLAGS;
#else
445
static int slub_debug;
446
#endif
447
448

static char *slub_debug_slabs;
449
static int disable_higher_order_debug;
450

Christoph Lameter's avatar
Christoph Lameter committed
451
452
453
454
455
/*
 * Object debugging
 */
static void print_section(char *text, u8 *addr, unsigned int length)
{
456
457
	print_hex_dump(KERN_ERR, text, DUMP_PREFIX_ADDRESS, 16, 1, addr,
			length, 1);
Christoph Lameter's avatar
Christoph Lameter committed
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
}

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,
474
			enum track_item alloc, unsigned long addr)
Christoph Lameter's avatar
Christoph Lameter committed
475
{
Akinobu Mita's avatar
Akinobu Mita committed
476
	struct track *p = get_track(s, object, alloc);
Christoph Lameter's avatar
Christoph Lameter committed
477
478

	if (addr) {
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
#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
497
498
		p->addr = addr;
		p->cpu = smp_processor_id();
499
		p->pid = current->pid;
Christoph Lameter's avatar
Christoph Lameter committed
500
501
502
503
504
505
506
		p->when = jiffies;
	} else
		memset(p, 0, sizeof(struct track));
}

static void init_tracking(struct kmem_cache *s, void *object)
{
507
508
509
	if (!(s->flags & SLAB_STORE_USER))
		return;

510
511
	set_track(s, object, TRACK_FREE, 0UL);
	set_track(s, object, TRACK_ALLOC, 0UL);
Christoph Lameter's avatar
Christoph Lameter committed
512
513
514
515
516
517
518
}

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

519
	printk(KERN_ERR "INFO: %s in %pS age=%lu cpu=%u pid=%d\n",
520
		s, (void *)t->addr, jiffies - t->when, t->cpu, t->pid);
521
522
523
524
525
526
527
528
529
530
#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
531
532
533
534
535
536
537
538
539
540
541
542
543
}

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)
{
544
545
	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);
546
547
548
549
550
551
552
553
554
555
556
557
558

}

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");
559
	printk(KERN_ERR "BUG %s (%s): %s\n", s->name, print_tainted(), buf);
560
561
	printk(KERN_ERR "----------------------------------------"
			"-------------------------------------\n\n");
562
563

	add_taint(TAINT_BAD_PAGE);
Christoph Lameter's avatar
Christoph Lameter committed
564
565
}

566
567
568
569
570
571
572
573
574
575
576
577
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
578
579
{
	unsigned int off;	/* Offset of last byte */
580
	u8 *addr = page_address(page);
581
582
583
584
585
586
587
588
589

	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)
590
		print_section("Bytes b4 ", p - 16, 16);
Christoph Lameter's avatar
Christoph Lameter committed
591

592
	print_section("Object ", p, min_t(unsigned long, s->object_size,
593
				PAGE_SIZE));
Christoph Lameter's avatar
Christoph Lameter committed
594
	if (s->flags & SLAB_RED_ZONE)
595
596
		print_section("Redzone ", p + s->object_size,
			s->inuse - s->object_size);
Christoph Lameter's avatar
Christoph Lameter committed
597
598
599
600
601
602

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

603
	if (s->flags & SLAB_STORE_USER)
Christoph Lameter's avatar
Christoph Lameter committed
604
605
606
607
		off += 2 * sizeof(struct track);

	if (off != s->size)
		/* Beginning of the filler is the free pointer */
608
		print_section("Padding ", p + off, s->size - off);
609
610

	dump_stack();
Christoph Lameter's avatar
Christoph Lameter committed
611
612
613
614
615
}

static void object_err(struct kmem_cache *s, struct page *page,
			u8 *object, char *reason)
{
616
	slab_bug(s, "%s", reason);
617
	print_trailer(s, page, object);
Christoph Lameter's avatar
Christoph Lameter committed
618
619
}

620
static void slab_err(struct kmem_cache *s, struct page *page, const char *fmt, ...)
Christoph Lameter's avatar
Christoph Lameter committed
621
622
623
624
{
	va_list args;
	char buf[100];

625
626
	va_start(args, fmt);
	vsnprintf(buf, sizeof(buf), fmt, args);
Christoph Lameter's avatar
Christoph Lameter committed
627
	va_end(args);
628
	slab_bug(s, "%s", buf);
629
	print_page_info(page);
Christoph Lameter's avatar
Christoph Lameter committed
630
631
632
	dump_stack();
}

633
static void init_object(struct kmem_cache *s, void *object, u8 val)
Christoph Lameter's avatar
Christoph Lameter committed
634
635
636
637
{
	u8 *p = object;

	if (s->flags & __OBJECT_POISON) {
638
639
		memset(p, POISON_FREE, s->object_size - 1);
		p[s->object_size - 1] = POISON_END;
Christoph Lameter's avatar
Christoph Lameter committed
640
641
642
	}

	if (s->flags & SLAB_RED_ZONE)
643
		memset(p + s->object_size, val, s->inuse - s->object_size);
Christoph Lameter's avatar
Christoph Lameter committed
644
645
}

646
647
648
649
650
651
652
653
654
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,
655
			u8 *start, unsigned int value, unsigned int bytes)
656
657
658
659
{
	u8 *fault;
	u8 *end;

660
	fault = memchr_inv(start, value, bytes);
661
662
663
664
665
666
667
668
669
670
671
672
673
674
	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
675
676
677
678
679
680
681
682
683
}

/*
 * 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
684
 *
Christoph Lameter's avatar
Christoph Lameter committed
685
686
687
 * 	Poisoning uses 0x6b (POISON_FREE) and the last byte is
 * 	0xa5 (POISON_END)
 *
688
 * object + s->object_size
Christoph Lameter's avatar
Christoph Lameter committed
689
 * 	Padding to reach word boundary. This is also used for Redzoning.
Christoph Lameter's avatar
Christoph Lameter committed
690
 * 	Padding is extended by another word if Redzoning is enabled and
691
 * 	object_size == inuse.
Christoph Lameter's avatar
Christoph Lameter committed
692
 *
Christoph Lameter's avatar
Christoph Lameter committed
693
694
695
696
 * 	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
697
698
 * 	Meta data starts here.
 *
Christoph Lameter's avatar
Christoph Lameter committed
699
700
 * 	A. Free pointer (if we cannot overwrite object on free)
 * 	B. Tracking data for SLAB_STORE_USER
Christoph Lameter's avatar
Christoph Lameter committed
701
 * 	C. Padding to reach required alignment boundary or at mininum
Christoph Lameter's avatar
Christoph Lameter committed
702
 * 		one word if debugging is on to be able to detect writes
Christoph Lameter's avatar
Christoph Lameter committed
703
704
705
 * 		before the word boundary.
 *
 *	Padding is done using 0x5a (POISON_INUSE)
Christoph Lameter's avatar
Christoph Lameter committed
706
707
 *
 * object + s->size
Christoph Lameter's avatar
Christoph Lameter committed
708
 * 	Nothing is used beyond s->size.
Christoph Lameter's avatar
Christoph Lameter committed
709
 *
710
 * If slabcaches are merged then the object_size and inuse boundaries are mostly
Christoph Lameter's avatar
Christoph Lameter committed
711
 * ignored. And therefore no slab options that rely on these boundaries
Christoph Lameter's avatar
Christoph Lameter committed
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
 * 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;

730
731
	return check_bytes_and_report(s, page, p, "Object padding",
				p + off, POISON_INUSE, s->size - off);
Christoph Lameter's avatar
Christoph Lameter committed
732
733
}

734
/* Check the pad bytes at the end of a slab page */
Christoph Lameter's avatar
Christoph Lameter committed
735
736
static int slab_pad_check(struct kmem_cache *s, struct page *page)
{
737
738
739
740
741
	u8 *start;
	u8 *fault;
	u8 *end;
	int length;
	int remainder;
Christoph Lameter's avatar
Christoph Lameter committed
742
743
744
745

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

746
	start = page_address(page);
747
	length = (PAGE_SIZE << compound_order(page)) - s->reserved;
748
749
	end = start + length;
	remainder = length % s->size;
Christoph Lameter's avatar
Christoph Lameter committed
750
751
752
	if (!remainder)
		return 1;

753
	fault = memchr_inv(end - remainder, POISON_INUSE, remainder);
754
755
756
757
758
759
	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);
760
	print_section("Padding ", end - remainder, remainder);
761

Eric Dumazet's avatar
Eric Dumazet committed
762
	restore_bytes(s, "slab padding", POISON_INUSE, end - remainder, end);
763
	return 0;
Christoph Lameter's avatar
Christoph Lameter committed
764
765
766
}

static int check_object(struct kmem_cache *s, struct page *page,
767
					void *object, u8 val)
Christoph Lameter's avatar
Christoph Lameter committed
768
769
{
	u8 *p = object;
770
	u8 *endobject = object + s->object_size;
Christoph Lameter's avatar
Christoph Lameter committed
771
772

	if (s->flags & SLAB_RED_ZONE) {
773
		if (!check_bytes_and_report(s, page, object, "Redzone",
774
			endobject, val, s->inuse - s->object_size))
Christoph Lameter's avatar
Christoph Lameter committed
775
776
			return 0;
	} else {
777
		if ((s->flags & SLAB_POISON) && s->object_size < s->inuse) {
Ingo Molnar's avatar
Ingo Molnar committed
778
			check_bytes_and_report(s, page, p, "Alignment padding",
779
				endobject, POISON_INUSE, s->inuse - s->object_size);
Ingo Molnar's avatar
Ingo Molnar committed
780
		}
Christoph Lameter's avatar
Christoph Lameter committed
781
782
783
	}

	if (s->flags & SLAB_POISON) {
784
		if (val != SLUB_RED_ACTIVE && (s->flags & __OBJECT_POISON) &&
785
			(!check_bytes_and_report(s, page, p, "Poison", p,
786
					POISON_FREE, s->object_size - 1) ||
787
			 !check_bytes_and_report(s, page, p, "Poison",
788
				p + s->object_size - 1, POISON_END, 1)))
Christoph Lameter's avatar
Christoph Lameter committed
789
790
791
792
793
794
795
			return 0;
		/*
		 * check_pad_bytes cleans up on its own.
		 */
		check_pad_bytes(s, page, p);
	}

796
	if (!s->offset && val == SLUB_RED_ACTIVE)
Christoph Lameter's avatar
Christoph Lameter committed
797
798
799
800
801
802
803
804
805
806
		/*
		 * 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
807
		 * No choice but to zap it and thus lose the remainder
Christoph Lameter's avatar
Christoph Lameter committed
808
		 * of the free objects in this slab. May cause
Christoph Lameter's avatar
Christoph Lameter committed
809
		 * another error because the object count is now wrong.
Christoph Lameter's avatar
Christoph Lameter committed
810
		 */
811
		set_freepointer(s, p, NULL);
Christoph Lameter's avatar
Christoph Lameter committed
812
813
814
815
816
817
818
		return 0;
	}
	return 1;
}

static int check_slab(struct kmem_cache *s, struct page *page)
{
819
820
	int maxobj;

Christoph Lameter's avatar
Christoph Lameter committed
821
822
823
	VM_BUG_ON(!irqs_disabled());

	if (!PageSlab(page)) {
824
		slab_err(s, page, "Not a valid slab page");
Christoph Lameter's avatar
Christoph Lameter committed
825
826
		return 0;
	}
827

828
	maxobj = order_objects(compound_order(page), s->size, s->reserved);
829
830
831
832
833
834
	if (page->objects > maxobj) {
		slab_err(s, page, "objects %u > max %u",
			s->name, page->objects, maxobj);
		return 0;
	}
	if (page->inuse > page->objects) {
835
		slab_err(s, page, "inuse %u > max %u",
836
			s->name, page->inuse, page->objects);
Christoph Lameter's avatar
Christoph Lameter committed
837
838
839
840
841
842
843
844
		return 0;
	}
	/* Slab_pad_check fixes things up after itself */
	slab_pad_check(s, page);
	return 1;
}

/*
Christoph Lameter's avatar
Christoph Lameter committed
845
846
 * 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
847
848
849
850
 */
static int on_freelist(struct kmem_cache *s, struct page *page, void *search)
{
	int nr = 0;
851
	void *fp;
Christoph Lameter's avatar
Christoph Lameter committed
852
	void *object = NULL;
853
	unsigned long max_objects;
Christoph Lameter's avatar
Christoph Lameter committed
854

855
	fp = page->freelist;
856
	while (fp && nr <= page->objects) {
Christoph Lameter's avatar
Christoph Lameter committed
857
858
859
860
861
862
		if (fp == search)
			return 1;
		if (!check_valid_pointer(s, page, fp)) {
			if (object) {
				object_err(s, page, object,
					"Freechain corrupt");
863
				set_freepointer(s, object, NULL);
Christoph Lameter's avatar
Christoph Lameter committed
864
865
				break;
			} else {
866
				slab_err(s, page, "Freepointer corrupt");
867
				page->freelist = NULL;
868
				page->inuse = page->objects;
869
				slab_fix(s, "Freelist cleared");
Christoph Lameter's avatar
Christoph Lameter committed
870
871
872
873
874
875
876
877
878
				return 0;
			}
			break;
		}
		object = fp;
		fp = get_freepointer(s, object);
		nr++;
	}

879
	max_objects = order_objects(compound_order(page), s->size, s->reserved);
880
881
	if (max_objects > MAX_OBJS_PER_PAGE)
		max_objects = MAX_OBJS_PER_PAGE;
882
883
884
885
886
887
888

	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.");
	}
889
	if (page->inuse != page->objects - nr) {
890
		slab_err(s, page, "Wrong object count. Counter is %d but "
891
892
			"counted were %d", page->inuse, page->objects - nr);
		page->inuse = page->objects - nr;
893
		slab_fix(s, "Object count adjusted.");
Christoph Lameter's avatar
Christoph Lameter committed
894
895
896
897
	}
	return search == NULL;
}

898
899
static void trace(struct kmem_cache *s, struct page *page, void *object,
								int alloc)
Christoph Lameter's avatar
Christoph Lameter committed
900
901
902
903
904
905
906
907
908
{
	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)
909
			print_section("Object ", (void *)object, s->object_size);
Christoph Lameter's avatar
Christoph Lameter committed
910
911
912
913
914

		dump_stack();
	}
}

915
916
917
918
919
920
/*
 * 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)
{
921
	flags &= gfp_allowed_mask;
922
923
924
	lockdep_trace_alloc(flags);
	might_sleep_if(flags & __GFP_WAIT);

925
	return should_failslab(s->object_size, flags, s->flags);
926
927
928
929
}

static inline void slab_post_alloc_hook(struct kmem_cache *s, gfp_t flags, void *object)
{
930
	flags &= gfp_allowed_mask;
931
	kmemcheck_slab_alloc(s, flags, object, slab_ksize(s));
932
	kmemleak_alloc_recursive(object, s->object_size, 1, s->flags, flags);
933
934
935
936
937
938
}

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

939
940
941
942
943
944
945
946
947
948
	/*
	 * 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);
949
950
		kmemcheck_slab_free(s, x, s->object_size);
		debug_check_no_locks_freed(x, s->object_size);
951
952
953
		local_irq_restore(flags);
	}
#endif
954
	if (!(s->flags & SLAB_DEBUG_OBJECTS))
955
		debug_check_no_obj_freed(x, s->object_size);
956
957
}

958
/*
Christoph Lameter's avatar
Christoph Lameter committed
959
 * Tracking of fully allocated slabs for debugging purposes.
960
961
 *
 * list_lock must be held.
962
 */
963
964
static void add_full(struct kmem_cache *s,
	struct kmem_cache_node *n, struct page *page)
965
{
966
967
968
	if (!(s->flags & SLAB_STORE_USER))
		return;

969
970
971
	list_add(&page->lru, &n->full);
}

972
973
974
/*
 * list_lock must be held.
 */
975
976
977
978
979
980
981
982
static void remove_full(struct kmem_cache *s, struct page *page)
{
	if (!(s->flags & SLAB_STORE_USER))
		return;

	list_del(&page->lru);
}

983
984
985
986
987
988
989
990
/* 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);
}

991
992
993
994
995
static inline unsigned long node_nr_slabs(struct kmem_cache_node *n)
{
	return atomic_long_read(&n->nr_slabs);
}

996
static inline void inc_slabs_node(struct kmem_cache *s, int node, int objects)
997
998
999
1000
1001
1002
1003
1004
1005
{
	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).
	 */
1006
	if (n) {
1007
		atomic_long_inc(&n->nr_slabs);
1008
1009
		atomic_long_add(objects, &n->total_objects);
	}
1010
}
1011
static inline void dec_slabs_node(struct kmem_cache *s, int node, int objects)
1012
1013
1014
1015
{
	struct kmem_cache_node *n = get_node(s, node);

	atomic_long_dec(&n->nr_slabs);
1016
	atomic_long_sub(objects, &n->total_objects);
1017
1018
1019
}

/* Object debug checks for alloc/free paths */
Christoph Lameter's avatar
Christoph Lameter committed
1020
1021
1022
1023
1024
1025
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;

1026
	init_object(s, object, SLUB_RED_INACTIVE);
Christoph Lameter's avatar
Christoph Lameter committed
1027
1028
1029
	init_tracking(s, object);
}

1030
static noinline int alloc_debug_processing(struct kmem_cache *s, struct page *page,
1031
					void *object, unsigned long addr)
Christoph Lameter's avatar
Christoph Lameter committed
1032
1033
1034
1035
1036
1037
{
	if (!check_slab(s, page))
		goto bad;

	if (!check_valid_pointer(s, page, object)) {
		object_err(s, page, object, "Freelist Pointer check fails");
1038
		goto bad;
Christoph Lameter's avatar
Christoph Lameter committed
1039
1040
	}

1041
	if (!check_object(s, page, object, SLUB_RED_INACTIVE))
Christoph Lameter's avatar
Christoph Lameter committed
1042
1043
		goto bad;

Christoph Lameter's avatar
Christoph Lameter committed
1044
1045
1046
1047
	/* 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);
1048
	init_object(s, object, SLUB_RED_ACTIVE);
Christoph Lameter's avatar
Christoph Lameter committed
1049
	return 1;
Christoph Lameter's avatar
Christoph Lameter committed
1050

Christoph Lameter's avatar
Christoph Lameter committed
1051
1052
1053
1054
1055
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
1056
		 * as used avoids touching the remaining objects.
Christoph Lameter's avatar
Christoph Lameter committed
1057
		 */
1058
		slab_fix(s, "Marking all objects used");
1059
		page->inuse = page->objects;
1060
		page->freelist = NULL;
Christoph Lameter's avatar
Christoph Lameter committed
1061
1062
1063
1064
	}
	return 0;
}

1065
1066
1067
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
1068
{
1069
	struct kmem_cache_node *n = get_node(s, page_to_nid(page));
1070

1071
	spin_lock_irqsave(&n->list_lock, *flags);
1072
1073
	slab_lock(page);

Christoph Lameter's avatar
Christoph Lameter committed
1074
1075
1076
1077
	if (!check_slab(s, page))
		goto fail;

	if (!check_valid_pointer(s, page, object)) {
1078
		slab_err(s, page, "Invalid object pointer 0x%p", object);
Christoph Lameter's avatar
Christoph Lameter committed
1079
1080
1081
1082
		goto fail;
	}

	if (on_freelist(s, page, object)) {
1083
		object_err(s, page, object, "Object already free");
Christoph Lameter's avatar
Christoph Lameter committed
1084
1085
1086
		goto fail;
	}

1087
	if (!check_object(s, page, object, SLUB_RED_ACTIVE))
1088
		goto out;
Christoph Lameter's avatar
Christoph Lameter committed
1089

1090
	if (unlikely(s != page->slab_cache)) {
Ingo Molnar's avatar
Ingo Molnar committed
1091
		if (!PageSlab(page)) {
1092
1093
			slab_err(s, page, "Attempt to free object(0x%p) "
				"outside of slab", object);
1094
		} else if (!page->slab_cache) {
Christoph Lameter's avatar
Christoph Lameter committed
1095
			printk(KERN_ERR
1096
				"SLUB <none>: no slab for object 0x%p.\n",
Christoph Lameter's avatar
Christoph Lameter committed
1097
						object);
1098
			dump_stack();
1099
		} else
1100
1101
			object_err(s, page, object,
					"page slab pointer corrupt.");
Christoph Lameter's avatar
Christoph Lameter committed
1102
1103
		goto fail;
	}
Christoph Lameter's avatar
Christoph Lameter committed
1104
1105
1106
1107

	if (s->flags & SLAB_STORE_USER)
		set_track(s, object, TRACK_FREE, addr);
	trace(s, page, object, 0);
1108
	init_object(s, object, SLUB_RED_INACTIVE);
1109
out:
1110
	slab_unlock(page);
1111
1112
1113
1114
1115
	/*
	 * Keep node_lock to preserve integrity
	 * until the object is actually freed
	 */
	return n;
Christoph Lameter's avatar
Christoph Lameter committed
1116

Christoph Lameter's avatar
Christoph Lameter committed
1117
fail:
1118
1119
	slab_unlock(page);
	spin_unlock_irqrestore(&n->list_lock, *flags);
1120
	slab_fix(s, "Object at 0x%p not freed", object);
1121
	return NULL;
Christoph Lameter's avatar
Christoph Lameter committed
1122
1123
}

1124
1125
static int __init setup_slub_debug(char *str)
{
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
	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;

1140
1141
1142
1143
1144
1145
1146
1147
1148
	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;
	}

1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
	slub_debug = 0;
	if (*str == '-')
		/*
		 * Switch off all debugging measures.
		 */
		goto out;

	/*
	 * Determine which debug features should be switched on
	 */
1159
	for (; *str && *str != ','; str++) {
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
		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;
1176
1177
1178
		case 'a':
			slub_debug |= SLAB_FAILSLAB;
			break;
1179
1180
		default:
			printk(KERN_ERR "slub_debug option '%c' "
1181
				"unknown. skipped\n", *str);
1182
		}
1183
1184
	}

1185