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>
21
#include <linux/notifier.h>
Christoph Lameter's avatar
Christoph Lameter committed
22
#include <linux/seq_file.h>
Vegard Nossum's avatar
Vegard Nossum committed
23
#include <linux/kmemcheck.h>
Christoph Lameter's avatar
Christoph Lameter committed
24
25
26
27
#include <linux/cpu.h>
#include <linux/cpuset.h>
#include <linux/mempolicy.h>
#include <linux/ctype.h>
28
#include <linux/debugobjects.h>
Christoph Lameter's avatar
Christoph Lameter committed
29
#include <linux/kallsyms.h>
30
#include <linux/memory.h>
Roman Zippel's avatar
Roman Zippel committed
31
#include <linux/math64.h>
Akinobu Mita's avatar
Akinobu Mita committed
32
#include <linux/fault-inject.h>
33
#include <linux/stacktrace.h>
34
#include <linux/prefetch.h>
35
#include <linux/memcontrol.h>
Christoph Lameter's avatar
Christoph Lameter committed
36

37
38
#include <trace/events/kmem.h>

39
40
#include "internal.h"

Christoph Lameter's avatar
Christoph Lameter committed
41
42
/*
 * Lock order:
43
 *   1. slab_mutex (Global Mutex)
44
45
 *   2. node->list_lock
 *   3. slab_lock(page) (Only on some arches and for debugging)
Christoph Lameter's avatar
Christoph Lameter committed
46
 *
47
 *   slab_mutex
48
 *
49
 *   The role of the slab_mutex is to protect the list of all the slabs
50
51
52
53
54
55
56
57
58
59
60
61
62
63
 *   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
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
 *
 *   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
84
85
 * 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
86
 * freed then the slab will show up again on the partial lists.
Christoph Lameter's avatar
Christoph Lameter committed
87
88
 * We track full slabs for debugging purposes though because otherwise we
 * cannot scan all objects.
Christoph Lameter's avatar
Christoph Lameter committed
89
90
91
92
93
94
95
 *
 * 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.
 *
96
97
98
99
100
101
102
103
104
105
106
107
 * 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
108
 * 			freelist that allows lockless access to
109
110
 * 			free objects in addition to the regular freelist
 * 			that requires the slab lock.
Christoph Lameter's avatar
Christoph Lameter committed
111
112
113
 *
 * PageError		Slab requires special handling due to debug
 * 			options set. This moves	slab handling out of
114
 * 			the fast path and disables lockless freelists.
Christoph Lameter's avatar
Christoph Lameter committed
115
116
 */

117
118
static inline int kmem_cache_debug(struct kmem_cache *s)
{
119
#ifdef CONFIG_SLUB_DEBUG
120
	return unlikely(s->flags & SLAB_DEBUG_FLAGS);
121
#else
122
	return 0;
123
#endif
124
}
125

126
127
128
129
130
131
132
133
134
static inline bool kmem_cache_has_cpu_partial(struct kmem_cache *s)
{
#ifdef CONFIG_SLUB_CPU_PARTIAL
	return !kmem_cache_debug(s);
#else
	return false;
#endif
}

Christoph Lameter's avatar
Christoph Lameter committed
135
136
137
138
139
140
141
142
143
144
145
/*
 * 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

146
147
148
/* Enable to log cmpxchg failures */
#undef SLUB_DEBUG_CMPXCHG

149
150
151
152
/*
 * Mininum number of partial slabs. These will be left on the partial
 * lists even if they are empty. kmem_cache_shrink may reclaim them.
 */
153
#define MIN_PARTIAL 5
Christoph Lameter's avatar
Christoph Lameter committed
154

155
156
157
/*
 * Maximum number of desirable partial slabs.
 * The existence of more partial slabs makes kmem_cache_shrink
158
 * sort the partial list by the number of objects in use.
159
160
161
 */
#define MAX_PARTIAL 10

Christoph Lameter's avatar
Christoph Lameter committed
162
163
#define DEBUG_DEFAULT_FLAGS (SLAB_DEBUG_FREE | SLAB_RED_ZONE | \
				SLAB_POISON | SLAB_STORE_USER)
Christoph Lameter's avatar
Christoph Lameter committed
164

165
/*
166
167
168
 * 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.
169
 */
170
#define DEBUG_METADATA_FLAGS (SLAB_RED_ZONE | SLAB_POISON | SLAB_STORE_USER)
171

Christoph Lameter's avatar
Christoph Lameter committed
172
173
174
175
/*
 * Set of flags that will prevent slab merging
 */
#define SLUB_NEVER_MERGE (SLAB_RED_ZONE | SLAB_POISON | SLAB_STORE_USER | \
176
177
		SLAB_TRACE | SLAB_DESTROY_BY_RCU | SLAB_NOLEAKTRACE | \
		SLAB_FAILSLAB)
Christoph Lameter's avatar
Christoph Lameter committed
178
179

#define SLUB_MERGE_SAME (SLAB_DEBUG_FREE | SLAB_RECLAIM_ACCOUNT | \
Vegard Nossum's avatar
Vegard Nossum committed
180
		SLAB_CACHE_DMA | SLAB_NOTRACK)
Christoph Lameter's avatar
Christoph Lameter committed
181

182
183
#define OO_SHIFT	16
#define OO_MASK		((1 << OO_SHIFT) - 1)
184
#define MAX_OBJS_PER_PAGE	32767 /* since page.objects is u15 */
185

Christoph Lameter's avatar
Christoph Lameter committed
186
/* Internal SLUB flags */
Christoph Lameter's avatar
Christoph Lameter committed
187
#define __OBJECT_POISON		0x80000000UL /* Poison object */
188
#define __CMPXCHG_DOUBLE	0x40000000UL /* Use cmpxchg_double */
Christoph Lameter's avatar
Christoph Lameter committed
189
190
191
192
193

#ifdef CONFIG_SMP
static struct notifier_block slab_notifier;
#endif

194
195
196
/*
 * Tracking user of a slab.
 */
197
#define TRACK_ADDRS_COUNT 16
198
struct track {
199
	unsigned long addr;	/* Called from address */
200
201
202
#ifdef CONFIG_STACKTRACE
	unsigned long addrs[TRACK_ADDRS_COUNT];	/* Called from address */
#endif
203
204
205
206
207
208
209
	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 };

210
#ifdef CONFIG_SYSFS
Christoph Lameter's avatar
Christoph Lameter committed
211
212
213
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 *);
214
static void memcg_propagate_slab_attrs(struct kmem_cache *s);
Christoph Lameter's avatar
Christoph Lameter committed
215
#else
216
217
218
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; }
219
static inline void sysfs_slab_remove(struct kmem_cache *s) { }
220

221
static inline void memcg_propagate_slab_attrs(struct kmem_cache *s) { }
Christoph Lameter's avatar
Christoph Lameter committed
222
223
#endif

224
static inline void stat(const struct kmem_cache *s, enum stat_item si)
225
226
{
#ifdef CONFIG_SLUB_STATS
227
	__this_cpu_inc(s->cpu_slab->stat[si]);
228
229
230
#endif
}

Christoph Lameter's avatar
Christoph Lameter committed
231
232
233
234
235
236
237
238
239
/********************************************************************
 * 			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
240
/* Verify that a pointer has an address that is valid within a slab page */
241
242
243
244
245
static inline int check_valid_pointer(struct kmem_cache *s,
				struct page *page, const void *object)
{
	void *base;

246
	if (!object)
247
248
		return 1;

249
	base = page_address(page);
250
	if (object < base || object >= base + page->objects * s->size ||
251
252
253
254
255
256
257
		(object - base) % s->size) {
		return 0;
	}

	return 1;
}

258
259
260
261
262
static inline void *get_freepointer(struct kmem_cache *s, void *object)
{
	return *(void **)(object + s->offset);
}

263
264
265
266
267
static void prefetch_freepointer(const struct kmem_cache *s, void *object)
{
	prefetch(object + s->offset);
}

268
269
270
271
272
273
274
275
276
277
278
279
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;
}

280
281
282
283
284
285
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 */
286
287
#define for_each_object(__p, __s, __addr, __objects) \
	for (__p = (__addr); __p < (__addr) + (__objects) * (__s)->size;\
288
289
290
291
292
293
294
295
			__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;
}

296
297
298
299
300
301
302
303
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))
304
		return s->object_size;
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319

#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;
}

320
321
322
323
324
static inline int order_objects(int order, unsigned long size, int reserved)
{
	return ((PAGE_SIZE << order) - reserved) / size;
}

325
static inline struct kmem_cache_order_objects oo_make(int order,
326
		unsigned long size, int reserved)
327
328
{
	struct kmem_cache_order_objects x = {
329
		(order << OO_SHIFT) + order_objects(order, size, reserved)
330
331
332
333
334
335
336
	};

	return x;
}

static inline int oo_order(struct kmem_cache_order_objects x)
{
337
	return x.x >> OO_SHIFT;
338
339
340
341
}

static inline int oo_objects(struct kmem_cache_order_objects x)
{
342
	return x.x & OO_MASK;
343
344
}

345
346
347
348
349
350
351
352
353
354
355
356
357
/*
 * 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);
}

358
359
360
361
362
363
364
/* 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());
365
366
#if defined(CONFIG_HAVE_CMPXCHG_DOUBLE) && \
    defined(CONFIG_HAVE_ALIGNED_STRUCT_PAGE)
367
	if (s->flags & __CMPXCHG_DOUBLE) {
368
		if (cmpxchg_double(&page->freelist, &page->counters,
369
370
371
372
373
374
375
			freelist_old, counters_old,
			freelist_new, counters_new))
		return 1;
	} else
#endif
	{
		slab_lock(page);
376
377
		if (page->freelist == freelist_old &&
					page->counters == counters_old) {
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
			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;
}

396
397
398
399
400
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)
{
401
402
#if defined(CONFIG_HAVE_CMPXCHG_DOUBLE) && \
    defined(CONFIG_HAVE_ALIGNED_STRUCT_PAGE)
403
	if (s->flags & __CMPXCHG_DOUBLE) {
404
		if (cmpxchg_double(&page->freelist, &page->counters,
405
406
407
408
409
410
			freelist_old, counters_old,
			freelist_new, counters_new))
		return 1;
	} else
#endif
	{
411
412
413
		unsigned long flags;

		local_irq_save(flags);
414
		slab_lock(page);
415
416
		if (page->freelist == freelist_old &&
					page->counters == counters_old) {
417
418
			page->freelist = freelist_new;
			page->counters = counters_new;
419
			slab_unlock(page);
420
			local_irq_restore(flags);
421
422
			return 1;
		}
423
		slab_unlock(page);
424
		local_irq_restore(flags);
425
426
427
428
429
430
431
432
433
434
435
436
	}

	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;
}

437
#ifdef CONFIG_SLUB_DEBUG
438
439
440
/*
 * Determine a map of object in use on a page.
 *
441
 * Node listlock must be held to guarantee that the page does
442
443
444
445
446
447
448
449
450
451
452
 * 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);
}

453
454
455
/*
 * Debug settings:
 */
456
457
458
#ifdef CONFIG_SLUB_DEBUG_ON
static int slub_debug = DEBUG_DEFAULT_FLAGS;
#else
459
static int slub_debug;
460
#endif
461
462

static char *slub_debug_slabs;
463
static int disable_higher_order_debug;
464

Christoph Lameter's avatar
Christoph Lameter committed
465
466
467
468
469
/*
 * Object debugging
 */
static void print_section(char *text, u8 *addr, unsigned int length)
{
470
471
	print_hex_dump(KERN_ERR, text, DUMP_PREFIX_ADDRESS, 16, 1, addr,
			length, 1);
Christoph Lameter's avatar
Christoph Lameter committed
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
}

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,
488
			enum track_item alloc, unsigned long addr)
Christoph Lameter's avatar
Christoph Lameter committed
489
{
Akinobu Mita's avatar
Akinobu Mita committed
490
	struct track *p = get_track(s, object, alloc);
Christoph Lameter's avatar
Christoph Lameter committed
491
492

	if (addr) {
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
#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
511
512
		p->addr = addr;
		p->cpu = smp_processor_id();
513
		p->pid = current->pid;
Christoph Lameter's avatar
Christoph Lameter committed
514
515
516
517
518
519
520
		p->when = jiffies;
	} else
		memset(p, 0, sizeof(struct track));
}

static void init_tracking(struct kmem_cache *s, void *object)
{
521
522
523
	if (!(s->flags & SLAB_STORE_USER))
		return;

524
525
	set_track(s, object, TRACK_FREE, 0UL);
	set_track(s, object, TRACK_ALLOC, 0UL);
Christoph Lameter's avatar
Christoph Lameter committed
526
527
528
529
530
531
532
}

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

533
	printk(KERN_ERR "INFO: %s in %pS age=%lu cpu=%u pid=%d\n",
534
		s, (void *)t->addr, jiffies - t->when, t->cpu, t->pid);
535
536
537
538
539
540
541
542
543
544
#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
545
546
547
548
549
550
551
552
553
554
555
556
557
}

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)
{
558
559
560
	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);
561
562
563
564
565
566
567
568
569
570
571
572
573

}

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");
574
	printk(KERN_ERR "BUG %s (%s): %s\n", s->name, print_tainted(), buf);
575
576
	printk(KERN_ERR "----------------------------------------"
			"-------------------------------------\n\n");
577

578
	add_taint(TAINT_BAD_PAGE, LOCKDEP_NOW_UNRELIABLE);
Christoph Lameter's avatar
Christoph Lameter committed
579
580
}

581
582
583
584
585
586
587
588
589
590
591
592
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
593
594
{
	unsigned int off;	/* Offset of last byte */
595
	u8 *addr = page_address(page);
596
597
598
599
600
601
602
603
604

	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)
605
		print_section("Bytes b4 ", p - 16, 16);
Christoph Lameter's avatar
Christoph Lameter committed
606

607
	print_section("Object ", p, min_t(unsigned long, s->object_size,
608
				PAGE_SIZE));
Christoph Lameter's avatar
Christoph Lameter committed
609
	if (s->flags & SLAB_RED_ZONE)
610
611
		print_section("Redzone ", p + s->object_size,
			s->inuse - s->object_size);
Christoph Lameter's avatar
Christoph Lameter committed
612
613
614
615
616
617

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

618
	if (s->flags & SLAB_STORE_USER)
Christoph Lameter's avatar
Christoph Lameter committed
619
620
621
622
		off += 2 * sizeof(struct track);

	if (off != s->size)
		/* Beginning of the filler is the free pointer */
623
		print_section("Padding ", p + off, s->size - off);
624
625

	dump_stack();
Christoph Lameter's avatar
Christoph Lameter committed
626
627
628
629
630
}

static void object_err(struct kmem_cache *s, struct page *page,
			u8 *object, char *reason)
{
631
	slab_bug(s, "%s", reason);
632
	print_trailer(s, page, object);
Christoph Lameter's avatar
Christoph Lameter committed
633
634
}

635
636
static void slab_err(struct kmem_cache *s, struct page *page,
			const char *fmt, ...)
Christoph Lameter's avatar
Christoph Lameter committed
637
638
639
640
{
	va_list args;
	char buf[100];

641
642
	va_start(args, fmt);
	vsnprintf(buf, sizeof(buf), fmt, args);
Christoph Lameter's avatar
Christoph Lameter committed
643
	va_end(args);
644
	slab_bug(s, "%s", buf);
645
	print_page_info(page);
Christoph Lameter's avatar
Christoph Lameter committed
646
647
648
	dump_stack();
}

649
static void init_object(struct kmem_cache *s, void *object, u8 val)
Christoph Lameter's avatar
Christoph Lameter committed
650
651
652
653
{
	u8 *p = object;

	if (s->flags & __OBJECT_POISON) {
654
655
		memset(p, POISON_FREE, s->object_size - 1);
		p[s->object_size - 1] = POISON_END;
Christoph Lameter's avatar
Christoph Lameter committed
656
657
658
	}

	if (s->flags & SLAB_RED_ZONE)
659
		memset(p + s->object_size, val, s->inuse - s->object_size);
Christoph Lameter's avatar
Christoph Lameter committed
660
661
}

662
663
664
665
666
667
668
669
670
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,
671
			u8 *start, unsigned int value, unsigned int bytes)
672
673
674
675
{
	u8 *fault;
	u8 *end;

676
	fault = memchr_inv(start, value, bytes);
677
678
679
680
681
682
683
684
685
686
687
688
689
690
	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
691
692
693
694
695
696
697
698
699
}

/*
 * 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
700
 *
Christoph Lameter's avatar
Christoph Lameter committed
701
702
703
 * 	Poisoning uses 0x6b (POISON_FREE) and the last byte is
 * 	0xa5 (POISON_END)
 *
704
 * object + s->object_size
Christoph Lameter's avatar
Christoph Lameter committed
705
 * 	Padding to reach word boundary. This is also used for Redzoning.
Christoph Lameter's avatar
Christoph Lameter committed
706
 * 	Padding is extended by another word if Redzoning is enabled and
707
 * 	object_size == inuse.
Christoph Lameter's avatar
Christoph Lameter committed
708
 *
Christoph Lameter's avatar
Christoph Lameter committed
709
710
711
712
 * 	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
713
714
 * 	Meta data starts here.
 *
Christoph Lameter's avatar
Christoph Lameter committed
715
716
 * 	A. Free pointer (if we cannot overwrite object on free)
 * 	B. Tracking data for SLAB_STORE_USER
Christoph Lameter's avatar
Christoph Lameter committed
717
 * 	C. Padding to reach required alignment boundary or at mininum
Christoph Lameter's avatar
Christoph Lameter committed
718
 * 		one word if debugging is on to be able to detect writes
Christoph Lameter's avatar
Christoph Lameter committed
719
720
721
 * 		before the word boundary.
 *
 *	Padding is done using 0x5a (POISON_INUSE)
Christoph Lameter's avatar
Christoph Lameter committed
722
723
 *
 * object + s->size
Christoph Lameter's avatar
Christoph Lameter committed
724
 * 	Nothing is used beyond s->size.
Christoph Lameter's avatar
Christoph Lameter committed
725
 *
726
 * If slabcaches are merged then the object_size and inuse boundaries are mostly
Christoph Lameter's avatar
Christoph Lameter committed
727
 * ignored. And therefore no slab options that rely on these boundaries
Christoph Lameter's avatar
Christoph Lameter committed
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
 * 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;

746
747
	return check_bytes_and_report(s, page, p, "Object padding",
				p + off, POISON_INUSE, s->size - off);
Christoph Lameter's avatar
Christoph Lameter committed
748
749
}

750
/* Check the pad bytes at the end of a slab page */
Christoph Lameter's avatar
Christoph Lameter committed
751
752
static int slab_pad_check(struct kmem_cache *s, struct page *page)
{
753
754
755
756
757
	u8 *start;
	u8 *fault;
	u8 *end;
	int length;
	int remainder;
Christoph Lameter's avatar
Christoph Lameter committed
758
759
760
761

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

762
	start = page_address(page);
763
	length = (PAGE_SIZE << compound_order(page)) - s->reserved;
764
765
	end = start + length;
	remainder = length % s->size;
Christoph Lameter's avatar
Christoph Lameter committed
766
767
768
	if (!remainder)
		return 1;

769
	fault = memchr_inv(end - remainder, POISON_INUSE, remainder);
770
771
772
773
774
775
	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);
776
	print_section("Padding ", end - remainder, remainder);
777

Eric Dumazet's avatar
Eric Dumazet committed
778
	restore_bytes(s, "slab padding", POISON_INUSE, end - remainder, end);
779
	return 0;
Christoph Lameter's avatar
Christoph Lameter committed
780
781
782
}

static int check_object(struct kmem_cache *s, struct page *page,
783
					void *object, u8 val)
Christoph Lameter's avatar
Christoph Lameter committed
784
785
{
	u8 *p = object;
786
	u8 *endobject = object + s->object_size;
Christoph Lameter's avatar
Christoph Lameter committed
787
788

	if (s->flags & SLAB_RED_ZONE) {
789
		if (!check_bytes_and_report(s, page, object, "Redzone",
790
			endobject, val, s->inuse - s->object_size))
Christoph Lameter's avatar
Christoph Lameter committed
791
792
			return 0;
	} else {
793
		if ((s->flags & SLAB_POISON) && s->object_size < s->inuse) {
Ingo Molnar's avatar
Ingo Molnar committed
794
			check_bytes_and_report(s, page, p, "Alignment padding",
795
796
				endobject, POISON_INUSE,
				s->inuse - s->object_size);
Ingo Molnar's avatar
Ingo Molnar committed
797
		}
Christoph Lameter's avatar
Christoph Lameter committed
798
799
800
	}

	if (s->flags & SLAB_POISON) {
801
		if (val != SLUB_RED_ACTIVE && (s->flags & __OBJECT_POISON) &&
802
			(!check_bytes_and_report(s, page, p, "Poison", p,
803
					POISON_FREE, s->object_size - 1) ||
804
			 !check_bytes_and_report(s, page, p, "Poison",
805
				p + s->object_size - 1, POISON_END, 1)))
Christoph Lameter's avatar
Christoph Lameter committed
806
807
808
809
810
811
812
			return 0;
		/*
		 * check_pad_bytes cleans up on its own.
		 */
		check_pad_bytes(s, page, p);
	}

813
	if (!s->offset && val == SLUB_RED_ACTIVE)
Christoph Lameter's avatar
Christoph Lameter committed
814
815
816
817
818
819
820
821
822
823
		/*
		 * 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
824
		 * No choice but to zap it and thus lose the remainder
Christoph Lameter's avatar
Christoph Lameter committed
825
		 * of the free objects in this slab. May cause
Christoph Lameter's avatar
Christoph Lameter committed
826
		 * another error because the object count is now wrong.
Christoph Lameter's avatar
Christoph Lameter committed
827
		 */
828
		set_freepointer(s, p, NULL);
Christoph Lameter's avatar
Christoph Lameter committed
829
830
831
832
833
834
835
		return 0;
	}
	return 1;
}

static int check_slab(struct kmem_cache *s, struct page *page)
{
836
837
	int maxobj;

Christoph Lameter's avatar
Christoph Lameter committed
838
839
840
	VM_BUG_ON(!irqs_disabled());

	if (!PageSlab(page)) {
841
		slab_err(s, page, "Not a valid slab page");
Christoph Lameter's avatar
Christoph Lameter committed
842
843
		return 0;
	}
844

845
	maxobj = order_objects(compound_order(page), s->size, s->reserved);
846
847
848
849
850
851
	if (page->objects > maxobj) {
		slab_err(s, page, "objects %u > max %u",
			s->name, page->objects, maxobj);
		return 0;
	}
	if (page->inuse > page->objects) {
852
		slab_err(s, page, "inuse %u > max %u",
853
			s->name, page->inuse, page->objects);
Christoph Lameter's avatar
Christoph Lameter committed
854
855
856
857
858
859
860
861
		return 0;
	}
	/* Slab_pad_check fixes things up after itself */
	slab_pad_check(s, page);
	return 1;
}

/*
Christoph Lameter's avatar
Christoph Lameter committed
862
863
 * 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
864
865
866
867
 */
static int on_freelist(struct kmem_cache *s, struct page *page, void *search)
{
	int nr = 0;
868
	void *fp;
Christoph Lameter's avatar
Christoph Lameter committed
869
	void *object = NULL;
870
	unsigned long max_objects;
Christoph Lameter's avatar
Christoph Lameter committed
871

872
	fp = page->freelist;
873
	while (fp && nr <= page->objects) {
Christoph Lameter's avatar
Christoph Lameter committed
874
875
876
877
878
879
		if (fp == search)
			return 1;
		if (!check_valid_pointer(s, page, fp)) {
			if (object) {
				object_err(s, page, object,
					"Freechain corrupt");
880
				set_freepointer(s, object, NULL);
Christoph Lameter's avatar
Christoph Lameter committed
881
			} else {
882
				slab_err(s, page, "Freepointer corrupt");
883
				page->freelist = NULL;
884
				page->inuse = page->objects;
885
				slab_fix(s, "Freelist cleared");
Christoph Lameter's avatar
Christoph Lameter committed
886
887
888
889
890
891
892
893
894
				return 0;
			}
			break;
		}
		object = fp;
		fp = get_freepointer(s, object);
		nr++;
	}

895
	max_objects = order_objects(compound_order(page), s->size, s->reserved);
896
897
	if (max_objects > MAX_OBJS_PER_PAGE)
		max_objects = MAX_OBJS_PER_PAGE;
898
899
900
901
902
903
904

	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.");
	}
905
	if (page->inuse != page->objects - nr) {
906
		slab_err(s, page, "Wrong object count. Counter is %d but "
907
908
			"counted were %d", page->inuse, page->objects - nr);
		page->inuse = page->objects - nr;
909
		slab_fix(s, "Object count adjusted.");
Christoph Lameter's avatar
Christoph Lameter committed
910
911
912
913
	}
	return search == NULL;
}

914
915
static void trace(struct kmem_cache *s, struct page *page, void *object,
								int alloc)
Christoph Lameter's avatar
Christoph Lameter committed
916
917
918
919
920
921
922
923
924
{
	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)
925
926
			print_section("Object ", (void *)object,
					s->object_size);
Christoph Lameter's avatar
Christoph Lameter committed
927
928
929
930
931

		dump_stack();
	}
}

932
933
934
935
/*
 * Hooks for other subsystems that check memory allocations. In a typical
 * production configuration these hooks all should produce no code at all.
 */
936
937
938
939
940
941
942
943
944
945
static inline void kmalloc_large_node_hook(void *ptr, size_t size, gfp_t flags)
{
	kmemleak_alloc(ptr, size, 1, flags);
}

static inline void kfree_hook(const void *x)
{
	kmemleak_free(x);
}

946
947
static inline int slab_pre_alloc_hook(struct kmem_cache *s, gfp_t flags)
{
948
	flags &= gfp_allowed_mask;
949
950
951
	lockdep_trace_alloc(flags);
	might_sleep_if(flags & __GFP_WAIT);

952
	return should_failslab(s->object_size, flags, s->flags);
953
954
}

955
956
static inline void slab_post_alloc_hook(struct kmem_cache *s,
					gfp_t flags, void *object)
957
{
958
	flags &= gfp_allowed_mask;
959
	kmemcheck_slab_alloc(s, flags, object, slab_ksize(s));
960
	kmemleak_alloc_recursive(object, s->object_size, 1, s->flags, flags);
961
962
963
964
965
966
}

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

967
	/*
968
	 * Trouble is that we may no longer disable interrupts in the fast path
969
970
971
972
973
974
975
976
	 * 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);
977
978
		kmemcheck_slab_free(s, x, s->object_size);
		debug_check_no_locks_freed(x, s->object_size);
979
980
981
		local_irq_restore(flags);
	}
#endif
982
	if (!(s->flags & SLAB_DEBUG_OBJECTS))
983
		debug_check_no_obj_freed(x, s->object_size);
984
985
}

986
/*
Christoph Lameter's avatar
Christoph Lameter committed
987
 * Tracking of fully allocated slabs for debugging purposes.
988
989
 *
 * list_lock must be held.
990
 */
991
992
static void add_full(struct kmem_cache *s,
	struct kmem_cache_node *n, struct page *page)
993
{
994
995
996
	if (!(s->flags & SLAB_STORE_USER))
		return;

997
998
999
	list_add(&page->lru, &n->full);
}

1000
/*