slub.c 112 KB
Newer Older
Christoph Lameter's avatar
Christoph Lameter committed
1
2
3
4
5
6
7
/*
 * SLUB: A slab allocator that limits cache line use instead of queuing
 * objects in per cpu and per node lists.
 *
 * The allocator synchronizes using per slab locks and only
 * uses a centralized lock to manage a pool of partial slabs.
 *
Christoph Lameter's avatar
Christoph Lameter committed
8
 * (C) 2007 SGI, Christoph Lameter
Christoph Lameter's avatar
Christoph Lameter committed
9
10
11
 */

#include <linux/mm.h>
Nick Piggin's avatar
Nick Piggin committed
12
#include <linux/swap.h> /* struct reclaim_state */
Christoph Lameter's avatar
Christoph Lameter committed
13
14
15
16
17
#include <linux/module.h>
#include <linux/bit_spinlock.h>
#include <linux/interrupt.h>
#include <linux/bitops.h>
#include <linux/slab.h>
18
#include <linux/proc_fs.h>
Christoph Lameter's avatar
Christoph Lameter committed
19
#include <linux/seq_file.h>
20
#include <linux/kmemtrace.h>
Vegard Nossum's avatar
Vegard Nossum committed
21
#include <linux/kmemcheck.h>
Christoph Lameter's avatar
Christoph Lameter committed
22
23
#include <linux/cpu.h>
#include <linux/cpuset.h>
24
#include <linux/kmemleak.h>
Christoph Lameter's avatar
Christoph Lameter committed
25
26
#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>
Christoph Lameter's avatar
Christoph Lameter committed
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77

/*
 * Lock order:
 *   1. slab_lock(page)
 *   2. slab->list_lock
 *
 *   The slab_lock protects operations on the object of a particular
 *   slab and its metadata in the page struct. If the slab lock
 *   has been taken then no allocations nor frees can be performed
 *   on the objects in the slab nor can the slab be added or removed
 *   from the partial or full lists since this would mean modifying
 *   the page_struct of the slab.
 *
 *   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.
 *
 *   The lock order is sometimes inverted when we are trying to get a slab
 *   off a list. We take the list_lock and then look for a page on the list
 *   to use. While we do that objects in the slabs may be freed. We can
 *   only operate on the slab if we have also taken the slab_lock. So we use
 *   a slab_trylock() on the slab. If trylock was successful then no frees
 *   can occur anymore and we can use the slab for allocations etc. If the
 *   slab_trylock() does not succeed then frees are in progress in the slab and
 *   we must stay away from it for a while since we may cause a bouncing
 *   cacheline if we try to acquire the lock. So go onto the next slab.
 *   If all pages are busy then we may allocate a new slab instead of reusing
 *   a partial slab. A new slab has noone operating on it and thus there is
 *   no danger of cacheline contention.
 *
 *   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
78
79
 * 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
80
 * freed then the slab will show up again on the partial lists.
Christoph Lameter's avatar
Christoph Lameter committed
81
82
 * We track full slabs for debugging purposes though because otherwise we
 * cannot scan all objects.
Christoph Lameter's avatar
Christoph Lameter committed
83
84
85
86
87
88
89
 *
 * 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.
 *
90
91
92
93
94
95
96
97
98
99
100
101
 * 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
102
 * 			freelist that allows lockless access to
103
104
 * 			free objects in addition to the regular freelist
 * 			that requires the slab lock.
Christoph Lameter's avatar
Christoph Lameter committed
105
106
107
 *
 * PageError		Slab requires special handling due to debug
 * 			options set. This moves	slab handling out of
108
 * 			the fast path and disables lockless freelists.
Christoph Lameter's avatar
Christoph Lameter committed
109
110
 */

111
#ifdef CONFIG_SLUB_DEBUG
112
#define SLABDEBUG 1
113
114
115
116
#else
#define SLABDEBUG 0
#endif

Christoph Lameter's avatar
Christoph Lameter committed
117
118
119
120
121
122
123
124
125
126
127
/*
 * 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

128
129
130
131
/*
 * Mininum number of partial slabs. These will be left on the partial
 * lists even if they are empty. kmem_cache_shrink may reclaim them.
 */
132
#define MIN_PARTIAL 5
Christoph Lameter's avatar
Christoph Lameter committed
133

134
135
136
137
138
139
140
/*
 * 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
141
142
#define DEBUG_DEFAULT_FLAGS (SLAB_DEBUG_FREE | SLAB_RED_ZONE | \
				SLAB_POISON | SLAB_STORE_USER)
Christoph Lameter's avatar
Christoph Lameter committed
143

144
/*
145
146
147
 * 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.
148
 */
149
#define DEBUG_METADATA_FLAGS (SLAB_RED_ZONE | SLAB_POISON | SLAB_STORE_USER)
150

Christoph Lameter's avatar
Christoph Lameter committed
151
152
153
154
/*
 * Set of flags that will prevent slab merging
 */
#define SLUB_NEVER_MERGE (SLAB_RED_ZONE | SLAB_POISON | SLAB_STORE_USER | \
155
		SLAB_TRACE | SLAB_DESTROY_BY_RCU | SLAB_NOLEAKTRACE)
Christoph Lameter's avatar
Christoph Lameter committed
156
157

#define SLUB_MERGE_SAME (SLAB_DEBUG_FREE | SLAB_RECLAIM_ACCOUNT | \
Vegard Nossum's avatar
Vegard Nossum committed
158
		SLAB_CACHE_DMA | SLAB_NOTRACK)
Christoph Lameter's avatar
Christoph Lameter committed
159
160

#ifndef ARCH_KMALLOC_MINALIGN
161
#define ARCH_KMALLOC_MINALIGN __alignof__(unsigned long long)
Christoph Lameter's avatar
Christoph Lameter committed
162
163
164
#endif

#ifndef ARCH_SLAB_MINALIGN
165
#define ARCH_SLAB_MINALIGN __alignof__(unsigned long long)
Christoph Lameter's avatar
Christoph Lameter committed
166
167
#endif

168
169
170
171
#define OO_SHIFT	16
#define OO_MASK		((1 << OO_SHIFT) - 1)
#define MAX_OBJS_PER_PAGE	65535 /* since page.objects is u16 */

Christoph Lameter's avatar
Christoph Lameter committed
172
/* Internal SLUB flags */
173
174
#define __OBJECT_POISON		0x80000000 /* Poison object */
#define __SYSFS_ADD_DEFERRED	0x40000000 /* Not yet visible via sysfs */
Christoph Lameter's avatar
Christoph Lameter committed
175
176
177
178
179
180
181
182
183
184

static int kmem_size = sizeof(struct kmem_cache);

#ifdef CONFIG_SMP
static struct notifier_block slab_notifier;
#endif

static enum {
	DOWN,		/* No slab functionality available */
	PARTIAL,	/* kmem_cache_open() works but kmalloc does not */
Christoph Lameter's avatar
Christoph Lameter committed
185
	UP,		/* Everything works but does not show up in sysfs */
Christoph Lameter's avatar
Christoph Lameter committed
186
187
188
189
190
	SYSFS		/* Sysfs up */
} slab_state = DOWN;

/* A list of all slab caches on the system */
static DECLARE_RWSEM(slub_lock);
Adrian Bunk's avatar
Adrian Bunk committed
191
static LIST_HEAD(slab_caches);
Christoph Lameter's avatar
Christoph Lameter committed
192

193
194
195
196
/*
 * Tracking user of a slab.
 */
struct track {
197
	unsigned long addr;	/* Called from address */
198
199
200
201
202
203
204
	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 };

205
#ifdef CONFIG_SLUB_DEBUG
Christoph Lameter's avatar
Christoph Lameter committed
206
207
208
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 *);
209

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

Christoph Lameter's avatar
Christoph Lameter committed
219
220
#endif

221
222
223
224
225
226
227
static inline void stat(struct kmem_cache_cpu *c, enum stat_item si)
{
#ifdef CONFIG_SLUB_STATS
	c->stat[si]++;
#endif
}

Christoph Lameter's avatar
Christoph Lameter committed
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
/********************************************************************
 * 			Core slab cache functions
 *******************************************************************/

int slab_is_available(void)
{
	return slab_state >= UP;
}

static inline struct kmem_cache_node *get_node(struct kmem_cache *s, int node)
{
#ifdef CONFIG_NUMA
	return s->node[node];
#else
	return &s->local_node;
#endif
}

246
247
static inline struct kmem_cache_cpu *get_cpu_slab(struct kmem_cache *s, int cpu)
{
248
249
250
251
252
#ifdef CONFIG_SMP
	return s->cpu_slab[cpu];
#else
	return &s->cpu_slab;
#endif
253
254
}

Christoph Lameter's avatar
Christoph Lameter committed
255
/* Verify that a pointer has an address that is valid within a slab page */
256
257
258
259
260
static inline int check_valid_pointer(struct kmem_cache *s,
				struct page *page, const void *object)
{
	void *base;

261
	if (!object)
262
263
		return 1;

264
	base = page_address(page);
265
	if (object < base || object >= base + page->objects * s->size ||
266
267
268
269
270
271
272
		(object - base) % s->size) {
		return 0;
	}

	return 1;
}

273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
/*
 * Slow version of get and set free pointer.
 *
 * This version requires touching the cache lines of kmem_cache which
 * we avoid to do in the fast alloc free paths. There we obtain the offset
 * from the page struct.
 */
static inline void *get_freepointer(struct kmem_cache *s, void *object)
{
	return *(void **)(object + s->offset);
}

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 */
291
292
#define for_each_object(__p, __s, __addr, __objects) \
	for (__p = (__addr); __p < (__addr) + (__objects) * (__s)->size;\
293
294
295
296
			__p += (__s)->size)

/* Scan freelist */
#define for_each_free_object(__p, __s, __free) \
297
	for (__p = (__free); __p; __p = get_freepointer((__s), __p))
298
299
300
301
302
303
304

/* 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;
}

305
306
307
308
static inline struct kmem_cache_order_objects oo_make(int order,
						unsigned long size)
{
	struct kmem_cache_order_objects x = {
309
		(order << OO_SHIFT) + (PAGE_SIZE << order) / size
310
311
312
313
314
315
316
	};

	return x;
}

static inline int oo_order(struct kmem_cache_order_objects x)
{
317
	return x.x >> OO_SHIFT;
318
319
320
321
}

static inline int oo_objects(struct kmem_cache_order_objects x)
{
322
	return x.x & OO_MASK;
323
324
}

325
326
327
328
#ifdef CONFIG_SLUB_DEBUG
/*
 * Debug settings:
 */
329
330
331
#ifdef CONFIG_SLUB_DEBUG_ON
static int slub_debug = DEBUG_DEFAULT_FLAGS;
#else
332
static int slub_debug;
333
#endif
334
335

static char *slub_debug_slabs;
336
static int disable_higher_order_debug;
337

Christoph Lameter's avatar
Christoph Lameter committed
338
339
340
341
342
343
344
345
346
347
348
349
350
/*
 * Object debugging
 */
static void print_section(char *text, u8 *addr, unsigned int length)
{
	int i, offset;
	int newline = 1;
	char ascii[17];

	ascii[16] = 0;

	for (i = 0; i < length; i++) {
		if (newline) {
351
			printk(KERN_ERR "%8s 0x%p: ", text, addr + i);
Christoph Lameter's avatar
Christoph Lameter committed
352
353
			newline = 0;
		}
354
		printk(KERN_CONT " %02x", addr[i]);
Christoph Lameter's avatar
Christoph Lameter committed
355
356
357
		offset = i % 16;
		ascii[offset] = isgraph(addr[i]) ? addr[i] : '.';
		if (offset == 15) {
358
			printk(KERN_CONT " %s\n", ascii);
Christoph Lameter's avatar
Christoph Lameter committed
359
360
361
362
363
364
			newline = 1;
		}
	}
	if (!newline) {
		i %= 16;
		while (i < 16) {
365
			printk(KERN_CONT "   ");
Christoph Lameter's avatar
Christoph Lameter committed
366
367
368
			ascii[i] = ' ';
			i++;
		}
369
		printk(KERN_CONT " %s\n", ascii);
Christoph Lameter's avatar
Christoph Lameter committed
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
	}
}

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,
387
			enum track_item alloc, unsigned long addr)
Christoph Lameter's avatar
Christoph Lameter committed
388
{
Akinobu Mita's avatar
Akinobu Mita committed
389
	struct track *p = get_track(s, object, alloc);
Christoph Lameter's avatar
Christoph Lameter committed
390
391
392
393

	if (addr) {
		p->addr = addr;
		p->cpu = smp_processor_id();
394
		p->pid = current->pid;
Christoph Lameter's avatar
Christoph Lameter committed
395
396
397
398
399
400
401
		p->when = jiffies;
	} else
		memset(p, 0, sizeof(struct track));
}

static void init_tracking(struct kmem_cache *s, void *object)
{
402
403
404
	if (!(s->flags & SLAB_STORE_USER))
		return;

405
406
	set_track(s, object, TRACK_FREE, 0UL);
	set_track(s, object, TRACK_ALLOC, 0UL);
Christoph Lameter's avatar
Christoph Lameter committed
407
408
409
410
411
412
413
}

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

414
	printk(KERN_ERR "INFO: %s in %pS age=%lu cpu=%u pid=%d\n",
415
		s, (void *)t->addr, jiffies - t->when, t->cpu, t->pid);
416
417
418
419
420
421
422
423
424
425
426
427
428
}

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)
{
429
430
	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);
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446

}

static void slab_bug(struct kmem_cache *s, char *fmt, ...)
{
	va_list args;
	char buf[100];

	va_start(args, fmt);
	vsnprintf(buf, sizeof(buf), fmt, args);
	va_end(args);
	printk(KERN_ERR "========================================"
			"=====================================\n");
	printk(KERN_ERR "BUG %s: %s\n", s->name, buf);
	printk(KERN_ERR "----------------------------------------"
			"-------------------------------------\n\n");
Christoph Lameter's avatar
Christoph Lameter committed
447
448
}

449
450
451
452
453
454
455
456
457
458
459
460
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
461
462
{
	unsigned int off;	/* Offset of last byte */
463
	u8 *addr = page_address(page);
464
465
466
467
468
469
470
471
472
473
474

	print_tracking(s, p);

	print_page_info(page);

	printk(KERN_ERR "INFO: Object 0x%p @offset=%tu fp=0x%p\n\n",
			p, p - addr, get_freepointer(s, p));

	if (p > addr + 16)
		print_section("Bytes b4", p - 16, 16);

475
	print_section("Object", p, min_t(unsigned long, s->objsize, PAGE_SIZE));
Christoph Lameter's avatar
Christoph Lameter committed
476
477
478
479
480
481
482
483
484
485

	if (s->flags & SLAB_RED_ZONE)
		print_section("Redzone", p + s->objsize,
			s->inuse - s->objsize);

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

486
	if (s->flags & SLAB_STORE_USER)
Christoph Lameter's avatar
Christoph Lameter committed
487
488
489
490
		off += 2 * sizeof(struct track);

	if (off != s->size)
		/* Beginning of the filler is the free pointer */
491
492
493
		print_section("Padding", p + off, s->size - off);

	dump_stack();
Christoph Lameter's avatar
Christoph Lameter committed
494
495
496
497
498
}

static void object_err(struct kmem_cache *s, struct page *page,
			u8 *object, char *reason)
{
499
	slab_bug(s, "%s", reason);
500
	print_trailer(s, page, object);
Christoph Lameter's avatar
Christoph Lameter committed
501
502
}

503
static void slab_err(struct kmem_cache *s, struct page *page, char *fmt, ...)
Christoph Lameter's avatar
Christoph Lameter committed
504
505
506
507
{
	va_list args;
	char buf[100];

508
509
	va_start(args, fmt);
	vsnprintf(buf, sizeof(buf), fmt, args);
Christoph Lameter's avatar
Christoph Lameter committed
510
	va_end(args);
511
	slab_bug(s, "%s", buf);
512
	print_page_info(page);
Christoph Lameter's avatar
Christoph Lameter committed
513
514
515
516
517
518
519
520
521
	dump_stack();
}

static void init_object(struct kmem_cache *s, void *object, int active)
{
	u8 *p = object;

	if (s->flags & __OBJECT_POISON) {
		memset(p, POISON_FREE, s->objsize - 1);
522
		p[s->objsize - 1] = POISON_END;
Christoph Lameter's avatar
Christoph Lameter committed
523
524
525
526
527
528
529
530
	}

	if (s->flags & SLAB_RED_ZONE)
		memset(p + s->objsize,
			active ? SLUB_RED_ACTIVE : SLUB_RED_INACTIVE,
			s->inuse - s->objsize);
}

531
static u8 *check_bytes(u8 *start, unsigned int value, unsigned int bytes)
Christoph Lameter's avatar
Christoph Lameter committed
532
533
534
{
	while (bytes) {
		if (*start != (u8)value)
535
			return start;
Christoph Lameter's avatar
Christoph Lameter committed
536
537
538
		start++;
		bytes--;
	}
539
540
541
542
543
544
545
546
547
548
549
550
	return NULL;
}

static void restore_bytes(struct kmem_cache *s, char *message, u8 data,
						void *from, void *to)
{
	slab_fix(s, "Restoring 0x%p-0x%p=0x%x\n", from, to - 1, data);
	memset(from, data, to - from);
}

static int check_bytes_and_report(struct kmem_cache *s, struct page *page,
			u8 *object, char *what,
551
			u8 *start, unsigned int value, unsigned int bytes)
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
{
	u8 *fault;
	u8 *end;

	fault = check_bytes(start, value, bytes);
	if (!fault)
		return 1;

	end = start + bytes;
	while (end > fault && end[-1] == value)
		end--;

	slab_bug(s, "%s overwritten", what);
	printk(KERN_ERR "INFO: 0x%p-0x%p. First byte 0x%x instead of 0x%x\n",
					fault, end - 1, fault[0], value);
	print_trailer(s, page, object);

	restore_bytes(s, what, value, fault, end);
	return 0;
Christoph Lameter's avatar
Christoph Lameter committed
571
572
573
574
575
576
577
578
579
}

/*
 * 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
580
 *
Christoph Lameter's avatar
Christoph Lameter committed
581
582
583
584
585
 * 	Poisoning uses 0x6b (POISON_FREE) and the last byte is
 * 	0xa5 (POISON_END)
 *
 * object + s->objsize
 * 	Padding to reach word boundary. This is also used for Redzoning.
Christoph Lameter's avatar
Christoph Lameter committed
586
587
588
 * 	Padding is extended by another word if Redzoning is enabled and
 * 	objsize == inuse.
 *
Christoph Lameter's avatar
Christoph Lameter committed
589
590
591
592
 * 	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
593
594
 * 	Meta data starts here.
 *
Christoph Lameter's avatar
Christoph Lameter committed
595
596
 * 	A. Free pointer (if we cannot overwrite object on free)
 * 	B. Tracking data for SLAB_STORE_USER
Christoph Lameter's avatar
Christoph Lameter committed
597
 * 	C. Padding to reach required alignment boundary or at mininum
Christoph Lameter's avatar
Christoph Lameter committed
598
 * 		one word if debugging is on to be able to detect writes
Christoph Lameter's avatar
Christoph Lameter committed
599
600
601
 * 		before the word boundary.
 *
 *	Padding is done using 0x5a (POISON_INUSE)
Christoph Lameter's avatar
Christoph Lameter committed
602
603
 *
 * object + s->size
Christoph Lameter's avatar
Christoph Lameter committed
604
 * 	Nothing is used beyond s->size.
Christoph Lameter's avatar
Christoph Lameter committed
605
 *
Christoph Lameter's avatar
Christoph Lameter committed
606
607
 * If slabcaches are merged then the objsize and inuse boundaries are mostly
 * ignored. And therefore no slab options that rely on these boundaries
Christoph Lameter's avatar
Christoph Lameter committed
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
 * 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;

626
627
	return check_bytes_and_report(s, page, p, "Object padding",
				p + off, POISON_INUSE, s->size - off);
Christoph Lameter's avatar
Christoph Lameter committed
628
629
}

630
/* Check the pad bytes at the end of a slab page */
Christoph Lameter's avatar
Christoph Lameter committed
631
632
static int slab_pad_check(struct kmem_cache *s, struct page *page)
{
633
634
635
636
637
	u8 *start;
	u8 *fault;
	u8 *end;
	int length;
	int remainder;
Christoph Lameter's avatar
Christoph Lameter committed
638
639
640
641

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

642
	start = page_address(page);
643
	length = (PAGE_SIZE << compound_order(page));
644
645
	end = start + length;
	remainder = length % s->size;
Christoph Lameter's avatar
Christoph Lameter committed
646
647
648
	if (!remainder)
		return 1;

649
	fault = check_bytes(end - remainder, POISON_INUSE, remainder);
650
651
652
653
654
655
	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);
656
	print_section("Padding", end - remainder, remainder);
657
658
659

	restore_bytes(s, "slab padding", POISON_INUSE, start, end);
	return 0;
Christoph Lameter's avatar
Christoph Lameter committed
660
661
662
663
664
665
666
667
668
669
670
671
}

static int check_object(struct kmem_cache *s, struct page *page,
					void *object, int active)
{
	u8 *p = object;
	u8 *endobject = object + s->objsize;

	if (s->flags & SLAB_RED_ZONE) {
		unsigned int red =
			active ? SLUB_RED_ACTIVE : SLUB_RED_INACTIVE;

672
673
		if (!check_bytes_and_report(s, page, object, "Redzone",
			endobject, red, s->inuse - s->objsize))
Christoph Lameter's avatar
Christoph Lameter committed
674
675
			return 0;
	} else {
Ingo Molnar's avatar
Ingo Molnar committed
676
677
678
679
		if ((s->flags & SLAB_POISON) && s->objsize < s->inuse) {
			check_bytes_and_report(s, page, p, "Alignment padding",
				endobject, POISON_INUSE, s->inuse - s->objsize);
		}
Christoph Lameter's avatar
Christoph Lameter committed
680
681
682
683
	}

	if (s->flags & SLAB_POISON) {
		if (!active && (s->flags & __OBJECT_POISON) &&
684
685
686
			(!check_bytes_and_report(s, page, p, "Poison", p,
					POISON_FREE, s->objsize - 1) ||
			 !check_bytes_and_report(s, page, p, "Poison",
687
				p + s->objsize - 1, POISON_END, 1)))
Christoph Lameter's avatar
Christoph Lameter committed
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
			return 0;
		/*
		 * check_pad_bytes cleans up on its own.
		 */
		check_pad_bytes(s, page, p);
	}

	if (!s->offset && active)
		/*
		 * 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
706
		 * No choice but to zap it and thus lose the remainder
Christoph Lameter's avatar
Christoph Lameter committed
707
		 * of the free objects in this slab. May cause
Christoph Lameter's avatar
Christoph Lameter committed
708
		 * another error because the object count is now wrong.
Christoph Lameter's avatar
Christoph Lameter committed
709
		 */
710
		set_freepointer(s, p, NULL);
Christoph Lameter's avatar
Christoph Lameter committed
711
712
713
714
715
716
717
		return 0;
	}
	return 1;
}

static int check_slab(struct kmem_cache *s, struct page *page)
{
718
719
	int maxobj;

Christoph Lameter's avatar
Christoph Lameter committed
720
721
722
	VM_BUG_ON(!irqs_disabled());

	if (!PageSlab(page)) {
723
		slab_err(s, page, "Not a valid slab page");
Christoph Lameter's avatar
Christoph Lameter committed
724
725
		return 0;
	}
726
727
728
729
730
731
732
733

	maxobj = (PAGE_SIZE << compound_order(page)) / s->size;
	if (page->objects > maxobj) {
		slab_err(s, page, "objects %u > max %u",
			s->name, page->objects, maxobj);
		return 0;
	}
	if (page->inuse > page->objects) {
734
		slab_err(s, page, "inuse %u > max %u",
735
			s->name, page->inuse, page->objects);
Christoph Lameter's avatar
Christoph Lameter committed
736
737
738
739
740
741
742
743
		return 0;
	}
	/* Slab_pad_check fixes things up after itself */
	slab_pad_check(s, page);
	return 1;
}

/*
Christoph Lameter's avatar
Christoph Lameter committed
744
745
 * 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
746
747
748
749
750
751
 */
static int on_freelist(struct kmem_cache *s, struct page *page, void *search)
{
	int nr = 0;
	void *fp = page->freelist;
	void *object = NULL;
752
	unsigned long max_objects;
Christoph Lameter's avatar
Christoph Lameter committed
753

754
	while (fp && nr <= page->objects) {
Christoph Lameter's avatar
Christoph Lameter committed
755
756
757
758
759
760
		if (fp == search)
			return 1;
		if (!check_valid_pointer(s, page, fp)) {
			if (object) {
				object_err(s, page, object,
					"Freechain corrupt");
761
				set_freepointer(s, object, NULL);
Christoph Lameter's avatar
Christoph Lameter committed
762
763
				break;
			} else {
764
				slab_err(s, page, "Freepointer corrupt");
765
				page->freelist = NULL;
766
				page->inuse = page->objects;
767
				slab_fix(s, "Freelist cleared");
Christoph Lameter's avatar
Christoph Lameter committed
768
769
770
771
772
773
774
775
776
				return 0;
			}
			break;
		}
		object = fp;
		fp = get_freepointer(s, object);
		nr++;
	}

777
	max_objects = (PAGE_SIZE << compound_order(page)) / s->size;
778
779
	if (max_objects > MAX_OBJS_PER_PAGE)
		max_objects = MAX_OBJS_PER_PAGE;
780
781
782
783
784
785
786

	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.");
	}
787
	if (page->inuse != page->objects - nr) {
788
		slab_err(s, page, "Wrong object count. Counter is %d but "
789
790
			"counted were %d", page->inuse, page->objects - nr);
		page->inuse = page->objects - nr;
791
		slab_fix(s, "Object count adjusted.");
Christoph Lameter's avatar
Christoph Lameter committed
792
793
794
795
	}
	return search == NULL;
}

796
797
static void trace(struct kmem_cache *s, struct page *page, void *object,
								int alloc)
Christoph Lameter's avatar
Christoph Lameter committed
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
{
	if (s->flags & SLAB_TRACE) {
		printk(KERN_INFO "TRACE %s %s 0x%p inuse=%d fp=0x%p\n",
			s->name,
			alloc ? "alloc" : "free",
			object, page->inuse,
			page->freelist);

		if (!alloc)
			print_section("Object", (void *)object, s->objsize);

		dump_stack();
	}
}

813
/*
Christoph Lameter's avatar
Christoph Lameter committed
814
 * Tracking of fully allocated slabs for debugging purposes.
815
 */
Christoph Lameter's avatar
Christoph Lameter committed
816
static void add_full(struct kmem_cache_node *n, struct page *page)
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
{
	spin_lock(&n->list_lock);
	list_add(&page->lru, &n->full);
	spin_unlock(&n->list_lock);
}

static void remove_full(struct kmem_cache *s, struct page *page)
{
	struct kmem_cache_node *n;

	if (!(s->flags & SLAB_STORE_USER))
		return;

	n = get_node(s, page_to_nid(page));

	spin_lock(&n->list_lock);
	list_del(&page->lru);
	spin_unlock(&n->list_lock);
}

837
838
839
840
841
842
843
844
/* 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);
}

845
846
847
848
849
static inline unsigned long node_nr_slabs(struct kmem_cache_node *n)
{
	return atomic_long_read(&n->nr_slabs);
}

850
static inline void inc_slabs_node(struct kmem_cache *s, int node, int objects)
851
852
853
854
855
856
857
858
859
{
	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).
	 */
860
	if (!NUMA_BUILD || n) {
861
		atomic_long_inc(&n->nr_slabs);
862
863
		atomic_long_add(objects, &n->total_objects);
	}
864
}
865
static inline void dec_slabs_node(struct kmem_cache *s, int node, int objects)
866
867
868
869
{
	struct kmem_cache_node *n = get_node(s, node);

	atomic_long_dec(&n->nr_slabs);
870
	atomic_long_sub(objects, &n->total_objects);
871
872
873
}

/* Object debug checks for alloc/free paths */
Christoph Lameter's avatar
Christoph Lameter committed
874
875
876
877
878
879
880
881
882
883
884
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;

	init_object(s, object, 0);
	init_tracking(s, object);
}

static int alloc_debug_processing(struct kmem_cache *s, struct page *page,
885
					void *object, unsigned long addr)
Christoph Lameter's avatar
Christoph Lameter committed
886
887
888
889
{
	if (!check_slab(s, page))
		goto bad;

890
	if (!on_freelist(s, page, object)) {
891
		object_err(s, page, object, "Object already allocated");
892
		goto bad;
Christoph Lameter's avatar
Christoph Lameter committed
893
894
895
896
	}

	if (!check_valid_pointer(s, page, object)) {
		object_err(s, page, object, "Freelist Pointer check fails");
897
		goto bad;
Christoph Lameter's avatar
Christoph Lameter committed
898
899
	}

900
	if (!check_object(s, page, object, 0))
Christoph Lameter's avatar
Christoph Lameter committed
901
902
		goto bad;

Christoph Lameter's avatar
Christoph Lameter committed
903
904
905
906
907
	/* 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);
	init_object(s, object, 1);
Christoph Lameter's avatar
Christoph Lameter committed
908
	return 1;
Christoph Lameter's avatar
Christoph Lameter committed
909

Christoph Lameter's avatar
Christoph Lameter committed
910
911
912
913
914
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
915
		 * as used avoids touching the remaining objects.
Christoph Lameter's avatar
Christoph Lameter committed
916
		 */
917
		slab_fix(s, "Marking all objects used");
918
		page->inuse = page->objects;
919
		page->freelist = NULL;
Christoph Lameter's avatar
Christoph Lameter committed
920
921
922
923
	}
	return 0;
}

Christoph Lameter's avatar
Christoph Lameter committed
924
static int free_debug_processing(struct kmem_cache *s, struct page *page,
925
					void *object, unsigned long addr)
Christoph Lameter's avatar
Christoph Lameter committed
926
927
928
929
930
{
	if (!check_slab(s, page))
		goto fail;

	if (!check_valid_pointer(s, page, object)) {
931
		slab_err(s, page, "Invalid object pointer 0x%p", object);
Christoph Lameter's avatar
Christoph Lameter committed
932
933
934
935
		goto fail;
	}

	if (on_freelist(s, page, object)) {
936
		object_err(s, page, object, "Object already free");
Christoph Lameter's avatar
Christoph Lameter committed
937
938
939
940
941
942
943
		goto fail;
	}

	if (!check_object(s, page, object, 1))
		return 0;

	if (unlikely(s != page->slab)) {
Ingo Molnar's avatar
Ingo Molnar committed
944
		if (!PageSlab(page)) {
945
946
			slab_err(s, page, "Attempt to free object(0x%p) "
				"outside of slab", object);
Ingo Molnar's avatar
Ingo Molnar committed
947
		} else if (!page->slab) {
Christoph Lameter's avatar
Christoph Lameter committed
948
			printk(KERN_ERR
949
				"SLUB <none>: no slab for object 0x%p.\n",
Christoph Lameter's avatar
Christoph Lameter committed
950
						object);
951
			dump_stack();
952
		} else
953
954
			object_err(s, page, object,
					"page slab pointer corrupt.");
Christoph Lameter's avatar
Christoph Lameter committed
955
956
		goto fail;
	}
Christoph Lameter's avatar
Christoph Lameter committed
957
958

	/* Special debug activities for freeing objects */
959
	if (!PageSlubFrozen(page) && !page->freelist)
Christoph Lameter's avatar
Christoph Lameter committed
960
961
962
963
964
		remove_full(s, page);
	if (s->flags & SLAB_STORE_USER)
		set_track(s, object, TRACK_FREE, addr);
	trace(s, page, object, 0);
	init_object(s, object, 0);
Christoph Lameter's avatar
Christoph Lameter committed
965
	return 1;
Christoph Lameter's avatar
Christoph Lameter committed
966

Christoph Lameter's avatar
Christoph Lameter committed
967
fail:
968
	slab_fix(s, "Object at 0x%p not freed", object);
Christoph Lameter's avatar
Christoph Lameter committed
969
970
971
	return 0;
}

972
973
static int __init setup_slub_debug(char *str)
{
974
975
976
977
978
979
980
981
982
983
984
985
986
987
	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;

988
989
990
991
992
993
994
995
996
	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;
	}

997
998
999
1000
	slub_debug = 0;
	if (*str == '-')
		/*
		 * Switch off all debugging measures.