slub.c 105.3 KB
Newer Older
C
Christoph Lameter 已提交
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.
 *
C
Christoph Lameter 已提交
8
 * (C) 2007 SGI, Christoph Lameter
C
Christoph Lameter 已提交
9 10 11 12 13 14 15 16
 */

#include <linux/mm.h>
#include <linux/module.h>
#include <linux/bit_spinlock.h>
#include <linux/interrupt.h>
#include <linux/bitops.h>
#include <linux/slab.h>
17
#include <linux/proc_fs.h>
C
Christoph Lameter 已提交
18 19 20 21 22
#include <linux/seq_file.h>
#include <linux/cpu.h>
#include <linux/cpuset.h>
#include <linux/mempolicy.h>
#include <linux/ctype.h>
23
#include <linux/debugobjects.h>
C
Christoph Lameter 已提交
24
#include <linux/kallsyms.h>
25
#include <linux/memory.h>
R
Roman Zippel 已提交
26
#include <linux/math64.h>
C
Christoph Lameter 已提交
27 28 29 30 31 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

/*
 * 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.
 *
C
Christoph Lameter 已提交
73 74
 * 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
C
Christoph Lameter 已提交
75
 * freed then the slab will show up again on the partial lists.
C
Christoph Lameter 已提交
76 77
 * We track full slabs for debugging purposes though because otherwise we
 * cannot scan all objects.
C
Christoph Lameter 已提交
78 79 80 81 82 83 84
 *
 * 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.
 *
85 86 87 88 89 90 91 92 93 94 95 96
 * 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
97
 * 			freelist that allows lockless access to
98 99
 * 			free objects in addition to the regular freelist
 * 			that requires the slab lock.
C
Christoph Lameter 已提交
100 101 102
 *
 * PageError		Slab requires special handling due to debug
 * 			options set. This moves	slab handling out of
103
 * 			the fast path and disables lockless freelists.
C
Christoph Lameter 已提交
104 105
 */

106
#ifdef CONFIG_SLUB_DEBUG
107
#define SLABDEBUG 1
108 109 110 111
#else
#define SLABDEBUG 0
#endif

C
Christoph Lameter 已提交
112 113 114 115 116 117 118 119 120 121 122
/*
 * 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

123 124 125 126
/*
 * Mininum number of partial slabs. These will be left on the partial
 * lists even if they are empty. kmem_cache_shrink may reclaim them.
 */
C
Christoph Lameter 已提交
127
#define MIN_PARTIAL 5
C
Christoph Lameter 已提交
128

129 130 131 132 133 134 135
/*
 * 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

C
Christoph Lameter 已提交
136 137
#define DEBUG_DEFAULT_FLAGS (SLAB_DEBUG_FREE | SLAB_RED_ZONE | \
				SLAB_POISON | SLAB_STORE_USER)
C
Christoph Lameter 已提交
138

C
Christoph Lameter 已提交
139 140 141 142 143 144 145 146 147 148
/*
 * Set of flags that will prevent slab merging
 */
#define SLUB_NEVER_MERGE (SLAB_RED_ZONE | SLAB_POISON | SLAB_STORE_USER | \
		SLAB_TRACE | SLAB_DESTROY_BY_RCU)

#define SLUB_MERGE_SAME (SLAB_DEBUG_FREE | SLAB_RECLAIM_ACCOUNT | \
		SLAB_CACHE_DMA)

#ifndef ARCH_KMALLOC_MINALIGN
149
#define ARCH_KMALLOC_MINALIGN __alignof__(unsigned long long)
C
Christoph Lameter 已提交
150 151 152
#endif

#ifndef ARCH_SLAB_MINALIGN
153
#define ARCH_SLAB_MINALIGN __alignof__(unsigned long long)
C
Christoph Lameter 已提交
154 155 156
#endif

/* Internal SLUB flags */
157 158
#define __OBJECT_POISON		0x80000000 /* Poison object */
#define __SYSFS_ADD_DEFERRED	0x40000000 /* Not yet visible via sysfs */
C
Christoph Lameter 已提交
159 160 161 162 163 164 165 166 167 168

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 */
C
Christoph Lameter 已提交
169
	UP,		/* Everything works but does not show up in sysfs */
C
Christoph Lameter 已提交
170 171 172 173 174
	SYSFS		/* Sysfs up */
} slab_state = DOWN;

/* A list of all slab caches on the system */
static DECLARE_RWSEM(slub_lock);
A
Adrian Bunk 已提交
175
static LIST_HEAD(slab_caches);
C
Christoph Lameter 已提交
176

177 178 179 180 181 182 183 184 185 186 187 188
/*
 * Tracking user of a slab.
 */
struct track {
	void *addr;		/* Called from address */
	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 };

C
Christoph Lameter 已提交
189
#ifdef CONFIG_SLUB_DEBUG
C
Christoph Lameter 已提交
190 191 192
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 *);
193

C
Christoph Lameter 已提交
194
#else
195 196 197
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; }
C
Christoph Lameter 已提交
198 199 200 201
static inline void sysfs_slab_remove(struct kmem_cache *s)
{
	kfree(s);
}
202

C
Christoph Lameter 已提交
203 204
#endif

205 206 207 208 209 210 211
static inline void stat(struct kmem_cache_cpu *c, enum stat_item si)
{
#ifdef CONFIG_SLUB_STATS
	c->stat[si]++;
#endif
}

C
Christoph Lameter 已提交
212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229
/********************************************************************
 * 			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
}

230 231
static inline struct kmem_cache_cpu *get_cpu_slab(struct kmem_cache *s, int cpu)
{
232 233 234 235 236
#ifdef CONFIG_SMP
	return s->cpu_slab[cpu];
#else
	return &s->cpu_slab;
#endif
237 238
}

C
Christoph Lameter 已提交
239
/* Verify that a pointer has an address that is valid within a slab page */
240 241 242 243 244
static inline int check_valid_pointer(struct kmem_cache *s,
				struct page *page, const void *object)
{
	void *base;

245
	if (!object)
246 247
		return 1;

248
	base = page_address(page);
249
	if (object < base || object >= base + page->objects * s->size ||
250 251 252 253 254 255 256
		(object - base) % s->size) {
		return 0;
	}

	return 1;
}

257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274
/*
 * 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 */
275 276
#define for_each_object(__p, __s, __addr, __objects) \
	for (__p = (__addr); __p < (__addr) + (__objects) * (__s)->size;\
277 278 279 280
			__p += (__s)->size)

/* Scan freelist */
#define for_each_free_object(__p, __s, __free) \
281
	for (__p = (__free); __p; __p = get_freepointer((__s), __p))
282 283 284 285 286 287 288

/* 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;
}

289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308
static inline struct kmem_cache_order_objects oo_make(int order,
						unsigned long size)
{
	struct kmem_cache_order_objects x = {
		(order << 16) + (PAGE_SIZE << order) / size
	};

	return x;
}

static inline int oo_order(struct kmem_cache_order_objects x)
{
	return x.x >> 16;
}

static inline int oo_objects(struct kmem_cache_order_objects x)
{
	return x.x & ((1 << 16) - 1);
}

C
Christoph Lameter 已提交
309 310 311 312
#ifdef CONFIG_SLUB_DEBUG
/*
 * Debug settings:
 */
313 314 315
#ifdef CONFIG_SLUB_DEBUG_ON
static int slub_debug = DEBUG_DEFAULT_FLAGS;
#else
C
Christoph Lameter 已提交
316
static int slub_debug;
317
#endif
C
Christoph Lameter 已提交
318 319 320

static char *slub_debug_slabs;

C
Christoph Lameter 已提交
321 322 323 324 325 326 327 328 329 330 331 332 333
/*
 * 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) {
334
			printk(KERN_ERR "%8s 0x%p: ", text, addr + i);
C
Christoph Lameter 已提交
335 336
			newline = 0;
		}
P
Pekka Enberg 已提交
337
		printk(KERN_CONT " %02x", addr[i]);
C
Christoph Lameter 已提交
338 339 340
		offset = i % 16;
		ascii[offset] = isgraph(addr[i]) ? addr[i] : '.';
		if (offset == 15) {
P
Pekka Enberg 已提交
341
			printk(KERN_CONT " %s\n", ascii);
C
Christoph Lameter 已提交
342 343 344 345 346 347
			newline = 1;
		}
	}
	if (!newline) {
		i %= 16;
		while (i < 16) {
P
Pekka Enberg 已提交
348
			printk(KERN_CONT "   ");
C
Christoph Lameter 已提交
349 350 351
			ascii[i] = ' ';
			i++;
		}
P
Pekka Enberg 已提交
352
		printk(KERN_CONT " %s\n", ascii);
C
Christoph Lameter 已提交
353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382
	}
}

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,
				enum track_item alloc, void *addr)
{
	struct track *p;

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

	p += alloc;
	if (addr) {
		p->addr = addr;
		p->cpu = smp_processor_id();
A
Alexey Dobriyan 已提交
383
		p->pid = current->pid;
C
Christoph Lameter 已提交
384 385 386 387 388 389 390
		p->when = jiffies;
	} else
		memset(p, 0, sizeof(struct track));
}

static void init_tracking(struct kmem_cache *s, void *object)
{
391 392 393 394 395
	if (!(s->flags & SLAB_STORE_USER))
		return;

	set_track(s, object, TRACK_FREE, NULL);
	set_track(s, object, TRACK_ALLOC, NULL);
C
Christoph Lameter 已提交
396 397 398 399 400 401 402
}

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

403 404
	printk(KERN_ERR "INFO: %s in %pS age=%lu cpu=%u pid=%d\n",
		s, t->addr, jiffies - t->when, t->cpu, t->pid);
405 406 407 408 409 410 411 412 413 414 415 416 417
}

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)
{
418 419
	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);
420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435

}

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");
C
Christoph Lameter 已提交
436 437
}

438 439 440 441 442 443 444 445 446 447 448 449
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)
C
Christoph Lameter 已提交
450 451
{
	unsigned int off;	/* Offset of last byte */
452
	u8 *addr = page_address(page);
453 454 455 456 457 458 459 460 461 462 463

	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);

464
	print_section("Object", p, min_t(unsigned long, s->objsize, PAGE_SIZE));
C
Christoph Lameter 已提交
465 466 467 468 469 470 471 472 473 474

	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;

475
	if (s->flags & SLAB_STORE_USER)
C
Christoph Lameter 已提交
476 477 478 479
		off += 2 * sizeof(struct track);

	if (off != s->size)
		/* Beginning of the filler is the free pointer */
480 481 482
		print_section("Padding", p + off, s->size - off);

	dump_stack();
C
Christoph Lameter 已提交
483 484 485 486 487
}

static void object_err(struct kmem_cache *s, struct page *page,
			u8 *object, char *reason)
{
488
	slab_bug(s, "%s", reason);
489
	print_trailer(s, page, object);
C
Christoph Lameter 已提交
490 491
}

492
static void slab_err(struct kmem_cache *s, struct page *page, char *fmt, ...)
C
Christoph Lameter 已提交
493 494 495 496
{
	va_list args;
	char buf[100];

497 498
	va_start(args, fmt);
	vsnprintf(buf, sizeof(buf), fmt, args);
C
Christoph Lameter 已提交
499
	va_end(args);
500
	slab_bug(s, "%s", buf);
501
	print_page_info(page);
C
Christoph Lameter 已提交
502 503 504 505 506 507 508 509 510
	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);
P
Pekka Enberg 已提交
511
		p[s->objsize - 1] = POISON_END;
C
Christoph Lameter 已提交
512 513 514 515 516 517 518 519
	}

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

520
static u8 *check_bytes(u8 *start, unsigned int value, unsigned int bytes)
C
Christoph Lameter 已提交
521 522 523
{
	while (bytes) {
		if (*start != (u8)value)
524
			return start;
C
Christoph Lameter 已提交
525 526 527
		start++;
		bytes--;
	}
528 529 530 531 532 533 534 535 536 537 538 539
	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,
P
Pekka Enberg 已提交
540
			u8 *start, unsigned int value, unsigned int bytes)
541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559
{
	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;
C
Christoph Lameter 已提交
560 561 562 563 564 565 566 567 568
}

/*
 * 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.
C
Christoph Lameter 已提交
569
 *
C
Christoph Lameter 已提交
570 571 572 573 574
 * 	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.
C
Christoph Lameter 已提交
575 576 577
 * 	Padding is extended by another word if Redzoning is enabled and
 * 	objsize == inuse.
 *
C
Christoph Lameter 已提交
578 579 580 581
 * 	We fill with 0xbb (RED_INACTIVE) for inactive objects and with
 * 	0xcc (RED_ACTIVE) for objects in use.
 *
 * object + s->inuse
C
Christoph Lameter 已提交
582 583
 * 	Meta data starts here.
 *
C
Christoph Lameter 已提交
584 585
 * 	A. Free pointer (if we cannot overwrite object on free)
 * 	B. Tracking data for SLAB_STORE_USER
C
Christoph Lameter 已提交
586
 * 	C. Padding to reach required alignment boundary or at mininum
C
Christoph Lameter 已提交
587
 * 		one word if debugging is on to be able to detect writes
C
Christoph Lameter 已提交
588 589 590
 * 		before the word boundary.
 *
 *	Padding is done using 0x5a (POISON_INUSE)
C
Christoph Lameter 已提交
591 592
 *
 * object + s->size
C
Christoph Lameter 已提交
593
 * 	Nothing is used beyond s->size.
C
Christoph Lameter 已提交
594
 *
C
Christoph Lameter 已提交
595 596
 * If slabcaches are merged then the objsize and inuse boundaries are mostly
 * ignored. And therefore no slab options that rely on these boundaries
C
Christoph Lameter 已提交
597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614
 * 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;

615 616
	return check_bytes_and_report(s, page, p, "Object padding",
				p + off, POISON_INUSE, s->size - off);
C
Christoph Lameter 已提交
617 618
}

619
/* Check the pad bytes at the end of a slab page */
C
Christoph Lameter 已提交
620 621
static int slab_pad_check(struct kmem_cache *s, struct page *page)
{
622 623 624 625 626
	u8 *start;
	u8 *fault;
	u8 *end;
	int length;
	int remainder;
C
Christoph Lameter 已提交
627 628 629 630

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

631
	start = page_address(page);
632
	length = (PAGE_SIZE << compound_order(page));
633 634
	end = start + length;
	remainder = length % s->size;
C
Christoph Lameter 已提交
635 636 637
	if (!remainder)
		return 1;

638
	fault = check_bytes(end - remainder, POISON_INUSE, remainder);
639 640 641 642 643 644
	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);
645
	print_section("Padding", end - remainder, remainder);
646 647 648

	restore_bytes(s, "slab padding", POISON_INUSE, start, end);
	return 0;
C
Christoph Lameter 已提交
649 650 651 652 653 654 655 656 657 658 659 660
}

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;

661 662
		if (!check_bytes_and_report(s, page, object, "Redzone",
			endobject, red, s->inuse - s->objsize))
C
Christoph Lameter 已提交
663 664
			return 0;
	} else {
I
Ingo Molnar 已提交
665 666 667 668
		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);
		}
C
Christoph Lameter 已提交
669 670 671 672
	}

	if (s->flags & SLAB_POISON) {
		if (!active && (s->flags & __OBJECT_POISON) &&
673 674 675
			(!check_bytes_and_report(s, page, p, "Poison", p,
					POISON_FREE, s->objsize - 1) ||
			 !check_bytes_and_report(s, page, p, "Poison",
P
Pekka Enberg 已提交
676
				p + s->objsize - 1, POISON_END, 1)))
C
Christoph Lameter 已提交
677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696
			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");
		/*
		 * No choice but to zap it and thus loose the remainder
		 * of the free objects in this slab. May cause
C
Christoph Lameter 已提交
697
		 * another error because the object count is now wrong.
C
Christoph Lameter 已提交
698
		 */
699
		set_freepointer(s, p, NULL);
C
Christoph Lameter 已提交
700 701 702 703 704 705 706
		return 0;
	}
	return 1;
}

static int check_slab(struct kmem_cache *s, struct page *page)
{
707 708
	int maxobj;

C
Christoph Lameter 已提交
709 710 711
	VM_BUG_ON(!irqs_disabled());

	if (!PageSlab(page)) {
712
		slab_err(s, page, "Not a valid slab page");
C
Christoph Lameter 已提交
713 714
		return 0;
	}
715 716 717 718 719 720 721 722

	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) {
723
		slab_err(s, page, "inuse %u > max %u",
724
			s->name, page->inuse, page->objects);
C
Christoph Lameter 已提交
725 726 727 728 729 730 731 732
		return 0;
	}
	/* Slab_pad_check fixes things up after itself */
	slab_pad_check(s, page);
	return 1;
}

/*
C
Christoph Lameter 已提交
733 734
 * 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.
C
Christoph Lameter 已提交
735 736 737 738 739 740
 */
static int on_freelist(struct kmem_cache *s, struct page *page, void *search)
{
	int nr = 0;
	void *fp = page->freelist;
	void *object = NULL;
741
	unsigned long max_objects;
C
Christoph Lameter 已提交
742

743
	while (fp && nr <= page->objects) {
C
Christoph Lameter 已提交
744 745 746 747 748 749
		if (fp == search)
			return 1;
		if (!check_valid_pointer(s, page, fp)) {
			if (object) {
				object_err(s, page, object,
					"Freechain corrupt");
750
				set_freepointer(s, object, NULL);
C
Christoph Lameter 已提交
751 752
				break;
			} else {
753
				slab_err(s, page, "Freepointer corrupt");
754
				page->freelist = NULL;
755
				page->inuse = page->objects;
756
				slab_fix(s, "Freelist cleared");
C
Christoph Lameter 已提交
757 758 759 760 761 762 763 764 765
				return 0;
			}
			break;
		}
		object = fp;
		fp = get_freepointer(s, object);
		nr++;
	}

766 767 768 769 770 771 772 773 774 775
	max_objects = (PAGE_SIZE << compound_order(page)) / s->size;
	if (max_objects > 65535)
		max_objects = 65535;

	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.");
	}
776
	if (page->inuse != page->objects - nr) {
777
		slab_err(s, page, "Wrong object count. Counter is %d but "
778 779
			"counted were %d", page->inuse, page->objects - nr);
		page->inuse = page->objects - nr;
780
		slab_fix(s, "Object count adjusted.");
C
Christoph Lameter 已提交
781 782 783 784
	}
	return search == NULL;
}

785 786
static void trace(struct kmem_cache *s, struct page *page, void *object,
								int alloc)
C
Christoph Lameter 已提交
787 788 789 790 791 792 793 794 795 796 797 798 799 800 801
{
	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();
	}
}

802
/*
C
Christoph Lameter 已提交
803
 * Tracking of fully allocated slabs for debugging purposes.
804
 */
C
Christoph Lameter 已提交
805
static void add_full(struct kmem_cache_node *n, struct page *page)
806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825
{
	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);
}

826 827 828 829 830 831 832 833
/* 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);
}

834
static inline void inc_slabs_node(struct kmem_cache *s, int node, int objects)
835 836 837 838 839 840 841 842 843
{
	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).
	 */
844
	if (!NUMA_BUILD || n) {
845
		atomic_long_inc(&n->nr_slabs);
846 847
		atomic_long_add(objects, &n->total_objects);
	}
848
}
849
static inline void dec_slabs_node(struct kmem_cache *s, int node, int objects)
850 851 852 853
{
	struct kmem_cache_node *n = get_node(s, node);

	atomic_long_dec(&n->nr_slabs);
854
	atomic_long_sub(objects, &n->total_objects);
855 856 857
}

/* Object debug checks for alloc/free paths */
C
Christoph Lameter 已提交
858 859 860 861 862 863 864 865 866 867 868 869
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,
						void *object, void *addr)
C
Christoph Lameter 已提交
870 871 872 873
{
	if (!check_slab(s, page))
		goto bad;

874
	if (!on_freelist(s, page, object)) {
875
		object_err(s, page, object, "Object already allocated");
876
		goto bad;
C
Christoph Lameter 已提交
877 878 879 880
	}

	if (!check_valid_pointer(s, page, object)) {
		object_err(s, page, object, "Freelist Pointer check fails");
881
		goto bad;
C
Christoph Lameter 已提交
882 883
	}

884
	if (!check_object(s, page, object, 0))
C
Christoph Lameter 已提交
885 886
		goto bad;

C
Christoph Lameter 已提交
887 888 889 890 891
	/* 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);
C
Christoph Lameter 已提交
892
	return 1;
C
Christoph Lameter 已提交
893

C
Christoph Lameter 已提交
894 895 896 897 898
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
C
Christoph Lameter 已提交
899
		 * as used avoids touching the remaining objects.
C
Christoph Lameter 已提交
900
		 */
901
		slab_fix(s, "Marking all objects used");
902
		page->inuse = page->objects;
903
		page->freelist = NULL;
C
Christoph Lameter 已提交
904 905 906 907
	}
	return 0;
}

C
Christoph Lameter 已提交
908 909
static int free_debug_processing(struct kmem_cache *s, struct page *page,
						void *object, void *addr)
C
Christoph Lameter 已提交
910 911 912 913 914
{
	if (!check_slab(s, page))
		goto fail;

	if (!check_valid_pointer(s, page, object)) {
915
		slab_err(s, page, "Invalid object pointer 0x%p", object);
C
Christoph Lameter 已提交
916 917 918 919
		goto fail;
	}

	if (on_freelist(s, page, object)) {
920
		object_err(s, page, object, "Object already free");
C
Christoph Lameter 已提交
921 922 923 924 925 926 927
		goto fail;
	}

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

	if (unlikely(s != page->slab)) {
I
Ingo Molnar 已提交
928
		if (!PageSlab(page)) {
929 930
			slab_err(s, page, "Attempt to free object(0x%p) "
				"outside of slab", object);
I
Ingo Molnar 已提交
931
		} else if (!page->slab) {
C
Christoph Lameter 已提交
932
			printk(KERN_ERR
933
				"SLUB <none>: no slab for object 0x%p.\n",
C
Christoph Lameter 已提交
934
						object);
935
			dump_stack();
P
Pekka Enberg 已提交
936
		} else
937 938
			object_err(s, page, object,
					"page slab pointer corrupt.");
C
Christoph Lameter 已提交
939 940
		goto fail;
	}
C
Christoph Lameter 已提交
941 942

	/* Special debug activities for freeing objects */
943
	if (!PageSlubFrozen(page) && !page->freelist)
C
Christoph Lameter 已提交
944 945 946 947 948
		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);
C
Christoph Lameter 已提交
949
	return 1;
C
Christoph Lameter 已提交
950

C
Christoph Lameter 已提交
951
fail:
952
	slab_fix(s, "Object at 0x%p not freed", object);
C
Christoph Lameter 已提交
953 954 955
	return 0;
}

C
Christoph Lameter 已提交
956 957
static int __init setup_slub_debug(char *str)
{
958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981
	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;

	slub_debug = 0;
	if (*str == '-')
		/*
		 * Switch off all debugging measures.
		 */
		goto out;

	/*
	 * Determine which debug features should be switched on
	 */
P
Pekka Enberg 已提交
982
	for (; *str && *str != ','; str++) {
983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000
		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;
		default:
			printk(KERN_ERR "slub_debug option '%c' "
P
Pekka Enberg 已提交
1001
				"unknown. skipped\n", *str);
1002
		}
C
Christoph Lameter 已提交
1003 1004
	}

1005
check_slabs:
C
Christoph Lameter 已提交
1006 1007
	if (*str == ',')
		slub_debug_slabs = str + 1;
1008
out:
C
Christoph Lameter 已提交
1009 1010 1011 1012 1013
	return 1;
}

__setup("slub_debug", setup_slub_debug);

1014 1015
static unsigned long kmem_cache_flags(unsigned long objsize,
	unsigned long flags, const char *name,
1016
	void (*ctor)(void *))
C
Christoph Lameter 已提交
1017 1018
{
	/*
1019
	 * Enable debugging if selected on the kernel commandline.
C
Christoph Lameter 已提交
1020
	 */
1021 1022 1023
	if (slub_debug && (!slub_debug_slabs ||
	    strncmp(slub_debug_slabs, name, strlen(slub_debug_slabs)) == 0))
			flags |= slub_debug;
1024 1025

	return flags;
C
Christoph Lameter 已提交
1026 1027
}
#else
C
Christoph Lameter 已提交
1028 1029
static inline void setup_object_debug(struct kmem_cache *s,
			struct page *page, void *object) {}
C
Christoph Lameter 已提交
1030

C
Christoph Lameter 已提交
1031 1032
static inline int alloc_debug_processing(struct kmem_cache *s,
	struct page *page, void *object, void *addr) { return 0; }
C
Christoph Lameter 已提交
1033

C
Christoph Lameter 已提交
1034 1035
static inline int free_debug_processing(struct kmem_cache *s,
	struct page *page, void *object, void *addr) { return 0; }
C
Christoph Lameter 已提交
1036 1037 1038 1039 1040

static inline int slab_pad_check(struct kmem_cache *s, struct page *page)
			{ return 1; }
static inline int check_object(struct kmem_cache *s, struct page *page,
			void *object, int active) { return 1; }
C
Christoph Lameter 已提交
1041
static inline void add_full(struct kmem_cache_node *n, struct page *page) {}
1042 1043
static inline unsigned long kmem_cache_flags(unsigned long objsize,
	unsigned long flags, const char *name,
1044
	void (*ctor)(void *))
1045 1046 1047
{
	return flags;
}
C
Christoph Lameter 已提交
1048
#define slub_debug 0
1049 1050 1051

static inline unsigned long slabs_node(struct kmem_cache *s, int node)
							{ return 0; }
1052 1053 1054 1055
static inline void inc_slabs_node(struct kmem_cache *s, int node,
							int objects) {}
static inline void dec_slabs_node(struct kmem_cache *s, int node,
							int objects) {}
C
Christoph Lameter 已提交
1056
#endif
1057

C
Christoph Lameter 已提交
1058 1059 1060
/*
 * Slab allocation and freeing
 */
1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071
static inline struct page *alloc_slab_page(gfp_t flags, int node,
					struct kmem_cache_order_objects oo)
{
	int order = oo_order(oo);

	if (node == -1)
		return alloc_pages(flags, order);
	else
		return alloc_pages_node(node, flags, order);
}

C
Christoph Lameter 已提交
1072 1073
static struct page *allocate_slab(struct kmem_cache *s, gfp_t flags, int node)
{
P
Pekka Enberg 已提交
1074
	struct page *page;
1075
	struct kmem_cache_order_objects oo = s->oo;
C
Christoph Lameter 已提交
1076

1077
	flags |= s->allocflags;
1078

1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089
	page = alloc_slab_page(flags | __GFP_NOWARN | __GFP_NORETRY, node,
									oo);
	if (unlikely(!page)) {
		oo = s->min;
		/*
		 * Allocation may have failed due to fragmentation.
		 * Try a lower order alloc if possible
		 */
		page = alloc_slab_page(flags, node, oo);
		if (!page)
			return NULL;
C
Christoph Lameter 已提交
1090

1091 1092
		stat(get_cpu_slab(s, raw_smp_processor_id()), ORDER_FALLBACK);
	}
1093
	page->objects = oo_objects(oo);
C
Christoph Lameter 已提交
1094 1095 1096
	mod_zone_page_state(page_zone(page),
		(s->flags & SLAB_RECLAIM_ACCOUNT) ?
		NR_SLAB_RECLAIMABLE : NR_SLAB_UNRECLAIMABLE,
1097
		1 << oo_order(oo));
C
Christoph Lameter 已提交
1098 1099 1100 1101 1102 1103 1104

	return page;
}

static void setup_object(struct kmem_cache *s, struct page *page,
				void *object)
{
C
Christoph Lameter 已提交
1105
	setup_object_debug(s, page, object);
1106
	if (unlikely(s->ctor))
1107
		s->ctor(object);
C
Christoph Lameter 已提交
1108 1109 1110 1111 1112 1113 1114 1115 1116
}

static struct page *new_slab(struct kmem_cache *s, gfp_t flags, int node)
{
	struct page *page;
	void *start;
	void *last;
	void *p;

C
Christoph Lameter 已提交
1117
	BUG_ON(flags & GFP_SLAB_BUG_MASK);
C
Christoph Lameter 已提交
1118

C
Christoph Lameter 已提交
1119 1120
	page = allocate_slab(s,
		flags & (GFP_RECLAIM_MASK | GFP_CONSTRAINT_MASK), node);
C
Christoph Lameter 已提交
1121 1122 1123
	if (!page)
		goto out;

1124
	inc_slabs_node(s, page_to_nid(page), page->objects);
C
Christoph Lameter 已提交
1125 1126 1127 1128
	page->slab = s;
	page->flags |= 1 << PG_slab;
	if (s->flags & (SLAB_DEBUG_FREE | SLAB_RED_ZONE | SLAB_POISON |
			SLAB_STORE_USER | SLAB_TRACE))
1129
		__SetPageSlubDebug(page);
C
Christoph Lameter 已提交
1130 1131 1132 1133

	start = page_address(page);

	if (unlikely(s->flags & SLAB_POISON))
1134
		memset(start, POISON_INUSE, PAGE_SIZE << compound_order(page));
C
Christoph Lameter 已提交
1135 1136

	last = start;
1137
	for_each_object(p, s, start, page->objects) {
C
Christoph Lameter 已提交
1138 1139 1140 1141 1142
		setup_object(s, page, last);
		set_freepointer(s, last, p);
		last = p;
	}
	setup_object(s, page, last);
1143
	set_freepointer(s, last, NULL);
C
Christoph Lameter 已提交
1144 1145 1146 1147 1148 1149 1150 1151 1152

	page->freelist = start;
	page->inuse = 0;
out:
	return page;
}

static void __free_slab(struct kmem_cache *s, struct page *page)
{
1153 1154
	int order = compound_order(page);
	int pages = 1 << order;
C
Christoph Lameter 已提交
1155

1156
	if (unlikely(SLABDEBUG && PageSlubDebug(page))) {
C
Christoph Lameter 已提交
1157 1158 1159
		void *p;

		slab_pad_check(s, page);
1160 1161
		for_each_object(p, s, page_address(page),
						page->objects)
C
Christoph Lameter 已提交
1162
			check_object(s, page, p, 0);
1163
		__ClearPageSlubDebug(page);
C
Christoph Lameter 已提交
1164 1165 1166 1167 1168
	}

	mod_zone_page_state(page_zone(page),
		(s->flags & SLAB_RECLAIM_ACCOUNT) ?
		NR_SLAB_RECLAIMABLE : NR_SLAB_UNRECLAIMABLE,
P
Pekka Enberg 已提交
1169
		-pages);
C
Christoph Lameter 已提交
1170

1171 1172
	__ClearPageSlab(page);
	reset_page_mapcount(page);
1173
	__free_pages(page, order);
C
Christoph Lameter 已提交
1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198
}

static void rcu_free_slab(struct rcu_head *h)
{
	struct page *page;

	page = container_of((struct list_head *)h, struct page, lru);
	__free_slab(page->slab, page);
}

static void free_slab(struct kmem_cache *s, struct page *page)
{
	if (unlikely(s->flags & SLAB_DESTROY_BY_RCU)) {
		/*
		 * RCU free overloads the RCU head over the LRU
		 */
		struct rcu_head *head = (void *)&page->lru;

		call_rcu(head, rcu_free_slab);
	} else
		__free_slab(s, page);
}

static void discard_slab(struct kmem_cache *s, struct page *page)
{
1199
	dec_slabs_node(s, page_to_nid(page), page->objects);
C
Christoph Lameter 已提交
1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212
	free_slab(s, page);
}

/*
 * 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)
{
N
Nick Piggin 已提交
1213
	__bit_spin_unlock(PG_locked, &page->flags);
C
Christoph Lameter 已提交
1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226
}

static __always_inline int slab_trylock(struct page *page)
{
	int rc = 1;

	rc = bit_spin_trylock(PG_locked, &page->flags);
	return rc;
}

/*
 * Management of partially allocated slabs
 */
1227 1228
static void add_partial(struct kmem_cache_node *n,
				struct page *page, int tail)
C
Christoph Lameter 已提交
1229
{
C
Christoph Lameter 已提交
1230 1231
	spin_lock(&n->list_lock);
	n->nr_partial++;
1232 1233 1234 1235
	if (tail)
		list_add_tail(&page->lru, &n->partial);
	else
		list_add(&page->lru, &n->partial);
C
Christoph Lameter 已提交
1236 1237 1238
	spin_unlock(&n->list_lock);
}

1239
static void remove_partial(struct kmem_cache *s, struct page *page)
C
Christoph Lameter 已提交
1240 1241 1242 1243 1244 1245 1246 1247 1248 1249
{
	struct kmem_cache_node *n = get_node(s, page_to_nid(page));

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

/*
C
Christoph Lameter 已提交
1250
 * Lock slab and remove from the partial list.
C
Christoph Lameter 已提交
1251
 *
C
Christoph Lameter 已提交
1252
 * Must hold list_lock.
C
Christoph Lameter 已提交
1253
 */
1254 1255
static inline int lock_and_freeze_slab(struct kmem_cache_node *n,
							struct page *page)
C
Christoph Lameter 已提交
1256 1257 1258 1259
{
	if (slab_trylock(page)) {
		list_del(&page->lru);
		n->nr_partial--;
1260
		__SetPageSlubFrozen(page);
C
Christoph Lameter 已提交
1261 1262 1263 1264 1265 1266
		return 1;
	}
	return 0;
}

/*
C
Christoph Lameter 已提交
1267
 * Try to allocate a partial slab from a specific node.
C
Christoph Lameter 已提交
1268 1269 1270 1271 1272 1273 1274 1275
 */
static struct page *get_partial_node(struct kmem_cache_node *n)
{
	struct page *page;

	/*
	 * Racy check. If we mistakenly see no partial slabs then we
	 * just allocate an empty slab. If we mistakenly try to get a
C
Christoph Lameter 已提交
1276 1277
	 * partial slab and there is none available then get_partials()
	 * will return NULL.
C
Christoph Lameter 已提交
1278 1279 1280 1281 1282 1283
	 */
	if (!n || !n->nr_partial)
		return NULL;

	spin_lock(&n->list_lock);
	list_for_each_entry(page, &n->partial, lru)
1284
		if (lock_and_freeze_slab(n, page))
C
Christoph Lameter 已提交
1285 1286 1287 1288 1289 1290 1291 1292
			goto out;
	page = NULL;
out:
	spin_unlock(&n->list_lock);
	return page;
}

/*
C
Christoph Lameter 已提交
1293
 * Get a page from somewhere. Search in increasing NUMA distances.
C
Christoph Lameter 已提交
1294 1295 1296 1297 1298
 */
static struct page *get_any_partial(struct kmem_cache *s, gfp_t flags)
{
#ifdef CONFIG_NUMA
	struct zonelist *zonelist;
1299
	struct zoneref *z;
1300 1301
	struct zone *zone;
	enum zone_type high_zoneidx = gfp_zone(flags);
C
Christoph Lameter 已提交
1302 1303 1304
	struct page *page;

	/*
C
Christoph Lameter 已提交
1305 1306 1307 1308
	 * The defrag ratio allows a configuration of the tradeoffs between
	 * inter node defragmentation and node local allocations. A lower
	 * defrag_ratio increases the tendency to do local allocations
	 * instead of attempting to obtain partial slabs from other nodes.
C
Christoph Lameter 已提交
1309
	 *
C
Christoph Lameter 已提交
1310 1311 1312 1313
	 * If the defrag_ratio is set to 0 then kmalloc() always
	 * returns node local objects. If the ratio is higher then kmalloc()
	 * may return off node objects because partial slabs are obtained
	 * from other nodes and filled up.
C
Christoph Lameter 已提交
1314
	 *
C
Christoph Lameter 已提交
1315
	 * If /sys/kernel/slab/xx/defrag_ratio is set to 100 (which makes
C
Christoph Lameter 已提交
1316 1317 1318 1319 1320
	 * defrag_ratio = 1000) then every (well almost) allocation will
	 * first attempt to defrag slab caches on other nodes. This means
	 * scanning over all nodes to look for partial slabs which may be
	 * expensive if we do it every time we are trying to find a slab
	 * with available objects.
C
Christoph Lameter 已提交
1321
	 */
1322 1323
	if (!s->remote_node_defrag_ratio ||
			get_cycles() % 1024 > s->remote_node_defrag_ratio)
C
Christoph Lameter 已提交
1324 1325
		return NULL;

1326
	zonelist = node_zonelist(slab_node(current->mempolicy), flags);
1327
	for_each_zone_zonelist(zone, z, zonelist, high_zoneidx) {
C
Christoph Lameter 已提交
1328 1329
		struct kmem_cache_node *n;

1330
		n = get_node(s, zone_to_nid(zone));
C
Christoph Lameter 已提交
1331

1332
		if (n && cpuset_zone_allowed_hardwall(zone, flags) &&
1333
				n->nr_partial > n->min_partial) {
C
Christoph Lameter 已提交
1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364
			page = get_partial_node(n);
			if (page)
				return page;
		}
	}
#endif
	return NULL;
}

/*
 * Get a partial page, lock it and return it.
 */
static struct page *get_partial(struct kmem_cache *s, gfp_t flags, int node)
{
	struct page *page;
	int searchnode = (node == -1) ? numa_node_id() : node;

	page = get_partial_node(get_node(s, searchnode));
	if (page || (flags & __GFP_THISNODE))
		return page;

	return get_any_partial(s, flags);
}

/*
 * Move a page back to the lists.
 *
 * Must be called with the slab lock held.
 *
 * On exit the slab lock will have been dropped.
 */
1365
static void unfreeze_slab(struct kmem_cache *s, struct page *page, int tail)
C
Christoph Lameter 已提交
1366
{
C
Christoph Lameter 已提交
1367
	struct kmem_cache_node *n = get_node(s, page_to_nid(page));
1368
	struct kmem_cache_cpu *c = get_cpu_slab(s, smp_processor_id());
C
Christoph Lameter 已提交
1369

1370
	__ClearPageSlubFrozen(page);
C
Christoph Lameter 已提交
1371
	if (page->inuse) {
C
Christoph Lameter 已提交
1372

1373
		if (page->freelist) {
1374
			add_partial(n, page, tail);
1375 1376 1377
			stat(c, tail ? DEACTIVATE_TO_TAIL : DEACTIVATE_TO_HEAD);
		} else {
			stat(c, DEACTIVATE_FULL);
1378 1379
			if (SLABDEBUG && PageSlubDebug(page) &&
						(s->flags & SLAB_STORE_USER))
1380 1381
				add_full(n, page);
		}
C
Christoph Lameter 已提交
1382 1383
		slab_unlock(page);
	} else {
1384
		stat(c, DEACTIVATE_EMPTY);
1385
		if (n->nr_partial < n->min_partial) {
C
Christoph Lameter 已提交
1386
			/*
C
Christoph Lameter 已提交
1387 1388 1389
			 * Adding an empty slab to the partial slabs in order
			 * to avoid page allocator overhead. This slab needs
			 * to come after the other slabs with objects in
C
Christoph Lameter 已提交
1390 1391 1392
			 * so that the others get filled first. That way the
			 * size of the partial list stays small.
			 *
1393 1394
			 * kmem_cache_shrink can reclaim any empty slabs from
			 * the partial list.
C
Christoph Lameter 已提交
1395
			 */
1396
			add_partial(n, page, 1);
C
Christoph Lameter 已提交
1397 1398 1399
			slab_unlock(page);
		} else {
			slab_unlock(page);
1400
			stat(get_cpu_slab(s, raw_smp_processor_id()), FREE_SLAB);
C
Christoph Lameter 已提交
1401 1402
			discard_slab(s, page);
		}
C
Christoph Lameter 已提交
1403 1404 1405 1406 1407 1408
	}
}

/*
 * Remove the cpu slab
 */
1409
static void deactivate_slab(struct kmem_cache *s, struct kmem_cache_cpu *c)
C
Christoph Lameter 已提交
1410
{
1411
	struct page *page = c->page;
1412
	int tail = 1;
1413

1414
	if (page->freelist)
1415
		stat(c, DEACTIVATE_REMOTE_FREES);
1416
	/*
C
Christoph Lameter 已提交
1417
	 * Merge cpu freelist into slab freelist. Typically we get here
1418 1419 1420
	 * because both freelists are empty. So this is unlikely
	 * to occur.
	 */
1421
	while (unlikely(c->freelist)) {
1422 1423
		void **object;

1424 1425
		tail = 0;	/* Hot objects. Put the slab first */

1426
		/* Retrieve object from cpu_freelist */
1427
		object = c->freelist;
1428
		c->freelist = c->freelist[c->offset];
1429 1430

		/* And put onto the regular freelist */
1431
		object[c->offset] = page->freelist;
1432 1433 1434
		page->freelist = object;
		page->inuse--;
	}
1435
	c->page = NULL;
1436
	unfreeze_slab(s, page, tail);
C
Christoph Lameter 已提交
1437 1438
}

1439
static inline void flush_slab(struct kmem_cache *s, struct kmem_cache_cpu *c)
C
Christoph Lameter 已提交
1440
{
1441
	stat(c, CPUSLAB_FLUSH);
1442 1443
	slab_lock(c->page);
	deactivate_slab(s, c);
C
Christoph Lameter 已提交
1444 1445 1446 1447
}

/*
 * Flush cpu slab.
C
Christoph Lameter 已提交
1448
 *
C
Christoph Lameter 已提交
1449 1450
 * Called from IPI handler with interrupts disabled.
 */
1451
static inline void __flush_cpu_slab(struct kmem_cache *s, int cpu)
C
Christoph Lameter 已提交
1452
{
1453
	struct kmem_cache_cpu *c = get_cpu_slab(s, cpu);
C
Christoph Lameter 已提交
1454

1455 1456
	if (likely(c && c->page))
		flush_slab(s, c);
C
Christoph Lameter 已提交
1457 1458 1459 1460 1461 1462
}

static void flush_cpu_slab(void *d)
{
	struct kmem_cache *s = d;

1463
	__flush_cpu_slab(s, smp_processor_id());
C
Christoph Lameter 已提交
1464 1465 1466 1467
}

static void flush_all(struct kmem_cache *s)
{
1468
	on_each_cpu(flush_cpu_slab, s, 1);
C
Christoph Lameter 已提交
1469 1470
}

1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483
/*
 * Check if the objects in a per cpu structure fit numa
 * locality expectations.
 */
static inline int node_match(struct kmem_cache_cpu *c, int node)
{
#ifdef CONFIG_NUMA
	if (node != -1 && c->node != node)
		return 0;
#endif
	return 1;
}

C
Christoph Lameter 已提交
1484
/*
1485 1486 1487 1488
 * Slow path. The lockless freelist is empty or we need to perform
 * debugging duties.
 *
 * Interrupts are disabled.
C
Christoph Lameter 已提交
1489
 *
1490 1491 1492
 * Processing is still very fast if new objects have been freed to the
 * regular freelist. In that case we simply take over the regular freelist
 * as the lockless freelist and zap the regular freelist.
C
Christoph Lameter 已提交
1493
 *
1494 1495 1496
 * If that is not working then we fall back to the partial lists. We take the
 * first element of the freelist as the object to allocate now and move the
 * rest of the freelist to the lockless freelist.
C
Christoph Lameter 已提交
1497
 *
1498
 * And if we were unable to get a new slab from the partial slab lists then
C
Christoph Lameter 已提交
1499 1500
 * we need to allocate a new slab. This is the slowest path since it involves
 * a call to the page allocator and the setup of a new slab.
C
Christoph Lameter 已提交
1501
 */
1502
static void *__slab_alloc(struct kmem_cache *s,
1503
		gfp_t gfpflags, int node, void *addr, struct kmem_cache_cpu *c)
C
Christoph Lameter 已提交
1504 1505
{
	void **object;
1506
	struct page *new;
C
Christoph Lameter 已提交
1507

1508 1509 1510
	/* We handle __GFP_ZERO in the caller */
	gfpflags &= ~__GFP_ZERO;

1511
	if (!c->page)
C
Christoph Lameter 已提交
1512 1513
		goto new_slab;

1514 1515
	slab_lock(c->page);
	if (unlikely(!node_match(c, node)))
C
Christoph Lameter 已提交
1516
		goto another_slab;
C
Christoph Lameter 已提交
1517

1518
	stat(c, ALLOC_REFILL);
C
Christoph Lameter 已提交
1519

1520
load_freelist:
1521
	object = c->page->freelist;
1522
	if (unlikely(!object))
C
Christoph Lameter 已提交
1523
		goto another_slab;
1524
	if (unlikely(SLABDEBUG && PageSlubDebug(c->page)))
C
Christoph Lameter 已提交
1525 1526
		goto debug;

1527
	c->freelist = object[c->offset];
1528
	c->page->inuse = c->page->objects;
1529
	c->page->freelist = NULL;
1530
	c->node = page_to_nid(c->page);
1531
unlock_out:
1532
	slab_unlock(c->page);
1533
	stat(c, ALLOC_SLOWPATH);
C
Christoph Lameter 已提交
1534 1535 1536
	return object;

another_slab:
1537
	deactivate_slab(s, c);
C
Christoph Lameter 已提交
1538 1539

new_slab:
1540 1541 1542
	new = get_partial(s, gfpflags, node);
	if (new) {
		c->page = new;
1543
		stat(c, ALLOC_FROM_PARTIAL);
1544
		goto load_freelist;
C
Christoph Lameter 已提交
1545 1546
	}

1547 1548 1549
	if (gfpflags & __GFP_WAIT)
		local_irq_enable();

1550
	new = new_slab(s, gfpflags, node);
1551 1552 1553 1554

	if (gfpflags & __GFP_WAIT)
		local_irq_disable();

1555 1556
	if (new) {
		c = get_cpu_slab(s, smp_processor_id());
1557
		stat(c, ALLOC_SLAB);
1558
		if (c->page)
1559 1560
			flush_slab(s, c);
		slab_lock(new);
1561
		__SetPageSlubFrozen(new);
1562
		c->page = new;
1563
		goto load_freelist;
C
Christoph Lameter 已提交
1564
	}
1565
	return NULL;
C
Christoph Lameter 已提交
1566
debug:
1567
	if (!alloc_debug_processing(s, c->page, object, addr))
C
Christoph Lameter 已提交
1568
		goto another_slab;
1569

1570
	c->page->inuse++;
1571
	c->page->freelist = object[c->offset];
1572
	c->node = -1;
1573
	goto unlock_out;
1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585
}

/*
 * Inlined fastpath so that allocation functions (kmalloc, kmem_cache_alloc)
 * have the fastpath folded into their functions. So no function call
 * overhead for requests that can be satisfied on the fastpath.
 *
 * The fastpath works by first checking if the lockless freelist can be used.
 * If not then __slab_alloc is called for slow processing.
 *
 * Otherwise we can simply pick the next object from the lockless free list.
 */
P
Pekka Enberg 已提交
1586
static __always_inline void *slab_alloc(struct kmem_cache *s,
1587
		gfp_t gfpflags, int node, void *addr)
1588 1589
{
	void **object;
1590
	struct kmem_cache_cpu *c;
1591
	unsigned long flags;
1592
	unsigned int objsize;
1593

1594
	local_irq_save(flags);
1595
	c = get_cpu_slab(s, smp_processor_id());
1596
	objsize = c->objsize;
1597
	if (unlikely(!c->freelist || !node_match(c, node)))
1598

1599
		object = __slab_alloc(s, gfpflags, node, addr, c);
1600 1601

	else {
1602
		object = c->freelist;
1603
		c->freelist = object[c->offset];
1604
		stat(c, ALLOC_FASTPATH);
1605 1606
	}
	local_irq_restore(flags);
1607 1608

	if (unlikely((gfpflags & __GFP_ZERO) && object))
1609
		memset(object, 0, objsize);
1610

1611
	return object;
C
Christoph Lameter 已提交
1612 1613 1614 1615
}

void *kmem_cache_alloc(struct kmem_cache *s, gfp_t gfpflags)
{
1616
	return slab_alloc(s, gfpflags, -1, __builtin_return_address(0));
C
Christoph Lameter 已提交
1617 1618 1619 1620 1621 1622
}
EXPORT_SYMBOL(kmem_cache_alloc);

#ifdef CONFIG_NUMA
void *kmem_cache_alloc_node(struct kmem_cache *s, gfp_t gfpflags, int node)
{
1623
	return slab_alloc(s, gfpflags, node, __builtin_return_address(0));
C
Christoph Lameter 已提交
1624 1625 1626 1627 1628
}
EXPORT_SYMBOL(kmem_cache_alloc_node);
#endif

/*
1629 1630
 * Slow patch handling. This may still be called frequently since objects
 * have a longer lifetime than the cpu slabs in most processing loads.
C
Christoph Lameter 已提交
1631
 *
1632 1633 1634
 * So we still attempt to reduce cache line usage. Just take the slab
 * lock and free the item. If there is no additional partial page
 * handling required then we can return immediately.
C
Christoph Lameter 已提交
1635
 */
1636
static void __slab_free(struct kmem_cache *s, struct page *page,
1637
				void *x, void *addr, unsigned int offset)
C
Christoph Lameter 已提交
1638 1639 1640
{
	void *prior;
	void **object = (void *)x;
1641
	struct kmem_cache_cpu *c;
C
Christoph Lameter 已提交
1642

1643 1644
	c = get_cpu_slab(s, raw_smp_processor_id());
	stat(c, FREE_SLOWPATH);
C
Christoph Lameter 已提交
1645 1646
	slab_lock(page);

1647
	if (unlikely(SLABDEBUG && PageSlubDebug(page)))
C
Christoph Lameter 已提交
1648
		goto debug;
C
Christoph Lameter 已提交
1649

C
Christoph Lameter 已提交
1650
checks_ok:
1651
	prior = object[offset] = page->freelist;
C
Christoph Lameter 已提交
1652 1653 1654
	page->freelist = object;
	page->inuse--;

1655
	if (unlikely(PageSlubFrozen(page))) {
1656
		stat(c, FREE_FROZEN);
C
Christoph Lameter 已提交
1657
		goto out_unlock;
1658
	}
C
Christoph Lameter 已提交
1659 1660 1661 1662 1663

	if (unlikely(!page->inuse))
		goto slab_empty;

	/*
C
Christoph Lameter 已提交
1664
	 * Objects left in the slab. If it was not on the partial list before
C
Christoph Lameter 已提交
1665 1666
	 * then add it.
	 */
1667
	if (unlikely(!prior)) {
1668
		add_partial(get_node(s, page_to_nid(page)), page, 1);
1669 1670
		stat(c, FREE_ADD_PARTIAL);
	}
C
Christoph Lameter 已提交
1671 1672 1673 1674 1675 1676

out_unlock:
	slab_unlock(page);
	return;

slab_empty:
1677
	if (prior) {
C
Christoph Lameter 已提交
1678
		/*
C
Christoph Lameter 已提交
1679
		 * Slab still on the partial list.
C
Christoph Lameter 已提交
1680 1681
		 */
		remove_partial(s, page);
1682 1683
		stat(c, FREE_REMOVE_PARTIAL);
	}
C
Christoph Lameter 已提交
1684
	slab_unlock(page);
1685
	stat(c, FREE_SLAB);
C
Christoph Lameter 已提交
1686 1687 1688 1689
	discard_slab(s, page);
	return;

debug:
C
Christoph Lameter 已提交
1690
	if (!free_debug_processing(s, page, x, addr))
C
Christoph Lameter 已提交
1691 1692
		goto out_unlock;
	goto checks_ok;
C
Christoph Lameter 已提交
1693 1694
}

1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705
/*
 * Fastpath with forced inlining to produce a kfree and kmem_cache_free that
 * can perform fastpath freeing without additional function calls.
 *
 * The fastpath is only possible if we are freeing to the current cpu slab
 * of this processor. This typically the case if we have just allocated
 * the item before.
 *
 * If fastpath is not possible then fall back to __slab_free where we deal
 * with all sorts of special processing.
 */
P
Pekka Enberg 已提交
1706
static __always_inline void slab_free(struct kmem_cache *s,
1707 1708 1709
			struct page *page, void *x, void *addr)
{
	void **object = (void *)x;
1710
	struct kmem_cache_cpu *c;
1711 1712
	unsigned long flags;

1713
	local_irq_save(flags);
1714
	c = get_cpu_slab(s, smp_processor_id());
1715
	debug_check_no_locks_freed(object, c->objsize);
1716 1717
	if (!(s->flags & SLAB_DEBUG_OBJECTS))
		debug_check_no_obj_freed(object, s->objsize);
1718
	if (likely(page == c->page && c->node >= 0)) {
1719
		object[c->offset] = c->freelist;
1720
		c->freelist = object;
1721
		stat(c, FREE_FASTPATH);
1722
	} else
1723
		__slab_free(s, page, x, addr, c->offset);
1724 1725 1726 1727

	local_irq_restore(flags);
}

C
Christoph Lameter 已提交
1728 1729
void kmem_cache_free(struct kmem_cache *s, void *x)
{
C
Christoph Lameter 已提交
1730
	struct page *page;
C
Christoph Lameter 已提交
1731

1732
	page = virt_to_head_page(x);
C
Christoph Lameter 已提交
1733

C
Christoph Lameter 已提交
1734
	slab_free(s, page, x, __builtin_return_address(0));
C
Christoph Lameter 已提交
1735 1736 1737 1738 1739 1740
}
EXPORT_SYMBOL(kmem_cache_free);

/* Figure out on which slab object the object resides */
static struct page *get_object_page(const void *x)
{
1741
	struct page *page = virt_to_head_page(x);
C
Christoph Lameter 已提交
1742 1743 1744 1745 1746 1747 1748 1749

	if (!PageSlab(page))
		return NULL;

	return page;
}

/*
C
Christoph Lameter 已提交
1750 1751 1752 1753
 * Object placement in a slab is made very easy because we always start at
 * offset 0. If we tune the size of the object to the alignment then we can
 * get the required alignment by putting one properly sized object after
 * another.
C
Christoph Lameter 已提交
1754 1755 1756 1757
 *
 * Notice that the allocation order determines the sizes of the per cpu
 * caches. Each processor has always one slab available for allocations.
 * Increasing the allocation order reduces the number of times that slabs
C
Christoph Lameter 已提交
1758
 * must be moved on and off the partial lists and is therefore a factor in
C
Christoph Lameter 已提交
1759 1760 1761 1762 1763 1764 1765 1766 1767 1768
 * locking overhead.
 */

/*
 * Mininum / Maximum order of slab pages. This influences locking overhead
 * and slab fragmentation. A higher order reduces the number of partial slabs
 * and increases the number of allocations possible without having to
 * take the list_lock.
 */
static int slub_min_order;
1769
static int slub_max_order = PAGE_ALLOC_COSTLY_ORDER;
1770
static int slub_min_objects;
C
Christoph Lameter 已提交
1771 1772 1773

/*
 * Merge control. If this is set then no merging of slab caches will occur.
C
Christoph Lameter 已提交
1774
 * (Could be removed. This was introduced to pacify the merge skeptics.)
C
Christoph Lameter 已提交
1775 1776 1777 1778 1779 1780
 */
static int slub_nomerge;

/*
 * Calculate the order of allocation given an slab object size.
 *
C
Christoph Lameter 已提交
1781 1782 1783 1784
 * The order of allocation has significant impact on performance and other
 * system components. Generally order 0 allocations should be preferred since
 * order 0 does not cause fragmentation in the page allocator. Larger objects
 * be problematic to put into order 0 slabs because there may be too much
C
Christoph Lameter 已提交
1785
 * unused space left. We go to a higher order if more than 1/16th of the slab
C
Christoph Lameter 已提交
1786 1787 1788 1789 1790 1791
 * would be wasted.
 *
 * In order to reach satisfactory performance we must ensure that a minimum
 * number of objects is in one slab. Otherwise we may generate too much
 * activity on the partial lists which requires taking the list_lock. This is
 * less a concern for large slabs though which are rarely used.
C
Christoph Lameter 已提交
1792
 *
C
Christoph Lameter 已提交
1793 1794 1795 1796
 * slub_max_order specifies the order where we begin to stop considering the
 * number of objects in a slab as critical. If we reach slub_max_order then
 * we try to keep the page order as low as possible. So we accept more waste
 * of space in favor of a small page order.
C
Christoph Lameter 已提交
1797
 *
C
Christoph Lameter 已提交
1798 1799 1800 1801
 * Higher order allocations also allow the placement of more objects in a
 * slab and thereby reduce object handling overhead. If the user has
 * requested a higher mininum order then we start with that one instead of
 * the smallest order which will fit the object.
C
Christoph Lameter 已提交
1802
 */
1803 1804
static inline int slab_order(int size, int min_objects,
				int max_order, int fract_leftover)
C
Christoph Lameter 已提交
1805 1806 1807
{
	int order;
	int rem;
1808
	int min_order = slub_min_order;
C
Christoph Lameter 已提交
1809

1810 1811 1812
	if ((PAGE_SIZE << min_order) / size > 65535)
		return get_order(size * 65535) - 1;

1813
	for (order = max(min_order,
1814 1815
				fls(min_objects * size - 1) - PAGE_SHIFT);
			order <= max_order; order++) {
C
Christoph Lameter 已提交
1816

1817
		unsigned long slab_size = PAGE_SIZE << order;
C
Christoph Lameter 已提交
1818

1819
		if (slab_size < min_objects * size)
C
Christoph Lameter 已提交
1820 1821 1822 1823
			continue;

		rem = slab_size % size;

1824
		if (rem <= slab_size / fract_leftover)
C
Christoph Lameter 已提交
1825 1826 1827
			break;

	}
C
Christoph Lameter 已提交
1828

C
Christoph Lameter 已提交
1829 1830 1831
	return order;
}

1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846
static inline int calculate_order(int size)
{
	int order;
	int min_objects;
	int fraction;

	/*
	 * Attempt to find best configuration for a slab. This
	 * works by first attempting to generate a layout with
	 * the best configuration and backing off gradually.
	 *
	 * First we reduce the acceptable waste in a slab. Then
	 * we reduce the minimum objects required in a slab.
	 */
	min_objects = slub_min_objects;
1847 1848
	if (!min_objects)
		min_objects = 4 * (fls(nr_cpu_ids) + 1);
1849
	while (min_objects > 1) {
C
Christoph Lameter 已提交
1850
		fraction = 16;
1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877
		while (fraction >= 4) {
			order = slab_order(size, min_objects,
						slub_max_order, fraction);
			if (order <= slub_max_order)
				return order;
			fraction /= 2;
		}
		min_objects /= 2;
	}

	/*
	 * We were unable to place multiple objects in a slab. Now
	 * lets see if we can place a single object there.
	 */
	order = slab_order(size, 1, slub_max_order, 1);
	if (order <= slub_max_order)
		return order;

	/*
	 * Doh this slab cannot be placed using slub_max_order.
	 */
	order = slab_order(size, 1, MAX_ORDER, 1);
	if (order <= MAX_ORDER)
		return order;
	return -ENOSYS;
}

C
Christoph Lameter 已提交
1878
/*
C
Christoph Lameter 已提交
1879
 * Figure out what the alignment of the objects will be.
C
Christoph Lameter 已提交
1880 1881 1882 1883 1884
 */
static unsigned long calculate_alignment(unsigned long flags,
		unsigned long align, unsigned long size)
{
	/*
C
Christoph Lameter 已提交
1885 1886
	 * If the user wants hardware cache aligned objects then follow that
	 * suggestion if the object is sufficiently large.
C
Christoph Lameter 已提交
1887
	 *
C
Christoph Lameter 已提交
1888 1889
	 * The hardware cache alignment cannot override the specified
	 * alignment though. If that is greater then use it.
C
Christoph Lameter 已提交
1890
	 */
1891 1892 1893 1894 1895 1896
	if (flags & SLAB_HWCACHE_ALIGN) {
		unsigned long ralign = cache_line_size();
		while (size <= ralign / 2)
			ralign /= 2;
		align = max(align, ralign);
	}
C
Christoph Lameter 已提交
1897 1898

	if (align < ARCH_SLAB_MINALIGN)
1899
		align = ARCH_SLAB_MINALIGN;
C
Christoph Lameter 已提交
1900 1901 1902 1903

	return ALIGN(align, sizeof(void *));
}

1904 1905 1906 1907
static void init_kmem_cache_cpu(struct kmem_cache *s,
			struct kmem_cache_cpu *c)
{
	c->page = NULL;
1908
	c->freelist = NULL;
1909
	c->node = 0;
1910 1911
	c->offset = s->offset / sizeof(void *);
	c->objsize = s->objsize;
P
Pekka Enberg 已提交
1912 1913 1914
#ifdef CONFIG_SLUB_STATS
	memset(c->stat, 0, NR_SLUB_STAT_ITEMS * sizeof(unsigned));
#endif
1915 1916
}

1917 1918
static void
init_kmem_cache_node(struct kmem_cache_node *n, struct kmem_cache *s)
C
Christoph Lameter 已提交
1919 1920
{
	n->nr_partial = 0;
1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931

	/*
	 * The larger the object size is, the more pages we want on the partial
	 * list to avoid pounding the page allocator excessively.
	 */
	n->min_partial = ilog2(s->size);
	if (n->min_partial < MIN_PARTIAL)
		n->min_partial = MIN_PARTIAL;
	else if (n->min_partial > MAX_PARTIAL)
		n->min_partial = MAX_PARTIAL;

C
Christoph Lameter 已提交
1932 1933
	spin_lock_init(&n->list_lock);
	INIT_LIST_HEAD(&n->partial);
1934
#ifdef CONFIG_SLUB_DEBUG
1935
	atomic_long_set(&n->nr_slabs, 0);
1936
	atomic_long_set(&n->total_objects, 0);
1937
	INIT_LIST_HEAD(&n->full);
1938
#endif
C
Christoph Lameter 已提交
1939 1940
}

1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065
#ifdef CONFIG_SMP
/*
 * Per cpu array for per cpu structures.
 *
 * The per cpu array places all kmem_cache_cpu structures from one processor
 * close together meaning that it becomes possible that multiple per cpu
 * structures are contained in one cacheline. This may be particularly
 * beneficial for the kmalloc caches.
 *
 * A desktop system typically has around 60-80 slabs. With 100 here we are
 * likely able to get per cpu structures for all caches from the array defined
 * here. We must be able to cover all kmalloc caches during bootstrap.
 *
 * If the per cpu array is exhausted then fall back to kmalloc
 * of individual cachelines. No sharing is possible then.
 */
#define NR_KMEM_CACHE_CPU 100

static DEFINE_PER_CPU(struct kmem_cache_cpu,
				kmem_cache_cpu)[NR_KMEM_CACHE_CPU];

static DEFINE_PER_CPU(struct kmem_cache_cpu *, kmem_cache_cpu_free);
static cpumask_t kmem_cach_cpu_free_init_once = CPU_MASK_NONE;

static struct kmem_cache_cpu *alloc_kmem_cache_cpu(struct kmem_cache *s,
							int cpu, gfp_t flags)
{
	struct kmem_cache_cpu *c = per_cpu(kmem_cache_cpu_free, cpu);

	if (c)
		per_cpu(kmem_cache_cpu_free, cpu) =
				(void *)c->freelist;
	else {
		/* Table overflow: So allocate ourselves */
		c = kmalloc_node(
			ALIGN(sizeof(struct kmem_cache_cpu), cache_line_size()),
			flags, cpu_to_node(cpu));
		if (!c)
			return NULL;
	}

	init_kmem_cache_cpu(s, c);
	return c;
}

static void free_kmem_cache_cpu(struct kmem_cache_cpu *c, int cpu)
{
	if (c < per_cpu(kmem_cache_cpu, cpu) ||
			c > per_cpu(kmem_cache_cpu, cpu) + NR_KMEM_CACHE_CPU) {
		kfree(c);
		return;
	}
	c->freelist = (void *)per_cpu(kmem_cache_cpu_free, cpu);
	per_cpu(kmem_cache_cpu_free, cpu) = c;
}

static void free_kmem_cache_cpus(struct kmem_cache *s)
{
	int cpu;

	for_each_online_cpu(cpu) {
		struct kmem_cache_cpu *c = get_cpu_slab(s, cpu);

		if (c) {
			s->cpu_slab[cpu] = NULL;
			free_kmem_cache_cpu(c, cpu);
		}
	}
}

static int alloc_kmem_cache_cpus(struct kmem_cache *s, gfp_t flags)
{
	int cpu;

	for_each_online_cpu(cpu) {
		struct kmem_cache_cpu *c = get_cpu_slab(s, cpu);

		if (c)
			continue;

		c = alloc_kmem_cache_cpu(s, cpu, flags);
		if (!c) {
			free_kmem_cache_cpus(s);
			return 0;
		}
		s->cpu_slab[cpu] = c;
	}
	return 1;
}

/*
 * Initialize the per cpu array.
 */
static void init_alloc_cpu_cpu(int cpu)
{
	int i;

	if (cpu_isset(cpu, kmem_cach_cpu_free_init_once))
		return;

	for (i = NR_KMEM_CACHE_CPU - 1; i >= 0; i--)
		free_kmem_cache_cpu(&per_cpu(kmem_cache_cpu, cpu)[i], cpu);

	cpu_set(cpu, kmem_cach_cpu_free_init_once);
}

static void __init init_alloc_cpu(void)
{
	int cpu;

	for_each_online_cpu(cpu)
		init_alloc_cpu_cpu(cpu);
  }

#else
static inline void free_kmem_cache_cpus(struct kmem_cache *s) {}
static inline void init_alloc_cpu(void) {}

static inline int alloc_kmem_cache_cpus(struct kmem_cache *s, gfp_t flags)
{
	init_kmem_cache_cpu(s, &s->cpu_slab);
	return 1;
}
#endif

C
Christoph Lameter 已提交
2066 2067 2068 2069 2070 2071 2072
#ifdef CONFIG_NUMA
/*
 * No kmalloc_node yet so do it by hand. We know that this is the first
 * slab on the node for this slabcache. There are no concurrent accesses
 * possible.
 *
 * Note that this function only works on the kmalloc_node_cache
2073 2074
 * when allocating for the kmalloc_node_cache. This is used for bootstrapping
 * memory on a fresh node that has no slab structures yet.
C
Christoph Lameter 已提交
2075
 */
2076 2077
static struct kmem_cache_node *early_kmem_cache_node_alloc(gfp_t gfpflags,
							   int node)
C
Christoph Lameter 已提交
2078 2079 2080
{
	struct page *page;
	struct kmem_cache_node *n;
R
root 已提交
2081
	unsigned long flags;
C
Christoph Lameter 已提交
2082 2083 2084

	BUG_ON(kmalloc_caches->size < sizeof(struct kmem_cache_node));

2085
	page = new_slab(kmalloc_caches, gfpflags, node);
C
Christoph Lameter 已提交
2086 2087

	BUG_ON(!page);
2088 2089 2090 2091 2092 2093 2094
	if (page_to_nid(page) != node) {
		printk(KERN_ERR "SLUB: Unable to allocate memory from "
				"node %d\n", node);
		printk(KERN_ERR "SLUB: Allocating a useless per node structure "
				"in order to be able to continue\n");
	}

C
Christoph Lameter 已提交
2095 2096 2097 2098 2099
	n = page->freelist;
	BUG_ON(!n);
	page->freelist = get_freepointer(kmalloc_caches, n);
	page->inuse++;
	kmalloc_caches->node[node] = n;
2100
#ifdef CONFIG_SLUB_DEBUG
2101 2102
	init_object(kmalloc_caches, n, 1);
	init_tracking(kmalloc_caches, n);
2103
#endif
2104
	init_kmem_cache_node(n, kmalloc_caches);
2105
	inc_slabs_node(kmalloc_caches, node, page->objects);
C
Christoph Lameter 已提交
2106

R
root 已提交
2107 2108 2109 2110 2111 2112
	/*
	 * lockdep requires consistent irq usage for each lock
	 * so even though there cannot be a race this early in
	 * the boot sequence, we still disable irqs.
	 */
	local_irq_save(flags);
2113
	add_partial(n, page, 0);
R
root 已提交
2114
	local_irq_restore(flags);
C
Christoph Lameter 已提交
2115 2116 2117 2118 2119 2120 2121
	return n;
}

static void free_kmem_cache_nodes(struct kmem_cache *s)
{
	int node;

C
Christoph Lameter 已提交
2122
	for_each_node_state(node, N_NORMAL_MEMORY) {
C
Christoph Lameter 已提交
2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139
		struct kmem_cache_node *n = s->node[node];
		if (n && n != &s->local_node)
			kmem_cache_free(kmalloc_caches, n);
		s->node[node] = NULL;
	}
}

static int init_kmem_cache_nodes(struct kmem_cache *s, gfp_t gfpflags)
{
	int node;
	int local_node;

	if (slab_state >= UP)
		local_node = page_to_nid(virt_to_page(s));
	else
		local_node = 0;

C
Christoph Lameter 已提交
2140
	for_each_node_state(node, N_NORMAL_MEMORY) {
C
Christoph Lameter 已提交
2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160
		struct kmem_cache_node *n;

		if (local_node == node)
			n = &s->local_node;
		else {
			if (slab_state == DOWN) {
				n = early_kmem_cache_node_alloc(gfpflags,
								node);
				continue;
			}
			n = kmem_cache_alloc_node(kmalloc_caches,
							gfpflags, node);

			if (!n) {
				free_kmem_cache_nodes(s);
				return 0;
			}

		}
		s->node[node] = n;
2161
		init_kmem_cache_node(n, s);
C
Christoph Lameter 已提交
2162 2163 2164 2165 2166 2167 2168 2169 2170 2171
	}
	return 1;
}
#else
static void free_kmem_cache_nodes(struct kmem_cache *s)
{
}

static int init_kmem_cache_nodes(struct kmem_cache *s, gfp_t gfpflags)
{
2172
	init_kmem_cache_node(&s->local_node, s);
C
Christoph Lameter 已提交
2173 2174 2175 2176 2177 2178 2179 2180
	return 1;
}
#endif

/*
 * calculate_sizes() determines the order and the distribution of data within
 * a slab object.
 */
2181
static int calculate_sizes(struct kmem_cache *s, int forced_order)
C
Christoph Lameter 已提交
2182 2183 2184 2185
{
	unsigned long flags = s->flags;
	unsigned long size = s->objsize;
	unsigned long align = s->align;
2186
	int order;
C
Christoph Lameter 已提交
2187

2188 2189 2190 2191 2192 2193 2194 2195
	/*
	 * Round up object size to the next word boundary. We can only
	 * place the free pointer at word boundaries and this determines
	 * the possible location of the free pointer.
	 */
	size = ALIGN(size, sizeof(void *));

#ifdef CONFIG_SLUB_DEBUG
C
Christoph Lameter 已提交
2196 2197 2198 2199 2200 2201
	/*
	 * Determine if we can poison the object itself. If the user of
	 * the slab may touch the object after free or before allocation
	 * then we should never poison the object itself.
	 */
	if ((flags & SLAB_POISON) && !(flags & SLAB_DESTROY_BY_RCU) &&
2202
			!s->ctor)
C
Christoph Lameter 已提交
2203 2204 2205 2206 2207 2208
		s->flags |= __OBJECT_POISON;
	else
		s->flags &= ~__OBJECT_POISON;


	/*
C
Christoph Lameter 已提交
2209
	 * If we are Redzoning then check if there is some space between the
C
Christoph Lameter 已提交
2210
	 * end of the object and the free pointer. If not then add an
C
Christoph Lameter 已提交
2211
	 * additional word to have some bytes to store Redzone information.
C
Christoph Lameter 已提交
2212 2213 2214
	 */
	if ((flags & SLAB_RED_ZONE) && size == s->objsize)
		size += sizeof(void *);
C
Christoph Lameter 已提交
2215
#endif
C
Christoph Lameter 已提交
2216 2217

	/*
C
Christoph Lameter 已提交
2218 2219
	 * With that we have determined the number of bytes in actual use
	 * by the object. This is the potential offset to the free pointer.
C
Christoph Lameter 已提交
2220 2221 2222 2223
	 */
	s->inuse = size;

	if (((flags & (SLAB_DESTROY_BY_RCU | SLAB_POISON)) ||
2224
		s->ctor)) {
C
Christoph Lameter 已提交
2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236
		/*
		 * Relocate free pointer after the object if it is not
		 * permitted to overwrite the first word of the object on
		 * kmem_cache_free.
		 *
		 * This is the case if we do RCU, have a constructor or
		 * destructor or are poisoning the objects.
		 */
		s->offset = size;
		size += sizeof(void *);
	}

2237
#ifdef CONFIG_SLUB_DEBUG
C
Christoph Lameter 已提交
2238 2239 2240 2241 2242 2243 2244
	if (flags & SLAB_STORE_USER)
		/*
		 * Need to store information about allocs and frees after
		 * the object.
		 */
		size += 2 * sizeof(struct track);

2245
	if (flags & SLAB_RED_ZONE)
C
Christoph Lameter 已提交
2246 2247 2248 2249 2250 2251 2252 2253
		/*
		 * Add some empty padding so that we can catch
		 * overwrites from earlier objects rather than let
		 * tracking information or the free pointer be
		 * corrupted if an user writes before the start
		 * of the object.
		 */
		size += sizeof(void *);
C
Christoph Lameter 已提交
2254
#endif
C
Christoph Lameter 已提交
2255

C
Christoph Lameter 已提交
2256 2257
	/*
	 * Determine the alignment based on various parameters that the
2258 2259
	 * user specified and the dynamic determination of cache line size
	 * on bootup.
C
Christoph Lameter 已提交
2260 2261 2262 2263 2264 2265 2266 2267 2268 2269
	 */
	align = calculate_alignment(flags, align, s->objsize);

	/*
	 * SLUB stores one object immediately after another beginning from
	 * offset 0. In order to align the objects we have to simply size
	 * each object to conform to the alignment.
	 */
	size = ALIGN(size, align);
	s->size = size;
2270 2271 2272 2273
	if (forced_order >= 0)
		order = forced_order;
	else
		order = calculate_order(size);
C
Christoph Lameter 已提交
2274

2275
	if (order < 0)
C
Christoph Lameter 已提交
2276 2277
		return 0;

2278
	s->allocflags = 0;
2279
	if (order)
2280 2281 2282 2283 2284 2285 2286 2287
		s->allocflags |= __GFP_COMP;

	if (s->flags & SLAB_CACHE_DMA)
		s->allocflags |= SLUB_DMA;

	if (s->flags & SLAB_RECLAIM_ACCOUNT)
		s->allocflags |= __GFP_RECLAIMABLE;

C
Christoph Lameter 已提交
2288 2289 2290
	/*
	 * Determine the number of objects per slab
	 */
2291
	s->oo = oo_make(order, size);
2292
	s->min = oo_make(get_order(size), size);
2293 2294
	if (oo_objects(s->oo) > oo_objects(s->max))
		s->max = s->oo;
C
Christoph Lameter 已提交
2295

2296
	return !!oo_objects(s->oo);
C
Christoph Lameter 已提交
2297 2298 2299 2300 2301 2302

}

static int kmem_cache_open(struct kmem_cache *s, gfp_t gfpflags,
		const char *name, size_t size,
		size_t align, unsigned long flags,
2303
		void (*ctor)(void *))
C
Christoph Lameter 已提交
2304 2305 2306 2307 2308 2309
{
	memset(s, 0, kmem_size);
	s->name = name;
	s->ctor = ctor;
	s->objsize = size;
	s->align = align;
2310
	s->flags = kmem_cache_flags(size, flags, name, ctor);
C
Christoph Lameter 已提交
2311

2312
	if (!calculate_sizes(s, -1))
C
Christoph Lameter 已提交
2313 2314 2315 2316
		goto error;

	s->refcount = 1;
#ifdef CONFIG_NUMA
2317
	s->remote_node_defrag_ratio = 1000;
C
Christoph Lameter 已提交
2318
#endif
2319 2320
	if (!init_kmem_cache_nodes(s, gfpflags & ~SLUB_DMA))
		goto error;
C
Christoph Lameter 已提交
2321

2322
	if (alloc_kmem_cache_cpus(s, gfpflags & ~SLUB_DMA))
C
Christoph Lameter 已提交
2323
		return 1;
2324
	free_kmem_cache_nodes(s);
C
Christoph Lameter 已提交
2325 2326 2327 2328
error:
	if (flags & SLAB_PANIC)
		panic("Cannot create slab %s size=%lu realsize=%u "
			"order=%u offset=%u flags=%lx\n",
2329
			s->name, (unsigned long)size, s->size, oo_order(s->oo),
C
Christoph Lameter 已提交
2330 2331 2332 2333 2334 2335 2336 2337 2338
			s->offset, flags);
	return 0;
}

/*
 * Check if a given pointer is valid
 */
int kmem_ptr_validate(struct kmem_cache *s, const void *object)
{
P
Pekka Enberg 已提交
2339
	struct page *page;
C
Christoph Lameter 已提交
2340 2341 2342 2343 2344 2345 2346

	page = get_object_page(object);

	if (!page || s != page->slab)
		/* No slab or wrong slab */
		return 0;

2347
	if (!check_valid_pointer(s, page, object))
C
Christoph Lameter 已提交
2348 2349 2350 2351 2352
		return 0;

	/*
	 * We could also check if the object is on the slabs freelist.
	 * But this would be too expensive and it seems that the main
C
Christoph Lameter 已提交
2353
	 * purpose of kmem_ptr_valid() is to check if the object belongs
C
Christoph Lameter 已提交
2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374
	 * to a certain slab.
	 */
	return 1;
}
EXPORT_SYMBOL(kmem_ptr_validate);

/*
 * Determine the size of a slab object
 */
unsigned int kmem_cache_size(struct kmem_cache *s)
{
	return s->objsize;
}
EXPORT_SYMBOL(kmem_cache_size);

const char *kmem_cache_name(struct kmem_cache *s)
{
	return s->name;
}
EXPORT_SYMBOL(kmem_cache_name);

2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400
static void list_slab_objects(struct kmem_cache *s, struct page *page,
							const char *text)
{
#ifdef CONFIG_SLUB_DEBUG
	void *addr = page_address(page);
	void *p;
	DECLARE_BITMAP(map, page->objects);

	bitmap_zero(map, page->objects);
	slab_err(s, page, "%s", text);
	slab_lock(page);
	for_each_free_object(p, s, page->freelist)
		set_bit(slab_index(p, s, addr), map);

	for_each_object(p, s, addr, page->objects) {

		if (!test_bit(slab_index(p, s, addr), map)) {
			printk(KERN_ERR "INFO: Object 0x%p @offset=%tu\n",
							p, p - addr);
			print_tracking(s, p);
		}
	}
	slab_unlock(page);
#endif
}

C
Christoph Lameter 已提交
2401
/*
C
Christoph Lameter 已提交
2402
 * Attempt to free all partial slabs on a node.
C
Christoph Lameter 已提交
2403
 */
C
Christoph Lameter 已提交
2404
static void free_partial(struct kmem_cache *s, struct kmem_cache_node *n)
C
Christoph Lameter 已提交
2405 2406 2407 2408 2409
{
	unsigned long flags;
	struct page *page, *h;

	spin_lock_irqsave(&n->list_lock, flags);
2410
	list_for_each_entry_safe(page, h, &n->partial, lru) {
C
Christoph Lameter 已提交
2411 2412 2413
		if (!page->inuse) {
			list_del(&page->lru);
			discard_slab(s, page);
C
Christoph Lameter 已提交
2414
			n->nr_partial--;
2415 2416 2417
		} else {
			list_slab_objects(s, page,
				"Objects remaining on kmem_cache_close()");
C
Christoph Lameter 已提交
2418
		}
2419
	}
C
Christoph Lameter 已提交
2420 2421 2422 2423
	spin_unlock_irqrestore(&n->list_lock, flags);
}

/*
C
Christoph Lameter 已提交
2424
 * Release all resources used by a slab cache.
C
Christoph Lameter 已提交
2425
 */
2426
static inline int kmem_cache_close(struct kmem_cache *s)
C
Christoph Lameter 已提交
2427 2428 2429 2430 2431 2432
{
	int node;

	flush_all(s);

	/* Attempt to free all objects */
2433
	free_kmem_cache_cpus(s);
C
Christoph Lameter 已提交
2434
	for_each_node_state(node, N_NORMAL_MEMORY) {
C
Christoph Lameter 已提交
2435 2436
		struct kmem_cache_node *n = get_node(s, node);

C
Christoph Lameter 已提交
2437 2438
		free_partial(s, n);
		if (n->nr_partial || slabs_node(s, node))
C
Christoph Lameter 已提交
2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450 2451 2452 2453 2454
			return 1;
	}
	free_kmem_cache_nodes(s);
	return 0;
}

/*
 * Close a cache and release the kmem_cache structure
 * (must be used for caches created using kmem_cache_create)
 */
void kmem_cache_destroy(struct kmem_cache *s)
{
	down_write(&slub_lock);
	s->refcount--;
	if (!s->refcount) {
		list_del(&s->list);
2455
		up_write(&slub_lock);
2456 2457 2458 2459 2460
		if (kmem_cache_close(s)) {
			printk(KERN_ERR "SLUB %s: %s called for cache that "
				"still has objects.\n", s->name, __func__);
			dump_stack();
		}
C
Christoph Lameter 已提交
2461
		sysfs_slab_remove(s);
2462 2463
	} else
		up_write(&slub_lock);
C
Christoph Lameter 已提交
2464 2465 2466 2467 2468 2469 2470
}
EXPORT_SYMBOL(kmem_cache_destroy);

/********************************************************************
 *		Kmalloc subsystem
 *******************************************************************/

2471
struct kmem_cache kmalloc_caches[PAGE_SHIFT + 1] __cacheline_aligned;
C
Christoph Lameter 已提交
2472 2473 2474 2475
EXPORT_SYMBOL(kmalloc_caches);

static int __init setup_slub_min_order(char *str)
{
P
Pekka Enberg 已提交
2476
	get_option(&str, &slub_min_order);
C
Christoph Lameter 已提交
2477 2478 2479 2480 2481 2482 2483 2484

	return 1;
}

__setup("slub_min_order=", setup_slub_min_order);

static int __init setup_slub_max_order(char *str)
{
P
Pekka Enberg 已提交
2485
	get_option(&str, &slub_max_order);
C
Christoph Lameter 已提交
2486 2487 2488 2489 2490 2491 2492 2493

	return 1;
}

__setup("slub_max_order=", setup_slub_max_order);

static int __init setup_slub_min_objects(char *str)
{
P
Pekka Enberg 已提交
2494
	get_option(&str, &slub_min_objects);
C
Christoph Lameter 已提交
2495 2496 2497 2498 2499 2500 2501 2502 2503 2504 2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515 2516 2517 2518

	return 1;
}

__setup("slub_min_objects=", setup_slub_min_objects);

static int __init setup_slub_nomerge(char *str)
{
	slub_nomerge = 1;
	return 1;
}

__setup("slub_nomerge", setup_slub_nomerge);

static struct kmem_cache *create_kmalloc_cache(struct kmem_cache *s,
		const char *name, int size, gfp_t gfp_flags)
{
	unsigned int flags = 0;

	if (gfp_flags & SLUB_DMA)
		flags = SLAB_CACHE_DMA;

	down_write(&slub_lock);
	if (!kmem_cache_open(s, gfp_flags, name, size, ARCH_KMALLOC_MINALIGN,
2519
								flags, NULL))
C
Christoph Lameter 已提交
2520 2521 2522 2523 2524 2525 2526 2527 2528 2529 2530 2531
		goto panic;

	list_add(&s->list, &slab_caches);
	up_write(&slub_lock);
	if (sysfs_slab_add(s))
		goto panic;
	return s;

panic:
	panic("Creation of kmalloc slab %s size=%d failed.\n", name, size);
}

2532
#ifdef CONFIG_ZONE_DMA
2533
static struct kmem_cache *kmalloc_caches_dma[PAGE_SHIFT + 1];
2534 2535 2536 2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548 2549 2550

static void sysfs_add_func(struct work_struct *w)
{
	struct kmem_cache *s;

	down_write(&slub_lock);
	list_for_each_entry(s, &slab_caches, list) {
		if (s->flags & __SYSFS_ADD_DEFERRED) {
			s->flags &= ~__SYSFS_ADD_DEFERRED;
			sysfs_slab_add(s);
		}
	}
	up_write(&slub_lock);
}

static DECLARE_WORK(sysfs_add_work, sysfs_add_func);

2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561
static noinline struct kmem_cache *dma_kmalloc_cache(int index, gfp_t flags)
{
	struct kmem_cache *s;
	char *text;
	size_t realsize;

	s = kmalloc_caches_dma[index];
	if (s)
		return s;

	/* Dynamically create dma cache */
2562 2563 2564 2565 2566 2567 2568 2569 2570
	if (flags & __GFP_WAIT)
		down_write(&slub_lock);
	else {
		if (!down_write_trylock(&slub_lock))
			goto out;
	}

	if (kmalloc_caches_dma[index])
		goto unlock_out;
2571

2572
	realsize = kmalloc_caches[index].objsize;
I
Ingo Molnar 已提交
2573 2574
	text = kasprintf(flags & ~SLUB_DMA, "kmalloc_dma-%d",
			 (unsigned int)realsize);
2575 2576 2577 2578 2579 2580 2581 2582
	s = kmalloc(kmem_size, flags & ~SLUB_DMA);

	if (!s || !text || !kmem_cache_open(s, flags, text,
			realsize, ARCH_KMALLOC_MINALIGN,
			SLAB_CACHE_DMA|__SYSFS_ADD_DEFERRED, NULL)) {
		kfree(s);
		kfree(text);
		goto unlock_out;
2583
	}
2584 2585 2586 2587 2588 2589 2590

	list_add(&s->list, &slab_caches);
	kmalloc_caches_dma[index] = s;

	schedule_work(&sysfs_add_work);

unlock_out:
2591
	up_write(&slub_lock);
2592
out:
2593
	return kmalloc_caches_dma[index];
2594 2595 2596
}
#endif

2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628 2629
/*
 * Conversion table for small slabs sizes / 8 to the index in the
 * kmalloc array. This is necessary for slabs < 192 since we have non power
 * of two cache sizes there. The size of larger slabs can be determined using
 * fls.
 */
static s8 size_index[24] = {
	3,	/* 8 */
	4,	/* 16 */
	5,	/* 24 */
	5,	/* 32 */
	6,	/* 40 */
	6,	/* 48 */
	6,	/* 56 */
	6,	/* 64 */
	1,	/* 72 */
	1,	/* 80 */
	1,	/* 88 */
	1,	/* 96 */
	7,	/* 104 */
	7,	/* 112 */
	7,	/* 120 */
	7,	/* 128 */
	2,	/* 136 */
	2,	/* 144 */
	2,	/* 152 */
	2,	/* 160 */
	2,	/* 168 */
	2,	/* 176 */
	2,	/* 184 */
	2	/* 192 */
};

C
Christoph Lameter 已提交
2630 2631
static struct kmem_cache *get_slab(size_t size, gfp_t flags)
{
2632
	int index;
C
Christoph Lameter 已提交
2633

2634 2635 2636
	if (size <= 192) {
		if (!size)
			return ZERO_SIZE_PTR;
C
Christoph Lameter 已提交
2637

2638
		index = size_index[(size - 1) / 8];
2639
	} else
2640
		index = fls(size - 1);
C
Christoph Lameter 已提交
2641 2642

#ifdef CONFIG_ZONE_DMA
2643
	if (unlikely((flags & SLUB_DMA)))
2644
		return dma_kmalloc_cache(index, flags);
2645

C
Christoph Lameter 已提交
2646 2647 2648 2649 2650 2651
#endif
	return &kmalloc_caches[index];
}

void *__kmalloc(size_t size, gfp_t flags)
{
2652
	struct kmem_cache *s;
C
Christoph Lameter 已提交
2653

2654
	if (unlikely(size > PAGE_SIZE))
2655
		return kmalloc_large(size, flags);
2656 2657 2658 2659

	s = get_slab(size, flags);

	if (unlikely(ZERO_OR_NULL_PTR(s)))
2660 2661
		return s;

2662
	return slab_alloc(s, flags, -1, __builtin_return_address(0));
C
Christoph Lameter 已提交
2663 2664 2665
}
EXPORT_SYMBOL(__kmalloc);

2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676
static void *kmalloc_large_node(size_t size, gfp_t flags, int node)
{
	struct page *page = alloc_pages_node(node, flags | __GFP_COMP,
						get_order(size));

	if (page)
		return page_address(page);
	else
		return NULL;
}

C
Christoph Lameter 已提交
2677 2678 2679
#ifdef CONFIG_NUMA
void *__kmalloc_node(size_t size, gfp_t flags, int node)
{
2680
	struct kmem_cache *s;
C
Christoph Lameter 已提交
2681

2682
	if (unlikely(size > PAGE_SIZE))
2683
		return kmalloc_large_node(size, flags, node);
2684 2685 2686 2687

	s = get_slab(size, flags);

	if (unlikely(ZERO_OR_NULL_PTR(s)))
2688 2689
		return s;

2690
	return slab_alloc(s, flags, node, __builtin_return_address(0));
C
Christoph Lameter 已提交
2691 2692 2693 2694 2695 2696
}
EXPORT_SYMBOL(__kmalloc_node);
#endif

size_t ksize(const void *object)
{
2697
	struct page *page;
C
Christoph Lameter 已提交
2698 2699
	struct kmem_cache *s;

2700
	if (unlikely(object == ZERO_SIZE_PTR))
2701 2702
		return 0;

2703 2704
	page = virt_to_head_page(object);

P
Pekka Enberg 已提交
2705 2706
	if (unlikely(!PageSlab(page))) {
		WARN_ON(!PageCompound(page));
2707
		return PAGE_SIZE << compound_order(page);
P
Pekka Enberg 已提交
2708
	}
C
Christoph Lameter 已提交
2709 2710
	s = page->slab;

2711
#ifdef CONFIG_SLUB_DEBUG
C
Christoph Lameter 已提交
2712 2713 2714 2715 2716 2717 2718
	/*
	 * Debugging requires use of the padding between object
	 * and whatever may come after it.
	 */
	if (s->flags & (SLAB_RED_ZONE | SLAB_POISON))
		return s->objsize;

2719
#endif
C
Christoph Lameter 已提交
2720 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735
	/*
	 * 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;
}

void kfree(const void *x)
{
	struct page *page;
2736
	void *object = (void *)x;
C
Christoph Lameter 已提交
2737

2738
	if (unlikely(ZERO_OR_NULL_PTR(x)))
C
Christoph Lameter 已提交
2739 2740
		return;

2741
	page = virt_to_head_page(x);
2742
	if (unlikely(!PageSlab(page))) {
2743
		BUG_ON(!PageCompound(page));
2744 2745 2746
		put_page(page);
		return;
	}
2747
	slab_free(page->slab, page, object, __builtin_return_address(0));
C
Christoph Lameter 已提交
2748 2749 2750
}
EXPORT_SYMBOL(kfree);

2751
/*
C
Christoph Lameter 已提交
2752 2753 2754 2755 2756 2757 2758 2759
 * kmem_cache_shrink removes empty slabs from the partial lists and sorts
 * the remaining slabs by the number of items in use. The slabs with the
 * most items in use come first. New allocations will then fill those up
 * and thus they can be removed from the partial lists.
 *
 * The slabs with the least items are placed last. This results in them
 * being allocated from last increasing the chance that the last objects
 * are freed in them.
2760 2761 2762 2763 2764 2765 2766 2767
 */
int kmem_cache_shrink(struct kmem_cache *s)
{
	int node;
	int i;
	struct kmem_cache_node *n;
	struct page *page;
	struct page *t;
2768
	int objects = oo_objects(s->max);
2769
	struct list_head *slabs_by_inuse =
2770
		kmalloc(sizeof(struct list_head) * objects, GFP_KERNEL);
2771 2772 2773 2774 2775 2776
	unsigned long flags;

	if (!slabs_by_inuse)
		return -ENOMEM;

	flush_all(s);
C
Christoph Lameter 已提交
2777
	for_each_node_state(node, N_NORMAL_MEMORY) {
2778 2779 2780 2781 2782
		n = get_node(s, node);

		if (!n->nr_partial)
			continue;

2783
		for (i = 0; i < objects; i++)
2784 2785 2786 2787 2788
			INIT_LIST_HEAD(slabs_by_inuse + i);

		spin_lock_irqsave(&n->list_lock, flags);

		/*
C
Christoph Lameter 已提交
2789
		 * Build lists indexed by the items in use in each slab.
2790
		 *
C
Christoph Lameter 已提交
2791 2792
		 * Note that concurrent frees may occur while we hold the
		 * list_lock. page->inuse here is the upper limit.
2793 2794 2795 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805
		 */
		list_for_each_entry_safe(page, t, &n->partial, lru) {
			if (!page->inuse && slab_trylock(page)) {
				/*
				 * Must hold slab lock here because slab_free
				 * may have freed the last object and be
				 * waiting to release the slab.
				 */
				list_del(&page->lru);
				n->nr_partial--;
				slab_unlock(page);
				discard_slab(s, page);
			} else {
2806 2807
				list_move(&page->lru,
				slabs_by_inuse + page->inuse);
2808 2809 2810 2811
			}
		}

		/*
C
Christoph Lameter 已提交
2812 2813
		 * Rebuild the partial list with the slabs filled up most
		 * first and the least used slabs at the end.
2814
		 */
2815
		for (i = objects - 1; i >= 0; i--)
2816 2817 2818 2819 2820 2821 2822 2823 2824 2825
			list_splice(slabs_by_inuse + i, n->partial.prev);

		spin_unlock_irqrestore(&n->list_lock, flags);
	}

	kfree(slabs_by_inuse);
	return 0;
}
EXPORT_SYMBOL(kmem_cache_shrink);

2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857 2858 2859 2860 2861 2862 2863 2864
#if defined(CONFIG_NUMA) && defined(CONFIG_MEMORY_HOTPLUG)
static int slab_mem_going_offline_callback(void *arg)
{
	struct kmem_cache *s;

	down_read(&slub_lock);
	list_for_each_entry(s, &slab_caches, list)
		kmem_cache_shrink(s);
	up_read(&slub_lock);

	return 0;
}

static void slab_mem_offline_callback(void *arg)
{
	struct kmem_cache_node *n;
	struct kmem_cache *s;
	struct memory_notify *marg = arg;
	int offline_node;

	offline_node = marg->status_change_nid;

	/*
	 * If the node still has available memory. we need kmem_cache_node
	 * for it yet.
	 */
	if (offline_node < 0)
		return;

	down_read(&slub_lock);
	list_for_each_entry(s, &slab_caches, list) {
		n = get_node(s, offline_node);
		if (n) {
			/*
			 * if n->nr_slabs > 0, slabs still exist on the node
			 * that is going down. We were unable to free them,
			 * and offline_pages() function shoudn't call this
			 * callback. So, we must fail.
			 */
2865
			BUG_ON(slabs_node(s, offline_node));
2866 2867 2868 2869 2870 2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889

			s->node[offline_node] = NULL;
			kmem_cache_free(kmalloc_caches, n);
		}
	}
	up_read(&slub_lock);
}

static int slab_mem_going_online_callback(void *arg)
{
	struct kmem_cache_node *n;
	struct kmem_cache *s;
	struct memory_notify *marg = arg;
	int nid = marg->status_change_nid;
	int ret = 0;

	/*
	 * If the node's memory is already available, then kmem_cache_node is
	 * already created. Nothing to do.
	 */
	if (nid < 0)
		return 0;

	/*
2890
	 * We are bringing a node online. No memory is available yet. We must
2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 2901 2902 2903 2904 2905
	 * allocate a kmem_cache_node structure in order to bring the node
	 * online.
	 */
	down_read(&slub_lock);
	list_for_each_entry(s, &slab_caches, list) {
		/*
		 * XXX: kmem_cache_alloc_node will fallback to other nodes
		 *      since memory is not yet available from the node that
		 *      is brought up.
		 */
		n = kmem_cache_alloc(kmalloc_caches, GFP_KERNEL);
		if (!n) {
			ret = -ENOMEM;
			goto out;
		}
2906
		init_kmem_cache_node(n, s);
2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925 2926 2927 2928 2929 2930 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940
		s->node[nid] = n;
	}
out:
	up_read(&slub_lock);
	return ret;
}

static int slab_memory_callback(struct notifier_block *self,
				unsigned long action, void *arg)
{
	int ret = 0;

	switch (action) {
	case MEM_GOING_ONLINE:
		ret = slab_mem_going_online_callback(arg);
		break;
	case MEM_GOING_OFFLINE:
		ret = slab_mem_going_offline_callback(arg);
		break;
	case MEM_OFFLINE:
	case MEM_CANCEL_ONLINE:
		slab_mem_offline_callback(arg);
		break;
	case MEM_ONLINE:
	case MEM_CANCEL_OFFLINE:
		break;
	}

	ret = notifier_from_errno(ret);
	return ret;
}

#endif /* CONFIG_MEMORY_HOTPLUG */

C
Christoph Lameter 已提交
2941 2942 2943 2944 2945 2946 2947
/********************************************************************
 *			Basic setup of slabs
 *******************************************************************/

void __init kmem_cache_init(void)
{
	int i;
2948
	int caches = 0;
C
Christoph Lameter 已提交
2949

2950 2951
	init_alloc_cpu();

C
Christoph Lameter 已提交
2952 2953 2954
#ifdef CONFIG_NUMA
	/*
	 * Must first have the slab cache available for the allocations of the
C
Christoph Lameter 已提交
2955
	 * struct kmem_cache_node's. There is special bootstrap code in
C
Christoph Lameter 已提交
2956 2957 2958 2959
	 * kmem_cache_open for slab_state == DOWN.
	 */
	create_kmalloc_cache(&kmalloc_caches[0], "kmem_cache_node",
		sizeof(struct kmem_cache_node), GFP_KERNEL);
2960
	kmalloc_caches[0].refcount = -1;
2961
	caches++;
2962

2963
	hotplug_memory_notifier(slab_memory_callback, SLAB_CALLBACK_PRI);
C
Christoph Lameter 已提交
2964 2965 2966 2967 2968 2969
#endif

	/* Able to allocate the per node structures */
	slab_state = PARTIAL;

	/* Caches that are not of the two-to-the-power-of size */
2970 2971
	if (KMALLOC_MIN_SIZE <= 64) {
		create_kmalloc_cache(&kmalloc_caches[1],
C
Christoph Lameter 已提交
2972
				"kmalloc-96", 96, GFP_KERNEL);
2973 2974
		caches++;
		create_kmalloc_cache(&kmalloc_caches[2],
C
Christoph Lameter 已提交
2975
				"kmalloc-192", 192, GFP_KERNEL);
2976 2977
		caches++;
	}
C
Christoph Lameter 已提交
2978

2979
	for (i = KMALLOC_SHIFT_LOW; i <= PAGE_SHIFT; i++) {
C
Christoph Lameter 已提交
2980 2981
		create_kmalloc_cache(&kmalloc_caches[i],
			"kmalloc", 1 << i, GFP_KERNEL);
2982 2983
		caches++;
	}
C
Christoph Lameter 已提交
2984

2985 2986 2987 2988

	/*
	 * Patch up the size_index table if we have strange large alignment
	 * requirements for the kmalloc array. This is only the case for
C
Christoph Lameter 已提交
2989
	 * MIPS it seems. The standard arches will not generate any code here.
2990 2991 2992 2993 2994 2995 2996 2997 2998 2999
	 *
	 * Largest permitted alignment is 256 bytes due to the way we
	 * handle the index determination for the smaller caches.
	 *
	 * Make sure that nothing crazy happens if someone starts tinkering
	 * around with ARCH_KMALLOC_MINALIGN
	 */
	BUILD_BUG_ON(KMALLOC_MIN_SIZE > 256 ||
		(KMALLOC_MIN_SIZE & (KMALLOC_MIN_SIZE - 1)));

3000
	for (i = 8; i < KMALLOC_MIN_SIZE; i += 8)
3001 3002
		size_index[(i - 1) / 8] = KMALLOC_SHIFT_LOW;

3003 3004 3005 3006 3007 3008 3009 3010 3011 3012
	if (KMALLOC_MIN_SIZE == 128) {
		/*
		 * The 192 byte sized cache is not used if the alignment
		 * is 128 byte. Redirect kmalloc to use the 256 byte cache
		 * instead.
		 */
		for (i = 128 + 8; i <= 192; i += 8)
			size_index[(i - 1) / 8] = 8;
	}

C
Christoph Lameter 已提交
3013 3014 3015
	slab_state = UP;

	/* Provide the correct kmalloc names now that the caches are up */
3016
	for (i = KMALLOC_SHIFT_LOW; i <= PAGE_SHIFT; i++)
C
Christoph Lameter 已提交
3017 3018 3019 3020 3021
		kmalloc_caches[i]. name =
			kasprintf(GFP_KERNEL, "kmalloc-%d", 1 << i);

#ifdef CONFIG_SMP
	register_cpu_notifier(&slab_notifier);
3022 3023 3024 3025
	kmem_size = offsetof(struct kmem_cache, cpu_slab) +
				nr_cpu_ids * sizeof(struct kmem_cache_cpu *);
#else
	kmem_size = sizeof(struct kmem_cache);
C
Christoph Lameter 已提交
3026 3027
#endif

I
Ingo Molnar 已提交
3028 3029
	printk(KERN_INFO
		"SLUB: Genslabs=%d, HWalign=%d, Order=%d-%d, MinObjects=%d,"
3030 3031
		" CPUs=%d, Nodes=%d\n",
		caches, cache_line_size(),
C
Christoph Lameter 已提交
3032 3033 3034 3035 3036 3037 3038 3039 3040 3041 3042 3043
		slub_min_order, slub_max_order, slub_min_objects,
		nr_cpu_ids, nr_node_ids);
}

/*
 * Find a mergeable slab cache
 */
static int slab_unmergeable(struct kmem_cache *s)
{
	if (slub_nomerge || (s->flags & SLUB_NEVER_MERGE))
		return 1;

3044
	if (s->ctor)
C
Christoph Lameter 已提交
3045 3046
		return 1;

3047 3048 3049 3050 3051 3052
	/*
	 * We may have set a slab to be unmergeable during bootstrap.
	 */
	if (s->refcount < 0)
		return 1;

C
Christoph Lameter 已提交
3053 3054 3055 3056
	return 0;
}

static struct kmem_cache *find_mergeable(size_t size,
3057
		size_t align, unsigned long flags, const char *name,
3058
		void (*ctor)(void *))
C
Christoph Lameter 已提交
3059
{
3060
	struct kmem_cache *s;
C
Christoph Lameter 已提交
3061 3062 3063 3064

	if (slub_nomerge || (flags & SLUB_NEVER_MERGE))
		return NULL;

3065
	if (ctor)
C
Christoph Lameter 已提交
3066 3067 3068 3069 3070
		return NULL;

	size = ALIGN(size, sizeof(void *));
	align = calculate_alignment(flags, align, size);
	size = ALIGN(size, align);
3071
	flags = kmem_cache_flags(size, flags, name, NULL);
C
Christoph Lameter 已提交
3072

3073
	list_for_each_entry(s, &slab_caches, list) {
C
Christoph Lameter 已提交
3074 3075 3076 3077 3078 3079
		if (slab_unmergeable(s))
			continue;

		if (size > s->size)
			continue;

3080
		if ((flags & SLUB_MERGE_SAME) != (s->flags & SLUB_MERGE_SAME))
C
Christoph Lameter 已提交
3081 3082 3083 3084 3085
				continue;
		/*
		 * Check if alignment is compatible.
		 * Courtesy of Adrian Drzewiecki
		 */
P
Pekka Enberg 已提交
3086
		if ((s->size & ~(align - 1)) != s->size)
C
Christoph Lameter 已提交
3087 3088 3089 3090 3091 3092 3093 3094 3095 3096 3097
			continue;

		if (s->size - size >= sizeof(void *))
			continue;

		return s;
	}
	return NULL;
}

struct kmem_cache *kmem_cache_create(const char *name, size_t size,
3098
		size_t align, unsigned long flags, void (*ctor)(void *))
C
Christoph Lameter 已提交
3099 3100 3101 3102
{
	struct kmem_cache *s;

	down_write(&slub_lock);
3103
	s = find_mergeable(size, align, flags, name, ctor);
C
Christoph Lameter 已提交
3104
	if (s) {
3105 3106
		int cpu;

C
Christoph Lameter 已提交
3107 3108 3109 3110 3111 3112
		s->refcount++;
		/*
		 * Adjust the object sizes so that we clear
		 * the complete object on kzalloc.
		 */
		s->objsize = max(s->objsize, (int)size);
3113 3114 3115 3116 3117 3118 3119

		/*
		 * And then we need to update the object size in the
		 * per cpu structures
		 */
		for_each_online_cpu(cpu)
			get_cpu_slab(s, cpu)->objsize = s->objsize;
C
Christoph Lameter 已提交
3120

C
Christoph Lameter 已提交
3121
		s->inuse = max_t(int, s->inuse, ALIGN(size, sizeof(void *)));
3122
		up_write(&slub_lock);
C
Christoph Lameter 已提交
3123

C
Christoph Lameter 已提交
3124 3125
		if (sysfs_slab_alias(s, name))
			goto err;
3126 3127
		return s;
	}
C
Christoph Lameter 已提交
3128

3129 3130 3131
	s = kmalloc(kmem_size, GFP_KERNEL);
	if (s) {
		if (kmem_cache_open(s, GFP_KERNEL, name,
3132
				size, align, flags, ctor)) {
C
Christoph Lameter 已提交
3133
			list_add(&s->list, &slab_caches);
3134 3135 3136 3137 3138 3139
			up_write(&slub_lock);
			if (sysfs_slab_add(s))
				goto err;
			return s;
		}
		kfree(s);
C
Christoph Lameter 已提交
3140 3141 3142 3143 3144 3145 3146 3147 3148 3149 3150 3151 3152 3153
	}
	up_write(&slub_lock);

err:
	if (flags & SLAB_PANIC)
		panic("Cannot create slabcache %s\n", name);
	else
		s = NULL;
	return s;
}
EXPORT_SYMBOL(kmem_cache_create);

#ifdef CONFIG_SMP
/*
C
Christoph Lameter 已提交
3154 3155
 * Use the cpu notifier to insure that the cpu slabs are flushed when
 * necessary.
C
Christoph Lameter 已提交
3156 3157 3158 3159 3160
 */
static int __cpuinit slab_cpuup_callback(struct notifier_block *nfb,
		unsigned long action, void *hcpu)
{
	long cpu = (long)hcpu;
3161 3162
	struct kmem_cache *s;
	unsigned long flags;
C
Christoph Lameter 已提交
3163 3164

	switch (action) {
3165 3166 3167 3168 3169 3170 3171 3172 3173 3174
	case CPU_UP_PREPARE:
	case CPU_UP_PREPARE_FROZEN:
		init_alloc_cpu_cpu(cpu);
		down_read(&slub_lock);
		list_for_each_entry(s, &slab_caches, list)
			s->cpu_slab[cpu] = alloc_kmem_cache_cpu(s, cpu,
							GFP_KERNEL);
		up_read(&slub_lock);
		break;

C
Christoph Lameter 已提交
3175
	case CPU_UP_CANCELED:
3176
	case CPU_UP_CANCELED_FROZEN:
C
Christoph Lameter 已提交
3177
	case CPU_DEAD:
3178
	case CPU_DEAD_FROZEN:
3179 3180
		down_read(&slub_lock);
		list_for_each_entry(s, &slab_caches, list) {
3181 3182
			struct kmem_cache_cpu *c = get_cpu_slab(s, cpu);

3183 3184 3185
			local_irq_save(flags);
			__flush_cpu_slab(s, cpu);
			local_irq_restore(flags);
3186 3187
			free_kmem_cache_cpu(c, cpu);
			s->cpu_slab[cpu] = NULL;
3188 3189
		}
		up_read(&slub_lock);
C
Christoph Lameter 已提交
3190 3191 3192 3193 3194 3195 3196
		break;
	default:
		break;
	}
	return NOTIFY_OK;
}

P
Pekka Enberg 已提交
3197
static struct notifier_block __cpuinitdata slab_notifier = {
I
Ingo Molnar 已提交
3198
	.notifier_call = slab_cpuup_callback
P
Pekka Enberg 已提交
3199
};
C
Christoph Lameter 已提交
3200 3201 3202 3203 3204

#endif

void *__kmalloc_track_caller(size_t size, gfp_t gfpflags, void *caller)
{
3205 3206
	struct kmem_cache *s;

3207
	if (unlikely(size > PAGE_SIZE))
3208 3209
		return kmalloc_large(size, gfpflags);

3210
	s = get_slab(size, gfpflags);
C
Christoph Lameter 已提交
3211

3212
	if (unlikely(ZERO_OR_NULL_PTR(s)))
3213
		return s;
C
Christoph Lameter 已提交
3214

3215
	return slab_alloc(s, gfpflags, -1, caller);
C
Christoph Lameter 已提交
3216 3217 3218 3219 3220
}

void *__kmalloc_node_track_caller(size_t size, gfp_t gfpflags,
					int node, void *caller)
{
3221 3222
	struct kmem_cache *s;

3223
	if (unlikely(size > PAGE_SIZE))
3224
		return kmalloc_large_node(size, gfpflags, node);
3225

3226
	s = get_slab(size, gfpflags);
C
Christoph Lameter 已提交
3227

3228
	if (unlikely(ZERO_OR_NULL_PTR(s)))
3229
		return s;
C
Christoph Lameter 已提交
3230

3231
	return slab_alloc(s, gfpflags, node, caller);
C
Christoph Lameter 已提交
3232 3233
}

C
Christoph Lameter 已提交
3234
#ifdef CONFIG_SLUB_DEBUG
3235 3236
static unsigned long count_partial(struct kmem_cache_node *n,
					int (*get_count)(struct page *))
3237 3238 3239 3240 3241 3242 3243
{
	unsigned long flags;
	unsigned long x = 0;
	struct page *page;

	spin_lock_irqsave(&n->list_lock, flags);
	list_for_each_entry(page, &n->partial, lru)
3244
		x += get_count(page);
3245 3246 3247
	spin_unlock_irqrestore(&n->list_lock, flags);
	return x;
}
3248 3249 3250 3251 3252 3253 3254 3255 3256 3257 3258 3259 3260 3261 3262

static int count_inuse(struct page *page)
{
	return page->inuse;
}

static int count_total(struct page *page)
{
	return page->objects;
}

static int count_free(struct page *page)
{
	return page->objects - page->inuse;
}
3263

3264 3265
static int validate_slab(struct kmem_cache *s, struct page *page,
						unsigned long *map)
3266 3267
{
	void *p;
3268
	void *addr = page_address(page);
3269 3270 3271 3272 3273 3274

	if (!check_slab(s, page) ||
			!on_freelist(s, page, NULL))
		return 0;

	/* Now we know that a valid freelist exists */
3275
	bitmap_zero(map, page->objects);
3276

3277 3278
	for_each_free_object(p, s, page->freelist) {
		set_bit(slab_index(p, s, addr), map);
3279 3280 3281 3282
		if (!check_object(s, page, p, 0))
			return 0;
	}

3283
	for_each_object(p, s, addr, page->objects)
3284
		if (!test_bit(slab_index(p, s, addr), map))
3285 3286 3287 3288 3289
			if (!check_object(s, page, p, 1))
				return 0;
	return 1;
}

3290 3291
static void validate_slab_slab(struct kmem_cache *s, struct page *page,
						unsigned long *map)
3292 3293
{
	if (slab_trylock(page)) {
3294
		validate_slab(s, page, map);
3295 3296 3297 3298 3299 3300
		slab_unlock(page);
	} else
		printk(KERN_INFO "SLUB %s: Skipped busy slab 0x%p\n",
			s->name, page);

	if (s->flags & DEBUG_DEFAULT_FLAGS) {
3301 3302
		if (!PageSlubDebug(page))
			printk(KERN_ERR "SLUB %s: SlubDebug not set "
3303 3304
				"on slab 0x%p\n", s->name, page);
	} else {
3305 3306
		if (PageSlubDebug(page))
			printk(KERN_ERR "SLUB %s: SlubDebug set on "
3307 3308 3309 3310
				"slab 0x%p\n", s->name, page);
	}
}

3311 3312
static int validate_slab_node(struct kmem_cache *s,
		struct kmem_cache_node *n, unsigned long *map)
3313 3314 3315 3316 3317 3318 3319 3320
{
	unsigned long count = 0;
	struct page *page;
	unsigned long flags;

	spin_lock_irqsave(&n->list_lock, flags);

	list_for_each_entry(page, &n->partial, lru) {
3321
		validate_slab_slab(s, page, map);
3322 3323 3324 3325 3326 3327 3328 3329 3330 3331
		count++;
	}
	if (count != n->nr_partial)
		printk(KERN_ERR "SLUB %s: %ld partial slabs counted but "
			"counter=%ld\n", s->name, count, n->nr_partial);

	if (!(s->flags & SLAB_STORE_USER))
		goto out;

	list_for_each_entry(page, &n->full, lru) {
3332
		validate_slab_slab(s, page, map);
3333 3334 3335 3336 3337 3338 3339 3340 3341 3342 3343 3344
		count++;
	}
	if (count != atomic_long_read(&n->nr_slabs))
		printk(KERN_ERR "SLUB: %s %ld slabs counted but "
			"counter=%ld\n", s->name, count,
			atomic_long_read(&n->nr_slabs));

out:
	spin_unlock_irqrestore(&n->list_lock, flags);
	return count;
}

3345
static long validate_slab_cache(struct kmem_cache *s)
3346 3347 3348
{
	int node;
	unsigned long count = 0;
3349
	unsigned long *map = kmalloc(BITS_TO_LONGS(oo_objects(s->max)) *
3350 3351 3352 3353
				sizeof(unsigned long), GFP_KERNEL);

	if (!map)
		return -ENOMEM;
3354 3355

	flush_all(s);
C
Christoph Lameter 已提交
3356
	for_each_node_state(node, N_NORMAL_MEMORY) {
3357 3358
		struct kmem_cache_node *n = get_node(s, node);

3359
		count += validate_slab_node(s, n, map);
3360
	}
3361
	kfree(map);
3362 3363 3364
	return count;
}

3365 3366 3367 3368 3369 3370 3371 3372 3373 3374 3375 3376 3377 3378 3379 3380 3381 3382 3383 3384
#ifdef SLUB_RESILIENCY_TEST
static void resiliency_test(void)
{
	u8 *p;

	printk(KERN_ERR "SLUB resiliency testing\n");
	printk(KERN_ERR "-----------------------\n");
	printk(KERN_ERR "A. Corruption after allocation\n");

	p = kzalloc(16, GFP_KERNEL);
	p[16] = 0x12;
	printk(KERN_ERR "\n1. kmalloc-16: Clobber Redzone/next pointer"
			" 0x12->0x%p\n\n", p + 16);

	validate_slab_cache(kmalloc_caches + 4);

	/* Hmmm... The next two are dangerous */
	p = kzalloc(32, GFP_KERNEL);
	p[32 + sizeof(void *)] = 0x34;
	printk(KERN_ERR "\n2. kmalloc-32: Clobber next pointer/next slab"
I
Ingo Molnar 已提交
3385 3386 3387
			" 0x34 -> -0x%p\n", p);
	printk(KERN_ERR
		"If allocated object is overwritten then not detectable\n\n");
3388 3389 3390 3391 3392 3393 3394

	validate_slab_cache(kmalloc_caches + 5);
	p = kzalloc(64, GFP_KERNEL);
	p += 64 + (get_cycles() & 0xff) * sizeof(void *);
	*p = 0x56;
	printk(KERN_ERR "\n3. kmalloc-64: corrupting random byte 0x56->0x%p\n",
									p);
I
Ingo Molnar 已提交
3395 3396
	printk(KERN_ERR
		"If allocated object is overwritten then not detectable\n\n");
3397 3398 3399 3400 3401 3402 3403 3404 3405 3406 3407 3408
	validate_slab_cache(kmalloc_caches + 6);

	printk(KERN_ERR "\nB. Corruption after free\n");
	p = kzalloc(128, GFP_KERNEL);
	kfree(p);
	*p = 0x78;
	printk(KERN_ERR "1. kmalloc-128: Clobber first word 0x78->0x%p\n\n", p);
	validate_slab_cache(kmalloc_caches + 7);

	p = kzalloc(256, GFP_KERNEL);
	kfree(p);
	p[50] = 0x9a;
I
Ingo Molnar 已提交
3409 3410
	printk(KERN_ERR "\n2. kmalloc-256: Clobber 50th byte 0x9a->0x%p\n\n",
			p);
3411 3412 3413 3414 3415 3416 3417 3418 3419 3420 3421 3422
	validate_slab_cache(kmalloc_caches + 8);

	p = kzalloc(512, GFP_KERNEL);
	kfree(p);
	p[512] = 0xab;
	printk(KERN_ERR "\n3. kmalloc-512: Clobber redzone 0xab->0x%p\n\n", p);
	validate_slab_cache(kmalloc_caches + 9);
}
#else
static void resiliency_test(void) {};
#endif

3423
/*
C
Christoph Lameter 已提交
3424
 * Generate lists of code addresses where slabcache objects are allocated
3425 3426 3427 3428 3429 3430
 * and freed.
 */

struct location {
	unsigned long count;
	void *addr;
3431 3432 3433 3434 3435 3436 3437
	long long sum_time;
	long min_time;
	long max_time;
	long min_pid;
	long max_pid;
	cpumask_t cpus;
	nodemask_t nodes;
3438 3439 3440 3441 3442 3443 3444 3445 3446 3447 3448 3449 3450 3451 3452
};

struct loc_track {
	unsigned long max;
	unsigned long count;
	struct location *loc;
};

static void free_loc_track(struct loc_track *t)
{
	if (t->max)
		free_pages((unsigned long)t->loc,
			get_order(sizeof(struct location) * t->max));
}

3453
static int alloc_loc_track(struct loc_track *t, unsigned long max, gfp_t flags)
3454 3455 3456 3457 3458 3459
{
	struct location *l;
	int order;

	order = get_order(sizeof(struct location) * max);

3460
	l = (void *)__get_free_pages(flags, order);
3461 3462 3463 3464 3465 3466 3467 3468 3469 3470 3471 3472 3473
	if (!l)
		return 0;

	if (t->count) {
		memcpy(l, t->loc, sizeof(struct location) * t->count);
		free_loc_track(t);
	}
	t->max = max;
	t->loc = l;
	return 1;
}

static int add_location(struct loc_track *t, struct kmem_cache *s,
3474
				const struct track *track)
3475 3476 3477 3478
{
	long start, end, pos;
	struct location *l;
	void *caddr;
3479
	unsigned long age = jiffies - track->when;
3480 3481 3482 3483 3484 3485 3486 3487 3488 3489 3490 3491 3492 3493 3494

	start = -1;
	end = t->count;

	for ( ; ; ) {
		pos = start + (end - start + 1) / 2;

		/*
		 * There is nothing at "end". If we end up there
		 * we need to add something to before end.
		 */
		if (pos == end)
			break;

		caddr = t->loc[pos].addr;
3495 3496 3497 3498 3499 3500 3501 3502 3503 3504 3505 3506 3507 3508 3509 3510 3511 3512 3513
		if (track->addr == caddr) {

			l = &t->loc[pos];
			l->count++;
			if (track->when) {
				l->sum_time += age;
				if (age < l->min_time)
					l->min_time = age;
				if (age > l->max_time)
					l->max_time = age;

				if (track->pid < l->min_pid)
					l->min_pid = track->pid;
				if (track->pid > l->max_pid)
					l->max_pid = track->pid;

				cpu_set(track->cpu, l->cpus);
			}
			node_set(page_to_nid(virt_to_page(track)), l->nodes);
3514 3515 3516
			return 1;
		}

3517
		if (track->addr < caddr)
3518 3519 3520 3521 3522 3523
			end = pos;
		else
			start = pos;
	}

	/*
C
Christoph Lameter 已提交
3524
	 * Not found. Insert new tracking element.
3525
	 */
3526
	if (t->count >= t->max && !alloc_loc_track(t, 2 * t->max, GFP_ATOMIC))
3527 3528 3529 3530 3531 3532 3533 3534
		return 0;

	l = t->loc + pos;
	if (pos < t->count)
		memmove(l + 1, l,
			(t->count - pos) * sizeof(struct location));
	t->count++;
	l->count = 1;
3535 3536 3537 3538 3539 3540 3541 3542 3543 3544
	l->addr = track->addr;
	l->sum_time = age;
	l->min_time = age;
	l->max_time = age;
	l->min_pid = track->pid;
	l->max_pid = track->pid;
	cpus_clear(l->cpus);
	cpu_set(track->cpu, l->cpus);
	nodes_clear(l->nodes);
	node_set(page_to_nid(virt_to_page(track)), l->nodes);
3545 3546 3547 3548 3549 3550
	return 1;
}

static void process_slab(struct loc_track *t, struct kmem_cache *s,
		struct page *page, enum track_item alloc)
{
3551
	void *addr = page_address(page);
3552
	DECLARE_BITMAP(map, page->objects);
3553 3554
	void *p;

3555
	bitmap_zero(map, page->objects);
3556 3557
	for_each_free_object(p, s, page->freelist)
		set_bit(slab_index(p, s, addr), map);
3558

3559
	for_each_object(p, s, addr, page->objects)
3560 3561
		if (!test_bit(slab_index(p, s, addr), map))
			add_location(t, s, get_track(s, p, alloc));
3562 3563 3564 3565 3566
}

static int list_locations(struct kmem_cache *s, char *buf,
					enum track_item alloc)
{
3567
	int len = 0;
3568
	unsigned long i;
3569
	struct loc_track t = { 0, 0, NULL };
3570 3571
	int node;

3572
	if (!alloc_loc_track(&t, PAGE_SIZE / sizeof(struct location),
3573
			GFP_TEMPORARY))
3574
		return sprintf(buf, "Out of memory\n");
3575 3576 3577 3578

	/* Push back cpu slabs */
	flush_all(s);

C
Christoph Lameter 已提交
3579
	for_each_node_state(node, N_NORMAL_MEMORY) {
3580 3581 3582 3583
		struct kmem_cache_node *n = get_node(s, node);
		unsigned long flags;
		struct page *page;

3584
		if (!atomic_long_read(&n->nr_slabs))
3585 3586 3587 3588 3589 3590 3591 3592 3593 3594 3595
			continue;

		spin_lock_irqsave(&n->list_lock, flags);
		list_for_each_entry(page, &n->partial, lru)
			process_slab(&t, s, page, alloc);
		list_for_each_entry(page, &n->full, lru)
			process_slab(&t, s, page, alloc);
		spin_unlock_irqrestore(&n->list_lock, flags);
	}

	for (i = 0; i < t.count; i++) {
3596
		struct location *l = &t.loc[i];
3597

3598
		if (len > PAGE_SIZE - 100)
3599
			break;
3600
		len += sprintf(buf + len, "%7ld ", l->count);
3601 3602

		if (l->addr)
3603
			len += sprint_symbol(buf + len, (unsigned long)l->addr);
3604
		else
3605
			len += sprintf(buf + len, "<not-available>");
3606 3607

		if (l->sum_time != l->min_time) {
3608
			len += sprintf(buf + len, " age=%ld/%ld/%ld",
R
Roman Zippel 已提交
3609 3610 3611
				l->min_time,
				(long)div_u64(l->sum_time, l->count),
				l->max_time);
3612
		} else
3613
			len += sprintf(buf + len, " age=%ld",
3614 3615 3616
				l->min_time);

		if (l->min_pid != l->max_pid)
3617
			len += sprintf(buf + len, " pid=%ld-%ld",
3618 3619
				l->min_pid, l->max_pid);
		else
3620
			len += sprintf(buf + len, " pid=%ld",
3621 3622
				l->min_pid);

3623
		if (num_online_cpus() > 1 && !cpus_empty(l->cpus) &&
3624 3625 3626
				len < PAGE_SIZE - 60) {
			len += sprintf(buf + len, " cpus=");
			len += cpulist_scnprintf(buf + len, PAGE_SIZE - len - 50,
3627 3628 3629
					l->cpus);
		}

3630
		if (num_online_nodes() > 1 && !nodes_empty(l->nodes) &&
3631 3632 3633
				len < PAGE_SIZE - 60) {
			len += sprintf(buf + len, " nodes=");
			len += nodelist_scnprintf(buf + len, PAGE_SIZE - len - 50,
3634 3635 3636
					l->nodes);
		}

3637
		len += sprintf(buf + len, "\n");
3638 3639 3640 3641
	}

	free_loc_track(&t);
	if (!t.count)
3642 3643
		len += sprintf(buf, "No data\n");
	return len;
3644 3645
}

C
Christoph Lameter 已提交
3646
enum slab_stat_type {
3647 3648 3649 3650 3651
	SL_ALL,			/* All slabs */
	SL_PARTIAL,		/* Only partially allocated slabs */
	SL_CPU,			/* Only slabs used for cpu caches */
	SL_OBJECTS,		/* Determine allocated objects not slabs */
	SL_TOTAL		/* Determine object capacity not slabs */
C
Christoph Lameter 已提交
3652 3653
};

3654
#define SO_ALL		(1 << SL_ALL)
C
Christoph Lameter 已提交
3655 3656 3657
#define SO_PARTIAL	(1 << SL_PARTIAL)
#define SO_CPU		(1 << SL_CPU)
#define SO_OBJECTS	(1 << SL_OBJECTS)
3658
#define SO_TOTAL	(1 << SL_TOTAL)
C
Christoph Lameter 已提交
3659

3660 3661
static ssize_t show_slab_objects(struct kmem_cache *s,
			    char *buf, unsigned long flags)
C
Christoph Lameter 已提交
3662 3663 3664 3665 3666 3667 3668 3669
{
	unsigned long total = 0;
	int node;
	int x;
	unsigned long *nodes;
	unsigned long *per_cpu;

	nodes = kzalloc(2 * sizeof(unsigned long) * nr_node_ids, GFP_KERNEL);
3670 3671
	if (!nodes)
		return -ENOMEM;
C
Christoph Lameter 已提交
3672 3673
	per_cpu = nodes + nr_node_ids;

3674 3675
	if (flags & SO_CPU) {
		int cpu;
C
Christoph Lameter 已提交
3676

3677 3678
		for_each_possible_cpu(cpu) {
			struct kmem_cache_cpu *c = get_cpu_slab(s, cpu);
3679

3680 3681 3682 3683 3684 3685 3686 3687
			if (!c || c->node < 0)
				continue;

			if (c->page) {
					if (flags & SO_TOTAL)
						x = c->page->objects;
				else if (flags & SO_OBJECTS)
					x = c->page->inuse;
C
Christoph Lameter 已提交
3688 3689
				else
					x = 1;
3690

C
Christoph Lameter 已提交
3691
				total += x;
3692
				nodes[c->node] += x;
C
Christoph Lameter 已提交
3693
			}
3694
			per_cpu[c->node]++;
C
Christoph Lameter 已提交
3695 3696 3697
		}
	}

3698 3699 3700 3701 3702 3703 3704 3705 3706
	if (flags & SO_ALL) {
		for_each_node_state(node, N_NORMAL_MEMORY) {
			struct kmem_cache_node *n = get_node(s, node);

		if (flags & SO_TOTAL)
			x = atomic_long_read(&n->total_objects);
		else if (flags & SO_OBJECTS)
			x = atomic_long_read(&n->total_objects) -
				count_partial(n, count_free);
C
Christoph Lameter 已提交
3707 3708

			else
3709
				x = atomic_long_read(&n->nr_slabs);
C
Christoph Lameter 已提交
3710 3711 3712 3713
			total += x;
			nodes[node] += x;
		}

3714 3715 3716
	} else if (flags & SO_PARTIAL) {
		for_each_node_state(node, N_NORMAL_MEMORY) {
			struct kmem_cache_node *n = get_node(s, node);
C
Christoph Lameter 已提交
3717

3718 3719 3720 3721
			if (flags & SO_TOTAL)
				x = count_partial(n, count_total);
			else if (flags & SO_OBJECTS)
				x = count_partial(n, count_inuse);
C
Christoph Lameter 已提交
3722
			else
3723
				x = n->nr_partial;
C
Christoph Lameter 已提交
3724 3725 3726 3727 3728 3729
			total += x;
			nodes[node] += x;
		}
	}
	x = sprintf(buf, "%lu", total);
#ifdef CONFIG_NUMA
C
Christoph Lameter 已提交
3730
	for_each_node_state(node, N_NORMAL_MEMORY)
C
Christoph Lameter 已提交
3731 3732 3733 3734 3735 3736 3737 3738 3739 3740 3741 3742
		if (nodes[node])
			x += sprintf(buf + x, " N%d=%lu",
					node, nodes[node]);
#endif
	kfree(nodes);
	return x + sprintf(buf + x, "\n");
}

static int any_slab_objects(struct kmem_cache *s)
{
	int node;

3743
	for_each_online_node(node) {
C
Christoph Lameter 已提交
3744 3745
		struct kmem_cache_node *n = get_node(s, node);

3746 3747 3748
		if (!n)
			continue;

3749
		if (atomic_long_read(&n->total_objects))
C
Christoph Lameter 已提交
3750 3751 3752 3753 3754 3755 3756 3757 3758 3759 3760 3761 3762 3763 3764 3765 3766 3767 3768 3769 3770 3771 3772 3773 3774 3775 3776 3777 3778 3779 3780 3781 3782 3783 3784 3785 3786 3787 3788 3789 3790
			return 1;
	}
	return 0;
}

#define to_slab_attr(n) container_of(n, struct slab_attribute, attr)
#define to_slab(n) container_of(n, struct kmem_cache, kobj);

struct slab_attribute {
	struct attribute attr;
	ssize_t (*show)(struct kmem_cache *s, char *buf);
	ssize_t (*store)(struct kmem_cache *s, const char *x, size_t count);
};

#define SLAB_ATTR_RO(_name) \
	static struct slab_attribute _name##_attr = __ATTR_RO(_name)

#define SLAB_ATTR(_name) \
	static struct slab_attribute _name##_attr =  \
	__ATTR(_name, 0644, _name##_show, _name##_store)

static ssize_t slab_size_show(struct kmem_cache *s, char *buf)
{
	return sprintf(buf, "%d\n", s->size);
}
SLAB_ATTR_RO(slab_size);

static ssize_t align_show(struct kmem_cache *s, char *buf)
{
	return sprintf(buf, "%d\n", s->align);
}
SLAB_ATTR_RO(align);

static ssize_t object_size_show(struct kmem_cache *s, char *buf)
{
	return sprintf(buf, "%d\n", s->objsize);
}
SLAB_ATTR_RO(object_size);

static ssize_t objs_per_slab_show(struct kmem_cache *s, char *buf)
{
3791
	return sprintf(buf, "%d\n", oo_objects(s->oo));
C
Christoph Lameter 已提交
3792 3793 3794
}
SLAB_ATTR_RO(objs_per_slab);

3795 3796 3797
static ssize_t order_store(struct kmem_cache *s,
				const char *buf, size_t length)
{
3798 3799 3800 3801 3802 3803
	unsigned long order;
	int err;

	err = strict_strtoul(buf, 10, &order);
	if (err)
		return err;
3804 3805 3806 3807 3808 3809 3810 3811

	if (order > slub_max_order || order < slub_min_order)
		return -EINVAL;

	calculate_sizes(s, order);
	return length;
}

C
Christoph Lameter 已提交
3812 3813
static ssize_t order_show(struct kmem_cache *s, char *buf)
{
3814
	return sprintf(buf, "%d\n", oo_order(s->oo));
C
Christoph Lameter 已提交
3815
}
3816
SLAB_ATTR(order);
C
Christoph Lameter 已提交
3817 3818 3819 3820 3821 3822 3823 3824 3825 3826 3827 3828 3829 3830 3831 3832 3833 3834 3835 3836

static ssize_t ctor_show(struct kmem_cache *s, char *buf)
{
	if (s->ctor) {
		int n = sprint_symbol(buf, (unsigned long)s->ctor);

		return n + sprintf(buf + n, "\n");
	}
	return 0;
}
SLAB_ATTR_RO(ctor);

static ssize_t aliases_show(struct kmem_cache *s, char *buf)
{
	return sprintf(buf, "%d\n", s->refcount - 1);
}
SLAB_ATTR_RO(aliases);

static ssize_t slabs_show(struct kmem_cache *s, char *buf)
{
3837
	return show_slab_objects(s, buf, SO_ALL);
C
Christoph Lameter 已提交
3838 3839 3840 3841 3842
}
SLAB_ATTR_RO(slabs);

static ssize_t partial_show(struct kmem_cache *s, char *buf)
{
3843
	return show_slab_objects(s, buf, SO_PARTIAL);
C
Christoph Lameter 已提交
3844 3845 3846 3847 3848
}
SLAB_ATTR_RO(partial);

static ssize_t cpu_slabs_show(struct kmem_cache *s, char *buf)
{
3849
	return show_slab_objects(s, buf, SO_CPU);
C
Christoph Lameter 已提交
3850 3851 3852 3853 3854
}
SLAB_ATTR_RO(cpu_slabs);

static ssize_t objects_show(struct kmem_cache *s, char *buf)
{
3855
	return show_slab_objects(s, buf, SO_ALL|SO_OBJECTS);
C
Christoph Lameter 已提交
3856 3857 3858
}
SLAB_ATTR_RO(objects);

3859 3860 3861 3862 3863 3864 3865 3866 3867 3868 3869 3870
static ssize_t objects_partial_show(struct kmem_cache *s, char *buf)
{
	return show_slab_objects(s, buf, SO_PARTIAL|SO_OBJECTS);
}
SLAB_ATTR_RO(objects_partial);

static ssize_t total_objects_show(struct kmem_cache *s, char *buf)
{
	return show_slab_objects(s, buf, SO_ALL|SO_TOTAL);
}
SLAB_ATTR_RO(total_objects);

C
Christoph Lameter 已提交
3871 3872 3873 3874 3875 3876 3877 3878 3879 3880 3881 3882 3883 3884 3885 3886 3887 3888 3889 3890 3891 3892 3893 3894 3895 3896 3897 3898 3899 3900 3901 3902 3903 3904 3905 3906 3907 3908 3909 3910 3911 3912 3913 3914 3915 3916 3917
static ssize_t sanity_checks_show(struct kmem_cache *s, char *buf)
{
	return sprintf(buf, "%d\n", !!(s->flags & SLAB_DEBUG_FREE));
}

static ssize_t sanity_checks_store(struct kmem_cache *s,
				const char *buf, size_t length)
{
	s->flags &= ~SLAB_DEBUG_FREE;
	if (buf[0] == '1')
		s->flags |= SLAB_DEBUG_FREE;
	return length;
}
SLAB_ATTR(sanity_checks);

static ssize_t trace_show(struct kmem_cache *s, char *buf)
{
	return sprintf(buf, "%d\n", !!(s->flags & SLAB_TRACE));
}

static ssize_t trace_store(struct kmem_cache *s, const char *buf,
							size_t length)
{
	s->flags &= ~SLAB_TRACE;
	if (buf[0] == '1')
		s->flags |= SLAB_TRACE;
	return length;
}
SLAB_ATTR(trace);

static ssize_t reclaim_account_show(struct kmem_cache *s, char *buf)
{
	return sprintf(buf, "%d\n", !!(s->flags & SLAB_RECLAIM_ACCOUNT));
}

static ssize_t reclaim_account_store(struct kmem_cache *s,
				const char *buf, size_t length)
{
	s->flags &= ~SLAB_RECLAIM_ACCOUNT;
	if (buf[0] == '1')
		s->flags |= SLAB_RECLAIM_ACCOUNT;
	return length;
}
SLAB_ATTR(reclaim_account);

static ssize_t hwcache_align_show(struct kmem_cache *s, char *buf)
{
3918
	return sprintf(buf, "%d\n", !!(s->flags & SLAB_HWCACHE_ALIGN));
C
Christoph Lameter 已提交
3919 3920 3921 3922 3923 3924 3925 3926 3927 3928 3929 3930 3931 3932 3933 3934 3935 3936 3937 3938 3939 3940 3941 3942 3943 3944 3945 3946 3947 3948 3949
}
SLAB_ATTR_RO(hwcache_align);

#ifdef CONFIG_ZONE_DMA
static ssize_t cache_dma_show(struct kmem_cache *s, char *buf)
{
	return sprintf(buf, "%d\n", !!(s->flags & SLAB_CACHE_DMA));
}
SLAB_ATTR_RO(cache_dma);
#endif

static ssize_t destroy_by_rcu_show(struct kmem_cache *s, char *buf)
{
	return sprintf(buf, "%d\n", !!(s->flags & SLAB_DESTROY_BY_RCU));
}
SLAB_ATTR_RO(destroy_by_rcu);

static ssize_t red_zone_show(struct kmem_cache *s, char *buf)
{
	return sprintf(buf, "%d\n", !!(s->flags & SLAB_RED_ZONE));
}

static ssize_t red_zone_store(struct kmem_cache *s,
				const char *buf, size_t length)
{
	if (any_slab_objects(s))
		return -EBUSY;

	s->flags &= ~SLAB_RED_ZONE;
	if (buf[0] == '1')
		s->flags |= SLAB_RED_ZONE;
3950
	calculate_sizes(s, -1);
C
Christoph Lameter 已提交
3951 3952 3953 3954 3955 3956 3957 3958 3959 3960 3961 3962 3963 3964 3965 3966 3967 3968
	return length;
}
SLAB_ATTR(red_zone);

static ssize_t poison_show(struct kmem_cache *s, char *buf)
{
	return sprintf(buf, "%d\n", !!(s->flags & SLAB_POISON));
}

static ssize_t poison_store(struct kmem_cache *s,
				const char *buf, size_t length)
{
	if (any_slab_objects(s))
		return -EBUSY;

	s->flags &= ~SLAB_POISON;
	if (buf[0] == '1')
		s->flags |= SLAB_POISON;
3969
	calculate_sizes(s, -1);
C
Christoph Lameter 已提交
3970 3971 3972 3973 3974 3975 3976 3977 3978 3979 3980 3981 3982 3983 3984 3985 3986 3987
	return length;
}
SLAB_ATTR(poison);

static ssize_t store_user_show(struct kmem_cache *s, char *buf)
{
	return sprintf(buf, "%d\n", !!(s->flags & SLAB_STORE_USER));
}

static ssize_t store_user_store(struct kmem_cache *s,
				const char *buf, size_t length)
{
	if (any_slab_objects(s))
		return -EBUSY;

	s->flags &= ~SLAB_STORE_USER;
	if (buf[0] == '1')
		s->flags |= SLAB_STORE_USER;
3988
	calculate_sizes(s, -1);
C
Christoph Lameter 已提交
3989 3990 3991 3992
	return length;
}
SLAB_ATTR(store_user);

3993 3994 3995 3996 3997 3998 3999 4000
static ssize_t validate_show(struct kmem_cache *s, char *buf)
{
	return 0;
}

static ssize_t validate_store(struct kmem_cache *s,
			const char *buf, size_t length)
{
4001 4002 4003 4004 4005 4006 4007 4008
	int ret = -EINVAL;

	if (buf[0] == '1') {
		ret = validate_slab_cache(s);
		if (ret >= 0)
			ret = length;
	}
	return ret;
4009 4010 4011
}
SLAB_ATTR(validate);

4012 4013 4014 4015 4016 4017 4018 4019 4020 4021 4022 4023 4024 4025 4026 4027 4028 4029 4030
static ssize_t shrink_show(struct kmem_cache *s, char *buf)
{
	return 0;
}

static ssize_t shrink_store(struct kmem_cache *s,
			const char *buf, size_t length)
{
	if (buf[0] == '1') {
		int rc = kmem_cache_shrink(s);

		if (rc)
			return rc;
	} else
		return -EINVAL;
	return length;
}
SLAB_ATTR(shrink);

4031 4032 4033 4034 4035 4036 4037 4038 4039 4040 4041 4042 4043 4044 4045 4046
static ssize_t alloc_calls_show(struct kmem_cache *s, char *buf)
{
	if (!(s->flags & SLAB_STORE_USER))
		return -ENOSYS;
	return list_locations(s, buf, TRACK_ALLOC);
}
SLAB_ATTR_RO(alloc_calls);

static ssize_t free_calls_show(struct kmem_cache *s, char *buf)
{
	if (!(s->flags & SLAB_STORE_USER))
		return -ENOSYS;
	return list_locations(s, buf, TRACK_FREE);
}
SLAB_ATTR_RO(free_calls);

C
Christoph Lameter 已提交
4047
#ifdef CONFIG_NUMA
4048
static ssize_t remote_node_defrag_ratio_show(struct kmem_cache *s, char *buf)
C
Christoph Lameter 已提交
4049
{
4050
	return sprintf(buf, "%d\n", s->remote_node_defrag_ratio / 10);
C
Christoph Lameter 已提交
4051 4052
}

4053
static ssize_t remote_node_defrag_ratio_store(struct kmem_cache *s,
C
Christoph Lameter 已提交
4054 4055
				const char *buf, size_t length)
{
4056 4057 4058 4059 4060 4061 4062
	unsigned long ratio;
	int err;

	err = strict_strtoul(buf, 10, &ratio);
	if (err)
		return err;

4063
	if (ratio <= 100)
4064
		s->remote_node_defrag_ratio = ratio * 10;
C
Christoph Lameter 已提交
4065 4066 4067

	return length;
}
4068
SLAB_ATTR(remote_node_defrag_ratio);
C
Christoph Lameter 已提交
4069 4070
#endif

4071 4072 4073 4074 4075 4076 4077 4078 4079 4080 4081 4082 4083 4084 4085 4086 4087 4088 4089 4090
#ifdef CONFIG_SLUB_STATS
static int show_stat(struct kmem_cache *s, char *buf, enum stat_item si)
{
	unsigned long sum  = 0;
	int cpu;
	int len;
	int *data = kmalloc(nr_cpu_ids * sizeof(int), GFP_KERNEL);

	if (!data)
		return -ENOMEM;

	for_each_online_cpu(cpu) {
		unsigned x = get_cpu_slab(s, cpu)->stat[si];

		data[cpu] = x;
		sum += x;
	}

	len = sprintf(buf, "%lu", sum);

4091
#ifdef CONFIG_SMP
4092 4093
	for_each_online_cpu(cpu) {
		if (data[cpu] && len < PAGE_SIZE - 20)
4094
			len += sprintf(buf + len, " C%d=%u", cpu, data[cpu]);
4095
	}
4096
#endif
4097 4098 4099 4100 4101 4102 4103 4104 4105 4106 4107 4108 4109 4110 4111 4112 4113 4114 4115 4116 4117 4118 4119 4120 4121 4122 4123 4124
	kfree(data);
	return len + sprintf(buf + len, "\n");
}

#define STAT_ATTR(si, text) 					\
static ssize_t text##_show(struct kmem_cache *s, char *buf)	\
{								\
	return show_stat(s, buf, si);				\
}								\
SLAB_ATTR_RO(text);						\

STAT_ATTR(ALLOC_FASTPATH, alloc_fastpath);
STAT_ATTR(ALLOC_SLOWPATH, alloc_slowpath);
STAT_ATTR(FREE_FASTPATH, free_fastpath);
STAT_ATTR(FREE_SLOWPATH, free_slowpath);
STAT_ATTR(FREE_FROZEN, free_frozen);
STAT_ATTR(FREE_ADD_PARTIAL, free_add_partial);
STAT_ATTR(FREE_REMOVE_PARTIAL, free_remove_partial);
STAT_ATTR(ALLOC_FROM_PARTIAL, alloc_from_partial);
STAT_ATTR(ALLOC_SLAB, alloc_slab);
STAT_ATTR(ALLOC_REFILL, alloc_refill);
STAT_ATTR(FREE_SLAB, free_slab);
STAT_ATTR(CPUSLAB_FLUSH, cpuslab_flush);
STAT_ATTR(DEACTIVATE_FULL, deactivate_full);
STAT_ATTR(DEACTIVATE_EMPTY, deactivate_empty);
STAT_ATTR(DEACTIVATE_TO_HEAD, deactivate_to_head);
STAT_ATTR(DEACTIVATE_TO_TAIL, deactivate_to_tail);
STAT_ATTR(DEACTIVATE_REMOTE_FREES, deactivate_remote_frees);
4125
STAT_ATTR(ORDER_FALLBACK, order_fallback);
4126 4127
#endif

P
Pekka Enberg 已提交
4128
static struct attribute *slab_attrs[] = {
C
Christoph Lameter 已提交
4129 4130 4131 4132 4133
	&slab_size_attr.attr,
	&object_size_attr.attr,
	&objs_per_slab_attr.attr,
	&order_attr.attr,
	&objects_attr.attr,
4134 4135
	&objects_partial_attr.attr,
	&total_objects_attr.attr,
C
Christoph Lameter 已提交
4136 4137 4138 4139 4140 4141 4142 4143 4144 4145 4146 4147 4148 4149
	&slabs_attr.attr,
	&partial_attr.attr,
	&cpu_slabs_attr.attr,
	&ctor_attr.attr,
	&aliases_attr.attr,
	&align_attr.attr,
	&sanity_checks_attr.attr,
	&trace_attr.attr,
	&hwcache_align_attr.attr,
	&reclaim_account_attr.attr,
	&destroy_by_rcu_attr.attr,
	&red_zone_attr.attr,
	&poison_attr.attr,
	&store_user_attr.attr,
4150
	&validate_attr.attr,
4151
	&shrink_attr.attr,
4152 4153
	&alloc_calls_attr.attr,
	&free_calls_attr.attr,
C
Christoph Lameter 已提交
4154 4155 4156 4157
#ifdef CONFIG_ZONE_DMA
	&cache_dma_attr.attr,
#endif
#ifdef CONFIG_NUMA
4158
	&remote_node_defrag_ratio_attr.attr,
4159 4160 4161 4162 4163 4164 4165 4166 4167 4168 4169 4170 4171 4172 4173 4174 4175 4176 4177
#endif
#ifdef CONFIG_SLUB_STATS
	&alloc_fastpath_attr.attr,
	&alloc_slowpath_attr.attr,
	&free_fastpath_attr.attr,
	&free_slowpath_attr.attr,
	&free_frozen_attr.attr,
	&free_add_partial_attr.attr,
	&free_remove_partial_attr.attr,
	&alloc_from_partial_attr.attr,
	&alloc_slab_attr.attr,
	&alloc_refill_attr.attr,
	&free_slab_attr.attr,
	&cpuslab_flush_attr.attr,
	&deactivate_full_attr.attr,
	&deactivate_empty_attr.attr,
	&deactivate_to_head_attr.attr,
	&deactivate_to_tail_attr.attr,
	&deactivate_remote_frees_attr.attr,
4178
	&order_fallback_attr.attr,
C
Christoph Lameter 已提交
4179 4180 4181 4182 4183 4184 4185 4186 4187 4188 4189 4190 4191 4192 4193 4194 4195 4196 4197 4198 4199 4200 4201 4202 4203 4204 4205 4206 4207 4208 4209 4210 4211 4212 4213 4214 4215 4216 4217 4218 4219 4220 4221 4222 4223 4224
#endif
	NULL
};

static struct attribute_group slab_attr_group = {
	.attrs = slab_attrs,
};

static ssize_t slab_attr_show(struct kobject *kobj,
				struct attribute *attr,
				char *buf)
{
	struct slab_attribute *attribute;
	struct kmem_cache *s;
	int err;

	attribute = to_slab_attr(attr);
	s = to_slab(kobj);

	if (!attribute->show)
		return -EIO;

	err = attribute->show(s, buf);

	return err;
}

static ssize_t slab_attr_store(struct kobject *kobj,
				struct attribute *attr,
				const char *buf, size_t len)
{
	struct slab_attribute *attribute;
	struct kmem_cache *s;
	int err;

	attribute = to_slab_attr(attr);
	s = to_slab(kobj);

	if (!attribute->store)
		return -EIO;

	err = attribute->store(s, buf, len);

	return err;
}

C
Christoph Lameter 已提交
4225 4226 4227 4228 4229 4230 4231
static void kmem_cache_release(struct kobject *kobj)
{
	struct kmem_cache *s = to_slab(kobj);

	kfree(s);
}

C
Christoph Lameter 已提交
4232 4233 4234 4235 4236 4237 4238
static struct sysfs_ops slab_sysfs_ops = {
	.show = slab_attr_show,
	.store = slab_attr_store,
};

static struct kobj_type slab_ktype = {
	.sysfs_ops = &slab_sysfs_ops,
C
Christoph Lameter 已提交
4239
	.release = kmem_cache_release
C
Christoph Lameter 已提交
4240 4241 4242 4243 4244 4245 4246 4247 4248 4249 4250 4251 4252 4253 4254
};

static int uevent_filter(struct kset *kset, struct kobject *kobj)
{
	struct kobj_type *ktype = get_ktype(kobj);

	if (ktype == &slab_ktype)
		return 1;
	return 0;
}

static struct kset_uevent_ops slab_uevent_ops = {
	.filter = uevent_filter,
};

4255
static struct kset *slab_kset;
C
Christoph Lameter 已提交
4256 4257 4258 4259

#define ID_STR_LENGTH 64

/* Create a unique string id for a slab cache:
C
Christoph Lameter 已提交
4260 4261
 *
 * Format	:[flags-]size
C
Christoph Lameter 已提交
4262 4263 4264 4265 4266 4267 4268 4269 4270 4271 4272 4273 4274 4275 4276 4277 4278 4279 4280 4281 4282 4283 4284 4285 4286 4287 4288 4289 4290 4291 4292 4293 4294 4295 4296 4297 4298 4299 4300 4301 4302 4303 4304 4305 4306 4307
 */
static char *create_unique_id(struct kmem_cache *s)
{
	char *name = kmalloc(ID_STR_LENGTH, GFP_KERNEL);
	char *p = name;

	BUG_ON(!name);

	*p++ = ':';
	/*
	 * First flags affecting slabcache operations. We will only
	 * get here for aliasable slabs so we do not need to support
	 * too many flags. The flags here must cover all flags that
	 * are matched during merging to guarantee that the id is
	 * unique.
	 */
	if (s->flags & SLAB_CACHE_DMA)
		*p++ = 'd';
	if (s->flags & SLAB_RECLAIM_ACCOUNT)
		*p++ = 'a';
	if (s->flags & SLAB_DEBUG_FREE)
		*p++ = 'F';
	if (p != name + 1)
		*p++ = '-';
	p += sprintf(p, "%07d", s->size);
	BUG_ON(p > name + ID_STR_LENGTH - 1);
	return name;
}

static int sysfs_slab_add(struct kmem_cache *s)
{
	int err;
	const char *name;
	int unmergeable;

	if (slab_state < SYSFS)
		/* Defer until later */
		return 0;

	unmergeable = slab_unmergeable(s);
	if (unmergeable) {
		/*
		 * Slabcache can never be merged so we can use the name proper.
		 * This is typically the case for debug situations. In that
		 * case we can catch duplicate names easily.
		 */
4308
		sysfs_remove_link(&slab_kset->kobj, s->name);
C
Christoph Lameter 已提交
4309 4310 4311 4312 4313 4314 4315 4316 4317
		name = s->name;
	} else {
		/*
		 * Create a unique name for the slab as a target
		 * for the symlinks.
		 */
		name = create_unique_id(s);
	}

4318
	s->kobj.kset = slab_kset;
4319 4320 4321
	err = kobject_init_and_add(&s->kobj, &slab_ktype, NULL, name);
	if (err) {
		kobject_put(&s->kobj);
C
Christoph Lameter 已提交
4322
		return err;
4323
	}
C
Christoph Lameter 已提交
4324 4325 4326 4327 4328 4329 4330 4331 4332 4333 4334 4335 4336 4337 4338 4339 4340

	err = sysfs_create_group(&s->kobj, &slab_attr_group);
	if (err)
		return err;
	kobject_uevent(&s->kobj, KOBJ_ADD);
	if (!unmergeable) {
		/* Setup first alias */
		sysfs_slab_alias(s, s->name);
		kfree(name);
	}
	return 0;
}

static void sysfs_slab_remove(struct kmem_cache *s)
{
	kobject_uevent(&s->kobj, KOBJ_REMOVE);
	kobject_del(&s->kobj);
C
Christoph Lameter 已提交
4341
	kobject_put(&s->kobj);
C
Christoph Lameter 已提交
4342 4343 4344 4345 4346 4347 4348 4349 4350 4351 4352 4353
}

/*
 * Need to buffer aliases during bootup until sysfs becomes
 * available lest we loose that information.
 */
struct saved_alias {
	struct kmem_cache *s;
	const char *name;
	struct saved_alias *next;
};

A
Adrian Bunk 已提交
4354
static struct saved_alias *alias_list;
C
Christoph Lameter 已提交
4355 4356 4357 4358 4359 4360 4361 4362 4363

static int sysfs_slab_alias(struct kmem_cache *s, const char *name)
{
	struct saved_alias *al;

	if (slab_state == SYSFS) {
		/*
		 * If we have a leftover link then remove it.
		 */
4364 4365
		sysfs_remove_link(&slab_kset->kobj, name);
		return sysfs_create_link(&slab_kset->kobj, &s->kobj, name);
C
Christoph Lameter 已提交
4366 4367 4368 4369 4370 4371 4372 4373 4374 4375 4376 4377 4378 4379 4380
	}

	al = kmalloc(sizeof(struct saved_alias), GFP_KERNEL);
	if (!al)
		return -ENOMEM;

	al->s = s;
	al->name = name;
	al->next = alias_list;
	alias_list = al;
	return 0;
}

static int __init slab_sysfs_init(void)
{
4381
	struct kmem_cache *s;
C
Christoph Lameter 已提交
4382 4383
	int err;

4384
	slab_kset = kset_create_and_add("slab", &slab_uevent_ops, kernel_kobj);
4385
	if (!slab_kset) {
C
Christoph Lameter 已提交
4386 4387 4388 4389
		printk(KERN_ERR "Cannot register slab subsystem.\n");
		return -ENOSYS;
	}

4390 4391
	slab_state = SYSFS;

4392
	list_for_each_entry(s, &slab_caches, list) {
4393
		err = sysfs_slab_add(s);
4394 4395 4396
		if (err)
			printk(KERN_ERR "SLUB: Unable to add boot slab %s"
						" to sysfs\n", s->name);
4397
	}
C
Christoph Lameter 已提交
4398 4399 4400 4401 4402 4403

	while (alias_list) {
		struct saved_alias *al = alias_list;

		alias_list = alias_list->next;
		err = sysfs_slab_alias(al->s, al->name);
4404 4405 4406
		if (err)
			printk(KERN_ERR "SLUB: Unable to add boot slab alias"
					" %s to sysfs\n", s->name);
C
Christoph Lameter 已提交
4407 4408 4409 4410 4411 4412 4413 4414 4415
		kfree(al);
	}

	resiliency_test();
	return 0;
}

__initcall(slab_sysfs_init);
#endif
P
Pekka J Enberg 已提交
4416 4417 4418 4419

/*
 * The /proc/slabinfo ABI
 */
4420
#ifdef CONFIG_SLABINFO
P
Pekka J Enberg 已提交
4421 4422 4423 4424 4425 4426 4427 4428 4429 4430 4431 4432 4433 4434 4435 4436 4437 4438 4439 4440 4441 4442 4443 4444 4445 4446 4447 4448 4449 4450 4451 4452 4453 4454 4455 4456
static void print_slabinfo_header(struct seq_file *m)
{
	seq_puts(m, "slabinfo - version: 2.1\n");
	seq_puts(m, "# name            <active_objs> <num_objs> <objsize> "
		 "<objperslab> <pagesperslab>");
	seq_puts(m, " : tunables <limit> <batchcount> <sharedfactor>");
	seq_puts(m, " : slabdata <active_slabs> <num_slabs> <sharedavail>");
	seq_putc(m, '\n');
}

static void *s_start(struct seq_file *m, loff_t *pos)
{
	loff_t n = *pos;

	down_read(&slub_lock);
	if (!n)
		print_slabinfo_header(m);

	return seq_list_start(&slab_caches, *pos);
}

static void *s_next(struct seq_file *m, void *p, loff_t *pos)
{
	return seq_list_next(p, &slab_caches, pos);
}

static void s_stop(struct seq_file *m, void *p)
{
	up_read(&slub_lock);
}

static int s_show(struct seq_file *m, void *p)
{
	unsigned long nr_partials = 0;
	unsigned long nr_slabs = 0;
	unsigned long nr_inuse = 0;
4457 4458
	unsigned long nr_objs = 0;
	unsigned long nr_free = 0;
P
Pekka J Enberg 已提交
4459 4460 4461 4462 4463 4464 4465 4466 4467 4468 4469 4470 4471
	struct kmem_cache *s;
	int node;

	s = list_entry(p, struct kmem_cache, list);

	for_each_online_node(node) {
		struct kmem_cache_node *n = get_node(s, node);

		if (!n)
			continue;

		nr_partials += n->nr_partial;
		nr_slabs += atomic_long_read(&n->nr_slabs);
4472 4473
		nr_objs += atomic_long_read(&n->total_objects);
		nr_free += count_partial(n, count_free);
P
Pekka J Enberg 已提交
4474 4475
	}

4476
	nr_inuse = nr_objs - nr_free;
P
Pekka J Enberg 已提交
4477 4478

	seq_printf(m, "%-17s %6lu %6lu %6u %4u %4d", s->name, nr_inuse,
4479 4480
		   nr_objs, s->size, oo_objects(s->oo),
		   (1 << oo_order(s->oo)));
P
Pekka J Enberg 已提交
4481 4482 4483 4484 4485 4486 4487
	seq_printf(m, " : tunables %4u %4u %4u", 0, 0, 0);
	seq_printf(m, " : slabdata %6lu %6lu %6lu", nr_slabs, nr_slabs,
		   0UL);
	seq_putc(m, '\n');
	return 0;
}

4488
static const struct seq_operations slabinfo_op = {
P
Pekka J Enberg 已提交
4489 4490 4491 4492 4493 4494
	.start = s_start,
	.next = s_next,
	.stop = s_stop,
	.show = s_show,
};

4495 4496 4497 4498 4499 4500 4501 4502 4503 4504 4505 4506 4507 4508 4509 4510 4511 4512
static int slabinfo_open(struct inode *inode, struct file *file)
{
	return seq_open(file, &slabinfo_op);
}

static const struct file_operations proc_slabinfo_operations = {
	.open		= slabinfo_open,
	.read		= seq_read,
	.llseek		= seq_lseek,
	.release	= seq_release,
};

static int __init slab_proc_init(void)
{
	proc_create("slabinfo",S_IWUSR|S_IRUGO,NULL,&proc_slabinfo_operations);
	return 0;
}
module_init(slab_proc_init);
4513
#endif /* CONFIG_SLABINFO */