Blame view
mm/slab.c
115 KB
1da177e4c
|
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 |
/* * linux/mm/slab.c * Written by Mark Hemment, 1996/97. * (markhe@nextd.demon.co.uk) * * kmem_cache_destroy() + some cleanup - 1999 Andrea Arcangeli * * Major cleanup, different bufctl logic, per-cpu arrays * (c) 2000 Manfred Spraul * * Cleanup, make the head arrays unconditional, preparation for NUMA * (c) 2002 Manfred Spraul * * An implementation of the Slab Allocator as described in outline in; * UNIX Internals: The New Frontiers by Uresh Vahalia * Pub: Prentice Hall ISBN 0-13-101908-2 * or with a little more detail in; * The Slab Allocator: An Object-Caching Kernel Memory Allocator * Jeff Bonwick (Sun Microsystems). * Presented at: USENIX Summer 1994 Technical Conference * * The memory is organized in caches, one cache for each object type. * (e.g. inode_cache, dentry_cache, buffer_head, vm_area_struct) * Each cache consists out of many slabs (they are small (usually one * page long) and always contiguous), and each slab contains multiple * initialized objects. * * This means, that your constructor is used only for newly allocated |
183ff22bb
|
29 |
* slabs and you must pass objects with the same initializations to |
1da177e4c
|
30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 |
* kmem_cache_free. * * Each cache can only support one memory type (GFP_DMA, GFP_HIGHMEM, * normal). If you need a special memory type, then must create a new * cache for that memory type. * * In order to reduce fragmentation, the slabs are sorted in 3 groups: * full slabs with 0 free objects * partial slabs * empty slabs with no allocated objects * * If partial slabs exist, then new allocations come from these slabs, * otherwise from empty slabs or new slabs are allocated. * * kmem_cache_destroy() CAN CRASH if you try to allocate from the cache * during kmem_cache_destroy(). The caller must prevent concurrent allocs. * * Each cache has a short per-cpu head array, most allocs * and frees go into that array, and if that array overflows, then 1/2 * of the entries in the array are given back into the global cache. * The head array is strictly LIFO and should improve the cache hit rates. * On SMP, it additionally reduces the spinlock operations. * |
a737b3e2f
|
53 |
* The c_cpuarray may not be read with enabled local interrupts - |
1da177e4c
|
54 55 56 57 |
* it's changed with a smp_call_function(). * * SMP synchronization: * constructors and destructors are called without any locking. |
343e0d7a9
|
58 |
* Several members in struct kmem_cache and struct slab never change, they |
1da177e4c
|
59 60 61 62 63 64 65 66 67 68 69 70 |
* are accessed without any locking. * The per-cpu arrays are never accessed from the wrong cpu, no locking, * and local interrupts are disabled so slab code is preempt-safe. * The non-constant members are protected with a per-cache irq spinlock. * * Many thanks to Mark Hemment, who wrote another per-cpu slab patch * in 2000 - many ideas in the current implementation are derived from * his patch. * * Further notes from the original documentation: * * 11 April '97. Started multi-threading - markhe |
fc0abb145
|
71 |
* The global cache-chain is protected by the mutex 'cache_chain_mutex'. |
1da177e4c
|
72 73 74 75 76 77 |
* The sem is only needed when accessing/extending the cache-chain, which * can never happen inside an interrupt (kmem_cache_create(), * kmem_cache_shrink() and kmem_cache_reap()). * * At present, each engine can be growing a cache. This should be blocked. * |
e498be7da
|
78 79 80 81 82 83 84 85 86 |
* 15 March 2005. NUMA slab allocator. * Shai Fultheim <shai@scalex86.org>. * Shobhit Dayal <shobhit@calsoftinc.com> * Alok N Kataria <alokk@calsoftinc.com> * Christoph Lameter <christoph@lameter.com> * * Modified the slab allocator to be node aware on NUMA systems. * Each node has its own list of partial, free and full slabs. * All object allocations for a node occur from node specific slab lists. |
1da177e4c
|
87 |
*/ |
1da177e4c
|
88 89 |
#include <linux/slab.h> #include <linux/mm.h> |
c9cf55285
|
90 |
#include <linux/poison.h> |
1da177e4c
|
91 92 93 94 95 |
#include <linux/swap.h> #include <linux/cache.h> #include <linux/interrupt.h> #include <linux/init.h> #include <linux/compiler.h> |
101a50019
|
96 |
#include <linux/cpuset.h> |
a0ec95a8e
|
97 |
#include <linux/proc_fs.h> |
1da177e4c
|
98 99 100 101 102 103 104 |
#include <linux/seq_file.h> #include <linux/notifier.h> #include <linux/kallsyms.h> #include <linux/cpu.h> #include <linux/sysctl.h> #include <linux/module.h> #include <linux/rcupdate.h> |
543537bd9
|
105 |
#include <linux/string.h> |
138ae6631
|
106 |
#include <linux/uaccess.h> |
e498be7da
|
107 |
#include <linux/nodemask.h> |
dc85da15d
|
108 |
#include <linux/mempolicy.h> |
fc0abb145
|
109 |
#include <linux/mutex.h> |
8a8b6502f
|
110 |
#include <linux/fault-inject.h> |
e7eebaf6a
|
111 |
#include <linux/rtmutex.h> |
6a2d7a955
|
112 |
#include <linux/reciprocal_div.h> |
3ac7fe5a4
|
113 |
#include <linux/debugobjects.h> |
1da177e4c
|
114 |
|
1da177e4c
|
115 116 117 118 119 |
#include <asm/cacheflush.h> #include <asm/tlbflush.h> #include <asm/page.h> /* |
50953fe9e
|
120 |
* DEBUG - 1 for kmem_cache_create() to honour; SLAB_RED_ZONE & SLAB_POISON. |
1da177e4c
|
121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 |
* 0 for faster, smaller code (especially in the critical paths). * * STATS - 1 to collect stats for /proc/slabinfo. * 0 for faster, smaller code (especially in the critical paths). * * FORCED_DEBUG - 1 enables SLAB_RED_ZONE and SLAB_POISON (if possible) */ #ifdef CONFIG_DEBUG_SLAB #define DEBUG 1 #define STATS 1 #define FORCED_DEBUG 1 #else #define DEBUG 0 #define STATS 0 #define FORCED_DEBUG 0 #endif |
1da177e4c
|
138 139 |
/* Shouldn't this be in a header file somewhere? */ #define BYTES_PER_WORD sizeof(void *) |
87a927c71
|
140 |
#define REDZONE_ALIGN max(BYTES_PER_WORD, __alignof__(unsigned long long)) |
1da177e4c
|
141 |
|
1da177e4c
|
142 143 144 145 146 147 |
#ifndef ARCH_KMALLOC_MINALIGN /* * Enforce a minimum alignment for the kmalloc caches. * Usually, the kmalloc caches are cache_line_size() aligned, except when * DEBUG and FORCED_DEBUG are enabled, then they are BYTES_PER_WORD aligned. * Some archs want to perform DMA into kmalloc caches and need a guaranteed |
b46b8f19c
|
148 149 150 |
* alignment larger than the alignment of a 64-bit integer. * ARCH_KMALLOC_MINALIGN allows that. * Note that increasing this value may disable some debug features. |
1da177e4c
|
151 |
*/ |
b46b8f19c
|
152 |
#define ARCH_KMALLOC_MINALIGN __alignof__(unsigned long long) |
1da177e4c
|
153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 |
#endif #ifndef ARCH_SLAB_MINALIGN /* * Enforce a minimum alignment for all caches. * Intended for archs that get misalignment faults even for BYTES_PER_WORD * aligned buffers. Includes ARCH_KMALLOC_MINALIGN. * If possible: Do not enable this flag for CONFIG_DEBUG_SLAB, it disables * some debug features. */ #define ARCH_SLAB_MINALIGN 0 #endif #ifndef ARCH_KMALLOC_FLAGS #define ARCH_KMALLOC_FLAGS SLAB_HWCACHE_ALIGN #endif /* Legal flag mask for kmem_cache_create(). */ #if DEBUG |
50953fe9e
|
172 |
# define CREATE_MASK (SLAB_RED_ZONE | \ |
1da177e4c
|
173 |
SLAB_POISON | SLAB_HWCACHE_ALIGN | \ |
ac2b898ca
|
174 |
SLAB_CACHE_DMA | \ |
5af608399
|
175 |
SLAB_STORE_USER | \ |
1da177e4c
|
176 |
SLAB_RECLAIM_ACCOUNT | SLAB_PANIC | \ |
3ac7fe5a4
|
177 178 |
SLAB_DESTROY_BY_RCU | SLAB_MEM_SPREAD | \ SLAB_DEBUG_OBJECTS) |
1da177e4c
|
179 |
#else |
ac2b898ca
|
180 |
# define CREATE_MASK (SLAB_HWCACHE_ALIGN | \ |
5af608399
|
181 |
SLAB_CACHE_DMA | \ |
1da177e4c
|
182 |
SLAB_RECLAIM_ACCOUNT | SLAB_PANIC | \ |
3ac7fe5a4
|
183 184 |
SLAB_DESTROY_BY_RCU | SLAB_MEM_SPREAD | \ SLAB_DEBUG_OBJECTS) |
1da177e4c
|
185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 |
#endif /* * kmem_bufctl_t: * * Bufctl's are used for linking objs within a slab * linked offsets. * * This implementation relies on "struct page" for locating the cache & * slab an object belongs to. * This allows the bufctl structure to be small (one int), but limits * the number of objects a slab (not a cache) can contain when off-slab * bufctls are used. The limit is the size of the largest general cache * that does not use off-slab slabs. * For 32bit archs with 4 kB pages, is this 56. * This is not serious, as it is only for large objects, when it is unwise * to have too many per slab. * Note: This limit can be raised by introducing a general cache whose size * is less than 512 (PAGE_SIZE<<3), but greater than 256. */ |
fa5b08d5f
|
205 |
typedef unsigned int kmem_bufctl_t; |
1da177e4c
|
206 207 |
#define BUFCTL_END (((kmem_bufctl_t)(~0U))-0) #define BUFCTL_FREE (((kmem_bufctl_t)(~0U))-1) |
871751e25
|
208 209 |
#define BUFCTL_ACTIVE (((kmem_bufctl_t)(~0U))-2) #define SLAB_LIMIT (((kmem_bufctl_t)(~0U))-3) |
1da177e4c
|
210 |
|
1da177e4c
|
211 212 213 214 215 216 217 218 |
/* * struct slab * * Manages the objs in a slab. Placed either at the beginning of mem allocated * for a slab, or allocated from an general cache. * Slabs are chained into three list: fully used, partial, fully free slabs. */ struct slab { |
b28a02de8
|
219 220 221 222 223 224 |
struct list_head list; unsigned long colouroff; void *s_mem; /* including colour offset */ unsigned int inuse; /* num of objs active in slab */ kmem_bufctl_t free; unsigned short nodeid; |
1da177e4c
|
225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 |
}; /* * struct slab_rcu * * slab_destroy on a SLAB_DESTROY_BY_RCU cache uses this structure to * arrange for kmem_freepages to be called via RCU. This is useful if * we need to approach a kernel structure obliquely, from its address * obtained without the usual locking. We can lock the structure to * stabilize it and check it's still at the given address, only if we * can be sure that the memory has not been meanwhile reused for some * other kind of object (which our subsystem's lock might corrupt). * * rcu_read_lock before reading the address, then rcu_read_unlock after * taking the spinlock within the structure expected at that address. * * We assume struct slab_rcu can overlay struct slab when destroying. */ struct slab_rcu { |
b28a02de8
|
244 |
struct rcu_head head; |
343e0d7a9
|
245 |
struct kmem_cache *cachep; |
b28a02de8
|
246 |
void *addr; |
1da177e4c
|
247 248 249 250 251 |
}; /* * struct array_cache * |
1da177e4c
|
252 253 254 255 256 257 258 259 260 261 262 263 264 265 |
* Purpose: * - LIFO ordering, to hand out cache-warm objects from _alloc * - reduce the number of linked list operations * - reduce spinlock operations * * The limit is stored in the per-cpu structure to reduce the data cache * footprint. * */ struct array_cache { unsigned int avail; unsigned int limit; unsigned int batchcount; unsigned int touched; |
e498be7da
|
266 |
spinlock_t lock; |
bda5b655f
|
267 |
void *entry[]; /* |
a737b3e2f
|
268 269 270 |
* Must have this definition in here for the proper * alignment of array_cache. Also simplifies accessing * the entries. |
a737b3e2f
|
271 |
*/ |
1da177e4c
|
272 |
}; |
a737b3e2f
|
273 274 275 |
/* * bootstrap: The caches do not work without cpuarrays anymore, but the * cpuarrays are allocated from the generic caches... |
1da177e4c
|
276 277 278 279 |
*/ #define BOOT_CPUCACHE_ENTRIES 1 struct arraycache_init { struct array_cache cache; |
b28a02de8
|
280 |
void *entries[BOOT_CPUCACHE_ENTRIES]; |
1da177e4c
|
281 282 283 |
}; /* |
e498be7da
|
284 |
* The slab lists for all objects. |
1da177e4c
|
285 286 |
*/ struct kmem_list3 { |
b28a02de8
|
287 288 289 290 |
struct list_head slabs_partial; /* partial list first, better asm code */ struct list_head slabs_full; struct list_head slabs_free; unsigned long free_objects; |
b28a02de8
|
291 |
unsigned int free_limit; |
2e1217cf9
|
292 |
unsigned int colour_next; /* Per-node cache coloring */ |
b28a02de8
|
293 294 295 |
spinlock_t list_lock; struct array_cache *shared; /* shared per node */ struct array_cache **alien; /* on other nodes */ |
35386e3b0
|
296 297 |
unsigned long next_reap; /* updated without locking */ int free_touched; /* updated without locking */ |
1da177e4c
|
298 |
}; |
e498be7da
|
299 300 301 |
/* * Need this for bootstrapping a per node allocator. */ |
556a169da
|
302 |
#define NUM_INIT_LISTS (3 * MAX_NUMNODES) |
e498be7da
|
303 304 |
struct kmem_list3 __initdata initkmem_list3[NUM_INIT_LISTS]; #define CACHE_CACHE 0 |
556a169da
|
305 306 |
#define SIZE_AC MAX_NUMNODES #define SIZE_L3 (2 * MAX_NUMNODES) |
e498be7da
|
307 |
|
ed11d9eb2
|
308 309 310 311 |
static int drain_freelist(struct kmem_cache *cache, struct kmem_list3 *l3, int tofree); static void free_block(struct kmem_cache *cachep, void **objpp, int len, int node); |
2ed3a4ef9
|
312 |
static int enable_cpucache(struct kmem_cache *cachep); |
65f27f384
|
313 |
static void cache_reap(struct work_struct *unused); |
ed11d9eb2
|
314 |
|
e498be7da
|
315 |
/* |
a737b3e2f
|
316 317 |
* This function must be completely optimized away if a constant is passed to * it. Mostly the same as what is in linux/slab.h except it returns an index. |
e498be7da
|
318 |
*/ |
7243cc05b
|
319 |
static __always_inline int index_of(const size_t size) |
e498be7da
|
320 |
{ |
5ec8a847b
|
321 |
extern void __bad_size(void); |
e498be7da
|
322 323 324 325 326 327 328 329 |
if (__builtin_constant_p(size)) { int i = 0; #define CACHE(x) \ if (size <=x) \ return i; \ else \ i++; |
1c61fc40f
|
330 |
#include <linux/kmalloc_sizes.h> |
e498be7da
|
331 |
#undef CACHE |
5ec8a847b
|
332 |
__bad_size(); |
7243cc05b
|
333 |
} else |
5ec8a847b
|
334 |
__bad_size(); |
e498be7da
|
335 336 |
return 0; } |
e0a427267
|
337 |
static int slab_early_init = 1; |
e498be7da
|
338 339 |
#define INDEX_AC index_of(sizeof(struct arraycache_init)) #define INDEX_L3 index_of(sizeof(struct kmem_list3)) |
1da177e4c
|
340 |
|
5295a74cc
|
341 |
static void kmem_list3_init(struct kmem_list3 *parent) |
e498be7da
|
342 343 344 345 346 347 |
{ INIT_LIST_HEAD(&parent->slabs_full); INIT_LIST_HEAD(&parent->slabs_partial); INIT_LIST_HEAD(&parent->slabs_free); parent->shared = NULL; parent->alien = NULL; |
2e1217cf9
|
348 |
parent->colour_next = 0; |
e498be7da
|
349 350 351 352 |
spin_lock_init(&parent->list_lock); parent->free_objects = 0; parent->free_touched = 0; } |
a737b3e2f
|
353 354 355 356 |
#define MAKE_LIST(cachep, listp, slab, nodeid) \ do { \ INIT_LIST_HEAD(listp); \ list_splice(&(cachep->nodelists[nodeid]->slab), listp); \ |
e498be7da
|
357 |
} while (0) |
a737b3e2f
|
358 359 |
#define MAKE_ALL_LISTS(cachep, ptr, nodeid) \ do { \ |
e498be7da
|
360 361 362 363 |
MAKE_LIST((cachep), (&(ptr)->slabs_full), slabs_full, nodeid); \ MAKE_LIST((cachep), (&(ptr)->slabs_partial), slabs_partial, nodeid); \ MAKE_LIST((cachep), (&(ptr)->slabs_free), slabs_free, nodeid); \ } while (0) |
1da177e4c
|
364 365 |
/* |
343e0d7a9
|
366 |
* struct kmem_cache |
1da177e4c
|
367 368 369 |
* * manages a cache. */ |
b28a02de8
|
370 |
|
2109a2d1b
|
371 |
struct kmem_cache { |
1da177e4c
|
372 |
/* 1) per-cpu data, touched during every alloc/free */ |
b28a02de8
|
373 |
struct array_cache *array[NR_CPUS]; |
b5d8ca7c5
|
374 |
/* 2) Cache tunables. Protected by cache_chain_mutex */ |
b28a02de8
|
375 376 377 |
unsigned int batchcount; unsigned int limit; unsigned int shared; |
b5d8ca7c5
|
378 |
|
3dafccf22
|
379 |
unsigned int buffer_size; |
6a2d7a955
|
380 |
u32 reciprocal_buffer_size; |
b5d8ca7c5
|
381 |
/* 3) touched by every alloc & free from the backend */ |
b5d8ca7c5
|
382 |
|
a737b3e2f
|
383 384 |
unsigned int flags; /* constant flags */ unsigned int num; /* # of objs per slab */ |
1da177e4c
|
385 |
|
b5d8ca7c5
|
386 |
/* 4) cache_grow/shrink */ |
1da177e4c
|
387 |
/* order of pgs per slab (2^n) */ |
b28a02de8
|
388 |
unsigned int gfporder; |
1da177e4c
|
389 390 |
/* force GFP flags, e.g. GFP_DMA */ |
b28a02de8
|
391 |
gfp_t gfpflags; |
1da177e4c
|
392 |
|
a737b3e2f
|
393 |
size_t colour; /* cache colouring range */ |
b28a02de8
|
394 |
unsigned int colour_off; /* colour offset */ |
343e0d7a9
|
395 |
struct kmem_cache *slabp_cache; |
b28a02de8
|
396 |
unsigned int slab_size; |
a737b3e2f
|
397 |
unsigned int dflags; /* dynamic flags */ |
1da177e4c
|
398 399 |
/* constructor func */ |
51cc50685
|
400 |
void (*ctor)(void *obj); |
1da177e4c
|
401 |
|
b5d8ca7c5
|
402 |
/* 5) cache creation/removal */ |
b28a02de8
|
403 404 |
const char *name; struct list_head next; |
1da177e4c
|
405 |
|
b5d8ca7c5
|
406 |
/* 6) statistics */ |
1da177e4c
|
407 |
#if STATS |
b28a02de8
|
408 409 410 411 412 413 414 415 416 |
unsigned long num_active; unsigned long num_allocations; unsigned long high_mark; unsigned long grown; unsigned long reaped; unsigned long errors; unsigned long max_freeable; unsigned long node_allocs; unsigned long node_frees; |
fb7faf331
|
417 |
unsigned long node_overflow; |
b28a02de8
|
418 419 420 421 |
atomic_t allochit; atomic_t allocmiss; atomic_t freehit; atomic_t freemiss; |
1da177e4c
|
422 423 |
#endif #if DEBUG |
3dafccf22
|
424 425 426 427 428 429 430 431 |
/* * If debugging is enabled, then the allocator can add additional * fields and/or padding to every object. buffer_size contains the total * object size including these internal fields, the following two * variables contain the offset to the user object and its size. */ int obj_offset; int obj_size; |
1da177e4c
|
432 |
#endif |
8da3430d8
|
433 434 435 436 437 438 439 440 441 442 443 |
/* * We put nodelists[] at the end of kmem_cache, because we want to size * this array to nr_node_ids slots instead of MAX_NUMNODES * (see kmem_cache_init()) * We still use [MAX_NUMNODES] and not [1] or [0] because cache_cache * is statically defined, so we reserve the max number of nodes. */ struct kmem_list3 *nodelists[MAX_NUMNODES]; /* * Do not add fields after nodelists[] */ |
1da177e4c
|
444 445 446 447 448 449 |
}; #define CFLGS_OFF_SLAB (0x80000000UL) #define OFF_SLAB(x) ((x)->flags & CFLGS_OFF_SLAB) #define BATCHREFILL_LIMIT 16 |
a737b3e2f
|
450 451 452 |
/* * Optimization question: fewer reaps means less probability for unnessary * cpucache drain/refill cycles. |
1da177e4c
|
453 |
* |
dc6f3f276
|
454 |
* OTOH the cpuarrays can contain lots of objects, |
1da177e4c
|
455 456 457 458 459 460 461 462 463 464 |
* which could lock up otherwise freeable slabs. */ #define REAPTIMEOUT_CPUC (2*HZ) #define REAPTIMEOUT_LIST3 (4*HZ) #if STATS #define STATS_INC_ACTIVE(x) ((x)->num_active++) #define STATS_DEC_ACTIVE(x) ((x)->num_active--) #define STATS_INC_ALLOCED(x) ((x)->num_allocations++) #define STATS_INC_GROWN(x) ((x)->grown++) |
ed11d9eb2
|
465 |
#define STATS_ADD_REAPED(x,y) ((x)->reaped += (y)) |
a737b3e2f
|
466 467 468 469 470 |
#define STATS_SET_HIGH(x) \ do { \ if ((x)->num_active > (x)->high_mark) \ (x)->high_mark = (x)->num_active; \ } while (0) |
1da177e4c
|
471 472 |
#define STATS_INC_ERR(x) ((x)->errors++) #define STATS_INC_NODEALLOCS(x) ((x)->node_allocs++) |
e498be7da
|
473 |
#define STATS_INC_NODEFREES(x) ((x)->node_frees++) |
fb7faf331
|
474 |
#define STATS_INC_ACOVERFLOW(x) ((x)->node_overflow++) |
a737b3e2f
|
475 476 477 478 479 |
#define STATS_SET_FREEABLE(x, i) \ do { \ if ((x)->max_freeable < i) \ (x)->max_freeable = i; \ } while (0) |
1da177e4c
|
480 481 482 483 484 485 486 487 488 |
#define STATS_INC_ALLOCHIT(x) atomic_inc(&(x)->allochit) #define STATS_INC_ALLOCMISS(x) atomic_inc(&(x)->allocmiss) #define STATS_INC_FREEHIT(x) atomic_inc(&(x)->freehit) #define STATS_INC_FREEMISS(x) atomic_inc(&(x)->freemiss) #else #define STATS_INC_ACTIVE(x) do { } while (0) #define STATS_DEC_ACTIVE(x) do { } while (0) #define STATS_INC_ALLOCED(x) do { } while (0) #define STATS_INC_GROWN(x) do { } while (0) |
ed11d9eb2
|
489 |
#define STATS_ADD_REAPED(x,y) do { } while (0) |
1da177e4c
|
490 491 492 |
#define STATS_SET_HIGH(x) do { } while (0) #define STATS_INC_ERR(x) do { } while (0) #define STATS_INC_NODEALLOCS(x) do { } while (0) |
e498be7da
|
493 |
#define STATS_INC_NODEFREES(x) do { } while (0) |
fb7faf331
|
494 |
#define STATS_INC_ACOVERFLOW(x) do { } while (0) |
a737b3e2f
|
495 |
#define STATS_SET_FREEABLE(x, i) do { } while (0) |
1da177e4c
|
496 497 498 499 500 501 502 |
#define STATS_INC_ALLOCHIT(x) do { } while (0) #define STATS_INC_ALLOCMISS(x) do { } while (0) #define STATS_INC_FREEHIT(x) do { } while (0) #define STATS_INC_FREEMISS(x) do { } while (0) #endif #if DEBUG |
1da177e4c
|
503 |
|
a737b3e2f
|
504 505 |
/* * memory layout of objects: |
1da177e4c
|
506 |
* 0 : objp |
3dafccf22
|
507 |
* 0 .. cachep->obj_offset - BYTES_PER_WORD - 1: padding. This ensures that |
1da177e4c
|
508 509 |
* the end of an object is aligned with the end of the real * allocation. Catches writes behind the end of the allocation. |
3dafccf22
|
510 |
* cachep->obj_offset - BYTES_PER_WORD .. cachep->obj_offset - 1: |
1da177e4c
|
511 |
* redzone word. |
3dafccf22
|
512 513 |
* cachep->obj_offset: The real object. * cachep->buffer_size - 2* BYTES_PER_WORD: redzone word [BYTES_PER_WORD long] |
a737b3e2f
|
514 515 |
* cachep->buffer_size - 1* BYTES_PER_WORD: last caller address * [BYTES_PER_WORD long] |
1da177e4c
|
516 |
*/ |
343e0d7a9
|
517 |
static int obj_offset(struct kmem_cache *cachep) |
1da177e4c
|
518 |
{ |
3dafccf22
|
519 |
return cachep->obj_offset; |
1da177e4c
|
520 |
} |
343e0d7a9
|
521 |
static int obj_size(struct kmem_cache *cachep) |
1da177e4c
|
522 |
{ |
3dafccf22
|
523 |
return cachep->obj_size; |
1da177e4c
|
524 |
} |
b46b8f19c
|
525 |
static unsigned long long *dbg_redzone1(struct kmem_cache *cachep, void *objp) |
1da177e4c
|
526 527 |
{ BUG_ON(!(cachep->flags & SLAB_RED_ZONE)); |
b46b8f19c
|
528 529 |
return (unsigned long long*) (objp + obj_offset(cachep) - sizeof(unsigned long long)); |
1da177e4c
|
530 |
} |
b46b8f19c
|
531 |
static unsigned long long *dbg_redzone2(struct kmem_cache *cachep, void *objp) |
1da177e4c
|
532 533 534 |
{ BUG_ON(!(cachep->flags & SLAB_RED_ZONE)); if (cachep->flags & SLAB_STORE_USER) |
b46b8f19c
|
535 536 |
return (unsigned long long *)(objp + cachep->buffer_size - sizeof(unsigned long long) - |
87a927c71
|
537 |
REDZONE_ALIGN); |
b46b8f19c
|
538 539 |
return (unsigned long long *) (objp + cachep->buffer_size - sizeof(unsigned long long)); |
1da177e4c
|
540 |
} |
343e0d7a9
|
541 |
static void **dbg_userword(struct kmem_cache *cachep, void *objp) |
1da177e4c
|
542 543 |
{ BUG_ON(!(cachep->flags & SLAB_STORE_USER)); |
3dafccf22
|
544 |
return (void **)(objp + cachep->buffer_size - BYTES_PER_WORD); |
1da177e4c
|
545 546 547 |
} #else |
3dafccf22
|
548 549 |
#define obj_offset(x) 0 #define obj_size(cachep) (cachep->buffer_size) |
b46b8f19c
|
550 551 |
#define dbg_redzone1(cachep, objp) ({BUG(); (unsigned long long *)NULL;}) #define dbg_redzone2(cachep, objp) ({BUG(); (unsigned long long *)NULL;}) |
1da177e4c
|
552 553 554 555 556 |
#define dbg_userword(cachep, objp) ({BUG(); (void **)NULL;}) #endif /* |
1da177e4c
|
557 558 559 560 561 |
* Do not go above this order unless 0 objects fit into the slab. */ #define BREAK_GFP_ORDER_HI 1 #define BREAK_GFP_ORDER_LO 0 static int slab_break_gfp_order = BREAK_GFP_ORDER_LO; |
a737b3e2f
|
562 563 564 565 |
/* * Functions for storing/retrieving the cachep and or slab from the page * allocator. These are used to find the slab an obj belongs to. With kfree(), * these are used to find the cache which an obj belongs to. |
1da177e4c
|
566 |
*/ |
065d41cb2
|
567 568 569 570 571 572 573 |
static inline void page_set_cache(struct page *page, struct kmem_cache *cache) { page->lru.next = (struct list_head *)cache; } static inline struct kmem_cache *page_get_cache(struct page *page) { |
d85f33855
|
574 |
page = compound_head(page); |
ddc2e812d
|
575 |
BUG_ON(!PageSlab(page)); |
065d41cb2
|
576 577 578 579 580 581 582 583 584 585 |
return (struct kmem_cache *)page->lru.next; } static inline void page_set_slab(struct page *page, struct slab *slab) { page->lru.prev = (struct list_head *)slab; } static inline struct slab *page_get_slab(struct page *page) { |
ddc2e812d
|
586 |
BUG_ON(!PageSlab(page)); |
065d41cb2
|
587 588 |
return (struct slab *)page->lru.prev; } |
1da177e4c
|
589 |
|
6ed5eb221
|
590 591 |
static inline struct kmem_cache *virt_to_cache(const void *obj) { |
b49af68ff
|
592 |
struct page *page = virt_to_head_page(obj); |
6ed5eb221
|
593 594 595 596 597 |
return page_get_cache(page); } static inline struct slab *virt_to_slab(const void *obj) { |
b49af68ff
|
598 |
struct page *page = virt_to_head_page(obj); |
6ed5eb221
|
599 600 |
return page_get_slab(page); } |
8fea4e96a
|
601 602 603 604 605 |
static inline void *index_to_obj(struct kmem_cache *cache, struct slab *slab, unsigned int idx) { return slab->s_mem + cache->buffer_size * idx; } |
6a2d7a955
|
606 607 608 609 610 611 612 613 |
/* * We want to avoid an expensive divide : (offset / cache->buffer_size) * Using the fact that buffer_size is a constant for a particular cache, * we can replace (offset / cache->buffer_size) by * reciprocal_divide(offset, cache->reciprocal_buffer_size) */ static inline unsigned int obj_to_index(const struct kmem_cache *cache, const struct slab *slab, void *obj) |
8fea4e96a
|
614 |
{ |
6a2d7a955
|
615 616 |
u32 offset = (obj - slab->s_mem); return reciprocal_divide(offset, cache->reciprocal_buffer_size); |
8fea4e96a
|
617 |
} |
a737b3e2f
|
618 619 620 |
/* * These are the default caches for kmalloc. Custom caches can have other sizes. */ |
1da177e4c
|
621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 |
struct cache_sizes malloc_sizes[] = { #define CACHE(x) { .cs_size = (x) }, #include <linux/kmalloc_sizes.h> CACHE(ULONG_MAX) #undef CACHE }; EXPORT_SYMBOL(malloc_sizes); /* Must match cache_sizes above. Out of line to keep cache footprint low. */ struct cache_names { char *name; char *name_dma; }; static struct cache_names __initdata cache_names[] = { #define CACHE(x) { .name = "size-" #x, .name_dma = "size-" #x "(DMA)" }, #include <linux/kmalloc_sizes.h> |
b28a02de8
|
638 |
{NULL,} |
1da177e4c
|
639 640 641 642 |
#undef CACHE }; static struct arraycache_init initarray_cache __initdata = |
b28a02de8
|
643 |
{ {0, BOOT_CPUCACHE_ENTRIES, 1, 0} }; |
1da177e4c
|
644 |
static struct arraycache_init initarray_generic = |
b28a02de8
|
645 |
{ {0, BOOT_CPUCACHE_ENTRIES, 1, 0} }; |
1da177e4c
|
646 647 |
/* internal cache of cache description objs */ |
343e0d7a9
|
648 |
static struct kmem_cache cache_cache = { |
b28a02de8
|
649 650 651 |
.batchcount = 1, .limit = BOOT_CPUCACHE_ENTRIES, .shared = 1, |
343e0d7a9
|
652 |
.buffer_size = sizeof(struct kmem_cache), |
b28a02de8
|
653 |
.name = "kmem_cache", |
1da177e4c
|
654 |
}; |
056c62418
|
655 |
#define BAD_ALIEN_MAGIC 0x01020304ul |
f1aaee53f
|
656 657 658 659 660 661 662 663 |
#ifdef CONFIG_LOCKDEP /* * Slab sometimes uses the kmalloc slabs to store the slab headers * for other slabs "off slab". * The locking for this is tricky in that it nests within the locks * of all other slabs in a few places; to deal with this special * locking we put on-slab caches into a separate lock-class. |
056c62418
|
664 665 666 667 |
* * We set lock class for alien array caches which are up during init. * The lock annotation will be lost if all cpus of a node goes down and * then comes back up during hotplug |
f1aaee53f
|
668 |
*/ |
056c62418
|
669 670 671 672 |
static struct lock_class_key on_slab_l3_key; static struct lock_class_key on_slab_alc_key; static inline void init_lock_keys(void) |
f1aaee53f
|
673 |
|
f1aaee53f
|
674 675 |
{ int q; |
056c62418
|
676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 |
struct cache_sizes *s = malloc_sizes; while (s->cs_size != ULONG_MAX) { for_each_node(q) { struct array_cache **alc; int r; struct kmem_list3 *l3 = s->cs_cachep->nodelists[q]; if (!l3 || OFF_SLAB(s->cs_cachep)) continue; lockdep_set_class(&l3->list_lock, &on_slab_l3_key); alc = l3->alien; /* * FIXME: This check for BAD_ALIEN_MAGIC * should go away when common slab code is taught to * work even without alien caches. * Currently, non NUMA code returns BAD_ALIEN_MAGIC * for alloc_alien_cache, */ if (!alc || (unsigned long)alc == BAD_ALIEN_MAGIC) continue; for_each_node(r) { if (alc[r]) lockdep_set_class(&alc[r]->lock, &on_slab_alc_key); } } s++; |
f1aaee53f
|
703 704 |
} } |
f1aaee53f
|
705 |
#else |
056c62418
|
706 |
static inline void init_lock_keys(void) |
f1aaee53f
|
707 708 709 |
{ } #endif |
8f5be20bf
|
710 |
/* |
95402b382
|
711 |
* Guard access to the cache-chain. |
8f5be20bf
|
712 |
*/ |
fc0abb145
|
713 |
static DEFINE_MUTEX(cache_chain_mutex); |
1da177e4c
|
714 715 716 |
static struct list_head cache_chain; /* |
1da177e4c
|
717 718 719 720 721 |
* chicken and egg problem: delay the per-cpu array allocation * until the general caches are up. */ static enum { NONE, |
e498be7da
|
722 723 |
PARTIAL_AC, PARTIAL_L3, |
1da177e4c
|
724 725 |
FULL } g_cpucache_up; |
39d24e642
|
726 727 728 729 730 731 732 |
/* * used by boot code to determine if it can use slab based allocator */ int slab_is_available(void) { return g_cpucache_up == FULL; } |
52bad64d9
|
733 |
static DEFINE_PER_CPU(struct delayed_work, reap_work); |
1da177e4c
|
734 |
|
343e0d7a9
|
735 |
static inline struct array_cache *cpu_cache_get(struct kmem_cache *cachep) |
1da177e4c
|
736 737 738 |
{ return cachep->array[smp_processor_id()]; } |
a737b3e2f
|
739 740 |
static inline struct kmem_cache *__find_general_cachep(size_t size, gfp_t gfpflags) |
1da177e4c
|
741 742 743 744 745 |
{ struct cache_sizes *csizep = malloc_sizes; #if DEBUG /* This happens if someone tries to call |
b28a02de8
|
746 747 748 |
* kmem_cache_create(), or __kmalloc(), before * the generic caches are initialized. */ |
c7e43c78a
|
749 |
BUG_ON(malloc_sizes[INDEX_AC].cs_cachep == NULL); |
1da177e4c
|
750 |
#endif |
6cb8f9132
|
751 752 |
if (!size) return ZERO_SIZE_PTR; |
1da177e4c
|
753 754 755 756 |
while (size > csizep->cs_size) csizep++; /* |
0abf40c1a
|
757 |
* Really subtle: The last entry with cs->cs_size==ULONG_MAX |
1da177e4c
|
758 759 760 |
* has cs_{dma,}cachep==NULL. Thus no special case * for large kmalloc calls required. */ |
4b51d6698
|
761 |
#ifdef CONFIG_ZONE_DMA |
1da177e4c
|
762 763 |
if (unlikely(gfpflags & GFP_DMA)) return csizep->cs_dmacachep; |
4b51d6698
|
764 |
#endif |
1da177e4c
|
765 766 |
return csizep->cs_cachep; } |
b221385bc
|
767 |
static struct kmem_cache *kmem_find_general_cachep(size_t size, gfp_t gfpflags) |
97e2bde47
|
768 769 770 |
{ return __find_general_cachep(size, gfpflags); } |
97e2bde47
|
771 |
|
fbaccacff
|
772 |
static size_t slab_mgmt_size(size_t nr_objs, size_t align) |
1da177e4c
|
773 |
{ |
fbaccacff
|
774 775 |
return ALIGN(sizeof(struct slab)+nr_objs*sizeof(kmem_bufctl_t), align); } |
1da177e4c
|
776 |
|
a737b3e2f
|
777 778 779 |
/* * Calculate the number of objects and left-over bytes for a given buffer size. */ |
fbaccacff
|
780 781 782 783 784 785 786 |
static void cache_estimate(unsigned long gfporder, size_t buffer_size, size_t align, int flags, size_t *left_over, unsigned int *num) { int nr_objs; size_t mgmt_size; size_t slab_size = PAGE_SIZE << gfporder; |
1da177e4c
|
787 |
|
fbaccacff
|
788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 |
/* * The slab management structure can be either off the slab or * on it. For the latter case, the memory allocated for a * slab is used for: * * - The struct slab * - One kmem_bufctl_t for each object * - Padding to respect alignment of @align * - @buffer_size bytes for each object * * If the slab management structure is off the slab, then the * alignment will already be calculated into the size. Because * the slabs are all pages aligned, the objects will be at the * correct alignment when allocated. */ if (flags & CFLGS_OFF_SLAB) { mgmt_size = 0; nr_objs = slab_size / buffer_size; if (nr_objs > SLAB_LIMIT) nr_objs = SLAB_LIMIT; } else { /* * Ignore padding for the initial guess. The padding * is at most @align-1 bytes, and @buffer_size is at * least @align. In the worst case, this result will * be one greater than the number of objects that fit * into the memory allocation when taking the padding * into account. */ nr_objs = (slab_size - sizeof(struct slab)) / (buffer_size + sizeof(kmem_bufctl_t)); /* * This calculated number will be either the right * amount, or one greater than what we want. */ if (slab_mgmt_size(nr_objs, align) + nr_objs*buffer_size > slab_size) nr_objs--; if (nr_objs > SLAB_LIMIT) nr_objs = SLAB_LIMIT; mgmt_size = slab_mgmt_size(nr_objs, align); } *num = nr_objs; *left_over = slab_size - nr_objs*buffer_size - mgmt_size; |
1da177e4c
|
836 |
} |
d40cee245
|
837 |
#define slab_error(cachep, msg) __slab_error(__func__, cachep, msg) |
1da177e4c
|
838 |
|
a737b3e2f
|
839 840 |
static void __slab_error(const char *function, struct kmem_cache *cachep, char *msg) |
1da177e4c
|
841 842 843 |
{ printk(KERN_ERR "slab error in %s(): cache `%s': %s ", |
b28a02de8
|
844 |
function, cachep->name, msg); |
1da177e4c
|
845 846 |
dump_stack(); } |
3395ee058
|
847 848 849 850 851 852 853 854 855 |
/* * By default on NUMA we use alien caches to stage the freeing of * objects allocated from other nodes. This causes massive memory * inefficiencies when using fake NUMA setup to split memory into a * large number of small nodes, so it can be disabled on the command * line */ static int use_alien_caches __read_mostly = 1; |
1807a1aaf
|
856 |
static int numa_platform __read_mostly = 1; |
3395ee058
|
857 858 859 860 861 862 |
static int __init noaliencache_setup(char *s) { use_alien_caches = 0; return 1; } __setup("noaliencache", noaliencache_setup); |
8fce4d8e3
|
863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 |
#ifdef CONFIG_NUMA /* * Special reaping functions for NUMA systems called from cache_reap(). * These take care of doing round robin flushing of alien caches (containing * objects freed on different nodes from which they were allocated) and the * flushing of remote pcps by calling drain_node_pages. */ static DEFINE_PER_CPU(unsigned long, reap_node); static void init_reap_node(int cpu) { int node; node = next_node(cpu_to_node(cpu), node_online_map); if (node == MAX_NUMNODES) |
442295c94
|
878 |
node = first_node(node_online_map); |
8fce4d8e3
|
879 |
|
7f6b8876c
|
880 |
per_cpu(reap_node, cpu) = node; |
8fce4d8e3
|
881 882 883 884 885 |
} static void next_reap_node(void) { int node = __get_cpu_var(reap_node); |
8fce4d8e3
|
886 887 888 889 890 891 892 893 894 895 |
node = next_node(node, node_online_map); if (unlikely(node >= MAX_NUMNODES)) node = first_node(node_online_map); __get_cpu_var(reap_node) = node; } #else #define init_reap_node(cpu) do { } while (0) #define next_reap_node(void) do { } while (0) #endif |
1da177e4c
|
896 897 898 899 900 901 902 |
/* * Initiate the reap timer running on the target CPU. We run at around 1 to 2Hz * via the workqueue/eventd. * Add the CPU number into the expiration time to minimize the possibility of * the CPUs getting into lockstep and contending for the global cache chain * lock. */ |
897e679b1
|
903 |
static void __cpuinit start_cpu_timer(int cpu) |
1da177e4c
|
904 |
{ |
52bad64d9
|
905 |
struct delayed_work *reap_work = &per_cpu(reap_work, cpu); |
1da177e4c
|
906 907 908 909 910 911 |
/* * When this gets called from do_initcalls via cpucache_init(), * init_workqueues() has already run, so keventd will be setup * at that time. */ |
52bad64d9
|
912 |
if (keventd_up() && reap_work->work.func == NULL) { |
8fce4d8e3
|
913 |
init_reap_node(cpu); |
65f27f384
|
914 |
INIT_DELAYED_WORK(reap_work, cache_reap); |
2b2842146
|
915 916 |
schedule_delayed_work_on(cpu, reap_work, __round_jiffies_relative(HZ, cpu)); |
1da177e4c
|
917 918 |
} } |
e498be7da
|
919 |
static struct array_cache *alloc_arraycache(int node, int entries, |
b28a02de8
|
920 |
int batchcount) |
1da177e4c
|
921 |
{ |
b28a02de8
|
922 |
int memsize = sizeof(void *) * entries + sizeof(struct array_cache); |
1da177e4c
|
923 |
struct array_cache *nc = NULL; |
e498be7da
|
924 |
nc = kmalloc_node(memsize, GFP_KERNEL, node); |
1da177e4c
|
925 926 927 928 929 |
if (nc) { nc->avail = 0; nc->limit = entries; nc->batchcount = batchcount; nc->touched = 0; |
e498be7da
|
930 |
spin_lock_init(&nc->lock); |
1da177e4c
|
931 932 933 |
} return nc; } |
3ded175a4
|
934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 |
/* * Transfer objects in one arraycache to another. * Locking must be handled by the caller. * * Return the number of entries transferred. */ static int transfer_objects(struct array_cache *to, struct array_cache *from, unsigned int max) { /* Figure out how many entries to transfer */ int nr = min(min(from->avail, max), to->limit - to->avail); if (!nr) return 0; memcpy(to->entry + to->avail, from->entry + from->avail -nr, sizeof(void *) *nr); from->avail -= nr; to->avail += nr; to->touched = 1; return nr; } |
765c4507a
|
957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 |
#ifndef CONFIG_NUMA #define drain_alien_cache(cachep, alien) do { } while (0) #define reap_alien(cachep, l3) do { } while (0) static inline struct array_cache **alloc_alien_cache(int node, int limit) { return (struct array_cache **)BAD_ALIEN_MAGIC; } static inline void free_alien_cache(struct array_cache **ac_ptr) { } static inline int cache_free_alien(struct kmem_cache *cachep, void *objp) { return 0; } static inline void *alternate_node_alloc(struct kmem_cache *cachep, gfp_t flags) { return NULL; } |
8b98c1699
|
981 |
static inline void *____cache_alloc_node(struct kmem_cache *cachep, |
765c4507a
|
982 983 984 985 986 987 |
gfp_t flags, int nodeid) { return NULL; } #else /* CONFIG_NUMA */ |
8b98c1699
|
988 |
static void *____cache_alloc_node(struct kmem_cache *, gfp_t, int); |
c61afb181
|
989 |
static void *alternate_node_alloc(struct kmem_cache *, gfp_t); |
dc85da15d
|
990 |
|
5295a74cc
|
991 |
static struct array_cache **alloc_alien_cache(int node, int limit) |
e498be7da
|
992 993 |
{ struct array_cache **ac_ptr; |
8ef828668
|
994 |
int memsize = sizeof(void *) * nr_node_ids; |
e498be7da
|
995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 |
int i; if (limit > 1) limit = 12; ac_ptr = kmalloc_node(memsize, GFP_KERNEL, node); if (ac_ptr) { for_each_node(i) { if (i == node || !node_online(i)) { ac_ptr[i] = NULL; continue; } ac_ptr[i] = alloc_arraycache(node, limit, 0xbaadf00d); if (!ac_ptr[i]) { |
cc550defe
|
1008 |
for (i--; i >= 0; i--) |
e498be7da
|
1009 1010 1011 1012 1013 1014 1015 1016 |
kfree(ac_ptr[i]); kfree(ac_ptr); return NULL; } } } return ac_ptr; } |
5295a74cc
|
1017 |
static void free_alien_cache(struct array_cache **ac_ptr) |
e498be7da
|
1018 1019 1020 1021 1022 |
{ int i; if (!ac_ptr) return; |
e498be7da
|
1023 |
for_each_node(i) |
b28a02de8
|
1024 |
kfree(ac_ptr[i]); |
e498be7da
|
1025 1026 |
kfree(ac_ptr); } |
343e0d7a9
|
1027 |
static void __drain_alien_cache(struct kmem_cache *cachep, |
5295a74cc
|
1028 |
struct array_cache *ac, int node) |
e498be7da
|
1029 1030 1031 1032 1033 |
{ struct kmem_list3 *rl3 = cachep->nodelists[node]; if (ac->avail) { spin_lock(&rl3->list_lock); |
e00946fe2
|
1034 1035 1036 1037 1038 |
/* * Stuff objects into the remote nodes shared array first. * That way we could avoid the overhead of putting the objects * into the free lists and getting them back later. */ |
693f7d362
|
1039 1040 |
if (rl3->shared) transfer_objects(rl3->shared, ac, ac->limit); |
e00946fe2
|
1041 |
|
ff69416e6
|
1042 |
free_block(cachep, ac->entry, ac->avail, node); |
e498be7da
|
1043 1044 1045 1046 |
ac->avail = 0; spin_unlock(&rl3->list_lock); } } |
8fce4d8e3
|
1047 1048 1049 1050 1051 1052 1053 1054 1055 |
/* * Called from cache_reap() to regularly drain alien caches round robin. */ static void reap_alien(struct kmem_cache *cachep, struct kmem_list3 *l3) { int node = __get_cpu_var(reap_node); if (l3->alien) { struct array_cache *ac = l3->alien[node]; |
e00946fe2
|
1056 1057 |
if (ac && ac->avail && spin_trylock_irq(&ac->lock)) { |
8fce4d8e3
|
1058 1059 1060 1061 1062 |
__drain_alien_cache(cachep, ac, node); spin_unlock_irq(&ac->lock); } } } |
a737b3e2f
|
1063 1064 |
static void drain_alien_cache(struct kmem_cache *cachep, struct array_cache **alien) |
e498be7da
|
1065 |
{ |
b28a02de8
|
1066 |
int i = 0; |
e498be7da
|
1067 1068 1069 1070 |
struct array_cache *ac; unsigned long flags; for_each_online_node(i) { |
4484ebf12
|
1071 |
ac = alien[i]; |
e498be7da
|
1072 1073 1074 1075 1076 1077 1078 |
if (ac) { spin_lock_irqsave(&ac->lock, flags); __drain_alien_cache(cachep, ac, i); spin_unlock_irqrestore(&ac->lock, flags); } } } |
729bd0b74
|
1079 |
|
873623dfa
|
1080 |
static inline int cache_free_alien(struct kmem_cache *cachep, void *objp) |
729bd0b74
|
1081 1082 1083 1084 1085 |
{ struct slab *slabp = virt_to_slab(objp); int nodeid = slabp->nodeid; struct kmem_list3 *l3; struct array_cache *alien = NULL; |
1ca4cb241
|
1086 1087 1088 |
int node; node = numa_node_id(); |
729bd0b74
|
1089 1090 1091 1092 1093 |
/* * Make sure we are not freeing a object from another node to the array * cache on this cpu. */ |
62918a036
|
1094 |
if (likely(slabp->nodeid == node)) |
729bd0b74
|
1095 |
return 0; |
1ca4cb241
|
1096 |
l3 = cachep->nodelists[node]; |
729bd0b74
|
1097 1098 1099 |
STATS_INC_NODEFREES(cachep); if (l3->alien && l3->alien[nodeid]) { alien = l3->alien[nodeid]; |
873623dfa
|
1100 |
spin_lock(&alien->lock); |
729bd0b74
|
1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 |
if (unlikely(alien->avail == alien->limit)) { STATS_INC_ACOVERFLOW(cachep); __drain_alien_cache(cachep, alien, nodeid); } alien->entry[alien->avail++] = objp; spin_unlock(&alien->lock); } else { spin_lock(&(cachep->nodelists[nodeid])->list_lock); free_block(cachep, &objp, 1, nodeid); spin_unlock(&(cachep->nodelists[nodeid])->list_lock); } return 1; } |
e498be7da
|
1114 |
#endif |
fbf1e473b
|
1115 1116 1117 1118 1119 |
static void __cpuinit cpuup_canceled(long cpu) { struct kmem_cache *cachep; struct kmem_list3 *l3 = NULL; int node = cpu_to_node(cpu); |
c5f59f083
|
1120 |
node_to_cpumask_ptr(mask, node); |
fbf1e473b
|
1121 1122 1123 1124 1125 |
list_for_each_entry(cachep, &cache_chain, next) { struct array_cache *nc; struct array_cache *shared; struct array_cache **alien; |
fbf1e473b
|
1126 |
|
fbf1e473b
|
1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 |
/* cpu is dead; no one can alloc from it. */ nc = cachep->array[cpu]; cachep->array[cpu] = NULL; l3 = cachep->nodelists[node]; if (!l3) goto free_array_cache; spin_lock_irq(&l3->list_lock); /* Free limit for this kmem_list3 */ l3->free_limit -= cachep->batchcount; if (nc) free_block(cachep, nc->entry, nc->avail, node); |
c5f59f083
|
1141 |
if (!cpus_empty(*mask)) { |
fbf1e473b
|
1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 |
spin_unlock_irq(&l3->list_lock); goto free_array_cache; } shared = l3->shared; if (shared) { free_block(cachep, shared->entry, shared->avail, node); l3->shared = NULL; } alien = l3->alien; l3->alien = NULL; spin_unlock_irq(&l3->list_lock); kfree(shared); if (alien) { drain_alien_cache(cachep, alien); free_alien_cache(alien); } free_array_cache: kfree(nc); } /* * In the previous loop, all the objects were freed to * the respective cache's slabs, now we can go ahead and * shrink each nodelist to its limit. */ list_for_each_entry(cachep, &cache_chain, next) { l3 = cachep->nodelists[node]; if (!l3) continue; drain_freelist(cachep, l3, l3->free_objects); } } static int __cpuinit cpuup_prepare(long cpu) |
1da177e4c
|
1180 |
{ |
343e0d7a9
|
1181 |
struct kmem_cache *cachep; |
e498be7da
|
1182 1183 |
struct kmem_list3 *l3 = NULL; int node = cpu_to_node(cpu); |
ea02e3dde
|
1184 |
const int memsize = sizeof(struct kmem_list3); |
1da177e4c
|
1185 |
|
fbf1e473b
|
1186 1187 1188 1189 1190 1191 1192 1193 |
/* * We need to do this right in the beginning since * alloc_arraycache's are going to use this list. * kmalloc_node allows us to add the slab to the right * kmem_list3 and not this cpu's kmem_list3 */ list_for_each_entry(cachep, &cache_chain, next) { |
a737b3e2f
|
1194 |
/* |
fbf1e473b
|
1195 1196 1197 |
* Set up the size64 kmemlist for cpu before we can * begin anything. Make sure some other cpu on this * node has not already allocated this |
e498be7da
|
1198 |
*/ |
fbf1e473b
|
1199 1200 1201 1202 1203 1204 1205 |
if (!cachep->nodelists[node]) { l3 = kmalloc_node(memsize, GFP_KERNEL, node); if (!l3) goto bad; kmem_list3_init(l3); l3->next_reap = jiffies + REAPTIMEOUT_LIST3 + ((unsigned long)cachep) % REAPTIMEOUT_LIST3; |
e498be7da
|
1206 |
|
a737b3e2f
|
1207 |
/* |
fbf1e473b
|
1208 1209 1210 |
* The l3s don't come and go as CPUs come and * go. cache_chain_mutex is sufficient * protection here. |
e498be7da
|
1211 |
*/ |
fbf1e473b
|
1212 |
cachep->nodelists[node] = l3; |
e498be7da
|
1213 |
} |
fbf1e473b
|
1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 |
spin_lock_irq(&cachep->nodelists[node]->list_lock); cachep->nodelists[node]->free_limit = (1 + nr_cpus_node(node)) * cachep->batchcount + cachep->num; spin_unlock_irq(&cachep->nodelists[node]->list_lock); } /* * Now we can go ahead with allocating the shared arrays and * array caches */ list_for_each_entry(cachep, &cache_chain, next) { struct array_cache *nc; struct array_cache *shared = NULL; struct array_cache **alien = NULL; nc = alloc_arraycache(node, cachep->limit, cachep->batchcount); if (!nc) goto bad; if (cachep->shared) { shared = alloc_arraycache(node, cachep->shared * cachep->batchcount, 0xbaadf00d); |
12d00f6a1
|
1238 1239 |
if (!shared) { kfree(nc); |
1da177e4c
|
1240 |
goto bad; |
12d00f6a1
|
1241 |
} |
fbf1e473b
|
1242 1243 1244 |
} if (use_alien_caches) { alien = alloc_alien_cache(node, cachep->limit); |
12d00f6a1
|
1245 1246 1247 |
if (!alien) { kfree(shared); kfree(nc); |
fbf1e473b
|
1248 |
goto bad; |
12d00f6a1
|
1249 |
} |
fbf1e473b
|
1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 |
} cachep->array[cpu] = nc; l3 = cachep->nodelists[node]; BUG_ON(!l3); spin_lock_irq(&l3->list_lock); if (!l3->shared) { /* * We are serialised from CPU_DEAD or * CPU_UP_CANCELLED by the cpucontrol lock */ l3->shared = shared; shared = NULL; } |
4484ebf12
|
1264 |
#ifdef CONFIG_NUMA |
fbf1e473b
|
1265 1266 1267 |
if (!l3->alien) { l3->alien = alien; alien = NULL; |
1da177e4c
|
1268 |
} |
fbf1e473b
|
1269 1270 1271 1272 1273 1274 1275 |
#endif spin_unlock_irq(&l3->list_lock); kfree(shared); free_alien_cache(alien); } return 0; bad: |
12d00f6a1
|
1276 |
cpuup_canceled(cpu); |
fbf1e473b
|
1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 |
return -ENOMEM; } static int __cpuinit cpuup_callback(struct notifier_block *nfb, unsigned long action, void *hcpu) { long cpu = (long)hcpu; int err = 0; switch (action) { |
fbf1e473b
|
1287 1288 |
case CPU_UP_PREPARE: case CPU_UP_PREPARE_FROZEN: |
95402b382
|
1289 |
mutex_lock(&cache_chain_mutex); |
fbf1e473b
|
1290 |
err = cpuup_prepare(cpu); |
95402b382
|
1291 |
mutex_unlock(&cache_chain_mutex); |
1da177e4c
|
1292 1293 |
break; case CPU_ONLINE: |
8bb784428
|
1294 |
case CPU_ONLINE_FROZEN: |
1da177e4c
|
1295 1296 1297 |
start_cpu_timer(cpu); break; #ifdef CONFIG_HOTPLUG_CPU |
5830c5902
|
1298 |
case CPU_DOWN_PREPARE: |
8bb784428
|
1299 |
case CPU_DOWN_PREPARE_FROZEN: |
5830c5902
|
1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 |
/* * Shutdown cache reaper. Note that the cache_chain_mutex is * held so that if cache_reap() is invoked it cannot do * anything expensive but will only modify reap_work * and reschedule the timer. */ cancel_rearming_delayed_work(&per_cpu(reap_work, cpu)); /* Now the cache_reaper is guaranteed to be not running. */ per_cpu(reap_work, cpu).work.func = NULL; break; case CPU_DOWN_FAILED: |
8bb784428
|
1311 |
case CPU_DOWN_FAILED_FROZEN: |
5830c5902
|
1312 1313 |
start_cpu_timer(cpu); break; |
1da177e4c
|
1314 |
case CPU_DEAD: |
8bb784428
|
1315 |
case CPU_DEAD_FROZEN: |
4484ebf12
|
1316 1317 1318 1319 1320 1321 1322 1323 |
/* * Even if all the cpus of a node are down, we don't free the * kmem_list3 of any cache. This to avoid a race between * cpu_down, and a kmalloc allocation from another cpu for * memory from the node of the cpu going down. The list3 * structure is usually allocated from kmem_cache_create() and * gets destroyed at kmem_cache_destroy(). */ |
183ff22bb
|
1324 |
/* fall through */ |
8f5be20bf
|
1325 |
#endif |
1da177e4c
|
1326 |
case CPU_UP_CANCELED: |
8bb784428
|
1327 |
case CPU_UP_CANCELED_FROZEN: |
95402b382
|
1328 |
mutex_lock(&cache_chain_mutex); |
fbf1e473b
|
1329 |
cpuup_canceled(cpu); |
fc0abb145
|
1330 |
mutex_unlock(&cache_chain_mutex); |
1da177e4c
|
1331 |
break; |
1da177e4c
|
1332 |
} |
fbf1e473b
|
1333 |
return err ? NOTIFY_BAD : NOTIFY_OK; |
1da177e4c
|
1334 |
} |
74b85f379
|
1335 1336 1337 |
static struct notifier_block __cpuinitdata cpucache_notifier = { &cpuup_callback, NULL, 0 }; |
1da177e4c
|
1338 |
|
e498be7da
|
1339 1340 1341 |
/* * swap the static kmem_list3 with kmalloced memory */ |
a737b3e2f
|
1342 1343 |
static void init_list(struct kmem_cache *cachep, struct kmem_list3 *list, int nodeid) |
e498be7da
|
1344 1345 |
{ struct kmem_list3 *ptr; |
e498be7da
|
1346 1347 1348 1349 1350 |
ptr = kmalloc_node(sizeof(struct kmem_list3), GFP_KERNEL, nodeid); BUG_ON(!ptr); local_irq_disable(); memcpy(ptr, list, sizeof(struct kmem_list3)); |
2b2d5493e
|
1351 1352 1353 1354 |
/* * Do not assume that spinlocks can be initialized via memcpy: */ spin_lock_init(&ptr->list_lock); |
e498be7da
|
1355 1356 1357 1358 |
MAKE_ALL_LISTS(cachep, ptr, nodeid); cachep->nodelists[nodeid] = ptr; local_irq_enable(); } |
a737b3e2f
|
1359 |
/* |
556a169da
|
1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 |
* For setting up all the kmem_list3s for cache whose buffer_size is same as * size of kmem_list3. */ static void __init set_up_list3s(struct kmem_cache *cachep, int index) { int node; for_each_online_node(node) { cachep->nodelists[node] = &initkmem_list3[index + node]; cachep->nodelists[node]->next_reap = jiffies + REAPTIMEOUT_LIST3 + ((unsigned long)cachep) % REAPTIMEOUT_LIST3; } } /* |
a737b3e2f
|
1376 1377 |
* Initialisation. Called after the page allocator have been initialised and * before smp_init(). |
1da177e4c
|
1378 1379 1380 1381 1382 1383 |
*/ void __init kmem_cache_init(void) { size_t left_over; struct cache_sizes *sizes; struct cache_names *names; |
e498be7da
|
1384 |
int i; |
07ed76b2a
|
1385 |
int order; |
1ca4cb241
|
1386 |
int node; |
e498be7da
|
1387 |
|
1807a1aaf
|
1388 |
if (num_possible_nodes() == 1) { |
62918a036
|
1389 |
use_alien_caches = 0; |
1807a1aaf
|
1390 1391 |
numa_platform = 0; } |
62918a036
|
1392 |
|
e498be7da
|
1393 1394 1395 1396 1397 |
for (i = 0; i < NUM_INIT_LISTS; i++) { kmem_list3_init(&initkmem_list3[i]); if (i < MAX_NUMNODES) cache_cache.nodelists[i] = NULL; } |
556a169da
|
1398 |
set_up_list3s(&cache_cache, CACHE_CACHE); |
1da177e4c
|
1399 1400 1401 1402 1403 1404 1405 |
/* * Fragmentation resistance on low memory - only use bigger * page orders on machines with more than 32MB of memory. */ if (num_physpages > (32 << 20) >> PAGE_SHIFT) slab_break_gfp_order = BREAK_GFP_ORDER_HI; |
1da177e4c
|
1406 1407 |
/* Bootstrap is tricky, because several objects are allocated * from caches that do not exist yet: |
a737b3e2f
|
1408 1409 1410 |
* 1) initialize the cache_cache cache: it contains the struct * kmem_cache structures of all caches, except cache_cache itself: * cache_cache is statically allocated. |
e498be7da
|
1411 1412 1413 |
* Initially an __init data area is used for the head array and the * kmem_list3 structures, it's replaced with a kmalloc allocated * array at the end of the bootstrap. |
1da177e4c
|
1414 |
* 2) Create the first kmalloc cache. |
343e0d7a9
|
1415 |
* The struct kmem_cache for the new cache is allocated normally. |
e498be7da
|
1416 1417 1418 |
* An __init data area is used for the head array. * 3) Create the remaining kmalloc caches, with minimally sized * head arrays. |
1da177e4c
|
1419 1420 |
* 4) Replace the __init data head arrays for cache_cache and the first * kmalloc cache with kmalloc allocated arrays. |
e498be7da
|
1421 1422 1423 |
* 5) Replace the __init data for kmem_list3 for cache_cache and * the other cache's with kmalloc allocated memory. * 6) Resize the head arrays of the kmalloc caches to their final sizes. |
1da177e4c
|
1424 |
*/ |
1ca4cb241
|
1425 |
node = numa_node_id(); |
1da177e4c
|
1426 |
/* 1) create the cache_cache */ |
1da177e4c
|
1427 1428 1429 1430 |
INIT_LIST_HEAD(&cache_chain); list_add(&cache_cache.next, &cache_chain); cache_cache.colour_off = cache_line_size(); cache_cache.array[smp_processor_id()] = &initarray_cache.cache; |
ec1f5eeeb
|
1431 |
cache_cache.nodelists[node] = &initkmem_list3[CACHE_CACHE + node]; |
1da177e4c
|
1432 |
|
8da3430d8
|
1433 1434 1435 1436 1437 1438 1439 1440 1441 |
/* * struct kmem_cache size depends on nr_node_ids, which * can be less than MAX_NUMNODES. */ cache_cache.buffer_size = offsetof(struct kmem_cache, nodelists) + nr_node_ids * sizeof(struct kmem_list3 *); #if DEBUG cache_cache.obj_size = cache_cache.buffer_size; #endif |
a737b3e2f
|
1442 1443 |
cache_cache.buffer_size = ALIGN(cache_cache.buffer_size, cache_line_size()); |
6a2d7a955
|
1444 1445 |
cache_cache.reciprocal_buffer_size = reciprocal_value(cache_cache.buffer_size); |
1da177e4c
|
1446 |
|
07ed76b2a
|
1447 1448 1449 1450 1451 1452 |
for (order = 0; order < MAX_ORDER; order++) { cache_estimate(order, cache_cache.buffer_size, cache_line_size(), 0, &left_over, &cache_cache.num); if (cache_cache.num) break; } |
40094fa65
|
1453 |
BUG_ON(!cache_cache.num); |
07ed76b2a
|
1454 |
cache_cache.gfporder = order; |
b28a02de8
|
1455 |
cache_cache.colour = left_over / cache_cache.colour_off; |
b28a02de8
|
1456 1457 |
cache_cache.slab_size = ALIGN(cache_cache.num * sizeof(kmem_bufctl_t) + sizeof(struct slab), cache_line_size()); |
1da177e4c
|
1458 1459 1460 1461 |
/* 2+3) create the kmalloc caches */ sizes = malloc_sizes; names = cache_names; |
a737b3e2f
|
1462 1463 1464 1465 |
/* * Initialize the caches that provide memory for the array cache and the * kmem_list3 structures first. Without this, further allocations will * bug. |
e498be7da
|
1466 1467 1468 |
*/ sizes[INDEX_AC].cs_cachep = kmem_cache_create(names[INDEX_AC].name, |
a737b3e2f
|
1469 1470 1471 |
sizes[INDEX_AC].cs_size, ARCH_KMALLOC_MINALIGN, ARCH_KMALLOC_FLAGS|SLAB_PANIC, |
20c2df83d
|
1472 |
NULL); |
e498be7da
|
1473 |
|
a737b3e2f
|
1474 |
if (INDEX_AC != INDEX_L3) { |
e498be7da
|
1475 |
sizes[INDEX_L3].cs_cachep = |
a737b3e2f
|
1476 1477 1478 1479 |
kmem_cache_create(names[INDEX_L3].name, sizes[INDEX_L3].cs_size, ARCH_KMALLOC_MINALIGN, ARCH_KMALLOC_FLAGS|SLAB_PANIC, |
20c2df83d
|
1480 |
NULL); |
a737b3e2f
|
1481 |
} |
e498be7da
|
1482 |
|
e0a427267
|
1483 |
slab_early_init = 0; |
1da177e4c
|
1484 |
while (sizes->cs_size != ULONG_MAX) { |
e498be7da
|
1485 1486 |
/* * For performance, all the general caches are L1 aligned. |
1da177e4c
|
1487 1488 1489 |
* This should be particularly beneficial on SMP boxes, as it * eliminates "false sharing". * Note for systems short on memory removing the alignment will |
e498be7da
|
1490 1491 |
* allow tighter packing of the smaller caches. */ |
a737b3e2f
|
1492 |
if (!sizes->cs_cachep) { |
e498be7da
|
1493 |
sizes->cs_cachep = kmem_cache_create(names->name, |
a737b3e2f
|
1494 1495 1496 |
sizes->cs_size, ARCH_KMALLOC_MINALIGN, ARCH_KMALLOC_FLAGS|SLAB_PANIC, |
20c2df83d
|
1497 |
NULL); |
a737b3e2f
|
1498 |
} |
4b51d6698
|
1499 1500 1501 |
#ifdef CONFIG_ZONE_DMA sizes->cs_dmacachep = kmem_cache_create( names->name_dma, |
a737b3e2f
|
1502 1503 1504 1505 |
sizes->cs_size, ARCH_KMALLOC_MINALIGN, ARCH_KMALLOC_FLAGS|SLAB_CACHE_DMA| SLAB_PANIC, |
20c2df83d
|
1506 |
NULL); |
4b51d6698
|
1507 |
#endif |
1da177e4c
|
1508 1509 1510 1511 1512 |
sizes++; names++; } /* 4) Replace the bootstrap head arrays */ { |
2b2d5493e
|
1513 |
struct array_cache *ptr; |
e498be7da
|
1514 |
|
1da177e4c
|
1515 |
ptr = kmalloc(sizeof(struct arraycache_init), GFP_KERNEL); |
e498be7da
|
1516 |
|
1da177e4c
|
1517 |
local_irq_disable(); |
9a2dba4b4
|
1518 1519 |
BUG_ON(cpu_cache_get(&cache_cache) != &initarray_cache.cache); memcpy(ptr, cpu_cache_get(&cache_cache), |
b28a02de8
|
1520 |
sizeof(struct arraycache_init)); |
2b2d5493e
|
1521 1522 1523 1524 |
/* * Do not assume that spinlocks can be initialized via memcpy: */ spin_lock_init(&ptr->lock); |
1da177e4c
|
1525 1526 |
cache_cache.array[smp_processor_id()] = ptr; local_irq_enable(); |
e498be7da
|
1527 |
|
1da177e4c
|
1528 |
ptr = kmalloc(sizeof(struct arraycache_init), GFP_KERNEL); |
e498be7da
|
1529 |
|
1da177e4c
|
1530 |
local_irq_disable(); |
9a2dba4b4
|
1531 |
BUG_ON(cpu_cache_get(malloc_sizes[INDEX_AC].cs_cachep) |
b28a02de8
|
1532 |
!= &initarray_generic.cache); |
9a2dba4b4
|
1533 |
memcpy(ptr, cpu_cache_get(malloc_sizes[INDEX_AC].cs_cachep), |
b28a02de8
|
1534 |
sizeof(struct arraycache_init)); |
2b2d5493e
|
1535 1536 1537 1538 |
/* * Do not assume that spinlocks can be initialized via memcpy: */ spin_lock_init(&ptr->lock); |
e498be7da
|
1539 |
malloc_sizes[INDEX_AC].cs_cachep->array[smp_processor_id()] = |
b28a02de8
|
1540 |
ptr; |
1da177e4c
|
1541 1542 |
local_irq_enable(); } |
e498be7da
|
1543 1544 |
/* 5) Replace the bootstrap kmem_list3's */ { |
1ca4cb241
|
1545 |
int nid; |
9c09a95cf
|
1546 |
for_each_online_node(nid) { |
ec1f5eeeb
|
1547 |
init_list(&cache_cache, &initkmem_list3[CACHE_CACHE + nid], nid); |
556a169da
|
1548 |
|
e498be7da
|
1549 |
init_list(malloc_sizes[INDEX_AC].cs_cachep, |
1ca4cb241
|
1550 |
&initkmem_list3[SIZE_AC + nid], nid); |
e498be7da
|
1551 1552 1553 |
if (INDEX_AC != INDEX_L3) { init_list(malloc_sizes[INDEX_L3].cs_cachep, |
1ca4cb241
|
1554 |
&initkmem_list3[SIZE_L3 + nid], nid); |
e498be7da
|
1555 1556 1557 |
} } } |
1da177e4c
|
1558 |
|
e498be7da
|
1559 |
/* 6) resize the head arrays to their final sizes */ |
1da177e4c
|
1560 |
{ |
343e0d7a9
|
1561 |
struct kmem_cache *cachep; |
fc0abb145
|
1562 |
mutex_lock(&cache_chain_mutex); |
1da177e4c
|
1563 |
list_for_each_entry(cachep, &cache_chain, next) |
2ed3a4ef9
|
1564 1565 |
if (enable_cpucache(cachep)) BUG(); |
fc0abb145
|
1566 |
mutex_unlock(&cache_chain_mutex); |
1da177e4c
|
1567 |
} |
056c62418
|
1568 1569 |
/* Annotate slab for lockdep -- annotate the malloc caches */ init_lock_keys(); |
1da177e4c
|
1570 1571 |
/* Done! */ g_cpucache_up = FULL; |
a737b3e2f
|
1572 1573 1574 |
/* * Register a cpu startup notifier callback that initializes * cpu_cache_get for all new cpus |
1da177e4c
|
1575 1576 |
*/ register_cpu_notifier(&cpucache_notifier); |
1da177e4c
|
1577 |
|
a737b3e2f
|
1578 1579 1580 |
/* * The reap timers are started later, with a module init call: That part * of the kernel is not yet operational. |
1da177e4c
|
1581 1582 1583 1584 1585 1586 |
*/ } static int __init cpucache_init(void) { int cpu; |
a737b3e2f
|
1587 1588 |
/* * Register the timers that return unneeded pages to the page allocator |
1da177e4c
|
1589 |
*/ |
e498be7da
|
1590 |
for_each_online_cpu(cpu) |
a737b3e2f
|
1591 |
start_cpu_timer(cpu); |
1da177e4c
|
1592 1593 |
return 0; } |
1da177e4c
|
1594 1595 1596 1597 1598 1599 1600 1601 1602 |
__initcall(cpucache_init); /* * Interface to system's page allocator. No need to hold the cache-lock. * * If we requested dmaable memory, we will get it. Even if we * did not request dmaable memory, we might get it, but that * would be relatively rare and ignorable. */ |
343e0d7a9
|
1603 |
static void *kmem_getpages(struct kmem_cache *cachep, gfp_t flags, int nodeid) |
1da177e4c
|
1604 1605 |
{ struct page *page; |
e1b6aa6f1
|
1606 |
int nr_pages; |
1da177e4c
|
1607 |
int i; |
d6fef9da1
|
1608 |
#ifndef CONFIG_MMU |
e1b6aa6f1
|
1609 1610 1611 |
/* * Nommu uses slab's for process anonymous memory allocations, and thus * requires __GFP_COMP to properly refcount higher order allocations |
d6fef9da1
|
1612 |
*/ |
e1b6aa6f1
|
1613 |
flags |= __GFP_COMP; |
d6fef9da1
|
1614 |
#endif |
765c4507a
|
1615 |
|
3c517a613
|
1616 |
flags |= cachep->gfpflags; |
e12ba74d8
|
1617 1618 |
if (cachep->flags & SLAB_RECLAIM_ACCOUNT) flags |= __GFP_RECLAIMABLE; |
e1b6aa6f1
|
1619 1620 |
page = alloc_pages_node(nodeid, flags, cachep->gfporder); |
1da177e4c
|
1621 1622 |
if (!page) return NULL; |
1da177e4c
|
1623 |
|
e1b6aa6f1
|
1624 |
nr_pages = (1 << cachep->gfporder); |
1da177e4c
|
1625 |
if (cachep->flags & SLAB_RECLAIM_ACCOUNT) |
972d1a7b1
|
1626 1627 1628 1629 1630 |
add_zone_page_state(page_zone(page), NR_SLAB_RECLAIMABLE, nr_pages); else add_zone_page_state(page_zone(page), NR_SLAB_UNRECLAIMABLE, nr_pages); |
e1b6aa6f1
|
1631 1632 1633 |
for (i = 0; i < nr_pages; i++) __SetPageSlab(page + i); return page_address(page); |
1da177e4c
|
1634 1635 1636 1637 1638 |
} /* * Interface to system's page release. */ |
343e0d7a9
|
1639 |
static void kmem_freepages(struct kmem_cache *cachep, void *addr) |
1da177e4c
|
1640 |
{ |
b28a02de8
|
1641 |
unsigned long i = (1 << cachep->gfporder); |
1da177e4c
|
1642 1643 |
struct page *page = virt_to_page(addr); const unsigned long nr_freed = i; |
972d1a7b1
|
1644 1645 1646 1647 1648 1649 |
if (cachep->flags & SLAB_RECLAIM_ACCOUNT) sub_zone_page_state(page_zone(page), NR_SLAB_RECLAIMABLE, nr_freed); else sub_zone_page_state(page_zone(page), NR_SLAB_UNRECLAIMABLE, nr_freed); |
1da177e4c
|
1650 |
while (i--) { |
f205b2fe6
|
1651 1652 |
BUG_ON(!PageSlab(page)); __ClearPageSlab(page); |
1da177e4c
|
1653 1654 |
page++; } |
1da177e4c
|
1655 1656 1657 |
if (current->reclaim_state) current->reclaim_state->reclaimed_slab += nr_freed; free_pages((unsigned long)addr, cachep->gfporder); |
1da177e4c
|
1658 1659 1660 1661 |
} static void kmem_rcu_free(struct rcu_head *head) { |
b28a02de8
|
1662 |
struct slab_rcu *slab_rcu = (struct slab_rcu *)head; |
343e0d7a9
|
1663 |
struct kmem_cache *cachep = slab_rcu->cachep; |
1da177e4c
|
1664 1665 1666 1667 1668 1669 1670 1671 1672 |
kmem_freepages(cachep, slab_rcu->addr); if (OFF_SLAB(cachep)) kmem_cache_free(cachep->slabp_cache, slab_rcu); } #if DEBUG #ifdef CONFIG_DEBUG_PAGEALLOC |
343e0d7a9
|
1673 |
static void store_stackinfo(struct kmem_cache *cachep, unsigned long *addr, |
b28a02de8
|
1674 |
unsigned long caller) |
1da177e4c
|
1675 |
{ |
3dafccf22
|
1676 |
int size = obj_size(cachep); |
1da177e4c
|
1677 |
|
3dafccf22
|
1678 |
addr = (unsigned long *)&((char *)addr)[obj_offset(cachep)]; |
1da177e4c
|
1679 |
|
b28a02de8
|
1680 |
if (size < 5 * sizeof(unsigned long)) |
1da177e4c
|
1681 |
return; |
b28a02de8
|
1682 1683 1684 1685 |
*addr++ = 0x12345678; *addr++ = caller; *addr++ = smp_processor_id(); size -= 3 * sizeof(unsigned long); |
1da177e4c
|
1686 1687 1688 1689 1690 1691 1692 |
{ unsigned long *sptr = &caller; unsigned long svalue; while (!kstack_end(sptr)) { svalue = *sptr++; if (kernel_text_address(svalue)) { |
b28a02de8
|
1693 |
*addr++ = svalue; |
1da177e4c
|
1694 1695 1696 1697 1698 1699 1700 |
size -= sizeof(unsigned long); if (size <= sizeof(unsigned long)) break; } } } |
b28a02de8
|
1701 |
*addr++ = 0x87654321; |
1da177e4c
|
1702 1703 |
} #endif |
343e0d7a9
|
1704 |
static void poison_obj(struct kmem_cache *cachep, void *addr, unsigned char val) |
1da177e4c
|
1705 |
{ |
3dafccf22
|
1706 1707 |
int size = obj_size(cachep); addr = &((char *)addr)[obj_offset(cachep)]; |
1da177e4c
|
1708 1709 |
memset(addr, val, size); |
b28a02de8
|
1710 |
*(unsigned char *)(addr + size - 1) = POISON_END; |
1da177e4c
|
1711 1712 1713 1714 1715 |
} static void dump_line(char *data, int offset, int limit) { int i; |
aa83aa40e
|
1716 1717 |
unsigned char error = 0; int bad_count = 0; |
1da177e4c
|
1718 |
printk(KERN_ERR "%03x:", offset); |
aa83aa40e
|
1719 1720 1721 1722 1723 |
for (i = 0; i < limit; i++) { if (data[offset + i] != POISON_FREE) { error = data[offset + i]; bad_count++; } |
b28a02de8
|
1724 |
printk(" %02x", (unsigned char)data[offset + i]); |
aa83aa40e
|
1725 |
} |
1da177e4c
|
1726 1727 |
printk(" "); |
aa83aa40e
|
1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 |
if (bad_count == 1) { error ^= POISON_FREE; if (!(error & (error - 1))) { printk(KERN_ERR "Single bit error detected. Probably " "bad RAM. "); #ifdef CONFIG_X86 printk(KERN_ERR "Run memtest86+ or a similar memory " "test tool. "); #else printk(KERN_ERR "Run a memory test tool. "); #endif } } |
1da177e4c
|
1745 1746 1747 1748 |
} #endif #if DEBUG |
343e0d7a9
|
1749 |
static void print_objinfo(struct kmem_cache *cachep, void *objp, int lines) |
1da177e4c
|
1750 1751 1752 1753 1754 |
{ int i, size; char *realobj; if (cachep->flags & SLAB_RED_ZONE) { |
b46b8f19c
|
1755 1756 |
printk(KERN_ERR "Redzone: 0x%llx/0x%llx. ", |
a737b3e2f
|
1757 1758 |
*dbg_redzone1(cachep, objp), *dbg_redzone2(cachep, objp)); |
1da177e4c
|
1759 1760 1761 1762 |
} if (cachep->flags & SLAB_STORE_USER) { printk(KERN_ERR "Last user: [<%p>]", |
a737b3e2f
|
1763 |
*dbg_userword(cachep, objp)); |
1da177e4c
|
1764 |
print_symbol("(%s)", |
a737b3e2f
|
1765 |
(unsigned long)*dbg_userword(cachep, objp)); |
1da177e4c
|
1766 1767 1768 |
printk(" "); } |
3dafccf22
|
1769 1770 |
realobj = (char *)objp + obj_offset(cachep); size = obj_size(cachep); |
b28a02de8
|
1771 |
for (i = 0; i < size && lines; i += 16, lines--) { |
1da177e4c
|
1772 1773 |
int limit; limit = 16; |
b28a02de8
|
1774 1775 |
if (i + limit > size) limit = size - i; |
1da177e4c
|
1776 1777 1778 |
dump_line(realobj, i, limit); } } |
343e0d7a9
|
1779 |
static void check_poison_obj(struct kmem_cache *cachep, void *objp) |
1da177e4c
|
1780 1781 1782 1783 |
{ char *realobj; int size, i; int lines = 0; |
3dafccf22
|
1784 1785 |
realobj = (char *)objp + obj_offset(cachep); size = obj_size(cachep); |
1da177e4c
|
1786 |
|
b28a02de8
|
1787 |
for (i = 0; i < size; i++) { |
1da177e4c
|
1788 |
char exp = POISON_FREE; |
b28a02de8
|
1789 |
if (i == size - 1) |
1da177e4c
|
1790 1791 1792 1793 1794 1795 |
exp = POISON_END; if (realobj[i] != exp) { int limit; /* Mismatch ! */ /* Print header */ if (lines == 0) { |
b28a02de8
|
1796 |
printk(KERN_ERR |
e94a40c50
|
1797 1798 1799 |
"Slab corruption: %s start=%p, len=%d ", cachep->name, realobj, size); |
1da177e4c
|
1800 1801 1802 |
print_objinfo(cachep, objp, 0); } /* Hexdump the affected line */ |
b28a02de8
|
1803 |
i = (i / 16) * 16; |
1da177e4c
|
1804 |
limit = 16; |
b28a02de8
|
1805 1806 |
if (i + limit > size) limit = size - i; |
1da177e4c
|
1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 |
dump_line(realobj, i, limit); i += 16; lines++; /* Limit to 5 lines */ if (lines > 5) break; } } if (lines != 0) { /* Print some data about the neighboring objects, if they * exist: */ |
6ed5eb221
|
1819 |
struct slab *slabp = virt_to_slab(objp); |
8fea4e96a
|
1820 |
unsigned int objnr; |
1da177e4c
|
1821 |
|
8fea4e96a
|
1822 |
objnr = obj_to_index(cachep, slabp, objp); |
1da177e4c
|
1823 |
if (objnr) { |
8fea4e96a
|
1824 |
objp = index_to_obj(cachep, slabp, objnr - 1); |
3dafccf22
|
1825 |
realobj = (char *)objp + obj_offset(cachep); |
1da177e4c
|
1826 1827 |
printk(KERN_ERR "Prev obj: start=%p, len=%d ", |
b28a02de8
|
1828 |
realobj, size); |
1da177e4c
|
1829 1830 |
print_objinfo(cachep, objp, 2); } |
b28a02de8
|
1831 |
if (objnr + 1 < cachep->num) { |
8fea4e96a
|
1832 |
objp = index_to_obj(cachep, slabp, objnr + 1); |
3dafccf22
|
1833 |
realobj = (char *)objp + obj_offset(cachep); |
1da177e4c
|
1834 1835 |
printk(KERN_ERR "Next obj: start=%p, len=%d ", |
b28a02de8
|
1836 |
realobj, size); |
1da177e4c
|
1837 1838 1839 1840 1841 |
print_objinfo(cachep, objp, 2); } } } #endif |
12dd36fae
|
1842 |
#if DEBUG |
e79aec291
|
1843 |
static void slab_destroy_debugcheck(struct kmem_cache *cachep, struct slab *slabp) |
1da177e4c
|
1844 |
{ |
1da177e4c
|
1845 1846 |
int i; for (i = 0; i < cachep->num; i++) { |
8fea4e96a
|
1847 |
void *objp = index_to_obj(cachep, slabp, i); |
1da177e4c
|
1848 1849 1850 |
if (cachep->flags & SLAB_POISON) { #ifdef CONFIG_DEBUG_PAGEALLOC |
a737b3e2f
|
1851 1852 |
if (cachep->buffer_size % PAGE_SIZE == 0 && OFF_SLAB(cachep)) |
b28a02de8
|
1853 |
kernel_map_pages(virt_to_page(objp), |
a737b3e2f
|
1854 |
cachep->buffer_size / PAGE_SIZE, 1); |
1da177e4c
|
1855 1856 1857 1858 1859 1860 1861 1862 1863 |
else check_poison_obj(cachep, objp); #else check_poison_obj(cachep, objp); #endif } if (cachep->flags & SLAB_RED_ZONE) { if (*dbg_redzone1(cachep, objp) != RED_INACTIVE) slab_error(cachep, "start of a freed object " |
b28a02de8
|
1864 |
"was overwritten"); |
1da177e4c
|
1865 1866 |
if (*dbg_redzone2(cachep, objp) != RED_INACTIVE) slab_error(cachep, "end of a freed object " |
b28a02de8
|
1867 |
"was overwritten"); |
1da177e4c
|
1868 |
} |
1da177e4c
|
1869 |
} |
12dd36fae
|
1870 |
} |
1da177e4c
|
1871 |
#else |
e79aec291
|
1872 |
static void slab_destroy_debugcheck(struct kmem_cache *cachep, struct slab *slabp) |
12dd36fae
|
1873 |
{ |
12dd36fae
|
1874 |
} |
1da177e4c
|
1875 |
#endif |
911851e6e
|
1876 1877 1878 1879 1880 |
/** * slab_destroy - destroy and release all objects in a slab * @cachep: cache pointer being destroyed * @slabp: slab pointer being destroyed * |
12dd36fae
|
1881 |
* Destroy all the objs in a slab, and release the mem back to the system. |
a737b3e2f
|
1882 1883 |
* Before calling the slab must have been unlinked from the cache. The * cache-lock is not held/needed. |
12dd36fae
|
1884 |
*/ |
343e0d7a9
|
1885 |
static void slab_destroy(struct kmem_cache *cachep, struct slab *slabp) |
12dd36fae
|
1886 1887 |
{ void *addr = slabp->s_mem - slabp->colouroff; |
e79aec291
|
1888 |
slab_destroy_debugcheck(cachep, slabp); |
1da177e4c
|
1889 1890 |
if (unlikely(cachep->flags & SLAB_DESTROY_BY_RCU)) { struct slab_rcu *slab_rcu; |
b28a02de8
|
1891 |
slab_rcu = (struct slab_rcu *)slabp; |
1da177e4c
|
1892 1893 1894 1895 1896 |
slab_rcu->cachep = cachep; slab_rcu->addr = addr; call_rcu(&slab_rcu->head, kmem_rcu_free); } else { kmem_freepages(cachep, addr); |
873623dfa
|
1897 1898 |
if (OFF_SLAB(cachep)) kmem_cache_free(cachep->slabp_cache, slabp); |
1da177e4c
|
1899 1900 |
} } |
117f6eb1d
|
1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 |
static void __kmem_cache_destroy(struct kmem_cache *cachep) { int i; struct kmem_list3 *l3; for_each_online_cpu(i) kfree(cachep->array[i]); /* NUMA: free the list3 structures */ for_each_online_node(i) { l3 = cachep->nodelists[i]; if (l3) { kfree(l3->shared); free_alien_cache(l3->alien); kfree(l3); } } kmem_cache_free(&cache_cache, cachep); } |
1da177e4c
|
1920 |
/** |
a70773ddb
|
1921 1922 1923 1924 1925 1926 1927 |
* calculate_slab_order - calculate size (page order) of slabs * @cachep: pointer to the cache that is being created * @size: size of objects to be created in this cache. * @align: required alignment for the objects. * @flags: slab allocation flags * * Also calculates the number of objects per slab. |
4d268eba1
|
1928 1929 1930 1931 1932 |
* * This could be made much more intelligent. For now, try to avoid using * high order pages for slabs. When the gfp() functions are more friendly * towards high-order requests, this should be changed. */ |
a737b3e2f
|
1933 |
static size_t calculate_slab_order(struct kmem_cache *cachep, |
ee13d785e
|
1934 |
size_t size, size_t align, unsigned long flags) |
4d268eba1
|
1935 |
{ |
b1ab41c49
|
1936 |
unsigned long offslab_limit; |
4d268eba1
|
1937 |
size_t left_over = 0; |
9888e6fa7
|
1938 |
int gfporder; |
4d268eba1
|
1939 |
|
0aa817f07
|
1940 |
for (gfporder = 0; gfporder <= KMALLOC_MAX_ORDER; gfporder++) { |
4d268eba1
|
1941 1942 |
unsigned int num; size_t remainder; |
9888e6fa7
|
1943 |
cache_estimate(gfporder, size, align, flags, &remainder, &num); |
4d268eba1
|
1944 1945 |
if (!num) continue; |
9888e6fa7
|
1946 |
|
b1ab41c49
|
1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 |
if (flags & CFLGS_OFF_SLAB) { /* * Max number of objs-per-slab for caches which * use off-slab slabs. Needed to avoid a possible * looping condition in cache_grow(). */ offslab_limit = size - sizeof(struct slab); offslab_limit /= sizeof(kmem_bufctl_t); if (num > offslab_limit) break; } |
4d268eba1
|
1959 |
|
9888e6fa7
|
1960 |
/* Found something acceptable - save it away */ |
4d268eba1
|
1961 |
cachep->num = num; |
9888e6fa7
|
1962 |
cachep->gfporder = gfporder; |
4d268eba1
|
1963 1964 1965 |
left_over = remainder; /* |
f78bb8ad4
|
1966 1967 1968 1969 1970 1971 1972 1973 |
* A VFS-reclaimable slab tends to have most allocations * as GFP_NOFS and we really don't want to have to be allocating * higher-order pages when we are unable to shrink dcache. */ if (flags & SLAB_RECLAIM_ACCOUNT) break; /* |
4d268eba1
|
1974 1975 1976 |
* Large number of objects is good, but very large slabs are * currently bad for the gfp()s. */ |
9888e6fa7
|
1977 |
if (gfporder >= slab_break_gfp_order) |
4d268eba1
|
1978 |
break; |
9888e6fa7
|
1979 1980 1981 |
/* * Acceptable internal fragmentation? */ |
a737b3e2f
|
1982 |
if (left_over * 8 <= (PAGE_SIZE << gfporder)) |
4d268eba1
|
1983 1984 1985 1986 |
break; } return left_over; } |
38bdc32af
|
1987 |
static int __init_refok setup_cpu_cache(struct kmem_cache *cachep) |
f30cf7d13
|
1988 |
{ |
2ed3a4ef9
|
1989 1990 |
if (g_cpucache_up == FULL) return enable_cpucache(cachep); |
f30cf7d13
|
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 |
if (g_cpucache_up == NONE) { /* * Note: the first kmem_cache_create must create the cache * that's used by kmalloc(24), otherwise the creation of * further caches will BUG(). */ cachep->array[smp_processor_id()] = &initarray_generic.cache; /* * If the cache that's used by kmalloc(sizeof(kmem_list3)) is * the first cache, then we need to set up all its list3s, * otherwise the creation of further caches will BUG(). */ set_up_list3s(cachep, SIZE_AC); if (INDEX_AC == INDEX_L3) g_cpucache_up = PARTIAL_L3; else g_cpucache_up = PARTIAL_AC; } else { cachep->array[smp_processor_id()] = kmalloc(sizeof(struct arraycache_init), GFP_KERNEL); if (g_cpucache_up == PARTIAL_AC) { set_up_list3s(cachep, SIZE_L3); g_cpucache_up = PARTIAL_L3; } else { int node; |
556a169da
|
2018 |
for_each_online_node(node) { |
f30cf7d13
|
2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 |
cachep->nodelists[node] = kmalloc_node(sizeof(struct kmem_list3), GFP_KERNEL, node); BUG_ON(!cachep->nodelists[node]); kmem_list3_init(cachep->nodelists[node]); } } } cachep->nodelists[numa_node_id()]->next_reap = jiffies + REAPTIMEOUT_LIST3 + ((unsigned long)cachep) % REAPTIMEOUT_LIST3; cpu_cache_get(cachep)->avail = 0; cpu_cache_get(cachep)->limit = BOOT_CPUCACHE_ENTRIES; cpu_cache_get(cachep)->batchcount = 1; cpu_cache_get(cachep)->touched = 0; cachep->batchcount = 1; cachep->limit = BOOT_CPUCACHE_ENTRIES; |
2ed3a4ef9
|
2037 |
return 0; |
f30cf7d13
|
2038 |
} |
4d268eba1
|
2039 |
/** |
1da177e4c
|
2040 2041 2042 2043 2044 2045 |
* kmem_cache_create - Create a cache. * @name: A string which is used in /proc/slabinfo to identify this cache. * @size: The size of objects to be created in this cache. * @align: The required alignment for the objects. * @flags: SLAB flags * @ctor: A constructor for the objects. |
1da177e4c
|
2046 2047 2048 |
* * Returns a ptr to the cache on success, NULL on failure. * Cannot be called within a int, but can be interrupted. |
20c2df83d
|
2049 |
* The @ctor is run when new pages are allocated by the cache. |
1da177e4c
|
2050 2051 |
* * @name must be valid until the cache is destroyed. This implies that |
a737b3e2f
|
2052 |
* the module calling this has to destroy the cache before getting unloaded. |
249da1665
|
2053 2054 |
* Note that kmem_cache_name() is not guaranteed to return the same pointer, * therefore applications must manage it themselves. |
a737b3e2f
|
2055 |
* |
1da177e4c
|
2056 2057 2058 2059 2060 2061 2062 2063 |
* The flags are * * %SLAB_POISON - Poison the slab with a known test pattern (a5a5a5a5) * to catch references to uninitialised memory. * * %SLAB_RED_ZONE - Insert `Red' zones around the allocated memory to check * for buffer overruns. * |
1da177e4c
|
2064 2065 2066 2067 |
* %SLAB_HWCACHE_ALIGN - Align the objects in this cache to a hardware * cacheline. This can be beneficial if you're counting cycles as closely * as davem. */ |
343e0d7a9
|
2068 |
struct kmem_cache * |
1da177e4c
|
2069 |
kmem_cache_create (const char *name, size_t size, size_t align, |
51cc50685
|
2070 |
unsigned long flags, void (*ctor)(void *)) |
1da177e4c
|
2071 2072 |
{ size_t left_over, slab_size, ralign; |
7a7c381d2
|
2073 |
struct kmem_cache *cachep = NULL, *pc; |
1da177e4c
|
2074 2075 2076 2077 |
/* * Sanity checks... these are all serious usage bugs. */ |
a737b3e2f
|
2078 |
if (!name || in_interrupt() || (size < BYTES_PER_WORD) || |
20c2df83d
|
2079 |
size > KMALLOC_MAX_SIZE) { |
d40cee245
|
2080 2081 |
printk(KERN_ERR "%s: Early error in slab %s ", __func__, |
a737b3e2f
|
2082 |
name); |
b28a02de8
|
2083 2084 |
BUG(); } |
1da177e4c
|
2085 |
|
f0188f474
|
2086 |
/* |
8f5be20bf
|
2087 |
* We use cache_chain_mutex to ensure a consistent view of |
174596a0b
|
2088 |
* cpu_online_mask as well. Please see cpuup_callback |
f0188f474
|
2089 |
*/ |
95402b382
|
2090 |
get_online_cpus(); |
fc0abb145
|
2091 |
mutex_lock(&cache_chain_mutex); |
4f12bb4f7
|
2092 |
|
7a7c381d2
|
2093 |
list_for_each_entry(pc, &cache_chain, next) { |
4f12bb4f7
|
2094 2095 2096 2097 2098 2099 2100 2101 |
char tmp; int res; /* * This happens when the module gets unloaded and doesn't * destroy its slab cache and no-one else reuses the vmalloc * area of the module. Print a warning. */ |
138ae6631
|
2102 |
res = probe_kernel_address(pc->name, tmp); |
4f12bb4f7
|
2103 |
if (res) { |
b4169525b
|
2104 2105 2106 |
printk(KERN_ERR "SLAB: cache with size %d has lost its name ", |
3dafccf22
|
2107 |
pc->buffer_size); |
4f12bb4f7
|
2108 2109 |
continue; } |
b28a02de8
|
2110 |
if (!strcmp(pc->name, name)) { |
b4169525b
|
2111 2112 2113 |
printk(KERN_ERR "kmem_cache_create: duplicate cache %s ", name); |
4f12bb4f7
|
2114 2115 2116 2117 |
dump_stack(); goto oops; } } |
1da177e4c
|
2118 2119 |
#if DEBUG WARN_ON(strchr(name, ' ')); /* It confuses parsers */ |
1da177e4c
|
2120 2121 2122 2123 2124 2125 2126 |
#if FORCED_DEBUG /* * Enable redzoning and last user accounting, except for caches with * large objects, if the increased size would increase the object size * above the next power of two: caches with object sizes just above a * power of two have a significant amount of internal fragmentation. */ |
87a927c71
|
2127 2128 |
if (size < 4096 || fls(size - 1) == fls(size-1 + REDZONE_ALIGN + 2 * sizeof(unsigned long long))) |
b28a02de8
|
2129 |
flags |= SLAB_RED_ZONE | SLAB_STORE_USER; |
1da177e4c
|
2130 2131 2132 2133 2134 2135 |
if (!(flags & SLAB_DESTROY_BY_RCU)) flags |= SLAB_POISON; #endif if (flags & SLAB_DESTROY_BY_RCU) BUG_ON(flags & SLAB_POISON); #endif |
1da177e4c
|
2136 |
/* |
a737b3e2f
|
2137 2138 |
* Always checks flags, a caller might be expecting debug support which * isn't available. |
1da177e4c
|
2139 |
*/ |
40094fa65
|
2140 |
BUG_ON(flags & ~CREATE_MASK); |
1da177e4c
|
2141 |
|
a737b3e2f
|
2142 2143 |
/* * Check that size is in terms of words. This is needed to avoid |
1da177e4c
|
2144 2145 2146 |
* unaligned accesses for some archs when redzoning is used, and makes * sure any on-slab bufctl's are also correctly aligned. */ |
b28a02de8
|
2147 2148 2149 |
if (size & (BYTES_PER_WORD - 1)) { size += (BYTES_PER_WORD - 1); size &= ~(BYTES_PER_WORD - 1); |
1da177e4c
|
2150 |
} |
a737b3e2f
|
2151 |
/* calculate the final buffer alignment: */ |
1da177e4c
|
2152 2153 |
/* 1) arch recommendation: can be overridden for debug */ if (flags & SLAB_HWCACHE_ALIGN) { |
a737b3e2f
|
2154 2155 2156 2157 |
/* * Default alignment: as specified by the arch code. Except if * an object is really small, then squeeze multiple objects into * one cacheline. |
1da177e4c
|
2158 2159 |
*/ ralign = cache_line_size(); |
b28a02de8
|
2160 |
while (size <= ralign / 2) |
1da177e4c
|
2161 2162 2163 2164 |
ralign /= 2; } else { ralign = BYTES_PER_WORD; } |
ca5f9703d
|
2165 2166 |
/* |
87a927c71
|
2167 2168 2169 |
* Redzoning and user store require word alignment or possibly larger. * Note this will be overridden by architecture or caller mandated * alignment if either is greater than BYTES_PER_WORD. |
ca5f9703d
|
2170 |
*/ |
87a927c71
|
2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 |
if (flags & SLAB_STORE_USER) ralign = BYTES_PER_WORD; if (flags & SLAB_RED_ZONE) { ralign = REDZONE_ALIGN; /* If redzoning, ensure that the second redzone is suitably * aligned, by adjusting the object size accordingly. */ size += REDZONE_ALIGN - 1; size &= ~(REDZONE_ALIGN - 1); } |
ca5f9703d
|
2181 |
|
a44b56d35
|
2182 |
/* 2) arch mandated alignment */ |
1da177e4c
|
2183 2184 |
if (ralign < ARCH_SLAB_MINALIGN) { ralign = ARCH_SLAB_MINALIGN; |
1da177e4c
|
2185 |
} |
a44b56d35
|
2186 |
/* 3) caller mandated alignment */ |
1da177e4c
|
2187 2188 |
if (ralign < align) { ralign = align; |
1da177e4c
|
2189 |
} |
a44b56d35
|
2190 |
/* disable debug if necessary */ |
b46b8f19c
|
2191 |
if (ralign > __alignof__(unsigned long long)) |
a44b56d35
|
2192 |
flags &= ~(SLAB_RED_ZONE | SLAB_STORE_USER); |
a737b3e2f
|
2193 |
/* |
ca5f9703d
|
2194 |
* 4) Store it. |
1da177e4c
|
2195 2196 2197 2198 |
*/ align = ralign; /* Get cache's description obj. */ |
e94b17660
|
2199 |
cachep = kmem_cache_zalloc(&cache_cache, GFP_KERNEL); |
1da177e4c
|
2200 |
if (!cachep) |
4f12bb4f7
|
2201 |
goto oops; |
1da177e4c
|
2202 2203 |
#if DEBUG |
3dafccf22
|
2204 |
cachep->obj_size = size; |
1da177e4c
|
2205 |
|
ca5f9703d
|
2206 2207 2208 2209 |
/* * Both debugging options require word-alignment which is calculated * into align above. */ |
1da177e4c
|
2210 |
if (flags & SLAB_RED_ZONE) { |
1da177e4c
|
2211 |
/* add space for red zone words */ |
b46b8f19c
|
2212 2213 |
cachep->obj_offset += sizeof(unsigned long long); size += 2 * sizeof(unsigned long long); |
1da177e4c
|
2214 2215 |
} if (flags & SLAB_STORE_USER) { |
ca5f9703d
|
2216 |
/* user store requires one word storage behind the end of |
87a927c71
|
2217 2218 |
* the real object. But if the second red zone needs to be * aligned to 64 bits, we must allow that much space. |
1da177e4c
|
2219 |
*/ |
87a927c71
|
2220 2221 2222 2223 |
if (flags & SLAB_RED_ZONE) size += REDZONE_ALIGN; else size += BYTES_PER_WORD; |
1da177e4c
|
2224 2225 |
} #if FORCED_DEBUG && defined(CONFIG_DEBUG_PAGEALLOC) |
b28a02de8
|
2226 |
if (size >= malloc_sizes[INDEX_L3 + 1].cs_size |
3dafccf22
|
2227 2228 |
&& cachep->obj_size > cache_line_size() && size < PAGE_SIZE) { cachep->obj_offset += PAGE_SIZE - size; |
1da177e4c
|
2229 2230 2231 2232 |
size = PAGE_SIZE; } #endif #endif |
e0a427267
|
2233 2234 2235 2236 2237 2238 |
/* * Determine if the slab management is 'on' or 'off' slab. * (bootstrapping cannot cope with offslab caches so don't do * it too early on.) */ if ((size >= (PAGE_SIZE >> 3)) && !slab_early_init) |
1da177e4c
|
2239 2240 2241 2242 2243 2244 2245 |
/* * Size is large, assume best to place the slab management obj * off-slab (should allow better packing of objs). */ flags |= CFLGS_OFF_SLAB; size = ALIGN(size, align); |
f78bb8ad4
|
2246 |
left_over = calculate_slab_order(cachep, size, align, flags); |
1da177e4c
|
2247 2248 |
if (!cachep->num) { |
b4169525b
|
2249 2250 2251 |
printk(KERN_ERR "kmem_cache_create: couldn't create cache %s. ", name); |
1da177e4c
|
2252 2253 |
kmem_cache_free(&cache_cache, cachep); cachep = NULL; |
4f12bb4f7
|
2254 |
goto oops; |
1da177e4c
|
2255 |
} |
b28a02de8
|
2256 2257 |
slab_size = ALIGN(cachep->num * sizeof(kmem_bufctl_t) + sizeof(struct slab), align); |
1da177e4c
|
2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 |
/* * If the slab has been placed off-slab, and we have enough space then * move it on-slab. This is at the expense of any extra colouring. */ if (flags & CFLGS_OFF_SLAB && left_over >= slab_size) { flags &= ~CFLGS_OFF_SLAB; left_over -= slab_size; } if (flags & CFLGS_OFF_SLAB) { /* really off slab. No need for manual alignment */ |
b28a02de8
|
2270 2271 |
slab_size = cachep->num * sizeof(kmem_bufctl_t) + sizeof(struct slab); |
1da177e4c
|
2272 2273 2274 2275 2276 2277 |
} cachep->colour_off = cache_line_size(); /* Offset must be a multiple of the alignment. */ if (cachep->colour_off < align) cachep->colour_off = align; |
b28a02de8
|
2278 |
cachep->colour = left_over / cachep->colour_off; |
1da177e4c
|
2279 2280 2281 |
cachep->slab_size = slab_size; cachep->flags = flags; cachep->gfpflags = 0; |
4b51d6698
|
2282 |
if (CONFIG_ZONE_DMA_FLAG && (flags & SLAB_CACHE_DMA)) |
1da177e4c
|
2283 |
cachep->gfpflags |= GFP_DMA; |
3dafccf22
|
2284 |
cachep->buffer_size = size; |
6a2d7a955
|
2285 |
cachep->reciprocal_buffer_size = reciprocal_value(size); |
1da177e4c
|
2286 |
|
e5ac9c5ae
|
2287 |
if (flags & CFLGS_OFF_SLAB) { |
b2d550736
|
2288 |
cachep->slabp_cache = kmem_find_general_cachep(slab_size, 0u); |
e5ac9c5ae
|
2289 2290 2291 2292 2293 2294 2295 |
/* * This is a possibility for one of the malloc_sizes caches. * But since we go off slab only for object size greater than * PAGE_SIZE/8, and malloc_sizes gets created in ascending order, * this should not happen at all. * But leave a BUG_ON for some lucky dude. */ |
6cb8f9132
|
2296 |
BUG_ON(ZERO_OR_NULL_PTR(cachep->slabp_cache)); |
e5ac9c5ae
|
2297 |
} |
1da177e4c
|
2298 |
cachep->ctor = ctor; |
1da177e4c
|
2299 |
cachep->name = name; |
2ed3a4ef9
|
2300 2301 2302 2303 2304 |
if (setup_cpu_cache(cachep)) { __kmem_cache_destroy(cachep); cachep = NULL; goto oops; } |
1da177e4c
|
2305 |
|
1da177e4c
|
2306 2307 |
/* cache setup completed, link it into the list */ list_add(&cachep->next, &cache_chain); |
a737b3e2f
|
2308 |
oops: |
1da177e4c
|
2309 2310 2311 |
if (!cachep && (flags & SLAB_PANIC)) panic("kmem_cache_create(): failed to create slab `%s' ", |
b28a02de8
|
2312 |
name); |
fc0abb145
|
2313 |
mutex_unlock(&cache_chain_mutex); |
95402b382
|
2314 |
put_online_cpus(); |
1da177e4c
|
2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 |
return cachep; } EXPORT_SYMBOL(kmem_cache_create); #if DEBUG static void check_irq_off(void) { BUG_ON(!irqs_disabled()); } static void check_irq_on(void) { BUG_ON(irqs_disabled()); } |
343e0d7a9
|
2329 |
static void check_spinlock_acquired(struct kmem_cache *cachep) |
1da177e4c
|
2330 2331 2332 |
{ #ifdef CONFIG_SMP check_irq_off(); |
e498be7da
|
2333 |
assert_spin_locked(&cachep->nodelists[numa_node_id()]->list_lock); |
1da177e4c
|
2334 2335 |
#endif } |
e498be7da
|
2336 |
|
343e0d7a9
|
2337 |
static void check_spinlock_acquired_node(struct kmem_cache *cachep, int node) |
e498be7da
|
2338 2339 2340 2341 2342 2343 |
{ #ifdef CONFIG_SMP check_irq_off(); assert_spin_locked(&cachep->nodelists[node]->list_lock); #endif } |
1da177e4c
|
2344 2345 2346 2347 |
#else #define check_irq_off() do { } while(0) #define check_irq_on() do { } while(0) #define check_spinlock_acquired(x) do { } while(0) |
e498be7da
|
2348 |
#define check_spinlock_acquired_node(x, y) do { } while(0) |
1da177e4c
|
2349 |
#endif |
aab2207cf
|
2350 2351 2352 |
static void drain_array(struct kmem_cache *cachep, struct kmem_list3 *l3, struct array_cache *ac, int force, int node); |
1da177e4c
|
2353 2354 |
static void do_drain(void *arg) { |
a737b3e2f
|
2355 |
struct kmem_cache *cachep = arg; |
1da177e4c
|
2356 |
struct array_cache *ac; |
ff69416e6
|
2357 |
int node = numa_node_id(); |
1da177e4c
|
2358 2359 |
check_irq_off(); |
9a2dba4b4
|
2360 |
ac = cpu_cache_get(cachep); |
ff69416e6
|
2361 2362 2363 |
spin_lock(&cachep->nodelists[node]->list_lock); free_block(cachep, ac->entry, ac->avail, node); spin_unlock(&cachep->nodelists[node]->list_lock); |
1da177e4c
|
2364 2365 |
ac->avail = 0; } |
343e0d7a9
|
2366 |
static void drain_cpu_caches(struct kmem_cache *cachep) |
1da177e4c
|
2367 |
{ |
e498be7da
|
2368 2369 |
struct kmem_list3 *l3; int node; |
15c8b6c1a
|
2370 |
on_each_cpu(do_drain, cachep, 1); |
1da177e4c
|
2371 |
check_irq_on(); |
b28a02de8
|
2372 |
for_each_online_node(node) { |
e498be7da
|
2373 |
l3 = cachep->nodelists[node]; |
a4523a8b3
|
2374 2375 2376 2377 2378 2379 2380 |
if (l3 && l3->alien) drain_alien_cache(cachep, l3->alien); } for_each_online_node(node) { l3 = cachep->nodelists[node]; if (l3) |
aab2207cf
|
2381 |
drain_array(cachep, l3, l3->shared, 1, node); |
e498be7da
|
2382 |
} |
1da177e4c
|
2383 |
} |
ed11d9eb2
|
2384 2385 2386 2387 2388 2389 2390 2391 |
/* * Remove slabs from the list of free slabs. * Specify the number of slabs to drain in tofree. * * Returns the actual number of slabs released. */ static int drain_freelist(struct kmem_cache *cache, struct kmem_list3 *l3, int tofree) |
1da177e4c
|
2392 |
{ |
ed11d9eb2
|
2393 2394 |
struct list_head *p; int nr_freed; |
1da177e4c
|
2395 |
struct slab *slabp; |
1da177e4c
|
2396 |
|
ed11d9eb2
|
2397 2398 |
nr_freed = 0; while (nr_freed < tofree && !list_empty(&l3->slabs_free)) { |
1da177e4c
|
2399 |
|
ed11d9eb2
|
2400 |
spin_lock_irq(&l3->list_lock); |
e498be7da
|
2401 |
p = l3->slabs_free.prev; |
ed11d9eb2
|
2402 2403 2404 2405 |
if (p == &l3->slabs_free) { spin_unlock_irq(&l3->list_lock); goto out; } |
1da177e4c
|
2406 |
|
ed11d9eb2
|
2407 |
slabp = list_entry(p, struct slab, list); |
1da177e4c
|
2408 |
#if DEBUG |
40094fa65
|
2409 |
BUG_ON(slabp->inuse); |
1da177e4c
|
2410 2411 |
#endif list_del(&slabp->list); |
ed11d9eb2
|
2412 2413 2414 2415 2416 |
/* * Safe to drop the lock. The slab is no longer linked * to the cache. */ l3->free_objects -= cache->num; |
e498be7da
|
2417 |
spin_unlock_irq(&l3->list_lock); |
ed11d9eb2
|
2418 2419 |
slab_destroy(cache, slabp); nr_freed++; |
1da177e4c
|
2420 |
} |
ed11d9eb2
|
2421 2422 |
out: return nr_freed; |
1da177e4c
|
2423 |
} |
8f5be20bf
|
2424 |
/* Called with cache_chain_mutex held to protect against cpu hotplug */ |
343e0d7a9
|
2425 |
static int __cache_shrink(struct kmem_cache *cachep) |
e498be7da
|
2426 2427 2428 2429 2430 2431 2432 2433 2434 |
{ int ret = 0, i = 0; struct kmem_list3 *l3; drain_cpu_caches(cachep); check_irq_on(); for_each_online_node(i) { l3 = cachep->nodelists[i]; |
ed11d9eb2
|
2435 2436 2437 2438 2439 2440 2441 |
if (!l3) continue; drain_freelist(cachep, l3, l3->free_objects); ret += !list_empty(&l3->slabs_full) || !list_empty(&l3->slabs_partial); |
e498be7da
|
2442 2443 2444 |
} return (ret ? 1 : 0); } |
1da177e4c
|
2445 2446 2447 2448 2449 2450 2451 |
/** * kmem_cache_shrink - Shrink a cache. * @cachep: The cache to shrink. * * Releases as many slabs as possible for a cache. * To help debugging, a zero exit status indicates all slabs were released. */ |
343e0d7a9
|
2452 |
int kmem_cache_shrink(struct kmem_cache *cachep) |
1da177e4c
|
2453 |
{ |
8f5be20bf
|
2454 |
int ret; |
40094fa65
|
2455 |
BUG_ON(!cachep || in_interrupt()); |
1da177e4c
|
2456 |
|
95402b382
|
2457 |
get_online_cpus(); |
8f5be20bf
|
2458 2459 2460 |
mutex_lock(&cache_chain_mutex); ret = __cache_shrink(cachep); mutex_unlock(&cache_chain_mutex); |
95402b382
|
2461 |
put_online_cpus(); |
8f5be20bf
|
2462 |
return ret; |
1da177e4c
|
2463 2464 2465 2466 2467 2468 2469 |
} EXPORT_SYMBOL(kmem_cache_shrink); /** * kmem_cache_destroy - delete a cache * @cachep: the cache to destroy * |
72fd4a35a
|
2470 |
* Remove a &struct kmem_cache object from the slab cache. |
1da177e4c
|
2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 |
* * It is expected this function will be called by a module when it is * unloaded. This will remove the cache completely, and avoid a duplicate * cache being allocated each time a module is loaded and unloaded, if the * module doesn't have persistent in-kernel storage across loads and unloads. * * The cache must be empty before calling this function. * * The caller must guarantee that noone will allocate memory from the cache * during the kmem_cache_destroy(). */ |
133d205a1
|
2482 |
void kmem_cache_destroy(struct kmem_cache *cachep) |
1da177e4c
|
2483 |
{ |
40094fa65
|
2484 |
BUG_ON(!cachep || in_interrupt()); |
1da177e4c
|
2485 |
|
1da177e4c
|
2486 |
/* Find the cache in the chain of caches. */ |
95402b382
|
2487 |
get_online_cpus(); |
fc0abb145
|
2488 |
mutex_lock(&cache_chain_mutex); |
1da177e4c
|
2489 2490 2491 2492 |
/* * the chain is never empty, cache_cache is never destroyed */ list_del(&cachep->next); |
1da177e4c
|
2493 2494 |
if (__cache_shrink(cachep)) { slab_error(cachep, "Can't free all objects"); |
b28a02de8
|
2495 |
list_add(&cachep->next, &cache_chain); |
fc0abb145
|
2496 |
mutex_unlock(&cache_chain_mutex); |
95402b382
|
2497 |
put_online_cpus(); |
133d205a1
|
2498 |
return; |
1da177e4c
|
2499 2500 2501 |
} if (unlikely(cachep->flags & SLAB_DESTROY_BY_RCU)) |
fbd568a3e
|
2502 |
synchronize_rcu(); |
1da177e4c
|
2503 |
|
117f6eb1d
|
2504 |
__kmem_cache_destroy(cachep); |
8f5be20bf
|
2505 |
mutex_unlock(&cache_chain_mutex); |
95402b382
|
2506 |
put_online_cpus(); |
1da177e4c
|
2507 2508 |
} EXPORT_SYMBOL(kmem_cache_destroy); |
e5ac9c5ae
|
2509 2510 2511 2512 2513 2514 2515 2516 2517 2518 2519 |
/* * Get the memory for a slab management obj. * For a slab cache when the slab descriptor is off-slab, slab descriptors * always come from malloc_sizes caches. The slab descriptor cannot * come from the same cache which is getting created because, * when we are searching for an appropriate cache for these * descriptors in kmem_cache_create, we search through the malloc_sizes array. * If we are creating a malloc_sizes cache here it would not be visible to * kmem_find_general_cachep till the initialization is complete. * Hence we cannot have slabp_cache same as the original cache. */ |
343e0d7a9
|
2520 |
static struct slab *alloc_slabmgmt(struct kmem_cache *cachep, void *objp, |
5b74ada7e
|
2521 2522 |
int colour_off, gfp_t local_flags, int nodeid) |
1da177e4c
|
2523 2524 |
{ struct slab *slabp; |
b28a02de8
|
2525 |
|
1da177e4c
|
2526 2527 |
if (OFF_SLAB(cachep)) { /* Slab management obj is off-slab. */ |
5b74ada7e
|
2528 |
slabp = kmem_cache_alloc_node(cachep->slabp_cache, |
8759ec50a
|
2529 |
local_flags, nodeid); |
1da177e4c
|
2530 2531 2532 |
if (!slabp) return NULL; } else { |
b28a02de8
|
2533 |
slabp = objp + colour_off; |
1da177e4c
|
2534 2535 2536 2537 |
colour_off += cachep->slab_size; } slabp->inuse = 0; slabp->colouroff = colour_off; |
b28a02de8
|
2538 |
slabp->s_mem = objp + colour_off; |
5b74ada7e
|
2539 |
slabp->nodeid = nodeid; |
e51bfd0ad
|
2540 |
slabp->free = 0; |
1da177e4c
|
2541 2542 2543 2544 2545 |
return slabp; } static inline kmem_bufctl_t *slab_bufctl(struct slab *slabp) { |
b28a02de8
|
2546 |
return (kmem_bufctl_t *) (slabp + 1); |
1da177e4c
|
2547 |
} |
343e0d7a9
|
2548 |
static void cache_init_objs(struct kmem_cache *cachep, |
a35afb830
|
2549 |
struct slab *slabp) |
1da177e4c
|
2550 2551 2552 2553 |
{ int i; for (i = 0; i < cachep->num; i++) { |
8fea4e96a
|
2554 |
void *objp = index_to_obj(cachep, slabp, i); |
1da177e4c
|
2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565 2566 |
#if DEBUG /* need to poison the objs? */ if (cachep->flags & SLAB_POISON) poison_obj(cachep, objp, POISON_FREE); if (cachep->flags & SLAB_STORE_USER) *dbg_userword(cachep, objp) = NULL; if (cachep->flags & SLAB_RED_ZONE) { *dbg_redzone1(cachep, objp) = RED_INACTIVE; *dbg_redzone2(cachep, objp) = RED_INACTIVE; } /* |
a737b3e2f
|
2567 2568 2569 |
* Constructors are not allowed to allocate memory from the same * cache which they are a constructor for. Otherwise, deadlock. * They must also be threaded. |
1da177e4c
|
2570 2571 |
*/ if (cachep->ctor && !(cachep->flags & SLAB_POISON)) |
51cc50685
|
2572 |
cachep->ctor(objp + obj_offset(cachep)); |
1da177e4c
|
2573 2574 2575 2576 |
if (cachep->flags & SLAB_RED_ZONE) { if (*dbg_redzone2(cachep, objp) != RED_INACTIVE) slab_error(cachep, "constructor overwrote the" |
b28a02de8
|
2577 |
" end of an object"); |
1da177e4c
|
2578 2579 |
if (*dbg_redzone1(cachep, objp) != RED_INACTIVE) slab_error(cachep, "constructor overwrote the" |
b28a02de8
|
2580 |
" start of an object"); |
1da177e4c
|
2581 |
} |
a737b3e2f
|
2582 2583 |
if ((cachep->buffer_size % PAGE_SIZE) == 0 && OFF_SLAB(cachep) && cachep->flags & SLAB_POISON) |
b28a02de8
|
2584 |
kernel_map_pages(virt_to_page(objp), |
3dafccf22
|
2585 |
cachep->buffer_size / PAGE_SIZE, 0); |
1da177e4c
|
2586 2587 |
#else if (cachep->ctor) |
51cc50685
|
2588 |
cachep->ctor(objp); |
1da177e4c
|
2589 |
#endif |
b28a02de8
|
2590 |
slab_bufctl(slabp)[i] = i + 1; |
1da177e4c
|
2591 |
} |
b28a02de8
|
2592 |
slab_bufctl(slabp)[i - 1] = BUFCTL_END; |
1da177e4c
|
2593 |
} |
343e0d7a9
|
2594 |
static void kmem_flagcheck(struct kmem_cache *cachep, gfp_t flags) |
1da177e4c
|
2595 |
{ |
4b51d6698
|
2596 2597 2598 2599 2600 2601 |
if (CONFIG_ZONE_DMA_FLAG) { if (flags & GFP_DMA) BUG_ON(!(cachep->gfpflags & GFP_DMA)); else BUG_ON(cachep->gfpflags & GFP_DMA); } |
1da177e4c
|
2602 |
} |
a737b3e2f
|
2603 2604 |
static void *slab_get_obj(struct kmem_cache *cachep, struct slab *slabp, int nodeid) |
78d382d77
|
2605 |
{ |
8fea4e96a
|
2606 |
void *objp = index_to_obj(cachep, slabp, slabp->free); |
78d382d77
|
2607 2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 |
kmem_bufctl_t next; slabp->inuse++; next = slab_bufctl(slabp)[slabp->free]; #if DEBUG slab_bufctl(slabp)[slabp->free] = BUFCTL_FREE; WARN_ON(slabp->nodeid != nodeid); #endif slabp->free = next; return objp; } |
a737b3e2f
|
2619 2620 |
static void slab_put_obj(struct kmem_cache *cachep, struct slab *slabp, void *objp, int nodeid) |
78d382d77
|
2621 |
{ |
8fea4e96a
|
2622 |
unsigned int objnr = obj_to_index(cachep, slabp, objp); |
78d382d77
|
2623 2624 2625 2626 |
#if DEBUG /* Verify that the slab belongs to the intended node */ WARN_ON(slabp->nodeid != nodeid); |
871751e25
|
2627 |
if (slab_bufctl(slabp)[objnr] + 1 <= SLAB_LIMIT + 1) { |
78d382d77
|
2628 |
printk(KERN_ERR "slab: double free detected in cache " |
a737b3e2f
|
2629 2630 |
"'%s', objp %p ", cachep->name, objp); |
78d382d77
|
2631 2632 2633 2634 2635 2636 2637 |
BUG(); } #endif slab_bufctl(slabp)[objnr] = slabp->free; slabp->free = objnr; slabp->inuse--; } |
4776874ff
|
2638 2639 2640 2641 2642 2643 2644 |
/* * Map pages beginning at addr to the given cache and slab. This is required * for the slab allocator to be able to lookup the cache and slab of a * virtual address for kfree, ksize, kmem_ptr_validate, and slab debugging. */ static void slab_map_pages(struct kmem_cache *cache, struct slab *slab, void *addr) |
1da177e4c
|
2645 |
{ |
4776874ff
|
2646 |
int nr_pages; |
1da177e4c
|
2647 |
struct page *page; |
4776874ff
|
2648 |
page = virt_to_page(addr); |
84097518d
|
2649 |
|
4776874ff
|
2650 |
nr_pages = 1; |
84097518d
|
2651 |
if (likely(!PageCompound(page))) |
4776874ff
|
2652 |
nr_pages <<= cache->gfporder; |
1da177e4c
|
2653 |
do { |
4776874ff
|
2654 2655 |
page_set_cache(page, cache); page_set_slab(page, slab); |
1da177e4c
|
2656 |
page++; |
4776874ff
|
2657 |
} while (--nr_pages); |
1da177e4c
|
2658 2659 2660 2661 2662 2663 |
} /* * Grow (by 1) the number of slabs within a cache. This is called by * kmem_cache_alloc() when there are no active objs left in a cache. */ |
3c517a613
|
2664 2665 |
static int cache_grow(struct kmem_cache *cachep, gfp_t flags, int nodeid, void *objp) |
1da177e4c
|
2666 |
{ |
b28a02de8
|
2667 |
struct slab *slabp; |
b28a02de8
|
2668 2669 |
size_t offset; gfp_t local_flags; |
e498be7da
|
2670 |
struct kmem_list3 *l3; |
1da177e4c
|
2671 |
|
a737b3e2f
|
2672 2673 2674 |
/* * Be lazy and only check for valid flags here, keeping it out of the * critical path in kmem_cache_alloc(). |
1da177e4c
|
2675 |
*/ |
6cb062296
|
2676 2677 |
BUG_ON(flags & GFP_SLAB_BUG_MASK); local_flags = flags & (GFP_CONSTRAINT_MASK|GFP_RECLAIM_MASK); |
1da177e4c
|
2678 |
|
2e1217cf9
|
2679 |
/* Take the l3 list lock to change the colour_next on this node */ |
1da177e4c
|
2680 |
check_irq_off(); |
2e1217cf9
|
2681 2682 |
l3 = cachep->nodelists[nodeid]; spin_lock(&l3->list_lock); |
1da177e4c
|
2683 2684 |
/* Get colour for the slab, and cal the next value. */ |
2e1217cf9
|
2685 2686 2687 2688 2689 |
offset = l3->colour_next; l3->colour_next++; if (l3->colour_next >= cachep->colour) l3->colour_next = 0; spin_unlock(&l3->list_lock); |
1da177e4c
|
2690 |
|
2e1217cf9
|
2691 |
offset *= cachep->colour_off; |
1da177e4c
|
2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 |
if (local_flags & __GFP_WAIT) local_irq_enable(); /* * The test for missing atomic flag is performed here, rather than * the more obvious place, simply to reduce the critical path length * in kmem_cache_alloc(). If a caller is seriously mis-behaving they * will eventually be caught here (where it matters). */ kmem_flagcheck(cachep, flags); |
a737b3e2f
|
2703 2704 2705 |
/* * Get mem for the objs. Attempt to allocate a physical page from * 'nodeid'. |
e498be7da
|
2706 |
*/ |
3c517a613
|
2707 |
if (!objp) |
b8c1c5da1
|
2708 |
objp = kmem_getpages(cachep, local_flags, nodeid); |
a737b3e2f
|
2709 |
if (!objp) |
1da177e4c
|
2710 2711 2712 |
goto failed; /* Get slab management. */ |
3c517a613
|
2713 |
slabp = alloc_slabmgmt(cachep, objp, offset, |
6cb062296
|
2714 |
local_flags & ~GFP_CONSTRAINT_MASK, nodeid); |
a737b3e2f
|
2715 |
if (!slabp) |
1da177e4c
|
2716 |
goto opps1; |
4776874ff
|
2717 |
slab_map_pages(cachep, slabp, objp); |
1da177e4c
|
2718 |
|
a35afb830
|
2719 |
cache_init_objs(cachep, slabp); |
1da177e4c
|
2720 2721 2722 2723 |
if (local_flags & __GFP_WAIT) local_irq_disable(); check_irq_off(); |
e498be7da
|
2724 |
spin_lock(&l3->list_lock); |
1da177e4c
|
2725 2726 |
/* Make slab active. */ |
e498be7da
|
2727 |
list_add_tail(&slabp->list, &(l3->slabs_free)); |
1da177e4c
|
2728 |
STATS_INC_GROWN(cachep); |
e498be7da
|
2729 2730 |
l3->free_objects += cachep->num; spin_unlock(&l3->list_lock); |
1da177e4c
|
2731 |
return 1; |
a737b3e2f
|
2732 |
opps1: |
1da177e4c
|
2733 |
kmem_freepages(cachep, objp); |
a737b3e2f
|
2734 |
failed: |
1da177e4c
|
2735 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 |
if (local_flags & __GFP_WAIT) local_irq_disable(); return 0; } #if DEBUG /* * Perform extra freeing checks: * - detect bad pointers. * - POISON/RED_ZONE checking |
1da177e4c
|
2746 2747 2748 |
*/ static void kfree_debugcheck(const void *objp) { |
1da177e4c
|
2749 2750 2751 |
if (!virt_addr_valid(objp)) { printk(KERN_ERR "kfree_debugcheck: out of range ptr %lxh. ", |
b28a02de8
|
2752 2753 |
(unsigned long)objp); BUG(); |
1da177e4c
|
2754 |
} |
1da177e4c
|
2755 |
} |
58ce1fd58
|
2756 2757 |
static inline void verify_redzone_free(struct kmem_cache *cache, void *obj) { |
b46b8f19c
|
2758 |
unsigned long long redzone1, redzone2; |
58ce1fd58
|
2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 |
redzone1 = *dbg_redzone1(cache, obj); redzone2 = *dbg_redzone2(cache, obj); /* * Redzone is ok. */ if (redzone1 == RED_ACTIVE && redzone2 == RED_ACTIVE) return; if (redzone1 == RED_INACTIVE && redzone2 == RED_INACTIVE) slab_error(cache, "double free detected"); else slab_error(cache, "memory outside object was overwritten"); |
b46b8f19c
|
2773 2774 |
printk(KERN_ERR "%p: redzone 1:0x%llx, redzone 2:0x%llx. ", |
58ce1fd58
|
2775 2776 |
obj, redzone1, redzone2); } |
343e0d7a9
|
2777 |
static void *cache_free_debugcheck(struct kmem_cache *cachep, void *objp, |
b28a02de8
|
2778 |
void *caller) |
1da177e4c
|
2779 2780 2781 2782 |
{ struct page *page; unsigned int objnr; struct slab *slabp; |
80cbd911c
|
2783 |
BUG_ON(virt_to_cache(objp) != cachep); |
3dafccf22
|
2784 |
objp -= obj_offset(cachep); |
1da177e4c
|
2785 |
kfree_debugcheck(objp); |
b49af68ff
|
2786 |
page = virt_to_head_page(objp); |
1da177e4c
|
2787 |
|
065d41cb2
|
2788 |
slabp = page_get_slab(page); |
1da177e4c
|
2789 2790 |
if (cachep->flags & SLAB_RED_ZONE) { |
58ce1fd58
|
2791 |
verify_redzone_free(cachep, objp); |
1da177e4c
|
2792 2793 2794 2795 2796 |
*dbg_redzone1(cachep, objp) = RED_INACTIVE; *dbg_redzone2(cachep, objp) = RED_INACTIVE; } if (cachep->flags & SLAB_STORE_USER) *dbg_userword(cachep, objp) = caller; |
8fea4e96a
|
2797 |
objnr = obj_to_index(cachep, slabp, objp); |
1da177e4c
|
2798 2799 |
BUG_ON(objnr >= cachep->num); |
8fea4e96a
|
2800 |
BUG_ON(objp != index_to_obj(cachep, slabp, objnr)); |
1da177e4c
|
2801 |
|
871751e25
|
2802 2803 2804 |
#ifdef CONFIG_DEBUG_SLAB_LEAK slab_bufctl(slabp)[objnr] = BUFCTL_FREE; #endif |
1da177e4c
|
2805 2806 |
if (cachep->flags & SLAB_POISON) { #ifdef CONFIG_DEBUG_PAGEALLOC |
a737b3e2f
|
2807 |
if ((cachep->buffer_size % PAGE_SIZE)==0 && OFF_SLAB(cachep)) { |
1da177e4c
|
2808 |
store_stackinfo(cachep, objp, (unsigned long)caller); |
b28a02de8
|
2809 |
kernel_map_pages(virt_to_page(objp), |
3dafccf22
|
2810 |
cachep->buffer_size / PAGE_SIZE, 0); |
1da177e4c
|
2811 2812 2813 2814 2815 2816 2817 2818 2819 |
} else { poison_obj(cachep, objp, POISON_FREE); } #else poison_obj(cachep, objp, POISON_FREE); #endif } return objp; } |
343e0d7a9
|
2820 |
static void check_slabp(struct kmem_cache *cachep, struct slab *slabp) |
1da177e4c
|
2821 2822 2823 |
{ kmem_bufctl_t i; int entries = 0; |
b28a02de8
|
2824 |
|
1da177e4c
|
2825 2826 2827 2828 2829 2830 2831 |
/* Check slab's freelist to see if this obj is there. */ for (i = slabp->free; i != BUFCTL_END; i = slab_bufctl(slabp)[i]) { entries++; if (entries > cachep->num || i >= cachep->num) goto bad; } if (entries != cachep->num - slabp->inuse) { |
a737b3e2f
|
2832 2833 2834 2835 2836 |
bad: printk(KERN_ERR "slab: Internal list corruption detected in " "cache '%s'(%d), slabp %p(%d). Hexdump: ", cachep->name, cachep->num, slabp, slabp->inuse); |
b28a02de8
|
2837 |
for (i = 0; |
264132bc6
|
2838 |
i < sizeof(*slabp) + cachep->num * sizeof(kmem_bufctl_t); |
b28a02de8
|
2839 |
i++) { |
a737b3e2f
|
2840 |
if (i % 16 == 0) |
1da177e4c
|
2841 2842 |
printk(" %03x:", i); |
b28a02de8
|
2843 |
printk(" %02x", ((unsigned char *)slabp)[i]); |
1da177e4c
|
2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 |
} printk(" "); BUG(); } } #else #define kfree_debugcheck(x) do { } while(0) #define cache_free_debugcheck(x,objp,z) (objp) #define check_slabp(x,y) do { } while(0) #endif |
343e0d7a9
|
2855 |
static void *cache_alloc_refill(struct kmem_cache *cachep, gfp_t flags) |
1da177e4c
|
2856 2857 2858 2859 |
{ int batchcount; struct kmem_list3 *l3; struct array_cache *ac; |
1ca4cb241
|
2860 |
int node; |
6d2144d35
|
2861 |
retry: |
1da177e4c
|
2862 |
check_irq_off(); |
6d2144d35
|
2863 |
node = numa_node_id(); |
9a2dba4b4
|
2864 |
ac = cpu_cache_get(cachep); |
1da177e4c
|
2865 2866 |
batchcount = ac->batchcount; if (!ac->touched && batchcount > BATCHREFILL_LIMIT) { |
a737b3e2f
|
2867 2868 2869 2870 |
/* * If there was little recent activity on this cache, then * perform only a partial refill. Otherwise we could generate * refill bouncing. |
1da177e4c
|
2871 2872 2873 |
*/ batchcount = BATCHREFILL_LIMIT; } |
1ca4cb241
|
2874 |
l3 = cachep->nodelists[node]; |
e498be7da
|
2875 2876 2877 |
BUG_ON(ac->avail > 0 || !l3); spin_lock(&l3->list_lock); |
1da177e4c
|
2878 |
|
3ded175a4
|
2879 2880 2881 |
/* See if we can refill from the shared array */ if (l3->shared && transfer_objects(ac, l3->shared, batchcount)) goto alloc_done; |
1da177e4c
|
2882 2883 2884 2885 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 |
while (batchcount > 0) { struct list_head *entry; struct slab *slabp; /* Get slab alloc is to come from. */ entry = l3->slabs_partial.next; if (entry == &l3->slabs_partial) { l3->free_touched = 1; entry = l3->slabs_free.next; if (entry == &l3->slabs_free) goto must_grow; } slabp = list_entry(entry, struct slab, list); check_slabp(cachep, slabp); check_spinlock_acquired(cachep); |
714b8171a
|
2897 2898 2899 2900 2901 2902 |
/* * The slab was either on partial or free list so * there must be at least one object available for * allocation. */ |
249b9f331
|
2903 |
BUG_ON(slabp->inuse >= cachep->num); |
714b8171a
|
2904 |
|
1da177e4c
|
2905 |
while (slabp->inuse < cachep->num && batchcount--) { |
1da177e4c
|
2906 2907 2908 |
STATS_INC_ALLOCED(cachep); STATS_INC_ACTIVE(cachep); STATS_SET_HIGH(cachep); |
78d382d77
|
2909 |
ac->entry[ac->avail++] = slab_get_obj(cachep, slabp, |
1ca4cb241
|
2910 |
node); |
1da177e4c
|
2911 2912 2913 2914 2915 2916 2917 2918 2919 2920 |
} check_slabp(cachep, slabp); /* move slabp to correct slabp list: */ list_del(&slabp->list); if (slabp->free == BUFCTL_END) list_add(&slabp->list, &l3->slabs_full); else list_add(&slabp->list, &l3->slabs_partial); } |
a737b3e2f
|
2921 |
must_grow: |
1da177e4c
|
2922 |
l3->free_objects -= ac->avail; |
a737b3e2f
|
2923 |
alloc_done: |
e498be7da
|
2924 |
spin_unlock(&l3->list_lock); |
1da177e4c
|
2925 2926 2927 |
if (unlikely(!ac->avail)) { int x; |
3c517a613
|
2928 |
x = cache_grow(cachep, flags | GFP_THISNODE, node, NULL); |
e498be7da
|
2929 |
|
a737b3e2f
|
2930 |
/* cache_grow can reenable interrupts, then ac could change. */ |
9a2dba4b4
|
2931 |
ac = cpu_cache_get(cachep); |
a737b3e2f
|
2932 |
if (!x && ac->avail == 0) /* no objects in sight? abort */ |
1da177e4c
|
2933 |
return NULL; |
a737b3e2f
|
2934 |
if (!ac->avail) /* objects refilled by interrupt? */ |
1da177e4c
|
2935 2936 2937 |
goto retry; } ac->touched = 1; |
e498be7da
|
2938 |
return ac->entry[--ac->avail]; |
1da177e4c
|
2939 |
} |
a737b3e2f
|
2940 2941 |
static inline void cache_alloc_debugcheck_before(struct kmem_cache *cachep, gfp_t flags) |
1da177e4c
|
2942 2943 2944 2945 2946 2947 2948 2949 |
{ might_sleep_if(flags & __GFP_WAIT); #if DEBUG kmem_flagcheck(cachep, flags); #endif } #if DEBUG |
a737b3e2f
|
2950 2951 |
static void *cache_alloc_debugcheck_after(struct kmem_cache *cachep, gfp_t flags, void *objp, void *caller) |
1da177e4c
|
2952 |
{ |
b28a02de8
|
2953 |
if (!objp) |
1da177e4c
|
2954 |
return objp; |
b28a02de8
|
2955 |
if (cachep->flags & SLAB_POISON) { |
1da177e4c
|
2956 |
#ifdef CONFIG_DEBUG_PAGEALLOC |
3dafccf22
|
2957 |
if ((cachep->buffer_size % PAGE_SIZE) == 0 && OFF_SLAB(cachep)) |
b28a02de8
|
2958 |
kernel_map_pages(virt_to_page(objp), |
3dafccf22
|
2959 |
cachep->buffer_size / PAGE_SIZE, 1); |
1da177e4c
|
2960 2961 2962 2963 2964 2965 2966 2967 2968 2969 2970 |
else check_poison_obj(cachep, objp); #else check_poison_obj(cachep, objp); #endif poison_obj(cachep, objp, POISON_INUSE); } if (cachep->flags & SLAB_STORE_USER) *dbg_userword(cachep, objp) = caller; if (cachep->flags & SLAB_RED_ZONE) { |
a737b3e2f
|
2971 2972 2973 2974 |
if (*dbg_redzone1(cachep, objp) != RED_INACTIVE || *dbg_redzone2(cachep, objp) != RED_INACTIVE) { slab_error(cachep, "double free, or memory outside" " object was overwritten"); |
b28a02de8
|
2975 |
printk(KERN_ERR |
b46b8f19c
|
2976 2977 |
"%p: redzone 1:0x%llx, redzone 2:0x%llx ", |
a737b3e2f
|
2978 2979 |
objp, *dbg_redzone1(cachep, objp), *dbg_redzone2(cachep, objp)); |
1da177e4c
|
2980 2981 2982 2983 |
} *dbg_redzone1(cachep, objp) = RED_ACTIVE; *dbg_redzone2(cachep, objp) = RED_ACTIVE; } |
871751e25
|
2984 2985 2986 2987 |
#ifdef CONFIG_DEBUG_SLAB_LEAK { struct slab *slabp; unsigned objnr; |
b49af68ff
|
2988 |
slabp = page_get_slab(virt_to_head_page(objp)); |
871751e25
|
2989 2990 2991 2992 |
objnr = (unsigned)(objp - slabp->s_mem) / cachep->buffer_size; slab_bufctl(slabp)[objnr] = BUFCTL_ACTIVE; } #endif |
3dafccf22
|
2993 |
objp += obj_offset(cachep); |
4f1049345
|
2994 |
if (cachep->ctor && cachep->flags & SLAB_POISON) |
51cc50685
|
2995 |
cachep->ctor(objp); |
a44b56d35
|
2996 2997 2998 2999 3000 3001 3002 |
#if ARCH_SLAB_MINALIGN if ((u32)objp & (ARCH_SLAB_MINALIGN-1)) { printk(KERN_ERR "0x%p: not aligned to ARCH_SLAB_MINALIGN=%d ", objp, ARCH_SLAB_MINALIGN); } #endif |
1da177e4c
|
3003 3004 3005 3006 3007 |
return objp; } #else #define cache_alloc_debugcheck_after(a,b,objp,d) (objp) #endif |
773ff60e8
|
3008 |
static bool slab_should_failslab(struct kmem_cache *cachep, gfp_t flags) |
8a8b6502f
|
3009 3010 |
{ if (cachep == &cache_cache) |
773ff60e8
|
3011 |
return false; |
8a8b6502f
|
3012 |
|
773ff60e8
|
3013 |
return should_failslab(obj_size(cachep), flags); |
8a8b6502f
|
3014 |
} |
343e0d7a9
|
3015 |
static inline void *____cache_alloc(struct kmem_cache *cachep, gfp_t flags) |
1da177e4c
|
3016 |
{ |
b28a02de8
|
3017 |
void *objp; |
1da177e4c
|
3018 |
struct array_cache *ac; |
5c3823008
|
3019 |
check_irq_off(); |
8a8b6502f
|
3020 |
|
9a2dba4b4
|
3021 |
ac = cpu_cache_get(cachep); |
1da177e4c
|
3022 3023 3024 |
if (likely(ac->avail)) { STATS_INC_ALLOCHIT(cachep); ac->touched = 1; |
e498be7da
|
3025 |
objp = ac->entry[--ac->avail]; |
1da177e4c
|
3026 3027 3028 3029 |
} else { STATS_INC_ALLOCMISS(cachep); objp = cache_alloc_refill(cachep, flags); } |
5c3823008
|
3030 3031 |
return objp; } |
e498be7da
|
3032 3033 |
#ifdef CONFIG_NUMA /* |
b2455396b
|
3034 |
* Try allocating on another node if PF_SPREAD_SLAB|PF_MEMPOLICY. |
c61afb181
|
3035 3036 3037 3038 3039 3040 3041 |
* * If we are in_interrupt, then process context, including cpusets and * mempolicy, may not apply and should not be used for allocation policy. */ static void *alternate_node_alloc(struct kmem_cache *cachep, gfp_t flags) { int nid_alloc, nid_here; |
765c4507a
|
3042 |
if (in_interrupt() || (flags & __GFP_THISNODE)) |
c61afb181
|
3043 3044 3045 3046 3047 3048 3049 |
return NULL; nid_alloc = nid_here = numa_node_id(); if (cpuset_do_slab_mem_spread() && (cachep->flags & SLAB_MEM_SPREAD)) nid_alloc = cpuset_mem_spread_node(); else if (current->mempolicy) nid_alloc = slab_node(current->mempolicy); if (nid_alloc != nid_here) |
8b98c1699
|
3050 |
return ____cache_alloc_node(cachep, flags, nid_alloc); |
c61afb181
|
3051 3052 3053 3054 |
return NULL; } /* |
765c4507a
|
3055 |
* Fallback function if there was no memory available and no objects on a |
3c517a613
|
3056 3057 3058 3059 3060 |
* certain node and fall back is permitted. First we scan all the * available nodelists for available objects. If that fails then we * perform an allocation without specifying a node. This allows the page * allocator to do its reclaim / fallback magic. We then insert the * slab into the proper nodelist and then allocate from it. |
765c4507a
|
3061 |
*/ |
8c8cc2c10
|
3062 |
static void *fallback_alloc(struct kmem_cache *cache, gfp_t flags) |
765c4507a
|
3063 |
{ |
8c8cc2c10
|
3064 3065 |
struct zonelist *zonelist; gfp_t local_flags; |
dd1a239f6
|
3066 |
struct zoneref *z; |
54a6eb5c4
|
3067 3068 |
struct zone *zone; enum zone_type high_zoneidx = gfp_zone(flags); |
765c4507a
|
3069 |
void *obj = NULL; |
3c517a613
|
3070 |
int nid; |
8c8cc2c10
|
3071 3072 3073 |
if (flags & __GFP_THISNODE) return NULL; |
0e88460da
|
3074 |
zonelist = node_zonelist(slab_node(current->mempolicy), flags); |
6cb062296
|
3075 |
local_flags = flags & (GFP_CONSTRAINT_MASK|GFP_RECLAIM_MASK); |
765c4507a
|
3076 |
|
3c517a613
|
3077 3078 3079 3080 3081 |
retry: /* * Look through allowed nodes for objects available * from existing per node queues. */ |
54a6eb5c4
|
3082 3083 |
for_each_zone_zonelist(zone, z, zonelist, high_zoneidx) { nid = zone_to_nid(zone); |
aedb0eb10
|
3084 |
|
54a6eb5c4
|
3085 |
if (cpuset_zone_allowed_hardwall(zone, flags) && |
3c517a613
|
3086 |
cache->nodelists[nid] && |
481c5346d
|
3087 |
cache->nodelists[nid]->free_objects) { |
3c517a613
|
3088 3089 |
obj = ____cache_alloc_node(cache, flags | GFP_THISNODE, nid); |
481c5346d
|
3090 3091 3092 |
if (obj) break; } |
3c517a613
|
3093 |
} |
cfce66047
|
3094 |
if (!obj) { |
3c517a613
|
3095 3096 3097 3098 3099 3100 |
/* * This allocation will be performed within the constraints * of the current cpuset / memory policy requirements. * We may trigger various forms of reclaim on the allowed * set and go into memory reserves if necessary. */ |
dd47ea755
|
3101 3102 3103 |
if (local_flags & __GFP_WAIT) local_irq_enable(); kmem_flagcheck(cache, flags); |
9ac33b2b7
|
3104 |
obj = kmem_getpages(cache, local_flags, -1); |
dd47ea755
|
3105 3106 |
if (local_flags & __GFP_WAIT) local_irq_disable(); |
3c517a613
|
3107 3108 3109 3110 3111 3112 3113 3114 3115 3116 3117 3118 3119 3120 3121 3122 |
if (obj) { /* * Insert into the appropriate per node queues */ nid = page_to_nid(virt_to_page(obj)); if (cache_grow(cache, flags, nid, obj)) { obj = ____cache_alloc_node(cache, flags | GFP_THISNODE, nid); if (!obj) /* * Another processor may allocate the * objects in the slab since we are * not holding any locks. */ goto retry; } else { |
b6a604518
|
3123 |
/* cache_grow already freed obj */ |
3c517a613
|
3124 3125 3126 |
obj = NULL; } } |
aedb0eb10
|
3127 |
} |
765c4507a
|
3128 3129 3130 3131 |
return obj; } /* |
e498be7da
|
3132 |
* A interface to enable slab creation on nodeid |
1da177e4c
|
3133 |
*/ |
8b98c1699
|
3134 |
static void *____cache_alloc_node(struct kmem_cache *cachep, gfp_t flags, |
a737b3e2f
|
3135 |
int nodeid) |
e498be7da
|
3136 3137 |
{ struct list_head *entry; |
b28a02de8
|
3138 3139 3140 |
struct slab *slabp; struct kmem_list3 *l3; void *obj; |
b28a02de8
|
3141 3142 3143 3144 |
int x; l3 = cachep->nodelists[nodeid]; BUG_ON(!l3); |
a737b3e2f
|
3145 |
retry: |
ca3b9b917
|
3146 |
check_irq_off(); |
b28a02de8
|
3147 3148 3149 3150 3151 3152 3153 3154 3155 3156 3157 3158 3159 3160 3161 3162 3163 3164 |
spin_lock(&l3->list_lock); entry = l3->slabs_partial.next; if (entry == &l3->slabs_partial) { l3->free_touched = 1; entry = l3->slabs_free.next; if (entry == &l3->slabs_free) goto must_grow; } slabp = list_entry(entry, struct slab, list); check_spinlock_acquired_node(cachep, nodeid); check_slabp(cachep, slabp); STATS_INC_NODEALLOCS(cachep); STATS_INC_ACTIVE(cachep); STATS_SET_HIGH(cachep); BUG_ON(slabp->inuse == cachep->num); |
78d382d77
|
3165 |
obj = slab_get_obj(cachep, slabp, nodeid); |
b28a02de8
|
3166 3167 3168 3169 |
check_slabp(cachep, slabp); l3->free_objects--; /* move slabp to correct slabp list: */ list_del(&slabp->list); |
a737b3e2f
|
3170 |
if (slabp->free == BUFCTL_END) |
b28a02de8
|
3171 |
list_add(&slabp->list, &l3->slabs_full); |
a737b3e2f
|
3172 |
else |
b28a02de8
|
3173 |
list_add(&slabp->list, &l3->slabs_partial); |
e498be7da
|
3174 |
|
b28a02de8
|
3175 3176 |
spin_unlock(&l3->list_lock); goto done; |
e498be7da
|
3177 |
|
a737b3e2f
|
3178 |
must_grow: |
b28a02de8
|
3179 |
spin_unlock(&l3->list_lock); |
3c517a613
|
3180 |
x = cache_grow(cachep, flags | GFP_THISNODE, nodeid, NULL); |
765c4507a
|
3181 3182 |
if (x) goto retry; |
1da177e4c
|
3183 |
|
8c8cc2c10
|
3184 |
return fallback_alloc(cachep, flags); |
e498be7da
|
3185 |
|
a737b3e2f
|
3186 |
done: |
b28a02de8
|
3187 |
return obj; |
e498be7da
|
3188 |
} |
8c8cc2c10
|
3189 3190 3191 3192 3193 3194 3195 3196 3197 3198 3199 3200 3201 3202 3203 3204 3205 3206 3207 |
/** * kmem_cache_alloc_node - Allocate an object on the specified node * @cachep: The cache to allocate from. * @flags: See kmalloc(). * @nodeid: node number of the target node. * @caller: return address of caller, used for debug information * * Identical to kmem_cache_alloc but it will allocate memory on the given * node, which can improve the performance for cpu bound structures. * * Fallback to other node is possible if __GFP_THISNODE is not set. */ static __always_inline void * __cache_alloc_node(struct kmem_cache *cachep, gfp_t flags, int nodeid, void *caller) { unsigned long save_flags; void *ptr; |
773ff60e8
|
3208 |
if (slab_should_failslab(cachep, flags)) |
824ebef12
|
3209 |
return NULL; |
8c8cc2c10
|
3210 3211 3212 3213 3214 3215 3216 3217 3218 3219 3220 3221 3222 3223 3224 3225 3226 3227 3228 3229 3230 3231 3232 3233 3234 3235 3236 3237 |
cache_alloc_debugcheck_before(cachep, flags); local_irq_save(save_flags); if (unlikely(nodeid == -1)) nodeid = numa_node_id(); if (unlikely(!cachep->nodelists[nodeid])) { /* Node not bootstrapped yet */ ptr = fallback_alloc(cachep, flags); goto out; } if (nodeid == numa_node_id()) { /* * Use the locally cached objects if possible. * However ____cache_alloc does not allow fallback * to other nodes. It may fail while we still have * objects on other nodes available. */ ptr = ____cache_alloc(cachep, flags); if (ptr) goto out; } /* ___cache_alloc_node can fall back to other nodes */ ptr = ____cache_alloc_node(cachep, flags, nodeid); out: local_irq_restore(save_flags); ptr = cache_alloc_debugcheck_after(cachep, flags, ptr, caller); |
d07dbea46
|
3238 3239 |
if (unlikely((flags & __GFP_ZERO) && ptr)) memset(ptr, 0, obj_size(cachep)); |
8c8cc2c10
|
3240 3241 3242 3243 3244 3245 3246 3247 3248 3249 3250 3251 3252 3253 3254 3255 3256 3257 3258 3259 3260 3261 3262 3263 3264 3265 3266 3267 3268 3269 3270 3271 3272 3273 3274 3275 3276 3277 3278 3279 |
return ptr; } static __always_inline void * __do_cache_alloc(struct kmem_cache *cache, gfp_t flags) { void *objp; if (unlikely(current->flags & (PF_SPREAD_SLAB | PF_MEMPOLICY))) { objp = alternate_node_alloc(cache, flags); if (objp) goto out; } objp = ____cache_alloc(cache, flags); /* * We may just have run out of memory on the local node. * ____cache_alloc_node() knows how to locate memory on other nodes */ if (!objp) objp = ____cache_alloc_node(cache, flags, numa_node_id()); out: return objp; } #else static __always_inline void * __do_cache_alloc(struct kmem_cache *cachep, gfp_t flags) { return ____cache_alloc(cachep, flags); } #endif /* CONFIG_NUMA */ static __always_inline void * __cache_alloc(struct kmem_cache *cachep, gfp_t flags, void *caller) { unsigned long save_flags; void *objp; |
773ff60e8
|
3280 |
if (slab_should_failslab(cachep, flags)) |
824ebef12
|
3281 |
return NULL; |
8c8cc2c10
|
3282 3283 3284 3285 3286 3287 |
cache_alloc_debugcheck_before(cachep, flags); local_irq_save(save_flags); objp = __do_cache_alloc(cachep, flags); local_irq_restore(save_flags); objp = cache_alloc_debugcheck_after(cachep, flags, objp, caller); prefetchw(objp); |
d07dbea46
|
3288 3289 |
if (unlikely((flags & __GFP_ZERO) && objp)) memset(objp, 0, obj_size(cachep)); |
8c8cc2c10
|
3290 3291 |
return objp; } |
e498be7da
|
3292 3293 3294 3295 |
/* * Caller needs to acquire correct kmem_list's list_lock */ |
343e0d7a9
|
3296 |
static void free_block(struct kmem_cache *cachep, void **objpp, int nr_objects, |
b28a02de8
|
3297 |
int node) |
1da177e4c
|
3298 3299 |
{ int i; |
e498be7da
|
3300 |
struct kmem_list3 *l3; |
1da177e4c
|
3301 3302 3303 3304 |
for (i = 0; i < nr_objects; i++) { void *objp = objpp[i]; struct slab *slabp; |
1da177e4c
|
3305 |
|
6ed5eb221
|
3306 |
slabp = virt_to_slab(objp); |
ff69416e6
|
3307 |
l3 = cachep->nodelists[node]; |
1da177e4c
|
3308 |
list_del(&slabp->list); |
ff69416e6
|
3309 |
check_spinlock_acquired_node(cachep, node); |
1da177e4c
|
3310 |
check_slabp(cachep, slabp); |
78d382d77
|
3311 |
slab_put_obj(cachep, slabp, objp, node); |
1da177e4c
|
3312 |
STATS_DEC_ACTIVE(cachep); |
e498be7da
|
3313 |
l3->free_objects++; |
1da177e4c
|
3314 3315 3316 3317 |
check_slabp(cachep, slabp); /* fixup slab chains */ if (slabp->inuse == 0) { |
e498be7da
|
3318 3319 |
if (l3->free_objects > l3->free_limit) { l3->free_objects -= cachep->num; |
e5ac9c5ae
|
3320 3321 3322 3323 3324 3325 |
/* No need to drop any previously held * lock here, even if we have a off-slab slab * descriptor it is guaranteed to come from * a different cache, refer to comments before * alloc_slabmgmt. */ |
1da177e4c
|
3326 3327 |
slab_destroy(cachep, slabp); } else { |
e498be7da
|
3328 |
list_add(&slabp->list, &l3->slabs_free); |
1da177e4c
|
3329 3330 3331 3332 3333 3334 |
} } else { /* Unconditionally move a slab to the end of the * partial list on free - maximum time for the * other objects to be freed, too. */ |
e498be7da
|
3335 |
list_add_tail(&slabp->list, &l3->slabs_partial); |
1da177e4c
|
3336 3337 3338 |
} } } |
343e0d7a9
|
3339 |
static void cache_flusharray(struct kmem_cache *cachep, struct array_cache *ac) |
1da177e4c
|
3340 3341 |
{ int batchcount; |
e498be7da
|
3342 |
struct kmem_list3 *l3; |
ff69416e6
|
3343 |
int node = numa_node_id(); |
1da177e4c
|
3344 3345 3346 3347 3348 3349 |
batchcount = ac->batchcount; #if DEBUG BUG_ON(!batchcount || batchcount > ac->avail); #endif check_irq_off(); |
ff69416e6
|
3350 |
l3 = cachep->nodelists[node]; |
873623dfa
|
3351 |
spin_lock(&l3->list_lock); |
e498be7da
|
3352 3353 |
if (l3->shared) { struct array_cache *shared_array = l3->shared; |
b28a02de8
|
3354 |
int max = shared_array->limit - shared_array->avail; |
1da177e4c
|
3355 3356 3357 |
if (max) { if (batchcount > max) batchcount = max; |
e498be7da
|
3358 |
memcpy(&(shared_array->entry[shared_array->avail]), |
b28a02de8
|
3359 |
ac->entry, sizeof(void *) * batchcount); |
1da177e4c
|
3360 3361 3362 3363 |
shared_array->avail += batchcount; goto free_done; } } |
ff69416e6
|
3364 |
free_block(cachep, ac->entry, batchcount, node); |
a737b3e2f
|
3365 |
free_done: |
1da177e4c
|
3366 3367 3368 3369 |
#if STATS { int i = 0; struct list_head *p; |
e498be7da
|
3370 3371 |
p = l3->slabs_free.next; while (p != &(l3->slabs_free)) { |
1da177e4c
|
3372 3373 3374 3375 3376 3377 3378 3379 3380 3381 3382 |
struct slab *slabp; slabp = list_entry(p, struct slab, list); BUG_ON(slabp->inuse); i++; p = p->next; } STATS_SET_FREEABLE(cachep, i); } #endif |
e498be7da
|
3383 |
spin_unlock(&l3->list_lock); |
1da177e4c
|
3384 |
ac->avail -= batchcount; |
a737b3e2f
|
3385 |
memmove(ac->entry, &(ac->entry[batchcount]), sizeof(void *)*ac->avail); |
1da177e4c
|
3386 3387 3388 |
} /* |
a737b3e2f
|
3389 3390 |
* Release an obj back to its cache. If the obj has a constructed state, it must * be in this state _before_ it is released. Called with disabled ints. |
1da177e4c
|
3391 |
*/ |
873623dfa
|
3392 |
static inline void __cache_free(struct kmem_cache *cachep, void *objp) |
1da177e4c
|
3393 |
{ |
9a2dba4b4
|
3394 |
struct array_cache *ac = cpu_cache_get(cachep); |
1da177e4c
|
3395 3396 3397 |
check_irq_off(); objp = cache_free_debugcheck(cachep, objp, __builtin_return_address(0)); |
1807a1aaf
|
3398 3399 3400 3401 3402 3403 3404 3405 |
/* * Skip calling cache_free_alien() when the platform is not numa. * This will avoid cache misses that happen while accessing slabp (which * is per page memory reference) to get nodeid. Instead use a global * variable to skip the call, which is mostly likely to be present in * the cache. */ if (numa_platform && cache_free_alien(cachep, objp)) |
729bd0b74
|
3406 |
return; |
1da177e4c
|
3407 3408 |
if (likely(ac->avail < ac->limit)) { STATS_INC_FREEHIT(cachep); |
e498be7da
|
3409 |
ac->entry[ac->avail++] = objp; |
1da177e4c
|
3410 3411 3412 3413 |
return; } else { STATS_INC_FREEMISS(cachep); cache_flusharray(cachep, ac); |
e498be7da
|
3414 |
ac->entry[ac->avail++] = objp; |
1da177e4c
|
3415 3416 3417 3418 3419 3420 3421 3422 3423 3424 3425 |
} } /** * kmem_cache_alloc - Allocate an object * @cachep: The cache to allocate from. * @flags: See kmalloc(). * * Allocate an object from this cache. The flags are only relevant * if the cache has no available objects. */ |
343e0d7a9
|
3426 |
void *kmem_cache_alloc(struct kmem_cache *cachep, gfp_t flags) |
1da177e4c
|
3427 |
{ |
7fd6b1413
|
3428 |
return __cache_alloc(cachep, flags, __builtin_return_address(0)); |
1da177e4c
|
3429 3430 3431 3432 |
} EXPORT_SYMBOL(kmem_cache_alloc); /** |
7682486b3
|
3433 |
* kmem_ptr_validate - check if an untrusted pointer might be a slab entry. |
1da177e4c
|
3434 3435 3436 |
* @cachep: the cache we're checking against * @ptr: pointer to validate * |
7682486b3
|
3437 |
* This verifies that the untrusted pointer looks sane; |
1da177e4c
|
3438 3439 3440 3441 3442 3443 3444 |
* it is _not_ a guarantee that the pointer is actually * part of the slab cache in question, but it at least * validates that the pointer can be dereferenced and * looks half-way sane. * * Currently only used for dentry validation. */ |
b7f869a28
|
3445 |
int kmem_ptr_validate(struct kmem_cache *cachep, const void *ptr) |
1da177e4c
|
3446 |
{ |
b28a02de8
|
3447 |
unsigned long addr = (unsigned long)ptr; |
1da177e4c
|
3448 |
unsigned long min_addr = PAGE_OFFSET; |
b28a02de8
|
3449 |
unsigned long align_mask = BYTES_PER_WORD - 1; |
3dafccf22
|
3450 |
unsigned long size = cachep->buffer_size; |
1da177e4c
|
3451 3452 3453 3454 3455 3456 3457 3458 3459 3460 3461 3462 3463 3464 3465 |
struct page *page; if (unlikely(addr < min_addr)) goto out; if (unlikely(addr > (unsigned long)high_memory - size)) goto out; if (unlikely(addr & align_mask)) goto out; if (unlikely(!kern_addr_valid(addr))) goto out; if (unlikely(!kern_addr_valid(addr + size - 1))) goto out; page = virt_to_page(ptr); if (unlikely(!PageSlab(page))) goto out; |
065d41cb2
|
3466 |
if (unlikely(page_get_cache(page) != cachep)) |
1da177e4c
|
3467 3468 |
goto out; return 1; |
a737b3e2f
|
3469 |
out: |
1da177e4c
|
3470 3471 3472 3473 |
return 0; } #ifdef CONFIG_NUMA |
8b98c1699
|
3474 3475 3476 3477 3478 |
void *kmem_cache_alloc_node(struct kmem_cache *cachep, gfp_t flags, int nodeid) { return __cache_alloc_node(cachep, flags, nodeid, __builtin_return_address(0)); } |
1da177e4c
|
3479 |
EXPORT_SYMBOL(kmem_cache_alloc_node); |
8b98c1699
|
3480 3481 |
static __always_inline void * __do_kmalloc_node(size_t size, gfp_t flags, int node, void *caller) |
97e2bde47
|
3482 |
{ |
343e0d7a9
|
3483 |
struct kmem_cache *cachep; |
97e2bde47
|
3484 3485 |
cachep = kmem_find_general_cachep(size, flags); |
6cb8f9132
|
3486 3487 |
if (unlikely(ZERO_OR_NULL_PTR(cachep))) return cachep; |
97e2bde47
|
3488 3489 |
return kmem_cache_alloc_node(cachep, flags, node); } |
8b98c1699
|
3490 3491 3492 3493 3494 3495 3496 |
#ifdef CONFIG_DEBUG_SLAB void *__kmalloc_node(size_t size, gfp_t flags, int node) { return __do_kmalloc_node(size, flags, node, __builtin_return_address(0)); } |
dbe5e69d2
|
3497 |
EXPORT_SYMBOL(__kmalloc_node); |
8b98c1699
|
3498 3499 |
void *__kmalloc_node_track_caller(size_t size, gfp_t flags, |
ce71e27c6
|
3500 |
int node, unsigned long caller) |
8b98c1699
|
3501 |
{ |
ce71e27c6
|
3502 |
return __do_kmalloc_node(size, flags, node, (void *)caller); |
8b98c1699
|
3503 3504 3505 3506 3507 3508 3509 3510 3511 3512 |
} EXPORT_SYMBOL(__kmalloc_node_track_caller); #else void *__kmalloc_node(size_t size, gfp_t flags, int node) { return __do_kmalloc_node(size, flags, node, NULL); } EXPORT_SYMBOL(__kmalloc_node); #endif /* CONFIG_DEBUG_SLAB */ #endif /* CONFIG_NUMA */ |
1da177e4c
|
3513 3514 |
/** |
800590f52
|
3515 |
* __do_kmalloc - allocate memory |
1da177e4c
|
3516 |
* @size: how many bytes of memory are required. |
800590f52
|
3517 |
* @flags: the type of memory to allocate (see kmalloc). |
911851e6e
|
3518 |
* @caller: function caller for debug tracking of the caller |
1da177e4c
|
3519 |
*/ |
7fd6b1413
|
3520 3521 |
static __always_inline void *__do_kmalloc(size_t size, gfp_t flags, void *caller) |
1da177e4c
|
3522 |
{ |
343e0d7a9
|
3523 |
struct kmem_cache *cachep; |
1da177e4c
|
3524 |
|
97e2bde47
|
3525 3526 3527 3528 3529 3530 |
/* If you want to save a few bytes .text space: replace * __ with kmem_. * Then kmalloc uses the uninlined functions instead of the inline * functions. */ cachep = __find_general_cachep(size, flags); |
a5c96d8a1
|
3531 3532 |
if (unlikely(ZERO_OR_NULL_PTR(cachep))) return cachep; |
7fd6b1413
|
3533 3534 |
return __cache_alloc(cachep, flags, caller); } |
7fd6b1413
|
3535 |
|
1d2c8eea6
|
3536 |
#ifdef CONFIG_DEBUG_SLAB |
7fd6b1413
|
3537 3538 |
void *__kmalloc(size_t size, gfp_t flags) { |
871751e25
|
3539 |
return __do_kmalloc(size, flags, __builtin_return_address(0)); |
1da177e4c
|
3540 3541 |
} EXPORT_SYMBOL(__kmalloc); |
ce71e27c6
|
3542 |
void *__kmalloc_track_caller(size_t size, gfp_t flags, unsigned long caller) |
7fd6b1413
|
3543 |
{ |
ce71e27c6
|
3544 |
return __do_kmalloc(size, flags, (void *)caller); |
7fd6b1413
|
3545 3546 |
} EXPORT_SYMBOL(__kmalloc_track_caller); |
1d2c8eea6
|
3547 3548 3549 3550 3551 3552 3553 |
#else void *__kmalloc(size_t size, gfp_t flags) { return __do_kmalloc(size, flags, NULL); } EXPORT_SYMBOL(__kmalloc); |
7fd6b1413
|
3554 |
#endif |
1da177e4c
|
3555 3556 3557 3558 3559 3560 3561 3562 |
/** * kmem_cache_free - Deallocate an object * @cachep: The cache the allocation was from. * @objp: The previously allocated object. * * Free an object which was previously allocated from this * cache. */ |
343e0d7a9
|
3563 |
void kmem_cache_free(struct kmem_cache *cachep, void *objp) |
1da177e4c
|
3564 3565 3566 3567 |
{ unsigned long flags; local_irq_save(flags); |
898552c9d
|
3568 |
debug_check_no_locks_freed(objp, obj_size(cachep)); |
3ac7fe5a4
|
3569 3570 |
if (!(cachep->flags & SLAB_DEBUG_OBJECTS)) debug_check_no_obj_freed(objp, obj_size(cachep)); |
873623dfa
|
3571 |
__cache_free(cachep, objp); |
1da177e4c
|
3572 3573 3574 3575 3576 |
local_irq_restore(flags); } EXPORT_SYMBOL(kmem_cache_free); /** |
1da177e4c
|
3577 3578 3579 |
* kfree - free previously allocated memory * @objp: pointer returned by kmalloc. * |
80e93effc
|
3580 3581 |
* If @objp is NULL, no operation is performed. * |
1da177e4c
|
3582 3583 3584 3585 3586 |
* Don't free memory not originally allocated by kmalloc() * or you will run into trouble. */ void kfree(const void *objp) { |
343e0d7a9
|
3587 |
struct kmem_cache *c; |
1da177e4c
|
3588 |
unsigned long flags; |
6cb8f9132
|
3589 |
if (unlikely(ZERO_OR_NULL_PTR(objp))) |
1da177e4c
|
3590 3591 3592 |
return; local_irq_save(flags); kfree_debugcheck(objp); |
6ed5eb221
|
3593 |
c = virt_to_cache(objp); |
f9b8404cf
|
3594 |
debug_check_no_locks_freed(objp, obj_size(c)); |
3ac7fe5a4
|
3595 |
debug_check_no_obj_freed(objp, obj_size(c)); |
873623dfa
|
3596 |
__cache_free(c, (void *)objp); |
1da177e4c
|
3597 3598 3599 |
local_irq_restore(flags); } EXPORT_SYMBOL(kfree); |
343e0d7a9
|
3600 |
unsigned int kmem_cache_size(struct kmem_cache *cachep) |
1da177e4c
|
3601 |
{ |
3dafccf22
|
3602 |
return obj_size(cachep); |
1da177e4c
|
3603 3604 |
} EXPORT_SYMBOL(kmem_cache_size); |
343e0d7a9
|
3605 |
const char *kmem_cache_name(struct kmem_cache *cachep) |
1944972d3
|
3606 3607 3608 3609 |
{ return cachep->name; } EXPORT_SYMBOL_GPL(kmem_cache_name); |
e498be7da
|
3610 |
/* |
183ff22bb
|
3611 |
* This initializes kmem_list3 or resizes various caches for all nodes. |
e498be7da
|
3612 |
*/ |
343e0d7a9
|
3613 |
static int alloc_kmemlist(struct kmem_cache *cachep) |
e498be7da
|
3614 3615 3616 |
{ int node; struct kmem_list3 *l3; |
cafeb02e0
|
3617 |
struct array_cache *new_shared; |
3395ee058
|
3618 |
struct array_cache **new_alien = NULL; |
e498be7da
|
3619 |
|
9c09a95cf
|
3620 |
for_each_online_node(node) { |
cafeb02e0
|
3621 |
|
3395ee058
|
3622 3623 3624 3625 3626 |
if (use_alien_caches) { new_alien = alloc_alien_cache(node, cachep->limit); if (!new_alien) goto fail; } |
cafeb02e0
|
3627 |
|
631098469
|
3628 3629 3630 |
new_shared = NULL; if (cachep->shared) { new_shared = alloc_arraycache(node, |
0718dc2a8
|
3631 |
cachep->shared*cachep->batchcount, |
a737b3e2f
|
3632 |
0xbaadf00d); |
631098469
|
3633 3634 3635 3636 |
if (!new_shared) { free_alien_cache(new_alien); goto fail; } |
0718dc2a8
|
3637 |
} |
cafeb02e0
|
3638 |
|
a737b3e2f
|
3639 3640 |
l3 = cachep->nodelists[node]; if (l3) { |
cafeb02e0
|
3641 |
struct array_cache *shared = l3->shared; |
e498be7da
|
3642 |
spin_lock_irq(&l3->list_lock); |
cafeb02e0
|
3643 |
if (shared) |
0718dc2a8
|
3644 3645 |
free_block(cachep, shared->entry, shared->avail, node); |
e498be7da
|
3646 |
|
cafeb02e0
|
3647 3648 |
l3->shared = new_shared; if (!l3->alien) { |
e498be7da
|
3649 3650 3651 |
l3->alien = new_alien; new_alien = NULL; } |
b28a02de8
|
3652 |
l3->free_limit = (1 + nr_cpus_node(node)) * |
a737b3e2f
|
3653 |
cachep->batchcount + cachep->num; |
e498be7da
|
3654 |
spin_unlock_irq(&l3->list_lock); |
cafeb02e0
|
3655 |
kfree(shared); |
e498be7da
|
3656 3657 3658 |
free_alien_cache(new_alien); continue; } |
a737b3e2f
|
3659 |
l3 = kmalloc_node(sizeof(struct kmem_list3), GFP_KERNEL, node); |
0718dc2a8
|
3660 3661 3662 |
if (!l3) { free_alien_cache(new_alien); kfree(new_shared); |
e498be7da
|
3663 |
goto fail; |
0718dc2a8
|
3664 |
} |
e498be7da
|
3665 3666 3667 |
kmem_list3_init(l3); l3->next_reap = jiffies + REAPTIMEOUT_LIST3 + |
a737b3e2f
|
3668 |
((unsigned long)cachep) % REAPTIMEOUT_LIST3; |
cafeb02e0
|
3669 |
l3->shared = new_shared; |
e498be7da
|
3670 |
l3->alien = new_alien; |
b28a02de8
|
3671 |
l3->free_limit = (1 + nr_cpus_node(node)) * |
a737b3e2f
|
3672 |
cachep->batchcount + cachep->num; |
e498be7da
|
3673 3674 |
cachep->nodelists[node] = l3; } |
cafeb02e0
|
3675 |
return 0; |
0718dc2a8
|
3676 |
|
a737b3e2f
|
3677 |
fail: |
0718dc2a8
|
3678 3679 3680 3681 3682 3683 3684 3685 3686 3687 3688 3689 3690 3691 3692 |
if (!cachep->next.next) { /* Cache is not active yet. Roll back what we did */ node--; while (node >= 0) { if (cachep->nodelists[node]) { l3 = cachep->nodelists[node]; kfree(l3->shared); free_alien_cache(l3->alien); kfree(l3); cachep->nodelists[node] = NULL; } node--; } } |
cafeb02e0
|
3693 |
return -ENOMEM; |
e498be7da
|
3694 |
} |
1da177e4c
|
3695 |
struct ccupdate_struct { |
343e0d7a9
|
3696 |
struct kmem_cache *cachep; |
1da177e4c
|
3697 3698 3699 3700 3701 |
struct array_cache *new[NR_CPUS]; }; static void do_ccupdate_local(void *info) { |
a737b3e2f
|
3702 |
struct ccupdate_struct *new = info; |
1da177e4c
|
3703 3704 3705 |
struct array_cache *old; check_irq_off(); |
9a2dba4b4
|
3706 |
old = cpu_cache_get(new->cachep); |
e498be7da
|
3707 |
|
1da177e4c
|
3708 3709 3710 |
new->cachep->array[smp_processor_id()] = new->new[smp_processor_id()]; new->new[smp_processor_id()] = old; } |
b5d8ca7c5
|
3711 |
/* Always called with the cache_chain_mutex held */ |
a737b3e2f
|
3712 3713 |
static int do_tune_cpucache(struct kmem_cache *cachep, int limit, int batchcount, int shared) |
1da177e4c
|
3714 |
{ |
d2e7b7d0a
|
3715 |
struct ccupdate_struct *new; |
2ed3a4ef9
|
3716 |
int i; |
1da177e4c
|
3717 |
|
d2e7b7d0a
|
3718 3719 3720 |
new = kzalloc(sizeof(*new), GFP_KERNEL); if (!new) return -ENOMEM; |
e498be7da
|
3721 |
for_each_online_cpu(i) { |
d2e7b7d0a
|
3722 |
new->new[i] = alloc_arraycache(cpu_to_node(i), limit, |
a737b3e2f
|
3723 |
batchcount); |
d2e7b7d0a
|
3724 |
if (!new->new[i]) { |
b28a02de8
|
3725 |
for (i--; i >= 0; i--) |
d2e7b7d0a
|
3726 3727 |
kfree(new->new[i]); kfree(new); |
e498be7da
|
3728 |
return -ENOMEM; |
1da177e4c
|
3729 3730 |
} } |
d2e7b7d0a
|
3731 |
new->cachep = cachep; |
1da177e4c
|
3732 |
|
15c8b6c1a
|
3733 |
on_each_cpu(do_ccupdate_local, (void *)new, 1); |
e498be7da
|
3734 |
|
1da177e4c
|
3735 |
check_irq_on(); |
1da177e4c
|
3736 3737 |
cachep->batchcount = batchcount; cachep->limit = limit; |
e498be7da
|
3738 |
cachep->shared = shared; |
1da177e4c
|
3739 |
|
e498be7da
|
3740 |
for_each_online_cpu(i) { |
d2e7b7d0a
|
3741 |
struct array_cache *ccold = new->new[i]; |
1da177e4c
|
3742 3743 |
if (!ccold) continue; |
e498be7da
|
3744 |
spin_lock_irq(&cachep->nodelists[cpu_to_node(i)]->list_lock); |
ff69416e6
|
3745 |
free_block(cachep, ccold->entry, ccold->avail, cpu_to_node(i)); |
e498be7da
|
3746 |
spin_unlock_irq(&cachep->nodelists[cpu_to_node(i)]->list_lock); |
1da177e4c
|
3747 3748 |
kfree(ccold); } |
d2e7b7d0a
|
3749 |
kfree(new); |
2ed3a4ef9
|
3750 |
return alloc_kmemlist(cachep); |
1da177e4c
|
3751 |
} |
b5d8ca7c5
|
3752 |
/* Called with cache_chain_mutex held always */ |
2ed3a4ef9
|
3753 |
static int enable_cpucache(struct kmem_cache *cachep) |
1da177e4c
|
3754 3755 3756 |
{ int err; int limit, shared; |
a737b3e2f
|
3757 3758 |
/* * The head array serves three purposes: |
1da177e4c
|
3759 3760 |
* - create a LIFO ordering, i.e. return objects that are cache-warm * - reduce the number of spinlock operations. |
a737b3e2f
|
3761 |
* - reduce the number of linked list operations on the slab and |
1da177e4c
|
3762 3763 3764 3765 |
* bufctl chains: array operations are cheaper. * The numbers are guessed, we should auto-tune as described by * Bonwick. */ |
3dafccf22
|
3766 |
if (cachep->buffer_size > 131072) |
1da177e4c
|
3767 |
limit = 1; |
3dafccf22
|
3768 |
else if (cachep->buffer_size > PAGE_SIZE) |
1da177e4c
|
3769 |
limit = 8; |
3dafccf22
|
3770 |
else if (cachep->buffer_size > 1024) |
1da177e4c
|
3771 |
limit = 24; |
3dafccf22
|
3772 |
else if (cachep->buffer_size > 256) |
1da177e4c
|
3773 3774 3775 |
limit = 54; else limit = 120; |
a737b3e2f
|
3776 3777 |
/* * CPU bound tasks (e.g. network routing) can exhibit cpu bound |
1da177e4c
|
3778 3779 3780 3781 3782 3783 3784 3785 |
* allocation behaviour: Most allocs on one cpu, most free operations * on another cpu. For these cases, an efficient object passing between * cpus is necessary. This is provided by a shared array. The array * replaces Bonwick's magazine layer. * On uniprocessor, it's functionally equivalent (but less efficient) * to a larger limit. Thus disabled by default. */ shared = 0; |
364fbb29a
|
3786 |
if (cachep->buffer_size <= PAGE_SIZE && num_possible_cpus() > 1) |
1da177e4c
|
3787 |
shared = 8; |
1da177e4c
|
3788 3789 |
#if DEBUG |
a737b3e2f
|
3790 3791 3792 |
/* * With debugging enabled, large batchcount lead to excessively long * periods with disabled local interrupts. Limit the batchcount |
1da177e4c
|
3793 3794 3795 3796 |
*/ if (limit > 32) limit = 32; #endif |
b28a02de8
|
3797 |
err = do_tune_cpucache(cachep, limit, (limit + 1) / 2, shared); |
1da177e4c
|
3798 3799 3800 |
if (err) printk(KERN_ERR "enable_cpucache failed for %s, error %d. ", |
b28a02de8
|
3801 |
cachep->name, -err); |
2ed3a4ef9
|
3802 |
return err; |
1da177e4c
|
3803 |
} |
1b55253a7
|
3804 3805 |
/* * Drain an array if it contains any elements taking the l3 lock only if |
b18e7e654
|
3806 3807 |
* necessary. Note that the l3 listlock also protects the array_cache * if drain_array() is used on the shared array. |
1b55253a7
|
3808 3809 3810 |
*/ void drain_array(struct kmem_cache *cachep, struct kmem_list3 *l3, struct array_cache *ac, int force, int node) |
1da177e4c
|
3811 3812 |
{ int tofree; |
1b55253a7
|
3813 3814 |
if (!ac || !ac->avail) return; |
1da177e4c
|
3815 3816 |
if (ac->touched && !force) { ac->touched = 0; |
b18e7e654
|
3817 |
} else { |
1b55253a7
|
3818 |
spin_lock_irq(&l3->list_lock); |
b18e7e654
|
3819 3820 3821 3822 3823 3824 3825 3826 3827 |
if (ac->avail) { tofree = force ? ac->avail : (ac->limit + 4) / 5; if (tofree > ac->avail) tofree = (ac->avail + 1) / 2; free_block(cachep, ac->entry, tofree, node); ac->avail -= tofree; memmove(ac->entry, &(ac->entry[tofree]), sizeof(void *) * ac->avail); } |
1b55253a7
|
3828 |
spin_unlock_irq(&l3->list_lock); |
1da177e4c
|
3829 3830 3831 3832 3833 |
} } /** * cache_reap - Reclaim memory from caches. |
05fb6bf0b
|
3834 |
* @w: work descriptor |
1da177e4c
|
3835 3836 3837 3838 3839 3840 |
* * Called from workqueue/eventd every few seconds. * Purpose: * - clear the per-cpu caches for this CPU. * - return freeable pages to the main free memory pool. * |
a737b3e2f
|
3841 3842 |
* If we cannot acquire the cache chain mutex then just give up - we'll try * again on the next iteration. |
1da177e4c
|
3843 |
*/ |
7c5cae368
|
3844 |
static void cache_reap(struct work_struct *w) |
1da177e4c
|
3845 |
{ |
7a7c381d2
|
3846 |
struct kmem_cache *searchp; |
e498be7da
|
3847 |
struct kmem_list3 *l3; |
aab2207cf
|
3848 |
int node = numa_node_id(); |
7c5cae368
|
3849 3850 |
struct delayed_work *work = container_of(w, struct delayed_work, work); |
1da177e4c
|
3851 |
|
7c5cae368
|
3852 |
if (!mutex_trylock(&cache_chain_mutex)) |
1da177e4c
|
3853 |
/* Give up. Setup the next iteration. */ |
7c5cae368
|
3854 |
goto out; |
1da177e4c
|
3855 |
|
7a7c381d2
|
3856 |
list_for_each_entry(searchp, &cache_chain, next) { |
1da177e4c
|
3857 |
check_irq_on(); |
35386e3b0
|
3858 3859 3860 3861 3862 |
/* * We only take the l3 lock if absolutely necessary and we * have established with reasonable certainty that * we can do some work if the lock was obtained. */ |
aab2207cf
|
3863 |
l3 = searchp->nodelists[node]; |
35386e3b0
|
3864 |
|
8fce4d8e3
|
3865 |
reap_alien(searchp, l3); |
1da177e4c
|
3866 |
|
aab2207cf
|
3867 |
drain_array(searchp, l3, cpu_cache_get(searchp), 0, node); |
1da177e4c
|
3868 |
|
35386e3b0
|
3869 3870 3871 3872 |
/* * These are racy checks but it does not matter * if we skip one check or scan twice. */ |
e498be7da
|
3873 |
if (time_after(l3->next_reap, jiffies)) |
35386e3b0
|
3874 |
goto next; |
1da177e4c
|
3875 |
|
e498be7da
|
3876 |
l3->next_reap = jiffies + REAPTIMEOUT_LIST3; |
1da177e4c
|
3877 |
|
aab2207cf
|
3878 |
drain_array(searchp, l3, l3->shared, 0, node); |
1da177e4c
|
3879 |
|
ed11d9eb2
|
3880 |
if (l3->free_touched) |
e498be7da
|
3881 |
l3->free_touched = 0; |
ed11d9eb2
|
3882 3883 |
else { int freed; |
1da177e4c
|
3884 |
|
ed11d9eb2
|
3885 3886 3887 3888 |
freed = drain_freelist(searchp, l3, (l3->free_limit + 5 * searchp->num - 1) / (5 * searchp->num)); STATS_ADD_REAPED(searchp, freed); } |
35386e3b0
|
3889 |
next: |
1da177e4c
|
3890 3891 3892 |
cond_resched(); } check_irq_on(); |
fc0abb145
|
3893 |
mutex_unlock(&cache_chain_mutex); |
8fce4d8e3
|
3894 |
next_reap_node(); |
7c5cae368
|
3895 |
out: |
a737b3e2f
|
3896 |
/* Set up the next iteration */ |
7c5cae368
|
3897 |
schedule_delayed_work(work, round_jiffies_relative(REAPTIMEOUT_CPUC)); |
1da177e4c
|
3898 |
} |
158a96242
|
3899 |
#ifdef CONFIG_SLABINFO |
1da177e4c
|
3900 |
|
85289f98d
|
3901 |
static void print_slabinfo_header(struct seq_file *m) |
1da177e4c
|
3902 |
{ |
85289f98d
|
3903 3904 3905 3906 |
/* * Output format version, so at least we can change it * without _too_ many complaints. */ |
1da177e4c
|
3907 |
#if STATS |
85289f98d
|
3908 3909 |
seq_puts(m, "slabinfo - version: 2.1 (statistics) "); |
1da177e4c
|
3910 |
#else |
85289f98d
|
3911 3912 |
seq_puts(m, "slabinfo - version: 2.1 "); |
1da177e4c
|
3913 |
#endif |
85289f98d
|
3914 3915 3916 3917 |
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>"); |
1da177e4c
|
3918 |
#if STATS |
85289f98d
|
3919 |
seq_puts(m, " : globalstat <listallocs> <maxobjs> <grown> <reaped> " |
fb7faf331
|
3920 |
"<error> <maxfreeable> <nodeallocs> <remotefrees> <alienoverflow>"); |
85289f98d
|
3921 |
seq_puts(m, " : cpustat <allochit> <allocmiss> <freehit> <freemiss>"); |
1da177e4c
|
3922 |
#endif |
85289f98d
|
3923 3924 3925 3926 3927 3928 3929 |
seq_putc(m, ' '); } static void *s_start(struct seq_file *m, loff_t *pos) { loff_t n = *pos; |
85289f98d
|
3930 |
|
fc0abb145
|
3931 |
mutex_lock(&cache_chain_mutex); |
85289f98d
|
3932 3933 |
if (!n) print_slabinfo_header(m); |
b92151bab
|
3934 3935 |
return seq_list_start(&cache_chain, *pos); |
1da177e4c
|
3936 3937 3938 3939 |
} static void *s_next(struct seq_file *m, void *p, loff_t *pos) { |
b92151bab
|
3940 |
return seq_list_next(p, &cache_chain, pos); |
1da177e4c
|
3941 3942 3943 3944 |
} static void s_stop(struct seq_file *m, void *p) { |
fc0abb145
|
3945 |
mutex_unlock(&cache_chain_mutex); |
1da177e4c
|
3946 3947 3948 3949 |
} static int s_show(struct seq_file *m, void *p) { |
b92151bab
|
3950 |
struct kmem_cache *cachep = list_entry(p, struct kmem_cache, next); |
b28a02de8
|
3951 3952 3953 3954 3955 |
struct slab *slabp; unsigned long active_objs; unsigned long num_objs; unsigned long active_slabs = 0; unsigned long num_slabs, free_objects = 0, shared_avail = 0; |
e498be7da
|
3956 |
const char *name; |
1da177e4c
|
3957 |
char *error = NULL; |
e498be7da
|
3958 3959 |
int node; struct kmem_list3 *l3; |
1da177e4c
|
3960 |
|
1da177e4c
|
3961 3962 |
active_objs = 0; num_slabs = 0; |
e498be7da
|
3963 3964 3965 3966 |
for_each_online_node(node) { l3 = cachep->nodelists[node]; if (!l3) continue; |
ca3b9b917
|
3967 3968 |
check_irq_on(); spin_lock_irq(&l3->list_lock); |
e498be7da
|
3969 |
|
7a7c381d2
|
3970 |
list_for_each_entry(slabp, &l3->slabs_full, list) { |
e498be7da
|
3971 3972 3973 3974 3975 |
if (slabp->inuse != cachep->num && !error) error = "slabs_full accounting error"; active_objs += cachep->num; active_slabs++; } |
7a7c381d2
|
3976 |
list_for_each_entry(slabp, &l3->slabs_partial, list) { |
e498be7da
|
3977 3978 3979 3980 3981 3982 3983 |
if (slabp->inuse == cachep->num && !error) error = "slabs_partial inuse accounting error"; if (!slabp->inuse && !error) error = "slabs_partial/inuse accounting error"; active_objs += slabp->inuse; active_slabs++; } |
7a7c381d2
|
3984 |
list_for_each_entry(slabp, &l3->slabs_free, list) { |
e498be7da
|
3985 3986 3987 3988 3989 |
if (slabp->inuse && !error) error = "slabs_free/inuse accounting error"; num_slabs++; } free_objects += l3->free_objects; |
4484ebf12
|
3990 3991 |
if (l3->shared) shared_avail += l3->shared->avail; |
e498be7da
|
3992 |
|
ca3b9b917
|
3993 |
spin_unlock_irq(&l3->list_lock); |
1da177e4c
|
3994 |
} |
b28a02de8
|
3995 3996 |
num_slabs += active_slabs; num_objs = num_slabs * cachep->num; |
e498be7da
|
3997 |
if (num_objs - active_objs != free_objects && !error) |
1da177e4c
|
3998 |
error = "free_objects accounting error"; |
b28a02de8
|
3999 |
name = cachep->name; |
1da177e4c
|
4000 4001 4002 4003 4004 |
if (error) printk(KERN_ERR "slab: cache %s error: %s ", name, error); seq_printf(m, "%-17s %6lu %6lu %6u %4u %4d", |
3dafccf22
|
4005 |
name, active_objs, num_objs, cachep->buffer_size, |
b28a02de8
|
4006 |
cachep->num, (1 << cachep->gfporder)); |
1da177e4c
|
4007 |
seq_printf(m, " : tunables %4u %4u %4u", |
b28a02de8
|
4008 |
cachep->limit, cachep->batchcount, cachep->shared); |
e498be7da
|
4009 |
seq_printf(m, " : slabdata %6lu %6lu %6lu", |
b28a02de8
|
4010 |
active_slabs, num_slabs, shared_avail); |
1da177e4c
|
4011 |
#if STATS |
b28a02de8
|
4012 |
{ /* list3 stats */ |
1da177e4c
|
4013 4014 4015 4016 4017 4018 |
unsigned long high = cachep->high_mark; unsigned long allocs = cachep->num_allocations; unsigned long grown = cachep->grown; unsigned long reaped = cachep->reaped; unsigned long errors = cachep->errors; unsigned long max_freeable = cachep->max_freeable; |
1da177e4c
|
4019 |
unsigned long node_allocs = cachep->node_allocs; |
e498be7da
|
4020 |
unsigned long node_frees = cachep->node_frees; |
fb7faf331
|
4021 |
unsigned long overflows = cachep->node_overflow; |
1da177e4c
|
4022 |
|
e498be7da
|
4023 |
seq_printf(m, " : globalstat %7lu %6lu %5lu %4lu \ |
fb7faf331
|
4024 |
%4lu %4lu %4lu %4lu %4lu", allocs, high, grown, |
a737b3e2f
|
4025 |
reaped, errors, max_freeable, node_allocs, |
fb7faf331
|
4026 |
node_frees, overflows); |
1da177e4c
|
4027 4028 4029 4030 4031 4032 4033 4034 4035 |
} /* cpu stats */ { unsigned long allochit = atomic_read(&cachep->allochit); unsigned long allocmiss = atomic_read(&cachep->allocmiss); unsigned long freehit = atomic_read(&cachep->freehit); unsigned long freemiss = atomic_read(&cachep->freemiss); seq_printf(m, " : cpustat %6lu %6lu %6lu %6lu", |
b28a02de8
|
4036 |
allochit, allocmiss, freehit, freemiss); |
1da177e4c
|
4037 4038 4039 4040 |
} #endif seq_putc(m, ' '); |
1da177e4c
|
4041 4042 4043 4044 4045 4046 4047 4048 4049 4050 4051 4052 4053 4054 4055 4056 |
return 0; } /* * slabinfo_op - iterator that generates /proc/slabinfo * * Output layout: * cache-name * num-active-objs * total-objs * object size * num-active-slabs * total-slabs * num-pages-per-slab * + further values on SMP and with statistics enabled */ |
7b3c3a50a
|
4057 |
static const struct seq_operations slabinfo_op = { |
b28a02de8
|
4058 4059 4060 4061 |
.start = s_start, .next = s_next, .stop = s_stop, .show = s_show, |
1da177e4c
|
4062 4063 4064 4065 4066 4067 4068 4069 4070 4071 |
}; #define MAX_SLABINFO_WRITE 128 /** * slabinfo_write - Tuning for the slab allocator * @file: unused * @buffer: user buffer * @count: data length * @ppos: unused */ |
b28a02de8
|
4072 4073 |
ssize_t slabinfo_write(struct file *file, const char __user * buffer, size_t count, loff_t *ppos) |
1da177e4c
|
4074 |
{ |
b28a02de8
|
4075 |
char kbuf[MAX_SLABINFO_WRITE + 1], *tmp; |
1da177e4c
|
4076 |
int limit, batchcount, shared, res; |
7a7c381d2
|
4077 |
struct kmem_cache *cachep; |
b28a02de8
|
4078 |
|
1da177e4c
|
4079 4080 4081 4082 |
if (count > MAX_SLABINFO_WRITE) return -EINVAL; if (copy_from_user(&kbuf, buffer, count)) return -EFAULT; |
b28a02de8
|
4083 |
kbuf[MAX_SLABINFO_WRITE] = '\0'; |
1da177e4c
|
4084 4085 4086 4087 4088 4089 4090 4091 4092 4093 |
tmp = strchr(kbuf, ' '); if (!tmp) return -EINVAL; *tmp = '\0'; tmp++; if (sscanf(tmp, " %d %d %d", &limit, &batchcount, &shared) != 3) return -EINVAL; /* Find the cache in the chain of caches. */ |
fc0abb145
|
4094 |
mutex_lock(&cache_chain_mutex); |
1da177e4c
|
4095 |
res = -EINVAL; |
7a7c381d2
|
4096 |
list_for_each_entry(cachep, &cache_chain, next) { |
1da177e4c
|
4097 |
if (!strcmp(cachep->name, kbuf)) { |
a737b3e2f
|
4098 4099 |
if (limit < 1 || batchcount < 1 || batchcount > limit || shared < 0) { |
e498be7da
|
4100 |
res = 0; |
1da177e4c
|
4101 |
} else { |
e498be7da
|
4102 |
res = do_tune_cpucache(cachep, limit, |
b28a02de8
|
4103 |
batchcount, shared); |
1da177e4c
|
4104 4105 4106 4107 |
} break; } } |
fc0abb145
|
4108 |
mutex_unlock(&cache_chain_mutex); |
1da177e4c
|
4109 4110 4111 4112 |
if (res >= 0) res = count; return res; } |
871751e25
|
4113 |
|
7b3c3a50a
|
4114 4115 4116 4117 4118 4119 4120 4121 4122 4123 4124 4125 |
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, .write = slabinfo_write, .llseek = seq_lseek, .release = seq_release, }; |
871751e25
|
4126 4127 4128 4129 |
#ifdef CONFIG_DEBUG_SLAB_LEAK static void *leaks_start(struct seq_file *m, loff_t *pos) { |
871751e25
|
4130 |
mutex_lock(&cache_chain_mutex); |
b92151bab
|
4131 |
return seq_list_start(&cache_chain, *pos); |
871751e25
|
4132 4133 4134 4135 4136 4137 4138 4139 4140 4141 4142 4143 4144 4145 4146 4147 4148 4149 4150 4151 4152 4153 4154 4155 4156 4157 4158 4159 4160 4161 4162 4163 4164 4165 4166 4167 4168 4169 4170 4171 4172 4173 4174 4175 4176 4177 4178 4179 4180 |
} static inline int add_caller(unsigned long *n, unsigned long v) { unsigned long *p; int l; if (!v) return 1; l = n[1]; p = n + 2; while (l) { int i = l/2; unsigned long *q = p + 2 * i; if (*q == v) { q[1]++; return 1; } if (*q > v) { l = i; } else { p = q + 2; l -= i + 1; } } if (++n[1] == n[0]) return 0; memmove(p + 2, p, n[1] * 2 * sizeof(unsigned long) - ((void *)p - (void *)n)); p[0] = v; p[1] = 1; return 1; } static void handle_slab(unsigned long *n, struct kmem_cache *c, struct slab *s) { void *p; int i; if (n[0] == n[1]) return; for (i = 0, p = s->s_mem; i < c->num; i++, p += c->buffer_size) { if (slab_bufctl(s)[i] != BUFCTL_ACTIVE) continue; if (!add_caller(n, (unsigned long)*dbg_userword(c, p))) return; } } static void show_symbol(struct seq_file *m, unsigned long address) { #ifdef CONFIG_KALLSYMS |
871751e25
|
4181 |
unsigned long offset, size; |
9281acea6
|
4182 |
char modname[MODULE_NAME_LEN], name[KSYM_NAME_LEN]; |
871751e25
|
4183 |
|
a5c43dae7
|
4184 |
if (lookup_symbol_attrs(address, &size, &offset, modname, name) == 0) { |
871751e25
|
4185 |
seq_printf(m, "%s+%#lx/%#lx", name, offset, size); |
a5c43dae7
|
4186 |
if (modname[0]) |
871751e25
|
4187 4188 4189 4190 4191 4192 4193 4194 4195 |
seq_printf(m, " [%s]", modname); return; } #endif seq_printf(m, "%p", (void *)address); } static int leaks_show(struct seq_file *m, void *p) { |
b92151bab
|
4196 |
struct kmem_cache *cachep = list_entry(p, struct kmem_cache, next); |
871751e25
|
4197 4198 4199 4200 4201 4202 4203 4204 4205 4206 4207 4208 4209 4210 4211 4212 4213 4214 4215 4216 4217 4218 4219 |
struct slab *slabp; struct kmem_list3 *l3; const char *name; unsigned long *n = m->private; int node; int i; if (!(cachep->flags & SLAB_STORE_USER)) return 0; if (!(cachep->flags & SLAB_RED_ZONE)) return 0; /* OK, we can do it */ n[1] = 0; for_each_online_node(node) { l3 = cachep->nodelists[node]; if (!l3) continue; check_irq_on(); spin_lock_irq(&l3->list_lock); |
7a7c381d2
|
4220 |
list_for_each_entry(slabp, &l3->slabs_full, list) |
871751e25
|
4221 |
handle_slab(n, cachep, slabp); |
7a7c381d2
|
4222 |
list_for_each_entry(slabp, &l3->slabs_partial, list) |
871751e25
|
4223 |
handle_slab(n, cachep, slabp); |
871751e25
|
4224 4225 4226 4227 4228 4229 4230 4231 4232 4233 4234 4235 4236 4237 4238 4239 4240 4241 4242 4243 4244 4245 4246 4247 4248 4249 |
spin_unlock_irq(&l3->list_lock); } name = cachep->name; if (n[0] == n[1]) { /* Increase the buffer size */ mutex_unlock(&cache_chain_mutex); m->private = kzalloc(n[0] * 4 * sizeof(unsigned long), GFP_KERNEL); if (!m->private) { /* Too bad, we are really out */ m->private = n; mutex_lock(&cache_chain_mutex); return -ENOMEM; } *(unsigned long *)m->private = n[0] * 2; kfree(n); mutex_lock(&cache_chain_mutex); /* Now make sure this entry will be retried */ m->count = m->size; return 0; } for (i = 0; i < n[1]; i++) { seq_printf(m, "%s: %lu ", name, n[2*i+3]); show_symbol(m, n[2*i+2]); seq_putc(m, ' '); } |
d2e7b7d0a
|
4250 |
|
871751e25
|
4251 4252 |
return 0; } |
a0ec95a8e
|
4253 |
static const struct seq_operations slabstats_op = { |
871751e25
|
4254 4255 4256 4257 4258 |
.start = leaks_start, .next = s_next, .stop = s_stop, .show = leaks_show, }; |
a0ec95a8e
|
4259 4260 4261 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 |
static int slabstats_open(struct inode *inode, struct file *file) { unsigned long *n = kzalloc(PAGE_SIZE, GFP_KERNEL); int ret = -ENOMEM; if (n) { ret = seq_open(file, &slabstats_op); if (!ret) { struct seq_file *m = file->private_data; *n = PAGE_SIZE / (2 * sizeof(unsigned long)); m->private = n; n = NULL; } kfree(n); } return ret; } static const struct file_operations proc_slabstats_operations = { .open = slabstats_open, .read = seq_read, .llseek = seq_lseek, .release = seq_release_private, }; #endif static int __init slab_proc_init(void) { |
7b3c3a50a
|
4287 |
proc_create("slabinfo",S_IWUSR|S_IRUGO,NULL,&proc_slabinfo_operations); |
a0ec95a8e
|
4288 4289 |
#ifdef CONFIG_DEBUG_SLAB_LEAK proc_create("slab_allocators", 0, NULL, &proc_slabstats_operations); |
871751e25
|
4290 |
#endif |
a0ec95a8e
|
4291 4292 4293 |
return 0; } module_init(slab_proc_init); |
1da177e4c
|
4294 |
#endif |
00e145b6d
|
4295 4296 4297 4298 4299 4300 4301 4302 4303 4304 4305 4306 |
/** * ksize - get the actual amount of memory allocated for a given object * @objp: Pointer to the object * * kmalloc may internally round up allocations and return more memory * than requested. ksize() can be used to determine the actual amount of * memory allocated. The caller may use this additional memory, even though * a smaller amount of memory was initially specified with the kmalloc call. * The caller must guarantee that objp points to a valid object previously * allocated with either kmalloc() or kmem_cache_alloc(). The object * must not be freed during the duration of the call. */ |
fd76bab2f
|
4307 |
size_t ksize(const void *objp) |
1da177e4c
|
4308 |
{ |
ef8b4520b
|
4309 4310 |
BUG_ON(!objp); if (unlikely(objp == ZERO_SIZE_PTR)) |
00e145b6d
|
4311 |
return 0; |
1da177e4c
|
4312 |
|
6ed5eb221
|
4313 |
return obj_size(virt_to_cache(objp)); |
1da177e4c
|
4314 |
} |
b1aabecd5
|
4315 |
EXPORT_SYMBOL(ksize); |