Blame view
mm/slab.c
119 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> |
d5cff6352
|
108 |
#include <linux/kmemleak.h> |
dc85da15d
|
109 |
#include <linux/mempolicy.h> |
fc0abb145
|
110 |
#include <linux/mutex.h> |
8a8b6502f
|
111 |
#include <linux/fault-inject.h> |
e7eebaf6a
|
112 |
#include <linux/rtmutex.h> |
6a2d7a955
|
113 |
#include <linux/reciprocal_div.h> |
3ac7fe5a4
|
114 |
#include <linux/debugobjects.h> |
c175eea46
|
115 |
#include <linux/kmemcheck.h> |
8f9f8d9e8
|
116 |
#include <linux/memory.h> |
268bb0ce3
|
117 |
#include <linux/prefetch.h> |
1da177e4c
|
118 |
|
1da177e4c
|
119 120 121 122 123 |
#include <asm/cacheflush.h> #include <asm/tlbflush.h> #include <asm/page.h> /* |
50953fe9e
|
124 |
* DEBUG - 1 for kmem_cache_create() to honour; SLAB_RED_ZONE & SLAB_POISON. |
1da177e4c
|
125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 |
* 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
|
142 143 |
/* Shouldn't this be in a header file somewhere? */ #define BYTES_PER_WORD sizeof(void *) |
87a927c71
|
144 |
#define REDZONE_ALIGN max(BYTES_PER_WORD, __alignof__(unsigned long long)) |
1da177e4c
|
145 |
|
1da177e4c
|
146 147 148 149 150 151 |
#ifndef ARCH_KMALLOC_FLAGS #define ARCH_KMALLOC_FLAGS SLAB_HWCACHE_ALIGN #endif /* Legal flag mask for kmem_cache_create(). */ #if DEBUG |
50953fe9e
|
152 |
# define CREATE_MASK (SLAB_RED_ZONE | \ |
1da177e4c
|
153 |
SLAB_POISON | SLAB_HWCACHE_ALIGN | \ |
ac2b898ca
|
154 |
SLAB_CACHE_DMA | \ |
5af608399
|
155 |
SLAB_STORE_USER | \ |
1da177e4c
|
156 |
SLAB_RECLAIM_ACCOUNT | SLAB_PANIC | \ |
3ac7fe5a4
|
157 |
SLAB_DESTROY_BY_RCU | SLAB_MEM_SPREAD | \ |
c175eea46
|
158 |
SLAB_DEBUG_OBJECTS | SLAB_NOLEAKTRACE | SLAB_NOTRACK) |
1da177e4c
|
159 |
#else |
ac2b898ca
|
160 |
# define CREATE_MASK (SLAB_HWCACHE_ALIGN | \ |
5af608399
|
161 |
SLAB_CACHE_DMA | \ |
1da177e4c
|
162 |
SLAB_RECLAIM_ACCOUNT | SLAB_PANIC | \ |
3ac7fe5a4
|
163 |
SLAB_DESTROY_BY_RCU | SLAB_MEM_SPREAD | \ |
c175eea46
|
164 |
SLAB_DEBUG_OBJECTS | SLAB_NOLEAKTRACE | SLAB_NOTRACK) |
1da177e4c
|
165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 |
#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
|
185 |
typedef unsigned int kmem_bufctl_t; |
1da177e4c
|
186 187 |
#define BUFCTL_END (((kmem_bufctl_t)(~0U))-0) #define BUFCTL_FREE (((kmem_bufctl_t)(~0U))-1) |
871751e25
|
188 189 |
#define BUFCTL_ACTIVE (((kmem_bufctl_t)(~0U))-2) #define SLAB_LIMIT (((kmem_bufctl_t)(~0U))-3) |
1da177e4c
|
190 |
|
1da177e4c
|
191 |
/* |
1da177e4c
|
192 193 194 195 196 197 198 199 200 201 202 203 |
* 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. |
1da177e4c
|
204 205 |
*/ struct slab_rcu { |
b28a02de8
|
206 |
struct rcu_head head; |
343e0d7a9
|
207 |
struct kmem_cache *cachep; |
b28a02de8
|
208 |
void *addr; |
1da177e4c
|
209 210 211 |
}; /* |
5bfe53a77
|
212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 |
* 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 { union { struct { 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; }; struct slab_rcu __slab_cover_slab_rcu; }; }; /* |
1da177e4c
|
233 234 |
* struct array_cache * |
1da177e4c
|
235 236 237 238 239 240 241 242 243 244 245 246 247 248 |
* 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
|
249 |
spinlock_t lock; |
bda5b655f
|
250 |
void *entry[]; /* |
a737b3e2f
|
251 252 253 |
* Must have this definition in here for the proper * alignment of array_cache. Also simplifies accessing * the entries. |
a737b3e2f
|
254 |
*/ |
1da177e4c
|
255 |
}; |
a737b3e2f
|
256 257 258 |
/* * bootstrap: The caches do not work without cpuarrays anymore, but the * cpuarrays are allocated from the generic caches... |
1da177e4c
|
259 260 261 262 |
*/ #define BOOT_CPUCACHE_ENTRIES 1 struct arraycache_init { struct array_cache cache; |
b28a02de8
|
263 |
void *entries[BOOT_CPUCACHE_ENTRIES]; |
1da177e4c
|
264 265 266 |
}; /* |
e498be7da
|
267 |
* The slab lists for all objects. |
1da177e4c
|
268 269 |
*/ struct kmem_list3 { |
b28a02de8
|
270 271 272 273 |
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
|
274 |
unsigned int free_limit; |
2e1217cf9
|
275 |
unsigned int colour_next; /* Per-node cache coloring */ |
b28a02de8
|
276 277 278 |
spinlock_t list_lock; struct array_cache *shared; /* shared per node */ struct array_cache **alien; /* on other nodes */ |
35386e3b0
|
279 280 |
unsigned long next_reap; /* updated without locking */ int free_touched; /* updated without locking */ |
1da177e4c
|
281 |
}; |
e498be7da
|
282 283 284 |
/* * Need this for bootstrapping a per node allocator. */ |
556a169da
|
285 |
#define NUM_INIT_LISTS (3 * MAX_NUMNODES) |
68a1b1955
|
286 |
static struct kmem_list3 __initdata initkmem_list3[NUM_INIT_LISTS]; |
e498be7da
|
287 |
#define CACHE_CACHE 0 |
556a169da
|
288 289 |
#define SIZE_AC MAX_NUMNODES #define SIZE_L3 (2 * MAX_NUMNODES) |
e498be7da
|
290 |
|
ed11d9eb2
|
291 292 293 294 |
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); |
83b519e8b
|
295 |
static int enable_cpucache(struct kmem_cache *cachep, gfp_t gfp); |
65f27f384
|
296 |
static void cache_reap(struct work_struct *unused); |
ed11d9eb2
|
297 |
|
e498be7da
|
298 |
/* |
a737b3e2f
|
299 300 |
* 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
|
301 |
*/ |
7243cc05b
|
302 |
static __always_inline int index_of(const size_t size) |
e498be7da
|
303 |
{ |
5ec8a847b
|
304 |
extern void __bad_size(void); |
e498be7da
|
305 306 307 308 309 310 311 312 |
if (__builtin_constant_p(size)) { int i = 0; #define CACHE(x) \ if (size <=x) \ return i; \ else \ i++; |
1c61fc40f
|
313 |
#include <linux/kmalloc_sizes.h> |
e498be7da
|
314 |
#undef CACHE |
5ec8a847b
|
315 |
__bad_size(); |
7243cc05b
|
316 |
} else |
5ec8a847b
|
317 |
__bad_size(); |
e498be7da
|
318 319 |
return 0; } |
e0a427267
|
320 |
static int slab_early_init = 1; |
e498be7da
|
321 322 |
#define INDEX_AC index_of(sizeof(struct arraycache_init)) #define INDEX_L3 index_of(sizeof(struct kmem_list3)) |
1da177e4c
|
323 |
|
5295a74cc
|
324 |
static void kmem_list3_init(struct kmem_list3 *parent) |
e498be7da
|
325 326 327 328 329 330 |
{ 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
|
331 |
parent->colour_next = 0; |
e498be7da
|
332 333 334 335 |
spin_lock_init(&parent->list_lock); parent->free_objects = 0; parent->free_touched = 0; } |
a737b3e2f
|
336 337 338 339 |
#define MAKE_LIST(cachep, listp, slab, nodeid) \ do { \ INIT_LIST_HEAD(listp); \ list_splice(&(cachep->nodelists[nodeid]->slab), listp); \ |
e498be7da
|
340 |
} while (0) |
a737b3e2f
|
341 342 |
#define MAKE_ALL_LISTS(cachep, ptr, nodeid) \ do { \ |
e498be7da
|
343 344 345 346 |
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
|
347 |
|
1da177e4c
|
348 349 350 351 |
#define CFLGS_OFF_SLAB (0x80000000UL) #define OFF_SLAB(x) ((x)->flags & CFLGS_OFF_SLAB) #define BATCHREFILL_LIMIT 16 |
a737b3e2f
|
352 353 354 |
/* * Optimization question: fewer reaps means less probability for unnessary * cpucache drain/refill cycles. |
1da177e4c
|
355 |
* |
dc6f3f276
|
356 |
* OTOH the cpuarrays can contain lots of objects, |
1da177e4c
|
357 358 359 360 361 362 363 364 365 366 |
* 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
|
367 |
#define STATS_ADD_REAPED(x,y) ((x)->reaped += (y)) |
a737b3e2f
|
368 369 370 371 372 |
#define STATS_SET_HIGH(x) \ do { \ if ((x)->num_active > (x)->high_mark) \ (x)->high_mark = (x)->num_active; \ } while (0) |
1da177e4c
|
373 374 |
#define STATS_INC_ERR(x) ((x)->errors++) #define STATS_INC_NODEALLOCS(x) ((x)->node_allocs++) |
e498be7da
|
375 |
#define STATS_INC_NODEFREES(x) ((x)->node_frees++) |
fb7faf331
|
376 |
#define STATS_INC_ACOVERFLOW(x) ((x)->node_overflow++) |
a737b3e2f
|
377 378 379 380 381 |
#define STATS_SET_FREEABLE(x, i) \ do { \ if ((x)->max_freeable < i) \ (x)->max_freeable = i; \ } while (0) |
1da177e4c
|
382 383 384 385 386 387 388 389 390 |
#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) |
4e60c86bd
|
391 |
#define STATS_ADD_REAPED(x,y) do { (void)(y); } while (0) |
1da177e4c
|
392 393 394 |
#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
|
395 |
#define STATS_INC_NODEFREES(x) do { } while (0) |
fb7faf331
|
396 |
#define STATS_INC_ACOVERFLOW(x) do { } while (0) |
a737b3e2f
|
397 |
#define STATS_SET_FREEABLE(x, i) do { } while (0) |
1da177e4c
|
398 399 400 401 402 403 404 |
#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
|
405 |
|
a737b3e2f
|
406 407 |
/* * memory layout of objects: |
1da177e4c
|
408 |
* 0 : objp |
3dafccf22
|
409 |
* 0 .. cachep->obj_offset - BYTES_PER_WORD - 1: padding. This ensures that |
1da177e4c
|
410 411 |
* the end of an object is aligned with the end of the real * allocation. Catches writes behind the end of the allocation. |
3dafccf22
|
412 |
* cachep->obj_offset - BYTES_PER_WORD .. cachep->obj_offset - 1: |
1da177e4c
|
413 |
* redzone word. |
3dafccf22
|
414 415 |
* cachep->obj_offset: The real object. * cachep->buffer_size - 2* BYTES_PER_WORD: redzone word [BYTES_PER_WORD long] |
a737b3e2f
|
416 417 |
* cachep->buffer_size - 1* BYTES_PER_WORD: last caller address * [BYTES_PER_WORD long] |
1da177e4c
|
418 |
*/ |
343e0d7a9
|
419 |
static int obj_offset(struct kmem_cache *cachep) |
1da177e4c
|
420 |
{ |
3dafccf22
|
421 |
return cachep->obj_offset; |
1da177e4c
|
422 |
} |
343e0d7a9
|
423 |
static int obj_size(struct kmem_cache *cachep) |
1da177e4c
|
424 |
{ |
3dafccf22
|
425 |
return cachep->obj_size; |
1da177e4c
|
426 |
} |
b46b8f19c
|
427 |
static unsigned long long *dbg_redzone1(struct kmem_cache *cachep, void *objp) |
1da177e4c
|
428 429 |
{ BUG_ON(!(cachep->flags & SLAB_RED_ZONE)); |
b46b8f19c
|
430 431 |
return (unsigned long long*) (objp + obj_offset(cachep) - sizeof(unsigned long long)); |
1da177e4c
|
432 |
} |
b46b8f19c
|
433 |
static unsigned long long *dbg_redzone2(struct kmem_cache *cachep, void *objp) |
1da177e4c
|
434 435 436 |
{ BUG_ON(!(cachep->flags & SLAB_RED_ZONE)); if (cachep->flags & SLAB_STORE_USER) |
b46b8f19c
|
437 438 |
return (unsigned long long *)(objp + cachep->buffer_size - sizeof(unsigned long long) - |
87a927c71
|
439 |
REDZONE_ALIGN); |
b46b8f19c
|
440 441 |
return (unsigned long long *) (objp + cachep->buffer_size - sizeof(unsigned long long)); |
1da177e4c
|
442 |
} |
343e0d7a9
|
443 |
static void **dbg_userword(struct kmem_cache *cachep, void *objp) |
1da177e4c
|
444 445 |
{ BUG_ON(!(cachep->flags & SLAB_STORE_USER)); |
3dafccf22
|
446 |
return (void **)(objp + cachep->buffer_size - BYTES_PER_WORD); |
1da177e4c
|
447 448 449 |
} #else |
3dafccf22
|
450 451 |
#define obj_offset(x) 0 #define obj_size(cachep) (cachep->buffer_size) |
b46b8f19c
|
452 453 |
#define dbg_redzone1(cachep, objp) ({BUG(); (unsigned long long *)NULL;}) #define dbg_redzone2(cachep, objp) ({BUG(); (unsigned long long *)NULL;}) |
1da177e4c
|
454 455 456 |
#define dbg_userword(cachep, objp) ({BUG(); (void **)NULL;}) #endif |
0f24f1287
|
457 |
#ifdef CONFIG_TRACING |
36555751c
|
458 459 460 461 462 463 |
size_t slab_buffer_size(struct kmem_cache *cachep) { return cachep->buffer_size; } EXPORT_SYMBOL(slab_buffer_size); #endif |
1da177e4c
|
464 |
/* |
1da177e4c
|
465 466 467 468 469 |
* 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
|
470 471 472 473 |
/* * 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
|
474 |
*/ |
065d41cb2
|
475 476 477 478 479 480 481 |
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
|
482 |
page = compound_head(page); |
ddc2e812d
|
483 |
BUG_ON(!PageSlab(page)); |
065d41cb2
|
484 485 486 487 488 489 490 491 492 493 |
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
|
494 |
BUG_ON(!PageSlab(page)); |
065d41cb2
|
495 496 |
return (struct slab *)page->lru.prev; } |
1da177e4c
|
497 |
|
6ed5eb221
|
498 499 |
static inline struct kmem_cache *virt_to_cache(const void *obj) { |
b49af68ff
|
500 |
struct page *page = virt_to_head_page(obj); |
6ed5eb221
|
501 502 503 504 505 |
return page_get_cache(page); } static inline struct slab *virt_to_slab(const void *obj) { |
b49af68ff
|
506 |
struct page *page = virt_to_head_page(obj); |
6ed5eb221
|
507 508 |
return page_get_slab(page); } |
8fea4e96a
|
509 510 511 512 513 |
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
|
514 515 516 517 518 519 520 521 |
/* * 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
|
522 |
{ |
6a2d7a955
|
523 524 |
u32 offset = (obj - slab->s_mem); return reciprocal_divide(offset, cache->reciprocal_buffer_size); |
8fea4e96a
|
525 |
} |
a737b3e2f
|
526 527 528 |
/* * These are the default caches for kmalloc. Custom caches can have other sizes. */ |
1da177e4c
|
529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 |
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
|
546 |
{NULL,} |
1da177e4c
|
547 548 549 550 |
#undef CACHE }; static struct arraycache_init initarray_cache __initdata = |
b28a02de8
|
551 |
{ {0, BOOT_CPUCACHE_ENTRIES, 1, 0} }; |
1da177e4c
|
552 |
static struct arraycache_init initarray_generic = |
b28a02de8
|
553 |
{ {0, BOOT_CPUCACHE_ENTRIES, 1, 0} }; |
1da177e4c
|
554 555 |
/* internal cache of cache description objs */ |
b56efcf0a
|
556 |
static struct kmem_list3 *cache_cache_nodelists[MAX_NUMNODES]; |
343e0d7a9
|
557 |
static struct kmem_cache cache_cache = { |
b56efcf0a
|
558 |
.nodelists = cache_cache_nodelists, |
b28a02de8
|
559 560 561 |
.batchcount = 1, .limit = BOOT_CPUCACHE_ENTRIES, .shared = 1, |
343e0d7a9
|
562 |
.buffer_size = sizeof(struct kmem_cache), |
b28a02de8
|
563 |
.name = "kmem_cache", |
1da177e4c
|
564 |
}; |
056c62418
|
565 |
#define BAD_ALIEN_MAGIC 0x01020304ul |
ce79ddc8e
|
566 567 568 569 570 571 572 573 574 |
/* * chicken and egg problem: delay the per-cpu array allocation * until the general caches are up. */ static enum { NONE, PARTIAL_AC, PARTIAL_L3, EARLY, |
52cef1891
|
575 |
LATE, |
ce79ddc8e
|
576 577 578 579 580 581 582 583 584 585 |
FULL } g_cpucache_up; /* * used by boot code to determine if it can use slab based allocator */ int slab_is_available(void) { return g_cpucache_up >= EARLY; } |
f1aaee53f
|
586 587 588 589 590 591 592 593 |
#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
|
594 595 596 597 |
* * 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
|
598 |
*/ |
056c62418
|
599 600 |
static struct lock_class_key on_slab_l3_key; static struct lock_class_key on_slab_alc_key; |
83835b3d9
|
601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 |
static struct lock_class_key debugobj_l3_key; static struct lock_class_key debugobj_alc_key; static void slab_set_lock_classes(struct kmem_cache *cachep, struct lock_class_key *l3_key, struct lock_class_key *alc_key, int q) { struct array_cache **alc; struct kmem_list3 *l3; int r; l3 = cachep->nodelists[q]; if (!l3) return; lockdep_set_class(&l3->list_lock, 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) return; for_each_node(r) { if (alc[r]) lockdep_set_class(&alc[r]->lock, alc_key); } } static void slab_set_debugobj_lock_classes_node(struct kmem_cache *cachep, int node) { slab_set_lock_classes(cachep, &debugobj_l3_key, &debugobj_alc_key, node); } static void slab_set_debugobj_lock_classes(struct kmem_cache *cachep) { int node; for_each_online_node(node) slab_set_debugobj_lock_classes_node(cachep, node); } |
ce79ddc8e
|
645 |
static void init_node_lock_keys(int q) |
f1aaee53f
|
646 |
{ |
056c62418
|
647 |
struct cache_sizes *s = malloc_sizes; |
52cef1891
|
648 |
if (g_cpucache_up < LATE) |
ce79ddc8e
|
649 650 651 |
return; for (s = malloc_sizes; s->cs_size != ULONG_MAX; s++) { |
ce79ddc8e
|
652 |
struct kmem_list3 *l3; |
ce79ddc8e
|
653 654 655 |
l3 = s->cs_cachep->nodelists[q]; if (!l3 || OFF_SLAB(s->cs_cachep)) |
00afa7580
|
656 |
continue; |
83835b3d9
|
657 658 659 |
slab_set_lock_classes(s->cs_cachep, &on_slab_l3_key, &on_slab_alc_key, q); |
f1aaee53f
|
660 661 |
} } |
ce79ddc8e
|
662 663 664 665 666 667 668 669 |
static inline void init_lock_keys(void) { int node; for_each_node(node) init_node_lock_keys(node); } |
f1aaee53f
|
670 |
#else |
ce79ddc8e
|
671 672 673 |
static void init_node_lock_keys(int q) { } |
056c62418
|
674 |
static inline void init_lock_keys(void) |
f1aaee53f
|
675 676 |
{ } |
83835b3d9
|
677 678 679 680 681 682 683 684 |
static void slab_set_debugobj_lock_classes_node(struct kmem_cache *cachep, int node) { } static void slab_set_debugobj_lock_classes(struct kmem_cache *cachep) { } |
f1aaee53f
|
685 |
#endif |
8f5be20bf
|
686 |
/* |
95402b382
|
687 |
* Guard access to the cache-chain. |
8f5be20bf
|
688 |
*/ |
fc0abb145
|
689 |
static DEFINE_MUTEX(cache_chain_mutex); |
1da177e4c
|
690 |
static struct list_head cache_chain; |
1871e52c7
|
691 |
static DEFINE_PER_CPU(struct delayed_work, slab_reap_work); |
1da177e4c
|
692 |
|
343e0d7a9
|
693 |
static inline struct array_cache *cpu_cache_get(struct kmem_cache *cachep) |
1da177e4c
|
694 695 696 |
{ return cachep->array[smp_processor_id()]; } |
a737b3e2f
|
697 698 |
static inline struct kmem_cache *__find_general_cachep(size_t size, gfp_t gfpflags) |
1da177e4c
|
699 700 701 702 703 |
{ struct cache_sizes *csizep = malloc_sizes; #if DEBUG /* This happens if someone tries to call |
b28a02de8
|
704 705 706 |
* kmem_cache_create(), or __kmalloc(), before * the generic caches are initialized. */ |
c7e43c78a
|
707 |
BUG_ON(malloc_sizes[INDEX_AC].cs_cachep == NULL); |
1da177e4c
|
708 |
#endif |
6cb8f9132
|
709 710 |
if (!size) return ZERO_SIZE_PTR; |
1da177e4c
|
711 712 713 714 |
while (size > csizep->cs_size) csizep++; /* |
0abf40c1a
|
715 |
* Really subtle: The last entry with cs->cs_size==ULONG_MAX |
1da177e4c
|
716 717 718 |
* has cs_{dma,}cachep==NULL. Thus no special case * for large kmalloc calls required. */ |
4b51d6698
|
719 |
#ifdef CONFIG_ZONE_DMA |
1da177e4c
|
720 721 |
if (unlikely(gfpflags & GFP_DMA)) return csizep->cs_dmacachep; |
4b51d6698
|
722 |
#endif |
1da177e4c
|
723 724 |
return csizep->cs_cachep; } |
b221385bc
|
725 |
static struct kmem_cache *kmem_find_general_cachep(size_t size, gfp_t gfpflags) |
97e2bde47
|
726 727 728 |
{ return __find_general_cachep(size, gfpflags); } |
97e2bde47
|
729 |
|
fbaccacff
|
730 |
static size_t slab_mgmt_size(size_t nr_objs, size_t align) |
1da177e4c
|
731 |
{ |
fbaccacff
|
732 733 |
return ALIGN(sizeof(struct slab)+nr_objs*sizeof(kmem_bufctl_t), align); } |
1da177e4c
|
734 |
|
a737b3e2f
|
735 736 737 |
/* * Calculate the number of objects and left-over bytes for a given buffer size. */ |
fbaccacff
|
738 739 740 741 742 743 744 |
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
|
745 |
|
fbaccacff
|
746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 |
/* * 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
|
794 |
} |
d40cee245
|
795 |
#define slab_error(cachep, msg) __slab_error(__func__, cachep, msg) |
1da177e4c
|
796 |
|
a737b3e2f
|
797 798 |
static void __slab_error(const char *function, struct kmem_cache *cachep, char *msg) |
1da177e4c
|
799 800 801 |
{ printk(KERN_ERR "slab error in %s(): cache `%s': %s ", |
b28a02de8
|
802 |
function, cachep->name, msg); |
1da177e4c
|
803 804 |
dump_stack(); } |
3395ee058
|
805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 |
/* * 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; static int __init noaliencache_setup(char *s) { use_alien_caches = 0; return 1; } __setup("noaliencache", noaliencache_setup); |
8fce4d8e3
|
820 821 822 823 824 825 826 |
#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. */ |
1871e52c7
|
827 |
static DEFINE_PER_CPU(unsigned long, slab_reap_node); |
8fce4d8e3
|
828 829 830 831 |
static void init_reap_node(int cpu) { int node; |
7d6e6d09d
|
832 |
node = next_node(cpu_to_mem(cpu), node_online_map); |
8fce4d8e3
|
833 |
if (node == MAX_NUMNODES) |
442295c94
|
834 |
node = first_node(node_online_map); |
8fce4d8e3
|
835 |
|
1871e52c7
|
836 |
per_cpu(slab_reap_node, cpu) = node; |
8fce4d8e3
|
837 838 839 840 |
} static void next_reap_node(void) { |
909ea9646
|
841 |
int node = __this_cpu_read(slab_reap_node); |
8fce4d8e3
|
842 |
|
8fce4d8e3
|
843 844 845 |
node = next_node(node, node_online_map); if (unlikely(node >= MAX_NUMNODES)) node = first_node(node_online_map); |
909ea9646
|
846 |
__this_cpu_write(slab_reap_node, node); |
8fce4d8e3
|
847 848 849 850 851 852 |
} #else #define init_reap_node(cpu) do { } while (0) #define next_reap_node(void) do { } while (0) #endif |
1da177e4c
|
853 854 855 856 857 858 859 |
/* * 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
|
860 |
static void __cpuinit start_cpu_timer(int cpu) |
1da177e4c
|
861 |
{ |
1871e52c7
|
862 |
struct delayed_work *reap_work = &per_cpu(slab_reap_work, cpu); |
1da177e4c
|
863 864 865 866 867 868 |
/* * When this gets called from do_initcalls via cpucache_init(), * init_workqueues() has already run, so keventd will be setup * at that time. */ |
52bad64d9
|
869 |
if (keventd_up() && reap_work->work.func == NULL) { |
8fce4d8e3
|
870 |
init_reap_node(cpu); |
78b435368
|
871 |
INIT_DELAYED_WORK_DEFERRABLE(reap_work, cache_reap); |
2b2842146
|
872 873 |
schedule_delayed_work_on(cpu, reap_work, __round_jiffies_relative(HZ, cpu)); |
1da177e4c
|
874 875 |
} } |
e498be7da
|
876 |
static struct array_cache *alloc_arraycache(int node, int entries, |
83b519e8b
|
877 |
int batchcount, gfp_t gfp) |
1da177e4c
|
878 |
{ |
b28a02de8
|
879 |
int memsize = sizeof(void *) * entries + sizeof(struct array_cache); |
1da177e4c
|
880 |
struct array_cache *nc = NULL; |
83b519e8b
|
881 |
nc = kmalloc_node(memsize, gfp, node); |
d5cff6352
|
882 883 |
/* * The array_cache structures contain pointers to free object. |
25985edce
|
884 |
* However, when such objects are allocated or transferred to another |
d5cff6352
|
885 886 887 888 889 |
* cache the pointers are not cleared and they could be counted as * valid references during a kmemleak scan. Therefore, kmemleak must * not scan such objects. */ kmemleak_no_scan(nc); |
1da177e4c
|
890 891 892 893 894 |
if (nc) { nc->avail = 0; nc->limit = entries; nc->batchcount = batchcount; nc->touched = 0; |
e498be7da
|
895 |
spin_lock_init(&nc->lock); |
1da177e4c
|
896 897 898 |
} return nc; } |
3ded175a4
|
899 900 901 902 903 904 905 906 907 908 |
/* * 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 */ |
732eacc05
|
909 |
int nr = min3(from->avail, max, to->limit - to->avail); |
3ded175a4
|
910 911 912 913 914 915 916 917 918 |
if (!nr) return 0; memcpy(to->entry + to->avail, from->entry + from->avail -nr, sizeof(void *) *nr); from->avail -= nr; to->avail += nr; |
3ded175a4
|
919 920 |
return nr; } |
765c4507a
|
921 922 923 924 |
#ifndef CONFIG_NUMA #define drain_alien_cache(cachep, alien) do { } while (0) #define reap_alien(cachep, l3) do { } while (0) |
83b519e8b
|
925 |
static inline struct array_cache **alloc_alien_cache(int node, int limit, gfp_t gfp) |
765c4507a
|
926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 |
{ 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
|
944 |
static inline void *____cache_alloc_node(struct kmem_cache *cachep, |
765c4507a
|
945 946 947 948 949 950 |
gfp_t flags, int nodeid) { return NULL; } #else /* CONFIG_NUMA */ |
8b98c1699
|
951 |
static void *____cache_alloc_node(struct kmem_cache *, gfp_t, int); |
c61afb181
|
952 |
static void *alternate_node_alloc(struct kmem_cache *, gfp_t); |
dc85da15d
|
953 |
|
83b519e8b
|
954 |
static struct array_cache **alloc_alien_cache(int node, int limit, gfp_t gfp) |
e498be7da
|
955 956 |
{ struct array_cache **ac_ptr; |
8ef828668
|
957 |
int memsize = sizeof(void *) * nr_node_ids; |
e498be7da
|
958 959 960 961 |
int i; if (limit > 1) limit = 12; |
f3186a9c5
|
962 |
ac_ptr = kzalloc_node(memsize, gfp, node); |
e498be7da
|
963 964 |
if (ac_ptr) { for_each_node(i) { |
f3186a9c5
|
965 |
if (i == node || !node_online(i)) |
e498be7da
|
966 |
continue; |
83b519e8b
|
967 |
ac_ptr[i] = alloc_arraycache(node, limit, 0xbaadf00d, gfp); |
e498be7da
|
968 |
if (!ac_ptr[i]) { |
cc550defe
|
969 |
for (i--; i >= 0; i--) |
e498be7da
|
970 971 972 973 974 975 976 977 |
kfree(ac_ptr[i]); kfree(ac_ptr); return NULL; } } } return ac_ptr; } |
5295a74cc
|
978 |
static void free_alien_cache(struct array_cache **ac_ptr) |
e498be7da
|
979 980 981 982 983 |
{ int i; if (!ac_ptr) return; |
e498be7da
|
984 |
for_each_node(i) |
b28a02de8
|
985 |
kfree(ac_ptr[i]); |
e498be7da
|
986 987 |
kfree(ac_ptr); } |
343e0d7a9
|
988 |
static void __drain_alien_cache(struct kmem_cache *cachep, |
5295a74cc
|
989 |
struct array_cache *ac, int node) |
e498be7da
|
990 991 992 993 994 |
{ struct kmem_list3 *rl3 = cachep->nodelists[node]; if (ac->avail) { spin_lock(&rl3->list_lock); |
e00946fe2
|
995 996 997 998 999 |
/* * 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
|
1000 1001 |
if (rl3->shared) transfer_objects(rl3->shared, ac, ac->limit); |
e00946fe2
|
1002 |
|
ff69416e6
|
1003 |
free_block(cachep, ac->entry, ac->avail, node); |
e498be7da
|
1004 1005 1006 1007 |
ac->avail = 0; spin_unlock(&rl3->list_lock); } } |
8fce4d8e3
|
1008 1009 1010 1011 1012 |
/* * Called from cache_reap() to regularly drain alien caches round robin. */ static void reap_alien(struct kmem_cache *cachep, struct kmem_list3 *l3) { |
909ea9646
|
1013 |
int node = __this_cpu_read(slab_reap_node); |
8fce4d8e3
|
1014 1015 1016 |
if (l3->alien) { struct array_cache *ac = l3->alien[node]; |
e00946fe2
|
1017 1018 |
if (ac && ac->avail && spin_trylock_irq(&ac->lock)) { |
8fce4d8e3
|
1019 1020 1021 1022 1023 |
__drain_alien_cache(cachep, ac, node); spin_unlock_irq(&ac->lock); } } } |
a737b3e2f
|
1024 1025 |
static void drain_alien_cache(struct kmem_cache *cachep, struct array_cache **alien) |
e498be7da
|
1026 |
{ |
b28a02de8
|
1027 |
int i = 0; |
e498be7da
|
1028 1029 1030 1031 |
struct array_cache *ac; unsigned long flags; for_each_online_node(i) { |
4484ebf12
|
1032 |
ac = alien[i]; |
e498be7da
|
1033 1034 1035 1036 1037 1038 1039 |
if (ac) { spin_lock_irqsave(&ac->lock, flags); __drain_alien_cache(cachep, ac, i); spin_unlock_irqrestore(&ac->lock, flags); } } } |
729bd0b74
|
1040 |
|
873623dfa
|
1041 |
static inline int cache_free_alien(struct kmem_cache *cachep, void *objp) |
729bd0b74
|
1042 1043 1044 1045 1046 |
{ struct slab *slabp = virt_to_slab(objp); int nodeid = slabp->nodeid; struct kmem_list3 *l3; struct array_cache *alien = NULL; |
1ca4cb241
|
1047 |
int node; |
7d6e6d09d
|
1048 |
node = numa_mem_id(); |
729bd0b74
|
1049 1050 1051 1052 1053 |
/* * Make sure we are not freeing a object from another node to the array * cache on this cpu. */ |
62918a036
|
1054 |
if (likely(slabp->nodeid == node)) |
729bd0b74
|
1055 |
return 0; |
1ca4cb241
|
1056 |
l3 = cachep->nodelists[node]; |
729bd0b74
|
1057 1058 1059 |
STATS_INC_NODEFREES(cachep); if (l3->alien && l3->alien[nodeid]) { alien = l3->alien[nodeid]; |
873623dfa
|
1060 |
spin_lock(&alien->lock); |
729bd0b74
|
1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 |
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
|
1074 |
#endif |
8f9f8d9e8
|
1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 |
/* * Allocates and initializes nodelists for a node on each slab cache, used for * either memory or cpu hotplug. If memory is being hot-added, the kmem_list3 * will be allocated off-node since memory is not yet online for the new node. * When hotplugging memory or a cpu, existing nodelists are not replaced if * already in use. * * Must hold cache_chain_mutex. */ static int init_cache_nodelists_node(int node) { struct kmem_cache *cachep; struct kmem_list3 *l3; const int memsize = sizeof(struct kmem_list3); list_for_each_entry(cachep, &cache_chain, next) { /* * 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 */ if (!cachep->nodelists[node]) { l3 = kmalloc_node(memsize, GFP_KERNEL, node); if (!l3) return -ENOMEM; kmem_list3_init(l3); l3->next_reap = jiffies + REAPTIMEOUT_LIST3 + ((unsigned long)cachep) % REAPTIMEOUT_LIST3; /* * The l3s don't come and go as CPUs come and * go. cache_chain_mutex is sufficient * protection here. */ cachep->nodelists[node] = l3; } 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); } return 0; } |
fbf1e473b
|
1120 1121 1122 1123 |
static void __cpuinit cpuup_canceled(long cpu) { struct kmem_cache *cachep; struct kmem_list3 *l3 = NULL; |
7d6e6d09d
|
1124 |
int node = cpu_to_mem(cpu); |
a70f73028
|
1125 |
const struct cpumask *mask = cpumask_of_node(node); |
fbf1e473b
|
1126 1127 1128 1129 1130 |
list_for_each_entry(cachep, &cache_chain, next) { struct array_cache *nc; struct array_cache *shared; struct array_cache **alien; |
fbf1e473b
|
1131 |
|
fbf1e473b
|
1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 |
/* 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); |
58463c1fe
|
1146 |
if (!cpumask_empty(mask)) { |
fbf1e473b
|
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 1180 1181 1182 1183 1184 |
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
|
1185 |
{ |
343e0d7a9
|
1186 |
struct kmem_cache *cachep; |
e498be7da
|
1187 |
struct kmem_list3 *l3 = NULL; |
7d6e6d09d
|
1188 |
int node = cpu_to_mem(cpu); |
8f9f8d9e8
|
1189 |
int err; |
1da177e4c
|
1190 |
|
fbf1e473b
|
1191 1192 1193 1194 1195 1196 |
/* * 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 */ |
8f9f8d9e8
|
1197 1198 1199 |
err = init_cache_nodelists_node(node); if (err < 0) goto bad; |
fbf1e473b
|
1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 |
/* * 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, |
83b519e8b
|
1211 |
cachep->batchcount, GFP_KERNEL); |
fbf1e473b
|
1212 1213 1214 1215 1216 |
if (!nc) goto bad; if (cachep->shared) { shared = alloc_arraycache(node, cachep->shared * cachep->batchcount, |
83b519e8b
|
1217 |
0xbaadf00d, GFP_KERNEL); |
12d00f6a1
|
1218 1219 |
if (!shared) { kfree(nc); |
1da177e4c
|
1220 |
goto bad; |
12d00f6a1
|
1221 |
} |
fbf1e473b
|
1222 1223 |
} if (use_alien_caches) { |
83b519e8b
|
1224 |
alien = alloc_alien_cache(node, cachep->limit, GFP_KERNEL); |
12d00f6a1
|
1225 1226 1227 |
if (!alien) { kfree(shared); kfree(nc); |
fbf1e473b
|
1228 |
goto bad; |
12d00f6a1
|
1229 |
} |
fbf1e473b
|
1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 |
} 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
|
1244 |
#ifdef CONFIG_NUMA |
fbf1e473b
|
1245 1246 1247 |
if (!l3->alien) { l3->alien = alien; alien = NULL; |
1da177e4c
|
1248 |
} |
fbf1e473b
|
1249 1250 1251 1252 |
#endif spin_unlock_irq(&l3->list_lock); kfree(shared); free_alien_cache(alien); |
83835b3d9
|
1253 1254 |
if (cachep->flags & SLAB_DEBUG_OBJECTS) slab_set_debugobj_lock_classes_node(cachep, node); |
fbf1e473b
|
1255 |
} |
ce79ddc8e
|
1256 |
init_node_lock_keys(node); |
fbf1e473b
|
1257 1258 |
return 0; bad: |
12d00f6a1
|
1259 |
cpuup_canceled(cpu); |
fbf1e473b
|
1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 |
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
|
1270 1271 |
case CPU_UP_PREPARE: case CPU_UP_PREPARE_FROZEN: |
95402b382
|
1272 |
mutex_lock(&cache_chain_mutex); |
fbf1e473b
|
1273 |
err = cpuup_prepare(cpu); |
95402b382
|
1274 |
mutex_unlock(&cache_chain_mutex); |
1da177e4c
|
1275 1276 |
break; case CPU_ONLINE: |
8bb784428
|
1277 |
case CPU_ONLINE_FROZEN: |
1da177e4c
|
1278 1279 1280 |
start_cpu_timer(cpu); break; #ifdef CONFIG_HOTPLUG_CPU |
5830c5902
|
1281 |
case CPU_DOWN_PREPARE: |
8bb784428
|
1282 |
case CPU_DOWN_PREPARE_FROZEN: |
5830c5902
|
1283 1284 1285 1286 1287 1288 |
/* * 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. */ |
afe2c511f
|
1289 |
cancel_delayed_work_sync(&per_cpu(slab_reap_work, cpu)); |
5830c5902
|
1290 |
/* Now the cache_reaper is guaranteed to be not running. */ |
1871e52c7
|
1291 |
per_cpu(slab_reap_work, cpu).work.func = NULL; |
5830c5902
|
1292 1293 |
break; case CPU_DOWN_FAILED: |
8bb784428
|
1294 |
case CPU_DOWN_FAILED_FROZEN: |
5830c5902
|
1295 1296 |
start_cpu_timer(cpu); break; |
1da177e4c
|
1297 |
case CPU_DEAD: |
8bb784428
|
1298 |
case CPU_DEAD_FROZEN: |
4484ebf12
|
1299 1300 1301 1302 1303 1304 1305 1306 |
/* * 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
|
1307 |
/* fall through */ |
8f5be20bf
|
1308 |
#endif |
1da177e4c
|
1309 |
case CPU_UP_CANCELED: |
8bb784428
|
1310 |
case CPU_UP_CANCELED_FROZEN: |
95402b382
|
1311 |
mutex_lock(&cache_chain_mutex); |
fbf1e473b
|
1312 |
cpuup_canceled(cpu); |
fc0abb145
|
1313 |
mutex_unlock(&cache_chain_mutex); |
1da177e4c
|
1314 |
break; |
1da177e4c
|
1315 |
} |
eac406801
|
1316 |
return notifier_from_errno(err); |
1da177e4c
|
1317 |
} |
74b85f379
|
1318 1319 1320 |
static struct notifier_block __cpuinitdata cpucache_notifier = { &cpuup_callback, NULL, 0 }; |
1da177e4c
|
1321 |
|
8f9f8d9e8
|
1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 |
#if defined(CONFIG_NUMA) && defined(CONFIG_MEMORY_HOTPLUG) /* * Drains freelist for a node on each slab cache, used for memory hot-remove. * Returns -EBUSY if all objects cannot be drained so that the node is not * removed. * * Must hold cache_chain_mutex. */ static int __meminit drain_cache_nodelists_node(int node) { struct kmem_cache *cachep; int ret = 0; list_for_each_entry(cachep, &cache_chain, next) { struct kmem_list3 *l3; l3 = cachep->nodelists[node]; if (!l3) continue; drain_freelist(cachep, l3, l3->free_objects); if (!list_empty(&l3->slabs_full) || !list_empty(&l3->slabs_partial)) { ret = -EBUSY; break; } } return ret; } static int __meminit slab_memory_callback(struct notifier_block *self, unsigned long action, void *arg) { struct memory_notify *mnb = arg; int ret = 0; int nid; nid = mnb->status_change_nid; if (nid < 0) goto out; switch (action) { case MEM_GOING_ONLINE: mutex_lock(&cache_chain_mutex); ret = init_cache_nodelists_node(nid); mutex_unlock(&cache_chain_mutex); break; case MEM_GOING_OFFLINE: mutex_lock(&cache_chain_mutex); ret = drain_cache_nodelists_node(nid); mutex_unlock(&cache_chain_mutex); break; case MEM_ONLINE: case MEM_OFFLINE: case MEM_CANCEL_ONLINE: case MEM_CANCEL_OFFLINE: break; } out: |
5fda1bd5b
|
1382 |
return notifier_from_errno(ret); |
8f9f8d9e8
|
1383 1384 |
} #endif /* CONFIG_NUMA && CONFIG_MEMORY_HOTPLUG */ |
e498be7da
|
1385 1386 1387 |
/* * swap the static kmem_list3 with kmalloced memory */ |
8f9f8d9e8
|
1388 1389 |
static void __init init_list(struct kmem_cache *cachep, struct kmem_list3 *list, int nodeid) |
e498be7da
|
1390 1391 |
{ struct kmem_list3 *ptr; |
83b519e8b
|
1392 |
ptr = kmalloc_node(sizeof(struct kmem_list3), GFP_NOWAIT, nodeid); |
e498be7da
|
1393 |
BUG_ON(!ptr); |
e498be7da
|
1394 |
memcpy(ptr, list, sizeof(struct kmem_list3)); |
2b2d5493e
|
1395 1396 1397 1398 |
/* * Do not assume that spinlocks can be initialized via memcpy: */ spin_lock_init(&ptr->list_lock); |
e498be7da
|
1399 1400 |
MAKE_ALL_LISTS(cachep, ptr, nodeid); cachep->nodelists[nodeid] = ptr; |
e498be7da
|
1401 |
} |
a737b3e2f
|
1402 |
/* |
556a169da
|
1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 |
* 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
|
1419 1420 |
* Initialisation. Called after the page allocator have been initialised and * before smp_init(). |
1da177e4c
|
1421 1422 1423 1424 1425 1426 |
*/ void __init kmem_cache_init(void) { size_t left_over; struct cache_sizes *sizes; struct cache_names *names; |
e498be7da
|
1427 |
int i; |
07ed76b2a
|
1428 |
int order; |
1ca4cb241
|
1429 |
int node; |
e498be7da
|
1430 |
|
b6e68bc1b
|
1431 |
if (num_possible_nodes() == 1) |
62918a036
|
1432 |
use_alien_caches = 0; |
e498be7da
|
1433 1434 1435 1436 1437 |
for (i = 0; i < NUM_INIT_LISTS; i++) { kmem_list3_init(&initkmem_list3[i]); if (i < MAX_NUMNODES) cache_cache.nodelists[i] = NULL; } |
556a169da
|
1438 |
set_up_list3s(&cache_cache, CACHE_CACHE); |
1da177e4c
|
1439 1440 1441 1442 1443 |
/* * Fragmentation resistance on low memory - only use bigger * page orders on machines with more than 32MB of memory. */ |
4481374ce
|
1444 |
if (totalram_pages > (32 << 20) >> PAGE_SHIFT) |
1da177e4c
|
1445 |
slab_break_gfp_order = BREAK_GFP_ORDER_HI; |
1da177e4c
|
1446 1447 |
/* Bootstrap is tricky, because several objects are allocated * from caches that do not exist yet: |
a737b3e2f
|
1448 1449 1450 |
* 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
|
1451 1452 1453 |
* 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
|
1454 |
* 2) Create the first kmalloc cache. |
343e0d7a9
|
1455 |
* The struct kmem_cache for the new cache is allocated normally. |
e498be7da
|
1456 1457 1458 |
* An __init data area is used for the head array. * 3) Create the remaining kmalloc caches, with minimally sized * head arrays. |
1da177e4c
|
1459 1460 |
* 4) Replace the __init data head arrays for cache_cache and the first * kmalloc cache with kmalloc allocated arrays. |
e498be7da
|
1461 1462 1463 |
* 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
|
1464 |
*/ |
7d6e6d09d
|
1465 |
node = numa_mem_id(); |
1ca4cb241
|
1466 |
|
1da177e4c
|
1467 |
/* 1) create the cache_cache */ |
1da177e4c
|
1468 1469 1470 1471 |
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
|
1472 |
cache_cache.nodelists[node] = &initkmem_list3[CACHE_CACHE + node]; |
1da177e4c
|
1473 |
|
8da3430d8
|
1474 |
/* |
b56efcf0a
|
1475 |
* struct kmem_cache size depends on nr_node_ids & nr_cpu_ids |
8da3430d8
|
1476 |
*/ |
b56efcf0a
|
1477 1478 |
cache_cache.buffer_size = offsetof(struct kmem_cache, array[nr_cpu_ids]) + nr_node_ids * sizeof(struct kmem_list3 *); |
8da3430d8
|
1479 1480 1481 |
#if DEBUG cache_cache.obj_size = cache_cache.buffer_size; #endif |
a737b3e2f
|
1482 1483 |
cache_cache.buffer_size = ALIGN(cache_cache.buffer_size, cache_line_size()); |
6a2d7a955
|
1484 1485 |
cache_cache.reciprocal_buffer_size = reciprocal_value(cache_cache.buffer_size); |
1da177e4c
|
1486 |
|
07ed76b2a
|
1487 1488 1489 1490 1491 1492 |
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
|
1493 |
BUG_ON(!cache_cache.num); |
07ed76b2a
|
1494 |
cache_cache.gfporder = order; |
b28a02de8
|
1495 |
cache_cache.colour = left_over / cache_cache.colour_off; |
b28a02de8
|
1496 1497 |
cache_cache.slab_size = ALIGN(cache_cache.num * sizeof(kmem_bufctl_t) + sizeof(struct slab), cache_line_size()); |
1da177e4c
|
1498 1499 1500 1501 |
/* 2+3) create the kmalloc caches */ sizes = malloc_sizes; names = cache_names; |
a737b3e2f
|
1502 1503 1504 1505 |
/* * Initialize the caches that provide memory for the array cache and the * kmem_list3 structures first. Without this, further allocations will * bug. |
e498be7da
|
1506 1507 1508 |
*/ sizes[INDEX_AC].cs_cachep = kmem_cache_create(names[INDEX_AC].name, |
a737b3e2f
|
1509 1510 1511 |
sizes[INDEX_AC].cs_size, ARCH_KMALLOC_MINALIGN, ARCH_KMALLOC_FLAGS|SLAB_PANIC, |
20c2df83d
|
1512 |
NULL); |
e498be7da
|
1513 |
|
a737b3e2f
|
1514 |
if (INDEX_AC != INDEX_L3) { |
e498be7da
|
1515 |
sizes[INDEX_L3].cs_cachep = |
a737b3e2f
|
1516 1517 1518 1519 |
kmem_cache_create(names[INDEX_L3].name, sizes[INDEX_L3].cs_size, ARCH_KMALLOC_MINALIGN, ARCH_KMALLOC_FLAGS|SLAB_PANIC, |
20c2df83d
|
1520 |
NULL); |
a737b3e2f
|
1521 |
} |
e498be7da
|
1522 |
|
e0a427267
|
1523 |
slab_early_init = 0; |
1da177e4c
|
1524 |
while (sizes->cs_size != ULONG_MAX) { |
e498be7da
|
1525 1526 |
/* * For performance, all the general caches are L1 aligned. |
1da177e4c
|
1527 1528 1529 |
* This should be particularly beneficial on SMP boxes, as it * eliminates "false sharing". * Note for systems short on memory removing the alignment will |
e498be7da
|
1530 1531 |
* allow tighter packing of the smaller caches. */ |
a737b3e2f
|
1532 |
if (!sizes->cs_cachep) { |
e498be7da
|
1533 |
sizes->cs_cachep = kmem_cache_create(names->name, |
a737b3e2f
|
1534 1535 1536 |
sizes->cs_size, ARCH_KMALLOC_MINALIGN, ARCH_KMALLOC_FLAGS|SLAB_PANIC, |
20c2df83d
|
1537 |
NULL); |
a737b3e2f
|
1538 |
} |
4b51d6698
|
1539 1540 1541 |
#ifdef CONFIG_ZONE_DMA sizes->cs_dmacachep = kmem_cache_create( names->name_dma, |
a737b3e2f
|
1542 1543 1544 1545 |
sizes->cs_size, ARCH_KMALLOC_MINALIGN, ARCH_KMALLOC_FLAGS|SLAB_CACHE_DMA| SLAB_PANIC, |
20c2df83d
|
1546 |
NULL); |
4b51d6698
|
1547 |
#endif |
1da177e4c
|
1548 1549 1550 1551 1552 |
sizes++; names++; } /* 4) Replace the bootstrap head arrays */ { |
2b2d5493e
|
1553 |
struct array_cache *ptr; |
e498be7da
|
1554 |
|
83b519e8b
|
1555 |
ptr = kmalloc(sizeof(struct arraycache_init), GFP_NOWAIT); |
e498be7da
|
1556 |
|
9a2dba4b4
|
1557 1558 |
BUG_ON(cpu_cache_get(&cache_cache) != &initarray_cache.cache); memcpy(ptr, cpu_cache_get(&cache_cache), |
b28a02de8
|
1559 |
sizeof(struct arraycache_init)); |
2b2d5493e
|
1560 1561 1562 1563 |
/* * Do not assume that spinlocks can be initialized via memcpy: */ spin_lock_init(&ptr->lock); |
1da177e4c
|
1564 |
cache_cache.array[smp_processor_id()] = ptr; |
e498be7da
|
1565 |
|
83b519e8b
|
1566 |
ptr = kmalloc(sizeof(struct arraycache_init), GFP_NOWAIT); |
e498be7da
|
1567 |
|
9a2dba4b4
|
1568 |
BUG_ON(cpu_cache_get(malloc_sizes[INDEX_AC].cs_cachep) |
b28a02de8
|
1569 |
!= &initarray_generic.cache); |
9a2dba4b4
|
1570 |
memcpy(ptr, cpu_cache_get(malloc_sizes[INDEX_AC].cs_cachep), |
b28a02de8
|
1571 |
sizeof(struct arraycache_init)); |
2b2d5493e
|
1572 1573 1574 1575 |
/* * Do not assume that spinlocks can be initialized via memcpy: */ spin_lock_init(&ptr->lock); |
e498be7da
|
1576 |
malloc_sizes[INDEX_AC].cs_cachep->array[smp_processor_id()] = |
b28a02de8
|
1577 |
ptr; |
1da177e4c
|
1578 |
} |
e498be7da
|
1579 1580 |
/* 5) Replace the bootstrap kmem_list3's */ { |
1ca4cb241
|
1581 |
int nid; |
9c09a95cf
|
1582 |
for_each_online_node(nid) { |
ec1f5eeeb
|
1583 |
init_list(&cache_cache, &initkmem_list3[CACHE_CACHE + nid], nid); |
556a169da
|
1584 |
|
e498be7da
|
1585 |
init_list(malloc_sizes[INDEX_AC].cs_cachep, |
1ca4cb241
|
1586 |
&initkmem_list3[SIZE_AC + nid], nid); |
e498be7da
|
1587 1588 1589 |
if (INDEX_AC != INDEX_L3) { init_list(malloc_sizes[INDEX_L3].cs_cachep, |
1ca4cb241
|
1590 |
&initkmem_list3[SIZE_L3 + nid], nid); |
e498be7da
|
1591 1592 1593 |
} } } |
1da177e4c
|
1594 |
|
8429db5c6
|
1595 |
g_cpucache_up = EARLY; |
8429db5c6
|
1596 1597 1598 1599 1600 |
} void __init kmem_cache_init_late(void) { struct kmem_cache *cachep; |
52cef1891
|
1601 |
g_cpucache_up = LATE; |
30765b92a
|
1602 1603 |
/* Annotate slab for lockdep -- annotate the malloc caches */ init_lock_keys(); |
8429db5c6
|
1604 1605 1606 1607 1608 1609 |
/* 6) resize the head arrays to their final sizes */ mutex_lock(&cache_chain_mutex); list_for_each_entry(cachep, &cache_chain, next) if (enable_cpucache(cachep, GFP_NOWAIT)) BUG(); mutex_unlock(&cache_chain_mutex); |
056c62418
|
1610 |
|
1da177e4c
|
1611 1612 |
/* Done! */ g_cpucache_up = FULL; |
a737b3e2f
|
1613 1614 1615 |
/* * Register a cpu startup notifier callback that initializes * cpu_cache_get for all new cpus |
1da177e4c
|
1616 1617 |
*/ register_cpu_notifier(&cpucache_notifier); |
1da177e4c
|
1618 |
|
8f9f8d9e8
|
1619 1620 1621 1622 1623 1624 1625 |
#ifdef CONFIG_NUMA /* * Register a memory hotplug callback that initializes and frees * nodelists. */ hotplug_memory_notifier(slab_memory_callback, SLAB_CALLBACK_PRI); #endif |
a737b3e2f
|
1626 1627 1628 |
/* * The reap timers are started later, with a module init call: That part * of the kernel is not yet operational. |
1da177e4c
|
1629 1630 1631 1632 1633 1634 |
*/ } static int __init cpucache_init(void) { int cpu; |
a737b3e2f
|
1635 1636 |
/* * Register the timers that return unneeded pages to the page allocator |
1da177e4c
|
1637 |
*/ |
e498be7da
|
1638 |
for_each_online_cpu(cpu) |
a737b3e2f
|
1639 |
start_cpu_timer(cpu); |
1da177e4c
|
1640 1641 |
return 0; } |
1da177e4c
|
1642 1643 1644 1645 1646 1647 1648 1649 1650 |
__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
|
1651 |
static void *kmem_getpages(struct kmem_cache *cachep, gfp_t flags, int nodeid) |
1da177e4c
|
1652 1653 |
{ struct page *page; |
e1b6aa6f1
|
1654 |
int nr_pages; |
1da177e4c
|
1655 |
int i; |
d6fef9da1
|
1656 |
#ifndef CONFIG_MMU |
e1b6aa6f1
|
1657 1658 1659 |
/* * Nommu uses slab's for process anonymous memory allocations, and thus * requires __GFP_COMP to properly refcount higher order allocations |
d6fef9da1
|
1660 |
*/ |
e1b6aa6f1
|
1661 |
flags |= __GFP_COMP; |
d6fef9da1
|
1662 |
#endif |
765c4507a
|
1663 |
|
3c517a613
|
1664 |
flags |= cachep->gfpflags; |
e12ba74d8
|
1665 1666 |
if (cachep->flags & SLAB_RECLAIM_ACCOUNT) flags |= __GFP_RECLAIMABLE; |
e1b6aa6f1
|
1667 |
|
517d08699
|
1668 |
page = alloc_pages_exact_node(nodeid, flags | __GFP_NOTRACK, cachep->gfporder); |
1da177e4c
|
1669 1670 |
if (!page) return NULL; |
1da177e4c
|
1671 |
|
e1b6aa6f1
|
1672 |
nr_pages = (1 << cachep->gfporder); |
1da177e4c
|
1673 |
if (cachep->flags & SLAB_RECLAIM_ACCOUNT) |
972d1a7b1
|
1674 1675 1676 1677 1678 |
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
|
1679 1680 |
for (i = 0; i < nr_pages; i++) __SetPageSlab(page + i); |
c175eea46
|
1681 |
|
b1eeab676
|
1682 1683 1684 1685 1686 1687 1688 1689 |
if (kmemcheck_enabled && !(cachep->flags & SLAB_NOTRACK)) { kmemcheck_alloc_shadow(page, cachep->gfporder, flags, nodeid); if (cachep->ctor) kmemcheck_mark_uninitialized_pages(page, nr_pages); else kmemcheck_mark_unallocated_pages(page, nr_pages); } |
c175eea46
|
1690 |
|
e1b6aa6f1
|
1691 |
return page_address(page); |
1da177e4c
|
1692 1693 1694 1695 1696 |
} /* * Interface to system's page release. */ |
343e0d7a9
|
1697 |
static void kmem_freepages(struct kmem_cache *cachep, void *addr) |
1da177e4c
|
1698 |
{ |
b28a02de8
|
1699 |
unsigned long i = (1 << cachep->gfporder); |
1da177e4c
|
1700 1701 |
struct page *page = virt_to_page(addr); const unsigned long nr_freed = i; |
b1eeab676
|
1702 |
kmemcheck_free_shadow(page, cachep->gfporder); |
c175eea46
|
1703 |
|
972d1a7b1
|
1704 1705 1706 1707 1708 1709 |
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
|
1710 |
while (i--) { |
f205b2fe6
|
1711 1712 |
BUG_ON(!PageSlab(page)); __ClearPageSlab(page); |
1da177e4c
|
1713 1714 |
page++; } |
1da177e4c
|
1715 1716 1717 |
if (current->reclaim_state) current->reclaim_state->reclaimed_slab += nr_freed; free_pages((unsigned long)addr, cachep->gfporder); |
1da177e4c
|
1718 1719 1720 1721 |
} static void kmem_rcu_free(struct rcu_head *head) { |
b28a02de8
|
1722 |
struct slab_rcu *slab_rcu = (struct slab_rcu *)head; |
343e0d7a9
|
1723 |
struct kmem_cache *cachep = slab_rcu->cachep; |
1da177e4c
|
1724 1725 1726 1727 1728 1729 1730 1731 1732 |
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
|
1733 |
static void store_stackinfo(struct kmem_cache *cachep, unsigned long *addr, |
b28a02de8
|
1734 |
unsigned long caller) |
1da177e4c
|
1735 |
{ |
3dafccf22
|
1736 |
int size = obj_size(cachep); |
1da177e4c
|
1737 |
|
3dafccf22
|
1738 |
addr = (unsigned long *)&((char *)addr)[obj_offset(cachep)]; |
1da177e4c
|
1739 |
|
b28a02de8
|
1740 |
if (size < 5 * sizeof(unsigned long)) |
1da177e4c
|
1741 |
return; |
b28a02de8
|
1742 1743 1744 1745 |
*addr++ = 0x12345678; *addr++ = caller; *addr++ = smp_processor_id(); size -= 3 * sizeof(unsigned long); |
1da177e4c
|
1746 1747 1748 1749 1750 1751 1752 |
{ unsigned long *sptr = &caller; unsigned long svalue; while (!kstack_end(sptr)) { svalue = *sptr++; if (kernel_text_address(svalue)) { |
b28a02de8
|
1753 |
*addr++ = svalue; |
1da177e4c
|
1754 1755 1756 1757 1758 1759 1760 |
size -= sizeof(unsigned long); if (size <= sizeof(unsigned long)) break; } } } |
b28a02de8
|
1761 |
*addr++ = 0x87654321; |
1da177e4c
|
1762 1763 |
} #endif |
343e0d7a9
|
1764 |
static void poison_obj(struct kmem_cache *cachep, void *addr, unsigned char val) |
1da177e4c
|
1765 |
{ |
3dafccf22
|
1766 1767 |
int size = obj_size(cachep); addr = &((char *)addr)[obj_offset(cachep)]; |
1da177e4c
|
1768 1769 |
memset(addr, val, size); |
b28a02de8
|
1770 |
*(unsigned char *)(addr + size - 1) = POISON_END; |
1da177e4c
|
1771 1772 1773 1774 1775 |
} static void dump_line(char *data, int offset, int limit) { int i; |
aa83aa40e
|
1776 1777 |
unsigned char error = 0; int bad_count = 0; |
fdde6abb3
|
1778 |
printk(KERN_ERR "%03x: ", offset); |
aa83aa40e
|
1779 1780 1781 1782 1783 |
for (i = 0; i < limit; i++) { if (data[offset + i] != POISON_FREE) { error = data[offset + i]; bad_count++; } |
aa83aa40e
|
1784 |
} |
fdde6abb3
|
1785 1786 |
print_hex_dump(KERN_CONT, "", 0, 16, 1, &data[offset], limit, 1); |
aa83aa40e
|
1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 |
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
|
1804 1805 1806 1807 |
} #endif #if DEBUG |
343e0d7a9
|
1808 |
static void print_objinfo(struct kmem_cache *cachep, void *objp, int lines) |
1da177e4c
|
1809 1810 1811 1812 1813 |
{ int i, size; char *realobj; if (cachep->flags & SLAB_RED_ZONE) { |
b46b8f19c
|
1814 1815 |
printk(KERN_ERR "Redzone: 0x%llx/0x%llx. ", |
a737b3e2f
|
1816 1817 |
*dbg_redzone1(cachep, objp), *dbg_redzone2(cachep, objp)); |
1da177e4c
|
1818 1819 1820 1821 |
} if (cachep->flags & SLAB_STORE_USER) { printk(KERN_ERR "Last user: [<%p>]", |
a737b3e2f
|
1822 |
*dbg_userword(cachep, objp)); |
1da177e4c
|
1823 |
print_symbol("(%s)", |
a737b3e2f
|
1824 |
(unsigned long)*dbg_userword(cachep, objp)); |
1da177e4c
|
1825 1826 1827 |
printk(" "); } |
3dafccf22
|
1828 1829 |
realobj = (char *)objp + obj_offset(cachep); size = obj_size(cachep); |
b28a02de8
|
1830 |
for (i = 0; i < size && lines; i += 16, lines--) { |
1da177e4c
|
1831 1832 |
int limit; limit = 16; |
b28a02de8
|
1833 1834 |
if (i + limit > size) limit = size - i; |
1da177e4c
|
1835 1836 1837 |
dump_line(realobj, i, limit); } } |
343e0d7a9
|
1838 |
static void check_poison_obj(struct kmem_cache *cachep, void *objp) |
1da177e4c
|
1839 1840 1841 1842 |
{ char *realobj; int size, i; int lines = 0; |
3dafccf22
|
1843 1844 |
realobj = (char *)objp + obj_offset(cachep); size = obj_size(cachep); |
1da177e4c
|
1845 |
|
b28a02de8
|
1846 |
for (i = 0; i < size; i++) { |
1da177e4c
|
1847 |
char exp = POISON_FREE; |
b28a02de8
|
1848 |
if (i == size - 1) |
1da177e4c
|
1849 1850 1851 1852 1853 1854 |
exp = POISON_END; if (realobj[i] != exp) { int limit; /* Mismatch ! */ /* Print header */ if (lines == 0) { |
b28a02de8
|
1855 |
printk(KERN_ERR |
e94a40c50
|
1856 1857 1858 |
"Slab corruption: %s start=%p, len=%d ", cachep->name, realobj, size); |
1da177e4c
|
1859 1860 1861 |
print_objinfo(cachep, objp, 0); } /* Hexdump the affected line */ |
b28a02de8
|
1862 |
i = (i / 16) * 16; |
1da177e4c
|
1863 |
limit = 16; |
b28a02de8
|
1864 1865 |
if (i + limit > size) limit = size - i; |
1da177e4c
|
1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 |
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
|
1878 |
struct slab *slabp = virt_to_slab(objp); |
8fea4e96a
|
1879 |
unsigned int objnr; |
1da177e4c
|
1880 |
|
8fea4e96a
|
1881 |
objnr = obj_to_index(cachep, slabp, objp); |
1da177e4c
|
1882 |
if (objnr) { |
8fea4e96a
|
1883 |
objp = index_to_obj(cachep, slabp, objnr - 1); |
3dafccf22
|
1884 |
realobj = (char *)objp + obj_offset(cachep); |
1da177e4c
|
1885 1886 |
printk(KERN_ERR "Prev obj: start=%p, len=%d ", |
b28a02de8
|
1887 |
realobj, size); |
1da177e4c
|
1888 1889 |
print_objinfo(cachep, objp, 2); } |
b28a02de8
|
1890 |
if (objnr + 1 < cachep->num) { |
8fea4e96a
|
1891 |
objp = index_to_obj(cachep, slabp, objnr + 1); |
3dafccf22
|
1892 |
realobj = (char *)objp + obj_offset(cachep); |
1da177e4c
|
1893 1894 |
printk(KERN_ERR "Next obj: start=%p, len=%d ", |
b28a02de8
|
1895 |
realobj, size); |
1da177e4c
|
1896 1897 1898 1899 1900 |
print_objinfo(cachep, objp, 2); } } } #endif |
12dd36fae
|
1901 |
#if DEBUG |
e79aec291
|
1902 |
static void slab_destroy_debugcheck(struct kmem_cache *cachep, struct slab *slabp) |
1da177e4c
|
1903 |
{ |
1da177e4c
|
1904 1905 |
int i; for (i = 0; i < cachep->num; i++) { |
8fea4e96a
|
1906 |
void *objp = index_to_obj(cachep, slabp, i); |
1da177e4c
|
1907 1908 1909 |
if (cachep->flags & SLAB_POISON) { #ifdef CONFIG_DEBUG_PAGEALLOC |
a737b3e2f
|
1910 1911 |
if (cachep->buffer_size % PAGE_SIZE == 0 && OFF_SLAB(cachep)) |
b28a02de8
|
1912 |
kernel_map_pages(virt_to_page(objp), |
a737b3e2f
|
1913 |
cachep->buffer_size / PAGE_SIZE, 1); |
1da177e4c
|
1914 1915 1916 1917 1918 1919 1920 1921 1922 |
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
|
1923 |
"was overwritten"); |
1da177e4c
|
1924 1925 |
if (*dbg_redzone2(cachep, objp) != RED_INACTIVE) slab_error(cachep, "end of a freed object " |
b28a02de8
|
1926 |
"was overwritten"); |
1da177e4c
|
1927 |
} |
1da177e4c
|
1928 |
} |
12dd36fae
|
1929 |
} |
1da177e4c
|
1930 |
#else |
e79aec291
|
1931 |
static void slab_destroy_debugcheck(struct kmem_cache *cachep, struct slab *slabp) |
12dd36fae
|
1932 |
{ |
12dd36fae
|
1933 |
} |
1da177e4c
|
1934 |
#endif |
911851e6e
|
1935 1936 1937 1938 1939 |
/** * slab_destroy - destroy and release all objects in a slab * @cachep: cache pointer being destroyed * @slabp: slab pointer being destroyed * |
12dd36fae
|
1940 |
* Destroy all the objs in a slab, and release the mem back to the system. |
a737b3e2f
|
1941 1942 |
* Before calling the slab must have been unlinked from the cache. The * cache-lock is not held/needed. |
12dd36fae
|
1943 |
*/ |
343e0d7a9
|
1944 |
static void slab_destroy(struct kmem_cache *cachep, struct slab *slabp) |
12dd36fae
|
1945 1946 |
{ void *addr = slabp->s_mem - slabp->colouroff; |
e79aec291
|
1947 |
slab_destroy_debugcheck(cachep, slabp); |
1da177e4c
|
1948 1949 |
if (unlikely(cachep->flags & SLAB_DESTROY_BY_RCU)) { struct slab_rcu *slab_rcu; |
b28a02de8
|
1950 |
slab_rcu = (struct slab_rcu *)slabp; |
1da177e4c
|
1951 1952 1953 1954 1955 |
slab_rcu->cachep = cachep; slab_rcu->addr = addr; call_rcu(&slab_rcu->head, kmem_rcu_free); } else { kmem_freepages(cachep, addr); |
873623dfa
|
1956 1957 |
if (OFF_SLAB(cachep)) kmem_cache_free(cachep->slabp_cache, slabp); |
1da177e4c
|
1958 1959 |
} } |
117f6eb1d
|
1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 |
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
|
1979 |
/** |
a70773ddb
|
1980 1981 1982 1983 1984 1985 1986 |
* 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
|
1987 1988 1989 1990 1991 |
* * 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
|
1992 |
static size_t calculate_slab_order(struct kmem_cache *cachep, |
ee13d785e
|
1993 |
size_t size, size_t align, unsigned long flags) |
4d268eba1
|
1994 |
{ |
b1ab41c49
|
1995 |
unsigned long offslab_limit; |
4d268eba1
|
1996 |
size_t left_over = 0; |
9888e6fa7
|
1997 |
int gfporder; |
4d268eba1
|
1998 |
|
0aa817f07
|
1999 |
for (gfporder = 0; gfporder <= KMALLOC_MAX_ORDER; gfporder++) { |
4d268eba1
|
2000 2001 |
unsigned int num; size_t remainder; |
9888e6fa7
|
2002 |
cache_estimate(gfporder, size, align, flags, &remainder, &num); |
4d268eba1
|
2003 2004 |
if (!num) continue; |
9888e6fa7
|
2005 |
|
b1ab41c49
|
2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 |
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
|
2018 |
|
9888e6fa7
|
2019 |
/* Found something acceptable - save it away */ |
4d268eba1
|
2020 |
cachep->num = num; |
9888e6fa7
|
2021 |
cachep->gfporder = gfporder; |
4d268eba1
|
2022 2023 2024 |
left_over = remainder; /* |
f78bb8ad4
|
2025 2026 2027 2028 2029 2030 2031 2032 |
* 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
|
2033 2034 2035 |
* Large number of objects is good, but very large slabs are * currently bad for the gfp()s. */ |
9888e6fa7
|
2036 |
if (gfporder >= slab_break_gfp_order) |
4d268eba1
|
2037 |
break; |
9888e6fa7
|
2038 2039 2040 |
/* * Acceptable internal fragmentation? */ |
a737b3e2f
|
2041 |
if (left_over * 8 <= (PAGE_SIZE << gfporder)) |
4d268eba1
|
2042 2043 2044 2045 |
break; } return left_over; } |
83b519e8b
|
2046 |
static int __init_refok setup_cpu_cache(struct kmem_cache *cachep, gfp_t gfp) |
f30cf7d13
|
2047 |
{ |
2ed3a4ef9
|
2048 |
if (g_cpucache_up == FULL) |
83b519e8b
|
2049 |
return enable_cpucache(cachep, gfp); |
2ed3a4ef9
|
2050 |
|
f30cf7d13
|
2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 |
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()] = |
83b519e8b
|
2071 |
kmalloc(sizeof(struct arraycache_init), gfp); |
f30cf7d13
|
2072 2073 2074 2075 2076 2077 |
if (g_cpucache_up == PARTIAL_AC) { set_up_list3s(cachep, SIZE_L3); g_cpucache_up = PARTIAL_L3; } else { int node; |
556a169da
|
2078 |
for_each_online_node(node) { |
f30cf7d13
|
2079 2080 |
cachep->nodelists[node] = kmalloc_node(sizeof(struct kmem_list3), |
eb91f1d0a
|
2081 |
gfp, node); |
f30cf7d13
|
2082 2083 2084 2085 2086 |
BUG_ON(!cachep->nodelists[node]); kmem_list3_init(cachep->nodelists[node]); } } } |
7d6e6d09d
|
2087 |
cachep->nodelists[numa_mem_id()]->next_reap = |
f30cf7d13
|
2088 2089 2090 2091 2092 2093 2094 2095 2096 |
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
|
2097 |
return 0; |
f30cf7d13
|
2098 |
} |
4d268eba1
|
2099 |
/** |
1da177e4c
|
2100 2101 2102 2103 2104 2105 |
* 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
|
2106 2107 2108 |
* * Returns a ptr to the cache on success, NULL on failure. * Cannot be called within a int, but can be interrupted. |
20c2df83d
|
2109 |
* The @ctor is run when new pages are allocated by the cache. |
1da177e4c
|
2110 2111 |
* * @name must be valid until the cache is destroyed. This implies that |
a737b3e2f
|
2112 2113 |
* the module calling this has to destroy the cache before getting unloaded. * |
1da177e4c
|
2114 2115 2116 2117 2118 2119 2120 2121 |
* 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
|
2122 2123 2124 2125 |
* %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
|
2126 |
struct kmem_cache * |
1da177e4c
|
2127 |
kmem_cache_create (const char *name, size_t size, size_t align, |
51cc50685
|
2128 |
unsigned long flags, void (*ctor)(void *)) |
1da177e4c
|
2129 2130 |
{ size_t left_over, slab_size, ralign; |
7a7c381d2
|
2131 |
struct kmem_cache *cachep = NULL, *pc; |
83b519e8b
|
2132 |
gfp_t gfp; |
1da177e4c
|
2133 2134 2135 2136 |
/* * Sanity checks... these are all serious usage bugs. */ |
a737b3e2f
|
2137 |
if (!name || in_interrupt() || (size < BYTES_PER_WORD) || |
20c2df83d
|
2138 |
size > KMALLOC_MAX_SIZE) { |
d40cee245
|
2139 2140 |
printk(KERN_ERR "%s: Early error in slab %s ", __func__, |
a737b3e2f
|
2141 |
name); |
b28a02de8
|
2142 2143 |
BUG(); } |
1da177e4c
|
2144 |
|
f0188f474
|
2145 |
/* |
8f5be20bf
|
2146 |
* We use cache_chain_mutex to ensure a consistent view of |
174596a0b
|
2147 |
* cpu_online_mask as well. Please see cpuup_callback |
f0188f474
|
2148 |
*/ |
83b519e8b
|
2149 2150 2151 2152 |
if (slab_is_available()) { get_online_cpus(); mutex_lock(&cache_chain_mutex); } |
4f12bb4f7
|
2153 |
|
7a7c381d2
|
2154 |
list_for_each_entry(pc, &cache_chain, next) { |
4f12bb4f7
|
2155 2156 2157 2158 2159 2160 2161 2162 |
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
|
2163 |
res = probe_kernel_address(pc->name, tmp); |
4f12bb4f7
|
2164 |
if (res) { |
b4169525b
|
2165 2166 2167 |
printk(KERN_ERR "SLAB: cache with size %d has lost its name ", |
3dafccf22
|
2168 |
pc->buffer_size); |
4f12bb4f7
|
2169 2170 |
continue; } |
b28a02de8
|
2171 |
if (!strcmp(pc->name, name)) { |
b4169525b
|
2172 2173 2174 |
printk(KERN_ERR "kmem_cache_create: duplicate cache %s ", name); |
4f12bb4f7
|
2175 2176 2177 2178 |
dump_stack(); goto oops; } } |
1da177e4c
|
2179 2180 |
#if DEBUG WARN_ON(strchr(name, ' ')); /* It confuses parsers */ |
1da177e4c
|
2181 2182 2183 2184 2185 2186 2187 |
#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
|
2188 2189 |
if (size < 4096 || fls(size - 1) == fls(size-1 + REDZONE_ALIGN + 2 * sizeof(unsigned long long))) |
b28a02de8
|
2190 |
flags |= SLAB_RED_ZONE | SLAB_STORE_USER; |
1da177e4c
|
2191 2192 2193 2194 2195 2196 |
if (!(flags & SLAB_DESTROY_BY_RCU)) flags |= SLAB_POISON; #endif if (flags & SLAB_DESTROY_BY_RCU) BUG_ON(flags & SLAB_POISON); #endif |
1da177e4c
|
2197 |
/* |
a737b3e2f
|
2198 2199 |
* Always checks flags, a caller might be expecting debug support which * isn't available. |
1da177e4c
|
2200 |
*/ |
40094fa65
|
2201 |
BUG_ON(flags & ~CREATE_MASK); |
1da177e4c
|
2202 |
|
a737b3e2f
|
2203 2204 |
/* * Check that size is in terms of words. This is needed to avoid |
1da177e4c
|
2205 2206 2207 |
* unaligned accesses for some archs when redzoning is used, and makes * sure any on-slab bufctl's are also correctly aligned. */ |
b28a02de8
|
2208 2209 2210 |
if (size & (BYTES_PER_WORD - 1)) { size += (BYTES_PER_WORD - 1); size &= ~(BYTES_PER_WORD - 1); |
1da177e4c
|
2211 |
} |
a737b3e2f
|
2212 |
/* calculate the final buffer alignment: */ |
1da177e4c
|
2213 2214 |
/* 1) arch recommendation: can be overridden for debug */ if (flags & SLAB_HWCACHE_ALIGN) { |
a737b3e2f
|
2215 2216 2217 2218 |
/* * Default alignment: as specified by the arch code. Except if * an object is really small, then squeeze multiple objects into * one cacheline. |
1da177e4c
|
2219 2220 |
*/ ralign = cache_line_size(); |
b28a02de8
|
2221 |
while (size <= ralign / 2) |
1da177e4c
|
2222 2223 2224 2225 |
ralign /= 2; } else { ralign = BYTES_PER_WORD; } |
ca5f9703d
|
2226 2227 |
/* |
87a927c71
|
2228 2229 2230 |
* 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
|
2231 |
*/ |
87a927c71
|
2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 |
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
|
2242 |
|
a44b56d35
|
2243 |
/* 2) arch mandated alignment */ |
1da177e4c
|
2244 2245 |
if (ralign < ARCH_SLAB_MINALIGN) { ralign = ARCH_SLAB_MINALIGN; |
1da177e4c
|
2246 |
} |
a44b56d35
|
2247 |
/* 3) caller mandated alignment */ |
1da177e4c
|
2248 2249 |
if (ralign < align) { ralign = align; |
1da177e4c
|
2250 |
} |
3ff84a7f3
|
2251 2252 |
/* disable debug if necessary */ if (ralign > __alignof__(unsigned long long)) |
a44b56d35
|
2253 |
flags &= ~(SLAB_RED_ZONE | SLAB_STORE_USER); |
a737b3e2f
|
2254 |
/* |
ca5f9703d
|
2255 |
* 4) Store it. |
1da177e4c
|
2256 2257 |
*/ align = ralign; |
83b519e8b
|
2258 2259 2260 2261 |
if (slab_is_available()) gfp = GFP_KERNEL; else gfp = GFP_NOWAIT; |
1da177e4c
|
2262 |
/* Get cache's description obj. */ |
83b519e8b
|
2263 |
cachep = kmem_cache_zalloc(&cache_cache, gfp); |
1da177e4c
|
2264 |
if (!cachep) |
4f12bb4f7
|
2265 |
goto oops; |
1da177e4c
|
2266 |
|
b56efcf0a
|
2267 |
cachep->nodelists = (struct kmem_list3 **)&cachep->array[nr_cpu_ids]; |
1da177e4c
|
2268 |
#if DEBUG |
3dafccf22
|
2269 |
cachep->obj_size = size; |
1da177e4c
|
2270 |
|
ca5f9703d
|
2271 2272 2273 2274 |
/* * Both debugging options require word-alignment which is calculated * into align above. */ |
1da177e4c
|
2275 |
if (flags & SLAB_RED_ZONE) { |
1da177e4c
|
2276 |
/* add space for red zone words */ |
3ff84a7f3
|
2277 2278 |
cachep->obj_offset += sizeof(unsigned long long); size += 2 * sizeof(unsigned long long); |
1da177e4c
|
2279 2280 |
} if (flags & SLAB_STORE_USER) { |
ca5f9703d
|
2281 |
/* user store requires one word storage behind the end of |
87a927c71
|
2282 2283 |
* the real object. But if the second red zone needs to be * aligned to 64 bits, we must allow that much space. |
1da177e4c
|
2284 |
*/ |
87a927c71
|
2285 2286 2287 2288 |
if (flags & SLAB_RED_ZONE) size += REDZONE_ALIGN; else size += BYTES_PER_WORD; |
1da177e4c
|
2289 2290 |
} #if FORCED_DEBUG && defined(CONFIG_DEBUG_PAGEALLOC) |
b28a02de8
|
2291 |
if (size >= malloc_sizes[INDEX_L3 + 1].cs_size |
1ab335d8f
|
2292 2293 |
&& cachep->obj_size > cache_line_size() && ALIGN(size, align) < PAGE_SIZE) { cachep->obj_offset += PAGE_SIZE - ALIGN(size, align); |
1da177e4c
|
2294 2295 2296 2297 |
size = PAGE_SIZE; } #endif #endif |
e0a427267
|
2298 2299 2300 |
/* * Determine if the slab management is 'on' or 'off' slab. * (bootstrapping cannot cope with offslab caches so don't do |
e7cb55b94
|
2301 2302 |
* it too early on. Always use on-slab management when * SLAB_NOLEAKTRACE to avoid recursive calls into kmemleak) |
e0a427267
|
2303 |
*/ |
e7cb55b94
|
2304 2305 |
if ((size >= (PAGE_SIZE >> 3)) && !slab_early_init && !(flags & SLAB_NOLEAKTRACE)) |
1da177e4c
|
2306 2307 2308 2309 2310 2311 2312 |
/* * 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
|
2313 |
left_over = calculate_slab_order(cachep, size, align, flags); |
1da177e4c
|
2314 2315 |
if (!cachep->num) { |
b4169525b
|
2316 2317 2318 |
printk(KERN_ERR "kmem_cache_create: couldn't create cache %s. ", name); |
1da177e4c
|
2319 2320 |
kmem_cache_free(&cache_cache, cachep); cachep = NULL; |
4f12bb4f7
|
2321 |
goto oops; |
1da177e4c
|
2322 |
} |
b28a02de8
|
2323 2324 |
slab_size = ALIGN(cachep->num * sizeof(kmem_bufctl_t) + sizeof(struct slab), align); |
1da177e4c
|
2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 |
/* * 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
|
2337 2338 |
slab_size = cachep->num * sizeof(kmem_bufctl_t) + sizeof(struct slab); |
674613652
|
2339 2340 2341 2342 2343 2344 2345 2346 2347 |
#ifdef CONFIG_PAGE_POISONING /* If we're going to use the generic kernel_map_pages() * poisoning, then it's going to smash the contents of * the redzone and userword anyhow, so switch them off. */ if (size % PAGE_SIZE == 0 && flags & SLAB_POISON) flags &= ~(SLAB_RED_ZONE | SLAB_STORE_USER); #endif |
1da177e4c
|
2348 2349 2350 2351 2352 2353 |
} cachep->colour_off = cache_line_size(); /* Offset must be a multiple of the alignment. */ if (cachep->colour_off < align) cachep->colour_off = align; |
b28a02de8
|
2354 |
cachep->colour = left_over / cachep->colour_off; |
1da177e4c
|
2355 2356 2357 |
cachep->slab_size = slab_size; cachep->flags = flags; cachep->gfpflags = 0; |
4b51d6698
|
2358 |
if (CONFIG_ZONE_DMA_FLAG && (flags & SLAB_CACHE_DMA)) |
1da177e4c
|
2359 |
cachep->gfpflags |= GFP_DMA; |
3dafccf22
|
2360 |
cachep->buffer_size = size; |
6a2d7a955
|
2361 |
cachep->reciprocal_buffer_size = reciprocal_value(size); |
1da177e4c
|
2362 |
|
e5ac9c5ae
|
2363 |
if (flags & CFLGS_OFF_SLAB) { |
b2d550736
|
2364 |
cachep->slabp_cache = kmem_find_general_cachep(slab_size, 0u); |
e5ac9c5ae
|
2365 2366 2367 2368 2369 2370 2371 |
/* * 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
|
2372 |
BUG_ON(ZERO_OR_NULL_PTR(cachep->slabp_cache)); |
e5ac9c5ae
|
2373 |
} |
1da177e4c
|
2374 |
cachep->ctor = ctor; |
1da177e4c
|
2375 |
cachep->name = name; |
83b519e8b
|
2376 |
if (setup_cpu_cache(cachep, gfp)) { |
2ed3a4ef9
|
2377 2378 2379 2380 |
__kmem_cache_destroy(cachep); cachep = NULL; goto oops; } |
1da177e4c
|
2381 |
|
83835b3d9
|
2382 2383 2384 2385 2386 2387 2388 2389 2390 |
if (flags & SLAB_DEBUG_OBJECTS) { /* * Would deadlock through slab_destroy()->call_rcu()-> * debug_object_activate()->kmem_cache_alloc(). */ WARN_ON_ONCE(flags & SLAB_DESTROY_BY_RCU); slab_set_debugobj_lock_classes(cachep); } |
1da177e4c
|
2391 2392 |
/* cache setup completed, link it into the list */ list_add(&cachep->next, &cache_chain); |
a737b3e2f
|
2393 |
oops: |
1da177e4c
|
2394 2395 2396 |
if (!cachep && (flags & SLAB_PANIC)) panic("kmem_cache_create(): failed to create slab `%s' ", |
b28a02de8
|
2397 |
name); |
83b519e8b
|
2398 2399 2400 2401 |
if (slab_is_available()) { mutex_unlock(&cache_chain_mutex); put_online_cpus(); } |
1da177e4c
|
2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 |
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
|
2416 |
static void check_spinlock_acquired(struct kmem_cache *cachep) |
1da177e4c
|
2417 2418 2419 |
{ #ifdef CONFIG_SMP check_irq_off(); |
7d6e6d09d
|
2420 |
assert_spin_locked(&cachep->nodelists[numa_mem_id()]->list_lock); |
1da177e4c
|
2421 2422 |
#endif } |
e498be7da
|
2423 |
|
343e0d7a9
|
2424 |
static void check_spinlock_acquired_node(struct kmem_cache *cachep, int node) |
e498be7da
|
2425 2426 2427 2428 2429 2430 |
{ #ifdef CONFIG_SMP check_irq_off(); assert_spin_locked(&cachep->nodelists[node]->list_lock); #endif } |
1da177e4c
|
2431 2432 2433 2434 |
#else #define check_irq_off() do { } while(0) #define check_irq_on() do { } while(0) #define check_spinlock_acquired(x) do { } while(0) |
e498be7da
|
2435 |
#define check_spinlock_acquired_node(x, y) do { } while(0) |
1da177e4c
|
2436 |
#endif |
aab2207cf
|
2437 2438 2439 |
static void drain_array(struct kmem_cache *cachep, struct kmem_list3 *l3, struct array_cache *ac, int force, int node); |
1da177e4c
|
2440 2441 |
static void do_drain(void *arg) { |
a737b3e2f
|
2442 |
struct kmem_cache *cachep = arg; |
1da177e4c
|
2443 |
struct array_cache *ac; |
7d6e6d09d
|
2444 |
int node = numa_mem_id(); |
1da177e4c
|
2445 2446 |
check_irq_off(); |
9a2dba4b4
|
2447 |
ac = cpu_cache_get(cachep); |
ff69416e6
|
2448 2449 2450 |
spin_lock(&cachep->nodelists[node]->list_lock); free_block(cachep, ac->entry, ac->avail, node); spin_unlock(&cachep->nodelists[node]->list_lock); |
1da177e4c
|
2451 2452 |
ac->avail = 0; } |
343e0d7a9
|
2453 |
static void drain_cpu_caches(struct kmem_cache *cachep) |
1da177e4c
|
2454 |
{ |
e498be7da
|
2455 2456 |
struct kmem_list3 *l3; int node; |
15c8b6c1a
|
2457 |
on_each_cpu(do_drain, cachep, 1); |
1da177e4c
|
2458 |
check_irq_on(); |
b28a02de8
|
2459 |
for_each_online_node(node) { |
e498be7da
|
2460 |
l3 = cachep->nodelists[node]; |
a4523a8b3
|
2461 2462 2463 2464 2465 2466 2467 |
if (l3 && l3->alien) drain_alien_cache(cachep, l3->alien); } for_each_online_node(node) { l3 = cachep->nodelists[node]; if (l3) |
aab2207cf
|
2468 |
drain_array(cachep, l3, l3->shared, 1, node); |
e498be7da
|
2469 |
} |
1da177e4c
|
2470 |
} |
ed11d9eb2
|
2471 2472 2473 2474 2475 2476 2477 2478 |
/* * 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
|
2479 |
{ |
ed11d9eb2
|
2480 2481 |
struct list_head *p; int nr_freed; |
1da177e4c
|
2482 |
struct slab *slabp; |
1da177e4c
|
2483 |
|
ed11d9eb2
|
2484 2485 |
nr_freed = 0; while (nr_freed < tofree && !list_empty(&l3->slabs_free)) { |
1da177e4c
|
2486 |
|
ed11d9eb2
|
2487 |
spin_lock_irq(&l3->list_lock); |
e498be7da
|
2488 |
p = l3->slabs_free.prev; |
ed11d9eb2
|
2489 2490 2491 2492 |
if (p == &l3->slabs_free) { spin_unlock_irq(&l3->list_lock); goto out; } |
1da177e4c
|
2493 |
|
ed11d9eb2
|
2494 |
slabp = list_entry(p, struct slab, list); |
1da177e4c
|
2495 |
#if DEBUG |
40094fa65
|
2496 |
BUG_ON(slabp->inuse); |
1da177e4c
|
2497 2498 |
#endif list_del(&slabp->list); |
ed11d9eb2
|
2499 2500 2501 2502 2503 |
/* * Safe to drop the lock. The slab is no longer linked * to the cache. */ l3->free_objects -= cache->num; |
e498be7da
|
2504 |
spin_unlock_irq(&l3->list_lock); |
ed11d9eb2
|
2505 2506 |
slab_destroy(cache, slabp); nr_freed++; |
1da177e4c
|
2507 |
} |
ed11d9eb2
|
2508 2509 |
out: return nr_freed; |
1da177e4c
|
2510 |
} |
8f5be20bf
|
2511 |
/* Called with cache_chain_mutex held to protect against cpu hotplug */ |
343e0d7a9
|
2512 |
static int __cache_shrink(struct kmem_cache *cachep) |
e498be7da
|
2513 2514 2515 2516 2517 2518 2519 2520 2521 |
{ 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
|
2522 2523 2524 2525 2526 2527 2528 |
if (!l3) continue; drain_freelist(cachep, l3, l3->free_objects); ret += !list_empty(&l3->slabs_full) || !list_empty(&l3->slabs_partial); |
e498be7da
|
2529 2530 2531 |
} return (ret ? 1 : 0); } |
1da177e4c
|
2532 2533 2534 2535 2536 2537 2538 |
/** * 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
|
2539 |
int kmem_cache_shrink(struct kmem_cache *cachep) |
1da177e4c
|
2540 |
{ |
8f5be20bf
|
2541 |
int ret; |
40094fa65
|
2542 |
BUG_ON(!cachep || in_interrupt()); |
1da177e4c
|
2543 |
|
95402b382
|
2544 |
get_online_cpus(); |
8f5be20bf
|
2545 2546 2547 |
mutex_lock(&cache_chain_mutex); ret = __cache_shrink(cachep); mutex_unlock(&cache_chain_mutex); |
95402b382
|
2548 |
put_online_cpus(); |
8f5be20bf
|
2549 |
return ret; |
1da177e4c
|
2550 2551 2552 2553 2554 2555 2556 |
} EXPORT_SYMBOL(kmem_cache_shrink); /** * kmem_cache_destroy - delete a cache * @cachep: the cache to destroy * |
72fd4a35a
|
2557 |
* Remove a &struct kmem_cache object from the slab cache. |
1da177e4c
|
2558 2559 2560 2561 2562 2563 2564 2565 |
* * 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. * |
25985edce
|
2566 |
* The caller must guarantee that no one will allocate memory from the cache |
1da177e4c
|
2567 2568 |
* during the kmem_cache_destroy(). */ |
133d205a1
|
2569 |
void kmem_cache_destroy(struct kmem_cache *cachep) |
1da177e4c
|
2570 |
{ |
40094fa65
|
2571 |
BUG_ON(!cachep || in_interrupt()); |
1da177e4c
|
2572 |
|
1da177e4c
|
2573 |
/* Find the cache in the chain of caches. */ |
95402b382
|
2574 |
get_online_cpus(); |
fc0abb145
|
2575 |
mutex_lock(&cache_chain_mutex); |
1da177e4c
|
2576 2577 2578 2579 |
/* * the chain is never empty, cache_cache is never destroyed */ list_del(&cachep->next); |
1da177e4c
|
2580 2581 |
if (__cache_shrink(cachep)) { slab_error(cachep, "Can't free all objects"); |
b28a02de8
|
2582 |
list_add(&cachep->next, &cache_chain); |
fc0abb145
|
2583 |
mutex_unlock(&cache_chain_mutex); |
95402b382
|
2584 |
put_online_cpus(); |
133d205a1
|
2585 |
return; |
1da177e4c
|
2586 2587 2588 |
} if (unlikely(cachep->flags & SLAB_DESTROY_BY_RCU)) |
7ed9f7e5d
|
2589 |
rcu_barrier(); |
1da177e4c
|
2590 |
|
117f6eb1d
|
2591 |
__kmem_cache_destroy(cachep); |
8f5be20bf
|
2592 |
mutex_unlock(&cache_chain_mutex); |
95402b382
|
2593 |
put_online_cpus(); |
1da177e4c
|
2594 2595 |
} EXPORT_SYMBOL(kmem_cache_destroy); |
e5ac9c5ae
|
2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 |
/* * 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
|
2607 |
static struct slab *alloc_slabmgmt(struct kmem_cache *cachep, void *objp, |
5b74ada7e
|
2608 2609 |
int colour_off, gfp_t local_flags, int nodeid) |
1da177e4c
|
2610 2611 |
{ struct slab *slabp; |
b28a02de8
|
2612 |
|
1da177e4c
|
2613 2614 |
if (OFF_SLAB(cachep)) { /* Slab management obj is off-slab. */ |
5b74ada7e
|
2615 |
slabp = kmem_cache_alloc_node(cachep->slabp_cache, |
8759ec50a
|
2616 |
local_flags, nodeid); |
d5cff6352
|
2617 2618 2619 2620 2621 2622 |
/* * If the first object in the slab is leaked (it's allocated * but no one has a reference to it), we want to make sure * kmemleak does not treat the ->s_mem pointer as a reference * to the object. Otherwise we will not report the leak. */ |
c017b4be3
|
2623 2624 |
kmemleak_scan_area(&slabp->list, sizeof(struct list_head), local_flags); |
1da177e4c
|
2625 2626 2627 |
if (!slabp) return NULL; } else { |
b28a02de8
|
2628 |
slabp = objp + colour_off; |
1da177e4c
|
2629 2630 2631 2632 |
colour_off += cachep->slab_size; } slabp->inuse = 0; slabp->colouroff = colour_off; |
b28a02de8
|
2633 |
slabp->s_mem = objp + colour_off; |
5b74ada7e
|
2634 |
slabp->nodeid = nodeid; |
e51bfd0ad
|
2635 |
slabp->free = 0; |
1da177e4c
|
2636 2637 2638 2639 2640 |
return slabp; } static inline kmem_bufctl_t *slab_bufctl(struct slab *slabp) { |
b28a02de8
|
2641 |
return (kmem_bufctl_t *) (slabp + 1); |
1da177e4c
|
2642 |
} |
343e0d7a9
|
2643 |
static void cache_init_objs(struct kmem_cache *cachep, |
a35afb830
|
2644 |
struct slab *slabp) |
1da177e4c
|
2645 2646 2647 2648 |
{ int i; for (i = 0; i < cachep->num; i++) { |
8fea4e96a
|
2649 |
void *objp = index_to_obj(cachep, slabp, i); |
1da177e4c
|
2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 |
#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
|
2662 2663 2664 |
* 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
|
2665 2666 |
*/ if (cachep->ctor && !(cachep->flags & SLAB_POISON)) |
51cc50685
|
2667 |
cachep->ctor(objp + obj_offset(cachep)); |
1da177e4c
|
2668 2669 2670 2671 |
if (cachep->flags & SLAB_RED_ZONE) { if (*dbg_redzone2(cachep, objp) != RED_INACTIVE) slab_error(cachep, "constructor overwrote the" |
b28a02de8
|
2672 |
" end of an object"); |
1da177e4c
|
2673 2674 |
if (*dbg_redzone1(cachep, objp) != RED_INACTIVE) slab_error(cachep, "constructor overwrote the" |
b28a02de8
|
2675 |
" start of an object"); |
1da177e4c
|
2676 |
} |
a737b3e2f
|
2677 2678 |
if ((cachep->buffer_size % PAGE_SIZE) == 0 && OFF_SLAB(cachep) && cachep->flags & SLAB_POISON) |
b28a02de8
|
2679 |
kernel_map_pages(virt_to_page(objp), |
3dafccf22
|
2680 |
cachep->buffer_size / PAGE_SIZE, 0); |
1da177e4c
|
2681 2682 |
#else if (cachep->ctor) |
51cc50685
|
2683 |
cachep->ctor(objp); |
1da177e4c
|
2684 |
#endif |
b28a02de8
|
2685 |
slab_bufctl(slabp)[i] = i + 1; |
1da177e4c
|
2686 |
} |
b28a02de8
|
2687 |
slab_bufctl(slabp)[i - 1] = BUFCTL_END; |
1da177e4c
|
2688 |
} |
343e0d7a9
|
2689 |
static void kmem_flagcheck(struct kmem_cache *cachep, gfp_t flags) |
1da177e4c
|
2690 |
{ |
4b51d6698
|
2691 2692 2693 2694 2695 2696 |
if (CONFIG_ZONE_DMA_FLAG) { if (flags & GFP_DMA) BUG_ON(!(cachep->gfpflags & GFP_DMA)); else BUG_ON(cachep->gfpflags & GFP_DMA); } |
1da177e4c
|
2697 |
} |
a737b3e2f
|
2698 2699 |
static void *slab_get_obj(struct kmem_cache *cachep, struct slab *slabp, int nodeid) |
78d382d77
|
2700 |
{ |
8fea4e96a
|
2701 |
void *objp = index_to_obj(cachep, slabp, slabp->free); |
78d382d77
|
2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 |
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
|
2714 2715 |
static void slab_put_obj(struct kmem_cache *cachep, struct slab *slabp, void *objp, int nodeid) |
78d382d77
|
2716 |
{ |
8fea4e96a
|
2717 |
unsigned int objnr = obj_to_index(cachep, slabp, objp); |
78d382d77
|
2718 2719 2720 2721 |
#if DEBUG /* Verify that the slab belongs to the intended node */ WARN_ON(slabp->nodeid != nodeid); |
871751e25
|
2722 |
if (slab_bufctl(slabp)[objnr] + 1 <= SLAB_LIMIT + 1) { |
78d382d77
|
2723 |
printk(KERN_ERR "slab: double free detected in cache " |
a737b3e2f
|
2724 2725 |
"'%s', objp %p ", cachep->name, objp); |
78d382d77
|
2726 2727 2728 2729 2730 2731 2732 |
BUG(); } #endif slab_bufctl(slabp)[objnr] = slabp->free; slabp->free = objnr; slabp->inuse--; } |
4776874ff
|
2733 2734 2735 |
/* * 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 |
ccd35fb9f
|
2736 |
* virtual address for kfree, ksize, and slab debugging. |
4776874ff
|
2737 2738 2739 |
*/ static void slab_map_pages(struct kmem_cache *cache, struct slab *slab, void *addr) |
1da177e4c
|
2740 |
{ |
4776874ff
|
2741 |
int nr_pages; |
1da177e4c
|
2742 |
struct page *page; |
4776874ff
|
2743 |
page = virt_to_page(addr); |
84097518d
|
2744 |
|
4776874ff
|
2745 |
nr_pages = 1; |
84097518d
|
2746 |
if (likely(!PageCompound(page))) |
4776874ff
|
2747 |
nr_pages <<= cache->gfporder; |
1da177e4c
|
2748 |
do { |
4776874ff
|
2749 2750 |
page_set_cache(page, cache); page_set_slab(page, slab); |
1da177e4c
|
2751 |
page++; |
4776874ff
|
2752 |
} while (--nr_pages); |
1da177e4c
|
2753 2754 2755 2756 2757 2758 |
} /* * 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
|
2759 2760 |
static int cache_grow(struct kmem_cache *cachep, gfp_t flags, int nodeid, void *objp) |
1da177e4c
|
2761 |
{ |
b28a02de8
|
2762 |
struct slab *slabp; |
b28a02de8
|
2763 2764 |
size_t offset; gfp_t local_flags; |
e498be7da
|
2765 |
struct kmem_list3 *l3; |
1da177e4c
|
2766 |
|
a737b3e2f
|
2767 2768 2769 |
/* * Be lazy and only check for valid flags here, keeping it out of the * critical path in kmem_cache_alloc(). |
1da177e4c
|
2770 |
*/ |
6cb062296
|
2771 2772 |
BUG_ON(flags & GFP_SLAB_BUG_MASK); local_flags = flags & (GFP_CONSTRAINT_MASK|GFP_RECLAIM_MASK); |
1da177e4c
|
2773 |
|
2e1217cf9
|
2774 |
/* Take the l3 list lock to change the colour_next on this node */ |
1da177e4c
|
2775 |
check_irq_off(); |
2e1217cf9
|
2776 2777 |
l3 = cachep->nodelists[nodeid]; spin_lock(&l3->list_lock); |
1da177e4c
|
2778 2779 |
/* Get colour for the slab, and cal the next value. */ |
2e1217cf9
|
2780 2781 2782 2783 2784 |
offset = l3->colour_next; l3->colour_next++; if (l3->colour_next >= cachep->colour) l3->colour_next = 0; spin_unlock(&l3->list_lock); |
1da177e4c
|
2785 |
|
2e1217cf9
|
2786 |
offset *= cachep->colour_off; |
1da177e4c
|
2787 2788 2789 2790 2791 2792 2793 2794 2795 2796 2797 |
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
|
2798 2799 2800 |
/* * Get mem for the objs. Attempt to allocate a physical page from * 'nodeid'. |
e498be7da
|
2801 |
*/ |
3c517a613
|
2802 |
if (!objp) |
b8c1c5da1
|
2803 |
objp = kmem_getpages(cachep, local_flags, nodeid); |
a737b3e2f
|
2804 |
if (!objp) |
1da177e4c
|
2805 2806 2807 |
goto failed; /* Get slab management. */ |
3c517a613
|
2808 |
slabp = alloc_slabmgmt(cachep, objp, offset, |
6cb062296
|
2809 |
local_flags & ~GFP_CONSTRAINT_MASK, nodeid); |
a737b3e2f
|
2810 |
if (!slabp) |
1da177e4c
|
2811 |
goto opps1; |
4776874ff
|
2812 |
slab_map_pages(cachep, slabp, objp); |
1da177e4c
|
2813 |
|
a35afb830
|
2814 |
cache_init_objs(cachep, slabp); |
1da177e4c
|
2815 2816 2817 2818 |
if (local_flags & __GFP_WAIT) local_irq_disable(); check_irq_off(); |
e498be7da
|
2819 |
spin_lock(&l3->list_lock); |
1da177e4c
|
2820 2821 |
/* Make slab active. */ |
e498be7da
|
2822 |
list_add_tail(&slabp->list, &(l3->slabs_free)); |
1da177e4c
|
2823 |
STATS_INC_GROWN(cachep); |
e498be7da
|
2824 2825 |
l3->free_objects += cachep->num; spin_unlock(&l3->list_lock); |
1da177e4c
|
2826 |
return 1; |
a737b3e2f
|
2827 |
opps1: |
1da177e4c
|
2828 |
kmem_freepages(cachep, objp); |
a737b3e2f
|
2829 |
failed: |
1da177e4c
|
2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 |
if (local_flags & __GFP_WAIT) local_irq_disable(); return 0; } #if DEBUG /* * Perform extra freeing checks: * - detect bad pointers. * - POISON/RED_ZONE checking |
1da177e4c
|
2841 2842 2843 |
*/ static void kfree_debugcheck(const void *objp) { |
1da177e4c
|
2844 2845 2846 |
if (!virt_addr_valid(objp)) { printk(KERN_ERR "kfree_debugcheck: out of range ptr %lxh. ", |
b28a02de8
|
2847 2848 |
(unsigned long)objp); BUG(); |
1da177e4c
|
2849 |
} |
1da177e4c
|
2850 |
} |
58ce1fd58
|
2851 2852 |
static inline void verify_redzone_free(struct kmem_cache *cache, void *obj) { |
b46b8f19c
|
2853 |
unsigned long long redzone1, redzone2; |
58ce1fd58
|
2854 2855 2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 |
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
|
2868 2869 |
printk(KERN_ERR "%p: redzone 1:0x%llx, redzone 2:0x%llx. ", |
58ce1fd58
|
2870 2871 |
obj, redzone1, redzone2); } |
343e0d7a9
|
2872 |
static void *cache_free_debugcheck(struct kmem_cache *cachep, void *objp, |
b28a02de8
|
2873 |
void *caller) |
1da177e4c
|
2874 2875 2876 2877 |
{ struct page *page; unsigned int objnr; struct slab *slabp; |
80cbd911c
|
2878 |
BUG_ON(virt_to_cache(objp) != cachep); |
3dafccf22
|
2879 |
objp -= obj_offset(cachep); |
1da177e4c
|
2880 |
kfree_debugcheck(objp); |
b49af68ff
|
2881 |
page = virt_to_head_page(objp); |
1da177e4c
|
2882 |
|
065d41cb2
|
2883 |
slabp = page_get_slab(page); |
1da177e4c
|
2884 2885 |
if (cachep->flags & SLAB_RED_ZONE) { |
58ce1fd58
|
2886 |
verify_redzone_free(cachep, objp); |
1da177e4c
|
2887 2888 2889 2890 2891 |
*dbg_redzone1(cachep, objp) = RED_INACTIVE; *dbg_redzone2(cachep, objp) = RED_INACTIVE; } if (cachep->flags & SLAB_STORE_USER) *dbg_userword(cachep, objp) = caller; |
8fea4e96a
|
2892 |
objnr = obj_to_index(cachep, slabp, objp); |
1da177e4c
|
2893 2894 |
BUG_ON(objnr >= cachep->num); |
8fea4e96a
|
2895 |
BUG_ON(objp != index_to_obj(cachep, slabp, objnr)); |
1da177e4c
|
2896 |
|
871751e25
|
2897 2898 2899 |
#ifdef CONFIG_DEBUG_SLAB_LEAK slab_bufctl(slabp)[objnr] = BUFCTL_FREE; #endif |
1da177e4c
|
2900 2901 |
if (cachep->flags & SLAB_POISON) { #ifdef CONFIG_DEBUG_PAGEALLOC |
a737b3e2f
|
2902 |
if ((cachep->buffer_size % PAGE_SIZE)==0 && OFF_SLAB(cachep)) { |
1da177e4c
|
2903 |
store_stackinfo(cachep, objp, (unsigned long)caller); |
b28a02de8
|
2904 |
kernel_map_pages(virt_to_page(objp), |
3dafccf22
|
2905 |
cachep->buffer_size / PAGE_SIZE, 0); |
1da177e4c
|
2906 2907 2908 2909 2910 2911 2912 2913 2914 |
} else { poison_obj(cachep, objp, POISON_FREE); } #else poison_obj(cachep, objp, POISON_FREE); #endif } return objp; } |
343e0d7a9
|
2915 |
static void check_slabp(struct kmem_cache *cachep, struct slab *slabp) |
1da177e4c
|
2916 2917 2918 |
{ kmem_bufctl_t i; int entries = 0; |
b28a02de8
|
2919 |
|
1da177e4c
|
2920 2921 2922 2923 2924 2925 2926 |
/* 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
|
2927 2928 2929 2930 2931 |
bad: printk(KERN_ERR "slab: Internal list corruption detected in " "cache '%s'(%d), slabp %p(%d). Hexdump: ", cachep->name, cachep->num, slabp, slabp->inuse); |
fdde6abb3
|
2932 2933 2934 |
print_hex_dump(KERN_ERR, "", DUMP_PREFIX_OFFSET, 16, 1, slabp, sizeof(*slabp) + cachep->num * sizeof(kmem_bufctl_t), 1); |
1da177e4c
|
2935 2936 2937 2938 2939 2940 2941 2942 |
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
|
2943 |
static void *cache_alloc_refill(struct kmem_cache *cachep, gfp_t flags) |
1da177e4c
|
2944 2945 2946 2947 |
{ int batchcount; struct kmem_list3 *l3; struct array_cache *ac; |
1ca4cb241
|
2948 |
int node; |
6d2144d35
|
2949 |
retry: |
1da177e4c
|
2950 |
check_irq_off(); |
7d6e6d09d
|
2951 |
node = numa_mem_id(); |
9a2dba4b4
|
2952 |
ac = cpu_cache_get(cachep); |
1da177e4c
|
2953 2954 |
batchcount = ac->batchcount; if (!ac->touched && batchcount > BATCHREFILL_LIMIT) { |
a737b3e2f
|
2955 2956 2957 2958 |
/* * If there was little recent activity on this cache, then * perform only a partial refill. Otherwise we could generate * refill bouncing. |
1da177e4c
|
2959 2960 2961 |
*/ batchcount = BATCHREFILL_LIMIT; } |
1ca4cb241
|
2962 |
l3 = cachep->nodelists[node]; |
e498be7da
|
2963 2964 2965 |
BUG_ON(ac->avail > 0 || !l3); spin_lock(&l3->list_lock); |
1da177e4c
|
2966 |
|
3ded175a4
|
2967 |
/* See if we can refill from the shared array */ |
44b57f1cc
|
2968 2969 |
if (l3->shared && transfer_objects(ac, l3->shared, batchcount)) { l3->shared->touched = 1; |
3ded175a4
|
2970 |
goto alloc_done; |
44b57f1cc
|
2971 |
} |
3ded175a4
|
2972 |
|
1da177e4c
|
2973 2974 2975 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 |
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
|
2988 2989 2990 2991 2992 2993 |
/* * The slab was either on partial or free list so * there must be at least one object available for * allocation. */ |
249b9f331
|
2994 |
BUG_ON(slabp->inuse >= cachep->num); |
714b8171a
|
2995 |
|
1da177e4c
|
2996 |
while (slabp->inuse < cachep->num && batchcount--) { |
1da177e4c
|
2997 2998 2999 |
STATS_INC_ALLOCED(cachep); STATS_INC_ACTIVE(cachep); STATS_SET_HIGH(cachep); |
78d382d77
|
3000 |
ac->entry[ac->avail++] = slab_get_obj(cachep, slabp, |
1ca4cb241
|
3001 |
node); |
1da177e4c
|
3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 |
} 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
|
3012 |
must_grow: |
1da177e4c
|
3013 |
l3->free_objects -= ac->avail; |
a737b3e2f
|
3014 |
alloc_done: |
e498be7da
|
3015 |
spin_unlock(&l3->list_lock); |
1da177e4c
|
3016 3017 3018 |
if (unlikely(!ac->avail)) { int x; |
3c517a613
|
3019 |
x = cache_grow(cachep, flags | GFP_THISNODE, node, NULL); |
e498be7da
|
3020 |
|
a737b3e2f
|
3021 |
/* cache_grow can reenable interrupts, then ac could change. */ |
9a2dba4b4
|
3022 |
ac = cpu_cache_get(cachep); |
a737b3e2f
|
3023 |
if (!x && ac->avail == 0) /* no objects in sight? abort */ |
1da177e4c
|
3024 |
return NULL; |
a737b3e2f
|
3025 |
if (!ac->avail) /* objects refilled by interrupt? */ |
1da177e4c
|
3026 3027 3028 |
goto retry; } ac->touched = 1; |
e498be7da
|
3029 |
return ac->entry[--ac->avail]; |
1da177e4c
|
3030 |
} |
a737b3e2f
|
3031 3032 |
static inline void cache_alloc_debugcheck_before(struct kmem_cache *cachep, gfp_t flags) |
1da177e4c
|
3033 3034 3035 3036 3037 3038 3039 3040 |
{ might_sleep_if(flags & __GFP_WAIT); #if DEBUG kmem_flagcheck(cachep, flags); #endif } #if DEBUG |
a737b3e2f
|
3041 3042 |
static void *cache_alloc_debugcheck_after(struct kmem_cache *cachep, gfp_t flags, void *objp, void *caller) |
1da177e4c
|
3043 |
{ |
b28a02de8
|
3044 |
if (!objp) |
1da177e4c
|
3045 |
return objp; |
b28a02de8
|
3046 |
if (cachep->flags & SLAB_POISON) { |
1da177e4c
|
3047 |
#ifdef CONFIG_DEBUG_PAGEALLOC |
3dafccf22
|
3048 |
if ((cachep->buffer_size % PAGE_SIZE) == 0 && OFF_SLAB(cachep)) |
b28a02de8
|
3049 |
kernel_map_pages(virt_to_page(objp), |
3dafccf22
|
3050 |
cachep->buffer_size / PAGE_SIZE, 1); |
1da177e4c
|
3051 3052 3053 3054 3055 3056 3057 3058 3059 3060 3061 |
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
|
3062 3063 3064 3065 |
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
|
3066 |
printk(KERN_ERR |
b46b8f19c
|
3067 3068 |
"%p: redzone 1:0x%llx, redzone 2:0x%llx ", |
a737b3e2f
|
3069 3070 |
objp, *dbg_redzone1(cachep, objp), *dbg_redzone2(cachep, objp)); |
1da177e4c
|
3071 3072 3073 3074 |
} *dbg_redzone1(cachep, objp) = RED_ACTIVE; *dbg_redzone2(cachep, objp) = RED_ACTIVE; } |
871751e25
|
3075 3076 3077 3078 |
#ifdef CONFIG_DEBUG_SLAB_LEAK { struct slab *slabp; unsigned objnr; |
b49af68ff
|
3079 |
slabp = page_get_slab(virt_to_head_page(objp)); |
871751e25
|
3080 3081 3082 3083 |
objnr = (unsigned)(objp - slabp->s_mem) / cachep->buffer_size; slab_bufctl(slabp)[objnr] = BUFCTL_ACTIVE; } #endif |
3dafccf22
|
3084 |
objp += obj_offset(cachep); |
4f1049345
|
3085 |
if (cachep->ctor && cachep->flags & SLAB_POISON) |
51cc50685
|
3086 |
cachep->ctor(objp); |
7ea466f22
|
3087 3088 |
if (ARCH_SLAB_MINALIGN && ((unsigned long)objp & (ARCH_SLAB_MINALIGN-1))) { |
a44b56d35
|
3089 3090 |
printk(KERN_ERR "0x%p: not aligned to ARCH_SLAB_MINALIGN=%d ", |
c225150b8
|
3091 |
objp, (int)ARCH_SLAB_MINALIGN); |
a44b56d35
|
3092 |
} |
1da177e4c
|
3093 3094 3095 3096 3097 |
return objp; } #else #define cache_alloc_debugcheck_after(a,b,objp,d) (objp) #endif |
773ff60e8
|
3098 |
static bool slab_should_failslab(struct kmem_cache *cachep, gfp_t flags) |
8a8b6502f
|
3099 3100 |
{ if (cachep == &cache_cache) |
773ff60e8
|
3101 |
return false; |
8a8b6502f
|
3102 |
|
4c13dd3b4
|
3103 |
return should_failslab(obj_size(cachep), flags, cachep->flags); |
8a8b6502f
|
3104 |
} |
343e0d7a9
|
3105 |
static inline void *____cache_alloc(struct kmem_cache *cachep, gfp_t flags) |
1da177e4c
|
3106 |
{ |
b28a02de8
|
3107 |
void *objp; |
1da177e4c
|
3108 |
struct array_cache *ac; |
5c3823008
|
3109 |
check_irq_off(); |
8a8b6502f
|
3110 |
|
9a2dba4b4
|
3111 |
ac = cpu_cache_get(cachep); |
1da177e4c
|
3112 3113 3114 |
if (likely(ac->avail)) { STATS_INC_ALLOCHIT(cachep); ac->touched = 1; |
e498be7da
|
3115 |
objp = ac->entry[--ac->avail]; |
1da177e4c
|
3116 3117 3118 |
} else { STATS_INC_ALLOCMISS(cachep); objp = cache_alloc_refill(cachep, flags); |
ddbf2e836
|
3119 3120 3121 3122 3123 |
/* * the 'ac' may be updated by cache_alloc_refill(), * and kmemleak_erase() requires its correct value. */ ac = cpu_cache_get(cachep); |
1da177e4c
|
3124 |
} |
d5cff6352
|
3125 3126 3127 3128 3129 |
/* * To avoid a false negative, if an object that is in one of the * per-CPU caches is leaked, we need to make sure kmemleak doesn't * treat the array pointers as a reference to the object. */ |
f3d8b53a3
|
3130 3131 |
if (objp) kmemleak_erase(&ac->entry[ac->avail]); |
5c3823008
|
3132 3133 |
return objp; } |
e498be7da
|
3134 3135 |
#ifdef CONFIG_NUMA /* |
b2455396b
|
3136 |
* Try allocating on another node if PF_SPREAD_SLAB|PF_MEMPOLICY. |
c61afb181
|
3137 3138 3139 3140 3141 3142 3143 |
* * 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
|
3144 |
if (in_interrupt() || (flags & __GFP_THISNODE)) |
c61afb181
|
3145 |
return NULL; |
7d6e6d09d
|
3146 |
nid_alloc = nid_here = numa_mem_id(); |
c0ff7453b
|
3147 |
get_mems_allowed(); |
c61afb181
|
3148 |
if (cpuset_do_slab_mem_spread() && (cachep->flags & SLAB_MEM_SPREAD)) |
6adef3ebe
|
3149 |
nid_alloc = cpuset_slab_spread_node(); |
c61afb181
|
3150 3151 |
else if (current->mempolicy) nid_alloc = slab_node(current->mempolicy); |
c0ff7453b
|
3152 |
put_mems_allowed(); |
c61afb181
|
3153 |
if (nid_alloc != nid_here) |
8b98c1699
|
3154 |
return ____cache_alloc_node(cachep, flags, nid_alloc); |
c61afb181
|
3155 3156 3157 3158 |
return NULL; } /* |
765c4507a
|
3159 |
* Fallback function if there was no memory available and no objects on a |
3c517a613
|
3160 3161 3162 3163 3164 |
* 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
|
3165 |
*/ |
8c8cc2c10
|
3166 |
static void *fallback_alloc(struct kmem_cache *cache, gfp_t flags) |
765c4507a
|
3167 |
{ |
8c8cc2c10
|
3168 3169 |
struct zonelist *zonelist; gfp_t local_flags; |
dd1a239f6
|
3170 |
struct zoneref *z; |
54a6eb5c4
|
3171 3172 |
struct zone *zone; enum zone_type high_zoneidx = gfp_zone(flags); |
765c4507a
|
3173 |
void *obj = NULL; |
3c517a613
|
3174 |
int nid; |
8c8cc2c10
|
3175 3176 3177 |
if (flags & __GFP_THISNODE) return NULL; |
c0ff7453b
|
3178 |
get_mems_allowed(); |
0e88460da
|
3179 |
zonelist = node_zonelist(slab_node(current->mempolicy), flags); |
6cb062296
|
3180 |
local_flags = flags & (GFP_CONSTRAINT_MASK|GFP_RECLAIM_MASK); |
765c4507a
|
3181 |
|
3c517a613
|
3182 3183 3184 3185 3186 |
retry: /* * Look through allowed nodes for objects available * from existing per node queues. */ |
54a6eb5c4
|
3187 3188 |
for_each_zone_zonelist(zone, z, zonelist, high_zoneidx) { nid = zone_to_nid(zone); |
aedb0eb10
|
3189 |
|
54a6eb5c4
|
3190 |
if (cpuset_zone_allowed_hardwall(zone, flags) && |
3c517a613
|
3191 |
cache->nodelists[nid] && |
481c5346d
|
3192 |
cache->nodelists[nid]->free_objects) { |
3c517a613
|
3193 3194 |
obj = ____cache_alloc_node(cache, flags | GFP_THISNODE, nid); |
481c5346d
|
3195 3196 3197 |
if (obj) break; } |
3c517a613
|
3198 |
} |
cfce66047
|
3199 |
if (!obj) { |
3c517a613
|
3200 3201 3202 3203 3204 3205 |
/* * 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
|
3206 3207 3208 |
if (local_flags & __GFP_WAIT) local_irq_enable(); kmem_flagcheck(cache, flags); |
7d6e6d09d
|
3209 |
obj = kmem_getpages(cache, local_flags, numa_mem_id()); |
dd47ea755
|
3210 3211 |
if (local_flags & __GFP_WAIT) local_irq_disable(); |
3c517a613
|
3212 3213 3214 3215 3216 3217 3218 3219 3220 3221 3222 3223 3224 3225 3226 3227 |
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
|
3228 |
/* cache_grow already freed obj */ |
3c517a613
|
3229 3230 3231 |
obj = NULL; } } |
aedb0eb10
|
3232 |
} |
c0ff7453b
|
3233 |
put_mems_allowed(); |
765c4507a
|
3234 3235 3236 3237 |
return obj; } /* |
e498be7da
|
3238 |
* A interface to enable slab creation on nodeid |
1da177e4c
|
3239 |
*/ |
8b98c1699
|
3240 |
static void *____cache_alloc_node(struct kmem_cache *cachep, gfp_t flags, |
a737b3e2f
|
3241 |
int nodeid) |
e498be7da
|
3242 3243 |
{ struct list_head *entry; |
b28a02de8
|
3244 3245 3246 |
struct slab *slabp; struct kmem_list3 *l3; void *obj; |
b28a02de8
|
3247 3248 3249 3250 |
int x; l3 = cachep->nodelists[nodeid]; BUG_ON(!l3); |
a737b3e2f
|
3251 |
retry: |
ca3b9b917
|
3252 |
check_irq_off(); |
b28a02de8
|
3253 3254 3255 3256 3257 3258 3259 3260 3261 3262 3263 3264 3265 3266 3267 3268 3269 3270 |
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
|
3271 |
obj = slab_get_obj(cachep, slabp, nodeid); |
b28a02de8
|
3272 3273 3274 3275 |
check_slabp(cachep, slabp); l3->free_objects--; /* move slabp to correct slabp list: */ list_del(&slabp->list); |
a737b3e2f
|
3276 |
if (slabp->free == BUFCTL_END) |
b28a02de8
|
3277 |
list_add(&slabp->list, &l3->slabs_full); |
a737b3e2f
|
3278 |
else |
b28a02de8
|
3279 |
list_add(&slabp->list, &l3->slabs_partial); |
e498be7da
|
3280 |
|
b28a02de8
|
3281 3282 |
spin_unlock(&l3->list_lock); goto done; |
e498be7da
|
3283 |
|
a737b3e2f
|
3284 |
must_grow: |
b28a02de8
|
3285 |
spin_unlock(&l3->list_lock); |
3c517a613
|
3286 |
x = cache_grow(cachep, flags | GFP_THISNODE, nodeid, NULL); |
765c4507a
|
3287 3288 |
if (x) goto retry; |
1da177e4c
|
3289 |
|
8c8cc2c10
|
3290 |
return fallback_alloc(cachep, flags); |
e498be7da
|
3291 |
|
a737b3e2f
|
3292 |
done: |
b28a02de8
|
3293 |
return obj; |
e498be7da
|
3294 |
} |
8c8cc2c10
|
3295 3296 3297 3298 3299 3300 3301 3302 3303 3304 3305 3306 3307 3308 3309 3310 3311 3312 3313 |
/** * 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; |
7d6e6d09d
|
3314 |
int slab_node = numa_mem_id(); |
8c8cc2c10
|
3315 |
|
dcce284a2
|
3316 |
flags &= gfp_allowed_mask; |
7e85ee0c1
|
3317 |
|
cf40bd16f
|
3318 |
lockdep_trace_alloc(flags); |
773ff60e8
|
3319 |
if (slab_should_failslab(cachep, flags)) |
824ebef12
|
3320 |
return NULL; |
8c8cc2c10
|
3321 3322 |
cache_alloc_debugcheck_before(cachep, flags); local_irq_save(save_flags); |
eacbbae38
|
3323 |
if (nodeid == NUMA_NO_NODE) |
7d6e6d09d
|
3324 |
nodeid = slab_node; |
8c8cc2c10
|
3325 3326 3327 3328 3329 3330 |
if (unlikely(!cachep->nodelists[nodeid])) { /* Node not bootstrapped yet */ ptr = fallback_alloc(cachep, flags); goto out; } |
7d6e6d09d
|
3331 |
if (nodeid == slab_node) { |
8c8cc2c10
|
3332 3333 3334 3335 3336 3337 3338 3339 3340 3341 3342 3343 3344 3345 3346 |
/* * 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); |
d5cff6352
|
3347 3348 |
kmemleak_alloc_recursive(ptr, obj_size(cachep), 1, cachep->flags, flags); |
8c8cc2c10
|
3349 |
|
c175eea46
|
3350 3351 |
if (likely(ptr)) kmemcheck_slab_alloc(cachep, flags, ptr, obj_size(cachep)); |
d07dbea46
|
3352 3353 |
if (unlikely((flags & __GFP_ZERO) && ptr)) memset(ptr, 0, obj_size(cachep)); |
8c8cc2c10
|
3354 3355 3356 3357 3358 3359 3360 3361 3362 3363 3364 3365 3366 3367 3368 3369 3370 3371 3372 |
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 */ |
7d6e6d09d
|
3373 3374 |
if (!objp) objp = ____cache_alloc_node(cache, flags, numa_mem_id()); |
8c8cc2c10
|
3375 3376 3377 3378 3379 3380 3381 3382 3383 3384 3385 3386 3387 3388 3389 3390 3391 3392 3393 |
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; |
dcce284a2
|
3394 |
flags &= gfp_allowed_mask; |
7e85ee0c1
|
3395 |
|
cf40bd16f
|
3396 |
lockdep_trace_alloc(flags); |
773ff60e8
|
3397 |
if (slab_should_failslab(cachep, flags)) |
824ebef12
|
3398 |
return NULL; |
8c8cc2c10
|
3399 3400 3401 3402 3403 |
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); |
d5cff6352
|
3404 3405 |
kmemleak_alloc_recursive(objp, obj_size(cachep), 1, cachep->flags, flags); |
8c8cc2c10
|
3406 |
prefetchw(objp); |
c175eea46
|
3407 3408 |
if (likely(objp)) kmemcheck_slab_alloc(cachep, flags, objp, obj_size(cachep)); |
d07dbea46
|
3409 3410 |
if (unlikely((flags & __GFP_ZERO) && objp)) memset(objp, 0, obj_size(cachep)); |
8c8cc2c10
|
3411 3412 |
return objp; } |
e498be7da
|
3413 3414 3415 3416 |
/* * Caller needs to acquire correct kmem_list's list_lock */ |
343e0d7a9
|
3417 |
static void free_block(struct kmem_cache *cachep, void **objpp, int nr_objects, |
b28a02de8
|
3418 |
int node) |
1da177e4c
|
3419 3420 |
{ int i; |
e498be7da
|
3421 |
struct kmem_list3 *l3; |
1da177e4c
|
3422 3423 3424 3425 |
for (i = 0; i < nr_objects; i++) { void *objp = objpp[i]; struct slab *slabp; |
1da177e4c
|
3426 |
|
6ed5eb221
|
3427 |
slabp = virt_to_slab(objp); |
ff69416e6
|
3428 |
l3 = cachep->nodelists[node]; |
1da177e4c
|
3429 |
list_del(&slabp->list); |
ff69416e6
|
3430 |
check_spinlock_acquired_node(cachep, node); |
1da177e4c
|
3431 |
check_slabp(cachep, slabp); |
78d382d77
|
3432 |
slab_put_obj(cachep, slabp, objp, node); |
1da177e4c
|
3433 |
STATS_DEC_ACTIVE(cachep); |
e498be7da
|
3434 |
l3->free_objects++; |
1da177e4c
|
3435 3436 3437 3438 |
check_slabp(cachep, slabp); /* fixup slab chains */ if (slabp->inuse == 0) { |
e498be7da
|
3439 3440 |
if (l3->free_objects > l3->free_limit) { l3->free_objects -= cachep->num; |
e5ac9c5ae
|
3441 3442 3443 3444 3445 3446 |
/* 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
|
3447 3448 |
slab_destroy(cachep, slabp); } else { |
e498be7da
|
3449 |
list_add(&slabp->list, &l3->slabs_free); |
1da177e4c
|
3450 3451 3452 3453 3454 3455 |
} } 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
|
3456 |
list_add_tail(&slabp->list, &l3->slabs_partial); |
1da177e4c
|
3457 3458 3459 |
} } } |
343e0d7a9
|
3460 |
static void cache_flusharray(struct kmem_cache *cachep, struct array_cache *ac) |
1da177e4c
|
3461 3462 |
{ int batchcount; |
e498be7da
|
3463 |
struct kmem_list3 *l3; |
7d6e6d09d
|
3464 |
int node = numa_mem_id(); |
1da177e4c
|
3465 3466 3467 3468 3469 3470 |
batchcount = ac->batchcount; #if DEBUG BUG_ON(!batchcount || batchcount > ac->avail); #endif check_irq_off(); |
ff69416e6
|
3471 |
l3 = cachep->nodelists[node]; |
873623dfa
|
3472 |
spin_lock(&l3->list_lock); |
e498be7da
|
3473 3474 |
if (l3->shared) { struct array_cache *shared_array = l3->shared; |
b28a02de8
|
3475 |
int max = shared_array->limit - shared_array->avail; |
1da177e4c
|
3476 3477 3478 |
if (max) { if (batchcount > max) batchcount = max; |
e498be7da
|
3479 |
memcpy(&(shared_array->entry[shared_array->avail]), |
b28a02de8
|
3480 |
ac->entry, sizeof(void *) * batchcount); |
1da177e4c
|
3481 3482 3483 3484 |
shared_array->avail += batchcount; goto free_done; } } |
ff69416e6
|
3485 |
free_block(cachep, ac->entry, batchcount, node); |
a737b3e2f
|
3486 |
free_done: |
1da177e4c
|
3487 3488 3489 3490 |
#if STATS { int i = 0; struct list_head *p; |
e498be7da
|
3491 3492 |
p = l3->slabs_free.next; while (p != &(l3->slabs_free)) { |
1da177e4c
|
3493 3494 3495 3496 3497 3498 3499 3500 3501 3502 3503 |
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
|
3504 |
spin_unlock(&l3->list_lock); |
1da177e4c
|
3505 |
ac->avail -= batchcount; |
a737b3e2f
|
3506 |
memmove(ac->entry, &(ac->entry[batchcount]), sizeof(void *)*ac->avail); |
1da177e4c
|
3507 3508 3509 |
} /* |
a737b3e2f
|
3510 3511 |
* 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
|
3512 |
*/ |
a947eb95e
|
3513 3514 |
static inline void __cache_free(struct kmem_cache *cachep, void *objp, void *caller) |
1da177e4c
|
3515 |
{ |
9a2dba4b4
|
3516 |
struct array_cache *ac = cpu_cache_get(cachep); |
1da177e4c
|
3517 3518 |
check_irq_off(); |
d5cff6352
|
3519 |
kmemleak_free_recursive(objp, cachep->flags); |
a947eb95e
|
3520 |
objp = cache_free_debugcheck(cachep, objp, caller); |
1da177e4c
|
3521 |
|
c175eea46
|
3522 |
kmemcheck_slab_free(cachep, objp, obj_size(cachep)); |
1807a1aaf
|
3523 3524 3525 3526 3527 3528 3529 |
/* * 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. */ |
b6e68bc1b
|
3530 |
if (nr_online_nodes > 1 && cache_free_alien(cachep, objp)) |
729bd0b74
|
3531 |
return; |
1da177e4c
|
3532 3533 |
if (likely(ac->avail < ac->limit)) { STATS_INC_FREEHIT(cachep); |
e498be7da
|
3534 |
ac->entry[ac->avail++] = objp; |
1da177e4c
|
3535 3536 3537 3538 |
return; } else { STATS_INC_FREEMISS(cachep); cache_flusharray(cachep, ac); |
e498be7da
|
3539 |
ac->entry[ac->avail++] = objp; |
1da177e4c
|
3540 3541 3542 3543 3544 3545 3546 3547 3548 3549 3550 |
} } /** * 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
|
3551 |
void *kmem_cache_alloc(struct kmem_cache *cachep, gfp_t flags) |
1da177e4c
|
3552 |
{ |
36555751c
|
3553 |
void *ret = __cache_alloc(cachep, flags, __builtin_return_address(0)); |
ca2b84cb3
|
3554 3555 |
trace_kmem_cache_alloc(_RET_IP_, ret, obj_size(cachep), cachep->buffer_size, flags); |
36555751c
|
3556 3557 |
return ret; |
1da177e4c
|
3558 3559 |
} EXPORT_SYMBOL(kmem_cache_alloc); |
0f24f1287
|
3560 |
#ifdef CONFIG_TRACING |
85beb5869
|
3561 3562 |
void * kmem_cache_alloc_trace(size_t size, struct kmem_cache *cachep, gfp_t flags) |
36555751c
|
3563 |
{ |
85beb5869
|
3564 3565 3566 3567 3568 3569 3570 |
void *ret; ret = __cache_alloc(cachep, flags, __builtin_return_address(0)); trace_kmalloc(_RET_IP_, ret, size, slab_buffer_size(cachep), flags); return ret; |
36555751c
|
3571 |
} |
85beb5869
|
3572 |
EXPORT_SYMBOL(kmem_cache_alloc_trace); |
36555751c
|
3573 |
#endif |
1da177e4c
|
3574 |
#ifdef CONFIG_NUMA |
8b98c1699
|
3575 3576 |
void *kmem_cache_alloc_node(struct kmem_cache *cachep, gfp_t flags, int nodeid) { |
36555751c
|
3577 3578 |
void *ret = __cache_alloc_node(cachep, flags, nodeid, __builtin_return_address(0)); |
ca2b84cb3
|
3579 3580 3581 |
trace_kmem_cache_alloc_node(_RET_IP_, ret, obj_size(cachep), cachep->buffer_size, flags, nodeid); |
36555751c
|
3582 3583 |
return ret; |
8b98c1699
|
3584 |
} |
1da177e4c
|
3585 |
EXPORT_SYMBOL(kmem_cache_alloc_node); |
0f24f1287
|
3586 |
#ifdef CONFIG_TRACING |
85beb5869
|
3587 3588 3589 3590 |
void *kmem_cache_alloc_node_trace(size_t size, struct kmem_cache *cachep, gfp_t flags, int nodeid) |
36555751c
|
3591 |
{ |
85beb5869
|
3592 3593 3594 |
void *ret; ret = __cache_alloc_node(cachep, flags, nodeid, |
36555751c
|
3595 |
__builtin_return_address(0)); |
85beb5869
|
3596 3597 3598 3599 |
trace_kmalloc_node(_RET_IP_, ret, size, slab_buffer_size(cachep), flags, nodeid); return ret; |
36555751c
|
3600 |
} |
85beb5869
|
3601 |
EXPORT_SYMBOL(kmem_cache_alloc_node_trace); |
36555751c
|
3602 |
#endif |
8b98c1699
|
3603 3604 |
static __always_inline void * __do_kmalloc_node(size_t size, gfp_t flags, int node, void *caller) |
97e2bde47
|
3605 |
{ |
343e0d7a9
|
3606 |
struct kmem_cache *cachep; |
97e2bde47
|
3607 3608 |
cachep = kmem_find_general_cachep(size, flags); |
6cb8f9132
|
3609 3610 |
if (unlikely(ZERO_OR_NULL_PTR(cachep))) return cachep; |
85beb5869
|
3611 |
return kmem_cache_alloc_node_trace(size, cachep, flags, node); |
97e2bde47
|
3612 |
} |
8b98c1699
|
3613 |
|
0bb38a5cd
|
3614 |
#if defined(CONFIG_DEBUG_SLAB) || defined(CONFIG_TRACING) |
8b98c1699
|
3615 3616 3617 3618 3619 |
void *__kmalloc_node(size_t size, gfp_t flags, int node) { return __do_kmalloc_node(size, flags, node, __builtin_return_address(0)); } |
dbe5e69d2
|
3620 |
EXPORT_SYMBOL(__kmalloc_node); |
8b98c1699
|
3621 3622 |
void *__kmalloc_node_track_caller(size_t size, gfp_t flags, |
ce71e27c6
|
3623 |
int node, unsigned long caller) |
8b98c1699
|
3624 |
{ |
ce71e27c6
|
3625 |
return __do_kmalloc_node(size, flags, node, (void *)caller); |
8b98c1699
|
3626 3627 3628 3629 3630 3631 3632 3633 |
} 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); |
0bb38a5cd
|
3634 |
#endif /* CONFIG_DEBUG_SLAB || CONFIG_TRACING */ |
8b98c1699
|
3635 |
#endif /* CONFIG_NUMA */ |
1da177e4c
|
3636 3637 |
/** |
800590f52
|
3638 |
* __do_kmalloc - allocate memory |
1da177e4c
|
3639 |
* @size: how many bytes of memory are required. |
800590f52
|
3640 |
* @flags: the type of memory to allocate (see kmalloc). |
911851e6e
|
3641 |
* @caller: function caller for debug tracking of the caller |
1da177e4c
|
3642 |
*/ |
7fd6b1413
|
3643 3644 |
static __always_inline void *__do_kmalloc(size_t size, gfp_t flags, void *caller) |
1da177e4c
|
3645 |
{ |
343e0d7a9
|
3646 |
struct kmem_cache *cachep; |
36555751c
|
3647 |
void *ret; |
1da177e4c
|
3648 |
|
97e2bde47
|
3649 3650 3651 3652 3653 3654 |
/* 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
|
3655 3656 |
if (unlikely(ZERO_OR_NULL_PTR(cachep))) return cachep; |
36555751c
|
3657 |
ret = __cache_alloc(cachep, flags, caller); |
ca2b84cb3
|
3658 3659 |
trace_kmalloc((unsigned long) caller, ret, size, cachep->buffer_size, flags); |
36555751c
|
3660 3661 |
return ret; |
7fd6b1413
|
3662 |
} |
7fd6b1413
|
3663 |
|
0bb38a5cd
|
3664 |
#if defined(CONFIG_DEBUG_SLAB) || defined(CONFIG_TRACING) |
7fd6b1413
|
3665 3666 |
void *__kmalloc(size_t size, gfp_t flags) { |
871751e25
|
3667 |
return __do_kmalloc(size, flags, __builtin_return_address(0)); |
1da177e4c
|
3668 3669 |
} EXPORT_SYMBOL(__kmalloc); |
ce71e27c6
|
3670 |
void *__kmalloc_track_caller(size_t size, gfp_t flags, unsigned long caller) |
7fd6b1413
|
3671 |
{ |
ce71e27c6
|
3672 |
return __do_kmalloc(size, flags, (void *)caller); |
7fd6b1413
|
3673 3674 |
} EXPORT_SYMBOL(__kmalloc_track_caller); |
1d2c8eea6
|
3675 3676 3677 3678 3679 3680 3681 |
#else void *__kmalloc(size_t size, gfp_t flags) { return __do_kmalloc(size, flags, NULL); } EXPORT_SYMBOL(__kmalloc); |
7fd6b1413
|
3682 |
#endif |
1da177e4c
|
3683 3684 3685 3686 3687 3688 3689 3690 |
/** * 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
|
3691 |
void kmem_cache_free(struct kmem_cache *cachep, void *objp) |
1da177e4c
|
3692 3693 3694 3695 |
{ unsigned long flags; local_irq_save(flags); |
898552c9d
|
3696 |
debug_check_no_locks_freed(objp, obj_size(cachep)); |
3ac7fe5a4
|
3697 3698 |
if (!(cachep->flags & SLAB_DEBUG_OBJECTS)) debug_check_no_obj_freed(objp, obj_size(cachep)); |
a947eb95e
|
3699 |
__cache_free(cachep, objp, __builtin_return_address(0)); |
1da177e4c
|
3700 |
local_irq_restore(flags); |
36555751c
|
3701 |
|
ca2b84cb3
|
3702 |
trace_kmem_cache_free(_RET_IP_, objp); |
1da177e4c
|
3703 3704 3705 3706 |
} EXPORT_SYMBOL(kmem_cache_free); /** |
1da177e4c
|
3707 3708 3709 |
* kfree - free previously allocated memory * @objp: pointer returned by kmalloc. * |
80e93effc
|
3710 3711 |
* If @objp is NULL, no operation is performed. * |
1da177e4c
|
3712 3713 3714 3715 3716 |
* Don't free memory not originally allocated by kmalloc() * or you will run into trouble. */ void kfree(const void *objp) { |
343e0d7a9
|
3717 |
struct kmem_cache *c; |
1da177e4c
|
3718 |
unsigned long flags; |
2121db74b
|
3719 |
trace_kfree(_RET_IP_, objp); |
6cb8f9132
|
3720 |
if (unlikely(ZERO_OR_NULL_PTR(objp))) |
1da177e4c
|
3721 3722 3723 |
return; local_irq_save(flags); kfree_debugcheck(objp); |
6ed5eb221
|
3724 |
c = virt_to_cache(objp); |
f9b8404cf
|
3725 |
debug_check_no_locks_freed(objp, obj_size(c)); |
3ac7fe5a4
|
3726 |
debug_check_no_obj_freed(objp, obj_size(c)); |
a947eb95e
|
3727 |
__cache_free(c, (void *)objp, __builtin_return_address(0)); |
1da177e4c
|
3728 3729 3730 |
local_irq_restore(flags); } EXPORT_SYMBOL(kfree); |
343e0d7a9
|
3731 |
unsigned int kmem_cache_size(struct kmem_cache *cachep) |
1da177e4c
|
3732 |
{ |
3dafccf22
|
3733 |
return obj_size(cachep); |
1da177e4c
|
3734 3735 |
} EXPORT_SYMBOL(kmem_cache_size); |
e498be7da
|
3736 |
/* |
183ff22bb
|
3737 |
* This initializes kmem_list3 or resizes various caches for all nodes. |
e498be7da
|
3738 |
*/ |
83b519e8b
|
3739 |
static int alloc_kmemlist(struct kmem_cache *cachep, gfp_t gfp) |
e498be7da
|
3740 3741 3742 |
{ int node; struct kmem_list3 *l3; |
cafeb02e0
|
3743 |
struct array_cache *new_shared; |
3395ee058
|
3744 |
struct array_cache **new_alien = NULL; |
e498be7da
|
3745 |
|
9c09a95cf
|
3746 |
for_each_online_node(node) { |
cafeb02e0
|
3747 |
|
3395ee058
|
3748 |
if (use_alien_caches) { |
83b519e8b
|
3749 |
new_alien = alloc_alien_cache(node, cachep->limit, gfp); |
3395ee058
|
3750 3751 3752 |
if (!new_alien) goto fail; } |
cafeb02e0
|
3753 |
|
631098469
|
3754 3755 3756 |
new_shared = NULL; if (cachep->shared) { new_shared = alloc_arraycache(node, |
0718dc2a8
|
3757 |
cachep->shared*cachep->batchcount, |
83b519e8b
|
3758 |
0xbaadf00d, gfp); |
631098469
|
3759 3760 3761 3762 |
if (!new_shared) { free_alien_cache(new_alien); goto fail; } |
0718dc2a8
|
3763 |
} |
cafeb02e0
|
3764 |
|
a737b3e2f
|
3765 3766 |
l3 = cachep->nodelists[node]; if (l3) { |
cafeb02e0
|
3767 |
struct array_cache *shared = l3->shared; |
e498be7da
|
3768 |
spin_lock_irq(&l3->list_lock); |
cafeb02e0
|
3769 |
if (shared) |
0718dc2a8
|
3770 3771 |
free_block(cachep, shared->entry, shared->avail, node); |
e498be7da
|
3772 |
|
cafeb02e0
|
3773 3774 |
l3->shared = new_shared; if (!l3->alien) { |
e498be7da
|
3775 3776 3777 |
l3->alien = new_alien; new_alien = NULL; } |
b28a02de8
|
3778 |
l3->free_limit = (1 + nr_cpus_node(node)) * |
a737b3e2f
|
3779 |
cachep->batchcount + cachep->num; |
e498be7da
|
3780 |
spin_unlock_irq(&l3->list_lock); |
cafeb02e0
|
3781 |
kfree(shared); |
e498be7da
|
3782 3783 3784 |
free_alien_cache(new_alien); continue; } |
83b519e8b
|
3785 |
l3 = kmalloc_node(sizeof(struct kmem_list3), gfp, node); |
0718dc2a8
|
3786 3787 3788 |
if (!l3) { free_alien_cache(new_alien); kfree(new_shared); |
e498be7da
|
3789 |
goto fail; |
0718dc2a8
|
3790 |
} |
e498be7da
|
3791 3792 3793 |
kmem_list3_init(l3); l3->next_reap = jiffies + REAPTIMEOUT_LIST3 + |
a737b3e2f
|
3794 |
((unsigned long)cachep) % REAPTIMEOUT_LIST3; |
cafeb02e0
|
3795 |
l3->shared = new_shared; |
e498be7da
|
3796 |
l3->alien = new_alien; |
b28a02de8
|
3797 |
l3->free_limit = (1 + nr_cpus_node(node)) * |
a737b3e2f
|
3798 |
cachep->batchcount + cachep->num; |
e498be7da
|
3799 3800 |
cachep->nodelists[node] = l3; } |
cafeb02e0
|
3801 |
return 0; |
0718dc2a8
|
3802 |
|
a737b3e2f
|
3803 |
fail: |
0718dc2a8
|
3804 3805 3806 3807 3808 3809 3810 3811 3812 3813 3814 3815 3816 3817 3818 |
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
|
3819 |
return -ENOMEM; |
e498be7da
|
3820 |
} |
1da177e4c
|
3821 |
struct ccupdate_struct { |
343e0d7a9
|
3822 |
struct kmem_cache *cachep; |
acfe7d744
|
3823 |
struct array_cache *new[0]; |
1da177e4c
|
3824 3825 3826 3827 |
}; static void do_ccupdate_local(void *info) { |
a737b3e2f
|
3828 |
struct ccupdate_struct *new = info; |
1da177e4c
|
3829 3830 3831 |
struct array_cache *old; check_irq_off(); |
9a2dba4b4
|
3832 |
old = cpu_cache_get(new->cachep); |
e498be7da
|
3833 |
|
1da177e4c
|
3834 3835 3836 |
new->cachep->array[smp_processor_id()] = new->new[smp_processor_id()]; new->new[smp_processor_id()] = old; } |
b5d8ca7c5
|
3837 |
/* Always called with the cache_chain_mutex held */ |
a737b3e2f
|
3838 |
static int do_tune_cpucache(struct kmem_cache *cachep, int limit, |
83b519e8b
|
3839 |
int batchcount, int shared, gfp_t gfp) |
1da177e4c
|
3840 |
{ |
d2e7b7d0a
|
3841 |
struct ccupdate_struct *new; |
2ed3a4ef9
|
3842 |
int i; |
1da177e4c
|
3843 |
|
acfe7d744
|
3844 3845 |
new = kzalloc(sizeof(*new) + nr_cpu_ids * sizeof(struct array_cache *), gfp); |
d2e7b7d0a
|
3846 3847 |
if (!new) return -ENOMEM; |
e498be7da
|
3848 |
for_each_online_cpu(i) { |
7d6e6d09d
|
3849 |
new->new[i] = alloc_arraycache(cpu_to_mem(i), limit, |
83b519e8b
|
3850 |
batchcount, gfp); |
d2e7b7d0a
|
3851 |
if (!new->new[i]) { |
b28a02de8
|
3852 |
for (i--; i >= 0; i--) |
d2e7b7d0a
|
3853 3854 |
kfree(new->new[i]); kfree(new); |
e498be7da
|
3855 |
return -ENOMEM; |
1da177e4c
|
3856 3857 |
} } |
d2e7b7d0a
|
3858 |
new->cachep = cachep; |
1da177e4c
|
3859 |
|
15c8b6c1a
|
3860 |
on_each_cpu(do_ccupdate_local, (void *)new, 1); |
e498be7da
|
3861 |
|
1da177e4c
|
3862 |
check_irq_on(); |
1da177e4c
|
3863 3864 |
cachep->batchcount = batchcount; cachep->limit = limit; |
e498be7da
|
3865 |
cachep->shared = shared; |
1da177e4c
|
3866 |
|
e498be7da
|
3867 |
for_each_online_cpu(i) { |
d2e7b7d0a
|
3868 |
struct array_cache *ccold = new->new[i]; |
1da177e4c
|
3869 3870 |
if (!ccold) continue; |
7d6e6d09d
|
3871 3872 3873 |
spin_lock_irq(&cachep->nodelists[cpu_to_mem(i)]->list_lock); free_block(cachep, ccold->entry, ccold->avail, cpu_to_mem(i)); spin_unlock_irq(&cachep->nodelists[cpu_to_mem(i)]->list_lock); |
1da177e4c
|
3874 3875 |
kfree(ccold); } |
d2e7b7d0a
|
3876 |
kfree(new); |
83b519e8b
|
3877 |
return alloc_kmemlist(cachep, gfp); |
1da177e4c
|
3878 |
} |
b5d8ca7c5
|
3879 |
/* Called with cache_chain_mutex held always */ |
83b519e8b
|
3880 |
static int enable_cpucache(struct kmem_cache *cachep, gfp_t gfp) |
1da177e4c
|
3881 3882 3883 |
{ int err; int limit, shared; |
a737b3e2f
|
3884 3885 |
/* * The head array serves three purposes: |
1da177e4c
|
3886 3887 |
* - create a LIFO ordering, i.e. return objects that are cache-warm * - reduce the number of spinlock operations. |
a737b3e2f
|
3888 |
* - reduce the number of linked list operations on the slab and |
1da177e4c
|
3889 3890 3891 3892 |
* bufctl chains: array operations are cheaper. * The numbers are guessed, we should auto-tune as described by * Bonwick. */ |
3dafccf22
|
3893 |
if (cachep->buffer_size > 131072) |
1da177e4c
|
3894 |
limit = 1; |
3dafccf22
|
3895 |
else if (cachep->buffer_size > PAGE_SIZE) |
1da177e4c
|
3896 |
limit = 8; |
3dafccf22
|
3897 |
else if (cachep->buffer_size > 1024) |
1da177e4c
|
3898 |
limit = 24; |
3dafccf22
|
3899 |
else if (cachep->buffer_size > 256) |
1da177e4c
|
3900 3901 3902 |
limit = 54; else limit = 120; |
a737b3e2f
|
3903 3904 |
/* * CPU bound tasks (e.g. network routing) can exhibit cpu bound |
1da177e4c
|
3905 3906 3907 3908 3909 3910 3911 3912 |
* 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
|
3913 |
if (cachep->buffer_size <= PAGE_SIZE && num_possible_cpus() > 1) |
1da177e4c
|
3914 |
shared = 8; |
1da177e4c
|
3915 3916 |
#if DEBUG |
a737b3e2f
|
3917 3918 3919 |
/* * With debugging enabled, large batchcount lead to excessively long * periods with disabled local interrupts. Limit the batchcount |
1da177e4c
|
3920 3921 3922 3923 |
*/ if (limit > 32) limit = 32; #endif |
83b519e8b
|
3924 |
err = do_tune_cpucache(cachep, limit, (limit + 1) / 2, shared, gfp); |
1da177e4c
|
3925 3926 3927 |
if (err) printk(KERN_ERR "enable_cpucache failed for %s, error %d. ", |
b28a02de8
|
3928 |
cachep->name, -err); |
2ed3a4ef9
|
3929 |
return err; |
1da177e4c
|
3930 |
} |
1b55253a7
|
3931 3932 |
/* * Drain an array if it contains any elements taking the l3 lock only if |
b18e7e654
|
3933 3934 |
* necessary. Note that the l3 listlock also protects the array_cache * if drain_array() is used on the shared array. |
1b55253a7
|
3935 |
*/ |
68a1b1955
|
3936 |
static void drain_array(struct kmem_cache *cachep, struct kmem_list3 *l3, |
1b55253a7
|
3937 |
struct array_cache *ac, int force, int node) |
1da177e4c
|
3938 3939 |
{ int tofree; |
1b55253a7
|
3940 3941 |
if (!ac || !ac->avail) return; |
1da177e4c
|
3942 3943 |
if (ac->touched && !force) { ac->touched = 0; |
b18e7e654
|
3944 |
} else { |
1b55253a7
|
3945 |
spin_lock_irq(&l3->list_lock); |
b18e7e654
|
3946 3947 3948 3949 3950 3951 3952 3953 3954 |
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
|
3955 |
spin_unlock_irq(&l3->list_lock); |
1da177e4c
|
3956 3957 3958 3959 3960 |
} } /** * cache_reap - Reclaim memory from caches. |
05fb6bf0b
|
3961 |
* @w: work descriptor |
1da177e4c
|
3962 3963 3964 3965 3966 3967 |
* * 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
|
3968 3969 |
* If we cannot acquire the cache chain mutex then just give up - we'll try * again on the next iteration. |
1da177e4c
|
3970 |
*/ |
7c5cae368
|
3971 |
static void cache_reap(struct work_struct *w) |
1da177e4c
|
3972 |
{ |
7a7c381d2
|
3973 |
struct kmem_cache *searchp; |
e498be7da
|
3974 |
struct kmem_list3 *l3; |
7d6e6d09d
|
3975 |
int node = numa_mem_id(); |
bf6aede71
|
3976 |
struct delayed_work *work = to_delayed_work(w); |
1da177e4c
|
3977 |
|
7c5cae368
|
3978 |
if (!mutex_trylock(&cache_chain_mutex)) |
1da177e4c
|
3979 |
/* Give up. Setup the next iteration. */ |
7c5cae368
|
3980 |
goto out; |
1da177e4c
|
3981 |
|
7a7c381d2
|
3982 |
list_for_each_entry(searchp, &cache_chain, next) { |
1da177e4c
|
3983 |
check_irq_on(); |
35386e3b0
|
3984 3985 3986 3987 3988 |
/* * 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
|
3989 |
l3 = searchp->nodelists[node]; |
35386e3b0
|
3990 |
|
8fce4d8e3
|
3991 |
reap_alien(searchp, l3); |
1da177e4c
|
3992 |
|
aab2207cf
|
3993 |
drain_array(searchp, l3, cpu_cache_get(searchp), 0, node); |
1da177e4c
|
3994 |
|
35386e3b0
|
3995 3996 3997 3998 |
/* * These are racy checks but it does not matter * if we skip one check or scan twice. */ |
e498be7da
|
3999 |
if (time_after(l3->next_reap, jiffies)) |
35386e3b0
|
4000 |
goto next; |
1da177e4c
|
4001 |
|
e498be7da
|
4002 |
l3->next_reap = jiffies + REAPTIMEOUT_LIST3; |
1da177e4c
|
4003 |
|
aab2207cf
|
4004 |
drain_array(searchp, l3, l3->shared, 0, node); |
1da177e4c
|
4005 |
|
ed11d9eb2
|
4006 |
if (l3->free_touched) |
e498be7da
|
4007 |
l3->free_touched = 0; |
ed11d9eb2
|
4008 4009 |
else { int freed; |
1da177e4c
|
4010 |
|
ed11d9eb2
|
4011 4012 4013 4014 |
freed = drain_freelist(searchp, l3, (l3->free_limit + 5 * searchp->num - 1) / (5 * searchp->num)); STATS_ADD_REAPED(searchp, freed); } |
35386e3b0
|
4015 |
next: |
1da177e4c
|
4016 4017 4018 |
cond_resched(); } check_irq_on(); |
fc0abb145
|
4019 |
mutex_unlock(&cache_chain_mutex); |
8fce4d8e3
|
4020 |
next_reap_node(); |
7c5cae368
|
4021 |
out: |
a737b3e2f
|
4022 |
/* Set up the next iteration */ |
7c5cae368
|
4023 |
schedule_delayed_work(work, round_jiffies_relative(REAPTIMEOUT_CPUC)); |
1da177e4c
|
4024 |
} |
158a96242
|
4025 |
#ifdef CONFIG_SLABINFO |
1da177e4c
|
4026 |
|
85289f98d
|
4027 |
static void print_slabinfo_header(struct seq_file *m) |
1da177e4c
|
4028 |
{ |
85289f98d
|
4029 4030 4031 4032 |
/* * Output format version, so at least we can change it * without _too_ many complaints. */ |
1da177e4c
|
4033 |
#if STATS |
85289f98d
|
4034 4035 |
seq_puts(m, "slabinfo - version: 2.1 (statistics) "); |
1da177e4c
|
4036 |
#else |
85289f98d
|
4037 4038 |
seq_puts(m, "slabinfo - version: 2.1 "); |
1da177e4c
|
4039 |
#endif |
85289f98d
|
4040 4041 4042 4043 |
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
|
4044 |
#if STATS |
85289f98d
|
4045 |
seq_puts(m, " : globalstat <listallocs> <maxobjs> <grown> <reaped> " |
fb7faf331
|
4046 |
"<error> <maxfreeable> <nodeallocs> <remotefrees> <alienoverflow>"); |
85289f98d
|
4047 |
seq_puts(m, " : cpustat <allochit> <allocmiss> <freehit> <freemiss>"); |
1da177e4c
|
4048 |
#endif |
85289f98d
|
4049 4050 4051 4052 4053 4054 4055 |
seq_putc(m, ' '); } static void *s_start(struct seq_file *m, loff_t *pos) { loff_t n = *pos; |
85289f98d
|
4056 |
|
fc0abb145
|
4057 |
mutex_lock(&cache_chain_mutex); |
85289f98d
|
4058 4059 |
if (!n) print_slabinfo_header(m); |
b92151bab
|
4060 4061 |
return seq_list_start(&cache_chain, *pos); |
1da177e4c
|
4062 4063 4064 4065 |
} static void *s_next(struct seq_file *m, void *p, loff_t *pos) { |
b92151bab
|
4066 |
return seq_list_next(p, &cache_chain, pos); |
1da177e4c
|
4067 4068 4069 4070 |
} static void s_stop(struct seq_file *m, void *p) { |
fc0abb145
|
4071 |
mutex_unlock(&cache_chain_mutex); |
1da177e4c
|
4072 4073 4074 4075 |
} static int s_show(struct seq_file *m, void *p) { |
b92151bab
|
4076 |
struct kmem_cache *cachep = list_entry(p, struct kmem_cache, next); |
b28a02de8
|
4077 4078 4079 4080 4081 |
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
|
4082 |
const char *name; |
1da177e4c
|
4083 |
char *error = NULL; |
e498be7da
|
4084 4085 |
int node; struct kmem_list3 *l3; |
1da177e4c
|
4086 |
|
1da177e4c
|
4087 4088 |
active_objs = 0; num_slabs = 0; |
e498be7da
|
4089 4090 4091 4092 |
for_each_online_node(node) { l3 = cachep->nodelists[node]; if (!l3) continue; |
ca3b9b917
|
4093 4094 |
check_irq_on(); spin_lock_irq(&l3->list_lock); |
e498be7da
|
4095 |
|
7a7c381d2
|
4096 |
list_for_each_entry(slabp, &l3->slabs_full, list) { |
e498be7da
|
4097 4098 4099 4100 4101 |
if (slabp->inuse != cachep->num && !error) error = "slabs_full accounting error"; active_objs += cachep->num; active_slabs++; } |
7a7c381d2
|
4102 |
list_for_each_entry(slabp, &l3->slabs_partial, list) { |
e498be7da
|
4103 4104 4105 4106 4107 4108 4109 |
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
|
4110 |
list_for_each_entry(slabp, &l3->slabs_free, list) { |
e498be7da
|
4111 4112 4113 4114 4115 |
if (slabp->inuse && !error) error = "slabs_free/inuse accounting error"; num_slabs++; } free_objects += l3->free_objects; |
4484ebf12
|
4116 4117 |
if (l3->shared) shared_avail += l3->shared->avail; |
e498be7da
|
4118 |
|
ca3b9b917
|
4119 |
spin_unlock_irq(&l3->list_lock); |
1da177e4c
|
4120 |
} |
b28a02de8
|
4121 4122 |
num_slabs += active_slabs; num_objs = num_slabs * cachep->num; |
e498be7da
|
4123 |
if (num_objs - active_objs != free_objects && !error) |
1da177e4c
|
4124 |
error = "free_objects accounting error"; |
b28a02de8
|
4125 |
name = cachep->name; |
1da177e4c
|
4126 4127 4128 4129 4130 |
if (error) printk(KERN_ERR "slab: cache %s error: %s ", name, error); seq_printf(m, "%-17s %6lu %6lu %6u %4u %4d", |
3dafccf22
|
4131 |
name, active_objs, num_objs, cachep->buffer_size, |
b28a02de8
|
4132 |
cachep->num, (1 << cachep->gfporder)); |
1da177e4c
|
4133 |
seq_printf(m, " : tunables %4u %4u %4u", |
b28a02de8
|
4134 |
cachep->limit, cachep->batchcount, cachep->shared); |
e498be7da
|
4135 |
seq_printf(m, " : slabdata %6lu %6lu %6lu", |
b28a02de8
|
4136 |
active_slabs, num_slabs, shared_avail); |
1da177e4c
|
4137 |
#if STATS |
b28a02de8
|
4138 |
{ /* list3 stats */ |
1da177e4c
|
4139 4140 4141 4142 4143 4144 |
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
|
4145 |
unsigned long node_allocs = cachep->node_allocs; |
e498be7da
|
4146 |
unsigned long node_frees = cachep->node_frees; |
fb7faf331
|
4147 |
unsigned long overflows = cachep->node_overflow; |
1da177e4c
|
4148 |
|
e92dd4fd1
|
4149 4150 4151 4152 4153 |
seq_printf(m, " : globalstat %7lu %6lu %5lu %4lu " "%4lu %4lu %4lu %4lu %4lu", allocs, high, grown, reaped, errors, max_freeable, node_allocs, node_frees, overflows); |
1da177e4c
|
4154 4155 4156 4157 4158 4159 4160 4161 4162 |
} /* 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
|
4163 |
allochit, allocmiss, freehit, freemiss); |
1da177e4c
|
4164 4165 4166 4167 |
} #endif seq_putc(m, ' '); |
1da177e4c
|
4168 4169 4170 4171 4172 4173 4174 4175 4176 4177 4178 4179 4180 4181 4182 4183 |
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
|
4184 |
static const struct seq_operations slabinfo_op = { |
b28a02de8
|
4185 4186 4187 4188 |
.start = s_start, .next = s_next, .stop = s_stop, .show = s_show, |
1da177e4c
|
4189 4190 4191 4192 4193 4194 4195 4196 4197 4198 |
}; #define MAX_SLABINFO_WRITE 128 /** * slabinfo_write - Tuning for the slab allocator * @file: unused * @buffer: user buffer * @count: data length * @ppos: unused */ |
68a1b1955
|
4199 |
static ssize_t slabinfo_write(struct file *file, const char __user *buffer, |
b28a02de8
|
4200 |
size_t count, loff_t *ppos) |
1da177e4c
|
4201 |
{ |
b28a02de8
|
4202 |
char kbuf[MAX_SLABINFO_WRITE + 1], *tmp; |
1da177e4c
|
4203 |
int limit, batchcount, shared, res; |
7a7c381d2
|
4204 |
struct kmem_cache *cachep; |
b28a02de8
|
4205 |
|
1da177e4c
|
4206 4207 4208 4209 |
if (count > MAX_SLABINFO_WRITE) return -EINVAL; if (copy_from_user(&kbuf, buffer, count)) return -EFAULT; |
b28a02de8
|
4210 |
kbuf[MAX_SLABINFO_WRITE] = '\0'; |
1da177e4c
|
4211 4212 4213 4214 4215 4216 4217 4218 4219 4220 |
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
|
4221 |
mutex_lock(&cache_chain_mutex); |
1da177e4c
|
4222 |
res = -EINVAL; |
7a7c381d2
|
4223 |
list_for_each_entry(cachep, &cache_chain, next) { |
1da177e4c
|
4224 |
if (!strcmp(cachep->name, kbuf)) { |
a737b3e2f
|
4225 4226 |
if (limit < 1 || batchcount < 1 || batchcount > limit || shared < 0) { |
e498be7da
|
4227 |
res = 0; |
1da177e4c
|
4228 |
} else { |
e498be7da
|
4229 |
res = do_tune_cpucache(cachep, limit, |
83b519e8b
|
4230 4231 |
batchcount, shared, GFP_KERNEL); |
1da177e4c
|
4232 4233 4234 4235 |
} break; } } |
fc0abb145
|
4236 |
mutex_unlock(&cache_chain_mutex); |
1da177e4c
|
4237 4238 4239 4240 |
if (res >= 0) res = count; return res; } |
871751e25
|
4241 |
|
7b3c3a50a
|
4242 4243 4244 4245 4246 4247 4248 4249 4250 4251 4252 4253 |
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
|
4254 4255 4256 4257 |
#ifdef CONFIG_DEBUG_SLAB_LEAK static void *leaks_start(struct seq_file *m, loff_t *pos) { |
871751e25
|
4258 |
mutex_lock(&cache_chain_mutex); |
b92151bab
|
4259 |
return seq_list_start(&cache_chain, *pos); |
871751e25
|
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 4287 4288 4289 4290 4291 4292 4293 4294 4295 4296 4297 4298 4299 4300 4301 4302 4303 4304 4305 4306 4307 4308 |
} 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
|
4309 |
unsigned long offset, size; |
9281acea6
|
4310 |
char modname[MODULE_NAME_LEN], name[KSYM_NAME_LEN]; |
871751e25
|
4311 |
|
a5c43dae7
|
4312 |
if (lookup_symbol_attrs(address, &size, &offset, modname, name) == 0) { |
871751e25
|
4313 |
seq_printf(m, "%s+%#lx/%#lx", name, offset, size); |
a5c43dae7
|
4314 |
if (modname[0]) |
871751e25
|
4315 4316 4317 4318 4319 4320 4321 4322 4323 |
seq_printf(m, " [%s]", modname); return; } #endif seq_printf(m, "%p", (void *)address); } static int leaks_show(struct seq_file *m, void *p) { |
b92151bab
|
4324 |
struct kmem_cache *cachep = list_entry(p, struct kmem_cache, next); |
871751e25
|
4325 4326 4327 4328 4329 4330 4331 4332 4333 4334 4335 4336 4337 4338 4339 4340 4341 4342 4343 4344 4345 4346 4347 |
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
|
4348 |
list_for_each_entry(slabp, &l3->slabs_full, list) |
871751e25
|
4349 |
handle_slab(n, cachep, slabp); |
7a7c381d2
|
4350 |
list_for_each_entry(slabp, &l3->slabs_partial, list) |
871751e25
|
4351 |
handle_slab(n, cachep, slabp); |
871751e25
|
4352 4353 4354 4355 4356 4357 4358 4359 4360 4361 4362 4363 4364 4365 4366 4367 4368 4369 4370 4371 4372 4373 4374 4375 4376 4377 |
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
|
4378 |
|
871751e25
|
4379 4380 |
return 0; } |
a0ec95a8e
|
4381 |
static const struct seq_operations slabstats_op = { |
871751e25
|
4382 4383 4384 4385 4386 |
.start = leaks_start, .next = s_next, .stop = s_stop, .show = leaks_show, }; |
a0ec95a8e
|
4387 4388 4389 4390 4391 4392 4393 4394 4395 4396 4397 4398 4399 4400 4401 4402 4403 4404 4405 4406 4407 4408 4409 4410 4411 4412 4413 4414 |
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) { |
ab067e99d
|
4415 |
proc_create("slabinfo",S_IWUSR|S_IRUSR,NULL,&proc_slabinfo_operations); |
a0ec95a8e
|
4416 4417 |
#ifdef CONFIG_DEBUG_SLAB_LEAK proc_create("slab_allocators", 0, NULL, &proc_slabstats_operations); |
871751e25
|
4418 |
#endif |
a0ec95a8e
|
4419 4420 4421 |
return 0; } module_init(slab_proc_init); |
1da177e4c
|
4422 |
#endif |
00e145b6d
|
4423 4424 4425 4426 4427 4428 4429 4430 4431 4432 4433 4434 |
/** * 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
|
4435 |
size_t ksize(const void *objp) |
1da177e4c
|
4436 |
{ |
ef8b4520b
|
4437 4438 |
BUG_ON(!objp); if (unlikely(objp == ZERO_SIZE_PTR)) |
00e145b6d
|
4439 |
return 0; |
1da177e4c
|
4440 |
|
6ed5eb221
|
4441 |
return obj_size(virt_to_cache(objp)); |
1da177e4c
|
4442 |
} |
b1aabecd5
|
4443 |
EXPORT_SYMBOL(ksize); |