Blame view
mm/slab.c
117 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 */ |
343e0d7a9
|
556 |
static struct kmem_cache cache_cache = { |
b28a02de8
|
557 558 559 |
.batchcount = 1, .limit = BOOT_CPUCACHE_ENTRIES, .shared = 1, |
343e0d7a9
|
560 |
.buffer_size = sizeof(struct kmem_cache), |
b28a02de8
|
561 |
.name = "kmem_cache", |
1da177e4c
|
562 |
}; |
056c62418
|
563 |
#define BAD_ALIEN_MAGIC 0x01020304ul |
ce79ddc8e
|
564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 |
/* * chicken and egg problem: delay the per-cpu array allocation * until the general caches are up. */ static enum { NONE, PARTIAL_AC, PARTIAL_L3, EARLY, 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
|
583 584 585 586 587 588 589 590 |
#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
|
591 592 593 594 |
* * 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
|
595 |
*/ |
056c62418
|
596 597 |
static struct lock_class_key on_slab_l3_key; static struct lock_class_key on_slab_alc_key; |
ce79ddc8e
|
598 |
static void init_node_lock_keys(int q) |
f1aaee53f
|
599 |
{ |
056c62418
|
600 |
struct cache_sizes *s = malloc_sizes; |
ce79ddc8e
|
601 602 603 604 605 606 607 608 609 610 |
if (g_cpucache_up != FULL) return; for (s = malloc_sizes; s->cs_size != ULONG_MAX; s++) { struct array_cache **alc; struct kmem_list3 *l3; int r; l3 = s->cs_cachep->nodelists[q]; if (!l3 || OFF_SLAB(s->cs_cachep)) |
00afa7580
|
611 |
continue; |
ce79ddc8e
|
612 613 614 615 616 617 618 619 620 621 |
lockdep_set_class(&l3->list_lock, &on_slab_l3_key); alc = l3->alien; /* * FIXME: This check for BAD_ALIEN_MAGIC * should go away when common slab code is taught to * work even without alien caches. * Currently, non NUMA code returns BAD_ALIEN_MAGIC * for alloc_alien_cache, */ if (!alc || (unsigned long)alc == BAD_ALIEN_MAGIC) |
00afa7580
|
622 |
continue; |
ce79ddc8e
|
623 624 625 626 |
for_each_node(r) { if (alc[r]) lockdep_set_class(&alc[r]->lock, &on_slab_alc_key); |
056c62418
|
627 |
} |
f1aaee53f
|
628 629 |
} } |
ce79ddc8e
|
630 631 632 633 634 635 636 637 |
static inline void init_lock_keys(void) { int node; for_each_node(node) init_node_lock_keys(node); } |
f1aaee53f
|
638 |
#else |
ce79ddc8e
|
639 640 641 |
static void init_node_lock_keys(int q) { } |
056c62418
|
642 |
static inline void init_lock_keys(void) |
f1aaee53f
|
643 644 645 |
{ } #endif |
8f5be20bf
|
646 |
/* |
95402b382
|
647 |
* Guard access to the cache-chain. |
8f5be20bf
|
648 |
*/ |
fc0abb145
|
649 |
static DEFINE_MUTEX(cache_chain_mutex); |
1da177e4c
|
650 |
static struct list_head cache_chain; |
1871e52c7
|
651 |
static DEFINE_PER_CPU(struct delayed_work, slab_reap_work); |
1da177e4c
|
652 |
|
343e0d7a9
|
653 |
static inline struct array_cache *cpu_cache_get(struct kmem_cache *cachep) |
1da177e4c
|
654 655 656 |
{ return cachep->array[smp_processor_id()]; } |
a737b3e2f
|
657 658 |
static inline struct kmem_cache *__find_general_cachep(size_t size, gfp_t gfpflags) |
1da177e4c
|
659 660 661 662 663 |
{ struct cache_sizes *csizep = malloc_sizes; #if DEBUG /* This happens if someone tries to call |
b28a02de8
|
664 665 666 |
* kmem_cache_create(), or __kmalloc(), before * the generic caches are initialized. */ |
c7e43c78a
|
667 |
BUG_ON(malloc_sizes[INDEX_AC].cs_cachep == NULL); |
1da177e4c
|
668 |
#endif |
6cb8f9132
|
669 670 |
if (!size) return ZERO_SIZE_PTR; |
1da177e4c
|
671 672 673 674 |
while (size > csizep->cs_size) csizep++; /* |
0abf40c1a
|
675 |
* Really subtle: The last entry with cs->cs_size==ULONG_MAX |
1da177e4c
|
676 677 678 |
* has cs_{dma,}cachep==NULL. Thus no special case * for large kmalloc calls required. */ |
4b51d6698
|
679 |
#ifdef CONFIG_ZONE_DMA |
1da177e4c
|
680 681 |
if (unlikely(gfpflags & GFP_DMA)) return csizep->cs_dmacachep; |
4b51d6698
|
682 |
#endif |
1da177e4c
|
683 684 |
return csizep->cs_cachep; } |
b221385bc
|
685 |
static struct kmem_cache *kmem_find_general_cachep(size_t size, gfp_t gfpflags) |
97e2bde47
|
686 687 688 |
{ return __find_general_cachep(size, gfpflags); } |
97e2bde47
|
689 |
|
fbaccacff
|
690 |
static size_t slab_mgmt_size(size_t nr_objs, size_t align) |
1da177e4c
|
691 |
{ |
fbaccacff
|
692 693 |
return ALIGN(sizeof(struct slab)+nr_objs*sizeof(kmem_bufctl_t), align); } |
1da177e4c
|
694 |
|
a737b3e2f
|
695 696 697 |
/* * Calculate the number of objects and left-over bytes for a given buffer size. */ |
fbaccacff
|
698 699 700 701 702 703 704 |
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
|
705 |
|
fbaccacff
|
706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 |
/* * 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
|
754 |
} |
d40cee245
|
755 |
#define slab_error(cachep, msg) __slab_error(__func__, cachep, msg) |
1da177e4c
|
756 |
|
a737b3e2f
|
757 758 |
static void __slab_error(const char *function, struct kmem_cache *cachep, char *msg) |
1da177e4c
|
759 760 761 |
{ printk(KERN_ERR "slab error in %s(): cache `%s': %s ", |
b28a02de8
|
762 |
function, cachep->name, msg); |
1da177e4c
|
763 764 |
dump_stack(); } |
3395ee058
|
765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 |
/* * 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
|
780 781 782 783 784 785 786 |
#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
|
787 |
static DEFINE_PER_CPU(unsigned long, slab_reap_node); |
8fce4d8e3
|
788 789 790 791 |
static void init_reap_node(int cpu) { int node; |
7d6e6d09d
|
792 |
node = next_node(cpu_to_mem(cpu), node_online_map); |
8fce4d8e3
|
793 |
if (node == MAX_NUMNODES) |
442295c94
|
794 |
node = first_node(node_online_map); |
8fce4d8e3
|
795 |
|
1871e52c7
|
796 |
per_cpu(slab_reap_node, cpu) = node; |
8fce4d8e3
|
797 798 799 800 |
} static void next_reap_node(void) { |
909ea9646
|
801 |
int node = __this_cpu_read(slab_reap_node); |
8fce4d8e3
|
802 |
|
8fce4d8e3
|
803 804 805 |
node = next_node(node, node_online_map); if (unlikely(node >= MAX_NUMNODES)) node = first_node(node_online_map); |
909ea9646
|
806 |
__this_cpu_write(slab_reap_node, node); |
8fce4d8e3
|
807 808 809 810 811 812 |
} #else #define init_reap_node(cpu) do { } while (0) #define next_reap_node(void) do { } while (0) #endif |
1da177e4c
|
813 814 815 816 817 818 819 |
/* * 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
|
820 |
static void __cpuinit start_cpu_timer(int cpu) |
1da177e4c
|
821 |
{ |
1871e52c7
|
822 |
struct delayed_work *reap_work = &per_cpu(slab_reap_work, cpu); |
1da177e4c
|
823 824 825 826 827 828 |
/* * When this gets called from do_initcalls via cpucache_init(), * init_workqueues() has already run, so keventd will be setup * at that time. */ |
52bad64d9
|
829 |
if (keventd_up() && reap_work->work.func == NULL) { |
8fce4d8e3
|
830 |
init_reap_node(cpu); |
78b435368
|
831 |
INIT_DELAYED_WORK_DEFERRABLE(reap_work, cache_reap); |
2b2842146
|
832 833 |
schedule_delayed_work_on(cpu, reap_work, __round_jiffies_relative(HZ, cpu)); |
1da177e4c
|
834 835 |
} } |
e498be7da
|
836 |
static struct array_cache *alloc_arraycache(int node, int entries, |
83b519e8b
|
837 |
int batchcount, gfp_t gfp) |
1da177e4c
|
838 |
{ |
b28a02de8
|
839 |
int memsize = sizeof(void *) * entries + sizeof(struct array_cache); |
1da177e4c
|
840 |
struct array_cache *nc = NULL; |
83b519e8b
|
841 |
nc = kmalloc_node(memsize, gfp, node); |
d5cff6352
|
842 843 |
/* * The array_cache structures contain pointers to free object. |
25985edce
|
844 |
* However, when such objects are allocated or transferred to another |
d5cff6352
|
845 846 847 848 849 |
* 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
|
850 851 852 853 854 |
if (nc) { nc->avail = 0; nc->limit = entries; nc->batchcount = batchcount; nc->touched = 0; |
e498be7da
|
855 |
spin_lock_init(&nc->lock); |
1da177e4c
|
856 857 858 |
} return nc; } |
3ded175a4
|
859 860 861 862 863 864 865 866 867 868 |
/* * 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
|
869 |
int nr = min3(from->avail, max, to->limit - to->avail); |
3ded175a4
|
870 871 872 873 874 875 876 877 878 |
if (!nr) return 0; memcpy(to->entry + to->avail, from->entry + from->avail -nr, sizeof(void *) *nr); from->avail -= nr; to->avail += nr; |
3ded175a4
|
879 880 |
return nr; } |
765c4507a
|
881 882 883 884 |
#ifndef CONFIG_NUMA #define drain_alien_cache(cachep, alien) do { } while (0) #define reap_alien(cachep, l3) do { } while (0) |
83b519e8b
|
885 |
static inline struct array_cache **alloc_alien_cache(int node, int limit, gfp_t gfp) |
765c4507a
|
886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 |
{ 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
|
904 |
static inline void *____cache_alloc_node(struct kmem_cache *cachep, |
765c4507a
|
905 906 907 908 909 910 |
gfp_t flags, int nodeid) { return NULL; } #else /* CONFIG_NUMA */ |
8b98c1699
|
911 |
static void *____cache_alloc_node(struct kmem_cache *, gfp_t, int); |
c61afb181
|
912 |
static void *alternate_node_alloc(struct kmem_cache *, gfp_t); |
dc85da15d
|
913 |
|
83b519e8b
|
914 |
static struct array_cache **alloc_alien_cache(int node, int limit, gfp_t gfp) |
e498be7da
|
915 916 |
{ struct array_cache **ac_ptr; |
8ef828668
|
917 |
int memsize = sizeof(void *) * nr_node_ids; |
e498be7da
|
918 919 920 921 |
int i; if (limit > 1) limit = 12; |
f3186a9c5
|
922 |
ac_ptr = kzalloc_node(memsize, gfp, node); |
e498be7da
|
923 924 |
if (ac_ptr) { for_each_node(i) { |
f3186a9c5
|
925 |
if (i == node || !node_online(i)) |
e498be7da
|
926 |
continue; |
83b519e8b
|
927 |
ac_ptr[i] = alloc_arraycache(node, limit, 0xbaadf00d, gfp); |
e498be7da
|
928 |
if (!ac_ptr[i]) { |
cc550defe
|
929 |
for (i--; i >= 0; i--) |
e498be7da
|
930 931 932 933 934 935 936 937 |
kfree(ac_ptr[i]); kfree(ac_ptr); return NULL; } } } return ac_ptr; } |
5295a74cc
|
938 |
static void free_alien_cache(struct array_cache **ac_ptr) |
e498be7da
|
939 940 941 942 943 |
{ int i; if (!ac_ptr) return; |
e498be7da
|
944 |
for_each_node(i) |
b28a02de8
|
945 |
kfree(ac_ptr[i]); |
e498be7da
|
946 947 |
kfree(ac_ptr); } |
343e0d7a9
|
948 |
static void __drain_alien_cache(struct kmem_cache *cachep, |
5295a74cc
|
949 |
struct array_cache *ac, int node) |
e498be7da
|
950 951 952 953 954 |
{ struct kmem_list3 *rl3 = cachep->nodelists[node]; if (ac->avail) { spin_lock(&rl3->list_lock); |
e00946fe2
|
955 956 957 958 959 |
/* * 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
|
960 961 |
if (rl3->shared) transfer_objects(rl3->shared, ac, ac->limit); |
e00946fe2
|
962 |
|
ff69416e6
|
963 |
free_block(cachep, ac->entry, ac->avail, node); |
e498be7da
|
964 965 966 967 |
ac->avail = 0; spin_unlock(&rl3->list_lock); } } |
8fce4d8e3
|
968 969 970 971 972 |
/* * Called from cache_reap() to regularly drain alien caches round robin. */ static void reap_alien(struct kmem_cache *cachep, struct kmem_list3 *l3) { |
909ea9646
|
973 |
int node = __this_cpu_read(slab_reap_node); |
8fce4d8e3
|
974 975 976 |
if (l3->alien) { struct array_cache *ac = l3->alien[node]; |
e00946fe2
|
977 978 |
if (ac && ac->avail && spin_trylock_irq(&ac->lock)) { |
8fce4d8e3
|
979 980 981 982 983 |
__drain_alien_cache(cachep, ac, node); spin_unlock_irq(&ac->lock); } } } |
a737b3e2f
|
984 985 |
static void drain_alien_cache(struct kmem_cache *cachep, struct array_cache **alien) |
e498be7da
|
986 |
{ |
b28a02de8
|
987 |
int i = 0; |
e498be7da
|
988 989 990 991 |
struct array_cache *ac; unsigned long flags; for_each_online_node(i) { |
4484ebf12
|
992 |
ac = alien[i]; |
e498be7da
|
993 994 995 996 997 998 999 |
if (ac) { spin_lock_irqsave(&ac->lock, flags); __drain_alien_cache(cachep, ac, i); spin_unlock_irqrestore(&ac->lock, flags); } } } |
729bd0b74
|
1000 |
|
873623dfa
|
1001 |
static inline int cache_free_alien(struct kmem_cache *cachep, void *objp) |
729bd0b74
|
1002 1003 1004 1005 1006 |
{ struct slab *slabp = virt_to_slab(objp); int nodeid = slabp->nodeid; struct kmem_list3 *l3; struct array_cache *alien = NULL; |
1ca4cb241
|
1007 |
int node; |
7d6e6d09d
|
1008 |
node = numa_mem_id(); |
729bd0b74
|
1009 1010 1011 1012 1013 |
/* * Make sure we are not freeing a object from another node to the array * cache on this cpu. */ |
62918a036
|
1014 |
if (likely(slabp->nodeid == node)) |
729bd0b74
|
1015 |
return 0; |
1ca4cb241
|
1016 |
l3 = cachep->nodelists[node]; |
729bd0b74
|
1017 1018 1019 |
STATS_INC_NODEFREES(cachep); if (l3->alien && l3->alien[nodeid]) { alien = l3->alien[nodeid]; |
873623dfa
|
1020 |
spin_lock(&alien->lock); |
729bd0b74
|
1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 |
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
|
1034 |
#endif |
8f9f8d9e8
|
1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 |
/* * 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
|
1080 1081 1082 1083 |
static void __cpuinit cpuup_canceled(long cpu) { struct kmem_cache *cachep; struct kmem_list3 *l3 = NULL; |
7d6e6d09d
|
1084 |
int node = cpu_to_mem(cpu); |
a70f73028
|
1085 |
const struct cpumask *mask = cpumask_of_node(node); |
fbf1e473b
|
1086 1087 1088 1089 1090 |
list_for_each_entry(cachep, &cache_chain, next) { struct array_cache *nc; struct array_cache *shared; struct array_cache **alien; |
fbf1e473b
|
1091 |
|
fbf1e473b
|
1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 |
/* 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
|
1106 |
if (!cpumask_empty(mask)) { |
fbf1e473b
|
1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 |
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
|
1145 |
{ |
343e0d7a9
|
1146 |
struct kmem_cache *cachep; |
e498be7da
|
1147 |
struct kmem_list3 *l3 = NULL; |
7d6e6d09d
|
1148 |
int node = cpu_to_mem(cpu); |
8f9f8d9e8
|
1149 |
int err; |
1da177e4c
|
1150 |
|
fbf1e473b
|
1151 1152 1153 1154 1155 1156 |
/* * 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
|
1157 1158 1159 |
err = init_cache_nodelists_node(node); if (err < 0) goto bad; |
fbf1e473b
|
1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 |
/* * 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
|
1171 |
cachep->batchcount, GFP_KERNEL); |
fbf1e473b
|
1172 1173 1174 1175 1176 |
if (!nc) goto bad; if (cachep->shared) { shared = alloc_arraycache(node, cachep->shared * cachep->batchcount, |
83b519e8b
|
1177 |
0xbaadf00d, GFP_KERNEL); |
12d00f6a1
|
1178 1179 |
if (!shared) { kfree(nc); |
1da177e4c
|
1180 |
goto bad; |
12d00f6a1
|
1181 |
} |
fbf1e473b
|
1182 1183 |
} if (use_alien_caches) { |
83b519e8b
|
1184 |
alien = alloc_alien_cache(node, cachep->limit, GFP_KERNEL); |
12d00f6a1
|
1185 1186 1187 |
if (!alien) { kfree(shared); kfree(nc); |
fbf1e473b
|
1188 |
goto bad; |
12d00f6a1
|
1189 |
} |
fbf1e473b
|
1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 |
} 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
|
1204 |
#ifdef CONFIG_NUMA |
fbf1e473b
|
1205 1206 1207 |
if (!l3->alien) { l3->alien = alien; alien = NULL; |
1da177e4c
|
1208 |
} |
fbf1e473b
|
1209 1210 1211 1212 1213 |
#endif spin_unlock_irq(&l3->list_lock); kfree(shared); free_alien_cache(alien); } |
ce79ddc8e
|
1214 |
init_node_lock_keys(node); |
fbf1e473b
|
1215 1216 |
return 0; bad: |
12d00f6a1
|
1217 |
cpuup_canceled(cpu); |
fbf1e473b
|
1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 |
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
|
1228 1229 |
case CPU_UP_PREPARE: case CPU_UP_PREPARE_FROZEN: |
95402b382
|
1230 |
mutex_lock(&cache_chain_mutex); |
fbf1e473b
|
1231 |
err = cpuup_prepare(cpu); |
95402b382
|
1232 |
mutex_unlock(&cache_chain_mutex); |
1da177e4c
|
1233 1234 |
break; case CPU_ONLINE: |
8bb784428
|
1235 |
case CPU_ONLINE_FROZEN: |
1da177e4c
|
1236 1237 1238 |
start_cpu_timer(cpu); break; #ifdef CONFIG_HOTPLUG_CPU |
5830c5902
|
1239 |
case CPU_DOWN_PREPARE: |
8bb784428
|
1240 |
case CPU_DOWN_PREPARE_FROZEN: |
5830c5902
|
1241 1242 1243 1244 1245 1246 |
/* * 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
|
1247 |
cancel_delayed_work_sync(&per_cpu(slab_reap_work, cpu)); |
5830c5902
|
1248 |
/* Now the cache_reaper is guaranteed to be not running. */ |
1871e52c7
|
1249 |
per_cpu(slab_reap_work, cpu).work.func = NULL; |
5830c5902
|
1250 1251 |
break; case CPU_DOWN_FAILED: |
8bb784428
|
1252 |
case CPU_DOWN_FAILED_FROZEN: |
5830c5902
|
1253 1254 |
start_cpu_timer(cpu); break; |
1da177e4c
|
1255 |
case CPU_DEAD: |
8bb784428
|
1256 |
case CPU_DEAD_FROZEN: |
4484ebf12
|
1257 1258 1259 1260 1261 1262 1263 1264 |
/* * 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
|
1265 |
/* fall through */ |
8f5be20bf
|
1266 |
#endif |
1da177e4c
|
1267 |
case CPU_UP_CANCELED: |
8bb784428
|
1268 |
case CPU_UP_CANCELED_FROZEN: |
95402b382
|
1269 |
mutex_lock(&cache_chain_mutex); |
fbf1e473b
|
1270 |
cpuup_canceled(cpu); |
fc0abb145
|
1271 |
mutex_unlock(&cache_chain_mutex); |
1da177e4c
|
1272 |
break; |
1da177e4c
|
1273 |
} |
eac406801
|
1274 |
return notifier_from_errno(err); |
1da177e4c
|
1275 |
} |
74b85f379
|
1276 1277 1278 |
static struct notifier_block __cpuinitdata cpucache_notifier = { &cpuup_callback, NULL, 0 }; |
1da177e4c
|
1279 |
|
8f9f8d9e8
|
1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 |
#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
|
1340 |
return notifier_from_errno(ret); |
8f9f8d9e8
|
1341 1342 |
} #endif /* CONFIG_NUMA && CONFIG_MEMORY_HOTPLUG */ |
e498be7da
|
1343 1344 1345 |
/* * swap the static kmem_list3 with kmalloced memory */ |
8f9f8d9e8
|
1346 1347 |
static void __init init_list(struct kmem_cache *cachep, struct kmem_list3 *list, int nodeid) |
e498be7da
|
1348 1349 |
{ struct kmem_list3 *ptr; |
83b519e8b
|
1350 |
ptr = kmalloc_node(sizeof(struct kmem_list3), GFP_NOWAIT, nodeid); |
e498be7da
|
1351 |
BUG_ON(!ptr); |
e498be7da
|
1352 |
memcpy(ptr, list, sizeof(struct kmem_list3)); |
2b2d5493e
|
1353 1354 1355 1356 |
/* * Do not assume that spinlocks can be initialized via memcpy: */ spin_lock_init(&ptr->list_lock); |
e498be7da
|
1357 1358 |
MAKE_ALL_LISTS(cachep, ptr, nodeid); cachep->nodelists[nodeid] = ptr; |
e498be7da
|
1359 |
} |
a737b3e2f
|
1360 |
/* |
556a169da
|
1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 |
* 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
|
1377 1378 |
* Initialisation. Called after the page allocator have been initialised and * before smp_init(). |
1da177e4c
|
1379 1380 1381 1382 1383 1384 |
*/ void __init kmem_cache_init(void) { size_t left_over; struct cache_sizes *sizes; struct cache_names *names; |
e498be7da
|
1385 |
int i; |
07ed76b2a
|
1386 |
int order; |
1ca4cb241
|
1387 |
int node; |
e498be7da
|
1388 |
|
b6e68bc1b
|
1389 |
if (num_possible_nodes() == 1) |
62918a036
|
1390 |
use_alien_caches = 0; |
e498be7da
|
1391 1392 1393 1394 1395 |
for (i = 0; i < NUM_INIT_LISTS; i++) { kmem_list3_init(&initkmem_list3[i]); if (i < MAX_NUMNODES) cache_cache.nodelists[i] = NULL; } |
556a169da
|
1396 |
set_up_list3s(&cache_cache, CACHE_CACHE); |
1da177e4c
|
1397 1398 1399 1400 1401 |
/* * Fragmentation resistance on low memory - only use bigger * page orders on machines with more than 32MB of memory. */ |
4481374ce
|
1402 |
if (totalram_pages > (32 << 20) >> PAGE_SHIFT) |
1da177e4c
|
1403 |
slab_break_gfp_order = BREAK_GFP_ORDER_HI; |
1da177e4c
|
1404 1405 |
/* Bootstrap is tricky, because several objects are allocated * from caches that do not exist yet: |
a737b3e2f
|
1406 1407 1408 |
* 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
|
1409 1410 1411 |
* 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
|
1412 |
* 2) Create the first kmalloc cache. |
343e0d7a9
|
1413 |
* The struct kmem_cache for the new cache is allocated normally. |
e498be7da
|
1414 1415 1416 |
* An __init data area is used for the head array. * 3) Create the remaining kmalloc caches, with minimally sized * head arrays. |
1da177e4c
|
1417 1418 |
* 4) Replace the __init data head arrays for cache_cache and the first * kmalloc cache with kmalloc allocated arrays. |
e498be7da
|
1419 1420 1421 |
* 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
|
1422 |
*/ |
7d6e6d09d
|
1423 |
node = numa_mem_id(); |
1ca4cb241
|
1424 |
|
1da177e4c
|
1425 |
/* 1) create the cache_cache */ |
1da177e4c
|
1426 1427 1428 1429 |
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
|
1430 |
cache_cache.nodelists[node] = &initkmem_list3[CACHE_CACHE + node]; |
1da177e4c
|
1431 |
|
8da3430d8
|
1432 1433 1434 1435 1436 1437 1438 1439 1440 |
/* * struct kmem_cache size depends on nr_node_ids, which * can be less than MAX_NUMNODES. */ cache_cache.buffer_size = offsetof(struct kmem_cache, nodelists) + nr_node_ids * sizeof(struct kmem_list3 *); #if DEBUG cache_cache.obj_size = cache_cache.buffer_size; #endif |
a737b3e2f
|
1441 1442 |
cache_cache.buffer_size = ALIGN(cache_cache.buffer_size, cache_line_size()); |
6a2d7a955
|
1443 1444 |
cache_cache.reciprocal_buffer_size = reciprocal_value(cache_cache.buffer_size); |
1da177e4c
|
1445 |
|
07ed76b2a
|
1446 1447 1448 1449 1450 1451 |
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
|
1452 |
BUG_ON(!cache_cache.num); |
07ed76b2a
|
1453 |
cache_cache.gfporder = order; |
b28a02de8
|
1454 |
cache_cache.colour = left_over / cache_cache.colour_off; |
b28a02de8
|
1455 1456 |
cache_cache.slab_size = ALIGN(cache_cache.num * sizeof(kmem_bufctl_t) + sizeof(struct slab), cache_line_size()); |
1da177e4c
|
1457 1458 1459 1460 |
/* 2+3) create the kmalloc caches */ sizes = malloc_sizes; names = cache_names; |
a737b3e2f
|
1461 1462 1463 1464 |
/* * Initialize the caches that provide memory for the array cache and the * kmem_list3 structures first. Without this, further allocations will * bug. |
e498be7da
|
1465 1466 1467 |
*/ sizes[INDEX_AC].cs_cachep = kmem_cache_create(names[INDEX_AC].name, |
a737b3e2f
|
1468 1469 1470 |
sizes[INDEX_AC].cs_size, ARCH_KMALLOC_MINALIGN, ARCH_KMALLOC_FLAGS|SLAB_PANIC, |
20c2df83d
|
1471 |
NULL); |
e498be7da
|
1472 |
|
a737b3e2f
|
1473 |
if (INDEX_AC != INDEX_L3) { |
e498be7da
|
1474 |
sizes[INDEX_L3].cs_cachep = |
a737b3e2f
|
1475 1476 1477 1478 |
kmem_cache_create(names[INDEX_L3].name, sizes[INDEX_L3].cs_size, ARCH_KMALLOC_MINALIGN, ARCH_KMALLOC_FLAGS|SLAB_PANIC, |
20c2df83d
|
1479 |
NULL); |
a737b3e2f
|
1480 |
} |
e498be7da
|
1481 |
|
e0a427267
|
1482 |
slab_early_init = 0; |
1da177e4c
|
1483 |
while (sizes->cs_size != ULONG_MAX) { |
e498be7da
|
1484 1485 |
/* * For performance, all the general caches are L1 aligned. |
1da177e4c
|
1486 1487 1488 |
* This should be particularly beneficial on SMP boxes, as it * eliminates "false sharing". * Note for systems short on memory removing the alignment will |
e498be7da
|
1489 1490 |
* allow tighter packing of the smaller caches. */ |
a737b3e2f
|
1491 |
if (!sizes->cs_cachep) { |
e498be7da
|
1492 |
sizes->cs_cachep = kmem_cache_create(names->name, |
a737b3e2f
|
1493 1494 1495 |
sizes->cs_size, ARCH_KMALLOC_MINALIGN, ARCH_KMALLOC_FLAGS|SLAB_PANIC, |
20c2df83d
|
1496 |
NULL); |
a737b3e2f
|
1497 |
} |
4b51d6698
|
1498 1499 1500 |
#ifdef CONFIG_ZONE_DMA sizes->cs_dmacachep = kmem_cache_create( names->name_dma, |
a737b3e2f
|
1501 1502 1503 1504 |
sizes->cs_size, ARCH_KMALLOC_MINALIGN, ARCH_KMALLOC_FLAGS|SLAB_CACHE_DMA| SLAB_PANIC, |
20c2df83d
|
1505 |
NULL); |
4b51d6698
|
1506 |
#endif |
1da177e4c
|
1507 1508 1509 1510 1511 |
sizes++; names++; } /* 4) Replace the bootstrap head arrays */ { |
2b2d5493e
|
1512 |
struct array_cache *ptr; |
e498be7da
|
1513 |
|
83b519e8b
|
1514 |
ptr = kmalloc(sizeof(struct arraycache_init), GFP_NOWAIT); |
e498be7da
|
1515 |
|
9a2dba4b4
|
1516 1517 |
BUG_ON(cpu_cache_get(&cache_cache) != &initarray_cache.cache); memcpy(ptr, cpu_cache_get(&cache_cache), |
b28a02de8
|
1518 |
sizeof(struct arraycache_init)); |
2b2d5493e
|
1519 1520 1521 1522 |
/* * Do not assume that spinlocks can be initialized via memcpy: */ spin_lock_init(&ptr->lock); |
1da177e4c
|
1523 |
cache_cache.array[smp_processor_id()] = ptr; |
e498be7da
|
1524 |
|
83b519e8b
|
1525 |
ptr = kmalloc(sizeof(struct arraycache_init), GFP_NOWAIT); |
e498be7da
|
1526 |
|
9a2dba4b4
|
1527 |
BUG_ON(cpu_cache_get(malloc_sizes[INDEX_AC].cs_cachep) |
b28a02de8
|
1528 |
!= &initarray_generic.cache); |
9a2dba4b4
|
1529 |
memcpy(ptr, cpu_cache_get(malloc_sizes[INDEX_AC].cs_cachep), |
b28a02de8
|
1530 |
sizeof(struct arraycache_init)); |
2b2d5493e
|
1531 1532 1533 1534 |
/* * Do not assume that spinlocks can be initialized via memcpy: */ spin_lock_init(&ptr->lock); |
e498be7da
|
1535 |
malloc_sizes[INDEX_AC].cs_cachep->array[smp_processor_id()] = |
b28a02de8
|
1536 |
ptr; |
1da177e4c
|
1537 |
} |
e498be7da
|
1538 1539 |
/* 5) Replace the bootstrap kmem_list3's */ { |
1ca4cb241
|
1540 |
int nid; |
9c09a95cf
|
1541 |
for_each_online_node(nid) { |
ec1f5eeeb
|
1542 |
init_list(&cache_cache, &initkmem_list3[CACHE_CACHE + nid], nid); |
556a169da
|
1543 |
|
e498be7da
|
1544 |
init_list(malloc_sizes[INDEX_AC].cs_cachep, |
1ca4cb241
|
1545 |
&initkmem_list3[SIZE_AC + nid], nid); |
e498be7da
|
1546 1547 1548 |
if (INDEX_AC != INDEX_L3) { init_list(malloc_sizes[INDEX_L3].cs_cachep, |
1ca4cb241
|
1549 |
&initkmem_list3[SIZE_L3 + nid], nid); |
e498be7da
|
1550 1551 1552 |
} } } |
1da177e4c
|
1553 |
|
8429db5c6
|
1554 |
g_cpucache_up = EARLY; |
8429db5c6
|
1555 1556 1557 1558 1559 |
} void __init kmem_cache_init_late(void) { struct kmem_cache *cachep; |
8429db5c6
|
1560 1561 1562 1563 1564 1565 |
/* 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
|
1566 |
|
1da177e4c
|
1567 1568 |
/* Done! */ g_cpucache_up = FULL; |
ec5a36f94
|
1569 1570 |
/* Annotate slab for lockdep -- annotate the malloc caches */ init_lock_keys(); |
a737b3e2f
|
1571 1572 1573 |
/* * Register a cpu startup notifier callback that initializes * cpu_cache_get for all new cpus |
1da177e4c
|
1574 1575 |
*/ register_cpu_notifier(&cpucache_notifier); |
1da177e4c
|
1576 |
|
8f9f8d9e8
|
1577 1578 1579 1580 1581 1582 1583 |
#ifdef CONFIG_NUMA /* * Register a memory hotplug callback that initializes and frees * nodelists. */ hotplug_memory_notifier(slab_memory_callback, SLAB_CALLBACK_PRI); #endif |
a737b3e2f
|
1584 1585 1586 |
/* * The reap timers are started later, with a module init call: That part * of the kernel is not yet operational. |
1da177e4c
|
1587 1588 1589 1590 1591 1592 |
*/ } static int __init cpucache_init(void) { int cpu; |
a737b3e2f
|
1593 1594 |
/* * Register the timers that return unneeded pages to the page allocator |
1da177e4c
|
1595 |
*/ |
e498be7da
|
1596 |
for_each_online_cpu(cpu) |
a737b3e2f
|
1597 |
start_cpu_timer(cpu); |
1da177e4c
|
1598 1599 |
return 0; } |
1da177e4c
|
1600 1601 1602 1603 1604 1605 1606 1607 1608 |
__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
|
1609 |
static void *kmem_getpages(struct kmem_cache *cachep, gfp_t flags, int nodeid) |
1da177e4c
|
1610 1611 |
{ struct page *page; |
e1b6aa6f1
|
1612 |
int nr_pages; |
1da177e4c
|
1613 |
int i; |
d6fef9da1
|
1614 |
#ifndef CONFIG_MMU |
e1b6aa6f1
|
1615 1616 1617 |
/* * Nommu uses slab's for process anonymous memory allocations, and thus * requires __GFP_COMP to properly refcount higher order allocations |
d6fef9da1
|
1618 |
*/ |
e1b6aa6f1
|
1619 |
flags |= __GFP_COMP; |
d6fef9da1
|
1620 |
#endif |
765c4507a
|
1621 |
|
3c517a613
|
1622 |
flags |= cachep->gfpflags; |
e12ba74d8
|
1623 1624 |
if (cachep->flags & SLAB_RECLAIM_ACCOUNT) flags |= __GFP_RECLAIMABLE; |
e1b6aa6f1
|
1625 |
|
517d08699
|
1626 |
page = alloc_pages_exact_node(nodeid, flags | __GFP_NOTRACK, cachep->gfporder); |
1da177e4c
|
1627 1628 |
if (!page) return NULL; |
1da177e4c
|
1629 |
|
e1b6aa6f1
|
1630 |
nr_pages = (1 << cachep->gfporder); |
1da177e4c
|
1631 |
if (cachep->flags & SLAB_RECLAIM_ACCOUNT) |
972d1a7b1
|
1632 1633 1634 1635 1636 |
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
|
1637 1638 |
for (i = 0; i < nr_pages; i++) __SetPageSlab(page + i); |
c175eea46
|
1639 |
|
b1eeab676
|
1640 1641 1642 1643 1644 1645 1646 1647 |
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
|
1648 |
|
e1b6aa6f1
|
1649 |
return page_address(page); |
1da177e4c
|
1650 1651 1652 1653 1654 |
} /* * Interface to system's page release. */ |
343e0d7a9
|
1655 |
static void kmem_freepages(struct kmem_cache *cachep, void *addr) |
1da177e4c
|
1656 |
{ |
b28a02de8
|
1657 |
unsigned long i = (1 << cachep->gfporder); |
1da177e4c
|
1658 1659 |
struct page *page = virt_to_page(addr); const unsigned long nr_freed = i; |
b1eeab676
|
1660 |
kmemcheck_free_shadow(page, cachep->gfporder); |
c175eea46
|
1661 |
|
972d1a7b1
|
1662 1663 1664 1665 1666 1667 |
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
|
1668 |
while (i--) { |
f205b2fe6
|
1669 1670 |
BUG_ON(!PageSlab(page)); __ClearPageSlab(page); |
1da177e4c
|
1671 1672 |
page++; } |
1da177e4c
|
1673 1674 1675 |
if (current->reclaim_state) current->reclaim_state->reclaimed_slab += nr_freed; free_pages((unsigned long)addr, cachep->gfporder); |
1da177e4c
|
1676 1677 1678 1679 |
} static void kmem_rcu_free(struct rcu_head *head) { |
b28a02de8
|
1680 |
struct slab_rcu *slab_rcu = (struct slab_rcu *)head; |
343e0d7a9
|
1681 |
struct kmem_cache *cachep = slab_rcu->cachep; |
1da177e4c
|
1682 1683 1684 1685 1686 1687 1688 1689 1690 |
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
|
1691 |
static void store_stackinfo(struct kmem_cache *cachep, unsigned long *addr, |
b28a02de8
|
1692 |
unsigned long caller) |
1da177e4c
|
1693 |
{ |
3dafccf22
|
1694 |
int size = obj_size(cachep); |
1da177e4c
|
1695 |
|
3dafccf22
|
1696 |
addr = (unsigned long *)&((char *)addr)[obj_offset(cachep)]; |
1da177e4c
|
1697 |
|
b28a02de8
|
1698 |
if (size < 5 * sizeof(unsigned long)) |
1da177e4c
|
1699 |
return; |
b28a02de8
|
1700 1701 1702 1703 |
*addr++ = 0x12345678; *addr++ = caller; *addr++ = smp_processor_id(); size -= 3 * sizeof(unsigned long); |
1da177e4c
|
1704 1705 1706 1707 1708 1709 1710 |
{ unsigned long *sptr = &caller; unsigned long svalue; while (!kstack_end(sptr)) { svalue = *sptr++; if (kernel_text_address(svalue)) { |
b28a02de8
|
1711 |
*addr++ = svalue; |
1da177e4c
|
1712 1713 1714 1715 1716 1717 1718 |
size -= sizeof(unsigned long); if (size <= sizeof(unsigned long)) break; } } } |
b28a02de8
|
1719 |
*addr++ = 0x87654321; |
1da177e4c
|
1720 1721 |
} #endif |
343e0d7a9
|
1722 |
static void poison_obj(struct kmem_cache *cachep, void *addr, unsigned char val) |
1da177e4c
|
1723 |
{ |
3dafccf22
|
1724 1725 |
int size = obj_size(cachep); addr = &((char *)addr)[obj_offset(cachep)]; |
1da177e4c
|
1726 1727 |
memset(addr, val, size); |
b28a02de8
|
1728 |
*(unsigned char *)(addr + size - 1) = POISON_END; |
1da177e4c
|
1729 1730 1731 1732 1733 |
} static void dump_line(char *data, int offset, int limit) { int i; |
aa83aa40e
|
1734 1735 |
unsigned char error = 0; int bad_count = 0; |
1da177e4c
|
1736 |
printk(KERN_ERR "%03x:", offset); |
aa83aa40e
|
1737 1738 1739 1740 1741 |
for (i = 0; i < limit; i++) { if (data[offset + i] != POISON_FREE) { error = data[offset + i]; bad_count++; } |
b28a02de8
|
1742 |
printk(" %02x", (unsigned char)data[offset + i]); |
aa83aa40e
|
1743 |
} |
1da177e4c
|
1744 1745 |
printk(" "); |
aa83aa40e
|
1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 |
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
|
1763 1764 1765 1766 |
} #endif #if DEBUG |
343e0d7a9
|
1767 |
static void print_objinfo(struct kmem_cache *cachep, void *objp, int lines) |
1da177e4c
|
1768 1769 1770 1771 1772 |
{ int i, size; char *realobj; if (cachep->flags & SLAB_RED_ZONE) { |
b46b8f19c
|
1773 1774 |
printk(KERN_ERR "Redzone: 0x%llx/0x%llx. ", |
a737b3e2f
|
1775 1776 |
*dbg_redzone1(cachep, objp), *dbg_redzone2(cachep, objp)); |
1da177e4c
|
1777 1778 1779 1780 |
} if (cachep->flags & SLAB_STORE_USER) { printk(KERN_ERR "Last user: [<%p>]", |
a737b3e2f
|
1781 |
*dbg_userword(cachep, objp)); |
1da177e4c
|
1782 |
print_symbol("(%s)", |
a737b3e2f
|
1783 |
(unsigned long)*dbg_userword(cachep, objp)); |
1da177e4c
|
1784 1785 1786 |
printk(" "); } |
3dafccf22
|
1787 1788 |
realobj = (char *)objp + obj_offset(cachep); size = obj_size(cachep); |
b28a02de8
|
1789 |
for (i = 0; i < size && lines; i += 16, lines--) { |
1da177e4c
|
1790 1791 |
int limit; limit = 16; |
b28a02de8
|
1792 1793 |
if (i + limit > size) limit = size - i; |
1da177e4c
|
1794 1795 1796 |
dump_line(realobj, i, limit); } } |
343e0d7a9
|
1797 |
static void check_poison_obj(struct kmem_cache *cachep, void *objp) |
1da177e4c
|
1798 1799 1800 1801 |
{ char *realobj; int size, i; int lines = 0; |
3dafccf22
|
1802 1803 |
realobj = (char *)objp + obj_offset(cachep); size = obj_size(cachep); |
1da177e4c
|
1804 |
|
b28a02de8
|
1805 |
for (i = 0; i < size; i++) { |
1da177e4c
|
1806 |
char exp = POISON_FREE; |
b28a02de8
|
1807 |
if (i == size - 1) |
1da177e4c
|
1808 1809 1810 1811 1812 1813 |
exp = POISON_END; if (realobj[i] != exp) { int limit; /* Mismatch ! */ /* Print header */ if (lines == 0) { |
b28a02de8
|
1814 |
printk(KERN_ERR |
e94a40c50
|
1815 1816 1817 |
"Slab corruption: %s start=%p, len=%d ", cachep->name, realobj, size); |
1da177e4c
|
1818 1819 1820 |
print_objinfo(cachep, objp, 0); } /* Hexdump the affected line */ |
b28a02de8
|
1821 |
i = (i / 16) * 16; |
1da177e4c
|
1822 |
limit = 16; |
b28a02de8
|
1823 1824 |
if (i + limit > size) limit = size - i; |
1da177e4c
|
1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 |
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
|
1837 |
struct slab *slabp = virt_to_slab(objp); |
8fea4e96a
|
1838 |
unsigned int objnr; |
1da177e4c
|
1839 |
|
8fea4e96a
|
1840 |
objnr = obj_to_index(cachep, slabp, objp); |
1da177e4c
|
1841 |
if (objnr) { |
8fea4e96a
|
1842 |
objp = index_to_obj(cachep, slabp, objnr - 1); |
3dafccf22
|
1843 |
realobj = (char *)objp + obj_offset(cachep); |
1da177e4c
|
1844 1845 |
printk(KERN_ERR "Prev obj: start=%p, len=%d ", |
b28a02de8
|
1846 |
realobj, size); |
1da177e4c
|
1847 1848 |
print_objinfo(cachep, objp, 2); } |
b28a02de8
|
1849 |
if (objnr + 1 < cachep->num) { |
8fea4e96a
|
1850 |
objp = index_to_obj(cachep, slabp, objnr + 1); |
3dafccf22
|
1851 |
realobj = (char *)objp + obj_offset(cachep); |
1da177e4c
|
1852 1853 |
printk(KERN_ERR "Next obj: start=%p, len=%d ", |
b28a02de8
|
1854 |
realobj, size); |
1da177e4c
|
1855 1856 1857 1858 1859 |
print_objinfo(cachep, objp, 2); } } } #endif |
12dd36fae
|
1860 |
#if DEBUG |
e79aec291
|
1861 |
static void slab_destroy_debugcheck(struct kmem_cache *cachep, struct slab *slabp) |
1da177e4c
|
1862 |
{ |
1da177e4c
|
1863 1864 |
int i; for (i = 0; i < cachep->num; i++) { |
8fea4e96a
|
1865 |
void *objp = index_to_obj(cachep, slabp, i); |
1da177e4c
|
1866 1867 1868 |
if (cachep->flags & SLAB_POISON) { #ifdef CONFIG_DEBUG_PAGEALLOC |
a737b3e2f
|
1869 1870 |
if (cachep->buffer_size % PAGE_SIZE == 0 && OFF_SLAB(cachep)) |
b28a02de8
|
1871 |
kernel_map_pages(virt_to_page(objp), |
a737b3e2f
|
1872 |
cachep->buffer_size / PAGE_SIZE, 1); |
1da177e4c
|
1873 1874 1875 1876 1877 1878 1879 1880 1881 |
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
|
1882 |
"was overwritten"); |
1da177e4c
|
1883 1884 |
if (*dbg_redzone2(cachep, objp) != RED_INACTIVE) slab_error(cachep, "end of a freed object " |
b28a02de8
|
1885 |
"was overwritten"); |
1da177e4c
|
1886 |
} |
1da177e4c
|
1887 |
} |
12dd36fae
|
1888 |
} |
1da177e4c
|
1889 |
#else |
e79aec291
|
1890 |
static void slab_destroy_debugcheck(struct kmem_cache *cachep, struct slab *slabp) |
12dd36fae
|
1891 |
{ |
12dd36fae
|
1892 |
} |
1da177e4c
|
1893 |
#endif |
911851e6e
|
1894 1895 1896 1897 1898 |
/** * slab_destroy - destroy and release all objects in a slab * @cachep: cache pointer being destroyed * @slabp: slab pointer being destroyed * |
12dd36fae
|
1899 |
* Destroy all the objs in a slab, and release the mem back to the system. |
a737b3e2f
|
1900 1901 |
* Before calling the slab must have been unlinked from the cache. The * cache-lock is not held/needed. |
12dd36fae
|
1902 |
*/ |
343e0d7a9
|
1903 |
static void slab_destroy(struct kmem_cache *cachep, struct slab *slabp) |
12dd36fae
|
1904 1905 |
{ void *addr = slabp->s_mem - slabp->colouroff; |
e79aec291
|
1906 |
slab_destroy_debugcheck(cachep, slabp); |
1da177e4c
|
1907 1908 |
if (unlikely(cachep->flags & SLAB_DESTROY_BY_RCU)) { struct slab_rcu *slab_rcu; |
b28a02de8
|
1909 |
slab_rcu = (struct slab_rcu *)slabp; |
1da177e4c
|
1910 1911 1912 1913 1914 |
slab_rcu->cachep = cachep; slab_rcu->addr = addr; call_rcu(&slab_rcu->head, kmem_rcu_free); } else { kmem_freepages(cachep, addr); |
873623dfa
|
1915 1916 |
if (OFF_SLAB(cachep)) kmem_cache_free(cachep->slabp_cache, slabp); |
1da177e4c
|
1917 1918 |
} } |
117f6eb1d
|
1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 |
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
|
1938 |
/** |
a70773ddb
|
1939 1940 1941 1942 1943 1944 1945 |
* 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
|
1946 1947 1948 1949 1950 |
* * 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
|
1951 |
static size_t calculate_slab_order(struct kmem_cache *cachep, |
ee13d785e
|
1952 |
size_t size, size_t align, unsigned long flags) |
4d268eba1
|
1953 |
{ |
b1ab41c49
|
1954 |
unsigned long offslab_limit; |
4d268eba1
|
1955 |
size_t left_over = 0; |
9888e6fa7
|
1956 |
int gfporder; |
4d268eba1
|
1957 |
|
0aa817f07
|
1958 |
for (gfporder = 0; gfporder <= KMALLOC_MAX_ORDER; gfporder++) { |
4d268eba1
|
1959 1960 |
unsigned int num; size_t remainder; |
9888e6fa7
|
1961 |
cache_estimate(gfporder, size, align, flags, &remainder, &num); |
4d268eba1
|
1962 1963 |
if (!num) continue; |
9888e6fa7
|
1964 |
|
b1ab41c49
|
1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 |
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
|
1977 |
|
9888e6fa7
|
1978 |
/* Found something acceptable - save it away */ |
4d268eba1
|
1979 |
cachep->num = num; |
9888e6fa7
|
1980 |
cachep->gfporder = gfporder; |
4d268eba1
|
1981 1982 1983 |
left_over = remainder; /* |
f78bb8ad4
|
1984 1985 1986 1987 1988 1989 1990 1991 |
* 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
|
1992 1993 1994 |
* Large number of objects is good, but very large slabs are * currently bad for the gfp()s. */ |
9888e6fa7
|
1995 |
if (gfporder >= slab_break_gfp_order) |
4d268eba1
|
1996 |
break; |
9888e6fa7
|
1997 1998 1999 |
/* * Acceptable internal fragmentation? */ |
a737b3e2f
|
2000 |
if (left_over * 8 <= (PAGE_SIZE << gfporder)) |
4d268eba1
|
2001 2002 2003 2004 |
break; } return left_over; } |
83b519e8b
|
2005 |
static int __init_refok setup_cpu_cache(struct kmem_cache *cachep, gfp_t gfp) |
f30cf7d13
|
2006 |
{ |
2ed3a4ef9
|
2007 |
if (g_cpucache_up == FULL) |
83b519e8b
|
2008 |
return enable_cpucache(cachep, gfp); |
2ed3a4ef9
|
2009 |
|
f30cf7d13
|
2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 |
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
|
2030 |
kmalloc(sizeof(struct arraycache_init), gfp); |
f30cf7d13
|
2031 2032 2033 2034 2035 2036 |
if (g_cpucache_up == PARTIAL_AC) { set_up_list3s(cachep, SIZE_L3); g_cpucache_up = PARTIAL_L3; } else { int node; |
556a169da
|
2037 |
for_each_online_node(node) { |
f30cf7d13
|
2038 2039 |
cachep->nodelists[node] = kmalloc_node(sizeof(struct kmem_list3), |
eb91f1d0a
|
2040 |
gfp, node); |
f30cf7d13
|
2041 2042 2043 2044 2045 |
BUG_ON(!cachep->nodelists[node]); kmem_list3_init(cachep->nodelists[node]); } } } |
7d6e6d09d
|
2046 |
cachep->nodelists[numa_mem_id()]->next_reap = |
f30cf7d13
|
2047 2048 2049 2050 2051 2052 2053 2054 2055 |
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
|
2056 |
return 0; |
f30cf7d13
|
2057 |
} |
4d268eba1
|
2058 |
/** |
1da177e4c
|
2059 2060 2061 2062 2063 2064 |
* 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
|
2065 2066 2067 |
* * Returns a ptr to the cache on success, NULL on failure. * Cannot be called within a int, but can be interrupted. |
20c2df83d
|
2068 |
* The @ctor is run when new pages are allocated by the cache. |
1da177e4c
|
2069 2070 |
* * @name must be valid until the cache is destroyed. This implies that |
a737b3e2f
|
2071 2072 |
* the module calling this has to destroy the cache before getting unloaded. * |
1da177e4c
|
2073 2074 2075 2076 2077 2078 2079 2080 |
* 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
|
2081 2082 2083 2084 |
* %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
|
2085 |
struct kmem_cache * |
1da177e4c
|
2086 |
kmem_cache_create (const char *name, size_t size, size_t align, |
51cc50685
|
2087 |
unsigned long flags, void (*ctor)(void *)) |
1da177e4c
|
2088 2089 |
{ size_t left_over, slab_size, ralign; |
7a7c381d2
|
2090 |
struct kmem_cache *cachep = NULL, *pc; |
83b519e8b
|
2091 |
gfp_t gfp; |
1da177e4c
|
2092 2093 2094 2095 |
/* * Sanity checks... these are all serious usage bugs. */ |
a737b3e2f
|
2096 |
if (!name || in_interrupt() || (size < BYTES_PER_WORD) || |
20c2df83d
|
2097 |
size > KMALLOC_MAX_SIZE) { |
d40cee245
|
2098 2099 |
printk(KERN_ERR "%s: Early error in slab %s ", __func__, |
a737b3e2f
|
2100 |
name); |
b28a02de8
|
2101 2102 |
BUG(); } |
1da177e4c
|
2103 |
|
f0188f474
|
2104 |
/* |
8f5be20bf
|
2105 |
* We use cache_chain_mutex to ensure a consistent view of |
174596a0b
|
2106 |
* cpu_online_mask as well. Please see cpuup_callback |
f0188f474
|
2107 |
*/ |
83b519e8b
|
2108 2109 2110 2111 |
if (slab_is_available()) { get_online_cpus(); mutex_lock(&cache_chain_mutex); } |
4f12bb4f7
|
2112 |
|
7a7c381d2
|
2113 |
list_for_each_entry(pc, &cache_chain, next) { |
4f12bb4f7
|
2114 2115 2116 2117 2118 2119 2120 2121 |
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
|
2122 |
res = probe_kernel_address(pc->name, tmp); |
4f12bb4f7
|
2123 |
if (res) { |
b4169525b
|
2124 2125 2126 |
printk(KERN_ERR "SLAB: cache with size %d has lost its name ", |
3dafccf22
|
2127 |
pc->buffer_size); |
4f12bb4f7
|
2128 2129 |
continue; } |
b28a02de8
|
2130 |
if (!strcmp(pc->name, name)) { |
b4169525b
|
2131 2132 2133 |
printk(KERN_ERR "kmem_cache_create: duplicate cache %s ", name); |
4f12bb4f7
|
2134 2135 2136 2137 |
dump_stack(); goto oops; } } |
1da177e4c
|
2138 2139 |
#if DEBUG WARN_ON(strchr(name, ' ')); /* It confuses parsers */ |
1da177e4c
|
2140 2141 2142 2143 2144 2145 2146 |
#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
|
2147 2148 |
if (size < 4096 || fls(size - 1) == fls(size-1 + REDZONE_ALIGN + 2 * sizeof(unsigned long long))) |
b28a02de8
|
2149 |
flags |= SLAB_RED_ZONE | SLAB_STORE_USER; |
1da177e4c
|
2150 2151 2152 2153 2154 2155 |
if (!(flags & SLAB_DESTROY_BY_RCU)) flags |= SLAB_POISON; #endif if (flags & SLAB_DESTROY_BY_RCU) BUG_ON(flags & SLAB_POISON); #endif |
1da177e4c
|
2156 |
/* |
a737b3e2f
|
2157 2158 |
* Always checks flags, a caller might be expecting debug support which * isn't available. |
1da177e4c
|
2159 |
*/ |
40094fa65
|
2160 |
BUG_ON(flags & ~CREATE_MASK); |
1da177e4c
|
2161 |
|
a737b3e2f
|
2162 2163 |
/* * Check that size is in terms of words. This is needed to avoid |
1da177e4c
|
2164 2165 2166 |
* unaligned accesses for some archs when redzoning is used, and makes * sure any on-slab bufctl's are also correctly aligned. */ |
b28a02de8
|
2167 2168 2169 |
if (size & (BYTES_PER_WORD - 1)) { size += (BYTES_PER_WORD - 1); size &= ~(BYTES_PER_WORD - 1); |
1da177e4c
|
2170 |
} |
a737b3e2f
|
2171 |
/* calculate the final buffer alignment: */ |
1da177e4c
|
2172 2173 |
/* 1) arch recommendation: can be overridden for debug */ if (flags & SLAB_HWCACHE_ALIGN) { |
a737b3e2f
|
2174 2175 2176 2177 |
/* * Default alignment: as specified by the arch code. Except if * an object is really small, then squeeze multiple objects into * one cacheline. |
1da177e4c
|
2178 2179 |
*/ ralign = cache_line_size(); |
b28a02de8
|
2180 |
while (size <= ralign / 2) |
1da177e4c
|
2181 2182 2183 2184 |
ralign /= 2; } else { ralign = BYTES_PER_WORD; } |
ca5f9703d
|
2185 2186 |
/* |
87a927c71
|
2187 2188 2189 |
* 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
|
2190 |
*/ |
87a927c71
|
2191 2192 2193 2194 2195 2196 2197 2198 2199 2200 |
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
|
2201 |
|
a44b56d35
|
2202 |
/* 2) arch mandated alignment */ |
1da177e4c
|
2203 2204 |
if (ralign < ARCH_SLAB_MINALIGN) { ralign = ARCH_SLAB_MINALIGN; |
1da177e4c
|
2205 |
} |
a44b56d35
|
2206 |
/* 3) caller mandated alignment */ |
1da177e4c
|
2207 2208 |
if (ralign < align) { ralign = align; |
1da177e4c
|
2209 |
} |
3ff84a7f3
|
2210 2211 |
/* disable debug if necessary */ if (ralign > __alignof__(unsigned long long)) |
a44b56d35
|
2212 |
flags &= ~(SLAB_RED_ZONE | SLAB_STORE_USER); |
a737b3e2f
|
2213 |
/* |
ca5f9703d
|
2214 |
* 4) Store it. |
1da177e4c
|
2215 2216 |
*/ align = ralign; |
83b519e8b
|
2217 2218 2219 2220 |
if (slab_is_available()) gfp = GFP_KERNEL; else gfp = GFP_NOWAIT; |
1da177e4c
|
2221 |
/* Get cache's description obj. */ |
83b519e8b
|
2222 |
cachep = kmem_cache_zalloc(&cache_cache, gfp); |
1da177e4c
|
2223 |
if (!cachep) |
4f12bb4f7
|
2224 |
goto oops; |
1da177e4c
|
2225 2226 |
#if DEBUG |
3dafccf22
|
2227 |
cachep->obj_size = size; |
1da177e4c
|
2228 |
|
ca5f9703d
|
2229 2230 2231 2232 |
/* * Both debugging options require word-alignment which is calculated * into align above. */ |
1da177e4c
|
2233 |
if (flags & SLAB_RED_ZONE) { |
1da177e4c
|
2234 |
/* add space for red zone words */ |
3ff84a7f3
|
2235 2236 |
cachep->obj_offset += sizeof(unsigned long long); size += 2 * sizeof(unsigned long long); |
1da177e4c
|
2237 2238 |
} if (flags & SLAB_STORE_USER) { |
ca5f9703d
|
2239 |
/* user store requires one word storage behind the end of |
87a927c71
|
2240 2241 |
* the real object. But if the second red zone needs to be * aligned to 64 bits, we must allow that much space. |
1da177e4c
|
2242 |
*/ |
87a927c71
|
2243 2244 2245 2246 |
if (flags & SLAB_RED_ZONE) size += REDZONE_ALIGN; else size += BYTES_PER_WORD; |
1da177e4c
|
2247 2248 |
} #if FORCED_DEBUG && defined(CONFIG_DEBUG_PAGEALLOC) |
b28a02de8
|
2249 |
if (size >= malloc_sizes[INDEX_L3 + 1].cs_size |
1ab335d8f
|
2250 2251 |
&& cachep->obj_size > cache_line_size() && ALIGN(size, align) < PAGE_SIZE) { cachep->obj_offset += PAGE_SIZE - ALIGN(size, align); |
1da177e4c
|
2252 2253 2254 2255 |
size = PAGE_SIZE; } #endif #endif |
e0a427267
|
2256 2257 2258 |
/* * Determine if the slab management is 'on' or 'off' slab. * (bootstrapping cannot cope with offslab caches so don't do |
e7cb55b94
|
2259 2260 |
* it too early on. Always use on-slab management when * SLAB_NOLEAKTRACE to avoid recursive calls into kmemleak) |
e0a427267
|
2261 |
*/ |
e7cb55b94
|
2262 2263 |
if ((size >= (PAGE_SIZE >> 3)) && !slab_early_init && !(flags & SLAB_NOLEAKTRACE)) |
1da177e4c
|
2264 2265 2266 2267 2268 2269 2270 |
/* * 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
|
2271 |
left_over = calculate_slab_order(cachep, size, align, flags); |
1da177e4c
|
2272 2273 |
if (!cachep->num) { |
b4169525b
|
2274 2275 2276 |
printk(KERN_ERR "kmem_cache_create: couldn't create cache %s. ", name); |
1da177e4c
|
2277 2278 |
kmem_cache_free(&cache_cache, cachep); cachep = NULL; |
4f12bb4f7
|
2279 |
goto oops; |
1da177e4c
|
2280 |
} |
b28a02de8
|
2281 2282 |
slab_size = ALIGN(cachep->num * sizeof(kmem_bufctl_t) + sizeof(struct slab), align); |
1da177e4c
|
2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 |
/* * 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
|
2295 2296 |
slab_size = cachep->num * sizeof(kmem_bufctl_t) + sizeof(struct slab); |
674613652
|
2297 2298 2299 2300 2301 2302 2303 2304 2305 |
#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
|
2306 2307 2308 2309 2310 2311 |
} cachep->colour_off = cache_line_size(); /* Offset must be a multiple of the alignment. */ if (cachep->colour_off < align) cachep->colour_off = align; |
b28a02de8
|
2312 |
cachep->colour = left_over / cachep->colour_off; |
1da177e4c
|
2313 2314 2315 |
cachep->slab_size = slab_size; cachep->flags = flags; cachep->gfpflags = 0; |
4b51d6698
|
2316 |
if (CONFIG_ZONE_DMA_FLAG && (flags & SLAB_CACHE_DMA)) |
1da177e4c
|
2317 |
cachep->gfpflags |= GFP_DMA; |
3dafccf22
|
2318 |
cachep->buffer_size = size; |
6a2d7a955
|
2319 |
cachep->reciprocal_buffer_size = reciprocal_value(size); |
1da177e4c
|
2320 |
|
e5ac9c5ae
|
2321 |
if (flags & CFLGS_OFF_SLAB) { |
b2d550736
|
2322 |
cachep->slabp_cache = kmem_find_general_cachep(slab_size, 0u); |
e5ac9c5ae
|
2323 2324 2325 2326 2327 2328 2329 |
/* * 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
|
2330 |
BUG_ON(ZERO_OR_NULL_PTR(cachep->slabp_cache)); |
e5ac9c5ae
|
2331 |
} |
1da177e4c
|
2332 |
cachep->ctor = ctor; |
1da177e4c
|
2333 |
cachep->name = name; |
83b519e8b
|
2334 |
if (setup_cpu_cache(cachep, gfp)) { |
2ed3a4ef9
|
2335 2336 2337 2338 |
__kmem_cache_destroy(cachep); cachep = NULL; goto oops; } |
1da177e4c
|
2339 |
|
1da177e4c
|
2340 2341 |
/* cache setup completed, link it into the list */ list_add(&cachep->next, &cache_chain); |
a737b3e2f
|
2342 |
oops: |
1da177e4c
|
2343 2344 2345 |
if (!cachep && (flags & SLAB_PANIC)) panic("kmem_cache_create(): failed to create slab `%s' ", |
b28a02de8
|
2346 |
name); |
83b519e8b
|
2347 2348 2349 2350 |
if (slab_is_available()) { mutex_unlock(&cache_chain_mutex); put_online_cpus(); } |
1da177e4c
|
2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 |
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
|
2365 |
static void check_spinlock_acquired(struct kmem_cache *cachep) |
1da177e4c
|
2366 2367 2368 |
{ #ifdef CONFIG_SMP check_irq_off(); |
7d6e6d09d
|
2369 |
assert_spin_locked(&cachep->nodelists[numa_mem_id()]->list_lock); |
1da177e4c
|
2370 2371 |
#endif } |
e498be7da
|
2372 |
|
343e0d7a9
|
2373 |
static void check_spinlock_acquired_node(struct kmem_cache *cachep, int node) |
e498be7da
|
2374 2375 2376 2377 2378 2379 |
{ #ifdef CONFIG_SMP check_irq_off(); assert_spin_locked(&cachep->nodelists[node]->list_lock); #endif } |
1da177e4c
|
2380 2381 2382 2383 |
#else #define check_irq_off() do { } while(0) #define check_irq_on() do { } while(0) #define check_spinlock_acquired(x) do { } while(0) |
e498be7da
|
2384 |
#define check_spinlock_acquired_node(x, y) do { } while(0) |
1da177e4c
|
2385 |
#endif |
aab2207cf
|
2386 2387 2388 |
static void drain_array(struct kmem_cache *cachep, struct kmem_list3 *l3, struct array_cache *ac, int force, int node); |
1da177e4c
|
2389 2390 |
static void do_drain(void *arg) { |
a737b3e2f
|
2391 |
struct kmem_cache *cachep = arg; |
1da177e4c
|
2392 |
struct array_cache *ac; |
7d6e6d09d
|
2393 |
int node = numa_mem_id(); |
1da177e4c
|
2394 2395 |
check_irq_off(); |
9a2dba4b4
|
2396 |
ac = cpu_cache_get(cachep); |
ff69416e6
|
2397 2398 2399 |
spin_lock(&cachep->nodelists[node]->list_lock); free_block(cachep, ac->entry, ac->avail, node); spin_unlock(&cachep->nodelists[node]->list_lock); |
1da177e4c
|
2400 2401 |
ac->avail = 0; } |
343e0d7a9
|
2402 |
static void drain_cpu_caches(struct kmem_cache *cachep) |
1da177e4c
|
2403 |
{ |
e498be7da
|
2404 2405 |
struct kmem_list3 *l3; int node; |
15c8b6c1a
|
2406 |
on_each_cpu(do_drain, cachep, 1); |
1da177e4c
|
2407 |
check_irq_on(); |
b28a02de8
|
2408 |
for_each_online_node(node) { |
e498be7da
|
2409 |
l3 = cachep->nodelists[node]; |
a4523a8b3
|
2410 2411 2412 2413 2414 2415 2416 |
if (l3 && l3->alien) drain_alien_cache(cachep, l3->alien); } for_each_online_node(node) { l3 = cachep->nodelists[node]; if (l3) |
aab2207cf
|
2417 |
drain_array(cachep, l3, l3->shared, 1, node); |
e498be7da
|
2418 |
} |
1da177e4c
|
2419 |
} |
ed11d9eb2
|
2420 2421 2422 2423 2424 2425 2426 2427 |
/* * 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
|
2428 |
{ |
ed11d9eb2
|
2429 2430 |
struct list_head *p; int nr_freed; |
1da177e4c
|
2431 |
struct slab *slabp; |
1da177e4c
|
2432 |
|
ed11d9eb2
|
2433 2434 |
nr_freed = 0; while (nr_freed < tofree && !list_empty(&l3->slabs_free)) { |
1da177e4c
|
2435 |
|
ed11d9eb2
|
2436 |
spin_lock_irq(&l3->list_lock); |
e498be7da
|
2437 |
p = l3->slabs_free.prev; |
ed11d9eb2
|
2438 2439 2440 2441 |
if (p == &l3->slabs_free) { spin_unlock_irq(&l3->list_lock); goto out; } |
1da177e4c
|
2442 |
|
ed11d9eb2
|
2443 |
slabp = list_entry(p, struct slab, list); |
1da177e4c
|
2444 |
#if DEBUG |
40094fa65
|
2445 |
BUG_ON(slabp->inuse); |
1da177e4c
|
2446 2447 |
#endif list_del(&slabp->list); |
ed11d9eb2
|
2448 2449 2450 2451 2452 |
/* * Safe to drop the lock. The slab is no longer linked * to the cache. */ l3->free_objects -= cache->num; |
e498be7da
|
2453 |
spin_unlock_irq(&l3->list_lock); |
ed11d9eb2
|
2454 2455 |
slab_destroy(cache, slabp); nr_freed++; |
1da177e4c
|
2456 |
} |
ed11d9eb2
|
2457 2458 |
out: return nr_freed; |
1da177e4c
|
2459 |
} |
8f5be20bf
|
2460 |
/* Called with cache_chain_mutex held to protect against cpu hotplug */ |
343e0d7a9
|
2461 |
static int __cache_shrink(struct kmem_cache *cachep) |
e498be7da
|
2462 2463 2464 2465 2466 2467 2468 2469 2470 |
{ 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
|
2471 2472 2473 2474 2475 2476 2477 |
if (!l3) continue; drain_freelist(cachep, l3, l3->free_objects); ret += !list_empty(&l3->slabs_full) || !list_empty(&l3->slabs_partial); |
e498be7da
|
2478 2479 2480 |
} return (ret ? 1 : 0); } |
1da177e4c
|
2481 2482 2483 2484 2485 2486 2487 |
/** * 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
|
2488 |
int kmem_cache_shrink(struct kmem_cache *cachep) |
1da177e4c
|
2489 |
{ |
8f5be20bf
|
2490 |
int ret; |
40094fa65
|
2491 |
BUG_ON(!cachep || in_interrupt()); |
1da177e4c
|
2492 |
|
95402b382
|
2493 |
get_online_cpus(); |
8f5be20bf
|
2494 2495 2496 |
mutex_lock(&cache_chain_mutex); ret = __cache_shrink(cachep); mutex_unlock(&cache_chain_mutex); |
95402b382
|
2497 |
put_online_cpus(); |
8f5be20bf
|
2498 |
return ret; |
1da177e4c
|
2499 2500 2501 2502 2503 2504 2505 |
} EXPORT_SYMBOL(kmem_cache_shrink); /** * kmem_cache_destroy - delete a cache * @cachep: the cache to destroy * |
72fd4a35a
|
2506 |
* Remove a &struct kmem_cache object from the slab cache. |
1da177e4c
|
2507 2508 2509 2510 2511 2512 2513 2514 |
* * 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
|
2515 |
* The caller must guarantee that no one will allocate memory from the cache |
1da177e4c
|
2516 2517 |
* during the kmem_cache_destroy(). */ |
133d205a1
|
2518 |
void kmem_cache_destroy(struct kmem_cache *cachep) |
1da177e4c
|
2519 |
{ |
40094fa65
|
2520 |
BUG_ON(!cachep || in_interrupt()); |
1da177e4c
|
2521 |
|
1da177e4c
|
2522 |
/* Find the cache in the chain of caches. */ |
95402b382
|
2523 |
get_online_cpus(); |
fc0abb145
|
2524 |
mutex_lock(&cache_chain_mutex); |
1da177e4c
|
2525 2526 2527 2528 |
/* * the chain is never empty, cache_cache is never destroyed */ list_del(&cachep->next); |
1da177e4c
|
2529 2530 |
if (__cache_shrink(cachep)) { slab_error(cachep, "Can't free all objects"); |
b28a02de8
|
2531 |
list_add(&cachep->next, &cache_chain); |
fc0abb145
|
2532 |
mutex_unlock(&cache_chain_mutex); |
95402b382
|
2533 |
put_online_cpus(); |
133d205a1
|
2534 |
return; |
1da177e4c
|
2535 2536 2537 |
} if (unlikely(cachep->flags & SLAB_DESTROY_BY_RCU)) |
7ed9f7e5d
|
2538 |
rcu_barrier(); |
1da177e4c
|
2539 |
|
117f6eb1d
|
2540 |
__kmem_cache_destroy(cachep); |
8f5be20bf
|
2541 |
mutex_unlock(&cache_chain_mutex); |
95402b382
|
2542 |
put_online_cpus(); |
1da177e4c
|
2543 2544 |
} EXPORT_SYMBOL(kmem_cache_destroy); |
e5ac9c5ae
|
2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 |
/* * 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
|
2556 |
static struct slab *alloc_slabmgmt(struct kmem_cache *cachep, void *objp, |
5b74ada7e
|
2557 2558 |
int colour_off, gfp_t local_flags, int nodeid) |
1da177e4c
|
2559 2560 |
{ struct slab *slabp; |
b28a02de8
|
2561 |
|
1da177e4c
|
2562 2563 |
if (OFF_SLAB(cachep)) { /* Slab management obj is off-slab. */ |
5b74ada7e
|
2564 |
slabp = kmem_cache_alloc_node(cachep->slabp_cache, |
8759ec50a
|
2565 |
local_flags, nodeid); |
d5cff6352
|
2566 2567 2568 2569 2570 2571 |
/* * 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
|
2572 2573 |
kmemleak_scan_area(&slabp->list, sizeof(struct list_head), local_flags); |
1da177e4c
|
2574 2575 2576 |
if (!slabp) return NULL; } else { |
b28a02de8
|
2577 |
slabp = objp + colour_off; |
1da177e4c
|
2578 2579 2580 2581 |
colour_off += cachep->slab_size; } slabp->inuse = 0; slabp->colouroff = colour_off; |
b28a02de8
|
2582 |
slabp->s_mem = objp + colour_off; |
5b74ada7e
|
2583 |
slabp->nodeid = nodeid; |
e51bfd0ad
|
2584 |
slabp->free = 0; |
1da177e4c
|
2585 2586 2587 2588 2589 |
return slabp; } static inline kmem_bufctl_t *slab_bufctl(struct slab *slabp) { |
b28a02de8
|
2590 |
return (kmem_bufctl_t *) (slabp + 1); |
1da177e4c
|
2591 |
} |
343e0d7a9
|
2592 |
static void cache_init_objs(struct kmem_cache *cachep, |
a35afb830
|
2593 |
struct slab *slabp) |
1da177e4c
|
2594 2595 2596 2597 |
{ int i; for (i = 0; i < cachep->num; i++) { |
8fea4e96a
|
2598 |
void *objp = index_to_obj(cachep, slabp, i); |
1da177e4c
|
2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 |
#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
|
2611 2612 2613 |
* 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
|
2614 2615 |
*/ if (cachep->ctor && !(cachep->flags & SLAB_POISON)) |
51cc50685
|
2616 |
cachep->ctor(objp + obj_offset(cachep)); |
1da177e4c
|
2617 2618 2619 2620 |
if (cachep->flags & SLAB_RED_ZONE) { if (*dbg_redzone2(cachep, objp) != RED_INACTIVE) slab_error(cachep, "constructor overwrote the" |
b28a02de8
|
2621 |
" end of an object"); |
1da177e4c
|
2622 2623 |
if (*dbg_redzone1(cachep, objp) != RED_INACTIVE) slab_error(cachep, "constructor overwrote the" |
b28a02de8
|
2624 |
" start of an object"); |
1da177e4c
|
2625 |
} |
a737b3e2f
|
2626 2627 |
if ((cachep->buffer_size % PAGE_SIZE) == 0 && OFF_SLAB(cachep) && cachep->flags & SLAB_POISON) |
b28a02de8
|
2628 |
kernel_map_pages(virt_to_page(objp), |
3dafccf22
|
2629 |
cachep->buffer_size / PAGE_SIZE, 0); |
1da177e4c
|
2630 2631 |
#else if (cachep->ctor) |
51cc50685
|
2632 |
cachep->ctor(objp); |
1da177e4c
|
2633 |
#endif |
b28a02de8
|
2634 |
slab_bufctl(slabp)[i] = i + 1; |
1da177e4c
|
2635 |
} |
b28a02de8
|
2636 |
slab_bufctl(slabp)[i - 1] = BUFCTL_END; |
1da177e4c
|
2637 |
} |
343e0d7a9
|
2638 |
static void kmem_flagcheck(struct kmem_cache *cachep, gfp_t flags) |
1da177e4c
|
2639 |
{ |
4b51d6698
|
2640 2641 2642 2643 2644 2645 |
if (CONFIG_ZONE_DMA_FLAG) { if (flags & GFP_DMA) BUG_ON(!(cachep->gfpflags & GFP_DMA)); else BUG_ON(cachep->gfpflags & GFP_DMA); } |
1da177e4c
|
2646 |
} |
a737b3e2f
|
2647 2648 |
static void *slab_get_obj(struct kmem_cache *cachep, struct slab *slabp, int nodeid) |
78d382d77
|
2649 |
{ |
8fea4e96a
|
2650 |
void *objp = index_to_obj(cachep, slabp, slabp->free); |
78d382d77
|
2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662 |
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
|
2663 2664 |
static void slab_put_obj(struct kmem_cache *cachep, struct slab *slabp, void *objp, int nodeid) |
78d382d77
|
2665 |
{ |
8fea4e96a
|
2666 |
unsigned int objnr = obj_to_index(cachep, slabp, objp); |
78d382d77
|
2667 2668 2669 2670 |
#if DEBUG /* Verify that the slab belongs to the intended node */ WARN_ON(slabp->nodeid != nodeid); |
871751e25
|
2671 |
if (slab_bufctl(slabp)[objnr] + 1 <= SLAB_LIMIT + 1) { |
78d382d77
|
2672 |
printk(KERN_ERR "slab: double free detected in cache " |
a737b3e2f
|
2673 2674 |
"'%s', objp %p ", cachep->name, objp); |
78d382d77
|
2675 2676 2677 2678 2679 2680 2681 |
BUG(); } #endif slab_bufctl(slabp)[objnr] = slabp->free; slabp->free = objnr; slabp->inuse--; } |
4776874ff
|
2682 2683 2684 |
/* * 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
|
2685 |
* virtual address for kfree, ksize, and slab debugging. |
4776874ff
|
2686 2687 2688 |
*/ static void slab_map_pages(struct kmem_cache *cache, struct slab *slab, void *addr) |
1da177e4c
|
2689 |
{ |
4776874ff
|
2690 |
int nr_pages; |
1da177e4c
|
2691 |
struct page *page; |
4776874ff
|
2692 |
page = virt_to_page(addr); |
84097518d
|
2693 |
|
4776874ff
|
2694 |
nr_pages = 1; |
84097518d
|
2695 |
if (likely(!PageCompound(page))) |
4776874ff
|
2696 |
nr_pages <<= cache->gfporder; |
1da177e4c
|
2697 |
do { |
4776874ff
|
2698 2699 |
page_set_cache(page, cache); page_set_slab(page, slab); |
1da177e4c
|
2700 |
page++; |
4776874ff
|
2701 |
} while (--nr_pages); |
1da177e4c
|
2702 2703 2704 2705 2706 2707 |
} /* * 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
|
2708 2709 |
static int cache_grow(struct kmem_cache *cachep, gfp_t flags, int nodeid, void *objp) |
1da177e4c
|
2710 |
{ |
b28a02de8
|
2711 |
struct slab *slabp; |
b28a02de8
|
2712 2713 |
size_t offset; gfp_t local_flags; |
e498be7da
|
2714 |
struct kmem_list3 *l3; |
1da177e4c
|
2715 |
|
a737b3e2f
|
2716 2717 2718 |
/* * Be lazy and only check for valid flags here, keeping it out of the * critical path in kmem_cache_alloc(). |
1da177e4c
|
2719 |
*/ |
6cb062296
|
2720 2721 |
BUG_ON(flags & GFP_SLAB_BUG_MASK); local_flags = flags & (GFP_CONSTRAINT_MASK|GFP_RECLAIM_MASK); |
1da177e4c
|
2722 |
|
2e1217cf9
|
2723 |
/* Take the l3 list lock to change the colour_next on this node */ |
1da177e4c
|
2724 |
check_irq_off(); |
2e1217cf9
|
2725 2726 |
l3 = cachep->nodelists[nodeid]; spin_lock(&l3->list_lock); |
1da177e4c
|
2727 2728 |
/* Get colour for the slab, and cal the next value. */ |
2e1217cf9
|
2729 2730 2731 2732 2733 |
offset = l3->colour_next; l3->colour_next++; if (l3->colour_next >= cachep->colour) l3->colour_next = 0; spin_unlock(&l3->list_lock); |
1da177e4c
|
2734 |
|
2e1217cf9
|
2735 |
offset *= cachep->colour_off; |
1da177e4c
|
2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 |
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
|
2747 2748 2749 |
/* * Get mem for the objs. Attempt to allocate a physical page from * 'nodeid'. |
e498be7da
|
2750 |
*/ |
3c517a613
|
2751 |
if (!objp) |
b8c1c5da1
|
2752 |
objp = kmem_getpages(cachep, local_flags, nodeid); |
a737b3e2f
|
2753 |
if (!objp) |
1da177e4c
|
2754 2755 2756 |
goto failed; /* Get slab management. */ |
3c517a613
|
2757 |
slabp = alloc_slabmgmt(cachep, objp, offset, |
6cb062296
|
2758 |
local_flags & ~GFP_CONSTRAINT_MASK, nodeid); |
a737b3e2f
|
2759 |
if (!slabp) |
1da177e4c
|
2760 |
goto opps1; |
4776874ff
|
2761 |
slab_map_pages(cachep, slabp, objp); |
1da177e4c
|
2762 |
|
a35afb830
|
2763 |
cache_init_objs(cachep, slabp); |
1da177e4c
|
2764 2765 2766 2767 |
if (local_flags & __GFP_WAIT) local_irq_disable(); check_irq_off(); |
e498be7da
|
2768 |
spin_lock(&l3->list_lock); |
1da177e4c
|
2769 2770 |
/* Make slab active. */ |
e498be7da
|
2771 |
list_add_tail(&slabp->list, &(l3->slabs_free)); |
1da177e4c
|
2772 |
STATS_INC_GROWN(cachep); |
e498be7da
|
2773 2774 |
l3->free_objects += cachep->num; spin_unlock(&l3->list_lock); |
1da177e4c
|
2775 |
return 1; |
a737b3e2f
|
2776 |
opps1: |
1da177e4c
|
2777 |
kmem_freepages(cachep, objp); |
a737b3e2f
|
2778 |
failed: |
1da177e4c
|
2779 2780 2781 2782 2783 2784 2785 2786 2787 2788 2789 |
if (local_flags & __GFP_WAIT) local_irq_disable(); return 0; } #if DEBUG /* * Perform extra freeing checks: * - detect bad pointers. * - POISON/RED_ZONE checking |
1da177e4c
|
2790 2791 2792 |
*/ static void kfree_debugcheck(const void *objp) { |
1da177e4c
|
2793 2794 2795 |
if (!virt_addr_valid(objp)) { printk(KERN_ERR "kfree_debugcheck: out of range ptr %lxh. ", |
b28a02de8
|
2796 2797 |
(unsigned long)objp); BUG(); |
1da177e4c
|
2798 |
} |
1da177e4c
|
2799 |
} |
58ce1fd58
|
2800 2801 |
static inline void verify_redzone_free(struct kmem_cache *cache, void *obj) { |
b46b8f19c
|
2802 |
unsigned long long redzone1, redzone2; |
58ce1fd58
|
2803 2804 2805 2806 2807 2808 2809 2810 2811 2812 2813 2814 2815 2816 |
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
|
2817 2818 |
printk(KERN_ERR "%p: redzone 1:0x%llx, redzone 2:0x%llx. ", |
58ce1fd58
|
2819 2820 |
obj, redzone1, redzone2); } |
343e0d7a9
|
2821 |
static void *cache_free_debugcheck(struct kmem_cache *cachep, void *objp, |
b28a02de8
|
2822 |
void *caller) |
1da177e4c
|
2823 2824 2825 2826 |
{ struct page *page; unsigned int objnr; struct slab *slabp; |
80cbd911c
|
2827 |
BUG_ON(virt_to_cache(objp) != cachep); |
3dafccf22
|
2828 |
objp -= obj_offset(cachep); |
1da177e4c
|
2829 |
kfree_debugcheck(objp); |
b49af68ff
|
2830 |
page = virt_to_head_page(objp); |
1da177e4c
|
2831 |
|
065d41cb2
|
2832 |
slabp = page_get_slab(page); |
1da177e4c
|
2833 2834 |
if (cachep->flags & SLAB_RED_ZONE) { |
58ce1fd58
|
2835 |
verify_redzone_free(cachep, objp); |
1da177e4c
|
2836 2837 2838 2839 2840 |
*dbg_redzone1(cachep, objp) = RED_INACTIVE; *dbg_redzone2(cachep, objp) = RED_INACTIVE; } if (cachep->flags & SLAB_STORE_USER) *dbg_userword(cachep, objp) = caller; |
8fea4e96a
|
2841 |
objnr = obj_to_index(cachep, slabp, objp); |
1da177e4c
|
2842 2843 |
BUG_ON(objnr >= cachep->num); |
8fea4e96a
|
2844 |
BUG_ON(objp != index_to_obj(cachep, slabp, objnr)); |
1da177e4c
|
2845 |
|
871751e25
|
2846 2847 2848 |
#ifdef CONFIG_DEBUG_SLAB_LEAK slab_bufctl(slabp)[objnr] = BUFCTL_FREE; #endif |
1da177e4c
|
2849 2850 |
if (cachep->flags & SLAB_POISON) { #ifdef CONFIG_DEBUG_PAGEALLOC |
a737b3e2f
|
2851 |
if ((cachep->buffer_size % PAGE_SIZE)==0 && OFF_SLAB(cachep)) { |
1da177e4c
|
2852 |
store_stackinfo(cachep, objp, (unsigned long)caller); |
b28a02de8
|
2853 |
kernel_map_pages(virt_to_page(objp), |
3dafccf22
|
2854 |
cachep->buffer_size / PAGE_SIZE, 0); |
1da177e4c
|
2855 2856 2857 2858 2859 2860 2861 2862 2863 |
} else { poison_obj(cachep, objp, POISON_FREE); } #else poison_obj(cachep, objp, POISON_FREE); #endif } return objp; } |
343e0d7a9
|
2864 |
static void check_slabp(struct kmem_cache *cachep, struct slab *slabp) |
1da177e4c
|
2865 2866 2867 |
{ kmem_bufctl_t i; int entries = 0; |
b28a02de8
|
2868 |
|
1da177e4c
|
2869 2870 2871 2872 2873 2874 2875 |
/* 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
|
2876 2877 2878 2879 2880 |
bad: printk(KERN_ERR "slab: Internal list corruption detected in " "cache '%s'(%d), slabp %p(%d). Hexdump: ", cachep->name, cachep->num, slabp, slabp->inuse); |
b28a02de8
|
2881 |
for (i = 0; |
264132bc6
|
2882 |
i < sizeof(*slabp) + cachep->num * sizeof(kmem_bufctl_t); |
b28a02de8
|
2883 |
i++) { |
a737b3e2f
|
2884 |
if (i % 16 == 0) |
1da177e4c
|
2885 2886 |
printk(" %03x:", i); |
b28a02de8
|
2887 |
printk(" %02x", ((unsigned char *)slabp)[i]); |
1da177e4c
|
2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 |
} printk(" "); BUG(); } } #else #define kfree_debugcheck(x) do { } while(0) #define cache_free_debugcheck(x,objp,z) (objp) #define check_slabp(x,y) do { } while(0) #endif |
343e0d7a9
|
2899 |
static void *cache_alloc_refill(struct kmem_cache *cachep, gfp_t flags) |
1da177e4c
|
2900 2901 2902 2903 |
{ int batchcount; struct kmem_list3 *l3; struct array_cache *ac; |
1ca4cb241
|
2904 |
int node; |
6d2144d35
|
2905 |
retry: |
1da177e4c
|
2906 |
check_irq_off(); |
7d6e6d09d
|
2907 |
node = numa_mem_id(); |
9a2dba4b4
|
2908 |
ac = cpu_cache_get(cachep); |
1da177e4c
|
2909 2910 |
batchcount = ac->batchcount; if (!ac->touched && batchcount > BATCHREFILL_LIMIT) { |
a737b3e2f
|
2911 2912 2913 2914 |
/* * If there was little recent activity on this cache, then * perform only a partial refill. Otherwise we could generate * refill bouncing. |
1da177e4c
|
2915 2916 2917 |
*/ batchcount = BATCHREFILL_LIMIT; } |
1ca4cb241
|
2918 |
l3 = cachep->nodelists[node]; |
e498be7da
|
2919 2920 2921 |
BUG_ON(ac->avail > 0 || !l3); spin_lock(&l3->list_lock); |
1da177e4c
|
2922 |
|
3ded175a4
|
2923 |
/* See if we can refill from the shared array */ |
44b57f1cc
|
2924 2925 |
if (l3->shared && transfer_objects(ac, l3->shared, batchcount)) { l3->shared->touched = 1; |
3ded175a4
|
2926 |
goto alloc_done; |
44b57f1cc
|
2927 |
} |
3ded175a4
|
2928 |
|
1da177e4c
|
2929 2930 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 |
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
|
2944 2945 2946 2947 2948 2949 |
/* * The slab was either on partial or free list so * there must be at least one object available for * allocation. */ |
249b9f331
|
2950 |
BUG_ON(slabp->inuse >= cachep->num); |
714b8171a
|
2951 |
|
1da177e4c
|
2952 |
while (slabp->inuse < cachep->num && batchcount--) { |
1da177e4c
|
2953 2954 2955 |
STATS_INC_ALLOCED(cachep); STATS_INC_ACTIVE(cachep); STATS_SET_HIGH(cachep); |
78d382d77
|
2956 |
ac->entry[ac->avail++] = slab_get_obj(cachep, slabp, |
1ca4cb241
|
2957 |
node); |
1da177e4c
|
2958 2959 2960 2961 2962 2963 2964 2965 2966 2967 |
} 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
|
2968 |
must_grow: |
1da177e4c
|
2969 |
l3->free_objects -= ac->avail; |
a737b3e2f
|
2970 |
alloc_done: |
e498be7da
|
2971 |
spin_unlock(&l3->list_lock); |
1da177e4c
|
2972 2973 2974 |
if (unlikely(!ac->avail)) { int x; |
3c517a613
|
2975 |
x = cache_grow(cachep, flags | GFP_THISNODE, node, NULL); |
e498be7da
|
2976 |
|
a737b3e2f
|
2977 |
/* cache_grow can reenable interrupts, then ac could change. */ |
9a2dba4b4
|
2978 |
ac = cpu_cache_get(cachep); |
a737b3e2f
|
2979 |
if (!x && ac->avail == 0) /* no objects in sight? abort */ |
1da177e4c
|
2980 |
return NULL; |
a737b3e2f
|
2981 |
if (!ac->avail) /* objects refilled by interrupt? */ |
1da177e4c
|
2982 2983 2984 |
goto retry; } ac->touched = 1; |
e498be7da
|
2985 |
return ac->entry[--ac->avail]; |
1da177e4c
|
2986 |
} |
a737b3e2f
|
2987 2988 |
static inline void cache_alloc_debugcheck_before(struct kmem_cache *cachep, gfp_t flags) |
1da177e4c
|
2989 2990 2991 2992 2993 2994 2995 2996 |
{ might_sleep_if(flags & __GFP_WAIT); #if DEBUG kmem_flagcheck(cachep, flags); #endif } #if DEBUG |
a737b3e2f
|
2997 2998 |
static void *cache_alloc_debugcheck_after(struct kmem_cache *cachep, gfp_t flags, void *objp, void *caller) |
1da177e4c
|
2999 |
{ |
b28a02de8
|
3000 |
if (!objp) |
1da177e4c
|
3001 |
return objp; |
b28a02de8
|
3002 |
if (cachep->flags & SLAB_POISON) { |
1da177e4c
|
3003 |
#ifdef CONFIG_DEBUG_PAGEALLOC |
3dafccf22
|
3004 |
if ((cachep->buffer_size % PAGE_SIZE) == 0 && OFF_SLAB(cachep)) |
b28a02de8
|
3005 |
kernel_map_pages(virt_to_page(objp), |
3dafccf22
|
3006 |
cachep->buffer_size / PAGE_SIZE, 1); |
1da177e4c
|
3007 3008 3009 3010 3011 3012 3013 3014 3015 3016 3017 |
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
|
3018 3019 3020 3021 |
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
|
3022 |
printk(KERN_ERR |
b46b8f19c
|
3023 3024 |
"%p: redzone 1:0x%llx, redzone 2:0x%llx ", |
a737b3e2f
|
3025 3026 |
objp, *dbg_redzone1(cachep, objp), *dbg_redzone2(cachep, objp)); |
1da177e4c
|
3027 3028 3029 3030 |
} *dbg_redzone1(cachep, objp) = RED_ACTIVE; *dbg_redzone2(cachep, objp) = RED_ACTIVE; } |
871751e25
|
3031 3032 3033 3034 |
#ifdef CONFIG_DEBUG_SLAB_LEAK { struct slab *slabp; unsigned objnr; |
b49af68ff
|
3035 |
slabp = page_get_slab(virt_to_head_page(objp)); |
871751e25
|
3036 3037 3038 3039 |
objnr = (unsigned)(objp - slabp->s_mem) / cachep->buffer_size; slab_bufctl(slabp)[objnr] = BUFCTL_ACTIVE; } #endif |
3dafccf22
|
3040 |
objp += obj_offset(cachep); |
4f1049345
|
3041 |
if (cachep->ctor && cachep->flags & SLAB_POISON) |
51cc50685
|
3042 |
cachep->ctor(objp); |
a44b56d35
|
3043 3044 3045 3046 3047 3048 3049 |
#if ARCH_SLAB_MINALIGN if ((u32)objp & (ARCH_SLAB_MINALIGN-1)) { printk(KERN_ERR "0x%p: not aligned to ARCH_SLAB_MINALIGN=%d ", objp, ARCH_SLAB_MINALIGN); } #endif |
1da177e4c
|
3050 3051 3052 3053 3054 |
return objp; } #else #define cache_alloc_debugcheck_after(a,b,objp,d) (objp) #endif |
773ff60e8
|
3055 |
static bool slab_should_failslab(struct kmem_cache *cachep, gfp_t flags) |
8a8b6502f
|
3056 3057 |
{ if (cachep == &cache_cache) |
773ff60e8
|
3058 |
return false; |
8a8b6502f
|
3059 |
|
4c13dd3b4
|
3060 |
return should_failslab(obj_size(cachep), flags, cachep->flags); |
8a8b6502f
|
3061 |
} |
343e0d7a9
|
3062 |
static inline void *____cache_alloc(struct kmem_cache *cachep, gfp_t flags) |
1da177e4c
|
3063 |
{ |
b28a02de8
|
3064 |
void *objp; |
1da177e4c
|
3065 |
struct array_cache *ac; |
5c3823008
|
3066 |
check_irq_off(); |
8a8b6502f
|
3067 |
|
9a2dba4b4
|
3068 |
ac = cpu_cache_get(cachep); |
1da177e4c
|
3069 3070 3071 |
if (likely(ac->avail)) { STATS_INC_ALLOCHIT(cachep); ac->touched = 1; |
e498be7da
|
3072 |
objp = ac->entry[--ac->avail]; |
1da177e4c
|
3073 3074 3075 |
} else { STATS_INC_ALLOCMISS(cachep); objp = cache_alloc_refill(cachep, flags); |
ddbf2e836
|
3076 3077 3078 3079 3080 |
/* * the 'ac' may be updated by cache_alloc_refill(), * and kmemleak_erase() requires its correct value. */ ac = cpu_cache_get(cachep); |
1da177e4c
|
3081 |
} |
d5cff6352
|
3082 3083 3084 3085 3086 |
/* * 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
|
3087 3088 |
if (objp) kmemleak_erase(&ac->entry[ac->avail]); |
5c3823008
|
3089 3090 |
return objp; } |
e498be7da
|
3091 3092 |
#ifdef CONFIG_NUMA /* |
b2455396b
|
3093 |
* Try allocating on another node if PF_SPREAD_SLAB|PF_MEMPOLICY. |
c61afb181
|
3094 3095 3096 3097 3098 3099 3100 |
* * 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
|
3101 |
if (in_interrupt() || (flags & __GFP_THISNODE)) |
c61afb181
|
3102 |
return NULL; |
7d6e6d09d
|
3103 |
nid_alloc = nid_here = numa_mem_id(); |
c0ff7453b
|
3104 |
get_mems_allowed(); |
c61afb181
|
3105 |
if (cpuset_do_slab_mem_spread() && (cachep->flags & SLAB_MEM_SPREAD)) |
6adef3ebe
|
3106 |
nid_alloc = cpuset_slab_spread_node(); |
c61afb181
|
3107 3108 |
else if (current->mempolicy) nid_alloc = slab_node(current->mempolicy); |
c0ff7453b
|
3109 |
put_mems_allowed(); |
c61afb181
|
3110 |
if (nid_alloc != nid_here) |
8b98c1699
|
3111 |
return ____cache_alloc_node(cachep, flags, nid_alloc); |
c61afb181
|
3112 3113 3114 3115 |
return NULL; } /* |
765c4507a
|
3116 |
* Fallback function if there was no memory available and no objects on a |
3c517a613
|
3117 3118 3119 3120 3121 |
* 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
|
3122 |
*/ |
8c8cc2c10
|
3123 |
static void *fallback_alloc(struct kmem_cache *cache, gfp_t flags) |
765c4507a
|
3124 |
{ |
8c8cc2c10
|
3125 3126 |
struct zonelist *zonelist; gfp_t local_flags; |
dd1a239f6
|
3127 |
struct zoneref *z; |
54a6eb5c4
|
3128 3129 |
struct zone *zone; enum zone_type high_zoneidx = gfp_zone(flags); |
765c4507a
|
3130 |
void *obj = NULL; |
3c517a613
|
3131 |
int nid; |
8c8cc2c10
|
3132 3133 3134 |
if (flags & __GFP_THISNODE) return NULL; |
c0ff7453b
|
3135 |
get_mems_allowed(); |
0e88460da
|
3136 |
zonelist = node_zonelist(slab_node(current->mempolicy), flags); |
6cb062296
|
3137 |
local_flags = flags & (GFP_CONSTRAINT_MASK|GFP_RECLAIM_MASK); |
765c4507a
|
3138 |
|
3c517a613
|
3139 3140 3141 3142 3143 |
retry: /* * Look through allowed nodes for objects available * from existing per node queues. */ |
54a6eb5c4
|
3144 3145 |
for_each_zone_zonelist(zone, z, zonelist, high_zoneidx) { nid = zone_to_nid(zone); |
aedb0eb10
|
3146 |
|
54a6eb5c4
|
3147 |
if (cpuset_zone_allowed_hardwall(zone, flags) && |
3c517a613
|
3148 |
cache->nodelists[nid] && |
481c5346d
|
3149 |
cache->nodelists[nid]->free_objects) { |
3c517a613
|
3150 3151 |
obj = ____cache_alloc_node(cache, flags | GFP_THISNODE, nid); |
481c5346d
|
3152 3153 3154 |
if (obj) break; } |
3c517a613
|
3155 |
} |
cfce66047
|
3156 |
if (!obj) { |
3c517a613
|
3157 3158 3159 3160 3161 3162 |
/* * 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
|
3163 3164 3165 |
if (local_flags & __GFP_WAIT) local_irq_enable(); kmem_flagcheck(cache, flags); |
7d6e6d09d
|
3166 |
obj = kmem_getpages(cache, local_flags, numa_mem_id()); |
dd47ea755
|
3167 3168 |
if (local_flags & __GFP_WAIT) local_irq_disable(); |
3c517a613
|
3169 3170 3171 3172 3173 3174 3175 3176 3177 3178 3179 3180 3181 3182 3183 3184 |
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
|
3185 |
/* cache_grow already freed obj */ |
3c517a613
|
3186 3187 3188 |
obj = NULL; } } |
aedb0eb10
|
3189 |
} |
c0ff7453b
|
3190 |
put_mems_allowed(); |
765c4507a
|
3191 3192 3193 3194 |
return obj; } /* |
e498be7da
|
3195 |
* A interface to enable slab creation on nodeid |
1da177e4c
|
3196 |
*/ |
8b98c1699
|
3197 |
static void *____cache_alloc_node(struct kmem_cache *cachep, gfp_t flags, |
a737b3e2f
|
3198 |
int nodeid) |
e498be7da
|
3199 3200 |
{ struct list_head *entry; |
b28a02de8
|
3201 3202 3203 |
struct slab *slabp; struct kmem_list3 *l3; void *obj; |
b28a02de8
|
3204 3205 3206 3207 |
int x; l3 = cachep->nodelists[nodeid]; BUG_ON(!l3); |
a737b3e2f
|
3208 |
retry: |
ca3b9b917
|
3209 |
check_irq_off(); |
b28a02de8
|
3210 3211 3212 3213 3214 3215 3216 3217 3218 3219 3220 3221 3222 3223 3224 3225 3226 3227 |
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
|
3228 |
obj = slab_get_obj(cachep, slabp, nodeid); |
b28a02de8
|
3229 3230 3231 3232 |
check_slabp(cachep, slabp); l3->free_objects--; /* move slabp to correct slabp list: */ list_del(&slabp->list); |
a737b3e2f
|
3233 |
if (slabp->free == BUFCTL_END) |
b28a02de8
|
3234 |
list_add(&slabp->list, &l3->slabs_full); |
a737b3e2f
|
3235 |
else |
b28a02de8
|
3236 |
list_add(&slabp->list, &l3->slabs_partial); |
e498be7da
|
3237 |
|
b28a02de8
|
3238 3239 |
spin_unlock(&l3->list_lock); goto done; |
e498be7da
|
3240 |
|
a737b3e2f
|
3241 |
must_grow: |
b28a02de8
|
3242 |
spin_unlock(&l3->list_lock); |
3c517a613
|
3243 |
x = cache_grow(cachep, flags | GFP_THISNODE, nodeid, NULL); |
765c4507a
|
3244 3245 |
if (x) goto retry; |
1da177e4c
|
3246 |
|
8c8cc2c10
|
3247 |
return fallback_alloc(cachep, flags); |
e498be7da
|
3248 |
|
a737b3e2f
|
3249 |
done: |
b28a02de8
|
3250 |
return obj; |
e498be7da
|
3251 |
} |
8c8cc2c10
|
3252 3253 3254 3255 3256 3257 3258 3259 3260 3261 3262 3263 3264 3265 3266 3267 3268 3269 3270 |
/** * 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
|
3271 |
int slab_node = numa_mem_id(); |
8c8cc2c10
|
3272 |
|
dcce284a2
|
3273 |
flags &= gfp_allowed_mask; |
7e85ee0c1
|
3274 |
|
cf40bd16f
|
3275 |
lockdep_trace_alloc(flags); |
773ff60e8
|
3276 |
if (slab_should_failslab(cachep, flags)) |
824ebef12
|
3277 |
return NULL; |
8c8cc2c10
|
3278 3279 |
cache_alloc_debugcheck_before(cachep, flags); local_irq_save(save_flags); |
8e15b79cf
|
3280 |
if (nodeid == -1) |
7d6e6d09d
|
3281 |
nodeid = slab_node; |
8c8cc2c10
|
3282 3283 3284 3285 3286 3287 |
if (unlikely(!cachep->nodelists[nodeid])) { /* Node not bootstrapped yet */ ptr = fallback_alloc(cachep, flags); goto out; } |
7d6e6d09d
|
3288 |
if (nodeid == slab_node) { |
8c8cc2c10
|
3289 3290 3291 3292 3293 3294 3295 3296 3297 3298 3299 3300 3301 3302 3303 |
/* * 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
|
3304 3305 |
kmemleak_alloc_recursive(ptr, obj_size(cachep), 1, cachep->flags, flags); |
8c8cc2c10
|
3306 |
|
c175eea46
|
3307 3308 |
if (likely(ptr)) kmemcheck_slab_alloc(cachep, flags, ptr, obj_size(cachep)); |
d07dbea46
|
3309 3310 |
if (unlikely((flags & __GFP_ZERO) && ptr)) memset(ptr, 0, obj_size(cachep)); |
8c8cc2c10
|
3311 3312 3313 3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325 3326 3327 3328 3329 |
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
|
3330 3331 |
if (!objp) objp = ____cache_alloc_node(cache, flags, numa_mem_id()); |
8c8cc2c10
|
3332 3333 3334 3335 3336 3337 3338 3339 3340 3341 3342 3343 3344 3345 3346 3347 3348 3349 3350 |
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
|
3351 |
flags &= gfp_allowed_mask; |
7e85ee0c1
|
3352 |
|
cf40bd16f
|
3353 |
lockdep_trace_alloc(flags); |
773ff60e8
|
3354 |
if (slab_should_failslab(cachep, flags)) |
824ebef12
|
3355 |
return NULL; |
8c8cc2c10
|
3356 3357 3358 3359 3360 |
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
|
3361 3362 |
kmemleak_alloc_recursive(objp, obj_size(cachep), 1, cachep->flags, flags); |
8c8cc2c10
|
3363 |
prefetchw(objp); |
c175eea46
|
3364 3365 |
if (likely(objp)) kmemcheck_slab_alloc(cachep, flags, objp, obj_size(cachep)); |
d07dbea46
|
3366 3367 |
if (unlikely((flags & __GFP_ZERO) && objp)) memset(objp, 0, obj_size(cachep)); |
8c8cc2c10
|
3368 3369 |
return objp; } |
e498be7da
|
3370 3371 3372 3373 |
/* * Caller needs to acquire correct kmem_list's list_lock */ |
343e0d7a9
|
3374 |
static void free_block(struct kmem_cache *cachep, void **objpp, int nr_objects, |
b28a02de8
|
3375 |
int node) |
1da177e4c
|
3376 3377 |
{ int i; |
e498be7da
|
3378 |
struct kmem_list3 *l3; |
1da177e4c
|
3379 3380 3381 3382 |
for (i = 0; i < nr_objects; i++) { void *objp = objpp[i]; struct slab *slabp; |
1da177e4c
|
3383 |
|
6ed5eb221
|
3384 |
slabp = virt_to_slab(objp); |
ff69416e6
|
3385 |
l3 = cachep->nodelists[node]; |
1da177e4c
|
3386 |
list_del(&slabp->list); |
ff69416e6
|
3387 |
check_spinlock_acquired_node(cachep, node); |
1da177e4c
|
3388 |
check_slabp(cachep, slabp); |
78d382d77
|
3389 |
slab_put_obj(cachep, slabp, objp, node); |
1da177e4c
|
3390 |
STATS_DEC_ACTIVE(cachep); |
e498be7da
|
3391 |
l3->free_objects++; |
1da177e4c
|
3392 3393 3394 3395 |
check_slabp(cachep, slabp); /* fixup slab chains */ if (slabp->inuse == 0) { |
e498be7da
|
3396 3397 |
if (l3->free_objects > l3->free_limit) { l3->free_objects -= cachep->num; |
e5ac9c5ae
|
3398 3399 3400 3401 3402 3403 |
/* 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
|
3404 3405 |
slab_destroy(cachep, slabp); } else { |
e498be7da
|
3406 |
list_add(&slabp->list, &l3->slabs_free); |
1da177e4c
|
3407 3408 3409 3410 3411 3412 |
} } 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
|
3413 |
list_add_tail(&slabp->list, &l3->slabs_partial); |
1da177e4c
|
3414 3415 3416 |
} } } |
343e0d7a9
|
3417 |
static void cache_flusharray(struct kmem_cache *cachep, struct array_cache *ac) |
1da177e4c
|
3418 3419 |
{ int batchcount; |
e498be7da
|
3420 |
struct kmem_list3 *l3; |
7d6e6d09d
|
3421 |
int node = numa_mem_id(); |
1da177e4c
|
3422 3423 3424 3425 3426 3427 |
batchcount = ac->batchcount; #if DEBUG BUG_ON(!batchcount || batchcount > ac->avail); #endif check_irq_off(); |
ff69416e6
|
3428 |
l3 = cachep->nodelists[node]; |
873623dfa
|
3429 |
spin_lock(&l3->list_lock); |
e498be7da
|
3430 3431 |
if (l3->shared) { struct array_cache *shared_array = l3->shared; |
b28a02de8
|
3432 |
int max = shared_array->limit - shared_array->avail; |
1da177e4c
|
3433 3434 3435 |
if (max) { if (batchcount > max) batchcount = max; |
e498be7da
|
3436 |
memcpy(&(shared_array->entry[shared_array->avail]), |
b28a02de8
|
3437 |
ac->entry, sizeof(void *) * batchcount); |
1da177e4c
|
3438 3439 3440 3441 |
shared_array->avail += batchcount; goto free_done; } } |
ff69416e6
|
3442 |
free_block(cachep, ac->entry, batchcount, node); |
a737b3e2f
|
3443 |
free_done: |
1da177e4c
|
3444 3445 3446 3447 |
#if STATS { int i = 0; struct list_head *p; |
e498be7da
|
3448 3449 |
p = l3->slabs_free.next; while (p != &(l3->slabs_free)) { |
1da177e4c
|
3450 3451 3452 3453 3454 3455 3456 3457 3458 3459 3460 |
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
|
3461 |
spin_unlock(&l3->list_lock); |
1da177e4c
|
3462 |
ac->avail -= batchcount; |
a737b3e2f
|
3463 |
memmove(ac->entry, &(ac->entry[batchcount]), sizeof(void *)*ac->avail); |
1da177e4c
|
3464 3465 3466 |
} /* |
a737b3e2f
|
3467 3468 |
* 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
|
3469 |
*/ |
a947eb95e
|
3470 3471 |
static inline void __cache_free(struct kmem_cache *cachep, void *objp, void *caller) |
1da177e4c
|
3472 |
{ |
9a2dba4b4
|
3473 |
struct array_cache *ac = cpu_cache_get(cachep); |
1da177e4c
|
3474 3475 |
check_irq_off(); |
d5cff6352
|
3476 |
kmemleak_free_recursive(objp, cachep->flags); |
a947eb95e
|
3477 |
objp = cache_free_debugcheck(cachep, objp, caller); |
1da177e4c
|
3478 |
|
c175eea46
|
3479 |
kmemcheck_slab_free(cachep, objp, obj_size(cachep)); |
1807a1aaf
|
3480 3481 3482 3483 3484 3485 3486 |
/* * 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
|
3487 |
if (nr_online_nodes > 1 && cache_free_alien(cachep, objp)) |
729bd0b74
|
3488 |
return; |
1da177e4c
|
3489 3490 |
if (likely(ac->avail < ac->limit)) { STATS_INC_FREEHIT(cachep); |
e498be7da
|
3491 |
ac->entry[ac->avail++] = objp; |
1da177e4c
|
3492 3493 3494 3495 |
return; } else { STATS_INC_FREEMISS(cachep); cache_flusharray(cachep, ac); |
e498be7da
|
3496 |
ac->entry[ac->avail++] = objp; |
1da177e4c
|
3497 3498 3499 3500 3501 3502 3503 3504 3505 3506 3507 |
} } /** * 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
|
3508 |
void *kmem_cache_alloc(struct kmem_cache *cachep, gfp_t flags) |
1da177e4c
|
3509 |
{ |
36555751c
|
3510 |
void *ret = __cache_alloc(cachep, flags, __builtin_return_address(0)); |
ca2b84cb3
|
3511 3512 |
trace_kmem_cache_alloc(_RET_IP_, ret, obj_size(cachep), cachep->buffer_size, flags); |
36555751c
|
3513 3514 |
return ret; |
1da177e4c
|
3515 3516 |
} EXPORT_SYMBOL(kmem_cache_alloc); |
0f24f1287
|
3517 |
#ifdef CONFIG_TRACING |
85beb5869
|
3518 3519 |
void * kmem_cache_alloc_trace(size_t size, struct kmem_cache *cachep, gfp_t flags) |
36555751c
|
3520 |
{ |
85beb5869
|
3521 3522 3523 3524 3525 3526 3527 |
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
|
3528 |
} |
85beb5869
|
3529 |
EXPORT_SYMBOL(kmem_cache_alloc_trace); |
36555751c
|
3530 |
#endif |
1da177e4c
|
3531 |
#ifdef CONFIG_NUMA |
8b98c1699
|
3532 3533 |
void *kmem_cache_alloc_node(struct kmem_cache *cachep, gfp_t flags, int nodeid) { |
36555751c
|
3534 3535 |
void *ret = __cache_alloc_node(cachep, flags, nodeid, __builtin_return_address(0)); |
ca2b84cb3
|
3536 3537 3538 |
trace_kmem_cache_alloc_node(_RET_IP_, ret, obj_size(cachep), cachep->buffer_size, flags, nodeid); |
36555751c
|
3539 3540 |
return ret; |
8b98c1699
|
3541 |
} |
1da177e4c
|
3542 |
EXPORT_SYMBOL(kmem_cache_alloc_node); |
0f24f1287
|
3543 |
#ifdef CONFIG_TRACING |
85beb5869
|
3544 3545 3546 3547 |
void *kmem_cache_alloc_node_trace(size_t size, struct kmem_cache *cachep, gfp_t flags, int nodeid) |
36555751c
|
3548 |
{ |
85beb5869
|
3549 3550 3551 |
void *ret; ret = __cache_alloc_node(cachep, flags, nodeid, |
36555751c
|
3552 |
__builtin_return_address(0)); |
85beb5869
|
3553 3554 3555 3556 |
trace_kmalloc_node(_RET_IP_, ret, size, slab_buffer_size(cachep), flags, nodeid); return ret; |
36555751c
|
3557 |
} |
85beb5869
|
3558 |
EXPORT_SYMBOL(kmem_cache_alloc_node_trace); |
36555751c
|
3559 |
#endif |
8b98c1699
|
3560 3561 |
static __always_inline void * __do_kmalloc_node(size_t size, gfp_t flags, int node, void *caller) |
97e2bde47
|
3562 |
{ |
343e0d7a9
|
3563 |
struct kmem_cache *cachep; |
97e2bde47
|
3564 3565 |
cachep = kmem_find_general_cachep(size, flags); |
6cb8f9132
|
3566 3567 |
if (unlikely(ZERO_OR_NULL_PTR(cachep))) return cachep; |
85beb5869
|
3568 |
return kmem_cache_alloc_node_trace(size, cachep, flags, node); |
97e2bde47
|
3569 |
} |
8b98c1699
|
3570 |
|
0bb38a5cd
|
3571 |
#if defined(CONFIG_DEBUG_SLAB) || defined(CONFIG_TRACING) |
8b98c1699
|
3572 3573 3574 3575 3576 |
void *__kmalloc_node(size_t size, gfp_t flags, int node) { return __do_kmalloc_node(size, flags, node, __builtin_return_address(0)); } |
dbe5e69d2
|
3577 |
EXPORT_SYMBOL(__kmalloc_node); |
8b98c1699
|
3578 3579 |
void *__kmalloc_node_track_caller(size_t size, gfp_t flags, |
ce71e27c6
|
3580 |
int node, unsigned long caller) |
8b98c1699
|
3581 |
{ |
ce71e27c6
|
3582 |
return __do_kmalloc_node(size, flags, node, (void *)caller); |
8b98c1699
|
3583 3584 3585 3586 3587 3588 3589 3590 |
} 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
|
3591 |
#endif /* CONFIG_DEBUG_SLAB || CONFIG_TRACING */ |
8b98c1699
|
3592 |
#endif /* CONFIG_NUMA */ |
1da177e4c
|
3593 3594 |
/** |
800590f52
|
3595 |
* __do_kmalloc - allocate memory |
1da177e4c
|
3596 |
* @size: how many bytes of memory are required. |
800590f52
|
3597 |
* @flags: the type of memory to allocate (see kmalloc). |
911851e6e
|
3598 |
* @caller: function caller for debug tracking of the caller |
1da177e4c
|
3599 |
*/ |
7fd6b1413
|
3600 3601 |
static __always_inline void *__do_kmalloc(size_t size, gfp_t flags, void *caller) |
1da177e4c
|
3602 |
{ |
343e0d7a9
|
3603 |
struct kmem_cache *cachep; |
36555751c
|
3604 |
void *ret; |
1da177e4c
|
3605 |
|
97e2bde47
|
3606 3607 3608 3609 3610 3611 |
/* 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
|
3612 3613 |
if (unlikely(ZERO_OR_NULL_PTR(cachep))) return cachep; |
36555751c
|
3614 |
ret = __cache_alloc(cachep, flags, caller); |
ca2b84cb3
|
3615 3616 |
trace_kmalloc((unsigned long) caller, ret, size, cachep->buffer_size, flags); |
36555751c
|
3617 3618 |
return ret; |
7fd6b1413
|
3619 |
} |
7fd6b1413
|
3620 |
|
0bb38a5cd
|
3621 |
#if defined(CONFIG_DEBUG_SLAB) || defined(CONFIG_TRACING) |
7fd6b1413
|
3622 3623 |
void *__kmalloc(size_t size, gfp_t flags) { |
871751e25
|
3624 |
return __do_kmalloc(size, flags, __builtin_return_address(0)); |
1da177e4c
|
3625 3626 |
} EXPORT_SYMBOL(__kmalloc); |
ce71e27c6
|
3627 |
void *__kmalloc_track_caller(size_t size, gfp_t flags, unsigned long caller) |
7fd6b1413
|
3628 |
{ |
ce71e27c6
|
3629 |
return __do_kmalloc(size, flags, (void *)caller); |
7fd6b1413
|
3630 3631 |
} EXPORT_SYMBOL(__kmalloc_track_caller); |
1d2c8eea6
|
3632 3633 3634 3635 3636 3637 3638 |
#else void *__kmalloc(size_t size, gfp_t flags) { return __do_kmalloc(size, flags, NULL); } EXPORT_SYMBOL(__kmalloc); |
7fd6b1413
|
3639 |
#endif |
1da177e4c
|
3640 3641 3642 3643 3644 3645 3646 3647 |
/** * 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
|
3648 |
void kmem_cache_free(struct kmem_cache *cachep, void *objp) |
1da177e4c
|
3649 3650 3651 3652 |
{ unsigned long flags; local_irq_save(flags); |
898552c9d
|
3653 |
debug_check_no_locks_freed(objp, obj_size(cachep)); |
3ac7fe5a4
|
3654 3655 |
if (!(cachep->flags & SLAB_DEBUG_OBJECTS)) debug_check_no_obj_freed(objp, obj_size(cachep)); |
a947eb95e
|
3656 |
__cache_free(cachep, objp, __builtin_return_address(0)); |
1da177e4c
|
3657 |
local_irq_restore(flags); |
36555751c
|
3658 |
|
ca2b84cb3
|
3659 |
trace_kmem_cache_free(_RET_IP_, objp); |
1da177e4c
|
3660 3661 3662 3663 |
} EXPORT_SYMBOL(kmem_cache_free); /** |
1da177e4c
|
3664 3665 3666 |
* kfree - free previously allocated memory * @objp: pointer returned by kmalloc. * |
80e93effc
|
3667 3668 |
* If @objp is NULL, no operation is performed. * |
1da177e4c
|
3669 3670 3671 3672 3673 |
* Don't free memory not originally allocated by kmalloc() * or you will run into trouble. */ void kfree(const void *objp) { |
343e0d7a9
|
3674 |
struct kmem_cache *c; |
1da177e4c
|
3675 |
unsigned long flags; |
2121db74b
|
3676 |
trace_kfree(_RET_IP_, objp); |
6cb8f9132
|
3677 |
if (unlikely(ZERO_OR_NULL_PTR(objp))) |
1da177e4c
|
3678 3679 3680 |
return; local_irq_save(flags); kfree_debugcheck(objp); |
6ed5eb221
|
3681 |
c = virt_to_cache(objp); |
f9b8404cf
|
3682 |
debug_check_no_locks_freed(objp, obj_size(c)); |
3ac7fe5a4
|
3683 |
debug_check_no_obj_freed(objp, obj_size(c)); |
a947eb95e
|
3684 |
__cache_free(c, (void *)objp, __builtin_return_address(0)); |
1da177e4c
|
3685 3686 3687 |
local_irq_restore(flags); } EXPORT_SYMBOL(kfree); |
343e0d7a9
|
3688 |
unsigned int kmem_cache_size(struct kmem_cache *cachep) |
1da177e4c
|
3689 |
{ |
3dafccf22
|
3690 |
return obj_size(cachep); |
1da177e4c
|
3691 3692 |
} EXPORT_SYMBOL(kmem_cache_size); |
e498be7da
|
3693 |
/* |
183ff22bb
|
3694 |
* This initializes kmem_list3 or resizes various caches for all nodes. |
e498be7da
|
3695 |
*/ |
83b519e8b
|
3696 |
static int alloc_kmemlist(struct kmem_cache *cachep, gfp_t gfp) |
e498be7da
|
3697 3698 3699 |
{ int node; struct kmem_list3 *l3; |
cafeb02e0
|
3700 |
struct array_cache *new_shared; |
3395ee058
|
3701 |
struct array_cache **new_alien = NULL; |
e498be7da
|
3702 |
|
9c09a95cf
|
3703 |
for_each_online_node(node) { |
cafeb02e0
|
3704 |
|
3395ee058
|
3705 |
if (use_alien_caches) { |
83b519e8b
|
3706 |
new_alien = alloc_alien_cache(node, cachep->limit, gfp); |
3395ee058
|
3707 3708 3709 |
if (!new_alien) goto fail; } |
cafeb02e0
|
3710 |
|
631098469
|
3711 3712 3713 |
new_shared = NULL; if (cachep->shared) { new_shared = alloc_arraycache(node, |
0718dc2a8
|
3714 |
cachep->shared*cachep->batchcount, |
83b519e8b
|
3715 |
0xbaadf00d, gfp); |
631098469
|
3716 3717 3718 3719 |
if (!new_shared) { free_alien_cache(new_alien); goto fail; } |
0718dc2a8
|
3720 |
} |
cafeb02e0
|
3721 |
|
a737b3e2f
|
3722 3723 |
l3 = cachep->nodelists[node]; if (l3) { |
cafeb02e0
|
3724 |
struct array_cache *shared = l3->shared; |
e498be7da
|
3725 |
spin_lock_irq(&l3->list_lock); |
cafeb02e0
|
3726 |
if (shared) |
0718dc2a8
|
3727 3728 |
free_block(cachep, shared->entry, shared->avail, node); |
e498be7da
|
3729 |
|
cafeb02e0
|
3730 3731 |
l3->shared = new_shared; if (!l3->alien) { |
e498be7da
|
3732 3733 3734 |
l3->alien = new_alien; new_alien = NULL; } |
b28a02de8
|
3735 |
l3->free_limit = (1 + nr_cpus_node(node)) * |
a737b3e2f
|
3736 |
cachep->batchcount + cachep->num; |
e498be7da
|
3737 |
spin_unlock_irq(&l3->list_lock); |
cafeb02e0
|
3738 |
kfree(shared); |
e498be7da
|
3739 3740 3741 |
free_alien_cache(new_alien); continue; } |
83b519e8b
|
3742 |
l3 = kmalloc_node(sizeof(struct kmem_list3), gfp, node); |
0718dc2a8
|
3743 3744 3745 |
if (!l3) { free_alien_cache(new_alien); kfree(new_shared); |
e498be7da
|
3746 |
goto fail; |
0718dc2a8
|
3747 |
} |
e498be7da
|
3748 3749 3750 |
kmem_list3_init(l3); l3->next_reap = jiffies + REAPTIMEOUT_LIST3 + |
a737b3e2f
|
3751 |
((unsigned long)cachep) % REAPTIMEOUT_LIST3; |
cafeb02e0
|
3752 |
l3->shared = new_shared; |
e498be7da
|
3753 |
l3->alien = new_alien; |
b28a02de8
|
3754 |
l3->free_limit = (1 + nr_cpus_node(node)) * |
a737b3e2f
|
3755 |
cachep->batchcount + cachep->num; |
e498be7da
|
3756 3757 |
cachep->nodelists[node] = l3; } |
cafeb02e0
|
3758 |
return 0; |
0718dc2a8
|
3759 |
|
a737b3e2f
|
3760 |
fail: |
0718dc2a8
|
3761 3762 3763 3764 3765 3766 3767 3768 3769 3770 3771 3772 3773 3774 3775 |
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
|
3776 |
return -ENOMEM; |
e498be7da
|
3777 |
} |
1da177e4c
|
3778 |
struct ccupdate_struct { |
343e0d7a9
|
3779 |
struct kmem_cache *cachep; |
1da177e4c
|
3780 3781 3782 3783 3784 |
struct array_cache *new[NR_CPUS]; }; static void do_ccupdate_local(void *info) { |
a737b3e2f
|
3785 |
struct ccupdate_struct *new = info; |
1da177e4c
|
3786 3787 3788 |
struct array_cache *old; check_irq_off(); |
9a2dba4b4
|
3789 |
old = cpu_cache_get(new->cachep); |
e498be7da
|
3790 |
|
1da177e4c
|
3791 3792 3793 |
new->cachep->array[smp_processor_id()] = new->new[smp_processor_id()]; new->new[smp_processor_id()] = old; } |
b5d8ca7c5
|
3794 |
/* Always called with the cache_chain_mutex held */ |
a737b3e2f
|
3795 |
static int do_tune_cpucache(struct kmem_cache *cachep, int limit, |
83b519e8b
|
3796 |
int batchcount, int shared, gfp_t gfp) |
1da177e4c
|
3797 |
{ |
d2e7b7d0a
|
3798 |
struct ccupdate_struct *new; |
2ed3a4ef9
|
3799 |
int i; |
1da177e4c
|
3800 |
|
83b519e8b
|
3801 |
new = kzalloc(sizeof(*new), gfp); |
d2e7b7d0a
|
3802 3803 |
if (!new) return -ENOMEM; |
e498be7da
|
3804 |
for_each_online_cpu(i) { |
7d6e6d09d
|
3805 |
new->new[i] = alloc_arraycache(cpu_to_mem(i), limit, |
83b519e8b
|
3806 |
batchcount, gfp); |
d2e7b7d0a
|
3807 |
if (!new->new[i]) { |
b28a02de8
|
3808 |
for (i--; i >= 0; i--) |
d2e7b7d0a
|
3809 3810 |
kfree(new->new[i]); kfree(new); |
e498be7da
|
3811 |
return -ENOMEM; |
1da177e4c
|
3812 3813 |
} } |
d2e7b7d0a
|
3814 |
new->cachep = cachep; |
1da177e4c
|
3815 |
|
15c8b6c1a
|
3816 |
on_each_cpu(do_ccupdate_local, (void *)new, 1); |
e498be7da
|
3817 |
|
1da177e4c
|
3818 |
check_irq_on(); |
1da177e4c
|
3819 3820 |
cachep->batchcount = batchcount; cachep->limit = limit; |
e498be7da
|
3821 |
cachep->shared = shared; |
1da177e4c
|
3822 |
|
e498be7da
|
3823 |
for_each_online_cpu(i) { |
d2e7b7d0a
|
3824 |
struct array_cache *ccold = new->new[i]; |
1da177e4c
|
3825 3826 |
if (!ccold) continue; |
7d6e6d09d
|
3827 3828 3829 |
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
|
3830 3831 |
kfree(ccold); } |
d2e7b7d0a
|
3832 |
kfree(new); |
83b519e8b
|
3833 |
return alloc_kmemlist(cachep, gfp); |
1da177e4c
|
3834 |
} |
b5d8ca7c5
|
3835 |
/* Called with cache_chain_mutex held always */ |
83b519e8b
|
3836 |
static int enable_cpucache(struct kmem_cache *cachep, gfp_t gfp) |
1da177e4c
|
3837 3838 3839 |
{ int err; int limit, shared; |
a737b3e2f
|
3840 3841 |
/* * The head array serves three purposes: |
1da177e4c
|
3842 3843 |
* - create a LIFO ordering, i.e. return objects that are cache-warm * - reduce the number of spinlock operations. |
a737b3e2f
|
3844 |
* - reduce the number of linked list operations on the slab and |
1da177e4c
|
3845 3846 3847 3848 |
* bufctl chains: array operations are cheaper. * The numbers are guessed, we should auto-tune as described by * Bonwick. */ |
3dafccf22
|
3849 |
if (cachep->buffer_size > 131072) |
1da177e4c
|
3850 |
limit = 1; |
3dafccf22
|
3851 |
else if (cachep->buffer_size > PAGE_SIZE) |
1da177e4c
|
3852 |
limit = 8; |
3dafccf22
|
3853 |
else if (cachep->buffer_size > 1024) |
1da177e4c
|
3854 |
limit = 24; |
3dafccf22
|
3855 |
else if (cachep->buffer_size > 256) |
1da177e4c
|
3856 3857 3858 |
limit = 54; else limit = 120; |
a737b3e2f
|
3859 3860 |
/* * CPU bound tasks (e.g. network routing) can exhibit cpu bound |
1da177e4c
|
3861 3862 3863 3864 3865 3866 3867 3868 |
* 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
|
3869 |
if (cachep->buffer_size <= PAGE_SIZE && num_possible_cpus() > 1) |
1da177e4c
|
3870 |
shared = 8; |
1da177e4c
|
3871 3872 |
#if DEBUG |
a737b3e2f
|
3873 3874 3875 |
/* * With debugging enabled, large batchcount lead to excessively long * periods with disabled local interrupts. Limit the batchcount |
1da177e4c
|
3876 3877 3878 3879 |
*/ if (limit > 32) limit = 32; #endif |
83b519e8b
|
3880 |
err = do_tune_cpucache(cachep, limit, (limit + 1) / 2, shared, gfp); |
1da177e4c
|
3881 3882 3883 |
if (err) printk(KERN_ERR "enable_cpucache failed for %s, error %d. ", |
b28a02de8
|
3884 |
cachep->name, -err); |
2ed3a4ef9
|
3885 |
return err; |
1da177e4c
|
3886 |
} |
1b55253a7
|
3887 3888 |
/* * Drain an array if it contains any elements taking the l3 lock only if |
b18e7e654
|
3889 3890 |
* necessary. Note that the l3 listlock also protects the array_cache * if drain_array() is used on the shared array. |
1b55253a7
|
3891 |
*/ |
68a1b1955
|
3892 |
static void drain_array(struct kmem_cache *cachep, struct kmem_list3 *l3, |
1b55253a7
|
3893 |
struct array_cache *ac, int force, int node) |
1da177e4c
|
3894 3895 |
{ int tofree; |
1b55253a7
|
3896 3897 |
if (!ac || !ac->avail) return; |
1da177e4c
|
3898 3899 |
if (ac->touched && !force) { ac->touched = 0; |
b18e7e654
|
3900 |
} else { |
1b55253a7
|
3901 |
spin_lock_irq(&l3->list_lock); |
b18e7e654
|
3902 3903 3904 3905 3906 3907 3908 3909 3910 |
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
|
3911 |
spin_unlock_irq(&l3->list_lock); |
1da177e4c
|
3912 3913 3914 3915 3916 |
} } /** * cache_reap - Reclaim memory from caches. |
05fb6bf0b
|
3917 |
* @w: work descriptor |
1da177e4c
|
3918 3919 3920 3921 3922 3923 |
* * 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
|
3924 3925 |
* If we cannot acquire the cache chain mutex then just give up - we'll try * again on the next iteration. |
1da177e4c
|
3926 |
*/ |
7c5cae368
|
3927 |
static void cache_reap(struct work_struct *w) |
1da177e4c
|
3928 |
{ |
7a7c381d2
|
3929 |
struct kmem_cache *searchp; |
e498be7da
|
3930 |
struct kmem_list3 *l3; |
7d6e6d09d
|
3931 |
int node = numa_mem_id(); |
bf6aede71
|
3932 |
struct delayed_work *work = to_delayed_work(w); |
1da177e4c
|
3933 |
|
7c5cae368
|
3934 |
if (!mutex_trylock(&cache_chain_mutex)) |
1da177e4c
|
3935 |
/* Give up. Setup the next iteration. */ |
7c5cae368
|
3936 |
goto out; |
1da177e4c
|
3937 |
|
7a7c381d2
|
3938 |
list_for_each_entry(searchp, &cache_chain, next) { |
1da177e4c
|
3939 |
check_irq_on(); |
35386e3b0
|
3940 3941 3942 3943 3944 |
/* * 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
|
3945 |
l3 = searchp->nodelists[node]; |
35386e3b0
|
3946 |
|
8fce4d8e3
|
3947 |
reap_alien(searchp, l3); |
1da177e4c
|
3948 |
|
aab2207cf
|
3949 |
drain_array(searchp, l3, cpu_cache_get(searchp), 0, node); |
1da177e4c
|
3950 |
|
35386e3b0
|
3951 3952 3953 3954 |
/* * These are racy checks but it does not matter * if we skip one check or scan twice. */ |
e498be7da
|
3955 |
if (time_after(l3->next_reap, jiffies)) |
35386e3b0
|
3956 |
goto next; |
1da177e4c
|
3957 |
|
e498be7da
|
3958 |
l3->next_reap = jiffies + REAPTIMEOUT_LIST3; |
1da177e4c
|
3959 |
|
aab2207cf
|
3960 |
drain_array(searchp, l3, l3->shared, 0, node); |
1da177e4c
|
3961 |
|
ed11d9eb2
|
3962 |
if (l3->free_touched) |
e498be7da
|
3963 |
l3->free_touched = 0; |
ed11d9eb2
|
3964 3965 |
else { int freed; |
1da177e4c
|
3966 |
|
ed11d9eb2
|
3967 3968 3969 3970 |
freed = drain_freelist(searchp, l3, (l3->free_limit + 5 * searchp->num - 1) / (5 * searchp->num)); STATS_ADD_REAPED(searchp, freed); } |
35386e3b0
|
3971 |
next: |
1da177e4c
|
3972 3973 3974 |
cond_resched(); } check_irq_on(); |
fc0abb145
|
3975 |
mutex_unlock(&cache_chain_mutex); |
8fce4d8e3
|
3976 |
next_reap_node(); |
7c5cae368
|
3977 |
out: |
a737b3e2f
|
3978 |
/* Set up the next iteration */ |
7c5cae368
|
3979 |
schedule_delayed_work(work, round_jiffies_relative(REAPTIMEOUT_CPUC)); |
1da177e4c
|
3980 |
} |
158a96242
|
3981 |
#ifdef CONFIG_SLABINFO |
1da177e4c
|
3982 |
|
85289f98d
|
3983 |
static void print_slabinfo_header(struct seq_file *m) |
1da177e4c
|
3984 |
{ |
85289f98d
|
3985 3986 3987 3988 |
/* * Output format version, so at least we can change it * without _too_ many complaints. */ |
1da177e4c
|
3989 |
#if STATS |
85289f98d
|
3990 3991 |
seq_puts(m, "slabinfo - version: 2.1 (statistics) "); |
1da177e4c
|
3992 |
#else |
85289f98d
|
3993 3994 |
seq_puts(m, "slabinfo - version: 2.1 "); |
1da177e4c
|
3995 |
#endif |
85289f98d
|
3996 3997 3998 3999 |
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
|
4000 |
#if STATS |
85289f98d
|
4001 |
seq_puts(m, " : globalstat <listallocs> <maxobjs> <grown> <reaped> " |
fb7faf331
|
4002 |
"<error> <maxfreeable> <nodeallocs> <remotefrees> <alienoverflow>"); |
85289f98d
|
4003 |
seq_puts(m, " : cpustat <allochit> <allocmiss> <freehit> <freemiss>"); |
1da177e4c
|
4004 |
#endif |
85289f98d
|
4005 4006 4007 4008 4009 4010 4011 |
seq_putc(m, ' '); } static void *s_start(struct seq_file *m, loff_t *pos) { loff_t n = *pos; |
85289f98d
|
4012 |
|
fc0abb145
|
4013 |
mutex_lock(&cache_chain_mutex); |
85289f98d
|
4014 4015 |
if (!n) print_slabinfo_header(m); |
b92151bab
|
4016 4017 |
return seq_list_start(&cache_chain, *pos); |
1da177e4c
|
4018 4019 4020 4021 |
} static void *s_next(struct seq_file *m, void *p, loff_t *pos) { |
b92151bab
|
4022 |
return seq_list_next(p, &cache_chain, pos); |
1da177e4c
|
4023 4024 4025 4026 |
} static void s_stop(struct seq_file *m, void *p) { |
fc0abb145
|
4027 |
mutex_unlock(&cache_chain_mutex); |
1da177e4c
|
4028 4029 4030 4031 |
} static int s_show(struct seq_file *m, void *p) { |
b92151bab
|
4032 |
struct kmem_cache *cachep = list_entry(p, struct kmem_cache, next); |
b28a02de8
|
4033 4034 4035 4036 4037 |
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
|
4038 |
const char *name; |
1da177e4c
|
4039 |
char *error = NULL; |
e498be7da
|
4040 4041 |
int node; struct kmem_list3 *l3; |
1da177e4c
|
4042 |
|
1da177e4c
|
4043 4044 |
active_objs = 0; num_slabs = 0; |
e498be7da
|
4045 4046 4047 4048 |
for_each_online_node(node) { l3 = cachep->nodelists[node]; if (!l3) continue; |
ca3b9b917
|
4049 4050 |
check_irq_on(); spin_lock_irq(&l3->list_lock); |
e498be7da
|
4051 |
|
7a7c381d2
|
4052 |
list_for_each_entry(slabp, &l3->slabs_full, list) { |
e498be7da
|
4053 4054 4055 4056 4057 |
if (slabp->inuse != cachep->num && !error) error = "slabs_full accounting error"; active_objs += cachep->num; active_slabs++; } |
7a7c381d2
|
4058 |
list_for_each_entry(slabp, &l3->slabs_partial, list) { |
e498be7da
|
4059 4060 4061 4062 4063 4064 4065 |
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
|
4066 |
list_for_each_entry(slabp, &l3->slabs_free, list) { |
e498be7da
|
4067 4068 4069 4070 4071 |
if (slabp->inuse && !error) error = "slabs_free/inuse accounting error"; num_slabs++; } free_objects += l3->free_objects; |
4484ebf12
|
4072 4073 |
if (l3->shared) shared_avail += l3->shared->avail; |
e498be7da
|
4074 |
|
ca3b9b917
|
4075 |
spin_unlock_irq(&l3->list_lock); |
1da177e4c
|
4076 |
} |
b28a02de8
|
4077 4078 |
num_slabs += active_slabs; num_objs = num_slabs * cachep->num; |
e498be7da
|
4079 |
if (num_objs - active_objs != free_objects && !error) |
1da177e4c
|
4080 |
error = "free_objects accounting error"; |
b28a02de8
|
4081 |
name = cachep->name; |
1da177e4c
|
4082 4083 4084 4085 4086 |
if (error) printk(KERN_ERR "slab: cache %s error: %s ", name, error); seq_printf(m, "%-17s %6lu %6lu %6u %4u %4d", |
3dafccf22
|
4087 |
name, active_objs, num_objs, cachep->buffer_size, |
b28a02de8
|
4088 |
cachep->num, (1 << cachep->gfporder)); |
1da177e4c
|
4089 |
seq_printf(m, " : tunables %4u %4u %4u", |
b28a02de8
|
4090 |
cachep->limit, cachep->batchcount, cachep->shared); |
e498be7da
|
4091 |
seq_printf(m, " : slabdata %6lu %6lu %6lu", |
b28a02de8
|
4092 |
active_slabs, num_slabs, shared_avail); |
1da177e4c
|
4093 |
#if STATS |
b28a02de8
|
4094 |
{ /* list3 stats */ |
1da177e4c
|
4095 4096 4097 4098 4099 4100 |
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
|
4101 |
unsigned long node_allocs = cachep->node_allocs; |
e498be7da
|
4102 |
unsigned long node_frees = cachep->node_frees; |
fb7faf331
|
4103 |
unsigned long overflows = cachep->node_overflow; |
1da177e4c
|
4104 |
|
e92dd4fd1
|
4105 4106 4107 4108 4109 |
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
|
4110 4111 4112 4113 4114 4115 4116 4117 4118 |
} /* 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
|
4119 |
allochit, allocmiss, freehit, freemiss); |
1da177e4c
|
4120 4121 4122 4123 |
} #endif seq_putc(m, ' '); |
1da177e4c
|
4124 4125 4126 4127 4128 4129 4130 4131 4132 4133 4134 4135 4136 4137 4138 4139 |
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
|
4140 |
static const struct seq_operations slabinfo_op = { |
b28a02de8
|
4141 4142 4143 4144 |
.start = s_start, .next = s_next, .stop = s_stop, .show = s_show, |
1da177e4c
|
4145 4146 4147 4148 4149 4150 4151 4152 4153 4154 |
}; #define MAX_SLABINFO_WRITE 128 /** * slabinfo_write - Tuning for the slab allocator * @file: unused * @buffer: user buffer * @count: data length * @ppos: unused */ |
68a1b1955
|
4155 |
static ssize_t slabinfo_write(struct file *file, const char __user *buffer, |
b28a02de8
|
4156 |
size_t count, loff_t *ppos) |
1da177e4c
|
4157 |
{ |
b28a02de8
|
4158 |
char kbuf[MAX_SLABINFO_WRITE + 1], *tmp; |
1da177e4c
|
4159 |
int limit, batchcount, shared, res; |
7a7c381d2
|
4160 |
struct kmem_cache *cachep; |
b28a02de8
|
4161 |
|
1da177e4c
|
4162 4163 4164 4165 |
if (count > MAX_SLABINFO_WRITE) return -EINVAL; if (copy_from_user(&kbuf, buffer, count)) return -EFAULT; |
b28a02de8
|
4166 |
kbuf[MAX_SLABINFO_WRITE] = '\0'; |
1da177e4c
|
4167 4168 4169 4170 4171 4172 4173 4174 4175 4176 |
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
|
4177 |
mutex_lock(&cache_chain_mutex); |
1da177e4c
|
4178 |
res = -EINVAL; |
7a7c381d2
|
4179 |
list_for_each_entry(cachep, &cache_chain, next) { |
1da177e4c
|
4180 |
if (!strcmp(cachep->name, kbuf)) { |
a737b3e2f
|
4181 4182 |
if (limit < 1 || batchcount < 1 || batchcount > limit || shared < 0) { |
e498be7da
|
4183 |
res = 0; |
1da177e4c
|
4184 |
} else { |
e498be7da
|
4185 |
res = do_tune_cpucache(cachep, limit, |
83b519e8b
|
4186 4187 |
batchcount, shared, GFP_KERNEL); |
1da177e4c
|
4188 4189 4190 4191 |
} break; } } |
fc0abb145
|
4192 |
mutex_unlock(&cache_chain_mutex); |
1da177e4c
|
4193 4194 4195 4196 |
if (res >= 0) res = count; return res; } |
871751e25
|
4197 |
|
7b3c3a50a
|
4198 4199 4200 4201 4202 4203 4204 4205 4206 4207 4208 4209 |
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
|
4210 4211 4212 4213 |
#ifdef CONFIG_DEBUG_SLAB_LEAK static void *leaks_start(struct seq_file *m, loff_t *pos) { |
871751e25
|
4214 |
mutex_lock(&cache_chain_mutex); |
b92151bab
|
4215 |
return seq_list_start(&cache_chain, *pos); |
871751e25
|
4216 4217 4218 4219 4220 4221 4222 4223 4224 4225 4226 4227 4228 4229 4230 4231 4232 4233 4234 4235 4236 4237 4238 4239 4240 4241 4242 4243 4244 4245 4246 4247 4248 4249 4250 4251 4252 4253 4254 4255 4256 4257 4258 4259 4260 4261 4262 4263 4264 |
} 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
|
4265 |
unsigned long offset, size; |
9281acea6
|
4266 |
char modname[MODULE_NAME_LEN], name[KSYM_NAME_LEN]; |
871751e25
|
4267 |
|
a5c43dae7
|
4268 |
if (lookup_symbol_attrs(address, &size, &offset, modname, name) == 0) { |
871751e25
|
4269 |
seq_printf(m, "%s+%#lx/%#lx", name, offset, size); |
a5c43dae7
|
4270 |
if (modname[0]) |
871751e25
|
4271 4272 4273 4274 4275 4276 4277 4278 4279 |
seq_printf(m, " [%s]", modname); return; } #endif seq_printf(m, "%p", (void *)address); } static int leaks_show(struct seq_file *m, void *p) { |
b92151bab
|
4280 |
struct kmem_cache *cachep = list_entry(p, struct kmem_cache, next); |
871751e25
|
4281 4282 4283 4284 4285 4286 4287 4288 4289 4290 4291 4292 4293 4294 4295 4296 4297 4298 4299 4300 4301 4302 4303 |
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
|
4304 |
list_for_each_entry(slabp, &l3->slabs_full, list) |
871751e25
|
4305 |
handle_slab(n, cachep, slabp); |
7a7c381d2
|
4306 |
list_for_each_entry(slabp, &l3->slabs_partial, list) |
871751e25
|
4307 |
handle_slab(n, cachep, slabp); |
871751e25
|
4308 4309 4310 4311 4312 4313 4314 4315 4316 4317 4318 4319 4320 4321 4322 4323 4324 4325 4326 4327 4328 4329 4330 4331 4332 4333 |
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
|
4334 |
|
871751e25
|
4335 4336 |
return 0; } |
a0ec95a8e
|
4337 |
static const struct seq_operations slabstats_op = { |
871751e25
|
4338 4339 4340 4341 4342 |
.start = leaks_start, .next = s_next, .stop = s_stop, .show = leaks_show, }; |
a0ec95a8e
|
4343 4344 4345 4346 4347 4348 4349 4350 4351 4352 4353 4354 4355 4356 4357 4358 4359 4360 4361 4362 4363 4364 4365 4366 4367 4368 4369 4370 |
static int slabstats_open(struct inode *inode, struct file *file) { unsigned long *n = kzalloc(PAGE_SIZE, GFP_KERNEL); int ret = -ENOMEM; if (n) { ret = seq_open(file, &slabstats_op); if (!ret) { struct seq_file *m = file->private_data; *n = PAGE_SIZE / (2 * sizeof(unsigned long)); m->private = n; n = NULL; } kfree(n); } return ret; } static const struct file_operations proc_slabstats_operations = { .open = slabstats_open, .read = seq_read, .llseek = seq_lseek, .release = seq_release_private, }; #endif static int __init slab_proc_init(void) { |
7b3c3a50a
|
4371 |
proc_create("slabinfo",S_IWUSR|S_IRUGO,NULL,&proc_slabinfo_operations); |
a0ec95a8e
|
4372 4373 |
#ifdef CONFIG_DEBUG_SLAB_LEAK proc_create("slab_allocators", 0, NULL, &proc_slabstats_operations); |
871751e25
|
4374 |
#endif |
a0ec95a8e
|
4375 4376 4377 |
return 0; } module_init(slab_proc_init); |
1da177e4c
|
4378 |
#endif |
00e145b6d
|
4379 4380 4381 4382 4383 4384 4385 4386 4387 4388 4389 4390 |
/** * 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
|
4391 |
size_t ksize(const void *objp) |
1da177e4c
|
4392 |
{ |
ef8b4520b
|
4393 4394 |
BUG_ON(!objp); if (unlikely(objp == ZERO_SIZE_PTR)) |
00e145b6d
|
4395 |
return 0; |
1da177e4c
|
4396 |
|
6ed5eb221
|
4397 |
return obj_size(virt_to_cache(objp)); |
1da177e4c
|
4398 |
} |
b1aabecd5
|
4399 |
EXPORT_SYMBOL(ksize); |