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 |
#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> |
02af61bb5
|
104 |
#include <linux/kmemtrace.h> |
1da177e4c
|
105 |
#include <linux/rcupdate.h> |
543537bd9
|
106 |
#include <linux/string.h> |
138ae6631
|
107 |
#include <linux/uaccess.h> |
e498be7da
|
108 |
#include <linux/nodemask.h> |
d5cff6352
|
109 |
#include <linux/kmemleak.h> |
dc85da15d
|
110 |
#include <linux/mempolicy.h> |
fc0abb145
|
111 |
#include <linux/mutex.h> |
8a8b6502f
|
112 |
#include <linux/fault-inject.h> |
e7eebaf6a
|
113 |
#include <linux/rtmutex.h> |
6a2d7a955
|
114 |
#include <linux/reciprocal_div.h> |
3ac7fe5a4
|
115 |
#include <linux/debugobjects.h> |
c175eea46
|
116 |
#include <linux/kmemcheck.h> |
1da177e4c
|
117 |
|
1da177e4c
|
118 119 120 121 122 |
#include <asm/cacheflush.h> #include <asm/tlbflush.h> #include <asm/page.h> /* |
50953fe9e
|
123 |
* DEBUG - 1 for kmem_cache_create() to honour; SLAB_RED_ZONE & SLAB_POISON. |
1da177e4c
|
124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 |
* 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
|
141 142 |
/* Shouldn't this be in a header file somewhere? */ #define BYTES_PER_WORD sizeof(void *) |
87a927c71
|
143 |
#define REDZONE_ALIGN max(BYTES_PER_WORD, __alignof__(unsigned long long)) |
1da177e4c
|
144 |
|
1da177e4c
|
145 146 147 148 149 150 |
#ifndef ARCH_KMALLOC_MINALIGN /* * Enforce a minimum alignment for the kmalloc caches. * Usually, the kmalloc caches are cache_line_size() aligned, except when * DEBUG and FORCED_DEBUG are enabled, then they are BYTES_PER_WORD aligned. * Some archs want to perform DMA into kmalloc caches and need a guaranteed |
b46b8f19c
|
151 152 153 |
* alignment larger than the alignment of a 64-bit integer. * ARCH_KMALLOC_MINALIGN allows that. * Note that increasing this value may disable some debug features. |
1da177e4c
|
154 |
*/ |
b46b8f19c
|
155 |
#define ARCH_KMALLOC_MINALIGN __alignof__(unsigned long long) |
1da177e4c
|
156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 |
#endif #ifndef ARCH_SLAB_MINALIGN /* * Enforce a minimum alignment for all caches. * Intended for archs that get misalignment faults even for BYTES_PER_WORD * aligned buffers. Includes ARCH_KMALLOC_MINALIGN. * If possible: Do not enable this flag for CONFIG_DEBUG_SLAB, it disables * some debug features. */ #define ARCH_SLAB_MINALIGN 0 #endif #ifndef ARCH_KMALLOC_FLAGS #define ARCH_KMALLOC_FLAGS SLAB_HWCACHE_ALIGN #endif /* Legal flag mask for kmem_cache_create(). */ #if DEBUG |
50953fe9e
|
175 |
# define CREATE_MASK (SLAB_RED_ZONE | \ |
1da177e4c
|
176 |
SLAB_POISON | SLAB_HWCACHE_ALIGN | \ |
ac2b898ca
|
177 |
SLAB_CACHE_DMA | \ |
5af608399
|
178 |
SLAB_STORE_USER | \ |
1da177e4c
|
179 |
SLAB_RECLAIM_ACCOUNT | SLAB_PANIC | \ |
3ac7fe5a4
|
180 |
SLAB_DESTROY_BY_RCU | SLAB_MEM_SPREAD | \ |
c175eea46
|
181 |
SLAB_DEBUG_OBJECTS | SLAB_NOLEAKTRACE | SLAB_NOTRACK) |
1da177e4c
|
182 |
#else |
ac2b898ca
|
183 |
# define CREATE_MASK (SLAB_HWCACHE_ALIGN | \ |
5af608399
|
184 |
SLAB_CACHE_DMA | \ |
1da177e4c
|
185 |
SLAB_RECLAIM_ACCOUNT | SLAB_PANIC | \ |
3ac7fe5a4
|
186 |
SLAB_DESTROY_BY_RCU | SLAB_MEM_SPREAD | \ |
c175eea46
|
187 |
SLAB_DEBUG_OBJECTS | SLAB_NOLEAKTRACE | SLAB_NOTRACK) |
1da177e4c
|
188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 |
#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
|
208 |
typedef unsigned int kmem_bufctl_t; |
1da177e4c
|
209 210 |
#define BUFCTL_END (((kmem_bufctl_t)(~0U))-0) #define BUFCTL_FREE (((kmem_bufctl_t)(~0U))-1) |
871751e25
|
211 212 |
#define BUFCTL_ACTIVE (((kmem_bufctl_t)(~0U))-2) #define SLAB_LIMIT (((kmem_bufctl_t)(~0U))-3) |
1da177e4c
|
213 |
|
1da177e4c
|
214 215 216 217 218 219 220 221 |
/* * struct slab * * Manages the objs in a slab. Placed either at the beginning of mem allocated * for a slab, or allocated from an general cache. * Slabs are chained into three list: fully used, partial, fully free slabs. */ struct slab { |
b28a02de8
|
222 223 224 225 226 227 |
struct list_head list; unsigned long colouroff; void *s_mem; /* including colour offset */ unsigned int inuse; /* num of objs active in slab */ kmem_bufctl_t free; unsigned short nodeid; |
1da177e4c
|
228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 |
}; /* * struct slab_rcu * * slab_destroy on a SLAB_DESTROY_BY_RCU cache uses this structure to * arrange for kmem_freepages to be called via RCU. This is useful if * we need to approach a kernel structure obliquely, from its address * obtained without the usual locking. We can lock the structure to * stabilize it and check it's still at the given address, only if we * can be sure that the memory has not been meanwhile reused for some * other kind of object (which our subsystem's lock might corrupt). * * rcu_read_lock before reading the address, then rcu_read_unlock after * taking the spinlock within the structure expected at that address. * * We assume struct slab_rcu can overlay struct slab when destroying. */ struct slab_rcu { |
b28a02de8
|
247 |
struct rcu_head head; |
343e0d7a9
|
248 |
struct kmem_cache *cachep; |
b28a02de8
|
249 |
void *addr; |
1da177e4c
|
250 251 252 253 254 |
}; /* * struct array_cache * |
1da177e4c
|
255 256 257 258 259 260 261 262 263 264 265 266 267 268 |
* 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
|
269 |
spinlock_t lock; |
bda5b655f
|
270 |
void *entry[]; /* |
a737b3e2f
|
271 272 273 |
* Must have this definition in here for the proper * alignment of array_cache. Also simplifies accessing * the entries. |
a737b3e2f
|
274 |
*/ |
1da177e4c
|
275 |
}; |
a737b3e2f
|
276 277 278 |
/* * bootstrap: The caches do not work without cpuarrays anymore, but the * cpuarrays are allocated from the generic caches... |
1da177e4c
|
279 280 281 282 |
*/ #define BOOT_CPUCACHE_ENTRIES 1 struct arraycache_init { struct array_cache cache; |
b28a02de8
|
283 |
void *entries[BOOT_CPUCACHE_ENTRIES]; |
1da177e4c
|
284 285 286 |
}; /* |
e498be7da
|
287 |
* The slab lists for all objects. |
1da177e4c
|
288 289 |
*/ struct kmem_list3 { |
b28a02de8
|
290 291 292 293 |
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
|
294 |
unsigned int free_limit; |
2e1217cf9
|
295 |
unsigned int colour_next; /* Per-node cache coloring */ |
b28a02de8
|
296 297 298 |
spinlock_t list_lock; struct array_cache *shared; /* shared per node */ struct array_cache **alien; /* on other nodes */ |
35386e3b0
|
299 300 |
unsigned long next_reap; /* updated without locking */ int free_touched; /* updated without locking */ |
1da177e4c
|
301 |
}; |
e498be7da
|
302 303 304 |
/* * Need this for bootstrapping a per node allocator. */ |
556a169da
|
305 |
#define NUM_INIT_LISTS (3 * MAX_NUMNODES) |
e498be7da
|
306 307 |
struct kmem_list3 __initdata initkmem_list3[NUM_INIT_LISTS]; #define CACHE_CACHE 0 |
556a169da
|
308 309 |
#define SIZE_AC MAX_NUMNODES #define SIZE_L3 (2 * MAX_NUMNODES) |
e498be7da
|
310 |
|
ed11d9eb2
|
311 312 313 314 |
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
|
315 |
static int enable_cpucache(struct kmem_cache *cachep, gfp_t gfp); |
65f27f384
|
316 |
static void cache_reap(struct work_struct *unused); |
ed11d9eb2
|
317 |
|
e498be7da
|
318 |
/* |
a737b3e2f
|
319 320 |
* 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
|
321 |
*/ |
7243cc05b
|
322 |
static __always_inline int index_of(const size_t size) |
e498be7da
|
323 |
{ |
5ec8a847b
|
324 |
extern void __bad_size(void); |
e498be7da
|
325 326 327 328 329 330 331 332 |
if (__builtin_constant_p(size)) { int i = 0; #define CACHE(x) \ if (size <=x) \ return i; \ else \ i++; |
1c61fc40f
|
333 |
#include <linux/kmalloc_sizes.h> |
e498be7da
|
334 |
#undef CACHE |
5ec8a847b
|
335 |
__bad_size(); |
7243cc05b
|
336 |
} else |
5ec8a847b
|
337 |
__bad_size(); |
e498be7da
|
338 339 |
return 0; } |
e0a427267
|
340 |
static int slab_early_init = 1; |
e498be7da
|
341 342 |
#define INDEX_AC index_of(sizeof(struct arraycache_init)) #define INDEX_L3 index_of(sizeof(struct kmem_list3)) |
1da177e4c
|
343 |
|
5295a74cc
|
344 |
static void kmem_list3_init(struct kmem_list3 *parent) |
e498be7da
|
345 346 347 348 349 350 |
{ 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
|
351 |
parent->colour_next = 0; |
e498be7da
|
352 353 354 355 |
spin_lock_init(&parent->list_lock); parent->free_objects = 0; parent->free_touched = 0; } |
a737b3e2f
|
356 357 358 359 |
#define MAKE_LIST(cachep, listp, slab, nodeid) \ do { \ INIT_LIST_HEAD(listp); \ list_splice(&(cachep->nodelists[nodeid]->slab), listp); \ |
e498be7da
|
360 |
} while (0) |
a737b3e2f
|
361 362 |
#define MAKE_ALL_LISTS(cachep, ptr, nodeid) \ do { \ |
e498be7da
|
363 364 365 366 |
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
|
367 |
|
1da177e4c
|
368 369 370 371 |
#define CFLGS_OFF_SLAB (0x80000000UL) #define OFF_SLAB(x) ((x)->flags & CFLGS_OFF_SLAB) #define BATCHREFILL_LIMIT 16 |
a737b3e2f
|
372 373 374 |
/* * Optimization question: fewer reaps means less probability for unnessary * cpucache drain/refill cycles. |
1da177e4c
|
375 |
* |
dc6f3f276
|
376 |
* OTOH the cpuarrays can contain lots of objects, |
1da177e4c
|
377 378 379 380 381 382 383 384 385 386 |
* 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
|
387 |
#define STATS_ADD_REAPED(x,y) ((x)->reaped += (y)) |
a737b3e2f
|
388 389 390 391 392 |
#define STATS_SET_HIGH(x) \ do { \ if ((x)->num_active > (x)->high_mark) \ (x)->high_mark = (x)->num_active; \ } while (0) |
1da177e4c
|
393 394 |
#define STATS_INC_ERR(x) ((x)->errors++) #define STATS_INC_NODEALLOCS(x) ((x)->node_allocs++) |
e498be7da
|
395 |
#define STATS_INC_NODEFREES(x) ((x)->node_frees++) |
fb7faf331
|
396 |
#define STATS_INC_ACOVERFLOW(x) ((x)->node_overflow++) |
a737b3e2f
|
397 398 399 400 401 |
#define STATS_SET_FREEABLE(x, i) \ do { \ if ((x)->max_freeable < i) \ (x)->max_freeable = i; \ } while (0) |
1da177e4c
|
402 403 404 405 406 407 408 409 410 |
#define STATS_INC_ALLOCHIT(x) atomic_inc(&(x)->allochit) #define STATS_INC_ALLOCMISS(x) atomic_inc(&(x)->allocmiss) #define STATS_INC_FREEHIT(x) atomic_inc(&(x)->freehit) #define STATS_INC_FREEMISS(x) atomic_inc(&(x)->freemiss) #else #define STATS_INC_ACTIVE(x) do { } while (0) #define STATS_DEC_ACTIVE(x) do { } while (0) #define STATS_INC_ALLOCED(x) do { } while (0) #define STATS_INC_GROWN(x) do { } while (0) |
ed11d9eb2
|
411 |
#define STATS_ADD_REAPED(x,y) do { } while (0) |
1da177e4c
|
412 413 414 |
#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
|
415 |
#define STATS_INC_NODEFREES(x) do { } while (0) |
fb7faf331
|
416 |
#define STATS_INC_ACOVERFLOW(x) do { } while (0) |
a737b3e2f
|
417 |
#define STATS_SET_FREEABLE(x, i) do { } while (0) |
1da177e4c
|
418 419 420 421 422 423 424 |
#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
|
425 |
|
a737b3e2f
|
426 427 |
/* * memory layout of objects: |
1da177e4c
|
428 |
* 0 : objp |
3dafccf22
|
429 |
* 0 .. cachep->obj_offset - BYTES_PER_WORD - 1: padding. This ensures that |
1da177e4c
|
430 431 |
* the end of an object is aligned with the end of the real * allocation. Catches writes behind the end of the allocation. |
3dafccf22
|
432 |
* cachep->obj_offset - BYTES_PER_WORD .. cachep->obj_offset - 1: |
1da177e4c
|
433 |
* redzone word. |
3dafccf22
|
434 435 |
* cachep->obj_offset: The real object. * cachep->buffer_size - 2* BYTES_PER_WORD: redzone word [BYTES_PER_WORD long] |
a737b3e2f
|
436 437 |
* cachep->buffer_size - 1* BYTES_PER_WORD: last caller address * [BYTES_PER_WORD long] |
1da177e4c
|
438 |
*/ |
343e0d7a9
|
439 |
static int obj_offset(struct kmem_cache *cachep) |
1da177e4c
|
440 |
{ |
3dafccf22
|
441 |
return cachep->obj_offset; |
1da177e4c
|
442 |
} |
343e0d7a9
|
443 |
static int obj_size(struct kmem_cache *cachep) |
1da177e4c
|
444 |
{ |
3dafccf22
|
445 |
return cachep->obj_size; |
1da177e4c
|
446 |
} |
b46b8f19c
|
447 |
static unsigned long long *dbg_redzone1(struct kmem_cache *cachep, void *objp) |
1da177e4c
|
448 449 |
{ BUG_ON(!(cachep->flags & SLAB_RED_ZONE)); |
b46b8f19c
|
450 451 |
return (unsigned long long*) (objp + obj_offset(cachep) - sizeof(unsigned long long)); |
1da177e4c
|
452 |
} |
b46b8f19c
|
453 |
static unsigned long long *dbg_redzone2(struct kmem_cache *cachep, void *objp) |
1da177e4c
|
454 455 456 |
{ BUG_ON(!(cachep->flags & SLAB_RED_ZONE)); if (cachep->flags & SLAB_STORE_USER) |
b46b8f19c
|
457 458 |
return (unsigned long long *)(objp + cachep->buffer_size - sizeof(unsigned long long) - |
87a927c71
|
459 |
REDZONE_ALIGN); |
b46b8f19c
|
460 461 |
return (unsigned long long *) (objp + cachep->buffer_size - sizeof(unsigned long long)); |
1da177e4c
|
462 |
} |
343e0d7a9
|
463 |
static void **dbg_userword(struct kmem_cache *cachep, void *objp) |
1da177e4c
|
464 465 |
{ BUG_ON(!(cachep->flags & SLAB_STORE_USER)); |
3dafccf22
|
466 |
return (void **)(objp + cachep->buffer_size - BYTES_PER_WORD); |
1da177e4c
|
467 468 469 |
} #else |
3dafccf22
|
470 471 |
#define obj_offset(x) 0 #define obj_size(cachep) (cachep->buffer_size) |
b46b8f19c
|
472 473 |
#define dbg_redzone1(cachep, objp) ({BUG(); (unsigned long long *)NULL;}) #define dbg_redzone2(cachep, objp) ({BUG(); (unsigned long long *)NULL;}) |
1da177e4c
|
474 475 476 |
#define dbg_userword(cachep, objp) ({BUG(); (void **)NULL;}) #endif |
0f24f1287
|
477 |
#ifdef CONFIG_TRACING |
36555751c
|
478 479 480 481 482 483 |
size_t slab_buffer_size(struct kmem_cache *cachep) { return cachep->buffer_size; } EXPORT_SYMBOL(slab_buffer_size); #endif |
1da177e4c
|
484 |
/* |
1da177e4c
|
485 486 487 488 489 |
* 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
|
490 491 492 493 |
/* * 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
|
494 |
*/ |
065d41cb2
|
495 496 497 498 499 500 501 |
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
|
502 |
page = compound_head(page); |
ddc2e812d
|
503 |
BUG_ON(!PageSlab(page)); |
065d41cb2
|
504 505 506 507 508 509 510 511 512 513 |
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
|
514 |
BUG_ON(!PageSlab(page)); |
065d41cb2
|
515 516 |
return (struct slab *)page->lru.prev; } |
1da177e4c
|
517 |
|
6ed5eb221
|
518 519 |
static inline struct kmem_cache *virt_to_cache(const void *obj) { |
b49af68ff
|
520 |
struct page *page = virt_to_head_page(obj); |
6ed5eb221
|
521 522 523 524 525 |
return page_get_cache(page); } static inline struct slab *virt_to_slab(const void *obj) { |
b49af68ff
|
526 |
struct page *page = virt_to_head_page(obj); |
6ed5eb221
|
527 528 |
return page_get_slab(page); } |
8fea4e96a
|
529 530 531 532 533 |
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
|
534 535 536 537 538 539 540 541 |
/* * 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
|
542 |
{ |
6a2d7a955
|
543 544 |
u32 offset = (obj - slab->s_mem); return reciprocal_divide(offset, cache->reciprocal_buffer_size); |
8fea4e96a
|
545 |
} |
a737b3e2f
|
546 547 548 |
/* * These are the default caches for kmalloc. Custom caches can have other sizes. */ |
1da177e4c
|
549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 |
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
|
566 |
{NULL,} |
1da177e4c
|
567 568 569 570 |
#undef CACHE }; static struct arraycache_init initarray_cache __initdata = |
b28a02de8
|
571 |
{ {0, BOOT_CPUCACHE_ENTRIES, 1, 0} }; |
1da177e4c
|
572 |
static struct arraycache_init initarray_generic = |
b28a02de8
|
573 |
{ {0, BOOT_CPUCACHE_ENTRIES, 1, 0} }; |
1da177e4c
|
574 575 |
/* internal cache of cache description objs */ |
343e0d7a9
|
576 |
static struct kmem_cache cache_cache = { |
b28a02de8
|
577 578 579 |
.batchcount = 1, .limit = BOOT_CPUCACHE_ENTRIES, .shared = 1, |
343e0d7a9
|
580 |
.buffer_size = sizeof(struct kmem_cache), |
b28a02de8
|
581 |
.name = "kmem_cache", |
1da177e4c
|
582 |
}; |
056c62418
|
583 |
#define BAD_ALIEN_MAGIC 0x01020304ul |
ce79ddc8e
|
584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 |
/* * 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
|
603 604 605 606 607 608 609 610 |
#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
|
611 612 613 614 |
* * 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
|
615 |
*/ |
056c62418
|
616 617 |
static struct lock_class_key on_slab_l3_key; static struct lock_class_key on_slab_alc_key; |
ce79ddc8e
|
618 |
static void init_node_lock_keys(int q) |
f1aaee53f
|
619 |
{ |
056c62418
|
620 |
struct cache_sizes *s = malloc_sizes; |
ce79ddc8e
|
621 622 623 624 625 626 627 628 629 630 |
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
|
631 |
continue; |
ce79ddc8e
|
632 633 634 635 636 637 638 639 640 641 |
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
|
642 |
continue; |
ce79ddc8e
|
643 644 645 646 |
for_each_node(r) { if (alc[r]) lockdep_set_class(&alc[r]->lock, &on_slab_alc_key); |
056c62418
|
647 |
} |
f1aaee53f
|
648 649 |
} } |
ce79ddc8e
|
650 651 652 653 654 655 656 657 |
static inline void init_lock_keys(void) { int node; for_each_node(node) init_node_lock_keys(node); } |
f1aaee53f
|
658 |
#else |
ce79ddc8e
|
659 660 661 |
static void init_node_lock_keys(int q) { } |
056c62418
|
662 |
static inline void init_lock_keys(void) |
f1aaee53f
|
663 664 665 |
{ } #endif |
8f5be20bf
|
666 |
/* |
95402b382
|
667 |
* Guard access to the cache-chain. |
8f5be20bf
|
668 |
*/ |
fc0abb145
|
669 |
static DEFINE_MUTEX(cache_chain_mutex); |
1da177e4c
|
670 |
static struct list_head cache_chain; |
1871e52c7
|
671 |
static DEFINE_PER_CPU(struct delayed_work, slab_reap_work); |
1da177e4c
|
672 |
|
343e0d7a9
|
673 |
static inline struct array_cache *cpu_cache_get(struct kmem_cache *cachep) |
1da177e4c
|
674 675 676 |
{ return cachep->array[smp_processor_id()]; } |
a737b3e2f
|
677 678 |
static inline struct kmem_cache *__find_general_cachep(size_t size, gfp_t gfpflags) |
1da177e4c
|
679 680 681 682 683 |
{ struct cache_sizes *csizep = malloc_sizes; #if DEBUG /* This happens if someone tries to call |
b28a02de8
|
684 685 686 |
* kmem_cache_create(), or __kmalloc(), before * the generic caches are initialized. */ |
c7e43c78a
|
687 |
BUG_ON(malloc_sizes[INDEX_AC].cs_cachep == NULL); |
1da177e4c
|
688 |
#endif |
6cb8f9132
|
689 690 |
if (!size) return ZERO_SIZE_PTR; |
1da177e4c
|
691 692 693 694 |
while (size > csizep->cs_size) csizep++; /* |
0abf40c1a
|
695 |
* Really subtle: The last entry with cs->cs_size==ULONG_MAX |
1da177e4c
|
696 697 698 |
* has cs_{dma,}cachep==NULL. Thus no special case * for large kmalloc calls required. */ |
4b51d6698
|
699 |
#ifdef CONFIG_ZONE_DMA |
1da177e4c
|
700 701 |
if (unlikely(gfpflags & GFP_DMA)) return csizep->cs_dmacachep; |
4b51d6698
|
702 |
#endif |
1da177e4c
|
703 704 |
return csizep->cs_cachep; } |
b221385bc
|
705 |
static struct kmem_cache *kmem_find_general_cachep(size_t size, gfp_t gfpflags) |
97e2bde47
|
706 707 708 |
{ return __find_general_cachep(size, gfpflags); } |
97e2bde47
|
709 |
|
fbaccacff
|
710 |
static size_t slab_mgmt_size(size_t nr_objs, size_t align) |
1da177e4c
|
711 |
{ |
fbaccacff
|
712 713 |
return ALIGN(sizeof(struct slab)+nr_objs*sizeof(kmem_bufctl_t), align); } |
1da177e4c
|
714 |
|
a737b3e2f
|
715 716 717 |
/* * Calculate the number of objects and left-over bytes for a given buffer size. */ |
fbaccacff
|
718 719 720 721 722 723 724 |
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
|
725 |
|
fbaccacff
|
726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 |
/* * 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
|
774 |
} |
d40cee245
|
775 |
#define slab_error(cachep, msg) __slab_error(__func__, cachep, msg) |
1da177e4c
|
776 |
|
a737b3e2f
|
777 778 |
static void __slab_error(const char *function, struct kmem_cache *cachep, char *msg) |
1da177e4c
|
779 780 781 |
{ printk(KERN_ERR "slab error in %s(): cache `%s': %s ", |
b28a02de8
|
782 |
function, cachep->name, msg); |
1da177e4c
|
783 784 |
dump_stack(); } |
3395ee058
|
785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 |
/* * 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
|
800 801 802 803 804 805 806 |
#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
|
807 |
static DEFINE_PER_CPU(unsigned long, slab_reap_node); |
8fce4d8e3
|
808 809 810 811 812 813 814 |
static void init_reap_node(int cpu) { int node; node = next_node(cpu_to_node(cpu), node_online_map); if (node == MAX_NUMNODES) |
442295c94
|
815 |
node = first_node(node_online_map); |
8fce4d8e3
|
816 |
|
1871e52c7
|
817 |
per_cpu(slab_reap_node, cpu) = node; |
8fce4d8e3
|
818 819 820 821 |
} static void next_reap_node(void) { |
1871e52c7
|
822 |
int node = __get_cpu_var(slab_reap_node); |
8fce4d8e3
|
823 |
|
8fce4d8e3
|
824 825 826 |
node = next_node(node, node_online_map); if (unlikely(node >= MAX_NUMNODES)) node = first_node(node_online_map); |
1871e52c7
|
827 |
__get_cpu_var(slab_reap_node) = node; |
8fce4d8e3
|
828 829 830 831 832 833 |
} #else #define init_reap_node(cpu) do { } while (0) #define next_reap_node(void) do { } while (0) #endif |
1da177e4c
|
834 835 836 837 838 839 840 |
/* * 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
|
841 |
static void __cpuinit start_cpu_timer(int cpu) |
1da177e4c
|
842 |
{ |
1871e52c7
|
843 |
struct delayed_work *reap_work = &per_cpu(slab_reap_work, cpu); |
1da177e4c
|
844 845 846 847 848 849 |
/* * When this gets called from do_initcalls via cpucache_init(), * init_workqueues() has already run, so keventd will be setup * at that time. */ |
52bad64d9
|
850 |
if (keventd_up() && reap_work->work.func == NULL) { |
8fce4d8e3
|
851 |
init_reap_node(cpu); |
65f27f384
|
852 |
INIT_DELAYED_WORK(reap_work, cache_reap); |
2b2842146
|
853 854 |
schedule_delayed_work_on(cpu, reap_work, __round_jiffies_relative(HZ, cpu)); |
1da177e4c
|
855 856 |
} } |
e498be7da
|
857 |
static struct array_cache *alloc_arraycache(int node, int entries, |
83b519e8b
|
858 |
int batchcount, gfp_t gfp) |
1da177e4c
|
859 |
{ |
b28a02de8
|
860 |
int memsize = sizeof(void *) * entries + sizeof(struct array_cache); |
1da177e4c
|
861 |
struct array_cache *nc = NULL; |
83b519e8b
|
862 |
nc = kmalloc_node(memsize, gfp, node); |
d5cff6352
|
863 864 865 866 867 868 869 870 |
/* * The array_cache structures contain pointers to free object. * However, when such objects are allocated or transfered to another * 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
|
871 872 873 874 875 |
if (nc) { nc->avail = 0; nc->limit = entries; nc->batchcount = batchcount; nc->touched = 0; |
e498be7da
|
876 |
spin_lock_init(&nc->lock); |
1da177e4c
|
877 878 879 |
} return nc; } |
3ded175a4
|
880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 |
/* * Transfer objects in one arraycache to another. * Locking must be handled by the caller. * * Return the number of entries transferred. */ static int transfer_objects(struct array_cache *to, struct array_cache *from, unsigned int max) { /* Figure out how many entries to transfer */ int nr = min(min(from->avail, max), to->limit - to->avail); if (!nr) return 0; memcpy(to->entry + to->avail, from->entry + from->avail -nr, sizeof(void *) *nr); from->avail -= nr; to->avail += nr; |
3ded175a4
|
900 901 |
return nr; } |
765c4507a
|
902 903 904 905 |
#ifndef CONFIG_NUMA #define drain_alien_cache(cachep, alien) do { } while (0) #define reap_alien(cachep, l3) do { } while (0) |
83b519e8b
|
906 |
static inline struct array_cache **alloc_alien_cache(int node, int limit, gfp_t gfp) |
765c4507a
|
907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 |
{ 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
|
925 |
static inline void *____cache_alloc_node(struct kmem_cache *cachep, |
765c4507a
|
926 927 928 929 930 931 |
gfp_t flags, int nodeid) { return NULL; } #else /* CONFIG_NUMA */ |
8b98c1699
|
932 |
static void *____cache_alloc_node(struct kmem_cache *, gfp_t, int); |
c61afb181
|
933 |
static void *alternate_node_alloc(struct kmem_cache *, gfp_t); |
dc85da15d
|
934 |
|
83b519e8b
|
935 |
static struct array_cache **alloc_alien_cache(int node, int limit, gfp_t gfp) |
e498be7da
|
936 937 |
{ struct array_cache **ac_ptr; |
8ef828668
|
938 |
int memsize = sizeof(void *) * nr_node_ids; |
e498be7da
|
939 940 941 942 |
int i; if (limit > 1) limit = 12; |
f3186a9c5
|
943 |
ac_ptr = kzalloc_node(memsize, gfp, node); |
e498be7da
|
944 945 |
if (ac_ptr) { for_each_node(i) { |
f3186a9c5
|
946 |
if (i == node || !node_online(i)) |
e498be7da
|
947 |
continue; |
83b519e8b
|
948 |
ac_ptr[i] = alloc_arraycache(node, limit, 0xbaadf00d, gfp); |
e498be7da
|
949 |
if (!ac_ptr[i]) { |
cc550defe
|
950 |
for (i--; i >= 0; i--) |
e498be7da
|
951 952 953 954 955 956 957 958 |
kfree(ac_ptr[i]); kfree(ac_ptr); return NULL; } } } return ac_ptr; } |
5295a74cc
|
959 |
static void free_alien_cache(struct array_cache **ac_ptr) |
e498be7da
|
960 961 962 963 964 |
{ int i; if (!ac_ptr) return; |
e498be7da
|
965 |
for_each_node(i) |
b28a02de8
|
966 |
kfree(ac_ptr[i]); |
e498be7da
|
967 968 |
kfree(ac_ptr); } |
343e0d7a9
|
969 |
static void __drain_alien_cache(struct kmem_cache *cachep, |
5295a74cc
|
970 |
struct array_cache *ac, int node) |
e498be7da
|
971 972 973 974 975 |
{ struct kmem_list3 *rl3 = cachep->nodelists[node]; if (ac->avail) { spin_lock(&rl3->list_lock); |
e00946fe2
|
976 977 978 979 980 |
/* * 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
|
981 982 |
if (rl3->shared) transfer_objects(rl3->shared, ac, ac->limit); |
e00946fe2
|
983 |
|
ff69416e6
|
984 |
free_block(cachep, ac->entry, ac->avail, node); |
e498be7da
|
985 986 987 988 |
ac->avail = 0; spin_unlock(&rl3->list_lock); } } |
8fce4d8e3
|
989 990 991 992 993 |
/* * Called from cache_reap() to regularly drain alien caches round robin. */ static void reap_alien(struct kmem_cache *cachep, struct kmem_list3 *l3) { |
1871e52c7
|
994 |
int node = __get_cpu_var(slab_reap_node); |
8fce4d8e3
|
995 996 997 |
if (l3->alien) { struct array_cache *ac = l3->alien[node]; |
e00946fe2
|
998 999 |
if (ac && ac->avail && spin_trylock_irq(&ac->lock)) { |
8fce4d8e3
|
1000 1001 1002 1003 1004 |
__drain_alien_cache(cachep, ac, node); spin_unlock_irq(&ac->lock); } } } |
a737b3e2f
|
1005 1006 |
static void drain_alien_cache(struct kmem_cache *cachep, struct array_cache **alien) |
e498be7da
|
1007 |
{ |
b28a02de8
|
1008 |
int i = 0; |
e498be7da
|
1009 1010 1011 1012 |
struct array_cache *ac; unsigned long flags; for_each_online_node(i) { |
4484ebf12
|
1013 |
ac = alien[i]; |
e498be7da
|
1014 1015 1016 1017 1018 1019 1020 |
if (ac) { spin_lock_irqsave(&ac->lock, flags); __drain_alien_cache(cachep, ac, i); spin_unlock_irqrestore(&ac->lock, flags); } } } |
729bd0b74
|
1021 |
|
873623dfa
|
1022 |
static inline int cache_free_alien(struct kmem_cache *cachep, void *objp) |
729bd0b74
|
1023 1024 1025 1026 1027 |
{ struct slab *slabp = virt_to_slab(objp); int nodeid = slabp->nodeid; struct kmem_list3 *l3; struct array_cache *alien = NULL; |
1ca4cb241
|
1028 1029 1030 |
int node; node = numa_node_id(); |
729bd0b74
|
1031 1032 1033 1034 1035 |
/* * Make sure we are not freeing a object from another node to the array * cache on this cpu. */ |
62918a036
|
1036 |
if (likely(slabp->nodeid == node)) |
729bd0b74
|
1037 |
return 0; |
1ca4cb241
|
1038 |
l3 = cachep->nodelists[node]; |
729bd0b74
|
1039 1040 1041 |
STATS_INC_NODEFREES(cachep); if (l3->alien && l3->alien[nodeid]) { alien = l3->alien[nodeid]; |
873623dfa
|
1042 |
spin_lock(&alien->lock); |
729bd0b74
|
1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 |
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
|
1056 |
#endif |
fbf1e473b
|
1057 1058 1059 1060 1061 |
static void __cpuinit cpuup_canceled(long cpu) { struct kmem_cache *cachep; struct kmem_list3 *l3 = NULL; int node = cpu_to_node(cpu); |
a70f73028
|
1062 |
const struct cpumask *mask = cpumask_of_node(node); |
fbf1e473b
|
1063 1064 1065 1066 1067 |
list_for_each_entry(cachep, &cache_chain, next) { struct array_cache *nc; struct array_cache *shared; struct array_cache **alien; |
fbf1e473b
|
1068 |
|
fbf1e473b
|
1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 |
/* 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
|
1083 |
if (!cpumask_empty(mask)) { |
fbf1e473b
|
1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 |
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
|
1122 |
{ |
343e0d7a9
|
1123 |
struct kmem_cache *cachep; |
e498be7da
|
1124 1125 |
struct kmem_list3 *l3 = NULL; int node = cpu_to_node(cpu); |
ea02e3dde
|
1126 |
const int memsize = sizeof(struct kmem_list3); |
1da177e4c
|
1127 |
|
fbf1e473b
|
1128 1129 1130 1131 1132 1133 1134 1135 |
/* * We need to do this right in the beginning since * alloc_arraycache's are going to use this list. * kmalloc_node allows us to add the slab to the right * kmem_list3 and not this cpu's kmem_list3 */ list_for_each_entry(cachep, &cache_chain, next) { |
a737b3e2f
|
1136 |
/* |
fbf1e473b
|
1137 1138 1139 |
* Set up the size64 kmemlist for cpu before we can * begin anything. Make sure some other cpu on this * node has not already allocated this |
e498be7da
|
1140 |
*/ |
fbf1e473b
|
1141 1142 1143 1144 1145 1146 1147 |
if (!cachep->nodelists[node]) { l3 = kmalloc_node(memsize, GFP_KERNEL, node); if (!l3) goto bad; kmem_list3_init(l3); l3->next_reap = jiffies + REAPTIMEOUT_LIST3 + ((unsigned long)cachep) % REAPTIMEOUT_LIST3; |
e498be7da
|
1148 |
|
a737b3e2f
|
1149 |
/* |
fbf1e473b
|
1150 1151 1152 |
* The l3s don't come and go as CPUs come and * go. cache_chain_mutex is sufficient * protection here. |
e498be7da
|
1153 |
*/ |
fbf1e473b
|
1154 |
cachep->nodelists[node] = l3; |
e498be7da
|
1155 |
} |
fbf1e473b
|
1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 |
spin_lock_irq(&cachep->nodelists[node]->list_lock); cachep->nodelists[node]->free_limit = (1 + nr_cpus_node(node)) * cachep->batchcount + cachep->num; spin_unlock_irq(&cachep->nodelists[node]->list_lock); } /* * Now we can go ahead with allocating the shared arrays and * array caches */ list_for_each_entry(cachep, &cache_chain, next) { struct array_cache *nc; struct array_cache *shared = NULL; struct array_cache **alien = NULL; nc = alloc_arraycache(node, cachep->limit, |
83b519e8b
|
1173 |
cachep->batchcount, GFP_KERNEL); |
fbf1e473b
|
1174 1175 1176 1177 1178 |
if (!nc) goto bad; if (cachep->shared) { shared = alloc_arraycache(node, cachep->shared * cachep->batchcount, |
83b519e8b
|
1179 |
0xbaadf00d, GFP_KERNEL); |
12d00f6a1
|
1180 1181 |
if (!shared) { kfree(nc); |
1da177e4c
|
1182 |
goto bad; |
12d00f6a1
|
1183 |
} |
fbf1e473b
|
1184 1185 |
} if (use_alien_caches) { |
83b519e8b
|
1186 |
alien = alloc_alien_cache(node, cachep->limit, GFP_KERNEL); |
12d00f6a1
|
1187 1188 1189 |
if (!alien) { kfree(shared); kfree(nc); |
fbf1e473b
|
1190 |
goto bad; |
12d00f6a1
|
1191 |
} |
fbf1e473b
|
1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 |
} 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
|
1206 |
#ifdef CONFIG_NUMA |
fbf1e473b
|
1207 1208 1209 |
if (!l3->alien) { l3->alien = alien; alien = NULL; |
1da177e4c
|
1210 |
} |
fbf1e473b
|
1211 1212 1213 1214 1215 |
#endif spin_unlock_irq(&l3->list_lock); kfree(shared); free_alien_cache(alien); } |
ce79ddc8e
|
1216 |
init_node_lock_keys(node); |
fbf1e473b
|
1217 1218 |
return 0; bad: |
12d00f6a1
|
1219 |
cpuup_canceled(cpu); |
fbf1e473b
|
1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 |
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
|
1230 1231 |
case CPU_UP_PREPARE: case CPU_UP_PREPARE_FROZEN: |
95402b382
|
1232 |
mutex_lock(&cache_chain_mutex); |
fbf1e473b
|
1233 |
err = cpuup_prepare(cpu); |
95402b382
|
1234 |
mutex_unlock(&cache_chain_mutex); |
1da177e4c
|
1235 1236 |
break; case CPU_ONLINE: |
8bb784428
|
1237 |
case CPU_ONLINE_FROZEN: |
1da177e4c
|
1238 1239 1240 |
start_cpu_timer(cpu); break; #ifdef CONFIG_HOTPLUG_CPU |
5830c5902
|
1241 |
case CPU_DOWN_PREPARE: |
8bb784428
|
1242 |
case CPU_DOWN_PREPARE_FROZEN: |
5830c5902
|
1243 1244 1245 1246 1247 1248 |
/* * 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. */ |
1871e52c7
|
1249 |
cancel_rearming_delayed_work(&per_cpu(slab_reap_work, cpu)); |
5830c5902
|
1250 |
/* Now the cache_reaper is guaranteed to be not running. */ |
1871e52c7
|
1251 |
per_cpu(slab_reap_work, cpu).work.func = NULL; |
5830c5902
|
1252 1253 |
break; case CPU_DOWN_FAILED: |
8bb784428
|
1254 |
case CPU_DOWN_FAILED_FROZEN: |
5830c5902
|
1255 1256 |
start_cpu_timer(cpu); break; |
1da177e4c
|
1257 |
case CPU_DEAD: |
8bb784428
|
1258 |
case CPU_DEAD_FROZEN: |
4484ebf12
|
1259 1260 1261 1262 1263 1264 1265 1266 |
/* * 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
|
1267 |
/* fall through */ |
8f5be20bf
|
1268 |
#endif |
1da177e4c
|
1269 |
case CPU_UP_CANCELED: |
8bb784428
|
1270 |
case CPU_UP_CANCELED_FROZEN: |
95402b382
|
1271 |
mutex_lock(&cache_chain_mutex); |
fbf1e473b
|
1272 |
cpuup_canceled(cpu); |
fc0abb145
|
1273 |
mutex_unlock(&cache_chain_mutex); |
1da177e4c
|
1274 |
break; |
1da177e4c
|
1275 |
} |
fbf1e473b
|
1276 |
return err ? NOTIFY_BAD : NOTIFY_OK; |
1da177e4c
|
1277 |
} |
74b85f379
|
1278 1279 1280 |
static struct notifier_block __cpuinitdata cpucache_notifier = { &cpuup_callback, NULL, 0 }; |
1da177e4c
|
1281 |
|
e498be7da
|
1282 1283 1284 |
/* * swap the static kmem_list3 with kmalloced memory */ |
a737b3e2f
|
1285 1286 |
static void init_list(struct kmem_cache *cachep, struct kmem_list3 *list, int nodeid) |
e498be7da
|
1287 1288 |
{ struct kmem_list3 *ptr; |
83b519e8b
|
1289 |
ptr = kmalloc_node(sizeof(struct kmem_list3), GFP_NOWAIT, nodeid); |
e498be7da
|
1290 |
BUG_ON(!ptr); |
e498be7da
|
1291 |
memcpy(ptr, list, sizeof(struct kmem_list3)); |
2b2d5493e
|
1292 1293 1294 1295 |
/* * Do not assume that spinlocks can be initialized via memcpy: */ spin_lock_init(&ptr->list_lock); |
e498be7da
|
1296 1297 |
MAKE_ALL_LISTS(cachep, ptr, nodeid); cachep->nodelists[nodeid] = ptr; |
e498be7da
|
1298 |
} |
a737b3e2f
|
1299 |
/* |
556a169da
|
1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 |
* 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
|
1316 1317 |
* Initialisation. Called after the page allocator have been initialised and * before smp_init(). |
1da177e4c
|
1318 1319 1320 1321 1322 1323 |
*/ void __init kmem_cache_init(void) { size_t left_over; struct cache_sizes *sizes; struct cache_names *names; |
e498be7da
|
1324 |
int i; |
07ed76b2a
|
1325 |
int order; |
1ca4cb241
|
1326 |
int node; |
e498be7da
|
1327 |
|
b6e68bc1b
|
1328 |
if (num_possible_nodes() == 1) |
62918a036
|
1329 |
use_alien_caches = 0; |
e498be7da
|
1330 1331 1332 1333 1334 |
for (i = 0; i < NUM_INIT_LISTS; i++) { kmem_list3_init(&initkmem_list3[i]); if (i < MAX_NUMNODES) cache_cache.nodelists[i] = NULL; } |
556a169da
|
1335 |
set_up_list3s(&cache_cache, CACHE_CACHE); |
1da177e4c
|
1336 1337 1338 1339 1340 |
/* * Fragmentation resistance on low memory - only use bigger * page orders on machines with more than 32MB of memory. */ |
4481374ce
|
1341 |
if (totalram_pages > (32 << 20) >> PAGE_SHIFT) |
1da177e4c
|
1342 |
slab_break_gfp_order = BREAK_GFP_ORDER_HI; |
1da177e4c
|
1343 1344 |
/* Bootstrap is tricky, because several objects are allocated * from caches that do not exist yet: |
a737b3e2f
|
1345 1346 1347 |
* 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
|
1348 1349 1350 |
* 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
|
1351 |
* 2) Create the first kmalloc cache. |
343e0d7a9
|
1352 |
* The struct kmem_cache for the new cache is allocated normally. |
e498be7da
|
1353 1354 1355 |
* An __init data area is used for the head array. * 3) Create the remaining kmalloc caches, with minimally sized * head arrays. |
1da177e4c
|
1356 1357 |
* 4) Replace the __init data head arrays for cache_cache and the first * kmalloc cache with kmalloc allocated arrays. |
e498be7da
|
1358 1359 1360 |
* 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
|
1361 |
*/ |
1ca4cb241
|
1362 |
node = numa_node_id(); |
1da177e4c
|
1363 |
/* 1) create the cache_cache */ |
1da177e4c
|
1364 1365 1366 1367 |
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
|
1368 |
cache_cache.nodelists[node] = &initkmem_list3[CACHE_CACHE + node]; |
1da177e4c
|
1369 |
|
8da3430d8
|
1370 1371 1372 1373 1374 1375 1376 1377 1378 |
/* * 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
|
1379 1380 |
cache_cache.buffer_size = ALIGN(cache_cache.buffer_size, cache_line_size()); |
6a2d7a955
|
1381 1382 |
cache_cache.reciprocal_buffer_size = reciprocal_value(cache_cache.buffer_size); |
1da177e4c
|
1383 |
|
07ed76b2a
|
1384 1385 1386 1387 1388 1389 |
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
|
1390 |
BUG_ON(!cache_cache.num); |
07ed76b2a
|
1391 |
cache_cache.gfporder = order; |
b28a02de8
|
1392 |
cache_cache.colour = left_over / cache_cache.colour_off; |
b28a02de8
|
1393 1394 |
cache_cache.slab_size = ALIGN(cache_cache.num * sizeof(kmem_bufctl_t) + sizeof(struct slab), cache_line_size()); |
1da177e4c
|
1395 1396 1397 1398 |
/* 2+3) create the kmalloc caches */ sizes = malloc_sizes; names = cache_names; |
a737b3e2f
|
1399 1400 1401 1402 |
/* * Initialize the caches that provide memory for the array cache and the * kmem_list3 structures first. Without this, further allocations will * bug. |
e498be7da
|
1403 1404 1405 |
*/ sizes[INDEX_AC].cs_cachep = kmem_cache_create(names[INDEX_AC].name, |
a737b3e2f
|
1406 1407 1408 |
sizes[INDEX_AC].cs_size, ARCH_KMALLOC_MINALIGN, ARCH_KMALLOC_FLAGS|SLAB_PANIC, |
20c2df83d
|
1409 |
NULL); |
e498be7da
|
1410 |
|
a737b3e2f
|
1411 |
if (INDEX_AC != INDEX_L3) { |
e498be7da
|
1412 |
sizes[INDEX_L3].cs_cachep = |
a737b3e2f
|
1413 1414 1415 1416 |
kmem_cache_create(names[INDEX_L3].name, sizes[INDEX_L3].cs_size, ARCH_KMALLOC_MINALIGN, ARCH_KMALLOC_FLAGS|SLAB_PANIC, |
20c2df83d
|
1417 |
NULL); |
a737b3e2f
|
1418 |
} |
e498be7da
|
1419 |
|
e0a427267
|
1420 |
slab_early_init = 0; |
1da177e4c
|
1421 |
while (sizes->cs_size != ULONG_MAX) { |
e498be7da
|
1422 1423 |
/* * For performance, all the general caches are L1 aligned. |
1da177e4c
|
1424 1425 1426 |
* This should be particularly beneficial on SMP boxes, as it * eliminates "false sharing". * Note for systems short on memory removing the alignment will |
e498be7da
|
1427 1428 |
* allow tighter packing of the smaller caches. */ |
a737b3e2f
|
1429 |
if (!sizes->cs_cachep) { |
e498be7da
|
1430 |
sizes->cs_cachep = kmem_cache_create(names->name, |
a737b3e2f
|
1431 1432 1433 |
sizes->cs_size, ARCH_KMALLOC_MINALIGN, ARCH_KMALLOC_FLAGS|SLAB_PANIC, |
20c2df83d
|
1434 |
NULL); |
a737b3e2f
|
1435 |
} |
4b51d6698
|
1436 1437 1438 |
#ifdef CONFIG_ZONE_DMA sizes->cs_dmacachep = kmem_cache_create( names->name_dma, |
a737b3e2f
|
1439 1440 1441 1442 |
sizes->cs_size, ARCH_KMALLOC_MINALIGN, ARCH_KMALLOC_FLAGS|SLAB_CACHE_DMA| SLAB_PANIC, |
20c2df83d
|
1443 |
NULL); |
4b51d6698
|
1444 |
#endif |
1da177e4c
|
1445 1446 1447 1448 1449 |
sizes++; names++; } /* 4) Replace the bootstrap head arrays */ { |
2b2d5493e
|
1450 |
struct array_cache *ptr; |
e498be7da
|
1451 |
|
83b519e8b
|
1452 |
ptr = kmalloc(sizeof(struct arraycache_init), GFP_NOWAIT); |
e498be7da
|
1453 |
|
9a2dba4b4
|
1454 1455 |
BUG_ON(cpu_cache_get(&cache_cache) != &initarray_cache.cache); memcpy(ptr, cpu_cache_get(&cache_cache), |
b28a02de8
|
1456 |
sizeof(struct arraycache_init)); |
2b2d5493e
|
1457 1458 1459 1460 |
/* * Do not assume that spinlocks can be initialized via memcpy: */ spin_lock_init(&ptr->lock); |
1da177e4c
|
1461 |
cache_cache.array[smp_processor_id()] = ptr; |
e498be7da
|
1462 |
|
83b519e8b
|
1463 |
ptr = kmalloc(sizeof(struct arraycache_init), GFP_NOWAIT); |
e498be7da
|
1464 |
|
9a2dba4b4
|
1465 |
BUG_ON(cpu_cache_get(malloc_sizes[INDEX_AC].cs_cachep) |
b28a02de8
|
1466 |
!= &initarray_generic.cache); |
9a2dba4b4
|
1467 |
memcpy(ptr, cpu_cache_get(malloc_sizes[INDEX_AC].cs_cachep), |
b28a02de8
|
1468 |
sizeof(struct arraycache_init)); |
2b2d5493e
|
1469 1470 1471 1472 |
/* * Do not assume that spinlocks can be initialized via memcpy: */ spin_lock_init(&ptr->lock); |
e498be7da
|
1473 |
malloc_sizes[INDEX_AC].cs_cachep->array[smp_processor_id()] = |
b28a02de8
|
1474 |
ptr; |
1da177e4c
|
1475 |
} |
e498be7da
|
1476 1477 |
/* 5) Replace the bootstrap kmem_list3's */ { |
1ca4cb241
|
1478 |
int nid; |
9c09a95cf
|
1479 |
for_each_online_node(nid) { |
ec1f5eeeb
|
1480 |
init_list(&cache_cache, &initkmem_list3[CACHE_CACHE + nid], nid); |
556a169da
|
1481 |
|
e498be7da
|
1482 |
init_list(malloc_sizes[INDEX_AC].cs_cachep, |
1ca4cb241
|
1483 |
&initkmem_list3[SIZE_AC + nid], nid); |
e498be7da
|
1484 1485 1486 |
if (INDEX_AC != INDEX_L3) { init_list(malloc_sizes[INDEX_L3].cs_cachep, |
1ca4cb241
|
1487 |
&initkmem_list3[SIZE_L3 + nid], nid); |
e498be7da
|
1488 1489 1490 |
} } } |
1da177e4c
|
1491 |
|
8429db5c6
|
1492 |
g_cpucache_up = EARLY; |
8429db5c6
|
1493 1494 1495 1496 1497 |
} void __init kmem_cache_init_late(void) { struct kmem_cache *cachep; |
8429db5c6
|
1498 1499 1500 1501 1502 1503 |
/* 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
|
1504 |
|
1da177e4c
|
1505 1506 |
/* Done! */ g_cpucache_up = FULL; |
ec5a36f94
|
1507 1508 |
/* Annotate slab for lockdep -- annotate the malloc caches */ init_lock_keys(); |
a737b3e2f
|
1509 1510 1511 |
/* * Register a cpu startup notifier callback that initializes * cpu_cache_get for all new cpus |
1da177e4c
|
1512 1513 |
*/ register_cpu_notifier(&cpucache_notifier); |
1da177e4c
|
1514 |
|
a737b3e2f
|
1515 1516 1517 |
/* * The reap timers are started later, with a module init call: That part * of the kernel is not yet operational. |
1da177e4c
|
1518 1519 1520 1521 1522 1523 |
*/ } static int __init cpucache_init(void) { int cpu; |
a737b3e2f
|
1524 1525 |
/* * Register the timers that return unneeded pages to the page allocator |
1da177e4c
|
1526 |
*/ |
e498be7da
|
1527 |
for_each_online_cpu(cpu) |
a737b3e2f
|
1528 |
start_cpu_timer(cpu); |
1da177e4c
|
1529 1530 |
return 0; } |
1da177e4c
|
1531 1532 1533 1534 1535 1536 1537 1538 1539 |
__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
|
1540 |
static void *kmem_getpages(struct kmem_cache *cachep, gfp_t flags, int nodeid) |
1da177e4c
|
1541 1542 |
{ struct page *page; |
e1b6aa6f1
|
1543 |
int nr_pages; |
1da177e4c
|
1544 |
int i; |
d6fef9da1
|
1545 |
#ifndef CONFIG_MMU |
e1b6aa6f1
|
1546 1547 1548 |
/* * Nommu uses slab's for process anonymous memory allocations, and thus * requires __GFP_COMP to properly refcount higher order allocations |
d6fef9da1
|
1549 |
*/ |
e1b6aa6f1
|
1550 |
flags |= __GFP_COMP; |
d6fef9da1
|
1551 |
#endif |
765c4507a
|
1552 |
|
3c517a613
|
1553 |
flags |= cachep->gfpflags; |
e12ba74d8
|
1554 1555 |
if (cachep->flags & SLAB_RECLAIM_ACCOUNT) flags |= __GFP_RECLAIMABLE; |
e1b6aa6f1
|
1556 |
|
517d08699
|
1557 |
page = alloc_pages_exact_node(nodeid, flags | __GFP_NOTRACK, cachep->gfporder); |
1da177e4c
|
1558 1559 |
if (!page) return NULL; |
1da177e4c
|
1560 |
|
e1b6aa6f1
|
1561 |
nr_pages = (1 << cachep->gfporder); |
1da177e4c
|
1562 |
if (cachep->flags & SLAB_RECLAIM_ACCOUNT) |
972d1a7b1
|
1563 1564 1565 1566 1567 |
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
|
1568 1569 |
for (i = 0; i < nr_pages; i++) __SetPageSlab(page + i); |
c175eea46
|
1570 |
|
b1eeab676
|
1571 1572 1573 1574 1575 1576 1577 1578 |
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
|
1579 |
|
e1b6aa6f1
|
1580 |
return page_address(page); |
1da177e4c
|
1581 1582 1583 1584 1585 |
} /* * Interface to system's page release. */ |
343e0d7a9
|
1586 |
static void kmem_freepages(struct kmem_cache *cachep, void *addr) |
1da177e4c
|
1587 |
{ |
b28a02de8
|
1588 |
unsigned long i = (1 << cachep->gfporder); |
1da177e4c
|
1589 1590 |
struct page *page = virt_to_page(addr); const unsigned long nr_freed = i; |
b1eeab676
|
1591 |
kmemcheck_free_shadow(page, cachep->gfporder); |
c175eea46
|
1592 |
|
972d1a7b1
|
1593 1594 1595 1596 1597 1598 |
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
|
1599 |
while (i--) { |
f205b2fe6
|
1600 1601 |
BUG_ON(!PageSlab(page)); __ClearPageSlab(page); |
1da177e4c
|
1602 1603 |
page++; } |
1da177e4c
|
1604 1605 1606 |
if (current->reclaim_state) current->reclaim_state->reclaimed_slab += nr_freed; free_pages((unsigned long)addr, cachep->gfporder); |
1da177e4c
|
1607 1608 1609 1610 |
} static void kmem_rcu_free(struct rcu_head *head) { |
b28a02de8
|
1611 |
struct slab_rcu *slab_rcu = (struct slab_rcu *)head; |
343e0d7a9
|
1612 |
struct kmem_cache *cachep = slab_rcu->cachep; |
1da177e4c
|
1613 1614 1615 1616 1617 1618 1619 1620 1621 |
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
|
1622 |
static void store_stackinfo(struct kmem_cache *cachep, unsigned long *addr, |
b28a02de8
|
1623 |
unsigned long caller) |
1da177e4c
|
1624 |
{ |
3dafccf22
|
1625 |
int size = obj_size(cachep); |
1da177e4c
|
1626 |
|
3dafccf22
|
1627 |
addr = (unsigned long *)&((char *)addr)[obj_offset(cachep)]; |
1da177e4c
|
1628 |
|
b28a02de8
|
1629 |
if (size < 5 * sizeof(unsigned long)) |
1da177e4c
|
1630 |
return; |
b28a02de8
|
1631 1632 1633 1634 |
*addr++ = 0x12345678; *addr++ = caller; *addr++ = smp_processor_id(); size -= 3 * sizeof(unsigned long); |
1da177e4c
|
1635 1636 1637 1638 1639 1640 1641 |
{ unsigned long *sptr = &caller; unsigned long svalue; while (!kstack_end(sptr)) { svalue = *sptr++; if (kernel_text_address(svalue)) { |
b28a02de8
|
1642 |
*addr++ = svalue; |
1da177e4c
|
1643 1644 1645 1646 1647 1648 1649 |
size -= sizeof(unsigned long); if (size <= sizeof(unsigned long)) break; } } } |
b28a02de8
|
1650 |
*addr++ = 0x87654321; |
1da177e4c
|
1651 1652 |
} #endif |
343e0d7a9
|
1653 |
static void poison_obj(struct kmem_cache *cachep, void *addr, unsigned char val) |
1da177e4c
|
1654 |
{ |
3dafccf22
|
1655 1656 |
int size = obj_size(cachep); addr = &((char *)addr)[obj_offset(cachep)]; |
1da177e4c
|
1657 1658 |
memset(addr, val, size); |
b28a02de8
|
1659 |
*(unsigned char *)(addr + size - 1) = POISON_END; |
1da177e4c
|
1660 1661 1662 1663 1664 |
} static void dump_line(char *data, int offset, int limit) { int i; |
aa83aa40e
|
1665 1666 |
unsigned char error = 0; int bad_count = 0; |
1da177e4c
|
1667 |
printk(KERN_ERR "%03x:", offset); |
aa83aa40e
|
1668 1669 1670 1671 1672 |
for (i = 0; i < limit; i++) { if (data[offset + i] != POISON_FREE) { error = data[offset + i]; bad_count++; } |
b28a02de8
|
1673 |
printk(" %02x", (unsigned char)data[offset + i]); |
aa83aa40e
|
1674 |
} |
1da177e4c
|
1675 1676 |
printk(" "); |
aa83aa40e
|
1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 |
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
|
1694 1695 1696 1697 |
} #endif #if DEBUG |
343e0d7a9
|
1698 |
static void print_objinfo(struct kmem_cache *cachep, void *objp, int lines) |
1da177e4c
|
1699 1700 1701 1702 1703 |
{ int i, size; char *realobj; if (cachep->flags & SLAB_RED_ZONE) { |
b46b8f19c
|
1704 1705 |
printk(KERN_ERR "Redzone: 0x%llx/0x%llx. ", |
a737b3e2f
|
1706 1707 |
*dbg_redzone1(cachep, objp), *dbg_redzone2(cachep, objp)); |
1da177e4c
|
1708 1709 1710 1711 |
} if (cachep->flags & SLAB_STORE_USER) { printk(KERN_ERR "Last user: [<%p>]", |
a737b3e2f
|
1712 |
*dbg_userword(cachep, objp)); |
1da177e4c
|
1713 |
print_symbol("(%s)", |
a737b3e2f
|
1714 |
(unsigned long)*dbg_userword(cachep, objp)); |
1da177e4c
|
1715 1716 1717 |
printk(" "); } |
3dafccf22
|
1718 1719 |
realobj = (char *)objp + obj_offset(cachep); size = obj_size(cachep); |
b28a02de8
|
1720 |
for (i = 0; i < size && lines; i += 16, lines--) { |
1da177e4c
|
1721 1722 |
int limit; limit = 16; |
b28a02de8
|
1723 1724 |
if (i + limit > size) limit = size - i; |
1da177e4c
|
1725 1726 1727 |
dump_line(realobj, i, limit); } } |
343e0d7a9
|
1728 |
static void check_poison_obj(struct kmem_cache *cachep, void *objp) |
1da177e4c
|
1729 1730 1731 1732 |
{ char *realobj; int size, i; int lines = 0; |
3dafccf22
|
1733 1734 |
realobj = (char *)objp + obj_offset(cachep); size = obj_size(cachep); |
1da177e4c
|
1735 |
|
b28a02de8
|
1736 |
for (i = 0; i < size; i++) { |
1da177e4c
|
1737 |
char exp = POISON_FREE; |
b28a02de8
|
1738 |
if (i == size - 1) |
1da177e4c
|
1739 1740 1741 1742 1743 1744 |
exp = POISON_END; if (realobj[i] != exp) { int limit; /* Mismatch ! */ /* Print header */ if (lines == 0) { |
b28a02de8
|
1745 |
printk(KERN_ERR |
e94a40c50
|
1746 1747 1748 |
"Slab corruption: %s start=%p, len=%d ", cachep->name, realobj, size); |
1da177e4c
|
1749 1750 1751 |
print_objinfo(cachep, objp, 0); } /* Hexdump the affected line */ |
b28a02de8
|
1752 |
i = (i / 16) * 16; |
1da177e4c
|
1753 |
limit = 16; |
b28a02de8
|
1754 1755 |
if (i + limit > size) limit = size - i; |
1da177e4c
|
1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 |
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
|
1768 |
struct slab *slabp = virt_to_slab(objp); |
8fea4e96a
|
1769 |
unsigned int objnr; |
1da177e4c
|
1770 |
|
8fea4e96a
|
1771 |
objnr = obj_to_index(cachep, slabp, objp); |
1da177e4c
|
1772 |
if (objnr) { |
8fea4e96a
|
1773 |
objp = index_to_obj(cachep, slabp, objnr - 1); |
3dafccf22
|
1774 |
realobj = (char *)objp + obj_offset(cachep); |
1da177e4c
|
1775 1776 |
printk(KERN_ERR "Prev obj: start=%p, len=%d ", |
b28a02de8
|
1777 |
realobj, size); |
1da177e4c
|
1778 1779 |
print_objinfo(cachep, objp, 2); } |
b28a02de8
|
1780 |
if (objnr + 1 < cachep->num) { |
8fea4e96a
|
1781 |
objp = index_to_obj(cachep, slabp, objnr + 1); |
3dafccf22
|
1782 |
realobj = (char *)objp + obj_offset(cachep); |
1da177e4c
|
1783 1784 |
printk(KERN_ERR "Next obj: start=%p, len=%d ", |
b28a02de8
|
1785 |
realobj, size); |
1da177e4c
|
1786 1787 1788 1789 1790 |
print_objinfo(cachep, objp, 2); } } } #endif |
12dd36fae
|
1791 |
#if DEBUG |
e79aec291
|
1792 |
static void slab_destroy_debugcheck(struct kmem_cache *cachep, struct slab *slabp) |
1da177e4c
|
1793 |
{ |
1da177e4c
|
1794 1795 |
int i; for (i = 0; i < cachep->num; i++) { |
8fea4e96a
|
1796 |
void *objp = index_to_obj(cachep, slabp, i); |
1da177e4c
|
1797 1798 1799 |
if (cachep->flags & SLAB_POISON) { #ifdef CONFIG_DEBUG_PAGEALLOC |
a737b3e2f
|
1800 1801 |
if (cachep->buffer_size % PAGE_SIZE == 0 && OFF_SLAB(cachep)) |
b28a02de8
|
1802 |
kernel_map_pages(virt_to_page(objp), |
a737b3e2f
|
1803 |
cachep->buffer_size / PAGE_SIZE, 1); |
1da177e4c
|
1804 1805 1806 1807 1808 1809 1810 1811 1812 |
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
|
1813 |
"was overwritten"); |
1da177e4c
|
1814 1815 |
if (*dbg_redzone2(cachep, objp) != RED_INACTIVE) slab_error(cachep, "end of a freed object " |
b28a02de8
|
1816 |
"was overwritten"); |
1da177e4c
|
1817 |
} |
1da177e4c
|
1818 |
} |
12dd36fae
|
1819 |
} |
1da177e4c
|
1820 |
#else |
e79aec291
|
1821 |
static void slab_destroy_debugcheck(struct kmem_cache *cachep, struct slab *slabp) |
12dd36fae
|
1822 |
{ |
12dd36fae
|
1823 |
} |
1da177e4c
|
1824 |
#endif |
911851e6e
|
1825 1826 1827 1828 1829 |
/** * slab_destroy - destroy and release all objects in a slab * @cachep: cache pointer being destroyed * @slabp: slab pointer being destroyed * |
12dd36fae
|
1830 |
* Destroy all the objs in a slab, and release the mem back to the system. |
a737b3e2f
|
1831 1832 |
* Before calling the slab must have been unlinked from the cache. The * cache-lock is not held/needed. |
12dd36fae
|
1833 |
*/ |
343e0d7a9
|
1834 |
static void slab_destroy(struct kmem_cache *cachep, struct slab *slabp) |
12dd36fae
|
1835 1836 |
{ void *addr = slabp->s_mem - slabp->colouroff; |
e79aec291
|
1837 |
slab_destroy_debugcheck(cachep, slabp); |
1da177e4c
|
1838 1839 |
if (unlikely(cachep->flags & SLAB_DESTROY_BY_RCU)) { struct slab_rcu *slab_rcu; |
b28a02de8
|
1840 |
slab_rcu = (struct slab_rcu *)slabp; |
1da177e4c
|
1841 1842 1843 1844 1845 |
slab_rcu->cachep = cachep; slab_rcu->addr = addr; call_rcu(&slab_rcu->head, kmem_rcu_free); } else { kmem_freepages(cachep, addr); |
873623dfa
|
1846 1847 |
if (OFF_SLAB(cachep)) kmem_cache_free(cachep->slabp_cache, slabp); |
1da177e4c
|
1848 1849 |
} } |
117f6eb1d
|
1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 |
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
|
1869 |
/** |
a70773ddb
|
1870 1871 1872 1873 1874 1875 1876 |
* 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
|
1877 1878 1879 1880 1881 |
* * 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
|
1882 |
static size_t calculate_slab_order(struct kmem_cache *cachep, |
ee13d785e
|
1883 |
size_t size, size_t align, unsigned long flags) |
4d268eba1
|
1884 |
{ |
b1ab41c49
|
1885 |
unsigned long offslab_limit; |
4d268eba1
|
1886 |
size_t left_over = 0; |
9888e6fa7
|
1887 |
int gfporder; |
4d268eba1
|
1888 |
|
0aa817f07
|
1889 |
for (gfporder = 0; gfporder <= KMALLOC_MAX_ORDER; gfporder++) { |
4d268eba1
|
1890 1891 |
unsigned int num; size_t remainder; |
9888e6fa7
|
1892 |
cache_estimate(gfporder, size, align, flags, &remainder, &num); |
4d268eba1
|
1893 1894 |
if (!num) continue; |
9888e6fa7
|
1895 |
|
b1ab41c49
|
1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 |
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
|
1908 |
|
9888e6fa7
|
1909 |
/* Found something acceptable - save it away */ |
4d268eba1
|
1910 |
cachep->num = num; |
9888e6fa7
|
1911 |
cachep->gfporder = gfporder; |
4d268eba1
|
1912 1913 1914 |
left_over = remainder; /* |
f78bb8ad4
|
1915 1916 1917 1918 1919 1920 1921 1922 |
* 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
|
1923 1924 1925 |
* Large number of objects is good, but very large slabs are * currently bad for the gfp()s. */ |
9888e6fa7
|
1926 |
if (gfporder >= slab_break_gfp_order) |
4d268eba1
|
1927 |
break; |
9888e6fa7
|
1928 1929 1930 |
/* * Acceptable internal fragmentation? */ |
a737b3e2f
|
1931 |
if (left_over * 8 <= (PAGE_SIZE << gfporder)) |
4d268eba1
|
1932 1933 1934 1935 |
break; } return left_over; } |
83b519e8b
|
1936 |
static int __init_refok setup_cpu_cache(struct kmem_cache *cachep, gfp_t gfp) |
f30cf7d13
|
1937 |
{ |
2ed3a4ef9
|
1938 |
if (g_cpucache_up == FULL) |
83b519e8b
|
1939 |
return enable_cpucache(cachep, gfp); |
2ed3a4ef9
|
1940 |
|
f30cf7d13
|
1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 |
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
|
1961 |
kmalloc(sizeof(struct arraycache_init), gfp); |
f30cf7d13
|
1962 1963 1964 1965 1966 1967 |
if (g_cpucache_up == PARTIAL_AC) { set_up_list3s(cachep, SIZE_L3); g_cpucache_up = PARTIAL_L3; } else { int node; |
556a169da
|
1968 |
for_each_online_node(node) { |
f30cf7d13
|
1969 1970 |
cachep->nodelists[node] = kmalloc_node(sizeof(struct kmem_list3), |
eb91f1d0a
|
1971 |
gfp, node); |
f30cf7d13
|
1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 |
BUG_ON(!cachep->nodelists[node]); kmem_list3_init(cachep->nodelists[node]); } } } cachep->nodelists[numa_node_id()]->next_reap = jiffies + REAPTIMEOUT_LIST3 + ((unsigned long)cachep) % REAPTIMEOUT_LIST3; cpu_cache_get(cachep)->avail = 0; cpu_cache_get(cachep)->limit = BOOT_CPUCACHE_ENTRIES; cpu_cache_get(cachep)->batchcount = 1; cpu_cache_get(cachep)->touched = 0; cachep->batchcount = 1; cachep->limit = BOOT_CPUCACHE_ENTRIES; |
2ed3a4ef9
|
1987 |
return 0; |
f30cf7d13
|
1988 |
} |
4d268eba1
|
1989 |
/** |
1da177e4c
|
1990 1991 1992 1993 1994 1995 |
* 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
|
1996 1997 1998 |
* * Returns a ptr to the cache on success, NULL on failure. * Cannot be called within a int, but can be interrupted. |
20c2df83d
|
1999 |
* The @ctor is run when new pages are allocated by the cache. |
1da177e4c
|
2000 2001 |
* * @name must be valid until the cache is destroyed. This implies that |
a737b3e2f
|
2002 |
* the module calling this has to destroy the cache before getting unloaded. |
249da1665
|
2003 2004 |
* Note that kmem_cache_name() is not guaranteed to return the same pointer, * therefore applications must manage it themselves. |
a737b3e2f
|
2005 |
* |
1da177e4c
|
2006 2007 2008 2009 2010 2011 2012 2013 |
* 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
|
2014 2015 2016 2017 |
* %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
|
2018 |
struct kmem_cache * |
1da177e4c
|
2019 |
kmem_cache_create (const char *name, size_t size, size_t align, |
51cc50685
|
2020 |
unsigned long flags, void (*ctor)(void *)) |
1da177e4c
|
2021 2022 |
{ size_t left_over, slab_size, ralign; |
7a7c381d2
|
2023 |
struct kmem_cache *cachep = NULL, *pc; |
83b519e8b
|
2024 |
gfp_t gfp; |
1da177e4c
|
2025 2026 2027 2028 |
/* * Sanity checks... these are all serious usage bugs. */ |
a737b3e2f
|
2029 |
if (!name || in_interrupt() || (size < BYTES_PER_WORD) || |
20c2df83d
|
2030 |
size > KMALLOC_MAX_SIZE) { |
d40cee245
|
2031 2032 |
printk(KERN_ERR "%s: Early error in slab %s ", __func__, |
a737b3e2f
|
2033 |
name); |
b28a02de8
|
2034 2035 |
BUG(); } |
1da177e4c
|
2036 |
|
f0188f474
|
2037 |
/* |
8f5be20bf
|
2038 |
* We use cache_chain_mutex to ensure a consistent view of |
174596a0b
|
2039 |
* cpu_online_mask as well. Please see cpuup_callback |
f0188f474
|
2040 |
*/ |
83b519e8b
|
2041 2042 2043 2044 |
if (slab_is_available()) { get_online_cpus(); mutex_lock(&cache_chain_mutex); } |
4f12bb4f7
|
2045 |
|
7a7c381d2
|
2046 |
list_for_each_entry(pc, &cache_chain, next) { |
4f12bb4f7
|
2047 2048 2049 2050 2051 2052 2053 2054 |
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
|
2055 |
res = probe_kernel_address(pc->name, tmp); |
4f12bb4f7
|
2056 |
if (res) { |
b4169525b
|
2057 2058 2059 |
printk(KERN_ERR "SLAB: cache with size %d has lost its name ", |
3dafccf22
|
2060 |
pc->buffer_size); |
4f12bb4f7
|
2061 2062 |
continue; } |
b28a02de8
|
2063 |
if (!strcmp(pc->name, name)) { |
b4169525b
|
2064 2065 2066 |
printk(KERN_ERR "kmem_cache_create: duplicate cache %s ", name); |
4f12bb4f7
|
2067 2068 2069 2070 |
dump_stack(); goto oops; } } |
1da177e4c
|
2071 2072 |
#if DEBUG WARN_ON(strchr(name, ' ')); /* It confuses parsers */ |
1da177e4c
|
2073 2074 2075 2076 2077 2078 2079 |
#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
|
2080 2081 |
if (size < 4096 || fls(size - 1) == fls(size-1 + REDZONE_ALIGN + 2 * sizeof(unsigned long long))) |
b28a02de8
|
2082 |
flags |= SLAB_RED_ZONE | SLAB_STORE_USER; |
1da177e4c
|
2083 2084 2085 2086 2087 2088 |
if (!(flags & SLAB_DESTROY_BY_RCU)) flags |= SLAB_POISON; #endif if (flags & SLAB_DESTROY_BY_RCU) BUG_ON(flags & SLAB_POISON); #endif |
1da177e4c
|
2089 |
/* |
a737b3e2f
|
2090 2091 |
* Always checks flags, a caller might be expecting debug support which * isn't available. |
1da177e4c
|
2092 |
*/ |
40094fa65
|
2093 |
BUG_ON(flags & ~CREATE_MASK); |
1da177e4c
|
2094 |
|
a737b3e2f
|
2095 2096 |
/* * Check that size is in terms of words. This is needed to avoid |
1da177e4c
|
2097 2098 2099 |
* unaligned accesses for some archs when redzoning is used, and makes * sure any on-slab bufctl's are also correctly aligned. */ |
b28a02de8
|
2100 2101 2102 |
if (size & (BYTES_PER_WORD - 1)) { size += (BYTES_PER_WORD - 1); size &= ~(BYTES_PER_WORD - 1); |
1da177e4c
|
2103 |
} |
a737b3e2f
|
2104 |
/* calculate the final buffer alignment: */ |
1da177e4c
|
2105 2106 |
/* 1) arch recommendation: can be overridden for debug */ if (flags & SLAB_HWCACHE_ALIGN) { |
a737b3e2f
|
2107 2108 2109 2110 |
/* * Default alignment: as specified by the arch code. Except if * an object is really small, then squeeze multiple objects into * one cacheline. |
1da177e4c
|
2111 2112 |
*/ ralign = cache_line_size(); |
b28a02de8
|
2113 |
while (size <= ralign / 2) |
1da177e4c
|
2114 2115 2116 2117 |
ralign /= 2; } else { ralign = BYTES_PER_WORD; } |
ca5f9703d
|
2118 2119 |
/* |
87a927c71
|
2120 2121 2122 |
* 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
|
2123 |
*/ |
87a927c71
|
2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 |
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
|
2134 |
|
a44b56d35
|
2135 |
/* 2) arch mandated alignment */ |
1da177e4c
|
2136 2137 |
if (ralign < ARCH_SLAB_MINALIGN) { ralign = ARCH_SLAB_MINALIGN; |
1da177e4c
|
2138 |
} |
a44b56d35
|
2139 |
/* 3) caller mandated alignment */ |
1da177e4c
|
2140 2141 |
if (ralign < align) { ralign = align; |
1da177e4c
|
2142 |
} |
a44b56d35
|
2143 |
/* disable debug if necessary */ |
b46b8f19c
|
2144 |
if (ralign > __alignof__(unsigned long long)) |
a44b56d35
|
2145 |
flags &= ~(SLAB_RED_ZONE | SLAB_STORE_USER); |
a737b3e2f
|
2146 |
/* |
ca5f9703d
|
2147 |
* 4) Store it. |
1da177e4c
|
2148 2149 |
*/ align = ralign; |
83b519e8b
|
2150 2151 2152 2153 |
if (slab_is_available()) gfp = GFP_KERNEL; else gfp = GFP_NOWAIT; |
1da177e4c
|
2154 |
/* Get cache's description obj. */ |
83b519e8b
|
2155 |
cachep = kmem_cache_zalloc(&cache_cache, gfp); |
1da177e4c
|
2156 |
if (!cachep) |
4f12bb4f7
|
2157 |
goto oops; |
1da177e4c
|
2158 2159 |
#if DEBUG |
3dafccf22
|
2160 |
cachep->obj_size = size; |
1da177e4c
|
2161 |
|
ca5f9703d
|
2162 2163 2164 2165 |
/* * Both debugging options require word-alignment which is calculated * into align above. */ |
1da177e4c
|
2166 |
if (flags & SLAB_RED_ZONE) { |
1da177e4c
|
2167 |
/* add space for red zone words */ |
b46b8f19c
|
2168 2169 |
cachep->obj_offset += sizeof(unsigned long long); size += 2 * sizeof(unsigned long long); |
1da177e4c
|
2170 2171 |
} if (flags & SLAB_STORE_USER) { |
ca5f9703d
|
2172 |
/* user store requires one word storage behind the end of |
87a927c71
|
2173 2174 |
* the real object. But if the second red zone needs to be * aligned to 64 bits, we must allow that much space. |
1da177e4c
|
2175 |
*/ |
87a927c71
|
2176 2177 2178 2179 |
if (flags & SLAB_RED_ZONE) size += REDZONE_ALIGN; else size += BYTES_PER_WORD; |
1da177e4c
|
2180 2181 |
} #if FORCED_DEBUG && defined(CONFIG_DEBUG_PAGEALLOC) |
b28a02de8
|
2182 |
if (size >= malloc_sizes[INDEX_L3 + 1].cs_size |
3dafccf22
|
2183 2184 |
&& cachep->obj_size > cache_line_size() && size < PAGE_SIZE) { cachep->obj_offset += PAGE_SIZE - size; |
1da177e4c
|
2185 2186 2187 2188 |
size = PAGE_SIZE; } #endif #endif |
e0a427267
|
2189 2190 2191 |
/* * Determine if the slab management is 'on' or 'off' slab. * (bootstrapping cannot cope with offslab caches so don't do |
e7cb55b94
|
2192 2193 |
* it too early on. Always use on-slab management when * SLAB_NOLEAKTRACE to avoid recursive calls into kmemleak) |
e0a427267
|
2194 |
*/ |
e7cb55b94
|
2195 2196 |
if ((size >= (PAGE_SIZE >> 3)) && !slab_early_init && !(flags & SLAB_NOLEAKTRACE)) |
1da177e4c
|
2197 2198 2199 2200 2201 2202 2203 |
/* * 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
|
2204 |
left_over = calculate_slab_order(cachep, size, align, flags); |
1da177e4c
|
2205 2206 |
if (!cachep->num) { |
b4169525b
|
2207 2208 2209 |
printk(KERN_ERR "kmem_cache_create: couldn't create cache %s. ", name); |
1da177e4c
|
2210 2211 |
kmem_cache_free(&cache_cache, cachep); cachep = NULL; |
4f12bb4f7
|
2212 |
goto oops; |
1da177e4c
|
2213 |
} |
b28a02de8
|
2214 2215 |
slab_size = ALIGN(cachep->num * sizeof(kmem_bufctl_t) + sizeof(struct slab), align); |
1da177e4c
|
2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 |
/* * 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
|
2228 2229 |
slab_size = cachep->num * sizeof(kmem_bufctl_t) + sizeof(struct slab); |
674613652
|
2230 2231 2232 2233 2234 2235 2236 2237 2238 |
#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
|
2239 2240 2241 2242 2243 2244 |
} cachep->colour_off = cache_line_size(); /* Offset must be a multiple of the alignment. */ if (cachep->colour_off < align) cachep->colour_off = align; |
b28a02de8
|
2245 |
cachep->colour = left_over / cachep->colour_off; |
1da177e4c
|
2246 2247 2248 |
cachep->slab_size = slab_size; cachep->flags = flags; cachep->gfpflags = 0; |
4b51d6698
|
2249 |
if (CONFIG_ZONE_DMA_FLAG && (flags & SLAB_CACHE_DMA)) |
1da177e4c
|
2250 |
cachep->gfpflags |= GFP_DMA; |
3dafccf22
|
2251 |
cachep->buffer_size = size; |
6a2d7a955
|
2252 |
cachep->reciprocal_buffer_size = reciprocal_value(size); |
1da177e4c
|
2253 |
|
e5ac9c5ae
|
2254 |
if (flags & CFLGS_OFF_SLAB) { |
b2d550736
|
2255 |
cachep->slabp_cache = kmem_find_general_cachep(slab_size, 0u); |
e5ac9c5ae
|
2256 2257 2258 2259 2260 2261 2262 |
/* * 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
|
2263 |
BUG_ON(ZERO_OR_NULL_PTR(cachep->slabp_cache)); |
e5ac9c5ae
|
2264 |
} |
1da177e4c
|
2265 |
cachep->ctor = ctor; |
1da177e4c
|
2266 |
cachep->name = name; |
83b519e8b
|
2267 |
if (setup_cpu_cache(cachep, gfp)) { |
2ed3a4ef9
|
2268 2269 2270 2271 |
__kmem_cache_destroy(cachep); cachep = NULL; goto oops; } |
1da177e4c
|
2272 |
|
1da177e4c
|
2273 2274 |
/* cache setup completed, link it into the list */ list_add(&cachep->next, &cache_chain); |
a737b3e2f
|
2275 |
oops: |
1da177e4c
|
2276 2277 2278 |
if (!cachep && (flags & SLAB_PANIC)) panic("kmem_cache_create(): failed to create slab `%s' ", |
b28a02de8
|
2279 |
name); |
83b519e8b
|
2280 2281 2282 2283 |
if (slab_is_available()) { mutex_unlock(&cache_chain_mutex); put_online_cpus(); } |
1da177e4c
|
2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 |
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
|
2298 |
static void check_spinlock_acquired(struct kmem_cache *cachep) |
1da177e4c
|
2299 2300 2301 |
{ #ifdef CONFIG_SMP check_irq_off(); |
e498be7da
|
2302 |
assert_spin_locked(&cachep->nodelists[numa_node_id()]->list_lock); |
1da177e4c
|
2303 2304 |
#endif } |
e498be7da
|
2305 |
|
343e0d7a9
|
2306 |
static void check_spinlock_acquired_node(struct kmem_cache *cachep, int node) |
e498be7da
|
2307 2308 2309 2310 2311 2312 |
{ #ifdef CONFIG_SMP check_irq_off(); assert_spin_locked(&cachep->nodelists[node]->list_lock); #endif } |
1da177e4c
|
2313 2314 2315 2316 |
#else #define check_irq_off() do { } while(0) #define check_irq_on() do { } while(0) #define check_spinlock_acquired(x) do { } while(0) |
e498be7da
|
2317 |
#define check_spinlock_acquired_node(x, y) do { } while(0) |
1da177e4c
|
2318 |
#endif |
aab2207cf
|
2319 2320 2321 |
static void drain_array(struct kmem_cache *cachep, struct kmem_list3 *l3, struct array_cache *ac, int force, int node); |
1da177e4c
|
2322 2323 |
static void do_drain(void *arg) { |
a737b3e2f
|
2324 |
struct kmem_cache *cachep = arg; |
1da177e4c
|
2325 |
struct array_cache *ac; |
ff69416e6
|
2326 |
int node = numa_node_id(); |
1da177e4c
|
2327 2328 |
check_irq_off(); |
9a2dba4b4
|
2329 |
ac = cpu_cache_get(cachep); |
ff69416e6
|
2330 2331 2332 |
spin_lock(&cachep->nodelists[node]->list_lock); free_block(cachep, ac->entry, ac->avail, node); spin_unlock(&cachep->nodelists[node]->list_lock); |
1da177e4c
|
2333 2334 |
ac->avail = 0; } |
343e0d7a9
|
2335 |
static void drain_cpu_caches(struct kmem_cache *cachep) |
1da177e4c
|
2336 |
{ |
e498be7da
|
2337 2338 |
struct kmem_list3 *l3; int node; |
15c8b6c1a
|
2339 |
on_each_cpu(do_drain, cachep, 1); |
1da177e4c
|
2340 |
check_irq_on(); |
b28a02de8
|
2341 |
for_each_online_node(node) { |
e498be7da
|
2342 |
l3 = cachep->nodelists[node]; |
a4523a8b3
|
2343 2344 2345 2346 2347 2348 2349 |
if (l3 && l3->alien) drain_alien_cache(cachep, l3->alien); } for_each_online_node(node) { l3 = cachep->nodelists[node]; if (l3) |
aab2207cf
|
2350 |
drain_array(cachep, l3, l3->shared, 1, node); |
e498be7da
|
2351 |
} |
1da177e4c
|
2352 |
} |
ed11d9eb2
|
2353 2354 2355 2356 2357 2358 2359 2360 |
/* * 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
|
2361 |
{ |
ed11d9eb2
|
2362 2363 |
struct list_head *p; int nr_freed; |
1da177e4c
|
2364 |
struct slab *slabp; |
1da177e4c
|
2365 |
|
ed11d9eb2
|
2366 2367 |
nr_freed = 0; while (nr_freed < tofree && !list_empty(&l3->slabs_free)) { |
1da177e4c
|
2368 |
|
ed11d9eb2
|
2369 |
spin_lock_irq(&l3->list_lock); |
e498be7da
|
2370 |
p = l3->slabs_free.prev; |
ed11d9eb2
|
2371 2372 2373 2374 |
if (p == &l3->slabs_free) { spin_unlock_irq(&l3->list_lock); goto out; } |
1da177e4c
|
2375 |
|
ed11d9eb2
|
2376 |
slabp = list_entry(p, struct slab, list); |
1da177e4c
|
2377 |
#if DEBUG |
40094fa65
|
2378 |
BUG_ON(slabp->inuse); |
1da177e4c
|
2379 2380 |
#endif list_del(&slabp->list); |
ed11d9eb2
|
2381 2382 2383 2384 2385 |
/* * Safe to drop the lock. The slab is no longer linked * to the cache. */ l3->free_objects -= cache->num; |
e498be7da
|
2386 |
spin_unlock_irq(&l3->list_lock); |
ed11d9eb2
|
2387 2388 |
slab_destroy(cache, slabp); nr_freed++; |
1da177e4c
|
2389 |
} |
ed11d9eb2
|
2390 2391 |
out: return nr_freed; |
1da177e4c
|
2392 |
} |
8f5be20bf
|
2393 |
/* Called with cache_chain_mutex held to protect against cpu hotplug */ |
343e0d7a9
|
2394 |
static int __cache_shrink(struct kmem_cache *cachep) |
e498be7da
|
2395 2396 2397 2398 2399 2400 2401 2402 2403 |
{ 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
|
2404 2405 2406 2407 2408 2409 2410 |
if (!l3) continue; drain_freelist(cachep, l3, l3->free_objects); ret += !list_empty(&l3->slabs_full) || !list_empty(&l3->slabs_partial); |
e498be7da
|
2411 2412 2413 |
} return (ret ? 1 : 0); } |
1da177e4c
|
2414 2415 2416 2417 2418 2419 2420 |
/** * 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
|
2421 |
int kmem_cache_shrink(struct kmem_cache *cachep) |
1da177e4c
|
2422 |
{ |
8f5be20bf
|
2423 |
int ret; |
40094fa65
|
2424 |
BUG_ON(!cachep || in_interrupt()); |
1da177e4c
|
2425 |
|
95402b382
|
2426 |
get_online_cpus(); |
8f5be20bf
|
2427 2428 2429 |
mutex_lock(&cache_chain_mutex); ret = __cache_shrink(cachep); mutex_unlock(&cache_chain_mutex); |
95402b382
|
2430 |
put_online_cpus(); |
8f5be20bf
|
2431 |
return ret; |
1da177e4c
|
2432 2433 2434 2435 2436 2437 2438 |
} EXPORT_SYMBOL(kmem_cache_shrink); /** * kmem_cache_destroy - delete a cache * @cachep: the cache to destroy * |
72fd4a35a
|
2439 |
* Remove a &struct kmem_cache object from the slab cache. |
1da177e4c
|
2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450 |
* * It is expected this function will be called by a module when it is * unloaded. This will remove the cache completely, and avoid a duplicate * cache being allocated each time a module is loaded and unloaded, if the * module doesn't have persistent in-kernel storage across loads and unloads. * * The cache must be empty before calling this function. * * The caller must guarantee that noone will allocate memory from the cache * during the kmem_cache_destroy(). */ |
133d205a1
|
2451 |
void kmem_cache_destroy(struct kmem_cache *cachep) |
1da177e4c
|
2452 |
{ |
40094fa65
|
2453 |
BUG_ON(!cachep || in_interrupt()); |
1da177e4c
|
2454 |
|
1da177e4c
|
2455 |
/* Find the cache in the chain of caches. */ |
95402b382
|
2456 |
get_online_cpus(); |
fc0abb145
|
2457 |
mutex_lock(&cache_chain_mutex); |
1da177e4c
|
2458 2459 2460 2461 |
/* * the chain is never empty, cache_cache is never destroyed */ list_del(&cachep->next); |
1da177e4c
|
2462 2463 |
if (__cache_shrink(cachep)) { slab_error(cachep, "Can't free all objects"); |
b28a02de8
|
2464 |
list_add(&cachep->next, &cache_chain); |
fc0abb145
|
2465 |
mutex_unlock(&cache_chain_mutex); |
95402b382
|
2466 |
put_online_cpus(); |
133d205a1
|
2467 |
return; |
1da177e4c
|
2468 2469 2470 |
} if (unlikely(cachep->flags & SLAB_DESTROY_BY_RCU)) |
7ed9f7e5d
|
2471 |
rcu_barrier(); |
1da177e4c
|
2472 |
|
117f6eb1d
|
2473 |
__kmem_cache_destroy(cachep); |
8f5be20bf
|
2474 |
mutex_unlock(&cache_chain_mutex); |
95402b382
|
2475 |
put_online_cpus(); |
1da177e4c
|
2476 2477 |
} EXPORT_SYMBOL(kmem_cache_destroy); |
e5ac9c5ae
|
2478 2479 2480 2481 2482 2483 2484 2485 2486 2487 2488 |
/* * 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
|
2489 |
static struct slab *alloc_slabmgmt(struct kmem_cache *cachep, void *objp, |
5b74ada7e
|
2490 2491 |
int colour_off, gfp_t local_flags, int nodeid) |
1da177e4c
|
2492 2493 |
{ struct slab *slabp; |
b28a02de8
|
2494 |
|
1da177e4c
|
2495 2496 |
if (OFF_SLAB(cachep)) { /* Slab management obj is off-slab. */ |
5b74ada7e
|
2497 |
slabp = kmem_cache_alloc_node(cachep->slabp_cache, |
8759ec50a
|
2498 |
local_flags, nodeid); |
d5cff6352
|
2499 2500 2501 2502 2503 2504 |
/* * 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
|
2505 2506 |
kmemleak_scan_area(&slabp->list, sizeof(struct list_head), local_flags); |
1da177e4c
|
2507 2508 2509 |
if (!slabp) return NULL; } else { |
b28a02de8
|
2510 |
slabp = objp + colour_off; |
1da177e4c
|
2511 2512 2513 2514 |
colour_off += cachep->slab_size; } slabp->inuse = 0; slabp->colouroff = colour_off; |
b28a02de8
|
2515 |
slabp->s_mem = objp + colour_off; |
5b74ada7e
|
2516 |
slabp->nodeid = nodeid; |
e51bfd0ad
|
2517 |
slabp->free = 0; |
1da177e4c
|
2518 2519 2520 2521 2522 |
return slabp; } static inline kmem_bufctl_t *slab_bufctl(struct slab *slabp) { |
b28a02de8
|
2523 |
return (kmem_bufctl_t *) (slabp + 1); |
1da177e4c
|
2524 |
} |
343e0d7a9
|
2525 |
static void cache_init_objs(struct kmem_cache *cachep, |
a35afb830
|
2526 |
struct slab *slabp) |
1da177e4c
|
2527 2528 2529 2530 |
{ int i; for (i = 0; i < cachep->num; i++) { |
8fea4e96a
|
2531 |
void *objp = index_to_obj(cachep, slabp, i); |
1da177e4c
|
2532 2533 2534 2535 2536 2537 2538 2539 2540 2541 2542 2543 |
#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
|
2544 2545 2546 |
* 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
|
2547 2548 |
*/ if (cachep->ctor && !(cachep->flags & SLAB_POISON)) |
51cc50685
|
2549 |
cachep->ctor(objp + obj_offset(cachep)); |
1da177e4c
|
2550 2551 2552 2553 |
if (cachep->flags & SLAB_RED_ZONE) { if (*dbg_redzone2(cachep, objp) != RED_INACTIVE) slab_error(cachep, "constructor overwrote the" |
b28a02de8
|
2554 |
" end of an object"); |
1da177e4c
|
2555 2556 |
if (*dbg_redzone1(cachep, objp) != RED_INACTIVE) slab_error(cachep, "constructor overwrote the" |
b28a02de8
|
2557 |
" start of an object"); |
1da177e4c
|
2558 |
} |
a737b3e2f
|
2559 2560 |
if ((cachep->buffer_size % PAGE_SIZE) == 0 && OFF_SLAB(cachep) && cachep->flags & SLAB_POISON) |
b28a02de8
|
2561 |
kernel_map_pages(virt_to_page(objp), |
3dafccf22
|
2562 |
cachep->buffer_size / PAGE_SIZE, 0); |
1da177e4c
|
2563 2564 |
#else if (cachep->ctor) |
51cc50685
|
2565 |
cachep->ctor(objp); |
1da177e4c
|
2566 |
#endif |
b28a02de8
|
2567 |
slab_bufctl(slabp)[i] = i + 1; |
1da177e4c
|
2568 |
} |
b28a02de8
|
2569 |
slab_bufctl(slabp)[i - 1] = BUFCTL_END; |
1da177e4c
|
2570 |
} |
343e0d7a9
|
2571 |
static void kmem_flagcheck(struct kmem_cache *cachep, gfp_t flags) |
1da177e4c
|
2572 |
{ |
4b51d6698
|
2573 2574 2575 2576 2577 2578 |
if (CONFIG_ZONE_DMA_FLAG) { if (flags & GFP_DMA) BUG_ON(!(cachep->gfpflags & GFP_DMA)); else BUG_ON(cachep->gfpflags & GFP_DMA); } |
1da177e4c
|
2579 |
} |
a737b3e2f
|
2580 2581 |
static void *slab_get_obj(struct kmem_cache *cachep, struct slab *slabp, int nodeid) |
78d382d77
|
2582 |
{ |
8fea4e96a
|
2583 |
void *objp = index_to_obj(cachep, slabp, slabp->free); |
78d382d77
|
2584 2585 2586 2587 2588 2589 2590 2591 2592 2593 2594 2595 |
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
|
2596 2597 |
static void slab_put_obj(struct kmem_cache *cachep, struct slab *slabp, void *objp, int nodeid) |
78d382d77
|
2598 |
{ |
8fea4e96a
|
2599 |
unsigned int objnr = obj_to_index(cachep, slabp, objp); |
78d382d77
|
2600 2601 2602 2603 |
#if DEBUG /* Verify that the slab belongs to the intended node */ WARN_ON(slabp->nodeid != nodeid); |
871751e25
|
2604 |
if (slab_bufctl(slabp)[objnr] + 1 <= SLAB_LIMIT + 1) { |
78d382d77
|
2605 |
printk(KERN_ERR "slab: double free detected in cache " |
a737b3e2f
|
2606 2607 |
"'%s', objp %p ", cachep->name, objp); |
78d382d77
|
2608 2609 2610 2611 2612 2613 2614 |
BUG(); } #endif slab_bufctl(slabp)[objnr] = slabp->free; slabp->free = objnr; slabp->inuse--; } |
4776874ff
|
2615 2616 2617 2618 2619 2620 2621 |
/* * Map pages beginning at addr to the given cache and slab. This is required * for the slab allocator to be able to lookup the cache and slab of a * virtual address for kfree, ksize, kmem_ptr_validate, and slab debugging. */ static void slab_map_pages(struct kmem_cache *cache, struct slab *slab, void *addr) |
1da177e4c
|
2622 |
{ |
4776874ff
|
2623 |
int nr_pages; |
1da177e4c
|
2624 |
struct page *page; |
4776874ff
|
2625 |
page = virt_to_page(addr); |
84097518d
|
2626 |
|
4776874ff
|
2627 |
nr_pages = 1; |
84097518d
|
2628 |
if (likely(!PageCompound(page))) |
4776874ff
|
2629 |
nr_pages <<= cache->gfporder; |
1da177e4c
|
2630 |
do { |
4776874ff
|
2631 2632 |
page_set_cache(page, cache); page_set_slab(page, slab); |
1da177e4c
|
2633 |
page++; |
4776874ff
|
2634 |
} while (--nr_pages); |
1da177e4c
|
2635 2636 2637 2638 2639 2640 |
} /* * 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
|
2641 2642 |
static int cache_grow(struct kmem_cache *cachep, gfp_t flags, int nodeid, void *objp) |
1da177e4c
|
2643 |
{ |
b28a02de8
|
2644 |
struct slab *slabp; |
b28a02de8
|
2645 2646 |
size_t offset; gfp_t local_flags; |
e498be7da
|
2647 |
struct kmem_list3 *l3; |
1da177e4c
|
2648 |
|
a737b3e2f
|
2649 2650 2651 |
/* * Be lazy and only check for valid flags here, keeping it out of the * critical path in kmem_cache_alloc(). |
1da177e4c
|
2652 |
*/ |
6cb062296
|
2653 2654 |
BUG_ON(flags & GFP_SLAB_BUG_MASK); local_flags = flags & (GFP_CONSTRAINT_MASK|GFP_RECLAIM_MASK); |
1da177e4c
|
2655 |
|
2e1217cf9
|
2656 |
/* Take the l3 list lock to change the colour_next on this node */ |
1da177e4c
|
2657 |
check_irq_off(); |
2e1217cf9
|
2658 2659 |
l3 = cachep->nodelists[nodeid]; spin_lock(&l3->list_lock); |
1da177e4c
|
2660 2661 |
/* Get colour for the slab, and cal the next value. */ |
2e1217cf9
|
2662 2663 2664 2665 2666 |
offset = l3->colour_next; l3->colour_next++; if (l3->colour_next >= cachep->colour) l3->colour_next = 0; spin_unlock(&l3->list_lock); |
1da177e4c
|
2667 |
|
2e1217cf9
|
2668 |
offset *= cachep->colour_off; |
1da177e4c
|
2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 |
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
|
2680 2681 2682 |
/* * Get mem for the objs. Attempt to allocate a physical page from * 'nodeid'. |
e498be7da
|
2683 |
*/ |
3c517a613
|
2684 |
if (!objp) |
b8c1c5da1
|
2685 |
objp = kmem_getpages(cachep, local_flags, nodeid); |
a737b3e2f
|
2686 |
if (!objp) |
1da177e4c
|
2687 2688 2689 |
goto failed; /* Get slab management. */ |
3c517a613
|
2690 |
slabp = alloc_slabmgmt(cachep, objp, offset, |
6cb062296
|
2691 |
local_flags & ~GFP_CONSTRAINT_MASK, nodeid); |
a737b3e2f
|
2692 |
if (!slabp) |
1da177e4c
|
2693 |
goto opps1; |
4776874ff
|
2694 |
slab_map_pages(cachep, slabp, objp); |
1da177e4c
|
2695 |
|
a35afb830
|
2696 |
cache_init_objs(cachep, slabp); |
1da177e4c
|
2697 2698 2699 2700 |
if (local_flags & __GFP_WAIT) local_irq_disable(); check_irq_off(); |
e498be7da
|
2701 |
spin_lock(&l3->list_lock); |
1da177e4c
|
2702 2703 |
/* Make slab active. */ |
e498be7da
|
2704 |
list_add_tail(&slabp->list, &(l3->slabs_free)); |
1da177e4c
|
2705 |
STATS_INC_GROWN(cachep); |
e498be7da
|
2706 2707 |
l3->free_objects += cachep->num; spin_unlock(&l3->list_lock); |
1da177e4c
|
2708 |
return 1; |
a737b3e2f
|
2709 |
opps1: |
1da177e4c
|
2710 |
kmem_freepages(cachep, objp); |
a737b3e2f
|
2711 |
failed: |
1da177e4c
|
2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 |
if (local_flags & __GFP_WAIT) local_irq_disable(); return 0; } #if DEBUG /* * Perform extra freeing checks: * - detect bad pointers. * - POISON/RED_ZONE checking |
1da177e4c
|
2723 2724 2725 |
*/ static void kfree_debugcheck(const void *objp) { |
1da177e4c
|
2726 2727 2728 |
if (!virt_addr_valid(objp)) { printk(KERN_ERR "kfree_debugcheck: out of range ptr %lxh. ", |
b28a02de8
|
2729 2730 |
(unsigned long)objp); BUG(); |
1da177e4c
|
2731 |
} |
1da177e4c
|
2732 |
} |
58ce1fd58
|
2733 2734 |
static inline void verify_redzone_free(struct kmem_cache *cache, void *obj) { |
b46b8f19c
|
2735 |
unsigned long long redzone1, redzone2; |
58ce1fd58
|
2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 |
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
|
2750 2751 |
printk(KERN_ERR "%p: redzone 1:0x%llx, redzone 2:0x%llx. ", |
58ce1fd58
|
2752 2753 |
obj, redzone1, redzone2); } |
343e0d7a9
|
2754 |
static void *cache_free_debugcheck(struct kmem_cache *cachep, void *objp, |
b28a02de8
|
2755 |
void *caller) |
1da177e4c
|
2756 2757 2758 2759 |
{ struct page *page; unsigned int objnr; struct slab *slabp; |
80cbd911c
|
2760 |
BUG_ON(virt_to_cache(objp) != cachep); |
3dafccf22
|
2761 |
objp -= obj_offset(cachep); |
1da177e4c
|
2762 |
kfree_debugcheck(objp); |
b49af68ff
|
2763 |
page = virt_to_head_page(objp); |
1da177e4c
|
2764 |
|
065d41cb2
|
2765 |
slabp = page_get_slab(page); |
1da177e4c
|
2766 2767 |
if (cachep->flags & SLAB_RED_ZONE) { |
58ce1fd58
|
2768 |
verify_redzone_free(cachep, objp); |
1da177e4c
|
2769 2770 2771 2772 2773 |
*dbg_redzone1(cachep, objp) = RED_INACTIVE; *dbg_redzone2(cachep, objp) = RED_INACTIVE; } if (cachep->flags & SLAB_STORE_USER) *dbg_userword(cachep, objp) = caller; |
8fea4e96a
|
2774 |
objnr = obj_to_index(cachep, slabp, objp); |
1da177e4c
|
2775 2776 |
BUG_ON(objnr >= cachep->num); |
8fea4e96a
|
2777 |
BUG_ON(objp != index_to_obj(cachep, slabp, objnr)); |
1da177e4c
|
2778 |
|
871751e25
|
2779 2780 2781 |
#ifdef CONFIG_DEBUG_SLAB_LEAK slab_bufctl(slabp)[objnr] = BUFCTL_FREE; #endif |
1da177e4c
|
2782 2783 |
if (cachep->flags & SLAB_POISON) { #ifdef CONFIG_DEBUG_PAGEALLOC |
a737b3e2f
|
2784 |
if ((cachep->buffer_size % PAGE_SIZE)==0 && OFF_SLAB(cachep)) { |
1da177e4c
|
2785 |
store_stackinfo(cachep, objp, (unsigned long)caller); |
b28a02de8
|
2786 |
kernel_map_pages(virt_to_page(objp), |
3dafccf22
|
2787 |
cachep->buffer_size / PAGE_SIZE, 0); |
1da177e4c
|
2788 2789 2790 2791 2792 2793 2794 2795 2796 |
} else { poison_obj(cachep, objp, POISON_FREE); } #else poison_obj(cachep, objp, POISON_FREE); #endif } return objp; } |
343e0d7a9
|
2797 |
static void check_slabp(struct kmem_cache *cachep, struct slab *slabp) |
1da177e4c
|
2798 2799 2800 |
{ kmem_bufctl_t i; int entries = 0; |
b28a02de8
|
2801 |
|
1da177e4c
|
2802 2803 2804 2805 2806 2807 2808 |
/* 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
|
2809 2810 2811 2812 2813 |
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
|
2814 |
for (i = 0; |
264132bc6
|
2815 |
i < sizeof(*slabp) + cachep->num * sizeof(kmem_bufctl_t); |
b28a02de8
|
2816 |
i++) { |
a737b3e2f
|
2817 |
if (i % 16 == 0) |
1da177e4c
|
2818 2819 |
printk(" %03x:", i); |
b28a02de8
|
2820 |
printk(" %02x", ((unsigned char *)slabp)[i]); |
1da177e4c
|
2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 |
} 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
|
2832 |
static void *cache_alloc_refill(struct kmem_cache *cachep, gfp_t flags) |
1da177e4c
|
2833 2834 2835 2836 |
{ int batchcount; struct kmem_list3 *l3; struct array_cache *ac; |
1ca4cb241
|
2837 |
int node; |
6d2144d35
|
2838 |
retry: |
1da177e4c
|
2839 |
check_irq_off(); |
6d2144d35
|
2840 |
node = numa_node_id(); |
9a2dba4b4
|
2841 |
ac = cpu_cache_get(cachep); |
1da177e4c
|
2842 2843 |
batchcount = ac->batchcount; if (!ac->touched && batchcount > BATCHREFILL_LIMIT) { |
a737b3e2f
|
2844 2845 2846 2847 |
/* * If there was little recent activity on this cache, then * perform only a partial refill. Otherwise we could generate * refill bouncing. |
1da177e4c
|
2848 2849 2850 |
*/ batchcount = BATCHREFILL_LIMIT; } |
1ca4cb241
|
2851 |
l3 = cachep->nodelists[node]; |
e498be7da
|
2852 2853 2854 |
BUG_ON(ac->avail > 0 || !l3); spin_lock(&l3->list_lock); |
1da177e4c
|
2855 |
|
3ded175a4
|
2856 |
/* See if we can refill from the shared array */ |
44b57f1cc
|
2857 2858 |
if (l3->shared && transfer_objects(ac, l3->shared, batchcount)) { l3->shared->touched = 1; |
3ded175a4
|
2859 |
goto alloc_done; |
44b57f1cc
|
2860 |
} |
3ded175a4
|
2861 |
|
1da177e4c
|
2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872 2873 2874 2875 2876 |
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
|
2877 2878 2879 2880 2881 2882 |
/* * The slab was either on partial or free list so * there must be at least one object available for * allocation. */ |
249b9f331
|
2883 |
BUG_ON(slabp->inuse >= cachep->num); |
714b8171a
|
2884 |
|
1da177e4c
|
2885 |
while (slabp->inuse < cachep->num && batchcount--) { |
1da177e4c
|
2886 2887 2888 |
STATS_INC_ALLOCED(cachep); STATS_INC_ACTIVE(cachep); STATS_SET_HIGH(cachep); |
78d382d77
|
2889 |
ac->entry[ac->avail++] = slab_get_obj(cachep, slabp, |
1ca4cb241
|
2890 |
node); |
1da177e4c
|
2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 |
} 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
|
2901 |
must_grow: |
1da177e4c
|
2902 |
l3->free_objects -= ac->avail; |
a737b3e2f
|
2903 |
alloc_done: |
e498be7da
|
2904 |
spin_unlock(&l3->list_lock); |
1da177e4c
|
2905 2906 2907 |
if (unlikely(!ac->avail)) { int x; |
3c517a613
|
2908 |
x = cache_grow(cachep, flags | GFP_THISNODE, node, NULL); |
e498be7da
|
2909 |
|
a737b3e2f
|
2910 |
/* cache_grow can reenable interrupts, then ac could change. */ |
9a2dba4b4
|
2911 |
ac = cpu_cache_get(cachep); |
a737b3e2f
|
2912 |
if (!x && ac->avail == 0) /* no objects in sight? abort */ |
1da177e4c
|
2913 |
return NULL; |
a737b3e2f
|
2914 |
if (!ac->avail) /* objects refilled by interrupt? */ |
1da177e4c
|
2915 2916 2917 |
goto retry; } ac->touched = 1; |
e498be7da
|
2918 |
return ac->entry[--ac->avail]; |
1da177e4c
|
2919 |
} |
a737b3e2f
|
2920 2921 |
static inline void cache_alloc_debugcheck_before(struct kmem_cache *cachep, gfp_t flags) |
1da177e4c
|
2922 2923 2924 2925 2926 2927 2928 2929 |
{ might_sleep_if(flags & __GFP_WAIT); #if DEBUG kmem_flagcheck(cachep, flags); #endif } #if DEBUG |
a737b3e2f
|
2930 2931 |
static void *cache_alloc_debugcheck_after(struct kmem_cache *cachep, gfp_t flags, void *objp, void *caller) |
1da177e4c
|
2932 |
{ |
b28a02de8
|
2933 |
if (!objp) |
1da177e4c
|
2934 |
return objp; |
b28a02de8
|
2935 |
if (cachep->flags & SLAB_POISON) { |
1da177e4c
|
2936 |
#ifdef CONFIG_DEBUG_PAGEALLOC |
3dafccf22
|
2937 |
if ((cachep->buffer_size % PAGE_SIZE) == 0 && OFF_SLAB(cachep)) |
b28a02de8
|
2938 |
kernel_map_pages(virt_to_page(objp), |
3dafccf22
|
2939 |
cachep->buffer_size / PAGE_SIZE, 1); |
1da177e4c
|
2940 2941 2942 2943 2944 2945 2946 2947 2948 2949 2950 |
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
|
2951 2952 2953 2954 |
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
|
2955 |
printk(KERN_ERR |
b46b8f19c
|
2956 2957 |
"%p: redzone 1:0x%llx, redzone 2:0x%llx ", |
a737b3e2f
|
2958 2959 |
objp, *dbg_redzone1(cachep, objp), *dbg_redzone2(cachep, objp)); |
1da177e4c
|
2960 2961 2962 2963 |
} *dbg_redzone1(cachep, objp) = RED_ACTIVE; *dbg_redzone2(cachep, objp) = RED_ACTIVE; } |
871751e25
|
2964 2965 2966 2967 |
#ifdef CONFIG_DEBUG_SLAB_LEAK { struct slab *slabp; unsigned objnr; |
b49af68ff
|
2968 |
slabp = page_get_slab(virt_to_head_page(objp)); |
871751e25
|
2969 2970 2971 2972 |
objnr = (unsigned)(objp - slabp->s_mem) / cachep->buffer_size; slab_bufctl(slabp)[objnr] = BUFCTL_ACTIVE; } #endif |
3dafccf22
|
2973 |
objp += obj_offset(cachep); |
4f1049345
|
2974 |
if (cachep->ctor && cachep->flags & SLAB_POISON) |
51cc50685
|
2975 |
cachep->ctor(objp); |
a44b56d35
|
2976 2977 2978 2979 2980 2981 2982 |
#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
|
2983 2984 2985 2986 2987 |
return objp; } #else #define cache_alloc_debugcheck_after(a,b,objp,d) (objp) #endif |
773ff60e8
|
2988 |
static bool slab_should_failslab(struct kmem_cache *cachep, gfp_t flags) |
8a8b6502f
|
2989 2990 |
{ if (cachep == &cache_cache) |
773ff60e8
|
2991 |
return false; |
8a8b6502f
|
2992 |
|
4c13dd3b4
|
2993 |
return should_failslab(obj_size(cachep), flags, cachep->flags); |
8a8b6502f
|
2994 |
} |
343e0d7a9
|
2995 |
static inline void *____cache_alloc(struct kmem_cache *cachep, gfp_t flags) |
1da177e4c
|
2996 |
{ |
b28a02de8
|
2997 |
void *objp; |
1da177e4c
|
2998 |
struct array_cache *ac; |
5c3823008
|
2999 |
check_irq_off(); |
8a8b6502f
|
3000 |
|
9a2dba4b4
|
3001 |
ac = cpu_cache_get(cachep); |
1da177e4c
|
3002 3003 3004 |
if (likely(ac->avail)) { STATS_INC_ALLOCHIT(cachep); ac->touched = 1; |
e498be7da
|
3005 |
objp = ac->entry[--ac->avail]; |
1da177e4c
|
3006 3007 3008 |
} else { STATS_INC_ALLOCMISS(cachep); objp = cache_alloc_refill(cachep, flags); |
ddbf2e836
|
3009 3010 3011 3012 3013 |
/* * the 'ac' may be updated by cache_alloc_refill(), * and kmemleak_erase() requires its correct value. */ ac = cpu_cache_get(cachep); |
1da177e4c
|
3014 |
} |
d5cff6352
|
3015 3016 3017 3018 3019 |
/* * 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
|
3020 3021 |
if (objp) kmemleak_erase(&ac->entry[ac->avail]); |
5c3823008
|
3022 3023 |
return objp; } |
e498be7da
|
3024 3025 |
#ifdef CONFIG_NUMA /* |
b2455396b
|
3026 |
* Try allocating on another node if PF_SPREAD_SLAB|PF_MEMPOLICY. |
c61afb181
|
3027 3028 3029 3030 3031 3032 3033 |
* * 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
|
3034 |
if (in_interrupt() || (flags & __GFP_THISNODE)) |
c61afb181
|
3035 3036 3037 3038 3039 3040 3041 |
return NULL; nid_alloc = nid_here = numa_node_id(); if (cpuset_do_slab_mem_spread() && (cachep->flags & SLAB_MEM_SPREAD)) nid_alloc = cpuset_mem_spread_node(); else if (current->mempolicy) nid_alloc = slab_node(current->mempolicy); if (nid_alloc != nid_here) |
8b98c1699
|
3042 |
return ____cache_alloc_node(cachep, flags, nid_alloc); |
c61afb181
|
3043 3044 3045 3046 |
return NULL; } /* |
765c4507a
|
3047 |
* Fallback function if there was no memory available and no objects on a |
3c517a613
|
3048 3049 3050 3051 3052 |
* 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
|
3053 |
*/ |
8c8cc2c10
|
3054 |
static void *fallback_alloc(struct kmem_cache *cache, gfp_t flags) |
765c4507a
|
3055 |
{ |
8c8cc2c10
|
3056 3057 |
struct zonelist *zonelist; gfp_t local_flags; |
dd1a239f6
|
3058 |
struct zoneref *z; |
54a6eb5c4
|
3059 3060 |
struct zone *zone; enum zone_type high_zoneidx = gfp_zone(flags); |
765c4507a
|
3061 |
void *obj = NULL; |
3c517a613
|
3062 |
int nid; |
8c8cc2c10
|
3063 3064 3065 |
if (flags & __GFP_THISNODE) return NULL; |
0e88460da
|
3066 |
zonelist = node_zonelist(slab_node(current->mempolicy), flags); |
6cb062296
|
3067 |
local_flags = flags & (GFP_CONSTRAINT_MASK|GFP_RECLAIM_MASK); |
765c4507a
|
3068 |
|
3c517a613
|
3069 3070 3071 3072 3073 |
retry: /* * Look through allowed nodes for objects available * from existing per node queues. */ |
54a6eb5c4
|
3074 3075 |
for_each_zone_zonelist(zone, z, zonelist, high_zoneidx) { nid = zone_to_nid(zone); |
aedb0eb10
|
3076 |
|
54a6eb5c4
|
3077 |
if (cpuset_zone_allowed_hardwall(zone, flags) && |
3c517a613
|
3078 |
cache->nodelists[nid] && |
481c5346d
|
3079 |
cache->nodelists[nid]->free_objects) { |
3c517a613
|
3080 3081 |
obj = ____cache_alloc_node(cache, flags | GFP_THISNODE, nid); |
481c5346d
|
3082 3083 3084 |
if (obj) break; } |
3c517a613
|
3085 |
} |
cfce66047
|
3086 |
if (!obj) { |
3c517a613
|
3087 3088 3089 3090 3091 3092 |
/* * 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
|
3093 3094 3095 |
if (local_flags & __GFP_WAIT) local_irq_enable(); kmem_flagcheck(cache, flags); |
6484eb3e2
|
3096 |
obj = kmem_getpages(cache, local_flags, numa_node_id()); |
dd47ea755
|
3097 3098 |
if (local_flags & __GFP_WAIT) local_irq_disable(); |
3c517a613
|
3099 3100 3101 3102 3103 3104 3105 3106 3107 3108 3109 3110 3111 3112 3113 3114 |
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
|
3115 |
/* cache_grow already freed obj */ |
3c517a613
|
3116 3117 3118 |
obj = NULL; } } |
aedb0eb10
|
3119 |
} |
765c4507a
|
3120 3121 3122 3123 |
return obj; } /* |
e498be7da
|
3124 |
* A interface to enable slab creation on nodeid |
1da177e4c
|
3125 |
*/ |
8b98c1699
|
3126 |
static void *____cache_alloc_node(struct kmem_cache *cachep, gfp_t flags, |
a737b3e2f
|
3127 |
int nodeid) |
e498be7da
|
3128 3129 |
{ struct list_head *entry; |
b28a02de8
|
3130 3131 3132 |
struct slab *slabp; struct kmem_list3 *l3; void *obj; |
b28a02de8
|
3133 3134 3135 3136 |
int x; l3 = cachep->nodelists[nodeid]; BUG_ON(!l3); |
a737b3e2f
|
3137 |
retry: |
ca3b9b917
|
3138 |
check_irq_off(); |
b28a02de8
|
3139 3140 3141 3142 3143 3144 3145 3146 3147 3148 3149 3150 3151 3152 3153 3154 3155 3156 |
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
|
3157 |
obj = slab_get_obj(cachep, slabp, nodeid); |
b28a02de8
|
3158 3159 3160 3161 |
check_slabp(cachep, slabp); l3->free_objects--; /* move slabp to correct slabp list: */ list_del(&slabp->list); |
a737b3e2f
|
3162 |
if (slabp->free == BUFCTL_END) |
b28a02de8
|
3163 |
list_add(&slabp->list, &l3->slabs_full); |
a737b3e2f
|
3164 |
else |
b28a02de8
|
3165 |
list_add(&slabp->list, &l3->slabs_partial); |
e498be7da
|
3166 |
|
b28a02de8
|
3167 3168 |
spin_unlock(&l3->list_lock); goto done; |
e498be7da
|
3169 |
|
a737b3e2f
|
3170 |
must_grow: |
b28a02de8
|
3171 |
spin_unlock(&l3->list_lock); |
3c517a613
|
3172 |
x = cache_grow(cachep, flags | GFP_THISNODE, nodeid, NULL); |
765c4507a
|
3173 3174 |
if (x) goto retry; |
1da177e4c
|
3175 |
|
8c8cc2c10
|
3176 |
return fallback_alloc(cachep, flags); |
e498be7da
|
3177 |
|
a737b3e2f
|
3178 |
done: |
b28a02de8
|
3179 |
return obj; |
e498be7da
|
3180 |
} |
8c8cc2c10
|
3181 3182 3183 3184 3185 3186 3187 3188 3189 3190 3191 3192 3193 3194 3195 3196 3197 3198 3199 |
/** * 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; |
dcce284a2
|
3200 |
flags &= gfp_allowed_mask; |
7e85ee0c1
|
3201 |
|
cf40bd16f
|
3202 |
lockdep_trace_alloc(flags); |
773ff60e8
|
3203 |
if (slab_should_failslab(cachep, flags)) |
824ebef12
|
3204 |
return NULL; |
8c8cc2c10
|
3205 3206 |
cache_alloc_debugcheck_before(cachep, flags); local_irq_save(save_flags); |
8e15b79cf
|
3207 |
if (nodeid == -1) |
8c8cc2c10
|
3208 3209 3210 3211 3212 3213 3214 3215 3216 3217 3218 3219 3220 3221 3222 3223 3224 3225 3226 3227 3228 3229 3230 3231 |
nodeid = numa_node_id(); if (unlikely(!cachep->nodelists[nodeid])) { /* Node not bootstrapped yet */ ptr = fallback_alloc(cachep, flags); goto out; } if (nodeid == numa_node_id()) { /* * Use the locally cached objects if possible. * However ____cache_alloc does not allow fallback * to other nodes. It may fail while we still have * objects on other nodes available. */ ptr = ____cache_alloc(cachep, flags); if (ptr) goto out; } /* ___cache_alloc_node can fall back to other nodes */ ptr = ____cache_alloc_node(cachep, flags, nodeid); out: local_irq_restore(save_flags); ptr = cache_alloc_debugcheck_after(cachep, flags, ptr, caller); |
d5cff6352
|
3232 3233 |
kmemleak_alloc_recursive(ptr, obj_size(cachep), 1, cachep->flags, flags); |
8c8cc2c10
|
3234 |
|
c175eea46
|
3235 3236 |
if (likely(ptr)) kmemcheck_slab_alloc(cachep, flags, ptr, obj_size(cachep)); |
d07dbea46
|
3237 3238 |
if (unlikely((flags & __GFP_ZERO) && ptr)) memset(ptr, 0, obj_size(cachep)); |
8c8cc2c10
|
3239 3240 3241 3242 3243 3244 3245 3246 3247 3248 3249 3250 3251 3252 3253 3254 3255 3256 3257 3258 3259 3260 3261 3262 3263 3264 3265 3266 3267 3268 3269 3270 3271 3272 3273 3274 3275 3276 3277 3278 |
return ptr; } static __always_inline void * __do_cache_alloc(struct kmem_cache *cache, gfp_t flags) { void *objp; if (unlikely(current->flags & (PF_SPREAD_SLAB | PF_MEMPOLICY))) { objp = alternate_node_alloc(cache, flags); if (objp) goto out; } objp = ____cache_alloc(cache, flags); /* * We may just have run out of memory on the local node. * ____cache_alloc_node() knows how to locate memory on other nodes */ if (!objp) objp = ____cache_alloc_node(cache, flags, numa_node_id()); out: return objp; } #else static __always_inline void * __do_cache_alloc(struct kmem_cache *cachep, gfp_t flags) { return ____cache_alloc(cachep, flags); } #endif /* CONFIG_NUMA */ static __always_inline void * __cache_alloc(struct kmem_cache *cachep, gfp_t flags, void *caller) { unsigned long save_flags; void *objp; |
dcce284a2
|
3279 |
flags &= gfp_allowed_mask; |
7e85ee0c1
|
3280 |
|
cf40bd16f
|
3281 |
lockdep_trace_alloc(flags); |
773ff60e8
|
3282 |
if (slab_should_failslab(cachep, flags)) |
824ebef12
|
3283 |
return NULL; |
8c8cc2c10
|
3284 3285 3286 3287 3288 |
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
|
3289 3290 |
kmemleak_alloc_recursive(objp, obj_size(cachep), 1, cachep->flags, flags); |
8c8cc2c10
|
3291 |
prefetchw(objp); |
c175eea46
|
3292 3293 |
if (likely(objp)) kmemcheck_slab_alloc(cachep, flags, objp, obj_size(cachep)); |
d07dbea46
|
3294 3295 |
if (unlikely((flags & __GFP_ZERO) && objp)) memset(objp, 0, obj_size(cachep)); |
8c8cc2c10
|
3296 3297 |
return objp; } |
e498be7da
|
3298 3299 3300 3301 |
/* * Caller needs to acquire correct kmem_list's list_lock */ |
343e0d7a9
|
3302 |
static void free_block(struct kmem_cache *cachep, void **objpp, int nr_objects, |
b28a02de8
|
3303 |
int node) |
1da177e4c
|
3304 3305 |
{ int i; |
e498be7da
|
3306 |
struct kmem_list3 *l3; |
1da177e4c
|
3307 3308 3309 3310 |
for (i = 0; i < nr_objects; i++) { void *objp = objpp[i]; struct slab *slabp; |
1da177e4c
|
3311 |
|
6ed5eb221
|
3312 |
slabp = virt_to_slab(objp); |
ff69416e6
|
3313 |
l3 = cachep->nodelists[node]; |
1da177e4c
|
3314 |
list_del(&slabp->list); |
ff69416e6
|
3315 |
check_spinlock_acquired_node(cachep, node); |
1da177e4c
|
3316 |
check_slabp(cachep, slabp); |
78d382d77
|
3317 |
slab_put_obj(cachep, slabp, objp, node); |
1da177e4c
|
3318 |
STATS_DEC_ACTIVE(cachep); |
e498be7da
|
3319 |
l3->free_objects++; |
1da177e4c
|
3320 3321 3322 3323 |
check_slabp(cachep, slabp); /* fixup slab chains */ if (slabp->inuse == 0) { |
e498be7da
|
3324 3325 |
if (l3->free_objects > l3->free_limit) { l3->free_objects -= cachep->num; |
e5ac9c5ae
|
3326 3327 3328 3329 3330 3331 |
/* 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
|
3332 3333 |
slab_destroy(cachep, slabp); } else { |
e498be7da
|
3334 |
list_add(&slabp->list, &l3->slabs_free); |
1da177e4c
|
3335 3336 3337 3338 3339 3340 |
} } 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
|
3341 |
list_add_tail(&slabp->list, &l3->slabs_partial); |
1da177e4c
|
3342 3343 3344 |
} } } |
343e0d7a9
|
3345 |
static void cache_flusharray(struct kmem_cache *cachep, struct array_cache *ac) |
1da177e4c
|
3346 3347 |
{ int batchcount; |
e498be7da
|
3348 |
struct kmem_list3 *l3; |
ff69416e6
|
3349 |
int node = numa_node_id(); |
1da177e4c
|
3350 3351 3352 3353 3354 3355 |
batchcount = ac->batchcount; #if DEBUG BUG_ON(!batchcount || batchcount > ac->avail); #endif check_irq_off(); |
ff69416e6
|
3356 |
l3 = cachep->nodelists[node]; |
873623dfa
|
3357 |
spin_lock(&l3->list_lock); |
e498be7da
|
3358 3359 |
if (l3->shared) { struct array_cache *shared_array = l3->shared; |
b28a02de8
|
3360 |
int max = shared_array->limit - shared_array->avail; |
1da177e4c
|
3361 3362 3363 |
if (max) { if (batchcount > max) batchcount = max; |
e498be7da
|
3364 |
memcpy(&(shared_array->entry[shared_array->avail]), |
b28a02de8
|
3365 |
ac->entry, sizeof(void *) * batchcount); |
1da177e4c
|
3366 3367 3368 3369 |
shared_array->avail += batchcount; goto free_done; } } |
ff69416e6
|
3370 |
free_block(cachep, ac->entry, batchcount, node); |
a737b3e2f
|
3371 |
free_done: |
1da177e4c
|
3372 3373 3374 3375 |
#if STATS { int i = 0; struct list_head *p; |
e498be7da
|
3376 3377 |
p = l3->slabs_free.next; while (p != &(l3->slabs_free)) { |
1da177e4c
|
3378 3379 3380 3381 3382 3383 3384 3385 3386 3387 3388 |
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
|
3389 |
spin_unlock(&l3->list_lock); |
1da177e4c
|
3390 |
ac->avail -= batchcount; |
a737b3e2f
|
3391 |
memmove(ac->entry, &(ac->entry[batchcount]), sizeof(void *)*ac->avail); |
1da177e4c
|
3392 3393 3394 |
} /* |
a737b3e2f
|
3395 3396 |
* 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
|
3397 |
*/ |
873623dfa
|
3398 |
static inline void __cache_free(struct kmem_cache *cachep, void *objp) |
1da177e4c
|
3399 |
{ |
9a2dba4b4
|
3400 |
struct array_cache *ac = cpu_cache_get(cachep); |
1da177e4c
|
3401 3402 |
check_irq_off(); |
d5cff6352
|
3403 |
kmemleak_free_recursive(objp, cachep->flags); |
1da177e4c
|
3404 |
objp = cache_free_debugcheck(cachep, objp, __builtin_return_address(0)); |
c175eea46
|
3405 |
kmemcheck_slab_free(cachep, objp, obj_size(cachep)); |
1807a1aaf
|
3406 3407 3408 3409 3410 3411 3412 |
/* * 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
|
3413 |
if (nr_online_nodes > 1 && cache_free_alien(cachep, objp)) |
729bd0b74
|
3414 |
return; |
1da177e4c
|
3415 3416 |
if (likely(ac->avail < ac->limit)) { STATS_INC_FREEHIT(cachep); |
e498be7da
|
3417 |
ac->entry[ac->avail++] = objp; |
1da177e4c
|
3418 3419 3420 3421 |
return; } else { STATS_INC_FREEMISS(cachep); cache_flusharray(cachep, ac); |
e498be7da
|
3422 |
ac->entry[ac->avail++] = objp; |
1da177e4c
|
3423 3424 3425 3426 3427 3428 3429 3430 3431 3432 3433 |
} } /** * 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
|
3434 |
void *kmem_cache_alloc(struct kmem_cache *cachep, gfp_t flags) |
1da177e4c
|
3435 |
{ |
36555751c
|
3436 |
void *ret = __cache_alloc(cachep, flags, __builtin_return_address(0)); |
ca2b84cb3
|
3437 3438 |
trace_kmem_cache_alloc(_RET_IP_, ret, obj_size(cachep), cachep->buffer_size, flags); |
36555751c
|
3439 3440 |
return ret; |
1da177e4c
|
3441 3442 |
} EXPORT_SYMBOL(kmem_cache_alloc); |
0f24f1287
|
3443 |
#ifdef CONFIG_TRACING |
36555751c
|
3444 3445 3446 3447 3448 3449 |
void *kmem_cache_alloc_notrace(struct kmem_cache *cachep, gfp_t flags) { return __cache_alloc(cachep, flags, __builtin_return_address(0)); } EXPORT_SYMBOL(kmem_cache_alloc_notrace); #endif |
1da177e4c
|
3450 |
/** |
7682486b3
|
3451 |
* kmem_ptr_validate - check if an untrusted pointer might be a slab entry. |
1da177e4c
|
3452 3453 3454 |
* @cachep: the cache we're checking against * @ptr: pointer to validate * |
7682486b3
|
3455 |
* This verifies that the untrusted pointer looks sane; |
1da177e4c
|
3456 3457 3458 3459 3460 3461 3462 |
* it is _not_ a guarantee that the pointer is actually * part of the slab cache in question, but it at least * validates that the pointer can be dereferenced and * looks half-way sane. * * Currently only used for dentry validation. */ |
b7f869a28
|
3463 |
int kmem_ptr_validate(struct kmem_cache *cachep, const void *ptr) |
1da177e4c
|
3464 |
{ |
3dafccf22
|
3465 |
unsigned long size = cachep->buffer_size; |
1da177e4c
|
3466 |
struct page *page; |
fc1c18335
|
3467 |
if (unlikely(!kern_ptr_validate(ptr, size))) |
1da177e4c
|
3468 3469 3470 3471 |
goto out; page = virt_to_page(ptr); if (unlikely(!PageSlab(page))) goto out; |
065d41cb2
|
3472 |
if (unlikely(page_get_cache(page) != cachep)) |
1da177e4c
|
3473 3474 |
goto out; return 1; |
a737b3e2f
|
3475 |
out: |
1da177e4c
|
3476 3477 3478 3479 |
return 0; } #ifdef CONFIG_NUMA |
8b98c1699
|
3480 3481 |
void *kmem_cache_alloc_node(struct kmem_cache *cachep, gfp_t flags, int nodeid) { |
36555751c
|
3482 3483 |
void *ret = __cache_alloc_node(cachep, flags, nodeid, __builtin_return_address(0)); |
ca2b84cb3
|
3484 3485 3486 |
trace_kmem_cache_alloc_node(_RET_IP_, ret, obj_size(cachep), cachep->buffer_size, flags, nodeid); |
36555751c
|
3487 3488 |
return ret; |
8b98c1699
|
3489 |
} |
1da177e4c
|
3490 |
EXPORT_SYMBOL(kmem_cache_alloc_node); |
0f24f1287
|
3491 |
#ifdef CONFIG_TRACING |
36555751c
|
3492 3493 3494 3495 3496 3497 3498 3499 3500 |
void *kmem_cache_alloc_node_notrace(struct kmem_cache *cachep, gfp_t flags, int nodeid) { return __cache_alloc_node(cachep, flags, nodeid, __builtin_return_address(0)); } EXPORT_SYMBOL(kmem_cache_alloc_node_notrace); #endif |
8b98c1699
|
3501 3502 |
static __always_inline void * __do_kmalloc_node(size_t size, gfp_t flags, int node, void *caller) |
97e2bde47
|
3503 |
{ |
343e0d7a9
|
3504 |
struct kmem_cache *cachep; |
36555751c
|
3505 |
void *ret; |
97e2bde47
|
3506 3507 |
cachep = kmem_find_general_cachep(size, flags); |
6cb8f9132
|
3508 3509 |
if (unlikely(ZERO_OR_NULL_PTR(cachep))) return cachep; |
36555751c
|
3510 |
ret = kmem_cache_alloc_node_notrace(cachep, flags, node); |
ca2b84cb3
|
3511 3512 |
trace_kmalloc_node((unsigned long) caller, ret, size, cachep->buffer_size, flags, node); |
36555751c
|
3513 3514 |
return ret; |
97e2bde47
|
3515 |
} |
8b98c1699
|
3516 |
|
0bb38a5cd
|
3517 |
#if defined(CONFIG_DEBUG_SLAB) || defined(CONFIG_TRACING) |
8b98c1699
|
3518 3519 3520 3521 3522 |
void *__kmalloc_node(size_t size, gfp_t flags, int node) { return __do_kmalloc_node(size, flags, node, __builtin_return_address(0)); } |
dbe5e69d2
|
3523 |
EXPORT_SYMBOL(__kmalloc_node); |
8b98c1699
|
3524 3525 |
void *__kmalloc_node_track_caller(size_t size, gfp_t flags, |
ce71e27c6
|
3526 |
int node, unsigned long caller) |
8b98c1699
|
3527 |
{ |
ce71e27c6
|
3528 |
return __do_kmalloc_node(size, flags, node, (void *)caller); |
8b98c1699
|
3529 3530 3531 3532 3533 3534 3535 3536 |
} 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
|
3537 |
#endif /* CONFIG_DEBUG_SLAB || CONFIG_TRACING */ |
8b98c1699
|
3538 |
#endif /* CONFIG_NUMA */ |
1da177e4c
|
3539 3540 |
/** |
800590f52
|
3541 |
* __do_kmalloc - allocate memory |
1da177e4c
|
3542 |
* @size: how many bytes of memory are required. |
800590f52
|
3543 |
* @flags: the type of memory to allocate (see kmalloc). |
911851e6e
|
3544 |
* @caller: function caller for debug tracking of the caller |
1da177e4c
|
3545 |
*/ |
7fd6b1413
|
3546 3547 |
static __always_inline void *__do_kmalloc(size_t size, gfp_t flags, void *caller) |
1da177e4c
|
3548 |
{ |
343e0d7a9
|
3549 |
struct kmem_cache *cachep; |
36555751c
|
3550 |
void *ret; |
1da177e4c
|
3551 |
|
97e2bde47
|
3552 3553 3554 3555 3556 3557 |
/* 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
|
3558 3559 |
if (unlikely(ZERO_OR_NULL_PTR(cachep))) return cachep; |
36555751c
|
3560 |
ret = __cache_alloc(cachep, flags, caller); |
ca2b84cb3
|
3561 3562 |
trace_kmalloc((unsigned long) caller, ret, size, cachep->buffer_size, flags); |
36555751c
|
3563 3564 |
return ret; |
7fd6b1413
|
3565 |
} |
7fd6b1413
|
3566 |
|
0bb38a5cd
|
3567 |
#if defined(CONFIG_DEBUG_SLAB) || defined(CONFIG_TRACING) |
7fd6b1413
|
3568 3569 |
void *__kmalloc(size_t size, gfp_t flags) { |
871751e25
|
3570 |
return __do_kmalloc(size, flags, __builtin_return_address(0)); |
1da177e4c
|
3571 3572 |
} EXPORT_SYMBOL(__kmalloc); |
ce71e27c6
|
3573 |
void *__kmalloc_track_caller(size_t size, gfp_t flags, unsigned long caller) |
7fd6b1413
|
3574 |
{ |
ce71e27c6
|
3575 |
return __do_kmalloc(size, flags, (void *)caller); |
7fd6b1413
|
3576 3577 |
} EXPORT_SYMBOL(__kmalloc_track_caller); |
1d2c8eea6
|
3578 3579 3580 3581 3582 3583 3584 |
#else void *__kmalloc(size_t size, gfp_t flags) { return __do_kmalloc(size, flags, NULL); } EXPORT_SYMBOL(__kmalloc); |
7fd6b1413
|
3585 |
#endif |
1da177e4c
|
3586 3587 3588 3589 3590 3591 3592 3593 |
/** * 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
|
3594 |
void kmem_cache_free(struct kmem_cache *cachep, void *objp) |
1da177e4c
|
3595 3596 3597 3598 |
{ unsigned long flags; local_irq_save(flags); |
898552c9d
|
3599 |
debug_check_no_locks_freed(objp, obj_size(cachep)); |
3ac7fe5a4
|
3600 3601 |
if (!(cachep->flags & SLAB_DEBUG_OBJECTS)) debug_check_no_obj_freed(objp, obj_size(cachep)); |
873623dfa
|
3602 |
__cache_free(cachep, objp); |
1da177e4c
|
3603 |
local_irq_restore(flags); |
36555751c
|
3604 |
|
ca2b84cb3
|
3605 |
trace_kmem_cache_free(_RET_IP_, objp); |
1da177e4c
|
3606 3607 3608 3609 |
} EXPORT_SYMBOL(kmem_cache_free); /** |
1da177e4c
|
3610 3611 3612 |
* kfree - free previously allocated memory * @objp: pointer returned by kmalloc. * |
80e93effc
|
3613 3614 |
* If @objp is NULL, no operation is performed. * |
1da177e4c
|
3615 3616 3617 3618 3619 |
* Don't free memory not originally allocated by kmalloc() * or you will run into trouble. */ void kfree(const void *objp) { |
343e0d7a9
|
3620 |
struct kmem_cache *c; |
1da177e4c
|
3621 |
unsigned long flags; |
2121db74b
|
3622 |
trace_kfree(_RET_IP_, objp); |
6cb8f9132
|
3623 |
if (unlikely(ZERO_OR_NULL_PTR(objp))) |
1da177e4c
|
3624 3625 3626 |
return; local_irq_save(flags); kfree_debugcheck(objp); |
6ed5eb221
|
3627 |
c = virt_to_cache(objp); |
f9b8404cf
|
3628 |
debug_check_no_locks_freed(objp, obj_size(c)); |
3ac7fe5a4
|
3629 |
debug_check_no_obj_freed(objp, obj_size(c)); |
873623dfa
|
3630 |
__cache_free(c, (void *)objp); |
1da177e4c
|
3631 3632 3633 |
local_irq_restore(flags); } EXPORT_SYMBOL(kfree); |
343e0d7a9
|
3634 |
unsigned int kmem_cache_size(struct kmem_cache *cachep) |
1da177e4c
|
3635 |
{ |
3dafccf22
|
3636 |
return obj_size(cachep); |
1da177e4c
|
3637 3638 |
} EXPORT_SYMBOL(kmem_cache_size); |
343e0d7a9
|
3639 |
const char *kmem_cache_name(struct kmem_cache *cachep) |
1944972d3
|
3640 3641 3642 3643 |
{ return cachep->name; } EXPORT_SYMBOL_GPL(kmem_cache_name); |
e498be7da
|
3644 |
/* |
183ff22bb
|
3645 |
* This initializes kmem_list3 or resizes various caches for all nodes. |
e498be7da
|
3646 |
*/ |
83b519e8b
|
3647 |
static int alloc_kmemlist(struct kmem_cache *cachep, gfp_t gfp) |
e498be7da
|
3648 3649 3650 |
{ int node; struct kmem_list3 *l3; |
cafeb02e0
|
3651 |
struct array_cache *new_shared; |
3395ee058
|
3652 |
struct array_cache **new_alien = NULL; |
e498be7da
|
3653 |
|
9c09a95cf
|
3654 |
for_each_online_node(node) { |
cafeb02e0
|
3655 |
|
3395ee058
|
3656 |
if (use_alien_caches) { |
83b519e8b
|
3657 |
new_alien = alloc_alien_cache(node, cachep->limit, gfp); |
3395ee058
|
3658 3659 3660 |
if (!new_alien) goto fail; } |
cafeb02e0
|
3661 |
|
631098469
|
3662 3663 3664 |
new_shared = NULL; if (cachep->shared) { new_shared = alloc_arraycache(node, |
0718dc2a8
|
3665 |
cachep->shared*cachep->batchcount, |
83b519e8b
|
3666 |
0xbaadf00d, gfp); |
631098469
|
3667 3668 3669 3670 |
if (!new_shared) { free_alien_cache(new_alien); goto fail; } |
0718dc2a8
|
3671 |
} |
cafeb02e0
|
3672 |
|
a737b3e2f
|
3673 3674 |
l3 = cachep->nodelists[node]; if (l3) { |
cafeb02e0
|
3675 |
struct array_cache *shared = l3->shared; |
e498be7da
|
3676 |
spin_lock_irq(&l3->list_lock); |
cafeb02e0
|
3677 |
if (shared) |
0718dc2a8
|
3678 3679 |
free_block(cachep, shared->entry, shared->avail, node); |
e498be7da
|
3680 |
|
cafeb02e0
|
3681 3682 |
l3->shared = new_shared; if (!l3->alien) { |
e498be7da
|
3683 3684 3685 |
l3->alien = new_alien; new_alien = NULL; } |
b28a02de8
|
3686 |
l3->free_limit = (1 + nr_cpus_node(node)) * |
a737b3e2f
|
3687 |
cachep->batchcount + cachep->num; |
e498be7da
|
3688 |
spin_unlock_irq(&l3->list_lock); |
cafeb02e0
|
3689 |
kfree(shared); |
e498be7da
|
3690 3691 3692 |
free_alien_cache(new_alien); continue; } |
83b519e8b
|
3693 |
l3 = kmalloc_node(sizeof(struct kmem_list3), gfp, node); |
0718dc2a8
|
3694 3695 3696 |
if (!l3) { free_alien_cache(new_alien); kfree(new_shared); |
e498be7da
|
3697 |
goto fail; |
0718dc2a8
|
3698 |
} |
e498be7da
|
3699 3700 3701 |
kmem_list3_init(l3); l3->next_reap = jiffies + REAPTIMEOUT_LIST3 + |
a737b3e2f
|
3702 |
((unsigned long)cachep) % REAPTIMEOUT_LIST3; |
cafeb02e0
|
3703 |
l3->shared = new_shared; |
e498be7da
|
3704 |
l3->alien = new_alien; |
b28a02de8
|
3705 |
l3->free_limit = (1 + nr_cpus_node(node)) * |
a737b3e2f
|
3706 |
cachep->batchcount + cachep->num; |
e498be7da
|
3707 3708 |
cachep->nodelists[node] = l3; } |
cafeb02e0
|
3709 |
return 0; |
0718dc2a8
|
3710 |
|
a737b3e2f
|
3711 |
fail: |
0718dc2a8
|
3712 3713 3714 3715 3716 3717 3718 3719 3720 3721 3722 3723 3724 3725 3726 |
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
|
3727 |
return -ENOMEM; |
e498be7da
|
3728 |
} |
1da177e4c
|
3729 |
struct ccupdate_struct { |
343e0d7a9
|
3730 |
struct kmem_cache *cachep; |
1da177e4c
|
3731 3732 3733 3734 3735 |
struct array_cache *new[NR_CPUS]; }; static void do_ccupdate_local(void *info) { |
a737b3e2f
|
3736 |
struct ccupdate_struct *new = info; |
1da177e4c
|
3737 3738 3739 |
struct array_cache *old; check_irq_off(); |
9a2dba4b4
|
3740 |
old = cpu_cache_get(new->cachep); |
e498be7da
|
3741 |
|
1da177e4c
|
3742 3743 3744 |
new->cachep->array[smp_processor_id()] = new->new[smp_processor_id()]; new->new[smp_processor_id()] = old; } |
b5d8ca7c5
|
3745 |
/* Always called with the cache_chain_mutex held */ |
a737b3e2f
|
3746 |
static int do_tune_cpucache(struct kmem_cache *cachep, int limit, |
83b519e8b
|
3747 |
int batchcount, int shared, gfp_t gfp) |
1da177e4c
|
3748 |
{ |
d2e7b7d0a
|
3749 |
struct ccupdate_struct *new; |
2ed3a4ef9
|
3750 |
int i; |
1da177e4c
|
3751 |
|
83b519e8b
|
3752 |
new = kzalloc(sizeof(*new), gfp); |
d2e7b7d0a
|
3753 3754 |
if (!new) return -ENOMEM; |
e498be7da
|
3755 |
for_each_online_cpu(i) { |
d2e7b7d0a
|
3756 |
new->new[i] = alloc_arraycache(cpu_to_node(i), limit, |
83b519e8b
|
3757 |
batchcount, gfp); |
d2e7b7d0a
|
3758 |
if (!new->new[i]) { |
b28a02de8
|
3759 |
for (i--; i >= 0; i--) |
d2e7b7d0a
|
3760 3761 |
kfree(new->new[i]); kfree(new); |
e498be7da
|
3762 |
return -ENOMEM; |
1da177e4c
|
3763 3764 |
} } |
d2e7b7d0a
|
3765 |
new->cachep = cachep; |
1da177e4c
|
3766 |
|
15c8b6c1a
|
3767 |
on_each_cpu(do_ccupdate_local, (void *)new, 1); |
e498be7da
|
3768 |
|
1da177e4c
|
3769 |
check_irq_on(); |
1da177e4c
|
3770 3771 |
cachep->batchcount = batchcount; cachep->limit = limit; |
e498be7da
|
3772 |
cachep->shared = shared; |
1da177e4c
|
3773 |
|
e498be7da
|
3774 |
for_each_online_cpu(i) { |
d2e7b7d0a
|
3775 |
struct array_cache *ccold = new->new[i]; |
1da177e4c
|
3776 3777 |
if (!ccold) continue; |
e498be7da
|
3778 |
spin_lock_irq(&cachep->nodelists[cpu_to_node(i)]->list_lock); |
ff69416e6
|
3779 |
free_block(cachep, ccold->entry, ccold->avail, cpu_to_node(i)); |
e498be7da
|
3780 |
spin_unlock_irq(&cachep->nodelists[cpu_to_node(i)]->list_lock); |
1da177e4c
|
3781 3782 |
kfree(ccold); } |
d2e7b7d0a
|
3783 |
kfree(new); |
83b519e8b
|
3784 |
return alloc_kmemlist(cachep, gfp); |
1da177e4c
|
3785 |
} |
b5d8ca7c5
|
3786 |
/* Called with cache_chain_mutex held always */ |
83b519e8b
|
3787 |
static int enable_cpucache(struct kmem_cache *cachep, gfp_t gfp) |
1da177e4c
|
3788 3789 3790 |
{ int err; int limit, shared; |
a737b3e2f
|
3791 3792 |
/* * The head array serves three purposes: |
1da177e4c
|
3793 3794 |
* - create a LIFO ordering, i.e. return objects that are cache-warm * - reduce the number of spinlock operations. |
a737b3e2f
|
3795 |
* - reduce the number of linked list operations on the slab and |
1da177e4c
|
3796 3797 3798 3799 |
* bufctl chains: array operations are cheaper. * The numbers are guessed, we should auto-tune as described by * Bonwick. */ |
3dafccf22
|
3800 |
if (cachep->buffer_size > 131072) |
1da177e4c
|
3801 |
limit = 1; |
3dafccf22
|
3802 |
else if (cachep->buffer_size > PAGE_SIZE) |
1da177e4c
|
3803 |
limit = 8; |
3dafccf22
|
3804 |
else if (cachep->buffer_size > 1024) |
1da177e4c
|
3805 |
limit = 24; |
3dafccf22
|
3806 |
else if (cachep->buffer_size > 256) |
1da177e4c
|
3807 3808 3809 |
limit = 54; else limit = 120; |
a737b3e2f
|
3810 3811 |
/* * CPU bound tasks (e.g. network routing) can exhibit cpu bound |
1da177e4c
|
3812 3813 3814 3815 3816 3817 3818 3819 |
* 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
|
3820 |
if (cachep->buffer_size <= PAGE_SIZE && num_possible_cpus() > 1) |
1da177e4c
|
3821 |
shared = 8; |
1da177e4c
|
3822 3823 |
#if DEBUG |
a737b3e2f
|
3824 3825 3826 |
/* * With debugging enabled, large batchcount lead to excessively long * periods with disabled local interrupts. Limit the batchcount |
1da177e4c
|
3827 3828 3829 3830 |
*/ if (limit > 32) limit = 32; #endif |
83b519e8b
|
3831 |
err = do_tune_cpucache(cachep, limit, (limit + 1) / 2, shared, gfp); |
1da177e4c
|
3832 3833 3834 |
if (err) printk(KERN_ERR "enable_cpucache failed for %s, error %d. ", |
b28a02de8
|
3835 |
cachep->name, -err); |
2ed3a4ef9
|
3836 |
return err; |
1da177e4c
|
3837 |
} |
1b55253a7
|
3838 3839 |
/* * Drain an array if it contains any elements taking the l3 lock only if |
b18e7e654
|
3840 3841 |
* necessary. Note that the l3 listlock also protects the array_cache * if drain_array() is used on the shared array. |
1b55253a7
|
3842 3843 3844 |
*/ void drain_array(struct kmem_cache *cachep, struct kmem_list3 *l3, struct array_cache *ac, int force, int node) |
1da177e4c
|
3845 3846 |
{ int tofree; |
1b55253a7
|
3847 3848 |
if (!ac || !ac->avail) return; |
1da177e4c
|
3849 3850 |
if (ac->touched && !force) { ac->touched = 0; |
b18e7e654
|
3851 |
} else { |
1b55253a7
|
3852 |
spin_lock_irq(&l3->list_lock); |
b18e7e654
|
3853 3854 3855 3856 3857 3858 3859 3860 3861 |
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
|
3862 |
spin_unlock_irq(&l3->list_lock); |
1da177e4c
|
3863 3864 3865 3866 3867 |
} } /** * cache_reap - Reclaim memory from caches. |
05fb6bf0b
|
3868 |
* @w: work descriptor |
1da177e4c
|
3869 3870 3871 3872 3873 3874 |
* * 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
|
3875 3876 |
* If we cannot acquire the cache chain mutex then just give up - we'll try * again on the next iteration. |
1da177e4c
|
3877 |
*/ |
7c5cae368
|
3878 |
static void cache_reap(struct work_struct *w) |
1da177e4c
|
3879 |
{ |
7a7c381d2
|
3880 |
struct kmem_cache *searchp; |
e498be7da
|
3881 |
struct kmem_list3 *l3; |
aab2207cf
|
3882 |
int node = numa_node_id(); |
bf6aede71
|
3883 |
struct delayed_work *work = to_delayed_work(w); |
1da177e4c
|
3884 |
|
7c5cae368
|
3885 |
if (!mutex_trylock(&cache_chain_mutex)) |
1da177e4c
|
3886 |
/* Give up. Setup the next iteration. */ |
7c5cae368
|
3887 |
goto out; |
1da177e4c
|
3888 |
|
7a7c381d2
|
3889 |
list_for_each_entry(searchp, &cache_chain, next) { |
1da177e4c
|
3890 |
check_irq_on(); |
35386e3b0
|
3891 3892 3893 3894 3895 |
/* * 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
|
3896 |
l3 = searchp->nodelists[node]; |
35386e3b0
|
3897 |
|
8fce4d8e3
|
3898 |
reap_alien(searchp, l3); |
1da177e4c
|
3899 |
|
aab2207cf
|
3900 |
drain_array(searchp, l3, cpu_cache_get(searchp), 0, node); |
1da177e4c
|
3901 |
|
35386e3b0
|
3902 3903 3904 3905 |
/* * These are racy checks but it does not matter * if we skip one check or scan twice. */ |
e498be7da
|
3906 |
if (time_after(l3->next_reap, jiffies)) |
35386e3b0
|
3907 |
goto next; |
1da177e4c
|
3908 |
|
e498be7da
|
3909 |
l3->next_reap = jiffies + REAPTIMEOUT_LIST3; |
1da177e4c
|
3910 |
|
aab2207cf
|
3911 |
drain_array(searchp, l3, l3->shared, 0, node); |
1da177e4c
|
3912 |
|
ed11d9eb2
|
3913 |
if (l3->free_touched) |
e498be7da
|
3914 |
l3->free_touched = 0; |
ed11d9eb2
|
3915 3916 |
else { int freed; |
1da177e4c
|
3917 |
|
ed11d9eb2
|
3918 3919 3920 3921 |
freed = drain_freelist(searchp, l3, (l3->free_limit + 5 * searchp->num - 1) / (5 * searchp->num)); STATS_ADD_REAPED(searchp, freed); } |
35386e3b0
|
3922 |
next: |
1da177e4c
|
3923 3924 3925 |
cond_resched(); } check_irq_on(); |
fc0abb145
|
3926 |
mutex_unlock(&cache_chain_mutex); |
8fce4d8e3
|
3927 |
next_reap_node(); |
7c5cae368
|
3928 |
out: |
a737b3e2f
|
3929 |
/* Set up the next iteration */ |
7c5cae368
|
3930 |
schedule_delayed_work(work, round_jiffies_relative(REAPTIMEOUT_CPUC)); |
1da177e4c
|
3931 |
} |
158a96242
|
3932 |
#ifdef CONFIG_SLABINFO |
1da177e4c
|
3933 |
|
85289f98d
|
3934 |
static void print_slabinfo_header(struct seq_file *m) |
1da177e4c
|
3935 |
{ |
85289f98d
|
3936 3937 3938 3939 |
/* * Output format version, so at least we can change it * without _too_ many complaints. */ |
1da177e4c
|
3940 |
#if STATS |
85289f98d
|
3941 3942 |
seq_puts(m, "slabinfo - version: 2.1 (statistics) "); |
1da177e4c
|
3943 |
#else |
85289f98d
|
3944 3945 |
seq_puts(m, "slabinfo - version: 2.1 "); |
1da177e4c
|
3946 |
#endif |
85289f98d
|
3947 3948 3949 3950 |
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
|
3951 |
#if STATS |
85289f98d
|
3952 |
seq_puts(m, " : globalstat <listallocs> <maxobjs> <grown> <reaped> " |
fb7faf331
|
3953 |
"<error> <maxfreeable> <nodeallocs> <remotefrees> <alienoverflow>"); |
85289f98d
|
3954 |
seq_puts(m, " : cpustat <allochit> <allocmiss> <freehit> <freemiss>"); |
1da177e4c
|
3955 |
#endif |
85289f98d
|
3956 3957 3958 3959 3960 3961 3962 |
seq_putc(m, ' '); } static void *s_start(struct seq_file *m, loff_t *pos) { loff_t n = *pos; |
85289f98d
|
3963 |
|
fc0abb145
|
3964 |
mutex_lock(&cache_chain_mutex); |
85289f98d
|
3965 3966 |
if (!n) print_slabinfo_header(m); |
b92151bab
|
3967 3968 |
return seq_list_start(&cache_chain, *pos); |
1da177e4c
|
3969 3970 3971 3972 |
} static void *s_next(struct seq_file *m, void *p, loff_t *pos) { |
b92151bab
|
3973 |
return seq_list_next(p, &cache_chain, pos); |
1da177e4c
|
3974 3975 3976 3977 |
} static void s_stop(struct seq_file *m, void *p) { |
fc0abb145
|
3978 |
mutex_unlock(&cache_chain_mutex); |
1da177e4c
|
3979 3980 3981 3982 |
} static int s_show(struct seq_file *m, void *p) { |
b92151bab
|
3983 |
struct kmem_cache *cachep = list_entry(p, struct kmem_cache, next); |
b28a02de8
|
3984 3985 3986 3987 3988 |
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
|
3989 |
const char *name; |
1da177e4c
|
3990 |
char *error = NULL; |
e498be7da
|
3991 3992 |
int node; struct kmem_list3 *l3; |
1da177e4c
|
3993 |
|
1da177e4c
|
3994 3995 |
active_objs = 0; num_slabs = 0; |
e498be7da
|
3996 3997 3998 3999 |
for_each_online_node(node) { l3 = cachep->nodelists[node]; if (!l3) continue; |
ca3b9b917
|
4000 4001 |
check_irq_on(); spin_lock_irq(&l3->list_lock); |
e498be7da
|
4002 |
|
7a7c381d2
|
4003 |
list_for_each_entry(slabp, &l3->slabs_full, list) { |
e498be7da
|
4004 4005 4006 4007 4008 |
if (slabp->inuse != cachep->num && !error) error = "slabs_full accounting error"; active_objs += cachep->num; active_slabs++; } |
7a7c381d2
|
4009 |
list_for_each_entry(slabp, &l3->slabs_partial, list) { |
e498be7da
|
4010 4011 4012 4013 4014 4015 4016 |
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
|
4017 |
list_for_each_entry(slabp, &l3->slabs_free, list) { |
e498be7da
|
4018 4019 4020 4021 4022 |
if (slabp->inuse && !error) error = "slabs_free/inuse accounting error"; num_slabs++; } free_objects += l3->free_objects; |
4484ebf12
|
4023 4024 |
if (l3->shared) shared_avail += l3->shared->avail; |
e498be7da
|
4025 |
|
ca3b9b917
|
4026 |
spin_unlock_irq(&l3->list_lock); |
1da177e4c
|
4027 |
} |
b28a02de8
|
4028 4029 |
num_slabs += active_slabs; num_objs = num_slabs * cachep->num; |
e498be7da
|
4030 |
if (num_objs - active_objs != free_objects && !error) |
1da177e4c
|
4031 |
error = "free_objects accounting error"; |
b28a02de8
|
4032 |
name = cachep->name; |
1da177e4c
|
4033 4034 4035 4036 4037 |
if (error) printk(KERN_ERR "slab: cache %s error: %s ", name, error); seq_printf(m, "%-17s %6lu %6lu %6u %4u %4d", |
3dafccf22
|
4038 |
name, active_objs, num_objs, cachep->buffer_size, |
b28a02de8
|
4039 |
cachep->num, (1 << cachep->gfporder)); |
1da177e4c
|
4040 |
seq_printf(m, " : tunables %4u %4u %4u", |
b28a02de8
|
4041 |
cachep->limit, cachep->batchcount, cachep->shared); |
e498be7da
|
4042 |
seq_printf(m, " : slabdata %6lu %6lu %6lu", |
b28a02de8
|
4043 |
active_slabs, num_slabs, shared_avail); |
1da177e4c
|
4044 |
#if STATS |
b28a02de8
|
4045 |
{ /* list3 stats */ |
1da177e4c
|
4046 4047 4048 4049 4050 4051 |
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
|
4052 |
unsigned long node_allocs = cachep->node_allocs; |
e498be7da
|
4053 |
unsigned long node_frees = cachep->node_frees; |
fb7faf331
|
4054 |
unsigned long overflows = cachep->node_overflow; |
1da177e4c
|
4055 |
|
e498be7da
|
4056 |
seq_printf(m, " : globalstat %7lu %6lu %5lu %4lu \ |
fb7faf331
|
4057 |
%4lu %4lu %4lu %4lu %4lu", allocs, high, grown, |
a737b3e2f
|
4058 |
reaped, errors, max_freeable, node_allocs, |
fb7faf331
|
4059 |
node_frees, overflows); |
1da177e4c
|
4060 4061 4062 4063 4064 4065 4066 4067 4068 |
} /* 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
|
4069 |
allochit, allocmiss, freehit, freemiss); |
1da177e4c
|
4070 4071 4072 4073 |
} #endif seq_putc(m, ' '); |
1da177e4c
|
4074 4075 4076 4077 4078 4079 4080 4081 4082 4083 4084 4085 4086 4087 4088 4089 |
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
|
4090 |
static const struct seq_operations slabinfo_op = { |
b28a02de8
|
4091 4092 4093 4094 |
.start = s_start, .next = s_next, .stop = s_stop, .show = s_show, |
1da177e4c
|
4095 4096 4097 4098 4099 4100 4101 4102 4103 4104 |
}; #define MAX_SLABINFO_WRITE 128 /** * slabinfo_write - Tuning for the slab allocator * @file: unused * @buffer: user buffer * @count: data length * @ppos: unused */ |
b28a02de8
|
4105 4106 |
ssize_t slabinfo_write(struct file *file, const char __user * buffer, size_t count, loff_t *ppos) |
1da177e4c
|
4107 |
{ |
b28a02de8
|
4108 |
char kbuf[MAX_SLABINFO_WRITE + 1], *tmp; |
1da177e4c
|
4109 |
int limit, batchcount, shared, res; |
7a7c381d2
|
4110 |
struct kmem_cache *cachep; |
b28a02de8
|
4111 |
|
1da177e4c
|
4112 4113 4114 4115 |
if (count > MAX_SLABINFO_WRITE) return -EINVAL; if (copy_from_user(&kbuf, buffer, count)) return -EFAULT; |
b28a02de8
|
4116 |
kbuf[MAX_SLABINFO_WRITE] = '\0'; |
1da177e4c
|
4117 4118 4119 4120 4121 4122 4123 4124 4125 4126 |
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
|
4127 |
mutex_lock(&cache_chain_mutex); |
1da177e4c
|
4128 |
res = -EINVAL; |
7a7c381d2
|
4129 |
list_for_each_entry(cachep, &cache_chain, next) { |
1da177e4c
|
4130 |
if (!strcmp(cachep->name, kbuf)) { |
a737b3e2f
|
4131 4132 |
if (limit < 1 || batchcount < 1 || batchcount > limit || shared < 0) { |
e498be7da
|
4133 |
res = 0; |
1da177e4c
|
4134 |
} else { |
e498be7da
|
4135 |
res = do_tune_cpucache(cachep, limit, |
83b519e8b
|
4136 4137 |
batchcount, shared, GFP_KERNEL); |
1da177e4c
|
4138 4139 4140 4141 |
} break; } } |
fc0abb145
|
4142 |
mutex_unlock(&cache_chain_mutex); |
1da177e4c
|
4143 4144 4145 4146 |
if (res >= 0) res = count; return res; } |
871751e25
|
4147 |
|
7b3c3a50a
|
4148 4149 4150 4151 4152 4153 4154 4155 4156 4157 4158 4159 |
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
|
4160 4161 4162 4163 |
#ifdef CONFIG_DEBUG_SLAB_LEAK static void *leaks_start(struct seq_file *m, loff_t *pos) { |
871751e25
|
4164 |
mutex_lock(&cache_chain_mutex); |
b92151bab
|
4165 |
return seq_list_start(&cache_chain, *pos); |
871751e25
|
4166 4167 4168 4169 4170 4171 4172 4173 4174 4175 4176 4177 4178 4179 4180 4181 4182 4183 4184 4185 4186 4187 4188 4189 4190 4191 4192 4193 4194 4195 4196 4197 4198 4199 4200 4201 4202 4203 4204 4205 4206 4207 4208 4209 4210 4211 4212 4213 4214 |
} 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
|
4215 |
unsigned long offset, size; |
9281acea6
|
4216 |
char modname[MODULE_NAME_LEN], name[KSYM_NAME_LEN]; |
871751e25
|
4217 |
|
a5c43dae7
|
4218 |
if (lookup_symbol_attrs(address, &size, &offset, modname, name) == 0) { |
871751e25
|
4219 |
seq_printf(m, "%s+%#lx/%#lx", name, offset, size); |
a5c43dae7
|
4220 |
if (modname[0]) |
871751e25
|
4221 4222 4223 4224 4225 4226 4227 4228 4229 |
seq_printf(m, " [%s]", modname); return; } #endif seq_printf(m, "%p", (void *)address); } static int leaks_show(struct seq_file *m, void *p) { |
b92151bab
|
4230 |
struct kmem_cache *cachep = list_entry(p, struct kmem_cache, next); |
871751e25
|
4231 4232 4233 4234 4235 4236 4237 4238 4239 4240 4241 4242 4243 4244 4245 4246 4247 4248 4249 4250 4251 4252 4253 |
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
|
4254 |
list_for_each_entry(slabp, &l3->slabs_full, list) |
871751e25
|
4255 |
handle_slab(n, cachep, slabp); |
7a7c381d2
|
4256 |
list_for_each_entry(slabp, &l3->slabs_partial, list) |
871751e25
|
4257 |
handle_slab(n, cachep, slabp); |
871751e25
|
4258 4259 4260 4261 4262 4263 4264 4265 4266 4267 4268 4269 4270 4271 4272 4273 4274 4275 4276 4277 4278 4279 4280 4281 4282 4283 |
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
|
4284 |
|
871751e25
|
4285 4286 |
return 0; } |
a0ec95a8e
|
4287 |
static const struct seq_operations slabstats_op = { |
871751e25
|
4288 4289 4290 4291 4292 |
.start = leaks_start, .next = s_next, .stop = s_stop, .show = leaks_show, }; |
a0ec95a8e
|
4293 4294 4295 4296 4297 4298 4299 4300 4301 4302 4303 4304 4305 4306 4307 4308 4309 4310 4311 4312 4313 4314 4315 4316 4317 4318 4319 4320 |
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
|
4321 |
proc_create("slabinfo",S_IWUSR|S_IRUGO,NULL,&proc_slabinfo_operations); |
a0ec95a8e
|
4322 4323 |
#ifdef CONFIG_DEBUG_SLAB_LEAK proc_create("slab_allocators", 0, NULL, &proc_slabstats_operations); |
871751e25
|
4324 |
#endif |
a0ec95a8e
|
4325 4326 4327 |
return 0; } module_init(slab_proc_init); |
1da177e4c
|
4328 |
#endif |
00e145b6d
|
4329 4330 4331 4332 4333 4334 4335 4336 4337 4338 4339 4340 |
/** * 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
|
4341 |
size_t ksize(const void *objp) |
1da177e4c
|
4342 |
{ |
ef8b4520b
|
4343 4344 |
BUG_ON(!objp); if (unlikely(objp == ZERO_SIZE_PTR)) |
00e145b6d
|
4345 |
return 0; |
1da177e4c
|
4346 |
|
6ed5eb221
|
4347 |
return obj_size(virt_to_cache(objp)); |
1da177e4c
|
4348 |
} |
b1aabecd5
|
4349 |
EXPORT_SYMBOL(ksize); |