Blame view
mm/slab.c
119 KB
1da177e4c Linux-2.6.12-rc2 |
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 spelling fixes: mm/ |
29 |
* slabs and you must pass objects with the same initializations to |
1da177e4c Linux-2.6.12-rc2 |
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 [PATCH] slab cleanup |
53 |
* The c_cpuarray may not be read with enabled local interrupts - |
1da177e4c Linux-2.6.12-rc2 |
54 55 56 57 |
* it's changed with a smp_call_function(). * * SMP synchronization: * constructors and destructors are called without any locking. |
343e0d7a9 [PATCH] slab: rep... |
58 |
* Several members in struct kmem_cache and struct slab never change, they |
1da177e4c Linux-2.6.12-rc2 |
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 [PATCH] sem2mutex... |
71 |
* The global cache-chain is protected by the mutex 'cache_chain_mutex'. |
1da177e4c Linux-2.6.12-rc2 |
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 [PATCH] Numa-awar... |
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 Linux-2.6.12-rc2 |
87 |
*/ |
1da177e4c Linux-2.6.12-rc2 |
88 89 |
#include <linux/slab.h> #include <linux/mm.h> |
c9cf55285 [PATCH] add poiso... |
90 |
#include <linux/poison.h> |
1da177e4c Linux-2.6.12-rc2 |
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 [PATCH] cpuset me... |
96 |
#include <linux/cpuset.h> |
a0ec95a8e proc: move /proc/... |
97 |
#include <linux/proc_fs.h> |
1da177e4c Linux-2.6.12-rc2 |
98 99 100 101 102 103 104 |
#include <linux/seq_file.h> #include <linux/notifier.h> #include <linux/kallsyms.h> #include <linux/cpu.h> #include <linux/sysctl.h> #include <linux/module.h> #include <linux/rcupdate.h> |
543537bd9 [PATCH] create a ... |
105 |
#include <linux/string.h> |
138ae6631 [PATCH] slab: use... |
106 |
#include <linux/uaccess.h> |
e498be7da [PATCH] Numa-awar... |
107 |
#include <linux/nodemask.h> |
d5cff6352 kmemleak: Add the... |
108 |
#include <linux/kmemleak.h> |
dc85da15d [PATCH] NUMA poli... |
109 |
#include <linux/mempolicy.h> |
fc0abb145 [PATCH] sem2mutex... |
110 |
#include <linux/mutex.h> |
8a8b6502f [PATCH] fault-inj... |
111 |
#include <linux/fault-inject.h> |
e7eebaf6a [PATCH] pi-futex:... |
112 |
#include <linux/rtmutex.h> |
6a2d7a955 [PATCH] SLAB: use... |
113 |
#include <linux/reciprocal_div.h> |
3ac7fe5a4 infrastructure to... |
114 |
#include <linux/debugobjects.h> |
c175eea46 slab: add hooks f... |
115 |
#include <linux/kmemcheck.h> |
8f9f8d9e8 slab: add memory ... |
116 |
#include <linux/memory.h> |
268bb0ce3 sanitize <linux/p... |
117 |
#include <linux/prefetch.h> |
1da177e4c Linux-2.6.12-rc2 |
118 |
|
1da177e4c Linux-2.6.12-rc2 |
119 120 121 |
#include <asm/cacheflush.h> #include <asm/tlbflush.h> #include <asm/page.h> |
4dee6b64e tracing/mm: Move ... |
122 |
#include <trace/events/kmem.h> |
1da177e4c Linux-2.6.12-rc2 |
123 |
/* |
50953fe9e slab allocators: ... |
124 |
* DEBUG - 1 for kmem_cache_create() to honour; SLAB_RED_ZONE & SLAB_POISON. |
1da177e4c Linux-2.6.12-rc2 |
125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 |
* 0 for faster, smaller code (especially in the critical paths). * * STATS - 1 to collect stats for /proc/slabinfo. * 0 for faster, smaller code (especially in the critical paths). * * FORCED_DEBUG - 1 enables SLAB_RED_ZONE and SLAB_POISON (if possible) */ #ifdef CONFIG_DEBUG_SLAB #define DEBUG 1 #define STATS 1 #define FORCED_DEBUG 1 #else #define DEBUG 0 #define STATS 0 #define FORCED_DEBUG 0 #endif |
1da177e4c Linux-2.6.12-rc2 |
142 143 |
/* Shouldn't this be in a header file somewhere? */ #define BYTES_PER_WORD sizeof(void *) |
87a927c71 Fix slab redzone ... |
144 |
#define REDZONE_ALIGN max(BYTES_PER_WORD, __alignof__(unsigned long long)) |
1da177e4c Linux-2.6.12-rc2 |
145 |
|
1da177e4c Linux-2.6.12-rc2 |
146 147 148 149 150 151 |
#ifndef ARCH_KMALLOC_FLAGS #define ARCH_KMALLOC_FLAGS SLAB_HWCACHE_ALIGN #endif /* Legal flag mask for kmem_cache_create(). */ #if DEBUG |
50953fe9e slab allocators: ... |
152 |
# define CREATE_MASK (SLAB_RED_ZONE | \ |
1da177e4c Linux-2.6.12-rc2 |
153 |
SLAB_POISON | SLAB_HWCACHE_ALIGN | \ |
ac2b898ca [PATCH] slab: Rem... |
154 |
SLAB_CACHE_DMA | \ |
5af608399 slab allocators: ... |
155 |
SLAB_STORE_USER | \ |
1da177e4c Linux-2.6.12-rc2 |
156 |
SLAB_RECLAIM_ACCOUNT | SLAB_PANIC | \ |
3ac7fe5a4 infrastructure to... |
157 |
SLAB_DESTROY_BY_RCU | SLAB_MEM_SPREAD | \ |
c175eea46 slab: add hooks f... |
158 |
SLAB_DEBUG_OBJECTS | SLAB_NOLEAKTRACE | SLAB_NOTRACK) |
1da177e4c Linux-2.6.12-rc2 |
159 |
#else |
ac2b898ca [PATCH] slab: Rem... |
160 |
# define CREATE_MASK (SLAB_HWCACHE_ALIGN | \ |
5af608399 slab allocators: ... |
161 |
SLAB_CACHE_DMA | \ |
1da177e4c Linux-2.6.12-rc2 |
162 |
SLAB_RECLAIM_ACCOUNT | SLAB_PANIC | \ |
3ac7fe5a4 infrastructure to... |
163 |
SLAB_DESTROY_BY_RCU | SLAB_MEM_SPREAD | \ |
c175eea46 slab: add hooks f... |
164 |
SLAB_DEBUG_OBJECTS | SLAB_NOLEAKTRACE | SLAB_NOTRACK) |
1da177e4c Linux-2.6.12-rc2 |
165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 |
#endif /* * kmem_bufctl_t: * * Bufctl's are used for linking objs within a slab * linked offsets. * * This implementation relies on "struct page" for locating the cache & * slab an object belongs to. * This allows the bufctl structure to be small (one int), but limits * the number of objects a slab (not a cache) can contain when off-slab * bufctls are used. The limit is the size of the largest general cache * that does not use off-slab slabs. * For 32bit archs with 4 kB pages, is this 56. * This is not serious, as it is only for large objects, when it is unwise * to have too many per slab. * Note: This limit can be raised by introducing a general cache whose size * is less than 512 (PAGE_SIZE<<3), but greater than 256. */ |
fa5b08d5f [PATCH] sab: cons... |
185 |
typedef unsigned int kmem_bufctl_t; |
1da177e4c Linux-2.6.12-rc2 |
186 187 |
#define BUFCTL_END (((kmem_bufctl_t)(~0U))-0) #define BUFCTL_FREE (((kmem_bufctl_t)(~0U))-1) |
871751e25 [PATCH] slab: imp... |
188 189 |
#define BUFCTL_ACTIVE (((kmem_bufctl_t)(~0U))-2) #define SLAB_LIMIT (((kmem_bufctl_t)(~0U))-3) |
1da177e4c Linux-2.6.12-rc2 |
190 |
|
1da177e4c Linux-2.6.12-rc2 |
191 |
/* |
1da177e4c Linux-2.6.12-rc2 |
192 193 194 195 196 197 198 199 200 201 202 203 |
* struct slab_rcu * * slab_destroy on a SLAB_DESTROY_BY_RCU cache uses this structure to * arrange for kmem_freepages to be called via RCU. This is useful if * we need to approach a kernel structure obliquely, from its address * obtained without the usual locking. We can lock the structure to * stabilize it and check it's still at the given address, only if we * can be sure that the memory has not been meanwhile reused for some * other kind of object (which our subsystem's lock might corrupt). * * rcu_read_lock before reading the address, then rcu_read_unlock after * taking the spinlock within the structure expected at that address. |
1da177e4c Linux-2.6.12-rc2 |
204 205 |
*/ struct slab_rcu { |
b28a02de8 [PATCH] slab: fix... |
206 |
struct rcu_head head; |
343e0d7a9 [PATCH] slab: rep... |
207 |
struct kmem_cache *cachep; |
b28a02de8 [PATCH] slab: fix... |
208 |
void *addr; |
1da177e4c Linux-2.6.12-rc2 |
209 210 211 |
}; /* |
5bfe53a77 slab,rcu: don't a... |
212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 |
* struct slab * * Manages the objs in a slab. Placed either at the beginning of mem allocated * for a slab, or allocated from an general cache. * Slabs are chained into three list: fully used, partial, fully free slabs. */ struct slab { union { struct { struct list_head list; unsigned long colouroff; void *s_mem; /* including colour offset */ unsigned int inuse; /* num of objs active in slab */ kmem_bufctl_t free; unsigned short nodeid; }; struct slab_rcu __slab_cover_slab_rcu; }; }; /* |
1da177e4c Linux-2.6.12-rc2 |
233 234 |
* struct array_cache * |
1da177e4c Linux-2.6.12-rc2 |
235 236 237 238 239 240 241 242 243 244 245 246 247 248 |
* Purpose: * - LIFO ordering, to hand out cache-warm objects from _alloc * - reduce the number of linked list operations * - reduce spinlock operations * * The limit is stored in the per-cpu structure to reduce the data cache * footprint. * */ struct array_cache { unsigned int avail; unsigned int limit; unsigned int batchcount; unsigned int touched; |
e498be7da [PATCH] Numa-awar... |
249 |
spinlock_t lock; |
bda5b655f Delete gcc-2.95 c... |
250 |
void *entry[]; /* |
a737b3e2f [PATCH] slab cleanup |
251 252 253 |
* Must have this definition in here for the proper * alignment of array_cache. Also simplifies accessing * the entries. |
a737b3e2f [PATCH] slab cleanup |
254 |
*/ |
1da177e4c Linux-2.6.12-rc2 |
255 |
}; |
a737b3e2f [PATCH] slab cleanup |
256 257 258 |
/* * bootstrap: The caches do not work without cpuarrays anymore, but the * cpuarrays are allocated from the generic caches... |
1da177e4c Linux-2.6.12-rc2 |
259 260 261 262 |
*/ #define BOOT_CPUCACHE_ENTRIES 1 struct arraycache_init { struct array_cache cache; |
b28a02de8 [PATCH] slab: fix... |
263 |
void *entries[BOOT_CPUCACHE_ENTRIES]; |
1da177e4c Linux-2.6.12-rc2 |
264 265 266 |
}; /* |
e498be7da [PATCH] Numa-awar... |
267 |
* The slab lists for all objects. |
1da177e4c Linux-2.6.12-rc2 |
268 269 |
*/ struct kmem_list3 { |
b28a02de8 [PATCH] slab: fix... |
270 271 272 273 |
struct list_head slabs_partial; /* partial list first, better asm code */ struct list_head slabs_full; struct list_head slabs_free; unsigned long free_objects; |
b28a02de8 [PATCH] slab: fix... |
274 |
unsigned int free_limit; |
2e1217cf9 [PATCH] NUMA slab... |
275 |
unsigned int colour_next; /* Per-node cache coloring */ |
b28a02de8 [PATCH] slab: fix... |
276 277 278 |
spinlock_t list_lock; struct array_cache *shared; /* shared per node */ struct array_cache **alien; /* on other nodes */ |
35386e3b0 [PATCH] slab: cac... |
279 280 |
unsigned long next_reap; /* updated without locking */ int free_touched; /* updated without locking */ |
1da177e4c Linux-2.6.12-rc2 |
281 |
}; |
e498be7da [PATCH] Numa-awar... |
282 283 284 |
/* * Need this for bootstrapping a per node allocator. */ |
556a169da slab: fix bootstr... |
285 |
#define NUM_INIT_LISTS (3 * MAX_NUMNODES) |
68a1b1955 mm/slab.c: make l... |
286 |
static struct kmem_list3 __initdata initkmem_list3[NUM_INIT_LISTS]; |
e498be7da [PATCH] Numa-awar... |
287 |
#define CACHE_CACHE 0 |
556a169da slab: fix bootstr... |
288 289 |
#define SIZE_AC MAX_NUMNODES #define SIZE_L3 (2 * MAX_NUMNODES) |
e498be7da [PATCH] Numa-awar... |
290 |
|
ed11d9eb2 [PATCH] slab: con... |
291 292 293 294 |
static int drain_freelist(struct kmem_cache *cache, struct kmem_list3 *l3, int tofree); static void free_block(struct kmem_cache *cachep, void **objpp, int len, int node); |
83b519e8b slab: setup alloc... |
295 |
static int enable_cpucache(struct kmem_cache *cachep, gfp_t gfp); |
65f27f384 WorkStruct: Pass ... |
296 |
static void cache_reap(struct work_struct *unused); |
ed11d9eb2 [PATCH] slab: con... |
297 |
|
e498be7da [PATCH] Numa-awar... |
298 |
/* |
a737b3e2f [PATCH] slab cleanup |
299 300 |
* This function must be completely optimized away if a constant is passed to * it. Mostly the same as what is in linux/slab.h except it returns an index. |
e498be7da [PATCH] Numa-awar... |
301 |
*/ |
7243cc05b [PATCH] slab: alp... |
302 |
static __always_inline int index_of(const size_t size) |
e498be7da [PATCH] Numa-awar... |
303 |
{ |
5ec8a847b [PATCH] slab: hav... |
304 |
extern void __bad_size(void); |
e498be7da [PATCH] Numa-awar... |
305 306 307 308 309 310 311 312 |
if (__builtin_constant_p(size)) { int i = 0; #define CACHE(x) \ if (size <=x) \ return i; \ else \ i++; |
1c61fc40f slab - use angle ... |
313 |
#include <linux/kmalloc_sizes.h> |
e498be7da [PATCH] Numa-awar... |
314 |
#undef CACHE |
5ec8a847b [PATCH] slab: hav... |
315 |
__bad_size(); |
7243cc05b [PATCH] slab: alp... |
316 |
} else |
5ec8a847b [PATCH] slab: hav... |
317 |
__bad_size(); |
e498be7da [PATCH] Numa-awar... |
318 319 |
return 0; } |
e0a427267 [PATCH] mm/slab.c... |
320 |
static int slab_early_init = 1; |
e498be7da [PATCH] Numa-awar... |
321 322 |
#define INDEX_AC index_of(sizeof(struct arraycache_init)) #define INDEX_L3 index_of(sizeof(struct kmem_list3)) |
1da177e4c Linux-2.6.12-rc2 |
323 |
|
5295a74cc [PATCH] slab: red... |
324 |
static void kmem_list3_init(struct kmem_list3 *parent) |
e498be7da [PATCH] Numa-awar... |
325 326 327 328 329 330 |
{ INIT_LIST_HEAD(&parent->slabs_full); INIT_LIST_HEAD(&parent->slabs_partial); INIT_LIST_HEAD(&parent->slabs_free); parent->shared = NULL; parent->alien = NULL; |
2e1217cf9 [PATCH] NUMA slab... |
331 |
parent->colour_next = 0; |
e498be7da [PATCH] Numa-awar... |
332 333 334 335 |
spin_lock_init(&parent->list_lock); parent->free_objects = 0; parent->free_touched = 0; } |
a737b3e2f [PATCH] slab cleanup |
336 337 338 339 |
#define MAKE_LIST(cachep, listp, slab, nodeid) \ do { \ INIT_LIST_HEAD(listp); \ list_splice(&(cachep->nodelists[nodeid]->slab), listp); \ |
e498be7da [PATCH] Numa-awar... |
340 |
} while (0) |
a737b3e2f [PATCH] slab cleanup |
341 342 |
#define MAKE_ALL_LISTS(cachep, ptr, nodeid) \ do { \ |
e498be7da [PATCH] Numa-awar... |
343 344 345 346 |
MAKE_LIST((cachep), (&(ptr)->slabs_full), slabs_full, nodeid); \ MAKE_LIST((cachep), (&(ptr)->slabs_partial), slabs_partial, nodeid); \ MAKE_LIST((cachep), (&(ptr)->slabs_free), slabs_free, nodeid); \ } while (0) |
1da177e4c Linux-2.6.12-rc2 |
347 |
|
1da177e4c Linux-2.6.12-rc2 |
348 349 350 351 |
#define CFLGS_OFF_SLAB (0x80000000UL) #define OFF_SLAB(x) ((x)->flags & CFLGS_OFF_SLAB) #define BATCHREFILL_LIMIT 16 |
a737b3e2f [PATCH] slab cleanup |
352 353 354 |
/* * Optimization question: fewer reaps means less probability for unnessary * cpucache drain/refill cycles. |
1da177e4c Linux-2.6.12-rc2 |
355 |
* |
dc6f3f276 mm/slab.c: fix a ... |
356 |
* OTOH the cpuarrays can contain lots of objects, |
1da177e4c Linux-2.6.12-rc2 |
357 358 359 360 361 362 363 364 365 366 |
* which could lock up otherwise freeable slabs. */ #define REAPTIMEOUT_CPUC (2*HZ) #define REAPTIMEOUT_LIST3 (4*HZ) #if STATS #define STATS_INC_ACTIVE(x) ((x)->num_active++) #define STATS_DEC_ACTIVE(x) ((x)->num_active--) #define STATS_INC_ALLOCED(x) ((x)->num_allocations++) #define STATS_INC_GROWN(x) ((x)->grown++) |
ed11d9eb2 [PATCH] slab: con... |
367 |
#define STATS_ADD_REAPED(x,y) ((x)->reaped += (y)) |
a737b3e2f [PATCH] slab cleanup |
368 369 370 371 372 |
#define STATS_SET_HIGH(x) \ do { \ if ((x)->num_active > (x)->high_mark) \ (x)->high_mark = (x)->num_active; \ } while (0) |
1da177e4c Linux-2.6.12-rc2 |
373 374 |
#define STATS_INC_ERR(x) ((x)->errors++) #define STATS_INC_NODEALLOCS(x) ((x)->node_allocs++) |
e498be7da [PATCH] Numa-awar... |
375 |
#define STATS_INC_NODEFREES(x) ((x)->node_frees++) |
fb7faf331 [PATCH] slab: add... |
376 |
#define STATS_INC_ACOVERFLOW(x) ((x)->node_overflow++) |
a737b3e2f [PATCH] slab cleanup |
377 378 379 380 381 |
#define STATS_SET_FREEABLE(x, i) \ do { \ if ((x)->max_freeable < i) \ (x)->max_freeable = i; \ } while (0) |
1da177e4c Linux-2.6.12-rc2 |
382 383 384 385 386 387 388 389 390 |
#define STATS_INC_ALLOCHIT(x) atomic_inc(&(x)->allochit) #define STATS_INC_ALLOCMISS(x) atomic_inc(&(x)->allocmiss) #define STATS_INC_FREEHIT(x) atomic_inc(&(x)->freehit) #define STATS_INC_FREEMISS(x) atomic_inc(&(x)->freemiss) #else #define STATS_INC_ACTIVE(x) do { } while (0) #define STATS_DEC_ACTIVE(x) do { } while (0) #define STATS_INC_ALLOCED(x) do { } while (0) #define STATS_INC_GROWN(x) do { } while (0) |
4e60c86bd gcc-4.6: mm: fix ... |
391 |
#define STATS_ADD_REAPED(x,y) do { (void)(y); } while (0) |
1da177e4c Linux-2.6.12-rc2 |
392 393 394 |
#define STATS_SET_HIGH(x) do { } while (0) #define STATS_INC_ERR(x) do { } while (0) #define STATS_INC_NODEALLOCS(x) do { } while (0) |
e498be7da [PATCH] Numa-awar... |
395 |
#define STATS_INC_NODEFREES(x) do { } while (0) |
fb7faf331 [PATCH] slab: add... |
396 |
#define STATS_INC_ACOVERFLOW(x) do { } while (0) |
a737b3e2f [PATCH] slab cleanup |
397 |
#define STATS_SET_FREEABLE(x, i) do { } while (0) |
1da177e4c Linux-2.6.12-rc2 |
398 399 400 401 402 403 404 |
#define STATS_INC_ALLOCHIT(x) do { } while (0) #define STATS_INC_ALLOCMISS(x) do { } while (0) #define STATS_INC_FREEHIT(x) do { } while (0) #define STATS_INC_FREEMISS(x) do { } while (0) #endif #if DEBUG |
1da177e4c Linux-2.6.12-rc2 |
405 |
|
a737b3e2f [PATCH] slab cleanup |
406 407 |
/* * memory layout of objects: |
1da177e4c Linux-2.6.12-rc2 |
408 |
* 0 : objp |
3dafccf22 [PATCH] slab: dis... |
409 |
* 0 .. cachep->obj_offset - BYTES_PER_WORD - 1: padding. This ensures that |
1da177e4c Linux-2.6.12-rc2 |
410 411 |
* the end of an object is aligned with the end of the real * allocation. Catches writes behind the end of the allocation. |
3dafccf22 [PATCH] slab: dis... |
412 |
* cachep->obj_offset - BYTES_PER_WORD .. cachep->obj_offset - 1: |
1da177e4c Linux-2.6.12-rc2 |
413 |
* redzone word. |
3dafccf22 [PATCH] slab: dis... |
414 415 |
* cachep->obj_offset: The real object. * cachep->buffer_size - 2* BYTES_PER_WORD: redzone word [BYTES_PER_WORD long] |
a737b3e2f [PATCH] slab cleanup |
416 417 |
* cachep->buffer_size - 1* BYTES_PER_WORD: last caller address * [BYTES_PER_WORD long] |
1da177e4c Linux-2.6.12-rc2 |
418 |
*/ |
343e0d7a9 [PATCH] slab: rep... |
419 |
static int obj_offset(struct kmem_cache *cachep) |
1da177e4c Linux-2.6.12-rc2 |
420 |
{ |
3dafccf22 [PATCH] slab: dis... |
421 |
return cachep->obj_offset; |
1da177e4c Linux-2.6.12-rc2 |
422 |
} |
343e0d7a9 [PATCH] slab: rep... |
423 |
static int obj_size(struct kmem_cache *cachep) |
1da177e4c Linux-2.6.12-rc2 |
424 |
{ |
3dafccf22 [PATCH] slab: dis... |
425 |
return cachep->obj_size; |
1da177e4c Linux-2.6.12-rc2 |
426 |
} |
b46b8f19c Increase slab red... |
427 |
static unsigned long long *dbg_redzone1(struct kmem_cache *cachep, void *objp) |
1da177e4c Linux-2.6.12-rc2 |
428 429 |
{ BUG_ON(!(cachep->flags & SLAB_RED_ZONE)); |
b46b8f19c Increase slab red... |
430 431 |
return (unsigned long long*) (objp + obj_offset(cachep) - sizeof(unsigned long long)); |
1da177e4c Linux-2.6.12-rc2 |
432 |
} |
b46b8f19c Increase slab red... |
433 |
static unsigned long long *dbg_redzone2(struct kmem_cache *cachep, void *objp) |
1da177e4c Linux-2.6.12-rc2 |
434 435 436 |
{ BUG_ON(!(cachep->flags & SLAB_RED_ZONE)); if (cachep->flags & SLAB_STORE_USER) |
b46b8f19c Increase slab red... |
437 438 |
return (unsigned long long *)(objp + cachep->buffer_size - sizeof(unsigned long long) - |
87a927c71 Fix slab redzone ... |
439 |
REDZONE_ALIGN); |
b46b8f19c Increase slab red... |
440 441 |
return (unsigned long long *) (objp + cachep->buffer_size - sizeof(unsigned long long)); |
1da177e4c Linux-2.6.12-rc2 |
442 |
} |
343e0d7a9 [PATCH] slab: rep... |
443 |
static void **dbg_userword(struct kmem_cache *cachep, void *objp) |
1da177e4c Linux-2.6.12-rc2 |
444 445 |
{ BUG_ON(!(cachep->flags & SLAB_STORE_USER)); |
3dafccf22 [PATCH] slab: dis... |
446 |
return (void **)(objp + cachep->buffer_size - BYTES_PER_WORD); |
1da177e4c Linux-2.6.12-rc2 |
447 448 449 |
} #else |
3dafccf22 [PATCH] slab: dis... |
450 451 |
#define obj_offset(x) 0 #define obj_size(cachep) (cachep->buffer_size) |
b46b8f19c Increase slab red... |
452 453 |
#define dbg_redzone1(cachep, objp) ({BUG(); (unsigned long long *)NULL;}) #define dbg_redzone2(cachep, objp) ({BUG(); (unsigned long long *)NULL;}) |
1da177e4c Linux-2.6.12-rc2 |
454 455 456 |
#define dbg_userword(cachep, objp) ({BUG(); (void **)NULL;}) #endif |
0f24f1287 tracing, slab: De... |
457 |
#ifdef CONFIG_TRACING |
36555751c kmemtrace: SLAB h... |
458 459 460 461 462 463 |
size_t slab_buffer_size(struct kmem_cache *cachep) { return cachep->buffer_size; } EXPORT_SYMBOL(slab_buffer_size); #endif |
1da177e4c Linux-2.6.12-rc2 |
464 |
/* |
3df1cccdf slab: introduce s... |
465 466 |
* Do not go above this order unless 0 objects fit into the slab or * overridden on the command line. |
1da177e4c Linux-2.6.12-rc2 |
467 |
*/ |
543585cc5 slab: rename slab... |
468 469 470 |
#define SLAB_MAX_ORDER_HI 1 #define SLAB_MAX_ORDER_LO 0 static int slab_max_order = SLAB_MAX_ORDER_LO; |
3df1cccdf slab: introduce s... |
471 |
static bool slab_max_order_set __initdata; |
1da177e4c Linux-2.6.12-rc2 |
472 |
|
a737b3e2f [PATCH] slab cleanup |
473 474 475 476 |
/* * 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 Linux-2.6.12-rc2 |
477 |
*/ |
065d41cb2 [PATCH] slab: con... |
478 479 480 481 482 483 484 |
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 Make page->privat... |
485 |
page = compound_head(page); |
ddc2e812d [PATCH] slab: ver... |
486 |
BUG_ON(!PageSlab(page)); |
065d41cb2 [PATCH] slab: con... |
487 488 489 490 491 492 493 494 495 496 |
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 [PATCH] slab: ver... |
497 |
BUG_ON(!PageSlab(page)); |
065d41cb2 [PATCH] slab: con... |
498 499 |
return (struct slab *)page->lru.prev; } |
1da177e4c Linux-2.6.12-rc2 |
500 |
|
6ed5eb221 [PATCH] slab: ext... |
501 502 |
static inline struct kmem_cache *virt_to_cache(const void *obj) { |
b49af68ff Add virt_to_head_... |
503 |
struct page *page = virt_to_head_page(obj); |
6ed5eb221 [PATCH] slab: ext... |
504 505 506 507 508 |
return page_get_cache(page); } static inline struct slab *virt_to_slab(const void *obj) { |
b49af68ff Add virt_to_head_... |
509 |
struct page *page = virt_to_head_page(obj); |
6ed5eb221 [PATCH] slab: ext... |
510 511 |
return page_get_slab(page); } |
8fea4e96a [PATCH] slab: obj... |
512 513 514 515 516 |
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 [PATCH] SLAB: use... |
517 518 519 520 521 522 523 524 |
/* * 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 [PATCH] slab: obj... |
525 |
{ |
6a2d7a955 [PATCH] SLAB: use... |
526 527 |
u32 offset = (obj - slab->s_mem); return reciprocal_divide(offset, cache->reciprocal_buffer_size); |
8fea4e96a [PATCH] slab: obj... |
528 |
} |
a737b3e2f [PATCH] slab cleanup |
529 530 531 |
/* * These are the default caches for kmalloc. Custom caches can have other sizes. */ |
1da177e4c Linux-2.6.12-rc2 |
532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 |
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 [PATCH] slab: fix... |
549 |
{NULL,} |
1da177e4c Linux-2.6.12-rc2 |
550 551 552 553 |
#undef CACHE }; static struct arraycache_init initarray_cache __initdata = |
b28a02de8 [PATCH] slab: fix... |
554 |
{ {0, BOOT_CPUCACHE_ENTRIES, 1, 0} }; |
1da177e4c Linux-2.6.12-rc2 |
555 |
static struct arraycache_init initarray_generic = |
b28a02de8 [PATCH] slab: fix... |
556 |
{ {0, BOOT_CPUCACHE_ENTRIES, 1, 0} }; |
1da177e4c Linux-2.6.12-rc2 |
557 558 |
/* internal cache of cache description objs */ |
b56efcf0a slab: shrink size... |
559 |
static struct kmem_list3 *cache_cache_nodelists[MAX_NUMNODES]; |
343e0d7a9 [PATCH] slab: rep... |
560 |
static struct kmem_cache cache_cache = { |
b56efcf0a slab: shrink size... |
561 |
.nodelists = cache_cache_nodelists, |
b28a02de8 [PATCH] slab: fix... |
562 563 564 |
.batchcount = 1, .limit = BOOT_CPUCACHE_ENTRIES, .shared = 1, |
343e0d7a9 [PATCH] slab: rep... |
565 |
.buffer_size = sizeof(struct kmem_cache), |
b28a02de8 [PATCH] slab: fix... |
566 |
.name = "kmem_cache", |
1da177e4c Linux-2.6.12-rc2 |
567 |
}; |
056c62418 [PATCH] slab: fix... |
568 |
#define BAD_ALIEN_MAGIC 0x01020304ul |
ce79ddc8e SLAB: Fix lockdep... |
569 570 571 572 573 574 575 576 577 |
/* * chicken and egg problem: delay the per-cpu array allocation * until the general caches are up. */ static enum { NONE, PARTIAL_AC, PARTIAL_L3, EARLY, |
52cef1891 slab, lockdep: Fi... |
578 |
LATE, |
ce79ddc8e SLAB: Fix lockdep... |
579 580 581 582 583 584 585 586 587 588 |
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 [PATCH] lockdep: ... |
589 590 591 592 593 594 595 596 |
#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 [PATCH] slab: fix... |
597 598 599 600 |
* * 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 [PATCH] lockdep: ... |
601 |
*/ |
056c62418 [PATCH] slab: fix... |
602 603 |
static struct lock_class_key on_slab_l3_key; static struct lock_class_key on_slab_alc_key; |
83835b3d9 slab, lockdep: An... |
604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 |
static struct lock_class_key debugobj_l3_key; static struct lock_class_key debugobj_alc_key; static void slab_set_lock_classes(struct kmem_cache *cachep, struct lock_class_key *l3_key, struct lock_class_key *alc_key, int q) { struct array_cache **alc; struct kmem_list3 *l3; int r; l3 = cachep->nodelists[q]; if (!l3) return; lockdep_set_class(&l3->list_lock, l3_key); alc = l3->alien; /* * FIXME: This check for BAD_ALIEN_MAGIC * should go away when common slab code is taught to * work even without alien caches. * Currently, non NUMA code returns BAD_ALIEN_MAGIC * for alloc_alien_cache, */ if (!alc || (unsigned long)alc == BAD_ALIEN_MAGIC) return; for_each_node(r) { if (alc[r]) lockdep_set_class(&alc[r]->lock, alc_key); } } static void slab_set_debugobj_lock_classes_node(struct kmem_cache *cachep, int node) { slab_set_lock_classes(cachep, &debugobj_l3_key, &debugobj_alc_key, node); } static void slab_set_debugobj_lock_classes(struct kmem_cache *cachep) { int node; for_each_online_node(node) slab_set_debugobj_lock_classes_node(cachep, node); } |
ce79ddc8e SLAB: Fix lockdep... |
648 |
static void init_node_lock_keys(int q) |
f1aaee53f [PATCH] lockdep: ... |
649 |
{ |
056c62418 [PATCH] slab: fix... |
650 |
struct cache_sizes *s = malloc_sizes; |
52cef1891 slab, lockdep: Fi... |
651 |
if (g_cpucache_up < LATE) |
ce79ddc8e SLAB: Fix lockdep... |
652 653 654 |
return; for (s = malloc_sizes; s->cs_size != ULONG_MAX; s++) { |
ce79ddc8e SLAB: Fix lockdep... |
655 |
struct kmem_list3 *l3; |
ce79ddc8e SLAB: Fix lockdep... |
656 657 658 |
l3 = s->cs_cachep->nodelists[q]; if (!l3 || OFF_SLAB(s->cs_cachep)) |
00afa7580 SLAB: Fix lockdep... |
659 |
continue; |
83835b3d9 slab, lockdep: An... |
660 661 662 |
slab_set_lock_classes(s->cs_cachep, &on_slab_l3_key, &on_slab_alc_key, q); |
f1aaee53f [PATCH] lockdep: ... |
663 664 |
} } |
ce79ddc8e SLAB: Fix lockdep... |
665 666 667 668 669 670 671 672 |
static inline void init_lock_keys(void) { int node; for_each_node(node) init_node_lock_keys(node); } |
f1aaee53f [PATCH] lockdep: ... |
673 |
#else |
ce79ddc8e SLAB: Fix lockdep... |
674 675 676 |
static void init_node_lock_keys(int q) { } |
056c62418 [PATCH] slab: fix... |
677 |
static inline void init_lock_keys(void) |
f1aaee53f [PATCH] lockdep: ... |
678 679 |
{ } |
83835b3d9 slab, lockdep: An... |
680 681 682 683 684 685 686 687 |
static void slab_set_debugobj_lock_classes_node(struct kmem_cache *cachep, int node) { } static void slab_set_debugobj_lock_classes(struct kmem_cache *cachep) { } |
f1aaee53f [PATCH] lockdep: ... |
688 |
#endif |
8f5be20bf [PATCH] mm: slab:... |
689 |
/* |
95402b382 cpu-hotplug: repl... |
690 |
* Guard access to the cache-chain. |
8f5be20bf [PATCH] mm: slab:... |
691 |
*/ |
fc0abb145 [PATCH] sem2mutex... |
692 |
static DEFINE_MUTEX(cache_chain_mutex); |
1da177e4c Linux-2.6.12-rc2 |
693 |
static struct list_head cache_chain; |
1871e52c7 percpu: make perc... |
694 |
static DEFINE_PER_CPU(struct delayed_work, slab_reap_work); |
1da177e4c Linux-2.6.12-rc2 |
695 |
|
343e0d7a9 [PATCH] slab: rep... |
696 |
static inline struct array_cache *cpu_cache_get(struct kmem_cache *cachep) |
1da177e4c Linux-2.6.12-rc2 |
697 698 699 |
{ return cachep->array[smp_processor_id()]; } |
a737b3e2f [PATCH] slab cleanup |
700 701 |
static inline struct kmem_cache *__find_general_cachep(size_t size, gfp_t gfpflags) |
1da177e4c Linux-2.6.12-rc2 |
702 703 704 705 706 |
{ struct cache_sizes *csizep = malloc_sizes; #if DEBUG /* This happens if someone tries to call |
b28a02de8 [PATCH] slab: fix... |
707 708 709 |
* kmem_cache_create(), or __kmalloc(), before * the generic caches are initialized. */ |
c7e43c78a [PATCH] Fix slab ... |
710 |
BUG_ON(malloc_sizes[INDEX_AC].cs_cachep == NULL); |
1da177e4c Linux-2.6.12-rc2 |
711 |
#endif |
6cb8f9132 Slab allocators: ... |
712 713 |
if (!size) return ZERO_SIZE_PTR; |
1da177e4c Linux-2.6.12-rc2 |
714 715 716 717 |
while (size > csizep->cs_size) csizep++; /* |
0abf40c1a [PATCH] vm: slab.... |
718 |
* Really subtle: The last entry with cs->cs_size==ULONG_MAX |
1da177e4c Linux-2.6.12-rc2 |
719 720 721 |
* has cs_{dma,}cachep==NULL. Thus no special case * for large kmalloc calls required. */ |
4b51d6698 [PATCH] optional ... |
722 |
#ifdef CONFIG_ZONE_DMA |
1da177e4c Linux-2.6.12-rc2 |
723 724 |
if (unlikely(gfpflags & GFP_DMA)) return csizep->cs_dmacachep; |
4b51d6698 [PATCH] optional ... |
725 |
#endif |
1da177e4c Linux-2.6.12-rc2 |
726 727 |
return csizep->cs_cachep; } |
b221385bc [PATCH] mm/: make... |
728 |
static struct kmem_cache *kmem_find_general_cachep(size_t size, gfp_t gfpflags) |
97e2bde47 [PATCH] add kmall... |
729 730 731 |
{ return __find_general_cachep(size, gfpflags); } |
97e2bde47 [PATCH] add kmall... |
732 |
|
fbaccacff [PATCH] slab: cac... |
733 |
static size_t slab_mgmt_size(size_t nr_objs, size_t align) |
1da177e4c Linux-2.6.12-rc2 |
734 |
{ |
fbaccacff [PATCH] slab: cac... |
735 736 |
return ALIGN(sizeof(struct slab)+nr_objs*sizeof(kmem_bufctl_t), align); } |
1da177e4c Linux-2.6.12-rc2 |
737 |
|
a737b3e2f [PATCH] slab cleanup |
738 739 740 |
/* * Calculate the number of objects and left-over bytes for a given buffer size. */ |
fbaccacff [PATCH] slab: cac... |
741 742 743 744 745 746 747 |
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 Linux-2.6.12-rc2 |
748 |
|
fbaccacff [PATCH] slab: cac... |
749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 |
/* * 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 Linux-2.6.12-rc2 |
797 |
} |
d40cee245 mm: remove remain... |
798 |
#define slab_error(cachep, msg) __slab_error(__func__, cachep, msg) |
1da177e4c Linux-2.6.12-rc2 |
799 |
|
a737b3e2f [PATCH] slab cleanup |
800 801 |
static void __slab_error(const char *function, struct kmem_cache *cachep, char *msg) |
1da177e4c Linux-2.6.12-rc2 |
802 803 804 |
{ printk(KERN_ERR "slab error in %s(): cache `%s': %s ", |
b28a02de8 [PATCH] slab: fix... |
805 |
function, cachep->name, msg); |
1da177e4c Linux-2.6.12-rc2 |
806 807 |
dump_stack(); } |
3395ee058 [PATCH] mm: add n... |
808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 |
/* * 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); |
3df1cccdf slab: introduce s... |
823 824 825 826 827 828 829 830 831 832 |
static int __init slab_max_order_setup(char *str) { get_option(&str, &slab_max_order); slab_max_order = slab_max_order < 0 ? 0 : min(slab_max_order, MAX_ORDER - 1); slab_max_order_set = true; return 1; } __setup("slab_max_order=", slab_max_order_setup); |
8fce4d8e3 [PATCH] slab: Nod... |
833 834 835 836 837 838 839 |
#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 percpu: make perc... |
840 |
static DEFINE_PER_CPU(unsigned long, slab_reap_node); |
8fce4d8e3 [PATCH] slab: Nod... |
841 842 843 844 |
static void init_reap_node(int cpu) { int node; |
7d6e6d09d numa: slab: use n... |
845 |
node = next_node(cpu_to_mem(cpu), node_online_map); |
8fce4d8e3 [PATCH] slab: Nod... |
846 |
if (node == MAX_NUMNODES) |
442295c94 [PATCH] mm: slab ... |
847 |
node = first_node(node_online_map); |
8fce4d8e3 [PATCH] slab: Nod... |
848 |
|
1871e52c7 percpu: make perc... |
849 |
per_cpu(slab_reap_node, cpu) = node; |
8fce4d8e3 [PATCH] slab: Nod... |
850 851 852 853 |
} static void next_reap_node(void) { |
909ea9646 core: Replace __g... |
854 |
int node = __this_cpu_read(slab_reap_node); |
8fce4d8e3 [PATCH] slab: Nod... |
855 |
|
8fce4d8e3 [PATCH] slab: Nod... |
856 857 858 |
node = next_node(node, node_online_map); if (unlikely(node >= MAX_NUMNODES)) node = first_node(node_online_map); |
909ea9646 core: Replace __g... |
859 |
__this_cpu_write(slab_reap_node, node); |
8fce4d8e3 [PATCH] slab: Nod... |
860 861 862 863 864 865 |
} #else #define init_reap_node(cpu) do { } while (0) #define next_reap_node(void) do { } while (0) #endif |
1da177e4c Linux-2.6.12-rc2 |
866 867 868 869 870 871 872 |
/* * 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 mm/slab.c: start_... |
873 |
static void __cpuinit start_cpu_timer(int cpu) |
1da177e4c Linux-2.6.12-rc2 |
874 |
{ |
1871e52c7 percpu: make perc... |
875 |
struct delayed_work *reap_work = &per_cpu(slab_reap_work, cpu); |
1da177e4c Linux-2.6.12-rc2 |
876 877 878 879 880 881 |
/* * When this gets called from do_initcalls via cpucache_init(), * init_workqueues() has already run, so keventd will be setup * at that time. */ |
52bad64d9 WorkStruct: Separ... |
882 |
if (keventd_up() && reap_work->work.func == NULL) { |
8fce4d8e3 [PATCH] slab: Nod... |
883 |
init_reap_node(cpu); |
78b435368 slab: use deferab... |
884 |
INIT_DELAYED_WORK_DEFERRABLE(reap_work, cache_reap); |
2b2842146 [PATCH] user of t... |
885 886 |
schedule_delayed_work_on(cpu, reap_work, __round_jiffies_relative(HZ, cpu)); |
1da177e4c Linux-2.6.12-rc2 |
887 888 |
} } |
e498be7da [PATCH] Numa-awar... |
889 |
static struct array_cache *alloc_arraycache(int node, int entries, |
83b519e8b slab: setup alloc... |
890 |
int batchcount, gfp_t gfp) |
1da177e4c Linux-2.6.12-rc2 |
891 |
{ |
b28a02de8 [PATCH] slab: fix... |
892 |
int memsize = sizeof(void *) * entries + sizeof(struct array_cache); |
1da177e4c Linux-2.6.12-rc2 |
893 |
struct array_cache *nc = NULL; |
83b519e8b slab: setup alloc... |
894 |
nc = kmalloc_node(memsize, gfp, node); |
d5cff6352 kmemleak: Add the... |
895 896 |
/* * The array_cache structures contain pointers to free object. |
25985edce Fix common misspe... |
897 |
* However, when such objects are allocated or transferred to another |
d5cff6352 kmemleak: Add the... |
898 899 900 901 902 |
* 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 Linux-2.6.12-rc2 |
903 904 905 906 907 |
if (nc) { nc->avail = 0; nc->limit = entries; nc->batchcount = batchcount; nc->touched = 0; |
e498be7da [PATCH] Numa-awar... |
908 |
spin_lock_init(&nc->lock); |
1da177e4c Linux-2.6.12-rc2 |
909 910 911 |
} return nc; } |
3ded175a4 [PATCH] slab: add... |
912 913 914 915 916 917 918 919 920 921 |
/* * Transfer objects in one arraycache to another. * Locking must be handled by the caller. * * Return the number of entries transferred. */ static int transfer_objects(struct array_cache *to, struct array_cache *from, unsigned int max) { /* Figure out how many entries to transfer */ |
732eacc05 replace nested ma... |
922 |
int nr = min3(from->avail, max, to->limit - to->avail); |
3ded175a4 [PATCH] slab: add... |
923 924 925 926 927 928 929 930 931 |
if (!nr) return 0; memcpy(to->entry + to->avail, from->entry + from->avail -nr, sizeof(void *) *nr); from->avail -= nr; to->avail += nr; |
3ded175a4 [PATCH] slab: add... |
932 933 |
return nr; } |
765c4507a [PATCH] GFP_THISN... |
934 935 936 937 |
#ifndef CONFIG_NUMA #define drain_alien_cache(cachep, alien) do { } while (0) #define reap_alien(cachep, l3) do { } while (0) |
83b519e8b slab: setup alloc... |
938 |
static inline struct array_cache **alloc_alien_cache(int node, int limit, gfp_t gfp) |
765c4507a [PATCH] GFP_THISN... |
939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 |
{ 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 [PATCH] leak trac... |
957 |
static inline void *____cache_alloc_node(struct kmem_cache *cachep, |
765c4507a [PATCH] GFP_THISN... |
958 959 960 961 962 963 |
gfp_t flags, int nodeid) { return NULL; } #else /* CONFIG_NUMA */ |
8b98c1699 [PATCH] leak trac... |
964 |
static void *____cache_alloc_node(struct kmem_cache *, gfp_t, int); |
c61afb181 [PATCH] cpuset me... |
965 |
static void *alternate_node_alloc(struct kmem_cache *, gfp_t); |
dc85da15d [PATCH] NUMA poli... |
966 |
|
83b519e8b slab: setup alloc... |
967 |
static struct array_cache **alloc_alien_cache(int node, int limit, gfp_t gfp) |
e498be7da [PATCH] Numa-awar... |
968 969 |
{ struct array_cache **ac_ptr; |
8ef828668 [PATCH] slab: red... |
970 |
int memsize = sizeof(void *) * nr_node_ids; |
e498be7da [PATCH] Numa-awar... |
971 972 973 974 |
int i; if (limit > 1) limit = 12; |
f3186a9c5 slab: initialize ... |
975 |
ac_ptr = kzalloc_node(memsize, gfp, node); |
e498be7da [PATCH] Numa-awar... |
976 977 |
if (ac_ptr) { for_each_node(i) { |
f3186a9c5 slab: initialize ... |
978 |
if (i == node || !node_online(i)) |
e498be7da [PATCH] Numa-awar... |
979 |
continue; |
83b519e8b slab: setup alloc... |
980 |
ac_ptr[i] = alloc_arraycache(node, limit, 0xbaadf00d, gfp); |
e498be7da [PATCH] Numa-awar... |
981 |
if (!ac_ptr[i]) { |
cc550defe slab: fix typo in... |
982 |
for (i--; i >= 0; i--) |
e498be7da [PATCH] Numa-awar... |
983 984 985 986 987 988 989 990 |
kfree(ac_ptr[i]); kfree(ac_ptr); return NULL; } } } return ac_ptr; } |
5295a74cc [PATCH] slab: red... |
991 |
static void free_alien_cache(struct array_cache **ac_ptr) |
e498be7da [PATCH] Numa-awar... |
992 993 994 995 996 |
{ int i; if (!ac_ptr) return; |
e498be7da [PATCH] Numa-awar... |
997 |
for_each_node(i) |
b28a02de8 [PATCH] slab: fix... |
998 |
kfree(ac_ptr[i]); |
e498be7da [PATCH] Numa-awar... |
999 1000 |
kfree(ac_ptr); } |
343e0d7a9 [PATCH] slab: rep... |
1001 |
static void __drain_alien_cache(struct kmem_cache *cachep, |
5295a74cc [PATCH] slab: red... |
1002 |
struct array_cache *ac, int node) |
e498be7da [PATCH] Numa-awar... |
1003 1004 1005 1006 1007 |
{ struct kmem_list3 *rl3 = cachep->nodelists[node]; if (ac->avail) { spin_lock(&rl3->list_lock); |
e00946fe2 [PATCH] slab: Byp... |
1008 1009 1010 1011 1012 |
/* * 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 [PATCH] slab: fix... |
1013 1014 |
if (rl3->shared) transfer_objects(rl3->shared, ac, ac->limit); |
e00946fe2 [PATCH] slab: Byp... |
1015 |
|
ff69416e6 [PATCH] slab: fix... |
1016 |
free_block(cachep, ac->entry, ac->avail, node); |
e498be7da [PATCH] Numa-awar... |
1017 1018 1019 1020 |
ac->avail = 0; spin_unlock(&rl3->list_lock); } } |
8fce4d8e3 [PATCH] slab: Nod... |
1021 1022 1023 1024 1025 |
/* * Called from cache_reap() to regularly drain alien caches round robin. */ static void reap_alien(struct kmem_cache *cachep, struct kmem_list3 *l3) { |
909ea9646 core: Replace __g... |
1026 |
int node = __this_cpu_read(slab_reap_node); |
8fce4d8e3 [PATCH] slab: Nod... |
1027 1028 1029 |
if (l3->alien) { struct array_cache *ac = l3->alien[node]; |
e00946fe2 [PATCH] slab: Byp... |
1030 1031 |
if (ac && ac->avail && spin_trylock_irq(&ac->lock)) { |
8fce4d8e3 [PATCH] slab: Nod... |
1032 1033 1034 1035 1036 |
__drain_alien_cache(cachep, ac, node); spin_unlock_irq(&ac->lock); } } } |
a737b3e2f [PATCH] slab cleanup |
1037 1038 |
static void drain_alien_cache(struct kmem_cache *cachep, struct array_cache **alien) |
e498be7da [PATCH] Numa-awar... |
1039 |
{ |
b28a02de8 [PATCH] slab: fix... |
1040 |
int i = 0; |
e498be7da [PATCH] Numa-awar... |
1041 1042 1043 1044 |
struct array_cache *ac; unsigned long flags; for_each_online_node(i) { |
4484ebf12 [PATCH] NUMA slab... |
1045 |
ac = alien[i]; |
e498be7da [PATCH] Numa-awar... |
1046 1047 1048 1049 1050 1051 1052 |
if (ac) { spin_lock_irqsave(&ac->lock, flags); __drain_alien_cache(cachep, ac, i); spin_unlock_irqrestore(&ac->lock, flags); } } } |
729bd0b74 [PATCH] slab: ext... |
1053 |
|
873623dfa [PATCH] lockdep: ... |
1054 |
static inline int cache_free_alien(struct kmem_cache *cachep, void *objp) |
729bd0b74 [PATCH] slab: ext... |
1055 1056 1057 1058 1059 |
{ struct slab *slabp = virt_to_slab(objp); int nodeid = slabp->nodeid; struct kmem_list3 *l3; struct array_cache *alien = NULL; |
1ca4cb241 [PATCH] slab: red... |
1060 |
int node; |
7d6e6d09d numa: slab: use n... |
1061 |
node = numa_mem_id(); |
729bd0b74 [PATCH] slab: ext... |
1062 1063 1064 1065 1066 |
/* * Make sure we are not freeing a object from another node to the array * cache on this cpu. */ |
62918a036 [PATCH] x86-64: s... |
1067 |
if (likely(slabp->nodeid == node)) |
729bd0b74 [PATCH] slab: ext... |
1068 |
return 0; |
1ca4cb241 [PATCH] slab: red... |
1069 |
l3 = cachep->nodelists[node]; |
729bd0b74 [PATCH] slab: ext... |
1070 1071 1072 |
STATS_INC_NODEFREES(cachep); if (l3->alien && l3->alien[nodeid]) { alien = l3->alien[nodeid]; |
873623dfa [PATCH] lockdep: ... |
1073 |
spin_lock(&alien->lock); |
729bd0b74 [PATCH] slab: ext... |
1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 |
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 [PATCH] Numa-awar... |
1087 |
#endif |
8f9f8d9e8 slab: add memory ... |
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 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 |
/* * Allocates and initializes nodelists for a node on each slab cache, used for * either memory or cpu hotplug. If memory is being hot-added, the kmem_list3 * will be allocated off-node since memory is not yet online for the new node. * When hotplugging memory or a cpu, existing nodelists are not replaced if * already in use. * * Must hold cache_chain_mutex. */ static int init_cache_nodelists_node(int node) { struct kmem_cache *cachep; struct kmem_list3 *l3; const int memsize = sizeof(struct kmem_list3); list_for_each_entry(cachep, &cache_chain, next) { /* * Set up the size64 kmemlist for cpu before we can * begin anything. Make sure some other cpu on this * node has not already allocated this */ if (!cachep->nodelists[node]) { l3 = kmalloc_node(memsize, GFP_KERNEL, node); if (!l3) return -ENOMEM; kmem_list3_init(l3); l3->next_reap = jiffies + REAPTIMEOUT_LIST3 + ((unsigned long)cachep) % REAPTIMEOUT_LIST3; /* * The l3s don't come and go as CPUs come and * go. cache_chain_mutex is sufficient * protection here. */ cachep->nodelists[node] = l3; } spin_lock_irq(&cachep->nodelists[node]->list_lock); cachep->nodelists[node]->free_limit = (1 + nr_cpus_node(node)) * cachep->batchcount + cachep->num; spin_unlock_irq(&cachep->nodelists[node]->list_lock); } return 0; } |
fbf1e473b cpu hotplug: slab... |
1133 1134 1135 1136 |
static void __cpuinit cpuup_canceled(long cpu) { struct kmem_cache *cachep; struct kmem_list3 *l3 = NULL; |
7d6e6d09d numa: slab: use n... |
1137 |
int node = cpu_to_mem(cpu); |
a70f73028 cpumask: replace ... |
1138 |
const struct cpumask *mask = cpumask_of_node(node); |
fbf1e473b cpu hotplug: slab... |
1139 1140 1141 1142 1143 |
list_for_each_entry(cachep, &cache_chain, next) { struct array_cache *nc; struct array_cache *shared; struct array_cache **alien; |
fbf1e473b cpu hotplug: slab... |
1144 |
|
fbf1e473b cpu hotplug: slab... |
1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 |
/* 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 cpumask: avoid de... |
1159 |
if (!cpumask_empty(mask)) { |
fbf1e473b cpu hotplug: slab... |
1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 |
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 Linux-2.6.12-rc2 |
1198 |
{ |
343e0d7a9 [PATCH] slab: rep... |
1199 |
struct kmem_cache *cachep; |
e498be7da [PATCH] Numa-awar... |
1200 |
struct kmem_list3 *l3 = NULL; |
7d6e6d09d numa: slab: use n... |
1201 |
int node = cpu_to_mem(cpu); |
8f9f8d9e8 slab: add memory ... |
1202 |
int err; |
1da177e4c Linux-2.6.12-rc2 |
1203 |
|
fbf1e473b cpu hotplug: slab... |
1204 1205 1206 1207 1208 1209 |
/* * We need to do this right in the beginning since * alloc_arraycache's are going to use this list. * kmalloc_node allows us to add the slab to the right * kmem_list3 and not this cpu's kmem_list3 */ |
8f9f8d9e8 slab: add memory ... |
1210 1211 1212 |
err = init_cache_nodelists_node(node); if (err < 0) goto bad; |
fbf1e473b cpu hotplug: slab... |
1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 |
/* * 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 slab: setup alloc... |
1224 |
cachep->batchcount, GFP_KERNEL); |
fbf1e473b cpu hotplug: slab... |
1225 1226 1227 1228 1229 |
if (!nc) goto bad; if (cachep->shared) { shared = alloc_arraycache(node, cachep->shared * cachep->batchcount, |
83b519e8b slab: setup alloc... |
1230 |
0xbaadf00d, GFP_KERNEL); |
12d00f6a1 cpu hotplug: slab... |
1231 1232 |
if (!shared) { kfree(nc); |
1da177e4c Linux-2.6.12-rc2 |
1233 |
goto bad; |
12d00f6a1 cpu hotplug: slab... |
1234 |
} |
fbf1e473b cpu hotplug: slab... |
1235 1236 |
} if (use_alien_caches) { |
83b519e8b slab: setup alloc... |
1237 |
alien = alloc_alien_cache(node, cachep->limit, GFP_KERNEL); |
12d00f6a1 cpu hotplug: slab... |
1238 1239 1240 |
if (!alien) { kfree(shared); kfree(nc); |
fbf1e473b cpu hotplug: slab... |
1241 |
goto bad; |
12d00f6a1 cpu hotplug: slab... |
1242 |
} |
fbf1e473b cpu hotplug: slab... |
1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 |
} 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 [PATCH] NUMA slab... |
1257 |
#ifdef CONFIG_NUMA |
fbf1e473b cpu hotplug: slab... |
1258 1259 1260 |
if (!l3->alien) { l3->alien = alien; alien = NULL; |
1da177e4c Linux-2.6.12-rc2 |
1261 |
} |
fbf1e473b cpu hotplug: slab... |
1262 1263 1264 1265 |
#endif spin_unlock_irq(&l3->list_lock); kfree(shared); free_alien_cache(alien); |
83835b3d9 slab, lockdep: An... |
1266 1267 |
if (cachep->flags & SLAB_DEBUG_OBJECTS) slab_set_debugobj_lock_classes_node(cachep, node); |
fbf1e473b cpu hotplug: slab... |
1268 |
} |
ce79ddc8e SLAB: Fix lockdep... |
1269 |
init_node_lock_keys(node); |
fbf1e473b cpu hotplug: slab... |
1270 1271 |
return 0; bad: |
12d00f6a1 cpu hotplug: slab... |
1272 |
cpuup_canceled(cpu); |
fbf1e473b cpu hotplug: slab... |
1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 |
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 cpu hotplug: slab... |
1283 1284 |
case CPU_UP_PREPARE: case CPU_UP_PREPARE_FROZEN: |
95402b382 cpu-hotplug: repl... |
1285 |
mutex_lock(&cache_chain_mutex); |
fbf1e473b cpu hotplug: slab... |
1286 |
err = cpuup_prepare(cpu); |
95402b382 cpu-hotplug: repl... |
1287 |
mutex_unlock(&cache_chain_mutex); |
1da177e4c Linux-2.6.12-rc2 |
1288 1289 |
break; case CPU_ONLINE: |
8bb784428 Add suspend-relat... |
1290 |
case CPU_ONLINE_FROZEN: |
1da177e4c Linux-2.6.12-rc2 |
1291 1292 1293 |
start_cpu_timer(cpu); break; #ifdef CONFIG_HOTPLUG_CPU |
5830c5902 slab: shut down c... |
1294 |
case CPU_DOWN_PREPARE: |
8bb784428 Add suspend-relat... |
1295 |
case CPU_DOWN_PREPARE_FROZEN: |
5830c5902 slab: shut down c... |
1296 1297 1298 1299 1300 1301 |
/* * Shutdown cache reaper. Note that the cache_chain_mutex is * held so that if cache_reap() is invoked it cannot do * anything expensive but will only modify reap_work * and reschedule the timer. */ |
afe2c511f workqueue: conver... |
1302 |
cancel_delayed_work_sync(&per_cpu(slab_reap_work, cpu)); |
5830c5902 slab: shut down c... |
1303 |
/* Now the cache_reaper is guaranteed to be not running. */ |
1871e52c7 percpu: make perc... |
1304 |
per_cpu(slab_reap_work, cpu).work.func = NULL; |
5830c5902 slab: shut down c... |
1305 1306 |
break; case CPU_DOWN_FAILED: |
8bb784428 Add suspend-relat... |
1307 |
case CPU_DOWN_FAILED_FROZEN: |
5830c5902 slab: shut down c... |
1308 1309 |
start_cpu_timer(cpu); break; |
1da177e4c Linux-2.6.12-rc2 |
1310 |
case CPU_DEAD: |
8bb784428 Add suspend-relat... |
1311 |
case CPU_DEAD_FROZEN: |
4484ebf12 [PATCH] NUMA slab... |
1312 1313 1314 1315 1316 1317 1318 1319 |
/* * 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 spelling fixes: mm/ |
1320 |
/* fall through */ |
8f5be20bf [PATCH] mm: slab:... |
1321 |
#endif |
1da177e4c Linux-2.6.12-rc2 |
1322 |
case CPU_UP_CANCELED: |
8bb784428 Add suspend-relat... |
1323 |
case CPU_UP_CANCELED_FROZEN: |
95402b382 cpu-hotplug: repl... |
1324 |
mutex_lock(&cache_chain_mutex); |
fbf1e473b cpu hotplug: slab... |
1325 |
cpuup_canceled(cpu); |
fc0abb145 [PATCH] sem2mutex... |
1326 |
mutex_unlock(&cache_chain_mutex); |
1da177e4c Linux-2.6.12-rc2 |
1327 |
break; |
1da177e4c Linux-2.6.12-rc2 |
1328 |
} |
eac406801 slab: convert cpu... |
1329 |
return notifier_from_errno(err); |
1da177e4c Linux-2.6.12-rc2 |
1330 |
} |
74b85f379 [PATCH] cpu hotpl... |
1331 1332 1333 |
static struct notifier_block __cpuinitdata cpucache_notifier = { &cpuup_callback, NULL, 0 }; |
1da177e4c Linux-2.6.12-rc2 |
1334 |
|
8f9f8d9e8 slab: add memory ... |
1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 |
#if defined(CONFIG_NUMA) && defined(CONFIG_MEMORY_HOTPLUG) /* * Drains freelist for a node on each slab cache, used for memory hot-remove. * Returns -EBUSY if all objects cannot be drained so that the node is not * removed. * * Must hold cache_chain_mutex. */ static int __meminit drain_cache_nodelists_node(int node) { struct kmem_cache *cachep; int ret = 0; list_for_each_entry(cachep, &cache_chain, next) { struct kmem_list3 *l3; l3 = cachep->nodelists[node]; if (!l3) continue; drain_freelist(cachep, l3, l3->free_objects); if (!list_empty(&l3->slabs_full) || !list_empty(&l3->slabs_partial)) { ret = -EBUSY; break; } } return ret; } static int __meminit slab_memory_callback(struct notifier_block *self, unsigned long action, void *arg) { struct memory_notify *mnb = arg; int ret = 0; int nid; nid = mnb->status_change_nid; if (nid < 0) goto out; switch (action) { case MEM_GOING_ONLINE: mutex_lock(&cache_chain_mutex); ret = init_cache_nodelists_node(nid); mutex_unlock(&cache_chain_mutex); break; case MEM_GOING_OFFLINE: mutex_lock(&cache_chain_mutex); ret = drain_cache_nodelists_node(nid); mutex_unlock(&cache_chain_mutex); break; case MEM_ONLINE: case MEM_OFFLINE: case MEM_CANCEL_ONLINE: case MEM_CANCEL_OFFLINE: break; } out: |
5fda1bd5b mm: notifier_from... |
1395 |
return notifier_from_errno(ret); |
8f9f8d9e8 slab: add memory ... |
1396 1397 |
} #endif /* CONFIG_NUMA && CONFIG_MEMORY_HOTPLUG */ |
e498be7da [PATCH] Numa-awar... |
1398 1399 1400 |
/* * swap the static kmem_list3 with kmalloced memory */ |
8f9f8d9e8 slab: add memory ... |
1401 1402 |
static void __init init_list(struct kmem_cache *cachep, struct kmem_list3 *list, int nodeid) |
e498be7da [PATCH] Numa-awar... |
1403 1404 |
{ struct kmem_list3 *ptr; |
83b519e8b slab: setup alloc... |
1405 |
ptr = kmalloc_node(sizeof(struct kmem_list3), GFP_NOWAIT, nodeid); |
e498be7da [PATCH] Numa-awar... |
1406 |
BUG_ON(!ptr); |
e498be7da [PATCH] Numa-awar... |
1407 |
memcpy(ptr, list, sizeof(struct kmem_list3)); |
2b2d5493e [PATCH] lockdep: ... |
1408 1409 1410 1411 |
/* * Do not assume that spinlocks can be initialized via memcpy: */ spin_lock_init(&ptr->list_lock); |
e498be7da [PATCH] Numa-awar... |
1412 1413 |
MAKE_ALL_LISTS(cachep, ptr, nodeid); cachep->nodelists[nodeid] = ptr; |
e498be7da [PATCH] Numa-awar... |
1414 |
} |
a737b3e2f [PATCH] slab cleanup |
1415 |
/* |
556a169da slab: fix bootstr... |
1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 |
* 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 [PATCH] slab cleanup |
1432 1433 |
* Initialisation. Called after the page allocator have been initialised and * before smp_init(). |
1da177e4c Linux-2.6.12-rc2 |
1434 1435 1436 1437 1438 1439 |
*/ void __init kmem_cache_init(void) { size_t left_over; struct cache_sizes *sizes; struct cache_names *names; |
e498be7da [PATCH] Numa-awar... |
1440 |
int i; |
07ed76b2a [PATCH] slab: all... |
1441 |
int order; |
1ca4cb241 [PATCH] slab: red... |
1442 |
int node; |
e498be7da [PATCH] Numa-awar... |
1443 |
|
b6e68bc1b page allocator: s... |
1444 |
if (num_possible_nodes() == 1) |
62918a036 [PATCH] x86-64: s... |
1445 |
use_alien_caches = 0; |
e498be7da [PATCH] Numa-awar... |
1446 1447 1448 1449 1450 |
for (i = 0; i < NUM_INIT_LISTS; i++) { kmem_list3_init(&initkmem_list3[i]); if (i < MAX_NUMNODES) cache_cache.nodelists[i] = NULL; } |
556a169da slab: fix bootstr... |
1451 |
set_up_list3s(&cache_cache, CACHE_CACHE); |
1da177e4c Linux-2.6.12-rc2 |
1452 1453 1454 |
/* * Fragmentation resistance on low memory - only use bigger |
3df1cccdf slab: introduce s... |
1455 1456 |
* page orders on machines with more than 32MB of memory if * not overridden on the command line. |
1da177e4c Linux-2.6.12-rc2 |
1457 |
*/ |
3df1cccdf slab: introduce s... |
1458 |
if (!slab_max_order_set && totalram_pages > (32 << 20) >> PAGE_SHIFT) |
543585cc5 slab: rename slab... |
1459 |
slab_max_order = SLAB_MAX_ORDER_HI; |
1da177e4c Linux-2.6.12-rc2 |
1460 |
|
1da177e4c Linux-2.6.12-rc2 |
1461 1462 |
/* Bootstrap is tricky, because several objects are allocated * from caches that do not exist yet: |
a737b3e2f [PATCH] slab cleanup |
1463 1464 1465 |
* 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 [PATCH] Numa-awar... |
1466 1467 1468 |
* 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 Linux-2.6.12-rc2 |
1469 |
* 2) Create the first kmalloc cache. |
343e0d7a9 [PATCH] slab: rep... |
1470 |
* The struct kmem_cache for the new cache is allocated normally. |
e498be7da [PATCH] Numa-awar... |
1471 1472 1473 |
* An __init data area is used for the head array. * 3) Create the remaining kmalloc caches, with minimally sized * head arrays. |
1da177e4c Linux-2.6.12-rc2 |
1474 1475 |
* 4) Replace the __init data head arrays for cache_cache and the first * kmalloc cache with kmalloc allocated arrays. |
e498be7da [PATCH] Numa-awar... |
1476 1477 1478 |
* 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 Linux-2.6.12-rc2 |
1479 |
*/ |
7d6e6d09d numa: slab: use n... |
1480 |
node = numa_mem_id(); |
1ca4cb241 [PATCH] slab: red... |
1481 |
|
1da177e4c Linux-2.6.12-rc2 |
1482 |
/* 1) create the cache_cache */ |
1da177e4c Linux-2.6.12-rc2 |
1483 1484 1485 1486 |
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 slab: fix cache_c... |
1487 |
cache_cache.nodelists[node] = &initkmem_list3[CACHE_CACHE + node]; |
1da177e4c Linux-2.6.12-rc2 |
1488 |
|
8da3430d8 slab: NUMA kmem_c... |
1489 |
/* |
b56efcf0a slab: shrink size... |
1490 |
* struct kmem_cache size depends on nr_node_ids & nr_cpu_ids |
8da3430d8 slab: NUMA kmem_c... |
1491 |
*/ |
b56efcf0a slab: shrink size... |
1492 1493 |
cache_cache.buffer_size = offsetof(struct kmem_cache, array[nr_cpu_ids]) + nr_node_ids * sizeof(struct kmem_list3 *); |
8da3430d8 slab: NUMA kmem_c... |
1494 1495 1496 |
#if DEBUG cache_cache.obj_size = cache_cache.buffer_size; #endif |
a737b3e2f [PATCH] slab cleanup |
1497 1498 |
cache_cache.buffer_size = ALIGN(cache_cache.buffer_size, cache_line_size()); |
6a2d7a955 [PATCH] SLAB: use... |
1499 1500 |
cache_cache.reciprocal_buffer_size = reciprocal_value(cache_cache.buffer_size); |
1da177e4c Linux-2.6.12-rc2 |
1501 |
|
07ed76b2a [PATCH] slab: all... |
1502 1503 1504 1505 1506 1507 |
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 BUG_ON() Conversi... |
1508 |
BUG_ON(!cache_cache.num); |
07ed76b2a [PATCH] slab: all... |
1509 |
cache_cache.gfporder = order; |
b28a02de8 [PATCH] slab: fix... |
1510 |
cache_cache.colour = left_over / cache_cache.colour_off; |
b28a02de8 [PATCH] slab: fix... |
1511 1512 |
cache_cache.slab_size = ALIGN(cache_cache.num * sizeof(kmem_bufctl_t) + sizeof(struct slab), cache_line_size()); |
1da177e4c Linux-2.6.12-rc2 |
1513 1514 1515 1516 |
/* 2+3) create the kmalloc caches */ sizes = malloc_sizes; names = cache_names; |
a737b3e2f [PATCH] slab cleanup |
1517 1518 1519 1520 |
/* * Initialize the caches that provide memory for the array cache and the * kmem_list3 structures first. Without this, further allocations will * bug. |
e498be7da [PATCH] Numa-awar... |
1521 1522 1523 |
*/ sizes[INDEX_AC].cs_cachep = kmem_cache_create(names[INDEX_AC].name, |
a737b3e2f [PATCH] slab cleanup |
1524 1525 1526 |
sizes[INDEX_AC].cs_size, ARCH_KMALLOC_MINALIGN, ARCH_KMALLOC_FLAGS|SLAB_PANIC, |
20c2df83d mm: Remove slab d... |
1527 |
NULL); |
e498be7da [PATCH] Numa-awar... |
1528 |
|
a737b3e2f [PATCH] slab cleanup |
1529 |
if (INDEX_AC != INDEX_L3) { |
e498be7da [PATCH] Numa-awar... |
1530 |
sizes[INDEX_L3].cs_cachep = |
a737b3e2f [PATCH] slab cleanup |
1531 1532 1533 1534 |
kmem_cache_create(names[INDEX_L3].name, sizes[INDEX_L3].cs_size, ARCH_KMALLOC_MINALIGN, ARCH_KMALLOC_FLAGS|SLAB_PANIC, |
20c2df83d mm: Remove slab d... |
1535 |
NULL); |
a737b3e2f [PATCH] slab cleanup |
1536 |
} |
e498be7da [PATCH] Numa-awar... |
1537 |
|
e0a427267 [PATCH] mm/slab.c... |
1538 |
slab_early_init = 0; |
1da177e4c Linux-2.6.12-rc2 |
1539 |
while (sizes->cs_size != ULONG_MAX) { |
e498be7da [PATCH] Numa-awar... |
1540 1541 |
/* * For performance, all the general caches are L1 aligned. |
1da177e4c Linux-2.6.12-rc2 |
1542 1543 1544 |
* This should be particularly beneficial on SMP boxes, as it * eliminates "false sharing". * Note for systems short on memory removing the alignment will |
e498be7da [PATCH] Numa-awar... |
1545 1546 |
* allow tighter packing of the smaller caches. */ |
a737b3e2f [PATCH] slab cleanup |
1547 |
if (!sizes->cs_cachep) { |
e498be7da [PATCH] Numa-awar... |
1548 |
sizes->cs_cachep = kmem_cache_create(names->name, |
a737b3e2f [PATCH] slab cleanup |
1549 1550 1551 |
sizes->cs_size, ARCH_KMALLOC_MINALIGN, ARCH_KMALLOC_FLAGS|SLAB_PANIC, |
20c2df83d mm: Remove slab d... |
1552 |
NULL); |
a737b3e2f [PATCH] slab cleanup |
1553 |
} |
4b51d6698 [PATCH] optional ... |
1554 1555 1556 |
#ifdef CONFIG_ZONE_DMA sizes->cs_dmacachep = kmem_cache_create( names->name_dma, |
a737b3e2f [PATCH] slab cleanup |
1557 1558 1559 1560 |
sizes->cs_size, ARCH_KMALLOC_MINALIGN, ARCH_KMALLOC_FLAGS|SLAB_CACHE_DMA| SLAB_PANIC, |
20c2df83d mm: Remove slab d... |
1561 |
NULL); |
4b51d6698 [PATCH] optional ... |
1562 |
#endif |
1da177e4c Linux-2.6.12-rc2 |
1563 1564 1565 1566 1567 |
sizes++; names++; } /* 4) Replace the bootstrap head arrays */ { |
2b2d5493e [PATCH] lockdep: ... |
1568 |
struct array_cache *ptr; |
e498be7da [PATCH] Numa-awar... |
1569 |
|
83b519e8b slab: setup alloc... |
1570 |
ptr = kmalloc(sizeof(struct arraycache_init), GFP_NOWAIT); |
e498be7da [PATCH] Numa-awar... |
1571 |
|
9a2dba4b4 [PATCH] slab: ren... |
1572 1573 |
BUG_ON(cpu_cache_get(&cache_cache) != &initarray_cache.cache); memcpy(ptr, cpu_cache_get(&cache_cache), |
b28a02de8 [PATCH] slab: fix... |
1574 |
sizeof(struct arraycache_init)); |
2b2d5493e [PATCH] lockdep: ... |
1575 1576 1577 1578 |
/* * Do not assume that spinlocks can be initialized via memcpy: */ spin_lock_init(&ptr->lock); |
1da177e4c Linux-2.6.12-rc2 |
1579 |
cache_cache.array[smp_processor_id()] = ptr; |
e498be7da [PATCH] Numa-awar... |
1580 |
|
83b519e8b slab: setup alloc... |
1581 |
ptr = kmalloc(sizeof(struct arraycache_init), GFP_NOWAIT); |
e498be7da [PATCH] Numa-awar... |
1582 |
|
9a2dba4b4 [PATCH] slab: ren... |
1583 |
BUG_ON(cpu_cache_get(malloc_sizes[INDEX_AC].cs_cachep) |
b28a02de8 [PATCH] slab: fix... |
1584 |
!= &initarray_generic.cache); |
9a2dba4b4 [PATCH] slab: ren... |
1585 |
memcpy(ptr, cpu_cache_get(malloc_sizes[INDEX_AC].cs_cachep), |
b28a02de8 [PATCH] slab: fix... |
1586 |
sizeof(struct arraycache_init)); |
2b2d5493e [PATCH] lockdep: ... |
1587 1588 1589 1590 |
/* * Do not assume that spinlocks can be initialized via memcpy: */ spin_lock_init(&ptr->lock); |
e498be7da [PATCH] Numa-awar... |
1591 |
malloc_sizes[INDEX_AC].cs_cachep->array[smp_processor_id()] = |
b28a02de8 [PATCH] slab: fix... |
1592 |
ptr; |
1da177e4c Linux-2.6.12-rc2 |
1593 |
} |
e498be7da [PATCH] Numa-awar... |
1594 1595 |
/* 5) Replace the bootstrap kmem_list3's */ { |
1ca4cb241 [PATCH] slab: red... |
1596 |
int nid; |
9c09a95cf slab: partially r... |
1597 |
for_each_online_node(nid) { |
ec1f5eeeb slab: fix cache_c... |
1598 |
init_list(&cache_cache, &initkmem_list3[CACHE_CACHE + nid], nid); |
556a169da slab: fix bootstr... |
1599 |
|
e498be7da [PATCH] Numa-awar... |
1600 |
init_list(malloc_sizes[INDEX_AC].cs_cachep, |
1ca4cb241 [PATCH] slab: red... |
1601 |
&initkmem_list3[SIZE_AC + nid], nid); |
e498be7da [PATCH] Numa-awar... |
1602 1603 1604 |
if (INDEX_AC != INDEX_L3) { init_list(malloc_sizes[INDEX_L3].cs_cachep, |
1ca4cb241 [PATCH] slab: red... |
1605 |
&initkmem_list3[SIZE_L3 + nid], nid); |
e498be7da [PATCH] Numa-awar... |
1606 1607 1608 |
} } } |
1da177e4c Linux-2.6.12-rc2 |
1609 |
|
8429db5c6 slab: setup cpu c... |
1610 |
g_cpucache_up = EARLY; |
8429db5c6 slab: setup cpu c... |
1611 1612 1613 1614 1615 |
} void __init kmem_cache_init_late(void) { struct kmem_cache *cachep; |
52cef1891 slab, lockdep: Fi... |
1616 |
g_cpucache_up = LATE; |
30765b92a slab, lockdep: An... |
1617 1618 |
/* Annotate slab for lockdep -- annotate the malloc caches */ init_lock_keys(); |
8429db5c6 slab: setup cpu c... |
1619 1620 1621 1622 1623 1624 |
/* 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 [PATCH] slab: fix... |
1625 |
|
1da177e4c Linux-2.6.12-rc2 |
1626 1627 |
/* Done! */ g_cpucache_up = FULL; |
a737b3e2f [PATCH] slab cleanup |
1628 1629 1630 |
/* * Register a cpu startup notifier callback that initializes * cpu_cache_get for all new cpus |
1da177e4c Linux-2.6.12-rc2 |
1631 1632 |
*/ register_cpu_notifier(&cpucache_notifier); |
1da177e4c Linux-2.6.12-rc2 |
1633 |
|
8f9f8d9e8 slab: add memory ... |
1634 1635 1636 1637 1638 1639 1640 |
#ifdef CONFIG_NUMA /* * Register a memory hotplug callback that initializes and frees * nodelists. */ hotplug_memory_notifier(slab_memory_callback, SLAB_CALLBACK_PRI); #endif |
a737b3e2f [PATCH] slab cleanup |
1641 1642 1643 |
/* * The reap timers are started later, with a module init call: That part * of the kernel is not yet operational. |
1da177e4c Linux-2.6.12-rc2 |
1644 1645 1646 1647 1648 1649 |
*/ } static int __init cpucache_init(void) { int cpu; |
a737b3e2f [PATCH] slab cleanup |
1650 1651 |
/* * Register the timers that return unneeded pages to the page allocator |
1da177e4c Linux-2.6.12-rc2 |
1652 |
*/ |
e498be7da [PATCH] Numa-awar... |
1653 |
for_each_online_cpu(cpu) |
a737b3e2f [PATCH] slab cleanup |
1654 |
start_cpu_timer(cpu); |
1da177e4c Linux-2.6.12-rc2 |
1655 1656 |
return 0; } |
1da177e4c Linux-2.6.12-rc2 |
1657 1658 1659 1660 1661 1662 1663 1664 1665 |
__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 [PATCH] slab: rep... |
1666 |
static void *kmem_getpages(struct kmem_cache *cachep, gfp_t flags, int nodeid) |
1da177e4c Linux-2.6.12-rc2 |
1667 1668 |
{ struct page *page; |
e1b6aa6f1 [PATCH] slab: cle... |
1669 |
int nr_pages; |
1da177e4c Linux-2.6.12-rc2 |
1670 |
int i; |
d6fef9da1 [PATCH] nommu: us... |
1671 |
#ifndef CONFIG_MMU |
e1b6aa6f1 [PATCH] slab: cle... |
1672 1673 1674 |
/* * Nommu uses slab's for process anonymous memory allocations, and thus * requires __GFP_COMP to properly refcount higher order allocations |
d6fef9da1 [PATCH] nommu: us... |
1675 |
*/ |
e1b6aa6f1 [PATCH] slab: cle... |
1676 |
flags |= __GFP_COMP; |
d6fef9da1 [PATCH] nommu: us... |
1677 |
#endif |
765c4507a [PATCH] GFP_THISN... |
1678 |
|
3c517a613 [PATCH] slab: bet... |
1679 |
flags |= cachep->gfpflags; |
e12ba74d8 Group short-lived... |
1680 1681 |
if (cachep->flags & SLAB_RECLAIM_ACCOUNT) flags |= __GFP_RECLAIMABLE; |
e1b6aa6f1 [PATCH] slab: cle... |
1682 |
|
517d08699 Merge branch 'akpm' |
1683 |
page = alloc_pages_exact_node(nodeid, flags | __GFP_NOTRACK, cachep->gfporder); |
1da177e4c Linux-2.6.12-rc2 |
1684 1685 |
if (!page) return NULL; |
1da177e4c Linux-2.6.12-rc2 |
1686 |
|
e1b6aa6f1 [PATCH] slab: cle... |
1687 |
nr_pages = (1 << cachep->gfporder); |
1da177e4c Linux-2.6.12-rc2 |
1688 |
if (cachep->flags & SLAB_RECLAIM_ACCOUNT) |
972d1a7b1 [PATCH] ZVC: Supp... |
1689 1690 1691 1692 1693 |
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 [PATCH] slab: cle... |
1694 1695 |
for (i = 0; i < nr_pages; i++) __SetPageSlab(page + i); |
c175eea46 slab: add hooks f... |
1696 |
|
b1eeab676 kmemcheck: add ho... |
1697 1698 1699 1700 1701 1702 1703 1704 |
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 slab: add hooks f... |
1705 |
|
e1b6aa6f1 [PATCH] slab: cle... |
1706 |
return page_address(page); |
1da177e4c Linux-2.6.12-rc2 |
1707 1708 1709 1710 1711 |
} /* * Interface to system's page release. */ |
343e0d7a9 [PATCH] slab: rep... |
1712 |
static void kmem_freepages(struct kmem_cache *cachep, void *addr) |
1da177e4c Linux-2.6.12-rc2 |
1713 |
{ |
b28a02de8 [PATCH] slab: fix... |
1714 |
unsigned long i = (1 << cachep->gfporder); |
1da177e4c Linux-2.6.12-rc2 |
1715 1716 |
struct page *page = virt_to_page(addr); const unsigned long nr_freed = i; |
b1eeab676 kmemcheck: add ho... |
1717 |
kmemcheck_free_shadow(page, cachep->gfporder); |
c175eea46 slab: add hooks f... |
1718 |
|
972d1a7b1 [PATCH] ZVC: Supp... |
1719 1720 1721 1722 1723 1724 |
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 Linux-2.6.12-rc2 |
1725 |
while (i--) { |
f205b2fe6 [PATCH] mm: slab ... |
1726 1727 |
BUG_ON(!PageSlab(page)); __ClearPageSlab(page); |
1da177e4c Linux-2.6.12-rc2 |
1728 1729 |
page++; } |
1da177e4c Linux-2.6.12-rc2 |
1730 1731 1732 |
if (current->reclaim_state) current->reclaim_state->reclaimed_slab += nr_freed; free_pages((unsigned long)addr, cachep->gfporder); |
1da177e4c Linux-2.6.12-rc2 |
1733 1734 1735 1736 |
} static void kmem_rcu_free(struct rcu_head *head) { |
b28a02de8 [PATCH] slab: fix... |
1737 |
struct slab_rcu *slab_rcu = (struct slab_rcu *)head; |
343e0d7a9 [PATCH] slab: rep... |
1738 |
struct kmem_cache *cachep = slab_rcu->cachep; |
1da177e4c Linux-2.6.12-rc2 |
1739 1740 1741 1742 1743 1744 1745 1746 1747 |
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 [PATCH] slab: rep... |
1748 |
static void store_stackinfo(struct kmem_cache *cachep, unsigned long *addr, |
b28a02de8 [PATCH] slab: fix... |
1749 |
unsigned long caller) |
1da177e4c Linux-2.6.12-rc2 |
1750 |
{ |
3dafccf22 [PATCH] slab: dis... |
1751 |
int size = obj_size(cachep); |
1da177e4c Linux-2.6.12-rc2 |
1752 |
|
3dafccf22 [PATCH] slab: dis... |
1753 |
addr = (unsigned long *)&((char *)addr)[obj_offset(cachep)]; |
1da177e4c Linux-2.6.12-rc2 |
1754 |
|
b28a02de8 [PATCH] slab: fix... |
1755 |
if (size < 5 * sizeof(unsigned long)) |
1da177e4c Linux-2.6.12-rc2 |
1756 |
return; |
b28a02de8 [PATCH] slab: fix... |
1757 1758 1759 1760 |
*addr++ = 0x12345678; *addr++ = caller; *addr++ = smp_processor_id(); size -= 3 * sizeof(unsigned long); |
1da177e4c Linux-2.6.12-rc2 |
1761 1762 1763 1764 1765 1766 1767 |
{ unsigned long *sptr = &caller; unsigned long svalue; while (!kstack_end(sptr)) { svalue = *sptr++; if (kernel_text_address(svalue)) { |
b28a02de8 [PATCH] slab: fix... |
1768 |
*addr++ = svalue; |
1da177e4c Linux-2.6.12-rc2 |
1769 1770 1771 1772 1773 1774 1775 |
size -= sizeof(unsigned long); if (size <= sizeof(unsigned long)) break; } } } |
b28a02de8 [PATCH] slab: fix... |
1776 |
*addr++ = 0x87654321; |
1da177e4c Linux-2.6.12-rc2 |
1777 1778 |
} #endif |
343e0d7a9 [PATCH] slab: rep... |
1779 |
static void poison_obj(struct kmem_cache *cachep, void *addr, unsigned char val) |
1da177e4c Linux-2.6.12-rc2 |
1780 |
{ |
3dafccf22 [PATCH] slab: dis... |
1781 1782 |
int size = obj_size(cachep); addr = &((char *)addr)[obj_offset(cachep)]; |
1da177e4c Linux-2.6.12-rc2 |
1783 1784 |
memset(addr, val, size); |
b28a02de8 [PATCH] slab: fix... |
1785 |
*(unsigned char *)(addr + size - 1) = POISON_END; |
1da177e4c Linux-2.6.12-rc2 |
1786 1787 1788 1789 1790 |
} static void dump_line(char *data, int offset, int limit) { int i; |
aa83aa40e [PATCH] single bi... |
1791 1792 |
unsigned char error = 0; int bad_count = 0; |
fdde6abb3 slab: use print_h... |
1793 |
printk(KERN_ERR "%03x: ", offset); |
aa83aa40e [PATCH] single bi... |
1794 1795 1796 1797 1798 |
for (i = 0; i < limit; i++) { if (data[offset + i] != POISON_FREE) { error = data[offset + i]; bad_count++; } |
aa83aa40e [PATCH] single bi... |
1799 |
} |
fdde6abb3 slab: use print_h... |
1800 1801 |
print_hex_dump(KERN_CONT, "", 0, 16, 1, &data[offset], limit, 1); |
aa83aa40e [PATCH] single bi... |
1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 |
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 Linux-2.6.12-rc2 |
1819 1820 1821 1822 |
} #endif #if DEBUG |
343e0d7a9 [PATCH] slab: rep... |
1823 |
static void print_objinfo(struct kmem_cache *cachep, void *objp, int lines) |
1da177e4c Linux-2.6.12-rc2 |
1824 1825 1826 1827 1828 |
{ int i, size; char *realobj; if (cachep->flags & SLAB_RED_ZONE) { |
b46b8f19c Increase slab red... |
1829 1830 |
printk(KERN_ERR "Redzone: 0x%llx/0x%llx. ", |
a737b3e2f [PATCH] slab cleanup |
1831 1832 |
*dbg_redzone1(cachep, objp), *dbg_redzone2(cachep, objp)); |
1da177e4c Linux-2.6.12-rc2 |
1833 1834 1835 1836 |
} if (cachep->flags & SLAB_STORE_USER) { printk(KERN_ERR "Last user: [<%p>]", |
a737b3e2f [PATCH] slab cleanup |
1837 |
*dbg_userword(cachep, objp)); |
1da177e4c Linux-2.6.12-rc2 |
1838 |
print_symbol("(%s)", |
a737b3e2f [PATCH] slab cleanup |
1839 |
(unsigned long)*dbg_userword(cachep, objp)); |
1da177e4c Linux-2.6.12-rc2 |
1840 1841 1842 |
printk(" "); } |
3dafccf22 [PATCH] slab: dis... |
1843 1844 |
realobj = (char *)objp + obj_offset(cachep); size = obj_size(cachep); |
b28a02de8 [PATCH] slab: fix... |
1845 |
for (i = 0; i < size && lines; i += 16, lines--) { |
1da177e4c Linux-2.6.12-rc2 |
1846 1847 |
int limit; limit = 16; |
b28a02de8 [PATCH] slab: fix... |
1848 1849 |
if (i + limit > size) limit = size - i; |
1da177e4c Linux-2.6.12-rc2 |
1850 1851 1852 |
dump_line(realobj, i, limit); } } |
343e0d7a9 [PATCH] slab: rep... |
1853 |
static void check_poison_obj(struct kmem_cache *cachep, void *objp) |
1da177e4c Linux-2.6.12-rc2 |
1854 1855 1856 1857 |
{ char *realobj; int size, i; int lines = 0; |
3dafccf22 [PATCH] slab: dis... |
1858 1859 |
realobj = (char *)objp + obj_offset(cachep); size = obj_size(cachep); |
1da177e4c Linux-2.6.12-rc2 |
1860 |
|
b28a02de8 [PATCH] slab: fix... |
1861 |
for (i = 0; i < size; i++) { |
1da177e4c Linux-2.6.12-rc2 |
1862 |
char exp = POISON_FREE; |
b28a02de8 [PATCH] slab: fix... |
1863 |
if (i == size - 1) |
1da177e4c Linux-2.6.12-rc2 |
1864 1865 1866 1867 1868 1869 |
exp = POISON_END; if (realobj[i] != exp) { int limit; /* Mismatch ! */ /* Print header */ if (lines == 0) { |
b28a02de8 [PATCH] slab: fix... |
1870 |
printk(KERN_ERR |
face37f5e slab: add taint f... |
1871 1872 1873 |
"Slab corruption (%s): %s start=%p, len=%d ", print_tainted(), cachep->name, realobj, size); |
1da177e4c Linux-2.6.12-rc2 |
1874 1875 1876 |
print_objinfo(cachep, objp, 0); } /* Hexdump the affected line */ |
b28a02de8 [PATCH] slab: fix... |
1877 |
i = (i / 16) * 16; |
1da177e4c Linux-2.6.12-rc2 |
1878 |
limit = 16; |
b28a02de8 [PATCH] slab: fix... |
1879 1880 |
if (i + limit > size) limit = size - i; |
1da177e4c Linux-2.6.12-rc2 |
1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 |
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 [PATCH] slab: ext... |
1893 |
struct slab *slabp = virt_to_slab(objp); |
8fea4e96a [PATCH] slab: obj... |
1894 |
unsigned int objnr; |
1da177e4c Linux-2.6.12-rc2 |
1895 |
|
8fea4e96a [PATCH] slab: obj... |
1896 |
objnr = obj_to_index(cachep, slabp, objp); |
1da177e4c Linux-2.6.12-rc2 |
1897 |
if (objnr) { |
8fea4e96a [PATCH] slab: obj... |
1898 |
objp = index_to_obj(cachep, slabp, objnr - 1); |
3dafccf22 [PATCH] slab: dis... |
1899 |
realobj = (char *)objp + obj_offset(cachep); |
1da177e4c Linux-2.6.12-rc2 |
1900 1901 |
printk(KERN_ERR "Prev obj: start=%p, len=%d ", |
b28a02de8 [PATCH] slab: fix... |
1902 |
realobj, size); |
1da177e4c Linux-2.6.12-rc2 |
1903 1904 |
print_objinfo(cachep, objp, 2); } |
b28a02de8 [PATCH] slab: fix... |
1905 |
if (objnr + 1 < cachep->num) { |
8fea4e96a [PATCH] slab: obj... |
1906 |
objp = index_to_obj(cachep, slabp, objnr + 1); |
3dafccf22 [PATCH] slab: dis... |
1907 |
realobj = (char *)objp + obj_offset(cachep); |
1da177e4c Linux-2.6.12-rc2 |
1908 1909 |
printk(KERN_ERR "Next obj: start=%p, len=%d ", |
b28a02de8 [PATCH] slab: fix... |
1910 |
realobj, size); |
1da177e4c Linux-2.6.12-rc2 |
1911 1912 1913 1914 1915 |
print_objinfo(cachep, objp, 2); } } } #endif |
12dd36fae [PATCH] slab: ext... |
1916 |
#if DEBUG |
e79aec291 slab: rename slab... |
1917 |
static void slab_destroy_debugcheck(struct kmem_cache *cachep, struct slab *slabp) |
1da177e4c Linux-2.6.12-rc2 |
1918 |
{ |
1da177e4c Linux-2.6.12-rc2 |
1919 1920 |
int i; for (i = 0; i < cachep->num; i++) { |
8fea4e96a [PATCH] slab: obj... |
1921 |
void *objp = index_to_obj(cachep, slabp, i); |
1da177e4c Linux-2.6.12-rc2 |
1922 1923 1924 |
if (cachep->flags & SLAB_POISON) { #ifdef CONFIG_DEBUG_PAGEALLOC |
a737b3e2f [PATCH] slab cleanup |
1925 1926 |
if (cachep->buffer_size % PAGE_SIZE == 0 && OFF_SLAB(cachep)) |
b28a02de8 [PATCH] slab: fix... |
1927 |
kernel_map_pages(virt_to_page(objp), |
a737b3e2f [PATCH] slab cleanup |
1928 |
cachep->buffer_size / PAGE_SIZE, 1); |
1da177e4c Linux-2.6.12-rc2 |
1929 1930 1931 1932 1933 1934 1935 1936 1937 |
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 [PATCH] slab: fix... |
1938 |
"was overwritten"); |
1da177e4c Linux-2.6.12-rc2 |
1939 1940 |
if (*dbg_redzone2(cachep, objp) != RED_INACTIVE) slab_error(cachep, "end of a freed object " |
b28a02de8 [PATCH] slab: fix... |
1941 |
"was overwritten"); |
1da177e4c Linux-2.6.12-rc2 |
1942 |
} |
1da177e4c Linux-2.6.12-rc2 |
1943 |
} |
12dd36fae [PATCH] slab: ext... |
1944 |
} |
1da177e4c Linux-2.6.12-rc2 |
1945 |
#else |
e79aec291 slab: rename slab... |
1946 |
static void slab_destroy_debugcheck(struct kmem_cache *cachep, struct slab *slabp) |
12dd36fae [PATCH] slab: ext... |
1947 |
{ |
12dd36fae [PATCH] slab: ext... |
1948 |
} |
1da177e4c Linux-2.6.12-rc2 |
1949 |
#endif |
911851e6e [PATCH] slab: fix... |
1950 1951 1952 1953 1954 |
/** * slab_destroy - destroy and release all objects in a slab * @cachep: cache pointer being destroyed * @slabp: slab pointer being destroyed * |
12dd36fae [PATCH] slab: ext... |
1955 |
* Destroy all the objs in a slab, and release the mem back to the system. |
a737b3e2f [PATCH] slab cleanup |
1956 1957 |
* Before calling the slab must have been unlinked from the cache. The * cache-lock is not held/needed. |
12dd36fae [PATCH] slab: ext... |
1958 |
*/ |
343e0d7a9 [PATCH] slab: rep... |
1959 |
static void slab_destroy(struct kmem_cache *cachep, struct slab *slabp) |
12dd36fae [PATCH] slab: ext... |
1960 1961 |
{ void *addr = slabp->s_mem - slabp->colouroff; |
e79aec291 slab: rename slab... |
1962 |
slab_destroy_debugcheck(cachep, slabp); |
1da177e4c Linux-2.6.12-rc2 |
1963 1964 |
if (unlikely(cachep->flags & SLAB_DESTROY_BY_RCU)) { struct slab_rcu *slab_rcu; |
b28a02de8 [PATCH] slab: fix... |
1965 |
slab_rcu = (struct slab_rcu *)slabp; |
1da177e4c Linux-2.6.12-rc2 |
1966 1967 1968 1969 1970 |
slab_rcu->cachep = cachep; slab_rcu->addr = addr; call_rcu(&slab_rcu->head, kmem_rcu_free); } else { kmem_freepages(cachep, addr); |
873623dfa [PATCH] lockdep: ... |
1971 1972 |
if (OFF_SLAB(cachep)) kmem_cache_free(cachep->slabp_cache, slabp); |
1da177e4c Linux-2.6.12-rc2 |
1973 1974 |
} } |
117f6eb1d [PATCH] slab: ext... |
1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 |
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 Linux-2.6.12-rc2 |
1994 |
/** |
a70773ddb [PATCH] mm/slab: ... |
1995 1996 1997 1998 1999 2000 2001 |
* 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 [PATCH] slab: ext... |
2002 2003 2004 2005 2006 |
* * 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 [PATCH] slab cleanup |
2007 |
static size_t calculate_slab_order(struct kmem_cache *cachep, |
ee13d785e [PATCH] slab: fix... |
2008 |
size_t size, size_t align, unsigned long flags) |
4d268eba1 [PATCH] slab: ext... |
2009 |
{ |
b1ab41c49 [PATCH] slab.c: f... |
2010 |
unsigned long offslab_limit; |
4d268eba1 [PATCH] slab: ext... |
2011 |
size_t left_over = 0; |
9888e6fa7 slab: clarify and... |
2012 |
int gfporder; |
4d268eba1 [PATCH] slab: ext... |
2013 |
|
0aa817f07 Slab allocators: ... |
2014 |
for (gfporder = 0; gfporder <= KMALLOC_MAX_ORDER; gfporder++) { |
4d268eba1 [PATCH] slab: ext... |
2015 2016 |
unsigned int num; size_t remainder; |
9888e6fa7 slab: clarify and... |
2017 |
cache_estimate(gfporder, size, align, flags, &remainder, &num); |
4d268eba1 [PATCH] slab: ext... |
2018 2019 |
if (!num) continue; |
9888e6fa7 slab: clarify and... |
2020 |
|
b1ab41c49 [PATCH] slab.c: f... |
2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 |
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 [PATCH] slab: ext... |
2033 |
|
9888e6fa7 slab: clarify and... |
2034 |
/* Found something acceptable - save it away */ |
4d268eba1 [PATCH] slab: ext... |
2035 |
cachep->num = num; |
9888e6fa7 slab: clarify and... |
2036 |
cachep->gfporder = gfporder; |
4d268eba1 [PATCH] slab: ext... |
2037 2038 2039 |
left_over = remainder; /* |
f78bb8ad4 slab: fix calcula... |
2040 2041 2042 2043 2044 2045 2046 2047 |
* 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 [PATCH] slab: ext... |
2048 2049 2050 |
* Large number of objects is good, but very large slabs are * currently bad for the gfp()s. */ |
543585cc5 slab: rename slab... |
2051 |
if (gfporder >= slab_max_order) |
4d268eba1 [PATCH] slab: ext... |
2052 |
break; |
9888e6fa7 slab: clarify and... |
2053 2054 2055 |
/* * Acceptable internal fragmentation? */ |
a737b3e2f [PATCH] slab cleanup |
2056 |
if (left_over * 8 <= (PAGE_SIZE << gfporder)) |
4d268eba1 [PATCH] slab: ext... |
2057 2058 2059 2060 |
break; } return left_over; } |
83b519e8b slab: setup alloc... |
2061 |
static int __init_refok setup_cpu_cache(struct kmem_cache *cachep, gfp_t gfp) |
f30cf7d13 [PATCH] slab: ext... |
2062 |
{ |
2ed3a4ef9 [PATCH] slab: do ... |
2063 |
if (g_cpucache_up == FULL) |
83b519e8b slab: setup alloc... |
2064 |
return enable_cpucache(cachep, gfp); |
2ed3a4ef9 [PATCH] slab: do ... |
2065 |
|
f30cf7d13 [PATCH] slab: ext... |
2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 |
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 slab: setup alloc... |
2086 |
kmalloc(sizeof(struct arraycache_init), gfp); |
f30cf7d13 [PATCH] slab: ext... |
2087 2088 2089 2090 2091 2092 |
if (g_cpucache_up == PARTIAL_AC) { set_up_list3s(cachep, SIZE_L3); g_cpucache_up = PARTIAL_L3; } else { int node; |
556a169da slab: fix bootstr... |
2093 |
for_each_online_node(node) { |
f30cf7d13 [PATCH] slab: ext... |
2094 2095 |
cachep->nodelists[node] = kmalloc_node(sizeof(struct kmem_list3), |
eb91f1d0a slab: fix gfp fla... |
2096 |
gfp, node); |
f30cf7d13 [PATCH] slab: ext... |
2097 2098 2099 2100 2101 |
BUG_ON(!cachep->nodelists[node]); kmem_list3_init(cachep->nodelists[node]); } } } |
7d6e6d09d numa: slab: use n... |
2102 |
cachep->nodelists[numa_mem_id()]->next_reap = |
f30cf7d13 [PATCH] slab: ext... |
2103 2104 2105 2106 2107 2108 2109 2110 2111 |
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 [PATCH] slab: do ... |
2112 |
return 0; |
f30cf7d13 [PATCH] slab: ext... |
2113 |
} |
4d268eba1 [PATCH] slab: ext... |
2114 |
/** |
1da177e4c Linux-2.6.12-rc2 |
2115 2116 2117 2118 2119 2120 |
* 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 Linux-2.6.12-rc2 |
2121 2122 2123 |
* * Returns a ptr to the cache on success, NULL on failure. * Cannot be called within a int, but can be interrupted. |
20c2df83d mm: Remove slab d... |
2124 |
* The @ctor is run when new pages are allocated by the cache. |
1da177e4c Linux-2.6.12-rc2 |
2125 2126 |
* * @name must be valid until the cache is destroyed. This implies that |
a737b3e2f [PATCH] slab cleanup |
2127 2128 |
* the module calling this has to destroy the cache before getting unloaded. * |
1da177e4c Linux-2.6.12-rc2 |
2129 2130 2131 2132 2133 2134 2135 2136 |
* 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 Linux-2.6.12-rc2 |
2137 2138 2139 2140 |
* %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 [PATCH] slab: rep... |
2141 |
struct kmem_cache * |
1da177e4c Linux-2.6.12-rc2 |
2142 |
kmem_cache_create (const char *name, size_t size, size_t align, |
51cc50685 SL*B: drop kmem c... |
2143 |
unsigned long flags, void (*ctor)(void *)) |
1da177e4c Linux-2.6.12-rc2 |
2144 2145 |
{ size_t left_over, slab_size, ralign; |
7a7c381d2 [PATCH] slab: sto... |
2146 |
struct kmem_cache *cachep = NULL, *pc; |
83b519e8b slab: setup alloc... |
2147 |
gfp_t gfp; |
1da177e4c Linux-2.6.12-rc2 |
2148 2149 2150 2151 |
/* * Sanity checks... these are all serious usage bugs. */ |
a737b3e2f [PATCH] slab cleanup |
2152 |
if (!name || in_interrupt() || (size < BYTES_PER_WORD) || |
20c2df83d mm: Remove slab d... |
2153 |
size > KMALLOC_MAX_SIZE) { |
d40cee245 mm: remove remain... |
2154 2155 |
printk(KERN_ERR "%s: Early error in slab %s ", __func__, |
a737b3e2f [PATCH] slab cleanup |
2156 |
name); |
b28a02de8 [PATCH] slab: fix... |
2157 2158 |
BUG(); } |
1da177e4c Linux-2.6.12-rc2 |
2159 |
|
f0188f474 [PATCH] slab: Avo... |
2160 |
/* |
8f5be20bf [PATCH] mm: slab:... |
2161 |
* We use cache_chain_mutex to ensure a consistent view of |
174596a0b cpumask: convert mm/ |
2162 |
* cpu_online_mask as well. Please see cpuup_callback |
f0188f474 [PATCH] slab: Avo... |
2163 |
*/ |
83b519e8b slab: setup alloc... |
2164 2165 2166 2167 |
if (slab_is_available()) { get_online_cpus(); mutex_lock(&cache_chain_mutex); } |
4f12bb4f7 [PATCH] slab: don... |
2168 |
|
7a7c381d2 [PATCH] slab: sto... |
2169 |
list_for_each_entry(pc, &cache_chain, next) { |
4f12bb4f7 [PATCH] slab: don... |
2170 2171 2172 2173 2174 2175 2176 2177 |
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 [PATCH] slab: use... |
2178 |
res = probe_kernel_address(pc->name, tmp); |
4f12bb4f7 [PATCH] slab: don... |
2179 |
if (res) { |
b4169525b include KERN_* co... |
2180 2181 2182 |
printk(KERN_ERR "SLAB: cache with size %d has lost its name ", |
3dafccf22 [PATCH] slab: dis... |
2183 |
pc->buffer_size); |
4f12bb4f7 [PATCH] slab: don... |
2184 2185 |
continue; } |
b28a02de8 [PATCH] slab: fix... |
2186 |
if (!strcmp(pc->name, name)) { |
b4169525b include KERN_* co... |
2187 2188 2189 |
printk(KERN_ERR "kmem_cache_create: duplicate cache %s ", name); |
4f12bb4f7 [PATCH] slab: don... |
2190 2191 2192 2193 |
dump_stack(); goto oops; } } |
1da177e4c Linux-2.6.12-rc2 |
2194 2195 |
#if DEBUG WARN_ON(strchr(name, ' ')); /* It confuses parsers */ |
1da177e4c Linux-2.6.12-rc2 |
2196 2197 2198 2199 2200 2201 2202 |
#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 Fix slab redzone ... |
2203 2204 |
if (size < 4096 || fls(size - 1) == fls(size-1 + REDZONE_ALIGN + 2 * sizeof(unsigned long long))) |
b28a02de8 [PATCH] slab: fix... |
2205 |
flags |= SLAB_RED_ZONE | SLAB_STORE_USER; |
1da177e4c Linux-2.6.12-rc2 |
2206 2207 2208 2209 2210 2211 |
if (!(flags & SLAB_DESTROY_BY_RCU)) flags |= SLAB_POISON; #endif if (flags & SLAB_DESTROY_BY_RCU) BUG_ON(flags & SLAB_POISON); #endif |
1da177e4c Linux-2.6.12-rc2 |
2212 |
/* |
a737b3e2f [PATCH] slab cleanup |
2213 2214 |
* Always checks flags, a caller might be expecting debug support which * isn't available. |
1da177e4c Linux-2.6.12-rc2 |
2215 |
*/ |
40094fa65 BUG_ON() Conversi... |
2216 |
BUG_ON(flags & ~CREATE_MASK); |
1da177e4c Linux-2.6.12-rc2 |
2217 |
|
a737b3e2f [PATCH] slab cleanup |
2218 2219 |
/* * Check that size is in terms of words. This is needed to avoid |
1da177e4c Linux-2.6.12-rc2 |
2220 2221 2222 |
* unaligned accesses for some archs when redzoning is used, and makes * sure any on-slab bufctl's are also correctly aligned. */ |
b28a02de8 [PATCH] slab: fix... |
2223 2224 2225 |
if (size & (BYTES_PER_WORD - 1)) { size += (BYTES_PER_WORD - 1); size &= ~(BYTES_PER_WORD - 1); |
1da177e4c Linux-2.6.12-rc2 |
2226 |
} |
a737b3e2f [PATCH] slab cleanup |
2227 |
/* calculate the final buffer alignment: */ |
1da177e4c Linux-2.6.12-rc2 |
2228 2229 |
/* 1) arch recommendation: can be overridden for debug */ if (flags & SLAB_HWCACHE_ALIGN) { |
a737b3e2f [PATCH] slab cleanup |
2230 2231 2232 2233 |
/* * Default alignment: as specified by the arch code. Except if * an object is really small, then squeeze multiple objects into * one cacheline. |
1da177e4c Linux-2.6.12-rc2 |
2234 2235 |
*/ ralign = cache_line_size(); |
b28a02de8 [PATCH] slab: fix... |
2236 |
while (size <= ralign / 2) |
1da177e4c Linux-2.6.12-rc2 |
2237 2238 2239 2240 |
ralign /= 2; } else { ralign = BYTES_PER_WORD; } |
ca5f9703d [PATCH] slab: res... |
2241 2242 |
/* |
87a927c71 Fix slab redzone ... |
2243 2244 2245 |
* 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 [PATCH] slab: res... |
2246 |
*/ |
87a927c71 Fix slab redzone ... |
2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 |
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 [PATCH] slab: res... |
2257 |
|
a44b56d35 [PATCH] slab debu... |
2258 |
/* 2) arch mandated alignment */ |
1da177e4c Linux-2.6.12-rc2 |
2259 2260 |
if (ralign < ARCH_SLAB_MINALIGN) { ralign = ARCH_SLAB_MINALIGN; |
1da177e4c Linux-2.6.12-rc2 |
2261 |
} |
a44b56d35 [PATCH] slab debu... |
2262 |
/* 3) caller mandated alignment */ |
1da177e4c Linux-2.6.12-rc2 |
2263 2264 |
if (ralign < align) { ralign = align; |
1da177e4c Linux-2.6.12-rc2 |
2265 |
} |
3ff84a7f3 Revert "slab: Fix... |
2266 2267 |
/* disable debug if necessary */ if (ralign > __alignof__(unsigned long long)) |
a44b56d35 [PATCH] slab debu... |
2268 |
flags &= ~(SLAB_RED_ZONE | SLAB_STORE_USER); |
a737b3e2f [PATCH] slab cleanup |
2269 |
/* |
ca5f9703d [PATCH] slab: res... |
2270 |
* 4) Store it. |
1da177e4c Linux-2.6.12-rc2 |
2271 2272 |
*/ align = ralign; |
83b519e8b slab: setup alloc... |
2273 2274 2275 2276 |
if (slab_is_available()) gfp = GFP_KERNEL; else gfp = GFP_NOWAIT; |
1da177e4c Linux-2.6.12-rc2 |
2277 |
/* Get cache's description obj. */ |
83b519e8b slab: setup alloc... |
2278 |
cachep = kmem_cache_zalloc(&cache_cache, gfp); |
1da177e4c Linux-2.6.12-rc2 |
2279 |
if (!cachep) |
4f12bb4f7 [PATCH] slab: don... |
2280 |
goto oops; |
1da177e4c Linux-2.6.12-rc2 |
2281 |
|
b56efcf0a slab: shrink size... |
2282 |
cachep->nodelists = (struct kmem_list3 **)&cachep->array[nr_cpu_ids]; |
1da177e4c Linux-2.6.12-rc2 |
2283 |
#if DEBUG |
3dafccf22 [PATCH] slab: dis... |
2284 |
cachep->obj_size = size; |
1da177e4c Linux-2.6.12-rc2 |
2285 |
|
ca5f9703d [PATCH] slab: res... |
2286 2287 2288 2289 |
/* * Both debugging options require word-alignment which is calculated * into align above. */ |
1da177e4c Linux-2.6.12-rc2 |
2290 |
if (flags & SLAB_RED_ZONE) { |
1da177e4c Linux-2.6.12-rc2 |
2291 |
/* add space for red zone words */ |
3ff84a7f3 Revert "slab: Fix... |
2292 2293 |
cachep->obj_offset += sizeof(unsigned long long); size += 2 * sizeof(unsigned long long); |
1da177e4c Linux-2.6.12-rc2 |
2294 2295 |
} if (flags & SLAB_STORE_USER) { |
ca5f9703d [PATCH] slab: res... |
2296 |
/* user store requires one word storage behind the end of |
87a927c71 Fix slab redzone ... |
2297 2298 |
* the real object. But if the second red zone needs to be * aligned to 64 bits, we must allow that much space. |
1da177e4c Linux-2.6.12-rc2 |
2299 |
*/ |
87a927c71 Fix slab redzone ... |
2300 2301 2302 2303 |
if (flags & SLAB_RED_ZONE) size += REDZONE_ALIGN; else size += BYTES_PER_WORD; |
1da177e4c Linux-2.6.12-rc2 |
2304 2305 |
} #if FORCED_DEBUG && defined(CONFIG_DEBUG_PAGEALLOC) |
b28a02de8 [PATCH] slab: fix... |
2306 |
if (size >= malloc_sizes[INDEX_L3 + 1].cs_size |
1ab335d8f slab: fix object ... |
2307 2308 |
&& cachep->obj_size > cache_line_size() && ALIGN(size, align) < PAGE_SIZE) { cachep->obj_offset += PAGE_SIZE - ALIGN(size, align); |
1da177e4c Linux-2.6.12-rc2 |
2309 2310 2311 2312 |
size = PAGE_SIZE; } #endif #endif |
e0a427267 [PATCH] mm/slab.c... |
2313 2314 2315 |
/* * Determine if the slab management is 'on' or 'off' slab. * (bootstrapping cannot cope with offslab caches so don't do |
e7cb55b94 kmemleak: Do not ... |
2316 2317 |
* it too early on. Always use on-slab management when * SLAB_NOLEAKTRACE to avoid recursive calls into kmemleak) |
e0a427267 [PATCH] mm/slab.c... |
2318 |
*/ |
e7cb55b94 kmemleak: Do not ... |
2319 2320 |
if ((size >= (PAGE_SIZE >> 3)) && !slab_early_init && !(flags & SLAB_NOLEAKTRACE)) |
1da177e4c Linux-2.6.12-rc2 |
2321 2322 2323 2324 2325 2326 2327 |
/* * 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 slab: fix calcula... |
2328 |
left_over = calculate_slab_order(cachep, size, align, flags); |
1da177e4c Linux-2.6.12-rc2 |
2329 2330 |
if (!cachep->num) { |
b4169525b include KERN_* co... |
2331 2332 2333 |
printk(KERN_ERR "kmem_cache_create: couldn't create cache %s. ", name); |
1da177e4c Linux-2.6.12-rc2 |
2334 2335 |
kmem_cache_free(&cache_cache, cachep); cachep = NULL; |
4f12bb4f7 [PATCH] slab: don... |
2336 |
goto oops; |
1da177e4c Linux-2.6.12-rc2 |
2337 |
} |
b28a02de8 [PATCH] slab: fix... |
2338 2339 |
slab_size = ALIGN(cachep->num * sizeof(kmem_bufctl_t) + sizeof(struct slab), align); |
1da177e4c Linux-2.6.12-rc2 |
2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 |
/* * 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 [PATCH] slab: fix... |
2352 2353 |
slab_size = cachep->num * sizeof(kmem_bufctl_t) + sizeof(struct slab); |
674613652 slab: fix generic... |
2354 2355 2356 2357 2358 2359 2360 2361 2362 |
#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 Linux-2.6.12-rc2 |
2363 2364 2365 2366 2367 2368 |
} cachep->colour_off = cache_line_size(); /* Offset must be a multiple of the alignment. */ if (cachep->colour_off < align) cachep->colour_off = align; |
b28a02de8 [PATCH] slab: fix... |
2369 |
cachep->colour = left_over / cachep->colour_off; |
1da177e4c Linux-2.6.12-rc2 |
2370 2371 2372 |
cachep->slab_size = slab_size; cachep->flags = flags; cachep->gfpflags = 0; |
4b51d6698 [PATCH] optional ... |
2373 |
if (CONFIG_ZONE_DMA_FLAG && (flags & SLAB_CACHE_DMA)) |
1da177e4c Linux-2.6.12-rc2 |
2374 |
cachep->gfpflags |= GFP_DMA; |
3dafccf22 [PATCH] slab: dis... |
2375 |
cachep->buffer_size = size; |
6a2d7a955 [PATCH] SLAB: use... |
2376 |
cachep->reciprocal_buffer_size = reciprocal_value(size); |
1da177e4c Linux-2.6.12-rc2 |
2377 |
|
e5ac9c5ae [PATCH] Add some ... |
2378 |
if (flags & CFLGS_OFF_SLAB) { |
b2d550736 [PATCH] mm/slab: ... |
2379 |
cachep->slabp_cache = kmem_find_general_cachep(slab_size, 0u); |
e5ac9c5ae [PATCH] Add some ... |
2380 2381 2382 2383 2384 2385 2386 |
/* * 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 Slab allocators: ... |
2387 |
BUG_ON(ZERO_OR_NULL_PTR(cachep->slabp_cache)); |
e5ac9c5ae [PATCH] Add some ... |
2388 |
} |
1da177e4c Linux-2.6.12-rc2 |
2389 |
cachep->ctor = ctor; |
1da177e4c Linux-2.6.12-rc2 |
2390 |
cachep->name = name; |
83b519e8b slab: setup alloc... |
2391 |
if (setup_cpu_cache(cachep, gfp)) { |
2ed3a4ef9 [PATCH] slab: do ... |
2392 2393 2394 2395 |
__kmem_cache_destroy(cachep); cachep = NULL; goto oops; } |
1da177e4c Linux-2.6.12-rc2 |
2396 |
|
83835b3d9 slab, lockdep: An... |
2397 2398 2399 2400 2401 2402 2403 2404 2405 |
if (flags & SLAB_DEBUG_OBJECTS) { /* * Would deadlock through slab_destroy()->call_rcu()-> * debug_object_activate()->kmem_cache_alloc(). */ WARN_ON_ONCE(flags & SLAB_DESTROY_BY_RCU); slab_set_debugobj_lock_classes(cachep); } |
1da177e4c Linux-2.6.12-rc2 |
2406 2407 |
/* cache setup completed, link it into the list */ list_add(&cachep->next, &cache_chain); |
a737b3e2f [PATCH] slab cleanup |
2408 |
oops: |
1da177e4c Linux-2.6.12-rc2 |
2409 2410 2411 |
if (!cachep && (flags & SLAB_PANIC)) panic("kmem_cache_create(): failed to create slab `%s' ", |
b28a02de8 [PATCH] slab: fix... |
2412 |
name); |
83b519e8b slab: setup alloc... |
2413 2414 2415 2416 |
if (slab_is_available()) { mutex_unlock(&cache_chain_mutex); put_online_cpus(); } |
1da177e4c Linux-2.6.12-rc2 |
2417 2418 2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430 |
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 [PATCH] slab: rep... |
2431 |
static void check_spinlock_acquired(struct kmem_cache *cachep) |
1da177e4c Linux-2.6.12-rc2 |
2432 2433 2434 |
{ #ifdef CONFIG_SMP check_irq_off(); |
7d6e6d09d numa: slab: use n... |
2435 |
assert_spin_locked(&cachep->nodelists[numa_mem_id()]->list_lock); |
1da177e4c Linux-2.6.12-rc2 |
2436 2437 |
#endif } |
e498be7da [PATCH] Numa-awar... |
2438 |
|
343e0d7a9 [PATCH] slab: rep... |
2439 |
static void check_spinlock_acquired_node(struct kmem_cache *cachep, int node) |
e498be7da [PATCH] Numa-awar... |
2440 2441 2442 2443 2444 2445 |
{ #ifdef CONFIG_SMP check_irq_off(); assert_spin_locked(&cachep->nodelists[node]->list_lock); #endif } |
1da177e4c Linux-2.6.12-rc2 |
2446 2447 2448 2449 |
#else #define check_irq_off() do { } while(0) #define check_irq_on() do { } while(0) #define check_spinlock_acquired(x) do { } while(0) |
e498be7da [PATCH] Numa-awar... |
2450 |
#define check_spinlock_acquired_node(x, y) do { } while(0) |
1da177e4c Linux-2.6.12-rc2 |
2451 |
#endif |
aab2207cf [PATCH] slab: mak... |
2452 2453 2454 |
static void drain_array(struct kmem_cache *cachep, struct kmem_list3 *l3, struct array_cache *ac, int force, int node); |
1da177e4c Linux-2.6.12-rc2 |
2455 2456 |
static void do_drain(void *arg) { |
a737b3e2f [PATCH] slab cleanup |
2457 |
struct kmem_cache *cachep = arg; |
1da177e4c Linux-2.6.12-rc2 |
2458 |
struct array_cache *ac; |
7d6e6d09d numa: slab: use n... |
2459 |
int node = numa_mem_id(); |
1da177e4c Linux-2.6.12-rc2 |
2460 2461 |
check_irq_off(); |
9a2dba4b4 [PATCH] slab: ren... |
2462 |
ac = cpu_cache_get(cachep); |
ff69416e6 [PATCH] slab: fix... |
2463 2464 2465 |
spin_lock(&cachep->nodelists[node]->list_lock); free_block(cachep, ac->entry, ac->avail, node); spin_unlock(&cachep->nodelists[node]->list_lock); |
1da177e4c Linux-2.6.12-rc2 |
2466 2467 |
ac->avail = 0; } |
343e0d7a9 [PATCH] slab: rep... |
2468 |
static void drain_cpu_caches(struct kmem_cache *cachep) |
1da177e4c Linux-2.6.12-rc2 |
2469 |
{ |
e498be7da [PATCH] Numa-awar... |
2470 2471 |
struct kmem_list3 *l3; int node; |
15c8b6c1a on_each_cpu(): ki... |
2472 |
on_each_cpu(do_drain, cachep, 1); |
1da177e4c Linux-2.6.12-rc2 |
2473 |
check_irq_on(); |
b28a02de8 [PATCH] slab: fix... |
2474 |
for_each_online_node(node) { |
e498be7da [PATCH] Numa-awar... |
2475 |
l3 = cachep->nodelists[node]; |
a4523a8b3 [PATCH] slab: Fix... |
2476 2477 2478 2479 2480 2481 2482 |
if (l3 && l3->alien) drain_alien_cache(cachep, l3->alien); } for_each_online_node(node) { l3 = cachep->nodelists[node]; if (l3) |
aab2207cf [PATCH] slab: mak... |
2483 |
drain_array(cachep, l3, l3->shared, 1, node); |
e498be7da [PATCH] Numa-awar... |
2484 |
} |
1da177e4c Linux-2.6.12-rc2 |
2485 |
} |
ed11d9eb2 [PATCH] slab: con... |
2486 2487 2488 2489 2490 2491 2492 2493 |
/* * 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 Linux-2.6.12-rc2 |
2494 |
{ |
ed11d9eb2 [PATCH] slab: con... |
2495 2496 |
struct list_head *p; int nr_freed; |
1da177e4c Linux-2.6.12-rc2 |
2497 |
struct slab *slabp; |
1da177e4c Linux-2.6.12-rc2 |
2498 |
|
ed11d9eb2 [PATCH] slab: con... |
2499 2500 |
nr_freed = 0; while (nr_freed < tofree && !list_empty(&l3->slabs_free)) { |
1da177e4c Linux-2.6.12-rc2 |
2501 |
|
ed11d9eb2 [PATCH] slab: con... |
2502 |
spin_lock_irq(&l3->list_lock); |
e498be7da [PATCH] Numa-awar... |
2503 |
p = l3->slabs_free.prev; |
ed11d9eb2 [PATCH] slab: con... |
2504 2505 2506 2507 |
if (p == &l3->slabs_free) { spin_unlock_irq(&l3->list_lock); goto out; } |
1da177e4c Linux-2.6.12-rc2 |
2508 |
|
ed11d9eb2 [PATCH] slab: con... |
2509 |
slabp = list_entry(p, struct slab, list); |
1da177e4c Linux-2.6.12-rc2 |
2510 |
#if DEBUG |
40094fa65 BUG_ON() Conversi... |
2511 |
BUG_ON(slabp->inuse); |
1da177e4c Linux-2.6.12-rc2 |
2512 2513 |
#endif list_del(&slabp->list); |
ed11d9eb2 [PATCH] slab: con... |
2514 2515 2516 2517 2518 |
/* * Safe to drop the lock. The slab is no longer linked * to the cache. */ l3->free_objects -= cache->num; |
e498be7da [PATCH] Numa-awar... |
2519 |
spin_unlock_irq(&l3->list_lock); |
ed11d9eb2 [PATCH] slab: con... |
2520 2521 |
slab_destroy(cache, slabp); nr_freed++; |
1da177e4c Linux-2.6.12-rc2 |
2522 |
} |
ed11d9eb2 [PATCH] slab: con... |
2523 2524 |
out: return nr_freed; |
1da177e4c Linux-2.6.12-rc2 |
2525 |
} |
8f5be20bf [PATCH] mm: slab:... |
2526 |
/* Called with cache_chain_mutex held to protect against cpu hotplug */ |
343e0d7a9 [PATCH] slab: rep... |
2527 |
static int __cache_shrink(struct kmem_cache *cachep) |
e498be7da [PATCH] Numa-awar... |
2528 2529 2530 2531 2532 2533 2534 2535 2536 |
{ 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 [PATCH] slab: con... |
2537 2538 2539 2540 2541 2542 2543 |
if (!l3) continue; drain_freelist(cachep, l3, l3->free_objects); ret += !list_empty(&l3->slabs_full) || !list_empty(&l3->slabs_partial); |
e498be7da [PATCH] Numa-awar... |
2544 2545 2546 |
} return (ret ? 1 : 0); } |
1da177e4c Linux-2.6.12-rc2 |
2547 2548 2549 2550 2551 2552 2553 |
/** * 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 [PATCH] slab: rep... |
2554 |
int kmem_cache_shrink(struct kmem_cache *cachep) |
1da177e4c Linux-2.6.12-rc2 |
2555 |
{ |
8f5be20bf [PATCH] mm: slab:... |
2556 |
int ret; |
40094fa65 BUG_ON() Conversi... |
2557 |
BUG_ON(!cachep || in_interrupt()); |
1da177e4c Linux-2.6.12-rc2 |
2558 |
|
95402b382 cpu-hotplug: repl... |
2559 |
get_online_cpus(); |
8f5be20bf [PATCH] mm: slab:... |
2560 2561 2562 |
mutex_lock(&cache_chain_mutex); ret = __cache_shrink(cachep); mutex_unlock(&cache_chain_mutex); |
95402b382 cpu-hotplug: repl... |
2563 |
put_online_cpus(); |
8f5be20bf [PATCH] mm: slab:... |
2564 |
return ret; |
1da177e4c Linux-2.6.12-rc2 |
2565 2566 2567 2568 2569 2570 2571 |
} EXPORT_SYMBOL(kmem_cache_shrink); /** * kmem_cache_destroy - delete a cache * @cachep: the cache to destroy * |
72fd4a35a [PATCH] Numerous ... |
2572 |
* Remove a &struct kmem_cache object from the slab cache. |
1da177e4c Linux-2.6.12-rc2 |
2573 2574 2575 2576 2577 2578 2579 2580 |
* * It is expected this function will be called by a module when it is * unloaded. This will remove the cache completely, and avoid a duplicate * cache being allocated each time a module is loaded and unloaded, if the * module doesn't have persistent in-kernel storage across loads and unloads. * * The cache must be empty before calling this function. * |
25985edce Fix common misspe... |
2581 |
* The caller must guarantee that no one will allocate memory from the cache |
1da177e4c Linux-2.6.12-rc2 |
2582 2583 |
* during the kmem_cache_destroy(). */ |
133d205a1 [PATCH] Make kmem... |
2584 |
void kmem_cache_destroy(struct kmem_cache *cachep) |
1da177e4c Linux-2.6.12-rc2 |
2585 |
{ |
40094fa65 BUG_ON() Conversi... |
2586 |
BUG_ON(!cachep || in_interrupt()); |
1da177e4c Linux-2.6.12-rc2 |
2587 |
|
1da177e4c Linux-2.6.12-rc2 |
2588 |
/* Find the cache in the chain of caches. */ |
95402b382 cpu-hotplug: repl... |
2589 |
get_online_cpus(); |
fc0abb145 [PATCH] sem2mutex... |
2590 |
mutex_lock(&cache_chain_mutex); |
1da177e4c Linux-2.6.12-rc2 |
2591 2592 2593 2594 |
/* * the chain is never empty, cache_cache is never destroyed */ list_del(&cachep->next); |
1da177e4c Linux-2.6.12-rc2 |
2595 2596 |
if (__cache_shrink(cachep)) { slab_error(cachep, "Can't free all objects"); |
b28a02de8 [PATCH] slab: fix... |
2597 |
list_add(&cachep->next, &cache_chain); |
fc0abb145 [PATCH] sem2mutex... |
2598 |
mutex_unlock(&cache_chain_mutex); |
95402b382 cpu-hotplug: repl... |
2599 |
put_online_cpus(); |
133d205a1 [PATCH] Make kmem... |
2600 |
return; |
1da177e4c Linux-2.6.12-rc2 |
2601 2602 2603 |
} if (unlikely(cachep->flags & SLAB_DESTROY_BY_RCU)) |
7ed9f7e5d fix RCU-callback-... |
2604 |
rcu_barrier(); |
1da177e4c Linux-2.6.12-rc2 |
2605 |
|
117f6eb1d [PATCH] slab: ext... |
2606 |
__kmem_cache_destroy(cachep); |
8f5be20bf [PATCH] mm: slab:... |
2607 |
mutex_unlock(&cache_chain_mutex); |
95402b382 cpu-hotplug: repl... |
2608 |
put_online_cpus(); |
1da177e4c Linux-2.6.12-rc2 |
2609 2610 |
} EXPORT_SYMBOL(kmem_cache_destroy); |
e5ac9c5ae [PATCH] Add some ... |
2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 |
/* * 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 [PATCH] slab: rep... |
2622 |
static struct slab *alloc_slabmgmt(struct kmem_cache *cachep, void *objp, |
5b74ada7e [PATCH] slab: all... |
2623 2624 |
int colour_off, gfp_t local_flags, int nodeid) |
1da177e4c Linux-2.6.12-rc2 |
2625 2626 |
{ struct slab *slabp; |
b28a02de8 [PATCH] slab: fix... |
2627 |
|
1da177e4c Linux-2.6.12-rc2 |
2628 2629 |
if (OFF_SLAB(cachep)) { /* Slab management obj is off-slab. */ |
5b74ada7e [PATCH] slab: all... |
2630 |
slabp = kmem_cache_alloc_node(cachep->slabp_cache, |
8759ec50a slab: remove GFP_... |
2631 |
local_flags, nodeid); |
d5cff6352 kmemleak: Add the... |
2632 2633 2634 2635 2636 2637 |
/* * 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 kmemleak: Simplif... |
2638 2639 |
kmemleak_scan_area(&slabp->list, sizeof(struct list_head), local_flags); |
1da177e4c Linux-2.6.12-rc2 |
2640 2641 2642 |
if (!slabp) return NULL; } else { |
b28a02de8 [PATCH] slab: fix... |
2643 |
slabp = objp + colour_off; |
1da177e4c Linux-2.6.12-rc2 |
2644 2645 2646 2647 |
colour_off += cachep->slab_size; } slabp->inuse = 0; slabp->colouroff = colour_off; |
b28a02de8 [PATCH] slab: fix... |
2648 |
slabp->s_mem = objp + colour_off; |
5b74ada7e [PATCH] slab: all... |
2649 |
slabp->nodeid = nodeid; |
e51bfd0ad slab: avoid doubl... |
2650 |
slabp->free = 0; |
1da177e4c Linux-2.6.12-rc2 |
2651 2652 2653 2654 2655 |
return slabp; } static inline kmem_bufctl_t *slab_bufctl(struct slab *slabp) { |
b28a02de8 [PATCH] slab: fix... |
2656 |
return (kmem_bufctl_t *) (slabp + 1); |
1da177e4c Linux-2.6.12-rc2 |
2657 |
} |
343e0d7a9 [PATCH] slab: rep... |
2658 |
static void cache_init_objs(struct kmem_cache *cachep, |
a35afb830 Remove SLAB_CTOR_... |
2659 |
struct slab *slabp) |
1da177e4c Linux-2.6.12-rc2 |
2660 2661 2662 2663 |
{ int i; for (i = 0; i < cachep->num; i++) { |
8fea4e96a [PATCH] slab: obj... |
2664 |
void *objp = index_to_obj(cachep, slabp, i); |
1da177e4c Linux-2.6.12-rc2 |
2665 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676 |
#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 [PATCH] slab cleanup |
2677 2678 2679 |
* 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 Linux-2.6.12-rc2 |
2680 2681 |
*/ if (cachep->ctor && !(cachep->flags & SLAB_POISON)) |
51cc50685 SL*B: drop kmem c... |
2682 |
cachep->ctor(objp + obj_offset(cachep)); |
1da177e4c Linux-2.6.12-rc2 |
2683 2684 2685 2686 |
if (cachep->flags & SLAB_RED_ZONE) { if (*dbg_redzone2(cachep, objp) != RED_INACTIVE) slab_error(cachep, "constructor overwrote the" |
b28a02de8 [PATCH] slab: fix... |
2687 |
" end of an object"); |
1da177e4c Linux-2.6.12-rc2 |
2688 2689 |
if (*dbg_redzone1(cachep, objp) != RED_INACTIVE) slab_error(cachep, "constructor overwrote the" |
b28a02de8 [PATCH] slab: fix... |
2690 |
" start of an object"); |
1da177e4c Linux-2.6.12-rc2 |
2691 |
} |
a737b3e2f [PATCH] slab cleanup |
2692 2693 |
if ((cachep->buffer_size % PAGE_SIZE) == 0 && OFF_SLAB(cachep) && cachep->flags & SLAB_POISON) |
b28a02de8 [PATCH] slab: fix... |
2694 |
kernel_map_pages(virt_to_page(objp), |
3dafccf22 [PATCH] slab: dis... |
2695 |
cachep->buffer_size / PAGE_SIZE, 0); |
1da177e4c Linux-2.6.12-rc2 |
2696 2697 |
#else if (cachep->ctor) |
51cc50685 SL*B: drop kmem c... |
2698 |
cachep->ctor(objp); |
1da177e4c Linux-2.6.12-rc2 |
2699 |
#endif |
b28a02de8 [PATCH] slab: fix... |
2700 |
slab_bufctl(slabp)[i] = i + 1; |
1da177e4c Linux-2.6.12-rc2 |
2701 |
} |
b28a02de8 [PATCH] slab: fix... |
2702 |
slab_bufctl(slabp)[i - 1] = BUFCTL_END; |
1da177e4c Linux-2.6.12-rc2 |
2703 |
} |
343e0d7a9 [PATCH] slab: rep... |
2704 |
static void kmem_flagcheck(struct kmem_cache *cachep, gfp_t flags) |
1da177e4c Linux-2.6.12-rc2 |
2705 |
{ |
4b51d6698 [PATCH] optional ... |
2706 2707 2708 2709 2710 2711 |
if (CONFIG_ZONE_DMA_FLAG) { if (flags & GFP_DMA) BUG_ON(!(cachep->gfpflags & GFP_DMA)); else BUG_ON(cachep->gfpflags & GFP_DMA); } |
1da177e4c Linux-2.6.12-rc2 |
2712 |
} |
a737b3e2f [PATCH] slab cleanup |
2713 2714 |
static void *slab_get_obj(struct kmem_cache *cachep, struct slab *slabp, int nodeid) |
78d382d77 [PATCH] slab: ext... |
2715 |
{ |
8fea4e96a [PATCH] slab: obj... |
2716 |
void *objp = index_to_obj(cachep, slabp, slabp->free); |
78d382d77 [PATCH] slab: ext... |
2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727 2728 |
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 [PATCH] slab cleanup |
2729 2730 |
static void slab_put_obj(struct kmem_cache *cachep, struct slab *slabp, void *objp, int nodeid) |
78d382d77 [PATCH] slab: ext... |
2731 |
{ |
8fea4e96a [PATCH] slab: obj... |
2732 |
unsigned int objnr = obj_to_index(cachep, slabp, objp); |
78d382d77 [PATCH] slab: ext... |
2733 2734 2735 2736 |
#if DEBUG /* Verify that the slab belongs to the intended node */ WARN_ON(slabp->nodeid != nodeid); |
871751e25 [PATCH] slab: imp... |
2737 |
if (slab_bufctl(slabp)[objnr] + 1 <= SLAB_LIMIT + 1) { |
78d382d77 [PATCH] slab: ext... |
2738 |
printk(KERN_ERR "slab: double free detected in cache " |
a737b3e2f [PATCH] slab cleanup |
2739 2740 |
"'%s', objp %p ", cachep->name, objp); |
78d382d77 [PATCH] slab: ext... |
2741 2742 2743 2744 2745 2746 2747 |
BUG(); } #endif slab_bufctl(slabp)[objnr] = slabp->free; slabp->free = objnr; slabp->inuse--; } |
4776874ff [PATCH] slab: pag... |
2748 2749 2750 |
/* * Map pages beginning at addr to the given cache and slab. This is required * for the slab allocator to be able to lookup the cache and slab of a |
ccd35fb9f kernel: kmem_ptr_... |
2751 |
* virtual address for kfree, ksize, and slab debugging. |
4776874ff [PATCH] slab: pag... |
2752 2753 2754 |
*/ static void slab_map_pages(struct kmem_cache *cache, struct slab *slab, void *addr) |
1da177e4c Linux-2.6.12-rc2 |
2755 |
{ |
4776874ff [PATCH] slab: pag... |
2756 |
int nr_pages; |
1da177e4c Linux-2.6.12-rc2 |
2757 |
struct page *page; |
4776874ff [PATCH] slab: pag... |
2758 |
page = virt_to_page(addr); |
84097518d [PATCH] mm: nommu... |
2759 |
|
4776874ff [PATCH] slab: pag... |
2760 |
nr_pages = 1; |
84097518d [PATCH] mm: nommu... |
2761 |
if (likely(!PageCompound(page))) |
4776874ff [PATCH] slab: pag... |
2762 |
nr_pages <<= cache->gfporder; |
1da177e4c Linux-2.6.12-rc2 |
2763 |
do { |
4776874ff [PATCH] slab: pag... |
2764 2765 |
page_set_cache(page, cache); page_set_slab(page, slab); |
1da177e4c Linux-2.6.12-rc2 |
2766 |
page++; |
4776874ff [PATCH] slab: pag... |
2767 |
} while (--nr_pages); |
1da177e4c Linux-2.6.12-rc2 |
2768 2769 2770 2771 2772 2773 |
} /* * 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 [PATCH] slab: bet... |
2774 2775 |
static int cache_grow(struct kmem_cache *cachep, gfp_t flags, int nodeid, void *objp) |
1da177e4c Linux-2.6.12-rc2 |
2776 |
{ |
b28a02de8 [PATCH] slab: fix... |
2777 |
struct slab *slabp; |
b28a02de8 [PATCH] slab: fix... |
2778 2779 |
size_t offset; gfp_t local_flags; |
e498be7da [PATCH] Numa-awar... |
2780 |
struct kmem_list3 *l3; |
1da177e4c Linux-2.6.12-rc2 |
2781 |
|
a737b3e2f [PATCH] slab cleanup |
2782 2783 2784 |
/* * Be lazy and only check for valid flags here, keeping it out of the * critical path in kmem_cache_alloc(). |
1da177e4c Linux-2.6.12-rc2 |
2785 |
*/ |
6cb062296 Categorize GFP flags |
2786 2787 |
BUG_ON(flags & GFP_SLAB_BUG_MASK); local_flags = flags & (GFP_CONSTRAINT_MASK|GFP_RECLAIM_MASK); |
1da177e4c Linux-2.6.12-rc2 |
2788 |
|
2e1217cf9 [PATCH] NUMA slab... |
2789 |
/* Take the l3 list lock to change the colour_next on this node */ |
1da177e4c Linux-2.6.12-rc2 |
2790 |
check_irq_off(); |
2e1217cf9 [PATCH] NUMA slab... |
2791 2792 |
l3 = cachep->nodelists[nodeid]; spin_lock(&l3->list_lock); |
1da177e4c Linux-2.6.12-rc2 |
2793 2794 |
/* Get colour for the slab, and cal the next value. */ |
2e1217cf9 [PATCH] NUMA slab... |
2795 2796 2797 2798 2799 |
offset = l3->colour_next; l3->colour_next++; if (l3->colour_next >= cachep->colour) l3->colour_next = 0; spin_unlock(&l3->list_lock); |
1da177e4c Linux-2.6.12-rc2 |
2800 |
|
2e1217cf9 [PATCH] NUMA slab... |
2801 |
offset *= cachep->colour_off; |
1da177e4c Linux-2.6.12-rc2 |
2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812 |
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 [PATCH] slab cleanup |
2813 2814 2815 |
/* * Get mem for the objs. Attempt to allocate a physical page from * 'nodeid'. |
e498be7da [PATCH] Numa-awar... |
2816 |
*/ |
3c517a613 [PATCH] slab: bet... |
2817 |
if (!objp) |
b8c1c5da1 slab: correctly h... |
2818 |
objp = kmem_getpages(cachep, local_flags, nodeid); |
a737b3e2f [PATCH] slab cleanup |
2819 |
if (!objp) |
1da177e4c Linux-2.6.12-rc2 |
2820 2821 2822 |
goto failed; /* Get slab management. */ |
3c517a613 [PATCH] slab: bet... |
2823 |
slabp = alloc_slabmgmt(cachep, objp, offset, |
6cb062296 Categorize GFP flags |
2824 |
local_flags & ~GFP_CONSTRAINT_MASK, nodeid); |
a737b3e2f [PATCH] slab cleanup |
2825 |
if (!slabp) |
1da177e4c Linux-2.6.12-rc2 |
2826 |
goto opps1; |
4776874ff [PATCH] slab: pag... |
2827 |
slab_map_pages(cachep, slabp, objp); |
1da177e4c Linux-2.6.12-rc2 |
2828 |
|
a35afb830 Remove SLAB_CTOR_... |
2829 |
cache_init_objs(cachep, slabp); |
1da177e4c Linux-2.6.12-rc2 |
2830 2831 2832 2833 |
if (local_flags & __GFP_WAIT) local_irq_disable(); check_irq_off(); |
e498be7da [PATCH] Numa-awar... |
2834 |
spin_lock(&l3->list_lock); |
1da177e4c Linux-2.6.12-rc2 |
2835 2836 |
/* Make slab active. */ |
e498be7da [PATCH] Numa-awar... |
2837 |
list_add_tail(&slabp->list, &(l3->slabs_free)); |
1da177e4c Linux-2.6.12-rc2 |
2838 |
STATS_INC_GROWN(cachep); |
e498be7da [PATCH] Numa-awar... |
2839 2840 |
l3->free_objects += cachep->num; spin_unlock(&l3->list_lock); |
1da177e4c Linux-2.6.12-rc2 |
2841 |
return 1; |
a737b3e2f [PATCH] slab cleanup |
2842 |
opps1: |
1da177e4c Linux-2.6.12-rc2 |
2843 |
kmem_freepages(cachep, objp); |
a737b3e2f [PATCH] slab cleanup |
2844 |
failed: |
1da177e4c Linux-2.6.12-rc2 |
2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 |
if (local_flags & __GFP_WAIT) local_irq_disable(); return 0; } #if DEBUG /* * Perform extra freeing checks: * - detect bad pointers. * - POISON/RED_ZONE checking |
1da177e4c Linux-2.6.12-rc2 |
2856 2857 2858 |
*/ static void kfree_debugcheck(const void *objp) { |
1da177e4c Linux-2.6.12-rc2 |
2859 2860 2861 |
if (!virt_addr_valid(objp)) { printk(KERN_ERR "kfree_debugcheck: out of range ptr %lxh. ", |
b28a02de8 [PATCH] slab: fix... |
2862 2863 |
(unsigned long)objp); BUG(); |
1da177e4c Linux-2.6.12-rc2 |
2864 |
} |
1da177e4c Linux-2.6.12-rc2 |
2865 |
} |
58ce1fd58 [PATCH] slab: red... |
2866 2867 |
static inline void verify_redzone_free(struct kmem_cache *cache, void *obj) { |
b46b8f19c Increase slab red... |
2868 |
unsigned long long redzone1, redzone2; |
58ce1fd58 [PATCH] slab: red... |
2869 2870 2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 |
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 Increase slab red... |
2883 2884 |
printk(KERN_ERR "%p: redzone 1:0x%llx, redzone 2:0x%llx. ", |
58ce1fd58 [PATCH] slab: red... |
2885 2886 |
obj, redzone1, redzone2); } |
343e0d7a9 [PATCH] slab: rep... |
2887 |
static void *cache_free_debugcheck(struct kmem_cache *cachep, void *objp, |
b28a02de8 [PATCH] slab: fix... |
2888 |
void *caller) |
1da177e4c Linux-2.6.12-rc2 |
2889 2890 2891 2892 |
{ struct page *page; unsigned int objnr; struct slab *slabp; |
80cbd911c Fix kmem_cache_fr... |
2893 |
BUG_ON(virt_to_cache(objp) != cachep); |
3dafccf22 [PATCH] slab: dis... |
2894 |
objp -= obj_offset(cachep); |
1da177e4c Linux-2.6.12-rc2 |
2895 |
kfree_debugcheck(objp); |
b49af68ff Add virt_to_head_... |
2896 |
page = virt_to_head_page(objp); |
1da177e4c Linux-2.6.12-rc2 |
2897 |
|
065d41cb2 [PATCH] slab: con... |
2898 |
slabp = page_get_slab(page); |
1da177e4c Linux-2.6.12-rc2 |
2899 2900 |
if (cachep->flags & SLAB_RED_ZONE) { |
58ce1fd58 [PATCH] slab: red... |
2901 |
verify_redzone_free(cachep, objp); |
1da177e4c Linux-2.6.12-rc2 |
2902 2903 2904 2905 2906 |
*dbg_redzone1(cachep, objp) = RED_INACTIVE; *dbg_redzone2(cachep, objp) = RED_INACTIVE; } if (cachep->flags & SLAB_STORE_USER) *dbg_userword(cachep, objp) = caller; |
8fea4e96a [PATCH] slab: obj... |
2907 |
objnr = obj_to_index(cachep, slabp, objp); |
1da177e4c Linux-2.6.12-rc2 |
2908 2909 |
BUG_ON(objnr >= cachep->num); |
8fea4e96a [PATCH] slab: obj... |
2910 |
BUG_ON(objp != index_to_obj(cachep, slabp, objnr)); |
1da177e4c Linux-2.6.12-rc2 |
2911 |
|
871751e25 [PATCH] slab: imp... |
2912 2913 2914 |
#ifdef CONFIG_DEBUG_SLAB_LEAK slab_bufctl(slabp)[objnr] = BUFCTL_FREE; #endif |
1da177e4c Linux-2.6.12-rc2 |
2915 2916 |
if (cachep->flags & SLAB_POISON) { #ifdef CONFIG_DEBUG_PAGEALLOC |
a737b3e2f [PATCH] slab cleanup |
2917 |
if ((cachep->buffer_size % PAGE_SIZE)==0 && OFF_SLAB(cachep)) { |
1da177e4c Linux-2.6.12-rc2 |
2918 |
store_stackinfo(cachep, objp, (unsigned long)caller); |
b28a02de8 [PATCH] slab: fix... |
2919 |
kernel_map_pages(virt_to_page(objp), |
3dafccf22 [PATCH] slab: dis... |
2920 |
cachep->buffer_size / PAGE_SIZE, 0); |
1da177e4c Linux-2.6.12-rc2 |
2921 2922 2923 2924 2925 2926 2927 2928 2929 |
} else { poison_obj(cachep, objp, POISON_FREE); } #else poison_obj(cachep, objp, POISON_FREE); #endif } return objp; } |
343e0d7a9 [PATCH] slab: rep... |
2930 |
static void check_slabp(struct kmem_cache *cachep, struct slab *slabp) |
1da177e4c Linux-2.6.12-rc2 |
2931 2932 2933 |
{ kmem_bufctl_t i; int entries = 0; |
b28a02de8 [PATCH] slab: fix... |
2934 |
|
1da177e4c Linux-2.6.12-rc2 |
2935 2936 2937 2938 2939 2940 2941 |
/* 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 [PATCH] slab cleanup |
2942 2943 |
bad: printk(KERN_ERR "slab: Internal list corruption detected in " |
face37f5e slab: add taint f... |
2944 2945 2946 2947 |
"cache '%s'(%d), slabp %p(%d). Tainted(%s). Hexdump: ", cachep->name, cachep->num, slabp, slabp->inuse, print_tainted()); |
fdde6abb3 slab: use print_h... |
2948 2949 2950 |
print_hex_dump(KERN_ERR, "", DUMP_PREFIX_OFFSET, 16, 1, slabp, sizeof(*slabp) + cachep->num * sizeof(kmem_bufctl_t), 1); |
1da177e4c Linux-2.6.12-rc2 |
2951 2952 2953 2954 2955 2956 2957 2958 |
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 [PATCH] slab: rep... |
2959 |
static void *cache_alloc_refill(struct kmem_cache *cachep, gfp_t flags) |
1da177e4c Linux-2.6.12-rc2 |
2960 2961 2962 2963 |
{ int batchcount; struct kmem_list3 *l3; struct array_cache *ac; |
1ca4cb241 [PATCH] slab: red... |
2964 |
int node; |
6d2144d35 slab: NUMA slab a... |
2965 |
retry: |
1da177e4c Linux-2.6.12-rc2 |
2966 |
check_irq_off(); |
7d6e6d09d numa: slab: use n... |
2967 |
node = numa_mem_id(); |
9a2dba4b4 [PATCH] slab: ren... |
2968 |
ac = cpu_cache_get(cachep); |
1da177e4c Linux-2.6.12-rc2 |
2969 2970 |
batchcount = ac->batchcount; if (!ac->touched && batchcount > BATCHREFILL_LIMIT) { |
a737b3e2f [PATCH] slab cleanup |
2971 2972 2973 2974 |
/* * If there was little recent activity on this cache, then * perform only a partial refill. Otherwise we could generate * refill bouncing. |
1da177e4c Linux-2.6.12-rc2 |
2975 2976 2977 |
*/ batchcount = BATCHREFILL_LIMIT; } |
1ca4cb241 [PATCH] slab: red... |
2978 |
l3 = cachep->nodelists[node]; |
e498be7da [PATCH] Numa-awar... |
2979 2980 2981 |
BUG_ON(ac->avail > 0 || !l3); spin_lock(&l3->list_lock); |
1da177e4c Linux-2.6.12-rc2 |
2982 |
|
3ded175a4 [PATCH] slab: add... |
2983 |
/* See if we can refill from the shared array */ |
44b57f1cc slab: fix regress... |
2984 2985 |
if (l3->shared && transfer_objects(ac, l3->shared, batchcount)) { l3->shared->touched = 1; |
3ded175a4 [PATCH] slab: add... |
2986 |
goto alloc_done; |
44b57f1cc slab: fix regress... |
2987 |
} |
3ded175a4 [PATCH] slab: add... |
2988 |
|
1da177e4c Linux-2.6.12-rc2 |
2989 2990 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 |
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 slab: ensure cach... |
3004 3005 3006 3007 3008 3009 |
/* * The slab was either on partial or free list so * there must be at least one object available for * allocation. */ |
249b9f331 slab: unsigned sl... |
3010 |
BUG_ON(slabp->inuse >= cachep->num); |
714b8171a slab: ensure cach... |
3011 |
|
1da177e4c Linux-2.6.12-rc2 |
3012 |
while (slabp->inuse < cachep->num && batchcount--) { |
1da177e4c Linux-2.6.12-rc2 |
3013 3014 3015 |
STATS_INC_ALLOCED(cachep); STATS_INC_ACTIVE(cachep); STATS_SET_HIGH(cachep); |
78d382d77 [PATCH] slab: ext... |
3016 |
ac->entry[ac->avail++] = slab_get_obj(cachep, slabp, |
1ca4cb241 [PATCH] slab: red... |
3017 |
node); |
1da177e4c Linux-2.6.12-rc2 |
3018 3019 3020 3021 3022 3023 3024 3025 3026 3027 |
} 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 [PATCH] slab cleanup |
3028 |
must_grow: |
1da177e4c Linux-2.6.12-rc2 |
3029 |
l3->free_objects -= ac->avail; |
a737b3e2f [PATCH] slab cleanup |
3030 |
alloc_done: |
e498be7da [PATCH] Numa-awar... |
3031 |
spin_unlock(&l3->list_lock); |
1da177e4c Linux-2.6.12-rc2 |
3032 3033 3034 |
if (unlikely(!ac->avail)) { int x; |
3c517a613 [PATCH] slab: bet... |
3035 |
x = cache_grow(cachep, flags | GFP_THISNODE, node, NULL); |
e498be7da [PATCH] Numa-awar... |
3036 |
|
a737b3e2f [PATCH] slab cleanup |
3037 |
/* cache_grow can reenable interrupts, then ac could change. */ |
9a2dba4b4 [PATCH] slab: ren... |
3038 |
ac = cpu_cache_get(cachep); |
a737b3e2f [PATCH] slab cleanup |
3039 |
if (!x && ac->avail == 0) /* no objects in sight? abort */ |
1da177e4c Linux-2.6.12-rc2 |
3040 |
return NULL; |
a737b3e2f [PATCH] slab cleanup |
3041 |
if (!ac->avail) /* objects refilled by interrupt? */ |
1da177e4c Linux-2.6.12-rc2 |
3042 3043 3044 |
goto retry; } ac->touched = 1; |
e498be7da [PATCH] Numa-awar... |
3045 |
return ac->entry[--ac->avail]; |
1da177e4c Linux-2.6.12-rc2 |
3046 |
} |
a737b3e2f [PATCH] slab cleanup |
3047 3048 |
static inline void cache_alloc_debugcheck_before(struct kmem_cache *cachep, gfp_t flags) |
1da177e4c Linux-2.6.12-rc2 |
3049 3050 3051 3052 3053 3054 3055 3056 |
{ might_sleep_if(flags & __GFP_WAIT); #if DEBUG kmem_flagcheck(cachep, flags); #endif } #if DEBUG |
a737b3e2f [PATCH] slab cleanup |
3057 3058 |
static void *cache_alloc_debugcheck_after(struct kmem_cache *cachep, gfp_t flags, void *objp, void *caller) |
1da177e4c Linux-2.6.12-rc2 |
3059 |
{ |
b28a02de8 [PATCH] slab: fix... |
3060 |
if (!objp) |
1da177e4c Linux-2.6.12-rc2 |
3061 |
return objp; |
b28a02de8 [PATCH] slab: fix... |
3062 |
if (cachep->flags & SLAB_POISON) { |
1da177e4c Linux-2.6.12-rc2 |
3063 |
#ifdef CONFIG_DEBUG_PAGEALLOC |
3dafccf22 [PATCH] slab: dis... |
3064 |
if ((cachep->buffer_size % PAGE_SIZE) == 0 && OFF_SLAB(cachep)) |
b28a02de8 [PATCH] slab: fix... |
3065 |
kernel_map_pages(virt_to_page(objp), |
3dafccf22 [PATCH] slab: dis... |
3066 |
cachep->buffer_size / PAGE_SIZE, 1); |
1da177e4c Linux-2.6.12-rc2 |
3067 3068 3069 3070 3071 3072 3073 3074 3075 3076 3077 |
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 [PATCH] slab cleanup |
3078 3079 3080 3081 |
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 [PATCH] slab: fix... |
3082 |
printk(KERN_ERR |
b46b8f19c Increase slab red... |
3083 3084 |
"%p: redzone 1:0x%llx, redzone 2:0x%llx ", |
a737b3e2f [PATCH] slab cleanup |
3085 3086 |
objp, *dbg_redzone1(cachep, objp), *dbg_redzone2(cachep, objp)); |
1da177e4c Linux-2.6.12-rc2 |
3087 3088 3089 3090 |
} *dbg_redzone1(cachep, objp) = RED_ACTIVE; *dbg_redzone2(cachep, objp) = RED_ACTIVE; } |
871751e25 [PATCH] slab: imp... |
3091 3092 3093 3094 |
#ifdef CONFIG_DEBUG_SLAB_LEAK { struct slab *slabp; unsigned objnr; |
b49af68ff Add virt_to_head_... |
3095 |
slabp = page_get_slab(virt_to_head_page(objp)); |
871751e25 [PATCH] slab: imp... |
3096 3097 3098 3099 |
objnr = (unsigned)(objp - slabp->s_mem) / cachep->buffer_size; slab_bufctl(slabp)[objnr] = BUFCTL_ACTIVE; } #endif |
3dafccf22 [PATCH] slab: dis... |
3100 |
objp += obj_offset(cachep); |
4f1049345 slab allocators: ... |
3101 |
if (cachep->ctor && cachep->flags & SLAB_POISON) |
51cc50685 SL*B: drop kmem c... |
3102 |
cachep->ctor(objp); |
7ea466f22 slab: fix DEBUG_S... |
3103 3104 |
if (ARCH_SLAB_MINALIGN && ((unsigned long)objp & (ARCH_SLAB_MINALIGN-1))) { |
a44b56d35 [PATCH] slab debu... |
3105 3106 |
printk(KERN_ERR "0x%p: not aligned to ARCH_SLAB_MINALIGN=%d ", |
c225150b8 slab: fix DEBUG_S... |
3107 |
objp, (int)ARCH_SLAB_MINALIGN); |
a44b56d35 [PATCH] slab debu... |
3108 |
} |
1da177e4c Linux-2.6.12-rc2 |
3109 3110 3111 3112 3113 |
return objp; } #else #define cache_alloc_debugcheck_after(a,b,objp,d) (objp) #endif |
773ff60e8 SLUB: failslab su... |
3114 |
static bool slab_should_failslab(struct kmem_cache *cachep, gfp_t flags) |
8a8b6502f [PATCH] fault-inj... |
3115 3116 |
{ if (cachep == &cache_cache) |
773ff60e8 SLUB: failslab su... |
3117 |
return false; |
8a8b6502f [PATCH] fault-inj... |
3118 |
|
4c13dd3b4 failslab: add abi... |
3119 |
return should_failslab(obj_size(cachep), flags, cachep->flags); |
8a8b6502f [PATCH] fault-inj... |
3120 |
} |
343e0d7a9 [PATCH] slab: rep... |
3121 |
static inline void *____cache_alloc(struct kmem_cache *cachep, gfp_t flags) |
1da177e4c Linux-2.6.12-rc2 |
3122 |
{ |
b28a02de8 [PATCH] slab: fix... |
3123 |
void *objp; |
1da177e4c Linux-2.6.12-rc2 |
3124 |
struct array_cache *ac; |
5c3823008 [PATCH] kmalloc_n... |
3125 |
check_irq_off(); |
8a8b6502f [PATCH] fault-inj... |
3126 |
|
9a2dba4b4 [PATCH] slab: ren... |
3127 |
ac = cpu_cache_get(cachep); |
1da177e4c Linux-2.6.12-rc2 |
3128 3129 3130 |
if (likely(ac->avail)) { STATS_INC_ALLOCHIT(cachep); ac->touched = 1; |
e498be7da [PATCH] Numa-awar... |
3131 |
objp = ac->entry[--ac->avail]; |
1da177e4c Linux-2.6.12-rc2 |
3132 3133 3134 |
} else { STATS_INC_ALLOCMISS(cachep); objp = cache_alloc_refill(cachep, flags); |
ddbf2e836 slab, kmemleak: p... |
3135 3136 3137 3138 3139 |
/* * the 'ac' may be updated by cache_alloc_refill(), * and kmemleak_erase() requires its correct value. */ ac = cpu_cache_get(cachep); |
1da177e4c Linux-2.6.12-rc2 |
3140 |
} |
d5cff6352 kmemleak: Add the... |
3141 3142 3143 3144 3145 |
/* * 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 slab, kmemleak: s... |
3146 3147 |
if (objp) kmemleak_erase(&ac->entry[ac->avail]); |
5c3823008 [PATCH] kmalloc_n... |
3148 3149 |
return objp; } |
e498be7da [PATCH] Numa-awar... |
3150 3151 |
#ifdef CONFIG_NUMA /* |
b2455396b [PATCH] cpuset: m... |
3152 |
* Try allocating on another node if PF_SPREAD_SLAB|PF_MEMPOLICY. |
c61afb181 [PATCH] cpuset me... |
3153 3154 3155 3156 3157 3158 3159 |
* * 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 [PATCH] GFP_THISN... |
3160 |
if (in_interrupt() || (flags & __GFP_THISNODE)) |
c61afb181 [PATCH] cpuset me... |
3161 |
return NULL; |
7d6e6d09d numa: slab: use n... |
3162 |
nid_alloc = nid_here = numa_mem_id(); |
c0ff7453b cpuset,mm: fix no... |
3163 |
get_mems_allowed(); |
c61afb181 [PATCH] cpuset me... |
3164 |
if (cpuset_do_slab_mem_spread() && (cachep->flags & SLAB_MEM_SPREAD)) |
6adef3ebe cpusets: new roun... |
3165 |
nid_alloc = cpuset_slab_spread_node(); |
c61afb181 [PATCH] cpuset me... |
3166 3167 |
else if (current->mempolicy) nid_alloc = slab_node(current->mempolicy); |
c0ff7453b cpuset,mm: fix no... |
3168 |
put_mems_allowed(); |
c61afb181 [PATCH] cpuset me... |
3169 |
if (nid_alloc != nid_here) |
8b98c1699 [PATCH] leak trac... |
3170 |
return ____cache_alloc_node(cachep, flags, nid_alloc); |
c61afb181 [PATCH] cpuset me... |
3171 3172 3173 3174 |
return NULL; } /* |
765c4507a [PATCH] GFP_THISN... |
3175 |
* Fallback function if there was no memory available and no objects on a |
3c517a613 [PATCH] slab: bet... |
3176 3177 3178 3179 3180 |
* 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 [PATCH] GFP_THISN... |
3181 |
*/ |
8c8cc2c10 [PATCH] slab: cac... |
3182 |
static void *fallback_alloc(struct kmem_cache *cache, gfp_t flags) |
765c4507a [PATCH] GFP_THISN... |
3183 |
{ |
8c8cc2c10 [PATCH] slab: cac... |
3184 3185 |
struct zonelist *zonelist; gfp_t local_flags; |
dd1a239f6 mm: have zonelist... |
3186 |
struct zoneref *z; |
54a6eb5c4 mm: use two zonel... |
3187 3188 |
struct zone *zone; enum zone_type high_zoneidx = gfp_zone(flags); |
765c4507a [PATCH] GFP_THISN... |
3189 |
void *obj = NULL; |
3c517a613 [PATCH] slab: bet... |
3190 |
int nid; |
8c8cc2c10 [PATCH] slab: cac... |
3191 3192 3193 |
if (flags & __GFP_THISNODE) return NULL; |
c0ff7453b cpuset,mm: fix no... |
3194 |
get_mems_allowed(); |
0e88460da mm: introduce nod... |
3195 |
zonelist = node_zonelist(slab_node(current->mempolicy), flags); |
6cb062296 Categorize GFP flags |
3196 |
local_flags = flags & (GFP_CONSTRAINT_MASK|GFP_RECLAIM_MASK); |
765c4507a [PATCH] GFP_THISN... |
3197 |
|
3c517a613 [PATCH] slab: bet... |
3198 3199 3200 3201 3202 |
retry: /* * Look through allowed nodes for objects available * from existing per node queues. */ |
54a6eb5c4 mm: use two zonel... |
3203 3204 |
for_each_zone_zonelist(zone, z, zonelist, high_zoneidx) { nid = zone_to_nid(zone); |
aedb0eb10 [PATCH] Slab: Do ... |
3205 |
|
54a6eb5c4 mm: use two zonel... |
3206 |
if (cpuset_zone_allowed_hardwall(zone, flags) && |
3c517a613 [PATCH] slab: bet... |
3207 |
cache->nodelists[nid] && |
481c5346d Slab: Fix memory ... |
3208 |
cache->nodelists[nid]->free_objects) { |
3c517a613 [PATCH] slab: bet... |
3209 3210 |
obj = ____cache_alloc_node(cache, flags | GFP_THISNODE, nid); |
481c5346d Slab: Fix memory ... |
3211 3212 3213 |
if (obj) break; } |
3c517a613 [PATCH] slab: bet... |
3214 |
} |
cfce66047 Slab allocators: ... |
3215 |
if (!obj) { |
3c517a613 [PATCH] slab: bet... |
3216 3217 3218 3219 3220 3221 |
/* * 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 [PATCH] slab: fix... |
3222 3223 3224 |
if (local_flags & __GFP_WAIT) local_irq_enable(); kmem_flagcheck(cache, flags); |
7d6e6d09d numa: slab: use n... |
3225 |
obj = kmem_getpages(cache, local_flags, numa_mem_id()); |
dd47ea755 [PATCH] slab: fix... |
3226 3227 |
if (local_flags & __GFP_WAIT) local_irq_disable(); |
3c517a613 [PATCH] slab: bet... |
3228 3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242 3243 |
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 [PATCH] fix BUG_O... |
3244 |
/* cache_grow already freed obj */ |
3c517a613 [PATCH] slab: bet... |
3245 3246 3247 |
obj = NULL; } } |
aedb0eb10 [PATCH] Slab: Do ... |
3248 |
} |
c0ff7453b cpuset,mm: fix no... |
3249 |
put_mems_allowed(); |
765c4507a [PATCH] GFP_THISN... |
3250 3251 3252 3253 |
return obj; } /* |
e498be7da [PATCH] Numa-awar... |
3254 |
* A interface to enable slab creation on nodeid |
1da177e4c Linux-2.6.12-rc2 |
3255 |
*/ |
8b98c1699 [PATCH] leak trac... |
3256 |
static void *____cache_alloc_node(struct kmem_cache *cachep, gfp_t flags, |
a737b3e2f [PATCH] slab cleanup |
3257 |
int nodeid) |
e498be7da [PATCH] Numa-awar... |
3258 3259 |
{ struct list_head *entry; |
b28a02de8 [PATCH] slab: fix... |
3260 3261 3262 |
struct slab *slabp; struct kmem_list3 *l3; void *obj; |
b28a02de8 [PATCH] slab: fix... |
3263 3264 3265 3266 |
int x; l3 = cachep->nodelists[nodeid]; BUG_ON(!l3); |
a737b3e2f [PATCH] slab cleanup |
3267 |
retry: |
ca3b9b917 [PATCH] NUMA slab... |
3268 |
check_irq_off(); |
b28a02de8 [PATCH] slab: fix... |
3269 3270 3271 3272 3273 3274 3275 3276 3277 3278 3279 3280 3281 3282 3283 3284 3285 3286 |
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 [PATCH] slab: ext... |
3287 |
obj = slab_get_obj(cachep, slabp, nodeid); |
b28a02de8 [PATCH] slab: fix... |
3288 3289 3290 3291 |
check_slabp(cachep, slabp); l3->free_objects--; /* move slabp to correct slabp list: */ list_del(&slabp->list); |
a737b3e2f [PATCH] slab cleanup |
3292 |
if (slabp->free == BUFCTL_END) |
b28a02de8 [PATCH] slab: fix... |
3293 |
list_add(&slabp->list, &l3->slabs_full); |
a737b3e2f [PATCH] slab cleanup |
3294 |
else |
b28a02de8 [PATCH] slab: fix... |
3295 |
list_add(&slabp->list, &l3->slabs_partial); |
e498be7da [PATCH] Numa-awar... |
3296 |
|
b28a02de8 [PATCH] slab: fix... |
3297 3298 |
spin_unlock(&l3->list_lock); goto done; |
e498be7da [PATCH] Numa-awar... |
3299 |
|
a737b3e2f [PATCH] slab cleanup |
3300 |
must_grow: |
b28a02de8 [PATCH] slab: fix... |
3301 |
spin_unlock(&l3->list_lock); |
3c517a613 [PATCH] slab: bet... |
3302 |
x = cache_grow(cachep, flags | GFP_THISNODE, nodeid, NULL); |
765c4507a [PATCH] GFP_THISN... |
3303 3304 |
if (x) goto retry; |
1da177e4c Linux-2.6.12-rc2 |
3305 |
|
8c8cc2c10 [PATCH] slab: cac... |
3306 |
return fallback_alloc(cachep, flags); |
e498be7da [PATCH] Numa-awar... |
3307 |
|
a737b3e2f [PATCH] slab cleanup |
3308 |
done: |
b28a02de8 [PATCH] slab: fix... |
3309 |
return obj; |
e498be7da [PATCH] Numa-awar... |
3310 |
} |
8c8cc2c10 [PATCH] slab: cac... |
3311 3312 3313 3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325 3326 3327 3328 3329 |
/** * kmem_cache_alloc_node - Allocate an object on the specified node * @cachep: The cache to allocate from. * @flags: See kmalloc(). * @nodeid: node number of the target node. * @caller: return address of caller, used for debug information * * Identical to kmem_cache_alloc but it will allocate memory on the given * node, which can improve the performance for cpu bound structures. * * Fallback to other node is possible if __GFP_THISNODE is not set. */ static __always_inline void * __cache_alloc_node(struct kmem_cache *cachep, gfp_t flags, int nodeid, void *caller) { unsigned long save_flags; void *ptr; |
7d6e6d09d numa: slab: use n... |
3330 |
int slab_node = numa_mem_id(); |
8c8cc2c10 [PATCH] slab: cac... |
3331 |
|
dcce284a2 mm: Extend gfp ma... |
3332 |
flags &= gfp_allowed_mask; |
7e85ee0c1 slab,slub: don't ... |
3333 |
|
cf40bd16f lockdep: annotate... |
3334 |
lockdep_trace_alloc(flags); |
773ff60e8 SLUB: failslab su... |
3335 |
if (slab_should_failslab(cachep, flags)) |
824ebef12 fault injection: ... |
3336 |
return NULL; |
8c8cc2c10 [PATCH] slab: cac... |
3337 3338 |
cache_alloc_debugcheck_before(cachep, flags); local_irq_save(save_flags); |
eacbbae38 slab: use NUMA_NO... |
3339 |
if (nodeid == NUMA_NO_NODE) |
7d6e6d09d numa: slab: use n... |
3340 |
nodeid = slab_node; |
8c8cc2c10 [PATCH] slab: cac... |
3341 3342 3343 3344 3345 3346 |
if (unlikely(!cachep->nodelists[nodeid])) { /* Node not bootstrapped yet */ ptr = fallback_alloc(cachep, flags); goto out; } |
7d6e6d09d numa: slab: use n... |
3347 |
if (nodeid == slab_node) { |
8c8cc2c10 [PATCH] slab: cac... |
3348 3349 3350 3351 3352 3353 3354 3355 3356 3357 3358 3359 3360 3361 3362 |
/* * 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 kmemleak: Add the... |
3363 3364 |
kmemleak_alloc_recursive(ptr, obj_size(cachep), 1, cachep->flags, flags); |
8c8cc2c10 [PATCH] slab: cac... |
3365 |
|
c175eea46 slab: add hooks f... |
3366 3367 |
if (likely(ptr)) kmemcheck_slab_alloc(cachep, flags, ptr, obj_size(cachep)); |
d07dbea46 Slab allocators: ... |
3368 3369 |
if (unlikely((flags & __GFP_ZERO) && ptr)) memset(ptr, 0, obj_size(cachep)); |
8c8cc2c10 [PATCH] slab: cac... |
3370 3371 3372 3373 3374 3375 3376 3377 3378 3379 3380 3381 3382 3383 3384 3385 3386 3387 3388 |
return ptr; } static __always_inline void * __do_cache_alloc(struct kmem_cache *cache, gfp_t flags) { void *objp; if (unlikely(current->flags & (PF_SPREAD_SLAB | PF_MEMPOLICY))) { objp = alternate_node_alloc(cache, flags); if (objp) goto out; } objp = ____cache_alloc(cache, flags); /* * We may just have run out of memory on the local node. * ____cache_alloc_node() knows how to locate memory on other nodes */ |
7d6e6d09d numa: slab: use n... |
3389 3390 |
if (!objp) objp = ____cache_alloc_node(cache, flags, numa_mem_id()); |
8c8cc2c10 [PATCH] slab: cac... |
3391 3392 3393 3394 3395 3396 3397 3398 3399 3400 3401 3402 3403 3404 3405 3406 3407 3408 3409 |
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 mm: Extend gfp ma... |
3410 |
flags &= gfp_allowed_mask; |
7e85ee0c1 slab,slub: don't ... |
3411 |
|
cf40bd16f lockdep: annotate... |
3412 |
lockdep_trace_alloc(flags); |
773ff60e8 SLUB: failslab su... |
3413 |
if (slab_should_failslab(cachep, flags)) |
824ebef12 fault injection: ... |
3414 |
return NULL; |
8c8cc2c10 [PATCH] slab: cac... |
3415 3416 3417 3418 3419 |
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 kmemleak: Add the... |
3420 3421 |
kmemleak_alloc_recursive(objp, obj_size(cachep), 1, cachep->flags, flags); |
8c8cc2c10 [PATCH] slab: cac... |
3422 |
prefetchw(objp); |
c175eea46 slab: add hooks f... |
3423 3424 |
if (likely(objp)) kmemcheck_slab_alloc(cachep, flags, objp, obj_size(cachep)); |
d07dbea46 Slab allocators: ... |
3425 3426 |
if (unlikely((flags & __GFP_ZERO) && objp)) memset(objp, 0, obj_size(cachep)); |
8c8cc2c10 [PATCH] slab: cac... |
3427 3428 |
return objp; } |
e498be7da [PATCH] Numa-awar... |
3429 3430 3431 3432 |
/* * Caller needs to acquire correct kmem_list's list_lock */ |
343e0d7a9 [PATCH] slab: rep... |
3433 |
static void free_block(struct kmem_cache *cachep, void **objpp, int nr_objects, |
b28a02de8 [PATCH] slab: fix... |
3434 |
int node) |
1da177e4c Linux-2.6.12-rc2 |
3435 3436 |
{ int i; |
e498be7da [PATCH] Numa-awar... |
3437 |
struct kmem_list3 *l3; |
1da177e4c Linux-2.6.12-rc2 |
3438 3439 3440 3441 |
for (i = 0; i < nr_objects; i++) { void *objp = objpp[i]; struct slab *slabp; |
1da177e4c Linux-2.6.12-rc2 |
3442 |
|
6ed5eb221 [PATCH] slab: ext... |
3443 |
slabp = virt_to_slab(objp); |
ff69416e6 [PATCH] slab: fix... |
3444 |
l3 = cachep->nodelists[node]; |
1da177e4c Linux-2.6.12-rc2 |
3445 |
list_del(&slabp->list); |
ff69416e6 [PATCH] slab: fix... |
3446 |
check_spinlock_acquired_node(cachep, node); |
1da177e4c Linux-2.6.12-rc2 |
3447 |
check_slabp(cachep, slabp); |
78d382d77 [PATCH] slab: ext... |
3448 |
slab_put_obj(cachep, slabp, objp, node); |
1da177e4c Linux-2.6.12-rc2 |
3449 |
STATS_DEC_ACTIVE(cachep); |
e498be7da [PATCH] Numa-awar... |
3450 |
l3->free_objects++; |
1da177e4c Linux-2.6.12-rc2 |
3451 3452 3453 3454 |
check_slabp(cachep, slabp); /* fixup slab chains */ if (slabp->inuse == 0) { |
e498be7da [PATCH] Numa-awar... |
3455 3456 |
if (l3->free_objects > l3->free_limit) { l3->free_objects -= cachep->num; |
e5ac9c5ae [PATCH] Add some ... |
3457 3458 3459 3460 3461 3462 |
/* 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 Linux-2.6.12-rc2 |
3463 3464 |
slab_destroy(cachep, slabp); } else { |
e498be7da [PATCH] Numa-awar... |
3465 |
list_add(&slabp->list, &l3->slabs_free); |
1da177e4c Linux-2.6.12-rc2 |
3466 3467 3468 3469 3470 3471 |
} } 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 [PATCH] Numa-awar... |
3472 |
list_add_tail(&slabp->list, &l3->slabs_partial); |
1da177e4c Linux-2.6.12-rc2 |
3473 3474 3475 |
} } } |
343e0d7a9 [PATCH] slab: rep... |
3476 |
static void cache_flusharray(struct kmem_cache *cachep, struct array_cache *ac) |
1da177e4c Linux-2.6.12-rc2 |
3477 3478 |
{ int batchcount; |
e498be7da [PATCH] Numa-awar... |
3479 |
struct kmem_list3 *l3; |
7d6e6d09d numa: slab: use n... |
3480 |
int node = numa_mem_id(); |
1da177e4c Linux-2.6.12-rc2 |
3481 3482 3483 3484 3485 3486 |
batchcount = ac->batchcount; #if DEBUG BUG_ON(!batchcount || batchcount > ac->avail); #endif check_irq_off(); |
ff69416e6 [PATCH] slab: fix... |
3487 |
l3 = cachep->nodelists[node]; |
873623dfa [PATCH] lockdep: ... |
3488 |
spin_lock(&l3->list_lock); |
e498be7da [PATCH] Numa-awar... |
3489 3490 |
if (l3->shared) { struct array_cache *shared_array = l3->shared; |
b28a02de8 [PATCH] slab: fix... |
3491 |
int max = shared_array->limit - shared_array->avail; |
1da177e4c Linux-2.6.12-rc2 |
3492 3493 3494 |
if (max) { if (batchcount > max) batchcount = max; |
e498be7da [PATCH] Numa-awar... |
3495 |
memcpy(&(shared_array->entry[shared_array->avail]), |
b28a02de8 [PATCH] slab: fix... |
3496 |
ac->entry, sizeof(void *) * batchcount); |
1da177e4c Linux-2.6.12-rc2 |
3497 3498 3499 3500 |
shared_array->avail += batchcount; goto free_done; } } |
ff69416e6 [PATCH] slab: fix... |
3501 |
free_block(cachep, ac->entry, batchcount, node); |
a737b3e2f [PATCH] slab cleanup |
3502 |
free_done: |
1da177e4c Linux-2.6.12-rc2 |
3503 3504 3505 3506 |
#if STATS { int i = 0; struct list_head *p; |
e498be7da [PATCH] Numa-awar... |
3507 3508 |
p = l3->slabs_free.next; while (p != &(l3->slabs_free)) { |
1da177e4c Linux-2.6.12-rc2 |
3509 3510 3511 3512 3513 3514 3515 3516 3517 3518 3519 |
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 [PATCH] Numa-awar... |
3520 |
spin_unlock(&l3->list_lock); |
1da177e4c Linux-2.6.12-rc2 |
3521 |
ac->avail -= batchcount; |
a737b3e2f [PATCH] slab cleanup |
3522 |
memmove(ac->entry, &(ac->entry[batchcount]), sizeof(void *)*ac->avail); |
1da177e4c Linux-2.6.12-rc2 |
3523 3524 3525 |
} /* |
a737b3e2f [PATCH] slab cleanup |
3526 3527 |
* 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 Linux-2.6.12-rc2 |
3528 |
*/ |
a947eb95e SLAB: Record actu... |
3529 3530 |
static inline void __cache_free(struct kmem_cache *cachep, void *objp, void *caller) |
1da177e4c Linux-2.6.12-rc2 |
3531 |
{ |
9a2dba4b4 [PATCH] slab: ren... |
3532 |
struct array_cache *ac = cpu_cache_get(cachep); |
1da177e4c Linux-2.6.12-rc2 |
3533 3534 |
check_irq_off(); |
d5cff6352 kmemleak: Add the... |
3535 |
kmemleak_free_recursive(objp, cachep->flags); |
a947eb95e SLAB: Record actu... |
3536 |
objp = cache_free_debugcheck(cachep, objp, caller); |
1da177e4c Linux-2.6.12-rc2 |
3537 |
|
c175eea46 slab: add hooks f... |
3538 |
kmemcheck_slab_free(cachep, objp, obj_size(cachep)); |
1807a1aaf slab: skip callin... |
3539 3540 3541 3542 3543 3544 3545 |
/* * 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 page allocator: s... |
3546 |
if (nr_online_nodes > 1 && cache_free_alien(cachep, objp)) |
729bd0b74 [PATCH] slab: ext... |
3547 |
return; |
1da177e4c Linux-2.6.12-rc2 |
3548 3549 |
if (likely(ac->avail < ac->limit)) { STATS_INC_FREEHIT(cachep); |
e498be7da [PATCH] Numa-awar... |
3550 |
ac->entry[ac->avail++] = objp; |
1da177e4c Linux-2.6.12-rc2 |
3551 3552 3553 3554 |
return; } else { STATS_INC_FREEMISS(cachep); cache_flusharray(cachep, ac); |
e498be7da [PATCH] Numa-awar... |
3555 |
ac->entry[ac->avail++] = objp; |
1da177e4c Linux-2.6.12-rc2 |
3556 3557 3558 3559 3560 3561 3562 3563 3564 3565 3566 |
} } /** * 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 [PATCH] slab: rep... |
3567 |
void *kmem_cache_alloc(struct kmem_cache *cachep, gfp_t flags) |
1da177e4c Linux-2.6.12-rc2 |
3568 |
{ |
36555751c kmemtrace: SLAB h... |
3569 |
void *ret = __cache_alloc(cachep, flags, __builtin_return_address(0)); |
ca2b84cb3 kmemtrace: use tr... |
3570 3571 |
trace_kmem_cache_alloc(_RET_IP_, ret, obj_size(cachep), cachep->buffer_size, flags); |
36555751c kmemtrace: SLAB h... |
3572 3573 |
return ret; |
1da177e4c Linux-2.6.12-rc2 |
3574 3575 |
} EXPORT_SYMBOL(kmem_cache_alloc); |
0f24f1287 tracing, slab: De... |
3576 |
#ifdef CONFIG_TRACING |
85beb5869 tracing/slab: Mov... |
3577 3578 |
void * kmem_cache_alloc_trace(size_t size, struct kmem_cache *cachep, gfp_t flags) |
36555751c kmemtrace: SLAB h... |
3579 |
{ |
85beb5869 tracing/slab: Mov... |
3580 3581 3582 3583 3584 3585 3586 |
void *ret; ret = __cache_alloc(cachep, flags, __builtin_return_address(0)); trace_kmalloc(_RET_IP_, ret, size, slab_buffer_size(cachep), flags); return ret; |
36555751c kmemtrace: SLAB h... |
3587 |
} |
85beb5869 tracing/slab: Mov... |
3588 |
EXPORT_SYMBOL(kmem_cache_alloc_trace); |
36555751c kmemtrace: SLAB h... |
3589 |
#endif |
1da177e4c Linux-2.6.12-rc2 |
3590 |
#ifdef CONFIG_NUMA |
8b98c1699 [PATCH] leak trac... |
3591 3592 |
void *kmem_cache_alloc_node(struct kmem_cache *cachep, gfp_t flags, int nodeid) { |
36555751c kmemtrace: SLAB h... |
3593 3594 |
void *ret = __cache_alloc_node(cachep, flags, nodeid, __builtin_return_address(0)); |
ca2b84cb3 kmemtrace: use tr... |
3595 3596 3597 |
trace_kmem_cache_alloc_node(_RET_IP_, ret, obj_size(cachep), cachep->buffer_size, flags, nodeid); |
36555751c kmemtrace: SLAB h... |
3598 3599 |
return ret; |
8b98c1699 [PATCH] leak trac... |
3600 |
} |
1da177e4c Linux-2.6.12-rc2 |
3601 |
EXPORT_SYMBOL(kmem_cache_alloc_node); |
0f24f1287 tracing, slab: De... |
3602 |
#ifdef CONFIG_TRACING |
85beb5869 tracing/slab: Mov... |
3603 3604 3605 3606 |
void *kmem_cache_alloc_node_trace(size_t size, struct kmem_cache *cachep, gfp_t flags, int nodeid) |
36555751c kmemtrace: SLAB h... |
3607 |
{ |
85beb5869 tracing/slab: Mov... |
3608 3609 3610 |
void *ret; ret = __cache_alloc_node(cachep, flags, nodeid, |
36555751c kmemtrace: SLAB h... |
3611 |
__builtin_return_address(0)); |
85beb5869 tracing/slab: Mov... |
3612 3613 3614 3615 |
trace_kmalloc_node(_RET_IP_, ret, size, slab_buffer_size(cachep), flags, nodeid); return ret; |
36555751c kmemtrace: SLAB h... |
3616 |
} |
85beb5869 tracing/slab: Mov... |
3617 |
EXPORT_SYMBOL(kmem_cache_alloc_node_trace); |
36555751c kmemtrace: SLAB h... |
3618 |
#endif |
8b98c1699 [PATCH] leak trac... |
3619 3620 |
static __always_inline void * __do_kmalloc_node(size_t size, gfp_t flags, int node, void *caller) |
97e2bde47 [PATCH] add kmall... |
3621 |
{ |
343e0d7a9 [PATCH] slab: rep... |
3622 |
struct kmem_cache *cachep; |
97e2bde47 [PATCH] add kmall... |
3623 3624 |
cachep = kmem_find_general_cachep(size, flags); |
6cb8f9132 Slab allocators: ... |
3625 3626 |
if (unlikely(ZERO_OR_NULL_PTR(cachep))) return cachep; |
85beb5869 tracing/slab: Mov... |
3627 |
return kmem_cache_alloc_node_trace(size, cachep, flags, node); |
97e2bde47 [PATCH] add kmall... |
3628 |
} |
8b98c1699 [PATCH] leak trac... |
3629 |
|
0bb38a5cd tracing, slab: Fi... |
3630 |
#if defined(CONFIG_DEBUG_SLAB) || defined(CONFIG_TRACING) |
8b98c1699 [PATCH] leak trac... |
3631 3632 3633 3634 3635 |
void *__kmalloc_node(size_t size, gfp_t flags, int node) { return __do_kmalloc_node(size, flags, node, __builtin_return_address(0)); } |
dbe5e69d2 [PATCH] slab: opt... |
3636 |
EXPORT_SYMBOL(__kmalloc_node); |
8b98c1699 [PATCH] leak trac... |
3637 3638 |
void *__kmalloc_node_track_caller(size_t size, gfp_t flags, |
ce71e27c6 SLUB: Replace __b... |
3639 |
int node, unsigned long caller) |
8b98c1699 [PATCH] leak trac... |
3640 |
{ |
ce71e27c6 SLUB: Replace __b... |
3641 |
return __do_kmalloc_node(size, flags, node, (void *)caller); |
8b98c1699 [PATCH] leak trac... |
3642 3643 3644 3645 3646 3647 3648 3649 |
} 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 tracing, slab: Fi... |
3650 |
#endif /* CONFIG_DEBUG_SLAB || CONFIG_TRACING */ |
8b98c1699 [PATCH] leak trac... |
3651 |
#endif /* CONFIG_NUMA */ |
1da177e4c Linux-2.6.12-rc2 |
3652 3653 |
/** |
800590f52 [PATCH] slab: kma... |
3654 |
* __do_kmalloc - allocate memory |
1da177e4c Linux-2.6.12-rc2 |
3655 |
* @size: how many bytes of memory are required. |
800590f52 [PATCH] slab: kma... |
3656 |
* @flags: the type of memory to allocate (see kmalloc). |
911851e6e [PATCH] slab: fix... |
3657 |
* @caller: function caller for debug tracking of the caller |
1da177e4c Linux-2.6.12-rc2 |
3658 |
*/ |
7fd6b1413 [PATCH] slab: fix... |
3659 3660 |
static __always_inline void *__do_kmalloc(size_t size, gfp_t flags, void *caller) |
1da177e4c Linux-2.6.12-rc2 |
3661 |
{ |
343e0d7a9 [PATCH] slab: rep... |
3662 |
struct kmem_cache *cachep; |
36555751c kmemtrace: SLAB h... |
3663 |
void *ret; |
1da177e4c Linux-2.6.12-rc2 |
3664 |
|
97e2bde47 [PATCH] add kmall... |
3665 3666 3667 3668 3669 3670 |
/* 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 Fix up non-NUMA S... |
3671 3672 |
if (unlikely(ZERO_OR_NULL_PTR(cachep))) return cachep; |
36555751c kmemtrace: SLAB h... |
3673 |
ret = __cache_alloc(cachep, flags, caller); |
ca2b84cb3 kmemtrace: use tr... |
3674 3675 |
trace_kmalloc((unsigned long) caller, ret, size, cachep->buffer_size, flags); |
36555751c kmemtrace: SLAB h... |
3676 3677 |
return ret; |
7fd6b1413 [PATCH] slab: fix... |
3678 |
} |
7fd6b1413 [PATCH] slab: fix... |
3679 |
|
0bb38a5cd tracing, slab: Fi... |
3680 |
#if defined(CONFIG_DEBUG_SLAB) || defined(CONFIG_TRACING) |
7fd6b1413 [PATCH] slab: fix... |
3681 3682 |
void *__kmalloc(size_t size, gfp_t flags) { |
871751e25 [PATCH] slab: imp... |
3683 |
return __do_kmalloc(size, flags, __builtin_return_address(0)); |
1da177e4c Linux-2.6.12-rc2 |
3684 3685 |
} EXPORT_SYMBOL(__kmalloc); |
ce71e27c6 SLUB: Replace __b... |
3686 |
void *__kmalloc_track_caller(size_t size, gfp_t flags, unsigned long caller) |
7fd6b1413 [PATCH] slab: fix... |
3687 |
{ |
ce71e27c6 SLUB: Replace __b... |
3688 |
return __do_kmalloc(size, flags, (void *)caller); |
7fd6b1413 [PATCH] slab: fix... |
3689 3690 |
} EXPORT_SYMBOL(__kmalloc_track_caller); |
1d2c8eea6 [PATCH] slab: cle... |
3691 3692 3693 3694 3695 3696 3697 |
#else void *__kmalloc(size_t size, gfp_t flags) { return __do_kmalloc(size, flags, NULL); } EXPORT_SYMBOL(__kmalloc); |
7fd6b1413 [PATCH] slab: fix... |
3698 |
#endif |
1da177e4c Linux-2.6.12-rc2 |
3699 3700 3701 3702 3703 3704 3705 3706 |
/** * 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 [PATCH] slab: rep... |
3707 |
void kmem_cache_free(struct kmem_cache *cachep, void *objp) |
1da177e4c Linux-2.6.12-rc2 |
3708 3709 3710 3711 |
{ unsigned long flags; local_irq_save(flags); |
898552c9d [PATCH] lockdep: ... |
3712 |
debug_check_no_locks_freed(objp, obj_size(cachep)); |
3ac7fe5a4 infrastructure to... |
3713 3714 |
if (!(cachep->flags & SLAB_DEBUG_OBJECTS)) debug_check_no_obj_freed(objp, obj_size(cachep)); |
a947eb95e SLAB: Record actu... |
3715 |
__cache_free(cachep, objp, __builtin_return_address(0)); |
1da177e4c Linux-2.6.12-rc2 |
3716 |
local_irq_restore(flags); |
36555751c kmemtrace: SLAB h... |
3717 |
|
ca2b84cb3 kmemtrace: use tr... |
3718 |
trace_kmem_cache_free(_RET_IP_, objp); |
1da177e4c Linux-2.6.12-rc2 |
3719 3720 3721 3722 |
} EXPORT_SYMBOL(kmem_cache_free); /** |
1da177e4c Linux-2.6.12-rc2 |
3723 3724 3725 |
* kfree - free previously allocated memory * @objp: pointer returned by kmalloc. * |
80e93effc [PATCH] update kf... |
3726 3727 |
* If @objp is NULL, no operation is performed. * |
1da177e4c Linux-2.6.12-rc2 |
3728 3729 3730 3731 3732 |
* Don't free memory not originally allocated by kmalloc() * or you will run into trouble. */ void kfree(const void *objp) { |
343e0d7a9 [PATCH] slab: rep... |
3733 |
struct kmem_cache *c; |
1da177e4c Linux-2.6.12-rc2 |
3734 |
unsigned long flags; |
2121db74b kmemtrace: trace ... |
3735 |
trace_kfree(_RET_IP_, objp); |
6cb8f9132 Slab allocators: ... |
3736 |
if (unlikely(ZERO_OR_NULL_PTR(objp))) |
1da177e4c Linux-2.6.12-rc2 |
3737 3738 3739 |
return; local_irq_save(flags); kfree_debugcheck(objp); |
6ed5eb221 [PATCH] slab: ext... |
3740 |
c = virt_to_cache(objp); |
f9b8404cf [PATCH] pi-futex:... |
3741 |
debug_check_no_locks_freed(objp, obj_size(c)); |
3ac7fe5a4 infrastructure to... |
3742 |
debug_check_no_obj_freed(objp, obj_size(c)); |
a947eb95e SLAB: Record actu... |
3743 |
__cache_free(c, (void *)objp, __builtin_return_address(0)); |
1da177e4c Linux-2.6.12-rc2 |
3744 3745 3746 |
local_irq_restore(flags); } EXPORT_SYMBOL(kfree); |
343e0d7a9 [PATCH] slab: rep... |
3747 |
unsigned int kmem_cache_size(struct kmem_cache *cachep) |
1da177e4c Linux-2.6.12-rc2 |
3748 |
{ |
3dafccf22 [PATCH] slab: dis... |
3749 |
return obj_size(cachep); |
1da177e4c Linux-2.6.12-rc2 |
3750 3751 |
} EXPORT_SYMBOL(kmem_cache_size); |
e498be7da [PATCH] Numa-awar... |
3752 |
/* |
183ff22bb spelling fixes: mm/ |
3753 |
* This initializes kmem_list3 or resizes various caches for all nodes. |
e498be7da [PATCH] Numa-awar... |
3754 |
*/ |
83b519e8b slab: setup alloc... |
3755 |
static int alloc_kmemlist(struct kmem_cache *cachep, gfp_t gfp) |
e498be7da [PATCH] Numa-awar... |
3756 3757 3758 |
{ int node; struct kmem_list3 *l3; |
cafeb02e0 [PATCH] alloc_kme... |
3759 |
struct array_cache *new_shared; |
3395ee058 [PATCH] mm: add n... |
3760 |
struct array_cache **new_alien = NULL; |
e498be7da [PATCH] Numa-awar... |
3761 |
|
9c09a95cf slab: partially r... |
3762 |
for_each_online_node(node) { |
cafeb02e0 [PATCH] alloc_kme... |
3763 |
|
3395ee058 [PATCH] mm: add n... |
3764 |
if (use_alien_caches) { |
83b519e8b slab: setup alloc... |
3765 |
new_alien = alloc_alien_cache(node, cachep->limit, gfp); |
3395ee058 [PATCH] mm: add n... |
3766 3767 3768 |
if (!new_alien) goto fail; } |
cafeb02e0 [PATCH] alloc_kme... |
3769 |
|
631098469 SLAB: don't alloc... |
3770 3771 3772 |
new_shared = NULL; if (cachep->shared) { new_shared = alloc_arraycache(node, |
0718dc2a8 [PATCH] slab: fix... |
3773 |
cachep->shared*cachep->batchcount, |
83b519e8b slab: setup alloc... |
3774 |
0xbaadf00d, gfp); |
631098469 SLAB: don't alloc... |
3775 3776 3777 3778 |
if (!new_shared) { free_alien_cache(new_alien); goto fail; } |
0718dc2a8 [PATCH] slab: fix... |
3779 |
} |
cafeb02e0 [PATCH] alloc_kme... |
3780 |
|
a737b3e2f [PATCH] slab cleanup |
3781 3782 |
l3 = cachep->nodelists[node]; if (l3) { |
cafeb02e0 [PATCH] alloc_kme... |
3783 |
struct array_cache *shared = l3->shared; |
e498be7da [PATCH] Numa-awar... |
3784 |
spin_lock_irq(&l3->list_lock); |
cafeb02e0 [PATCH] alloc_kme... |
3785 |
if (shared) |
0718dc2a8 [PATCH] slab: fix... |
3786 3787 |
free_block(cachep, shared->entry, shared->avail, node); |
e498be7da [PATCH] Numa-awar... |
3788 |
|
cafeb02e0 [PATCH] alloc_kme... |
3789 3790 |
l3->shared = new_shared; if (!l3->alien) { |
e498be7da [PATCH] Numa-awar... |
3791 3792 3793 |
l3->alien = new_alien; new_alien = NULL; } |
b28a02de8 [PATCH] slab: fix... |
3794 |
l3->free_limit = (1 + nr_cpus_node(node)) * |
a737b3e2f [PATCH] slab cleanup |
3795 |
cachep->batchcount + cachep->num; |
e498be7da [PATCH] Numa-awar... |
3796 |
spin_unlock_irq(&l3->list_lock); |
cafeb02e0 [PATCH] alloc_kme... |
3797 |
kfree(shared); |
e498be7da [PATCH] Numa-awar... |
3798 3799 3800 |
free_alien_cache(new_alien); continue; } |
83b519e8b slab: setup alloc... |
3801 |
l3 = kmalloc_node(sizeof(struct kmem_list3), gfp, node); |
0718dc2a8 [PATCH] slab: fix... |
3802 3803 3804 |
if (!l3) { free_alien_cache(new_alien); kfree(new_shared); |
e498be7da [PATCH] Numa-awar... |
3805 |
goto fail; |
0718dc2a8 [PATCH] slab: fix... |
3806 |
} |
e498be7da [PATCH] Numa-awar... |
3807 3808 3809 |
kmem_list3_init(l3); l3->next_reap = jiffies + REAPTIMEOUT_LIST3 + |
a737b3e2f [PATCH] slab cleanup |
3810 |
((unsigned long)cachep) % REAPTIMEOUT_LIST3; |
cafeb02e0 [PATCH] alloc_kme... |
3811 |
l3->shared = new_shared; |
e498be7da [PATCH] Numa-awar... |
3812 |
l3->alien = new_alien; |
b28a02de8 [PATCH] slab: fix... |
3813 |
l3->free_limit = (1 + nr_cpus_node(node)) * |
a737b3e2f [PATCH] slab cleanup |
3814 |
cachep->batchcount + cachep->num; |
e498be7da [PATCH] Numa-awar... |
3815 3816 |
cachep->nodelists[node] = l3; } |
cafeb02e0 [PATCH] alloc_kme... |
3817 |
return 0; |
0718dc2a8 [PATCH] slab: fix... |
3818 |
|
a737b3e2f [PATCH] slab cleanup |
3819 |
fail: |
0718dc2a8 [PATCH] slab: fix... |
3820 3821 3822 3823 3824 3825 3826 3827 3828 3829 3830 3831 3832 3833 3834 |
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 [PATCH] alloc_kme... |
3835 |
return -ENOMEM; |
e498be7da [PATCH] Numa-awar... |
3836 |
} |
1da177e4c Linux-2.6.12-rc2 |
3837 |
struct ccupdate_struct { |
343e0d7a9 [PATCH] slab: rep... |
3838 |
struct kmem_cache *cachep; |
acfe7d744 slab: remove one ... |
3839 |
struct array_cache *new[0]; |
1da177e4c Linux-2.6.12-rc2 |
3840 3841 3842 3843 |
}; static void do_ccupdate_local(void *info) { |
a737b3e2f [PATCH] slab cleanup |
3844 |
struct ccupdate_struct *new = info; |
1da177e4c Linux-2.6.12-rc2 |
3845 3846 3847 |
struct array_cache *old; check_irq_off(); |
9a2dba4b4 [PATCH] slab: ren... |
3848 |
old = cpu_cache_get(new->cachep); |
e498be7da [PATCH] Numa-awar... |
3849 |
|
1da177e4c Linux-2.6.12-rc2 |
3850 3851 3852 |
new->cachep->array[smp_processor_id()] = new->new[smp_processor_id()]; new->new[smp_processor_id()] = old; } |
b5d8ca7c5 [PATCH] slab: rem... |
3853 |
/* Always called with the cache_chain_mutex held */ |
a737b3e2f [PATCH] slab cleanup |
3854 |
static int do_tune_cpucache(struct kmem_cache *cachep, int limit, |
83b519e8b slab: setup alloc... |
3855 |
int batchcount, int shared, gfp_t gfp) |
1da177e4c Linux-2.6.12-rc2 |
3856 |
{ |
d2e7b7d0a [PATCH] fix poten... |
3857 |
struct ccupdate_struct *new; |
2ed3a4ef9 [PATCH] slab: do ... |
3858 |
int i; |
1da177e4c Linux-2.6.12-rc2 |
3859 |
|
acfe7d744 slab: remove one ... |
3860 3861 |
new = kzalloc(sizeof(*new) + nr_cpu_ids * sizeof(struct array_cache *), gfp); |
d2e7b7d0a [PATCH] fix poten... |
3862 3863 |
if (!new) return -ENOMEM; |
e498be7da [PATCH] Numa-awar... |
3864 |
for_each_online_cpu(i) { |
7d6e6d09d numa: slab: use n... |
3865 |
new->new[i] = alloc_arraycache(cpu_to_mem(i), limit, |
83b519e8b slab: setup alloc... |
3866 |
batchcount, gfp); |
d2e7b7d0a [PATCH] fix poten... |
3867 |
if (!new->new[i]) { |
b28a02de8 [PATCH] slab: fix... |
3868 |
for (i--; i >= 0; i--) |
d2e7b7d0a [PATCH] fix poten... |
3869 3870 |
kfree(new->new[i]); kfree(new); |
e498be7da [PATCH] Numa-awar... |
3871 |
return -ENOMEM; |
1da177e4c Linux-2.6.12-rc2 |
3872 3873 |
} } |
d2e7b7d0a [PATCH] fix poten... |
3874 |
new->cachep = cachep; |
1da177e4c Linux-2.6.12-rc2 |
3875 |
|
15c8b6c1a on_each_cpu(): ki... |
3876 |
on_each_cpu(do_ccupdate_local, (void *)new, 1); |
e498be7da [PATCH] Numa-awar... |
3877 |
|
1da177e4c Linux-2.6.12-rc2 |
3878 |
check_irq_on(); |
1da177e4c Linux-2.6.12-rc2 |
3879 3880 |
cachep->batchcount = batchcount; cachep->limit = limit; |
e498be7da [PATCH] Numa-awar... |
3881 |
cachep->shared = shared; |
1da177e4c Linux-2.6.12-rc2 |
3882 |
|
e498be7da [PATCH] Numa-awar... |
3883 |
for_each_online_cpu(i) { |
d2e7b7d0a [PATCH] fix poten... |
3884 |
struct array_cache *ccold = new->new[i]; |
1da177e4c Linux-2.6.12-rc2 |
3885 3886 |
if (!ccold) continue; |
7d6e6d09d numa: slab: use n... |
3887 3888 3889 |
spin_lock_irq(&cachep->nodelists[cpu_to_mem(i)]->list_lock); free_block(cachep, ccold->entry, ccold->avail, cpu_to_mem(i)); spin_unlock_irq(&cachep->nodelists[cpu_to_mem(i)]->list_lock); |
1da177e4c Linux-2.6.12-rc2 |
3890 3891 |
kfree(ccold); } |
d2e7b7d0a [PATCH] fix poten... |
3892 |
kfree(new); |
83b519e8b slab: setup alloc... |
3893 |
return alloc_kmemlist(cachep, gfp); |
1da177e4c Linux-2.6.12-rc2 |
3894 |
} |
b5d8ca7c5 [PATCH] slab: rem... |
3895 |
/* Called with cache_chain_mutex held always */ |
83b519e8b slab: setup alloc... |
3896 |
static int enable_cpucache(struct kmem_cache *cachep, gfp_t gfp) |
1da177e4c Linux-2.6.12-rc2 |
3897 3898 3899 |
{ int err; int limit, shared; |
a737b3e2f [PATCH] slab cleanup |
3900 3901 |
/* * The head array serves three purposes: |
1da177e4c Linux-2.6.12-rc2 |
3902 3903 |
* - create a LIFO ordering, i.e. return objects that are cache-warm * - reduce the number of spinlock operations. |
a737b3e2f [PATCH] slab cleanup |
3904 |
* - reduce the number of linked list operations on the slab and |
1da177e4c Linux-2.6.12-rc2 |
3905 3906 3907 3908 |
* bufctl chains: array operations are cheaper. * The numbers are guessed, we should auto-tune as described by * Bonwick. */ |
3dafccf22 [PATCH] slab: dis... |
3909 |
if (cachep->buffer_size > 131072) |
1da177e4c Linux-2.6.12-rc2 |
3910 |
limit = 1; |
3dafccf22 [PATCH] slab: dis... |
3911 |
else if (cachep->buffer_size > PAGE_SIZE) |
1da177e4c Linux-2.6.12-rc2 |
3912 |
limit = 8; |
3dafccf22 [PATCH] slab: dis... |
3913 |
else if (cachep->buffer_size > 1024) |
1da177e4c Linux-2.6.12-rc2 |
3914 |
limit = 24; |
3dafccf22 [PATCH] slab: dis... |
3915 |
else if (cachep->buffer_size > 256) |
1da177e4c Linux-2.6.12-rc2 |
3916 3917 3918 |
limit = 54; else limit = 120; |
a737b3e2f [PATCH] slab cleanup |
3919 3920 |
/* * CPU bound tasks (e.g. network routing) can exhibit cpu bound |
1da177e4c Linux-2.6.12-rc2 |
3921 3922 3923 3924 3925 3926 3927 3928 |
* 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 SLAB: use num_pos... |
3929 |
if (cachep->buffer_size <= PAGE_SIZE && num_possible_cpus() > 1) |
1da177e4c Linux-2.6.12-rc2 |
3930 |
shared = 8; |
1da177e4c Linux-2.6.12-rc2 |
3931 3932 |
#if DEBUG |
a737b3e2f [PATCH] slab cleanup |
3933 3934 3935 |
/* * With debugging enabled, large batchcount lead to excessively long * periods with disabled local interrupts. Limit the batchcount |
1da177e4c Linux-2.6.12-rc2 |
3936 3937 3938 3939 |
*/ if (limit > 32) limit = 32; #endif |
83b519e8b slab: setup alloc... |
3940 |
err = do_tune_cpucache(cachep, limit, (limit + 1) / 2, shared, gfp); |
1da177e4c Linux-2.6.12-rc2 |
3941 3942 3943 |
if (err) printk(KERN_ERR "enable_cpucache failed for %s, error %d. ", |
b28a02de8 [PATCH] slab: fix... |
3944 |
cachep->name, -err); |
2ed3a4ef9 [PATCH] slab: do ... |
3945 |
return err; |
1da177e4c Linux-2.6.12-rc2 |
3946 |
} |
1b55253a7 [PATCH] slab: rem... |
3947 3948 |
/* * Drain an array if it contains any elements taking the l3 lock only if |
b18e7e654 [PATCH] slab: fix... |
3949 3950 |
* necessary. Note that the l3 listlock also protects the array_cache * if drain_array() is used on the shared array. |
1b55253a7 [PATCH] slab: rem... |
3951 |
*/ |
68a1b1955 mm/slab.c: make l... |
3952 |
static void drain_array(struct kmem_cache *cachep, struct kmem_list3 *l3, |
1b55253a7 [PATCH] slab: rem... |
3953 |
struct array_cache *ac, int force, int node) |
1da177e4c Linux-2.6.12-rc2 |
3954 3955 |
{ int tofree; |
1b55253a7 [PATCH] slab: rem... |
3956 3957 |
if (!ac || !ac->avail) return; |
1da177e4c Linux-2.6.12-rc2 |
3958 3959 |
if (ac->touched && !force) { ac->touched = 0; |
b18e7e654 [PATCH] slab: fix... |
3960 |
} else { |
1b55253a7 [PATCH] slab: rem... |
3961 |
spin_lock_irq(&l3->list_lock); |
b18e7e654 [PATCH] slab: fix... |
3962 3963 3964 3965 3966 3967 3968 3969 3970 |
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 [PATCH] slab: rem... |
3971 |
spin_unlock_irq(&l3->list_lock); |
1da177e4c Linux-2.6.12-rc2 |
3972 3973 3974 3975 3976 |
} } /** * cache_reap - Reclaim memory from caches. |
05fb6bf0b [PATCH] kernel-do... |
3977 |
* @w: work descriptor |
1da177e4c Linux-2.6.12-rc2 |
3978 3979 3980 3981 3982 3983 |
* * 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 [PATCH] slab cleanup |
3984 3985 |
* If we cannot acquire the cache chain mutex then just give up - we'll try * again on the next iteration. |
1da177e4c Linux-2.6.12-rc2 |
3986 |
*/ |
7c5cae368 [PATCH] slab: use... |
3987 |
static void cache_reap(struct work_struct *w) |
1da177e4c Linux-2.6.12-rc2 |
3988 |
{ |
7a7c381d2 [PATCH] slab: sto... |
3989 |
struct kmem_cache *searchp; |
e498be7da [PATCH] Numa-awar... |
3990 |
struct kmem_list3 *l3; |
7d6e6d09d numa: slab: use n... |
3991 |
int node = numa_mem_id(); |
bf6aede71 workqueue: add to... |
3992 |
struct delayed_work *work = to_delayed_work(w); |
1da177e4c Linux-2.6.12-rc2 |
3993 |
|
7c5cae368 [PATCH] slab: use... |
3994 |
if (!mutex_trylock(&cache_chain_mutex)) |
1da177e4c Linux-2.6.12-rc2 |
3995 |
/* Give up. Setup the next iteration. */ |
7c5cae368 [PATCH] slab: use... |
3996 |
goto out; |
1da177e4c Linux-2.6.12-rc2 |
3997 |
|
7a7c381d2 [PATCH] slab: sto... |
3998 |
list_for_each_entry(searchp, &cache_chain, next) { |
1da177e4c Linux-2.6.12-rc2 |
3999 |
check_irq_on(); |
35386e3b0 [PATCH] slab: cac... |
4000 4001 4002 4003 4004 |
/* * 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 [PATCH] slab: mak... |
4005 |
l3 = searchp->nodelists[node]; |
35386e3b0 [PATCH] slab: cac... |
4006 |
|
8fce4d8e3 [PATCH] slab: Nod... |
4007 |
reap_alien(searchp, l3); |
1da177e4c Linux-2.6.12-rc2 |
4008 |
|
aab2207cf [PATCH] slab: mak... |
4009 |
drain_array(searchp, l3, cpu_cache_get(searchp), 0, node); |
1da177e4c Linux-2.6.12-rc2 |
4010 |
|
35386e3b0 [PATCH] slab: cac... |
4011 4012 4013 4014 |
/* * These are racy checks but it does not matter * if we skip one check or scan twice. */ |
e498be7da [PATCH] Numa-awar... |
4015 |
if (time_after(l3->next_reap, jiffies)) |
35386e3b0 [PATCH] slab: cac... |
4016 |
goto next; |
1da177e4c Linux-2.6.12-rc2 |
4017 |
|
e498be7da [PATCH] Numa-awar... |
4018 |
l3->next_reap = jiffies + REAPTIMEOUT_LIST3; |
1da177e4c Linux-2.6.12-rc2 |
4019 |
|
aab2207cf [PATCH] slab: mak... |
4020 |
drain_array(searchp, l3, l3->shared, 0, node); |
1da177e4c Linux-2.6.12-rc2 |
4021 |
|
ed11d9eb2 [PATCH] slab: con... |
4022 |
if (l3->free_touched) |
e498be7da [PATCH] Numa-awar... |
4023 |
l3->free_touched = 0; |
ed11d9eb2 [PATCH] slab: con... |
4024 4025 |
else { int freed; |
1da177e4c Linux-2.6.12-rc2 |
4026 |
|
ed11d9eb2 [PATCH] slab: con... |
4027 4028 4029 4030 |
freed = drain_freelist(searchp, l3, (l3->free_limit + 5 * searchp->num - 1) / (5 * searchp->num)); STATS_ADD_REAPED(searchp, freed); } |
35386e3b0 [PATCH] slab: cac... |
4031 |
next: |
1da177e4c Linux-2.6.12-rc2 |
4032 4033 4034 |
cond_resched(); } check_irq_on(); |
fc0abb145 [PATCH] sem2mutex... |
4035 |
mutex_unlock(&cache_chain_mutex); |
8fce4d8e3 [PATCH] slab: Nod... |
4036 |
next_reap_node(); |
7c5cae368 [PATCH] slab: use... |
4037 |
out: |
a737b3e2f [PATCH] slab cleanup |
4038 |
/* Set up the next iteration */ |
7c5cae368 [PATCH] slab: use... |
4039 |
schedule_delayed_work(work, round_jiffies_relative(REAPTIMEOUT_CPUC)); |
1da177e4c Linux-2.6.12-rc2 |
4040 |
} |
158a96242 Unify /proc/slabi... |
4041 |
#ifdef CONFIG_SLABINFO |
1da177e4c Linux-2.6.12-rc2 |
4042 |
|
85289f98d [PATCH] slab: ext... |
4043 |
static void print_slabinfo_header(struct seq_file *m) |
1da177e4c Linux-2.6.12-rc2 |
4044 |
{ |
85289f98d [PATCH] slab: ext... |
4045 4046 4047 4048 |
/* * Output format version, so at least we can change it * without _too_ many complaints. */ |
1da177e4c Linux-2.6.12-rc2 |
4049 |
#if STATS |
85289f98d [PATCH] slab: ext... |
4050 4051 |
seq_puts(m, "slabinfo - version: 2.1 (statistics) "); |
1da177e4c Linux-2.6.12-rc2 |
4052 |
#else |
85289f98d [PATCH] slab: ext... |
4053 4054 |
seq_puts(m, "slabinfo - version: 2.1 "); |
1da177e4c Linux-2.6.12-rc2 |
4055 |
#endif |
85289f98d [PATCH] slab: ext... |
4056 4057 4058 4059 |
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 Linux-2.6.12-rc2 |
4060 |
#if STATS |
85289f98d [PATCH] slab: ext... |
4061 |
seq_puts(m, " : globalstat <listallocs> <maxobjs> <grown> <reaped> " |
fb7faf331 [PATCH] slab: add... |
4062 |
"<error> <maxfreeable> <nodeallocs> <remotefrees> <alienoverflow>"); |
85289f98d [PATCH] slab: ext... |
4063 |
seq_puts(m, " : cpustat <allochit> <allocmiss> <freehit> <freemiss>"); |
1da177e4c Linux-2.6.12-rc2 |
4064 |
#endif |
85289f98d [PATCH] slab: ext... |
4065 4066 4067 4068 4069 4070 4071 |
seq_putc(m, ' '); } static void *s_start(struct seq_file *m, loff_t *pos) { loff_t n = *pos; |
85289f98d [PATCH] slab: ext... |
4072 |
|
fc0abb145 [PATCH] sem2mutex... |
4073 |
mutex_lock(&cache_chain_mutex); |
85289f98d [PATCH] slab: ext... |
4074 4075 |
if (!n) print_slabinfo_header(m); |
b92151bab Make /proc/slabin... |
4076 4077 |
return seq_list_start(&cache_chain, *pos); |
1da177e4c Linux-2.6.12-rc2 |
4078 4079 4080 4081 |
} static void *s_next(struct seq_file *m, void *p, loff_t *pos) { |
b92151bab Make /proc/slabin... |
4082 |
return seq_list_next(p, &cache_chain, pos); |
1da177e4c Linux-2.6.12-rc2 |
4083 4084 4085 4086 |
} static void s_stop(struct seq_file *m, void *p) { |
fc0abb145 [PATCH] sem2mutex... |
4087 |
mutex_unlock(&cache_chain_mutex); |
1da177e4c Linux-2.6.12-rc2 |
4088 4089 4090 4091 |
} static int s_show(struct seq_file *m, void *p) { |
b92151bab Make /proc/slabin... |
4092 |
struct kmem_cache *cachep = list_entry(p, struct kmem_cache, next); |
b28a02de8 [PATCH] slab: fix... |
4093 4094 4095 4096 4097 |
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 [PATCH] Numa-awar... |
4098 |
const char *name; |
1da177e4c Linux-2.6.12-rc2 |
4099 |
char *error = NULL; |
e498be7da [PATCH] Numa-awar... |
4100 4101 |
int node; struct kmem_list3 *l3; |
1da177e4c Linux-2.6.12-rc2 |
4102 |
|
1da177e4c Linux-2.6.12-rc2 |
4103 4104 |
active_objs = 0; num_slabs = 0; |
e498be7da [PATCH] Numa-awar... |
4105 4106 4107 4108 |
for_each_online_node(node) { l3 = cachep->nodelists[node]; if (!l3) continue; |
ca3b9b917 [PATCH] NUMA slab... |
4109 4110 |
check_irq_on(); spin_lock_irq(&l3->list_lock); |
e498be7da [PATCH] Numa-awar... |
4111 |
|
7a7c381d2 [PATCH] slab: sto... |
4112 |
list_for_each_entry(slabp, &l3->slabs_full, list) { |
e498be7da [PATCH] Numa-awar... |
4113 4114 4115 4116 4117 |
if (slabp->inuse != cachep->num && !error) error = "slabs_full accounting error"; active_objs += cachep->num; active_slabs++; } |
7a7c381d2 [PATCH] slab: sto... |
4118 |
list_for_each_entry(slabp, &l3->slabs_partial, list) { |
e498be7da [PATCH] Numa-awar... |
4119 4120 4121 4122 4123 4124 4125 |
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 [PATCH] slab: sto... |
4126 |
list_for_each_entry(slabp, &l3->slabs_free, list) { |
e498be7da [PATCH] Numa-awar... |
4127 4128 4129 4130 4131 |
if (slabp->inuse && !error) error = "slabs_free/inuse accounting error"; num_slabs++; } free_objects += l3->free_objects; |
4484ebf12 [PATCH] NUMA slab... |
4132 4133 |
if (l3->shared) shared_avail += l3->shared->avail; |
e498be7da [PATCH] Numa-awar... |
4134 |
|
ca3b9b917 [PATCH] NUMA slab... |
4135 |
spin_unlock_irq(&l3->list_lock); |
1da177e4c Linux-2.6.12-rc2 |
4136 |
} |
b28a02de8 [PATCH] slab: fix... |
4137 4138 |
num_slabs += active_slabs; num_objs = num_slabs * cachep->num; |
e498be7da [PATCH] Numa-awar... |
4139 |
if (num_objs - active_objs != free_objects && !error) |
1da177e4c Linux-2.6.12-rc2 |
4140 |
error = "free_objects accounting error"; |
b28a02de8 [PATCH] slab: fix... |
4141 |
name = cachep->name; |
1da177e4c Linux-2.6.12-rc2 |
4142 4143 4144 4145 4146 |
if (error) printk(KERN_ERR "slab: cache %s error: %s ", name, error); seq_printf(m, "%-17s %6lu %6lu %6u %4u %4d", |
3dafccf22 [PATCH] slab: dis... |
4147 |
name, active_objs, num_objs, cachep->buffer_size, |
b28a02de8 [PATCH] slab: fix... |
4148 |
cachep->num, (1 << cachep->gfporder)); |
1da177e4c Linux-2.6.12-rc2 |
4149 |
seq_printf(m, " : tunables %4u %4u %4u", |
b28a02de8 [PATCH] slab: fix... |
4150 |
cachep->limit, cachep->batchcount, cachep->shared); |
e498be7da [PATCH] Numa-awar... |
4151 |
seq_printf(m, " : slabdata %6lu %6lu %6lu", |
b28a02de8 [PATCH] slab: fix... |
4152 |
active_slabs, num_slabs, shared_avail); |
1da177e4c Linux-2.6.12-rc2 |
4153 |
#if STATS |
b28a02de8 [PATCH] slab: fix... |
4154 |
{ /* list3 stats */ |
1da177e4c Linux-2.6.12-rc2 |
4155 4156 4157 4158 4159 4160 |
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 Linux-2.6.12-rc2 |
4161 |
unsigned long node_allocs = cachep->node_allocs; |
e498be7da [PATCH] Numa-awar... |
4162 |
unsigned long node_frees = cachep->node_frees; |
fb7faf331 [PATCH] slab: add... |
4163 |
unsigned long overflows = cachep->node_overflow; |
1da177e4c Linux-2.6.12-rc2 |
4164 |
|
e92dd4fd1 slab: Fix continu... |
4165 4166 4167 4168 4169 |
seq_printf(m, " : globalstat %7lu %6lu %5lu %4lu " "%4lu %4lu %4lu %4lu %4lu", allocs, high, grown, reaped, errors, max_freeable, node_allocs, node_frees, overflows); |
1da177e4c Linux-2.6.12-rc2 |
4170 4171 4172 4173 4174 4175 4176 4177 4178 |
} /* 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 [PATCH] slab: fix... |
4179 |
allochit, allocmiss, freehit, freemiss); |
1da177e4c Linux-2.6.12-rc2 |
4180 4181 4182 4183 |
} #endif seq_putc(m, ' '); |
1da177e4c Linux-2.6.12-rc2 |
4184 4185 4186 4187 4188 4189 4190 4191 4192 4193 4194 4195 4196 4197 4198 4199 |
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 proc: move /proc/... |
4200 |
static const struct seq_operations slabinfo_op = { |
b28a02de8 [PATCH] slab: fix... |
4201 4202 4203 4204 |
.start = s_start, .next = s_next, .stop = s_stop, .show = s_show, |
1da177e4c Linux-2.6.12-rc2 |
4205 4206 4207 4208 4209 4210 4211 4212 4213 4214 |
}; #define MAX_SLABINFO_WRITE 128 /** * slabinfo_write - Tuning for the slab allocator * @file: unused * @buffer: user buffer * @count: data length * @ppos: unused */ |
68a1b1955 mm/slab.c: make l... |
4215 |
static ssize_t slabinfo_write(struct file *file, const char __user *buffer, |
b28a02de8 [PATCH] slab: fix... |
4216 |
size_t count, loff_t *ppos) |
1da177e4c Linux-2.6.12-rc2 |
4217 |
{ |
b28a02de8 [PATCH] slab: fix... |
4218 |
char kbuf[MAX_SLABINFO_WRITE + 1], *tmp; |
1da177e4c Linux-2.6.12-rc2 |
4219 |
int limit, batchcount, shared, res; |
7a7c381d2 [PATCH] slab: sto... |
4220 |
struct kmem_cache *cachep; |
b28a02de8 [PATCH] slab: fix... |
4221 |
|
1da177e4c Linux-2.6.12-rc2 |
4222 4223 4224 4225 |
if (count > MAX_SLABINFO_WRITE) return -EINVAL; if (copy_from_user(&kbuf, buffer, count)) return -EFAULT; |
b28a02de8 [PATCH] slab: fix... |
4226 |
kbuf[MAX_SLABINFO_WRITE] = '\0'; |
1da177e4c Linux-2.6.12-rc2 |
4227 4228 4229 4230 4231 4232 4233 4234 4235 4236 |
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 [PATCH] sem2mutex... |
4237 |
mutex_lock(&cache_chain_mutex); |
1da177e4c Linux-2.6.12-rc2 |
4238 |
res = -EINVAL; |
7a7c381d2 [PATCH] slab: sto... |
4239 |
list_for_each_entry(cachep, &cache_chain, next) { |
1da177e4c Linux-2.6.12-rc2 |
4240 |
if (!strcmp(cachep->name, kbuf)) { |
a737b3e2f [PATCH] slab cleanup |
4241 4242 |
if (limit < 1 || batchcount < 1 || batchcount > limit || shared < 0) { |
e498be7da [PATCH] Numa-awar... |
4243 |
res = 0; |
1da177e4c Linux-2.6.12-rc2 |
4244 |
} else { |
e498be7da [PATCH] Numa-awar... |
4245 |
res = do_tune_cpucache(cachep, limit, |
83b519e8b slab: setup alloc... |
4246 4247 |
batchcount, shared, GFP_KERNEL); |
1da177e4c Linux-2.6.12-rc2 |
4248 4249 4250 4251 |
} break; } } |
fc0abb145 [PATCH] sem2mutex... |
4252 |
mutex_unlock(&cache_chain_mutex); |
1da177e4c Linux-2.6.12-rc2 |
4253 4254 4255 4256 |
if (res >= 0) res = count; return res; } |
871751e25 [PATCH] slab: imp... |
4257 |
|
7b3c3a50a proc: move /proc/... |
4258 4259 4260 4261 4262 4263 4264 4265 4266 4267 4268 4269 |
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 [PATCH] slab: imp... |
4270 4271 4272 4273 |
#ifdef CONFIG_DEBUG_SLAB_LEAK static void *leaks_start(struct seq_file *m, loff_t *pos) { |
871751e25 [PATCH] slab: imp... |
4274 |
mutex_lock(&cache_chain_mutex); |
b92151bab Make /proc/slabin... |
4275 |
return seq_list_start(&cache_chain, *pos); |
871751e25 [PATCH] slab: imp... |
4276 4277 4278 4279 4280 4281 4282 4283 4284 4285 4286 4287 4288 4289 4290 4291 4292 4293 4294 4295 4296 4297 4298 4299 4300 4301 4302 4303 4304 4305 4306 4307 4308 4309 4310 4311 4312 4313 4314 4315 4316 4317 4318 4319 4320 4321 4322 4323 4324 |
} 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 [PATCH] slab: imp... |
4325 |
unsigned long offset, size; |
9281acea6 kallsyms: make KS... |
4326 |
char modname[MODULE_NAME_LEN], name[KSYM_NAME_LEN]; |
871751e25 [PATCH] slab: imp... |
4327 |
|
a5c43dae7 Fix race between ... |
4328 |
if (lookup_symbol_attrs(address, &size, &offset, modname, name) == 0) { |
871751e25 [PATCH] slab: imp... |
4329 |
seq_printf(m, "%s+%#lx/%#lx", name, offset, size); |
a5c43dae7 Fix race between ... |
4330 |
if (modname[0]) |
871751e25 [PATCH] slab: imp... |
4331 4332 4333 4334 4335 4336 4337 4338 4339 |
seq_printf(m, " [%s]", modname); return; } #endif seq_printf(m, "%p", (void *)address); } static int leaks_show(struct seq_file *m, void *p) { |
b92151bab Make /proc/slabin... |
4340 |
struct kmem_cache *cachep = list_entry(p, struct kmem_cache, next); |
871751e25 [PATCH] slab: imp... |
4341 4342 4343 4344 4345 4346 4347 4348 4349 4350 4351 4352 4353 4354 4355 4356 4357 4358 4359 4360 4361 4362 4363 |
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 [PATCH] slab: sto... |
4364 |
list_for_each_entry(slabp, &l3->slabs_full, list) |
871751e25 [PATCH] slab: imp... |
4365 |
handle_slab(n, cachep, slabp); |
7a7c381d2 [PATCH] slab: sto... |
4366 |
list_for_each_entry(slabp, &l3->slabs_partial, list) |
871751e25 [PATCH] slab: imp... |
4367 |
handle_slab(n, cachep, slabp); |
871751e25 [PATCH] slab: imp... |
4368 4369 4370 4371 4372 4373 4374 4375 4376 4377 4378 4379 4380 4381 4382 4383 4384 4385 4386 4387 4388 4389 4390 4391 4392 4393 |
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 [PATCH] fix poten... |
4394 |
|
871751e25 [PATCH] slab: imp... |
4395 4396 |
return 0; } |
a0ec95a8e proc: move /proc/... |
4397 |
static const struct seq_operations slabstats_op = { |
871751e25 [PATCH] slab: imp... |
4398 4399 4400 4401 4402 |
.start = leaks_start, .next = s_next, .stop = s_stop, .show = leaks_show, }; |
a0ec95a8e proc: move /proc/... |
4403 4404 4405 4406 4407 4408 4409 4410 4411 4412 4413 4414 4415 4416 4417 4418 4419 4420 4421 4422 4423 4424 4425 4426 4427 4428 4429 4430 |
static int slabstats_open(struct inode *inode, struct file *file) { unsigned long *n = kzalloc(PAGE_SIZE, GFP_KERNEL); int ret = -ENOMEM; if (n) { ret = seq_open(file, &slabstats_op); if (!ret) { struct seq_file *m = file->private_data; *n = PAGE_SIZE / (2 * sizeof(unsigned long)); m->private = n; n = NULL; } kfree(n); } return ret; } static const struct file_operations proc_slabstats_operations = { .open = slabstats_open, .read = seq_read, .llseek = seq_lseek, .release = seq_release_private, }; #endif static int __init slab_proc_init(void) { |
ab067e99d mm: restrict acce... |
4431 |
proc_create("slabinfo",S_IWUSR|S_IRUSR,NULL,&proc_slabinfo_operations); |
a0ec95a8e proc: move /proc/... |
4432 4433 |
#ifdef CONFIG_DEBUG_SLAB_LEAK proc_create("slab_allocators", 0, NULL, &proc_slabstats_operations); |
871751e25 [PATCH] slab: imp... |
4434 |
#endif |
a0ec95a8e proc: move /proc/... |
4435 4436 4437 |
return 0; } module_init(slab_proc_init); |
1da177e4c Linux-2.6.12-rc2 |
4438 |
#endif |
00e145b6d [PATCH] slab: rem... |
4439 4440 4441 4442 4443 4444 4445 4446 4447 4448 4449 4450 |
/** * 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 slab: introduce k... |
4451 |
size_t ksize(const void *objp) |
1da177e4c Linux-2.6.12-rc2 |
4452 |
{ |
ef8b4520b Slab allocators: ... |
4453 4454 |
BUG_ON(!objp); if (unlikely(objp == ZERO_SIZE_PTR)) |
00e145b6d [PATCH] slab: rem... |
4455 |
return 0; |
1da177e4c Linux-2.6.12-rc2 |
4456 |
|
6ed5eb221 [PATCH] slab: ext... |
4457 |
return obj_size(virt_to_cache(objp)); |
1da177e4c Linux-2.6.12-rc2 |
4458 |
} |
b1aabecd5 mm: Export symbol... |
4459 |
EXPORT_SYMBOL(ksize); |