Blame view
mm/slab.c
110 KB
b24413180
|
1 |
// SPDX-License-Identifier: GPL-2.0 |
1da177e4c
|
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 29 |
/* * 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
|
30 |
* slabs and you must pass objects with the same initializations to |
1da177e4c
|
31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 |
* 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
|
54 |
* The c_cpuarray may not be read with enabled local interrupts - |
1da177e4c
|
55 56 57 58 |
* it's changed with a smp_call_function(). * * SMP synchronization: * constructors and destructors are called without any locking. |
343e0d7a9
|
59 |
* Several members in struct kmem_cache and struct slab never change, they |
1da177e4c
|
60 61 62 63 64 65 66 67 68 69 70 71 |
* 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 |
18004c5d4
|
72 |
* The global cache-chain is protected by the mutex 'slab_mutex'. |
1da177e4c
|
73 74 75 76 77 78 |
* 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
|
79 80 81 82 83 84 85 86 87 |
* 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
|
88 |
*/ |
1da177e4c
|
89 90 |
#include <linux/slab.h> #include <linux/mm.h> |
c9cf55285
|
91 |
#include <linux/poison.h> |
1da177e4c
|
92 93 94 95 96 |
#include <linux/swap.h> #include <linux/cache.h> #include <linux/interrupt.h> #include <linux/init.h> #include <linux/compiler.h> |
101a50019
|
97 |
#include <linux/cpuset.h> |
a0ec95a8e
|
98 |
#include <linux/proc_fs.h> |
1da177e4c
|
99 100 101 102 103 104 105 |
#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
|
106 |
#include <linux/string.h> |
138ae6631
|
107 |
#include <linux/uaccess.h> |
e498be7da
|
108 |
#include <linux/nodemask.h> |
d5cff6352
|
109 |
#include <linux/kmemleak.h> |
dc85da15d
|
110 |
#include <linux/mempolicy.h> |
fc0abb145
|
111 |
#include <linux/mutex.h> |
8a8b6502f
|
112 |
#include <linux/fault-inject.h> |
e7eebaf6a
|
113 |
#include <linux/rtmutex.h> |
6a2d7a955
|
114 |
#include <linux/reciprocal_div.h> |
3ac7fe5a4
|
115 |
#include <linux/debugobjects.h> |
8f9f8d9e8
|
116 |
#include <linux/memory.h> |
268bb0ce3
|
117 |
#include <linux/prefetch.h> |
3f8c24529
|
118 |
#include <linux/sched/task_stack.h> |
1da177e4c
|
119 |
|
381760ead
|
120 |
#include <net/sock.h> |
1da177e4c
|
121 122 123 |
#include <asm/cacheflush.h> #include <asm/tlbflush.h> #include <asm/page.h> |
4dee6b64e
|
124 |
#include <trace/events/kmem.h> |
072bb0aa5
|
125 |
#include "internal.h" |
b9ce5ef49
|
126 |
#include "slab.h" |
1da177e4c
|
127 |
/* |
50953fe9e
|
128 |
* DEBUG - 1 for kmem_cache_create() to honour; SLAB_RED_ZONE & SLAB_POISON. |
1da177e4c
|
129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 |
* 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
|
146 147 |
/* Shouldn't this be in a header file somewhere? */ #define BYTES_PER_WORD sizeof(void *) |
87a927c71
|
148 |
#define REDZONE_ALIGN max(BYTES_PER_WORD, __alignof__(unsigned long long)) |
1da177e4c
|
149 |
|
1da177e4c
|
150 151 152 |
#ifndef ARCH_KMALLOC_FLAGS #define ARCH_KMALLOC_FLAGS SLAB_HWCACHE_ALIGN #endif |
f315e3fa1
|
153 154 155 156 157 158 159 160 |
#define FREELIST_BYTE_INDEX (((PAGE_SIZE >> BITS_PER_BYTE) \ <= SLAB_OBJ_MIN_SIZE) ? 1 : 0) #if FREELIST_BYTE_INDEX typedef unsigned char freelist_idx_t; #else typedef unsigned short freelist_idx_t; #endif |
30321c7b6
|
161 |
#define SLAB_OBJ_MAX_NUM ((1 << sizeof(freelist_idx_t) * BITS_PER_BYTE) - 1) |
f315e3fa1
|
162 |
|
072bb0aa5
|
163 |
/* |
1da177e4c
|
164 165 |
* struct array_cache * |
1da177e4c
|
166 167 168 169 170 171 172 173 174 175 176 177 178 179 |
* 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; |
bda5b655f
|
180 |
void *entry[]; /* |
a737b3e2f
|
181 182 183 |
* Must have this definition in here for the proper * alignment of array_cache. Also simplifies accessing * the entries. |
a737b3e2f
|
184 |
*/ |
1da177e4c
|
185 |
}; |
c8522a3a5
|
186 187 188 189 |
struct alien_cache { spinlock_t lock; struct array_cache ac; }; |
a737b3e2f
|
190 |
/* |
e498be7da
|
191 192 |
* Need this for bootstrapping a per node allocator. */ |
bf0dea23a
|
193 |
#define NUM_INIT_LISTS (2 * MAX_NUMNODES) |
ce8eb6c42
|
194 |
static struct kmem_cache_node __initdata init_kmem_cache_node[NUM_INIT_LISTS]; |
e498be7da
|
195 |
#define CACHE_CACHE 0 |
bf0dea23a
|
196 |
#define SIZE_NODE (MAX_NUMNODES) |
e498be7da
|
197 |
|
ed11d9eb2
|
198 |
static int drain_freelist(struct kmem_cache *cache, |
ce8eb6c42
|
199 |
struct kmem_cache_node *n, int tofree); |
ed11d9eb2
|
200 |
static void free_block(struct kmem_cache *cachep, void **objpp, int len, |
97654dfa2
|
201 202 |
int node, struct list_head *list); static void slabs_destroy(struct kmem_cache *cachep, struct list_head *list); |
83b519e8b
|
203 |
static int enable_cpucache(struct kmem_cache *cachep, gfp_t gfp); |
65f27f384
|
204 |
static void cache_reap(struct work_struct *unused); |
ed11d9eb2
|
205 |
|
76b342bdc
|
206 207 208 209 210 |
static inline void fixup_objfreelist_debug(struct kmem_cache *cachep, void **list); static inline void fixup_slab_list(struct kmem_cache *cachep, struct kmem_cache_node *n, struct page *page, void **list); |
e0a427267
|
211 |
static int slab_early_init = 1; |
ce8eb6c42
|
212 |
#define INDEX_NODE kmalloc_index(sizeof(struct kmem_cache_node)) |
1da177e4c
|
213 |
|
ce8eb6c42
|
214 |
static void kmem_cache_node_init(struct kmem_cache_node *parent) |
e498be7da
|
215 216 217 218 |
{ INIT_LIST_HEAD(&parent->slabs_full); INIT_LIST_HEAD(&parent->slabs_partial); INIT_LIST_HEAD(&parent->slabs_free); |
bf00bd345
|
219 |
parent->total_slabs = 0; |
f728b0a5d
|
220 |
parent->free_slabs = 0; |
e498be7da
|
221 222 |
parent->shared = NULL; parent->alien = NULL; |
2e1217cf9
|
223 |
parent->colour_next = 0; |
e498be7da
|
224 225 226 227 |
spin_lock_init(&parent->list_lock); parent->free_objects = 0; parent->free_touched = 0; } |
a737b3e2f
|
228 229 230 |
#define MAKE_LIST(cachep, listp, slab, nodeid) \ do { \ INIT_LIST_HEAD(listp); \ |
18bf85411
|
231 |
list_splice(&get_node(cachep, nodeid)->slab, listp); \ |
e498be7da
|
232 |
} while (0) |
a737b3e2f
|
233 234 |
#define MAKE_ALL_LISTS(cachep, ptr, nodeid) \ do { \ |
e498be7da
|
235 236 237 238 |
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
|
239 |
|
b03a017be
|
240 |
#define CFLGS_OBJFREELIST_SLAB (0x40000000UL) |
1da177e4c
|
241 |
#define CFLGS_OFF_SLAB (0x80000000UL) |
b03a017be
|
242 |
#define OBJFREELIST_SLAB(x) ((x)->flags & CFLGS_OBJFREELIST_SLAB) |
1da177e4c
|
243 244 245 |
#define OFF_SLAB(x) ((x)->flags & CFLGS_OFF_SLAB) #define BATCHREFILL_LIMIT 16 |
a737b3e2f
|
246 247 248 |
/* * Optimization question: fewer reaps means less probability for unnessary * cpucache drain/refill cycles. |
1da177e4c
|
249 |
* |
dc6f3f276
|
250 |
* OTOH the cpuarrays can contain lots of objects, |
1da177e4c
|
251 252 |
* which could lock up otherwise freeable slabs. */ |
5f0985bb1
|
253 254 |
#define REAPTIMEOUT_AC (2*HZ) #define REAPTIMEOUT_NODE (4*HZ) |
1da177e4c
|
255 256 257 258 259 260 |
#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
|
261 |
#define STATS_ADD_REAPED(x,y) ((x)->reaped += (y)) |
a737b3e2f
|
262 263 264 265 266 |
#define STATS_SET_HIGH(x) \ do { \ if ((x)->num_active > (x)->high_mark) \ (x)->high_mark = (x)->num_active; \ } while (0) |
1da177e4c
|
267 268 |
#define STATS_INC_ERR(x) ((x)->errors++) #define STATS_INC_NODEALLOCS(x) ((x)->node_allocs++) |
e498be7da
|
269 |
#define STATS_INC_NODEFREES(x) ((x)->node_frees++) |
fb7faf331
|
270 |
#define STATS_INC_ACOVERFLOW(x) ((x)->node_overflow++) |
a737b3e2f
|
271 272 273 274 275 |
#define STATS_SET_FREEABLE(x, i) \ do { \ if ((x)->max_freeable < i) \ (x)->max_freeable = i; \ } while (0) |
1da177e4c
|
276 277 278 279 280 281 282 283 284 |
#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
|
285 |
#define STATS_ADD_REAPED(x,y) do { (void)(y); } while (0) |
1da177e4c
|
286 287 288 |
#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
|
289 |
#define STATS_INC_NODEFREES(x) do { } while (0) |
fb7faf331
|
290 |
#define STATS_INC_ACOVERFLOW(x) do { } while (0) |
a737b3e2f
|
291 |
#define STATS_SET_FREEABLE(x, i) do { } while (0) |
1da177e4c
|
292 293 294 295 296 297 298 |
#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
|
299 |
|
a737b3e2f
|
300 301 |
/* * memory layout of objects: |
1da177e4c
|
302 |
* 0 : objp |
3dafccf22
|
303 |
* 0 .. cachep->obj_offset - BYTES_PER_WORD - 1: padding. This ensures that |
1da177e4c
|
304 305 |
* the end of an object is aligned with the end of the real * allocation. Catches writes behind the end of the allocation. |
3dafccf22
|
306 |
* cachep->obj_offset - BYTES_PER_WORD .. cachep->obj_offset - 1: |
1da177e4c
|
307 |
* redzone word. |
3dafccf22
|
308 |
* cachep->obj_offset: The real object. |
3b0efdfa1
|
309 310 |
* cachep->size - 2* BYTES_PER_WORD: redzone word [BYTES_PER_WORD long] * cachep->size - 1* BYTES_PER_WORD: last caller address |
a737b3e2f
|
311 |
* [BYTES_PER_WORD long] |
1da177e4c
|
312 |
*/ |
343e0d7a9
|
313 |
static int obj_offset(struct kmem_cache *cachep) |
1da177e4c
|
314 |
{ |
3dafccf22
|
315 |
return cachep->obj_offset; |
1da177e4c
|
316 |
} |
b46b8f19c
|
317 |
static unsigned long long *dbg_redzone1(struct kmem_cache *cachep, void *objp) |
1da177e4c
|
318 319 |
{ BUG_ON(!(cachep->flags & SLAB_RED_ZONE)); |
b46b8f19c
|
320 321 |
return (unsigned long long*) (objp + obj_offset(cachep) - sizeof(unsigned long long)); |
1da177e4c
|
322 |
} |
b46b8f19c
|
323 |
static unsigned long long *dbg_redzone2(struct kmem_cache *cachep, void *objp) |
1da177e4c
|
324 325 326 |
{ BUG_ON(!(cachep->flags & SLAB_RED_ZONE)); if (cachep->flags & SLAB_STORE_USER) |
3b0efdfa1
|
327 |
return (unsigned long long *)(objp + cachep->size - |
b46b8f19c
|
328 |
sizeof(unsigned long long) - |
87a927c71
|
329 |
REDZONE_ALIGN); |
3b0efdfa1
|
330 |
return (unsigned long long *) (objp + cachep->size - |
b46b8f19c
|
331 |
sizeof(unsigned long long)); |
1da177e4c
|
332 |
} |
343e0d7a9
|
333 |
static void **dbg_userword(struct kmem_cache *cachep, void *objp) |
1da177e4c
|
334 335 |
{ BUG_ON(!(cachep->flags & SLAB_STORE_USER)); |
3b0efdfa1
|
336 |
return (void **)(objp + cachep->size - BYTES_PER_WORD); |
1da177e4c
|
337 338 339 |
} #else |
3dafccf22
|
340 |
#define obj_offset(x) 0 |
b46b8f19c
|
341 342 |
#define dbg_redzone1(cachep, objp) ({BUG(); (unsigned long long *)NULL;}) #define dbg_redzone2(cachep, objp) ({BUG(); (unsigned long long *)NULL;}) |
1da177e4c
|
343 344 345 |
#define dbg_userword(cachep, objp) ({BUG(); (void **)NULL;}) #endif |
037873014
|
346 |
#ifdef CONFIG_DEBUG_SLAB_LEAK |
d31676dfd
|
347 |
static inline bool is_store_user_clean(struct kmem_cache *cachep) |
037873014
|
348 |
{ |
d31676dfd
|
349 350 |
return atomic_read(&cachep->store_user_clean) == 1; } |
037873014
|
351 |
|
d31676dfd
|
352 353 354 355 |
static inline void set_store_user_clean(struct kmem_cache *cachep) { atomic_set(&cachep->store_user_clean, 1); } |
037873014
|
356 |
|
d31676dfd
|
357 358 359 360 |
static inline void set_store_user_dirty(struct kmem_cache *cachep) { if (is_store_user_clean(cachep)) atomic_set(&cachep->store_user_clean, 0); |
037873014
|
361 362 363 |
} #else |
d31676dfd
|
364 |
static inline void set_store_user_dirty(struct kmem_cache *cachep) {} |
037873014
|
365 366 |
#endif |
1da177e4c
|
367 |
/* |
3df1cccdf
|
368 369 |
* Do not go above this order unless 0 objects fit into the slab or * overridden on the command line. |
1da177e4c
|
370 |
*/ |
543585cc5
|
371 372 373 |
#define SLAB_MAX_ORDER_HI 1 #define SLAB_MAX_ORDER_LO 0 static int slab_max_order = SLAB_MAX_ORDER_LO; |
3df1cccdf
|
374 |
static bool slab_max_order_set __initdata; |
1da177e4c
|
375 |
|
6ed5eb221
|
376 377 |
static inline struct kmem_cache *virt_to_cache(const void *obj) { |
b49af68ff
|
378 |
struct page *page = virt_to_head_page(obj); |
350260889
|
379 |
return page->slab_cache; |
6ed5eb221
|
380 |
} |
8456a648c
|
381 |
static inline void *index_to_obj(struct kmem_cache *cache, struct page *page, |
8fea4e96a
|
382 383 |
unsigned int idx) { |
8456a648c
|
384 |
return page->s_mem + cache->size * idx; |
8fea4e96a
|
385 |
} |
6a2d7a955
|
386 |
/* |
3b0efdfa1
|
387 388 389 |
* We want to avoid an expensive divide : (offset / cache->size) * Using the fact that size is a constant for a particular cache, * we can replace (offset / cache->size) by |
6a2d7a955
|
390 391 392 |
* reciprocal_divide(offset, cache->reciprocal_buffer_size) */ static inline unsigned int obj_to_index(const struct kmem_cache *cache, |
8456a648c
|
393 |
const struct page *page, void *obj) |
8fea4e96a
|
394 |
{ |
8456a648c
|
395 |
u32 offset = (obj - page->s_mem); |
6a2d7a955
|
396 |
return reciprocal_divide(offset, cache->reciprocal_buffer_size); |
8fea4e96a
|
397 |
} |
6fb924304
|
398 |
#define BOOT_CPUCACHE_ENTRIES 1 |
1da177e4c
|
399 |
/* internal cache of cache description objs */ |
9b030cb86
|
400 |
static struct kmem_cache kmem_cache_boot = { |
b28a02de8
|
401 402 403 |
.batchcount = 1, .limit = BOOT_CPUCACHE_ENTRIES, .shared = 1, |
3b0efdfa1
|
404 |
.size = sizeof(struct kmem_cache), |
b28a02de8
|
405 |
.name = "kmem_cache", |
1da177e4c
|
406 |
}; |
1871e52c7
|
407 |
static DEFINE_PER_CPU(struct delayed_work, slab_reap_work); |
1da177e4c
|
408 |
|
343e0d7a9
|
409 |
static inline struct array_cache *cpu_cache_get(struct kmem_cache *cachep) |
1da177e4c
|
410 |
{ |
bf0dea23a
|
411 |
return this_cpu_ptr(cachep->cpu_cache); |
1da177e4c
|
412 |
} |
a737b3e2f
|
413 414 415 |
/* * Calculate the number of objects and left-over bytes for a given buffer size. */ |
70f75067b
|
416 417 |
static unsigned int cache_estimate(unsigned long gfporder, size_t buffer_size, unsigned long flags, size_t *left_over) |
fbaccacff
|
418 |
{ |
70f75067b
|
419 |
unsigned int num; |
fbaccacff
|
420 |
size_t slab_size = PAGE_SIZE << gfporder; |
1da177e4c
|
421 |
|
fbaccacff
|
422 423 424 425 426 |
/* * 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: * |
fbaccacff
|
427 |
* - @buffer_size bytes for each object |
2e6b36021
|
428 429 430 431 432 |
* - One freelist_idx_t for each object * * We don't need to consider alignment of freelist because * freelist will be at the end of slab page. The objects will be * at the correct alignment. |
fbaccacff
|
433 434 435 436 437 438 |
* * 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. */ |
b03a017be
|
439 |
if (flags & (CFLGS_OBJFREELIST_SLAB | CFLGS_OFF_SLAB)) { |
70f75067b
|
440 |
num = slab_size / buffer_size; |
2e6b36021
|
441 |
*left_over = slab_size % buffer_size; |
fbaccacff
|
442 |
} else { |
70f75067b
|
443 |
num = slab_size / (buffer_size + sizeof(freelist_idx_t)); |
2e6b36021
|
444 445 |
*left_over = slab_size % (buffer_size + sizeof(freelist_idx_t)); |
fbaccacff
|
446 |
} |
70f75067b
|
447 448 |
return num; |
1da177e4c
|
449 |
} |
f28510d30
|
450 |
#if DEBUG |
d40cee245
|
451 |
#define slab_error(cachep, msg) __slab_error(__func__, cachep, msg) |
1da177e4c
|
452 |
|
a737b3e2f
|
453 454 |
static void __slab_error(const char *function, struct kmem_cache *cachep, char *msg) |
1da177e4c
|
455 |
{ |
1170532bb
|
456 457 |
pr_err("slab error in %s(): cache `%s': %s ", |
b28a02de8
|
458 |
function, cachep->name, msg); |
1da177e4c
|
459 |
dump_stack(); |
373d4d099
|
460 |
add_taint(TAINT_BAD_PAGE, LOCKDEP_NOW_UNRELIABLE); |
1da177e4c
|
461 |
} |
f28510d30
|
462 |
#endif |
1da177e4c
|
463 |
|
3395ee058
|
464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 |
/* * 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
|
479 480 481 482 483 484 485 486 487 488 |
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
|
489 490 491 492 493 494 495 |
#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
|
496 |
static DEFINE_PER_CPU(unsigned long, slab_reap_node); |
8fce4d8e3
|
497 498 499 |
static void init_reap_node(int cpu) { |
0edaf86cf
|
500 501 |
per_cpu(slab_reap_node, cpu) = next_node_in(cpu_to_mem(cpu), node_online_map); |
8fce4d8e3
|
502 503 504 505 |
} static void next_reap_node(void) { |
909ea9646
|
506 |
int node = __this_cpu_read(slab_reap_node); |
8fce4d8e3
|
507 |
|
0edaf86cf
|
508 |
node = next_node_in(node, node_online_map); |
909ea9646
|
509 |
__this_cpu_write(slab_reap_node, node); |
8fce4d8e3
|
510 511 512 513 514 515 |
} #else #define init_reap_node(cpu) do { } while (0) #define next_reap_node(void) do { } while (0) #endif |
1da177e4c
|
516 517 518 519 520 521 522 |
/* * 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. */ |
0db0628d9
|
523 |
static void start_cpu_timer(int cpu) |
1da177e4c
|
524 |
{ |
1871e52c7
|
525 |
struct delayed_work *reap_work = &per_cpu(slab_reap_work, cpu); |
1da177e4c
|
526 |
|
eac0337af
|
527 |
if (reap_work->work.func == NULL) { |
8fce4d8e3
|
528 |
init_reap_node(cpu); |
203b42f73
|
529 |
INIT_DEFERRABLE_WORK(reap_work, cache_reap); |
2b2842146
|
530 531 |
schedule_delayed_work_on(cpu, reap_work, __round_jiffies_relative(HZ, cpu)); |
1da177e4c
|
532 533 |
} } |
1fe00d50a
|
534 |
static void init_arraycache(struct array_cache *ac, int limit, int batch) |
1da177e4c
|
535 |
{ |
d5cff6352
|
536 537 |
/* * The array_cache structures contain pointers to free object. |
25985edce
|
538 |
* However, when such objects are allocated or transferred to another |
d5cff6352
|
539 540 541 542 |
* 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. */ |
1fe00d50a
|
543 544 545 546 547 548 |
kmemleak_no_scan(ac); if (ac) { ac->avail = 0; ac->limit = limit; ac->batchcount = batch; ac->touched = 0; |
1da177e4c
|
549 |
} |
1fe00d50a
|
550 551 552 553 554 |
} static struct array_cache *alloc_arraycache(int node, int entries, int batchcount, gfp_t gfp) { |
5e8047896
|
555 |
size_t memsize = sizeof(void *) * entries + sizeof(struct array_cache); |
1fe00d50a
|
556 557 558 559 560 |
struct array_cache *ac = NULL; ac = kmalloc_node(memsize, gfp, node); init_arraycache(ac, entries, batchcount); return ac; |
1da177e4c
|
561 |
} |
f68f8dddb
|
562 563 |
static noinline void cache_free_pfmemalloc(struct kmem_cache *cachep, struct page *page, void *objp) |
072bb0aa5
|
564 |
{ |
f68f8dddb
|
565 566 567 |
struct kmem_cache_node *n; int page_node; LIST_HEAD(list); |
072bb0aa5
|
568 |
|
f68f8dddb
|
569 570 |
page_node = page_to_nid(page); n = get_node(cachep, page_node); |
381760ead
|
571 |
|
f68f8dddb
|
572 573 574 |
spin_lock(&n->list_lock); free_block(cachep, &objp, 1, page_node, &list); spin_unlock(&n->list_lock); |
381760ead
|
575 |
|
f68f8dddb
|
576 |
slabs_destroy(cachep, &list); |
072bb0aa5
|
577 |
} |
3ded175a4
|
578 579 580 581 582 583 584 585 586 587 |
/* * 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
|
588 |
int nr = min3(from->avail, max, to->limit - to->avail); |
3ded175a4
|
589 590 591 592 593 594 595 596 597 |
if (!nr) return 0; memcpy(to->entry + to->avail, from->entry + from->avail -nr, sizeof(void *) *nr); from->avail -= nr; to->avail += nr; |
3ded175a4
|
598 599 |
return nr; } |
765c4507a
|
600 601 602 |
#ifndef CONFIG_NUMA #define drain_alien_cache(cachep, alien) do { } while (0) |
ce8eb6c42
|
603 |
#define reap_alien(cachep, n) do { } while (0) |
765c4507a
|
604 |
|
c8522a3a5
|
605 606 |
static inline struct alien_cache **alloc_alien_cache(int node, int limit, gfp_t gfp) |
765c4507a
|
607 |
{ |
8888177ea
|
608 |
return NULL; |
765c4507a
|
609 |
} |
c8522a3a5
|
610 |
static inline void free_alien_cache(struct alien_cache **ac_ptr) |
765c4507a
|
611 612 613 614 615 616 617 618 619 620 621 622 623 |
{ } 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
|
624 |
static inline void *____cache_alloc_node(struct kmem_cache *cachep, |
765c4507a
|
625 626 627 628 |
gfp_t flags, int nodeid) { return NULL; } |
4167e9b2c
|
629 630 |
static inline gfp_t gfp_exact_node(gfp_t flags) { |
444eb2a44
|
631 |
return flags & ~__GFP_NOFAIL; |
4167e9b2c
|
632 |
} |
765c4507a
|
633 |
#else /* CONFIG_NUMA */ |
8b98c1699
|
634 |
static void *____cache_alloc_node(struct kmem_cache *, gfp_t, int); |
c61afb181
|
635 |
static void *alternate_node_alloc(struct kmem_cache *, gfp_t); |
dc85da15d
|
636 |
|
c8522a3a5
|
637 638 639 |
static struct alien_cache *__alloc_alien_cache(int node, int entries, int batch, gfp_t gfp) { |
5e8047896
|
640 |
size_t memsize = sizeof(void *) * entries + sizeof(struct alien_cache); |
c8522a3a5
|
641 642 643 644 |
struct alien_cache *alc = NULL; alc = kmalloc_node(memsize, gfp, node); init_arraycache(&alc->ac, entries, batch); |
49dfc304b
|
645 |
spin_lock_init(&alc->lock); |
c8522a3a5
|
646 647 648 649 |
return alc; } static struct alien_cache **alloc_alien_cache(int node, int limit, gfp_t gfp) |
e498be7da
|
650 |
{ |
c8522a3a5
|
651 |
struct alien_cache **alc_ptr; |
5e8047896
|
652 |
size_t memsize = sizeof(void *) * nr_node_ids; |
e498be7da
|
653 654 655 656 |
int i; if (limit > 1) limit = 12; |
c8522a3a5
|
657 658 659 660 661 662 663 664 665 666 667 668 669 |
alc_ptr = kzalloc_node(memsize, gfp, node); if (!alc_ptr) return NULL; for_each_node(i) { if (i == node || !node_online(i)) continue; alc_ptr[i] = __alloc_alien_cache(node, limit, 0xbaadf00d, gfp); if (!alc_ptr[i]) { for (i--; i >= 0; i--) kfree(alc_ptr[i]); kfree(alc_ptr); return NULL; |
e498be7da
|
670 671 |
} } |
c8522a3a5
|
672 |
return alc_ptr; |
e498be7da
|
673 |
} |
c8522a3a5
|
674 |
static void free_alien_cache(struct alien_cache **alc_ptr) |
e498be7da
|
675 676 |
{ int i; |
c8522a3a5
|
677 |
if (!alc_ptr) |
e498be7da
|
678 |
return; |
e498be7da
|
679 |
for_each_node(i) |
c8522a3a5
|
680 681 |
kfree(alc_ptr[i]); kfree(alc_ptr); |
e498be7da
|
682 |
} |
343e0d7a9
|
683 |
static void __drain_alien_cache(struct kmem_cache *cachep, |
833b706cc
|
684 685 |
struct array_cache *ac, int node, struct list_head *list) |
e498be7da
|
686 |
{ |
18bf85411
|
687 |
struct kmem_cache_node *n = get_node(cachep, node); |
e498be7da
|
688 689 |
if (ac->avail) { |
ce8eb6c42
|
690 |
spin_lock(&n->list_lock); |
e00946fe2
|
691 692 693 694 695 |
/* * 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. */ |
ce8eb6c42
|
696 697 |
if (n->shared) transfer_objects(n->shared, ac, ac->limit); |
e00946fe2
|
698 |
|
833b706cc
|
699 |
free_block(cachep, ac->entry, ac->avail, node, list); |
e498be7da
|
700 |
ac->avail = 0; |
ce8eb6c42
|
701 |
spin_unlock(&n->list_lock); |
e498be7da
|
702 703 |
} } |
8fce4d8e3
|
704 705 706 |
/* * Called from cache_reap() to regularly drain alien caches round robin. */ |
ce8eb6c42
|
707 |
static void reap_alien(struct kmem_cache *cachep, struct kmem_cache_node *n) |
8fce4d8e3
|
708 |
{ |
909ea9646
|
709 |
int node = __this_cpu_read(slab_reap_node); |
8fce4d8e3
|
710 |
|
ce8eb6c42
|
711 |
if (n->alien) { |
c8522a3a5
|
712 713 714 715 716 |
struct alien_cache *alc = n->alien[node]; struct array_cache *ac; if (alc) { ac = &alc->ac; |
49dfc304b
|
717 |
if (ac->avail && spin_trylock_irq(&alc->lock)) { |
833b706cc
|
718 719 720 |
LIST_HEAD(list); __drain_alien_cache(cachep, ac, node, &list); |
49dfc304b
|
721 |
spin_unlock_irq(&alc->lock); |
833b706cc
|
722 |
slabs_destroy(cachep, &list); |
c8522a3a5
|
723 |
} |
8fce4d8e3
|
724 725 726 |
} } } |
a737b3e2f
|
727 |
static void drain_alien_cache(struct kmem_cache *cachep, |
c8522a3a5
|
728 |
struct alien_cache **alien) |
e498be7da
|
729 |
{ |
b28a02de8
|
730 |
int i = 0; |
c8522a3a5
|
731 |
struct alien_cache *alc; |
e498be7da
|
732 733 734 735 |
struct array_cache *ac; unsigned long flags; for_each_online_node(i) { |
c8522a3a5
|
736 737 |
alc = alien[i]; if (alc) { |
833b706cc
|
738 |
LIST_HEAD(list); |
c8522a3a5
|
739 |
ac = &alc->ac; |
49dfc304b
|
740 |
spin_lock_irqsave(&alc->lock, flags); |
833b706cc
|
741 |
__drain_alien_cache(cachep, ac, i, &list); |
49dfc304b
|
742 |
spin_unlock_irqrestore(&alc->lock, flags); |
833b706cc
|
743 |
slabs_destroy(cachep, &list); |
e498be7da
|
744 745 746 |
} } } |
729bd0b74
|
747 |
|
25c4f304b
|
748 749 |
static int __cache_free_alien(struct kmem_cache *cachep, void *objp, int node, int page_node) |
729bd0b74
|
750 |
{ |
ce8eb6c42
|
751 |
struct kmem_cache_node *n; |
c8522a3a5
|
752 753 |
struct alien_cache *alien = NULL; struct array_cache *ac; |
97654dfa2
|
754 |
LIST_HEAD(list); |
1ca4cb241
|
755 |
|
18bf85411
|
756 |
n = get_node(cachep, node); |
729bd0b74
|
757 |
STATS_INC_NODEFREES(cachep); |
25c4f304b
|
758 759 |
if (n->alien && n->alien[page_node]) { alien = n->alien[page_node]; |
c8522a3a5
|
760 |
ac = &alien->ac; |
49dfc304b
|
761 |
spin_lock(&alien->lock); |
c8522a3a5
|
762 |
if (unlikely(ac->avail == ac->limit)) { |
729bd0b74
|
763 |
STATS_INC_ACOVERFLOW(cachep); |
25c4f304b
|
764 |
__drain_alien_cache(cachep, ac, page_node, &list); |
729bd0b74
|
765 |
} |
f68f8dddb
|
766 |
ac->entry[ac->avail++] = objp; |
49dfc304b
|
767 |
spin_unlock(&alien->lock); |
833b706cc
|
768 |
slabs_destroy(cachep, &list); |
729bd0b74
|
769 |
} else { |
25c4f304b
|
770 |
n = get_node(cachep, page_node); |
18bf85411
|
771 |
spin_lock(&n->list_lock); |
25c4f304b
|
772 |
free_block(cachep, &objp, 1, page_node, &list); |
18bf85411
|
773 |
spin_unlock(&n->list_lock); |
97654dfa2
|
774 |
slabs_destroy(cachep, &list); |
729bd0b74
|
775 776 777 |
} return 1; } |
25c4f304b
|
778 779 780 781 782 783 784 785 786 787 788 789 790 791 |
static inline int cache_free_alien(struct kmem_cache *cachep, void *objp) { int page_node = page_to_nid(virt_to_page(objp)); int node = numa_mem_id(); /* * Make sure we are not freeing a object from another node to the array * cache on this cpu. */ if (likely(node == page_node)) return 0; return __cache_free_alien(cachep, objp, node, page_node); } |
4167e9b2c
|
792 793 |
/* |
444eb2a44
|
794 795 |
* Construct gfp mask to allocate from a specific node but do not reclaim or * warn about failures. |
4167e9b2c
|
796 797 798 |
*/ static inline gfp_t gfp_exact_node(gfp_t flags) { |
444eb2a44
|
799 |
return (flags | __GFP_THISNODE | __GFP_NOWARN) & ~(__GFP_RECLAIM|__GFP_NOFAIL); |
4167e9b2c
|
800 |
} |
e498be7da
|
801 |
#endif |
ded0ecf61
|
802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 |
static int init_cache_node(struct kmem_cache *cachep, int node, gfp_t gfp) { struct kmem_cache_node *n; /* * Set up the kmem_cache_node for cpu before we can * begin anything. Make sure some other cpu on this * node has not already allocated this */ n = get_node(cachep, node); if (n) { spin_lock_irq(&n->list_lock); n->free_limit = (1 + nr_cpus_node(node)) * cachep->batchcount + cachep->num; spin_unlock_irq(&n->list_lock); return 0; } n = kmalloc_node(sizeof(struct kmem_cache_node), gfp, node); if (!n) return -ENOMEM; kmem_cache_node_init(n); n->next_reap = jiffies + REAPTIMEOUT_NODE + ((unsigned long)cachep) % REAPTIMEOUT_NODE; n->free_limit = (1 + nr_cpus_node(node)) * cachep->batchcount + cachep->num; /* * The kmem_cache_nodes don't come and go as CPUs * come and go. slab_mutex is sufficient * protection here. */ cachep->node[node] = n; return 0; } |
6731d4f12
|
841 |
#if (defined(CONFIG_NUMA) && defined(CONFIG_MEMORY_HOTPLUG)) || defined(CONFIG_SMP) |
8f9f8d9e8
|
842 |
/* |
6a67368c3
|
843 |
* Allocates and initializes node for a node on each slab cache, used for |
ce8eb6c42
|
844 |
* either memory or cpu hotplug. If memory is being hot-added, the kmem_cache_node |
8f9f8d9e8
|
845 |
* will be allocated off-node since memory is not yet online for the new node. |
6a67368c3
|
846 |
* When hotplugging memory or a cpu, existing node are not replaced if |
8f9f8d9e8
|
847 848 |
* already in use. * |
18004c5d4
|
849 |
* Must hold slab_mutex. |
8f9f8d9e8
|
850 |
*/ |
6a67368c3
|
851 |
static int init_cache_node_node(int node) |
8f9f8d9e8
|
852 |
{ |
ded0ecf61
|
853 |
int ret; |
8f9f8d9e8
|
854 |
struct kmem_cache *cachep; |
8f9f8d9e8
|
855 |
|
18004c5d4
|
856 |
list_for_each_entry(cachep, &slab_caches, list) { |
ded0ecf61
|
857 858 859 |
ret = init_cache_node(cachep, node, GFP_KERNEL); if (ret) return ret; |
8f9f8d9e8
|
860 |
} |
ded0ecf61
|
861 |
|
8f9f8d9e8
|
862 863 |
return 0; } |
6731d4f12
|
864 |
#endif |
8f9f8d9e8
|
865 |
|
c3d332b6b
|
866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 |
static int setup_kmem_cache_node(struct kmem_cache *cachep, int node, gfp_t gfp, bool force_change) { int ret = -ENOMEM; struct kmem_cache_node *n; struct array_cache *old_shared = NULL; struct array_cache *new_shared = NULL; struct alien_cache **new_alien = NULL; LIST_HEAD(list); if (use_alien_caches) { new_alien = alloc_alien_cache(node, cachep->limit, gfp); if (!new_alien) goto fail; } if (cachep->shared) { new_shared = alloc_arraycache(node, cachep->shared * cachep->batchcount, 0xbaadf00d, gfp); if (!new_shared) goto fail; } ret = init_cache_node(cachep, node, gfp); if (ret) goto fail; n = get_node(cachep, node); spin_lock_irq(&n->list_lock); if (n->shared && force_change) { free_block(cachep, n->shared->entry, n->shared->avail, node, &list); n->shared->avail = 0; } if (!n->shared || force_change) { old_shared = n->shared; n->shared = new_shared; new_shared = NULL; } if (!n->alien) { n->alien = new_alien; new_alien = NULL; } spin_unlock_irq(&n->list_lock); slabs_destroy(cachep, &list); |
801faf0db
|
914 915 916 917 918 919 |
/* * To protect lockless access to n->shared during irq disabled context. * If n->shared isn't NULL in irq disabled context, accessing to it is * guaranteed to be valid until irq is re-enabled, because it will be * freed after synchronize_sched(). */ |
86d9f4853
|
920 |
if (old_shared && force_change) |
801faf0db
|
921 |
synchronize_sched(); |
c3d332b6b
|
922 923 924 925 926 927 928 |
fail: kfree(old_shared); kfree(new_shared); free_alien_cache(new_alien); return ret; } |
6731d4f12
|
929 |
#ifdef CONFIG_SMP |
0db0628d9
|
930 |
static void cpuup_canceled(long cpu) |
fbf1e473b
|
931 932 |
{ struct kmem_cache *cachep; |
ce8eb6c42
|
933 |
struct kmem_cache_node *n = NULL; |
7d6e6d09d
|
934 |
int node = cpu_to_mem(cpu); |
a70f73028
|
935 |
const struct cpumask *mask = cpumask_of_node(node); |
fbf1e473b
|
936 |
|
18004c5d4
|
937 |
list_for_each_entry(cachep, &slab_caches, list) { |
fbf1e473b
|
938 939 |
struct array_cache *nc; struct array_cache *shared; |
c8522a3a5
|
940 |
struct alien_cache **alien; |
97654dfa2
|
941 |
LIST_HEAD(list); |
fbf1e473b
|
942 |
|
18bf85411
|
943 |
n = get_node(cachep, node); |
ce8eb6c42
|
944 |
if (!n) |
bf0dea23a
|
945 |
continue; |
fbf1e473b
|
946 |
|
ce8eb6c42
|
947 |
spin_lock_irq(&n->list_lock); |
fbf1e473b
|
948 |
|
ce8eb6c42
|
949 950 |
/* Free limit for this kmem_cache_node */ n->free_limit -= cachep->batchcount; |
bf0dea23a
|
951 952 953 954 |
/* cpu is dead; no one can alloc from it. */ nc = per_cpu_ptr(cachep->cpu_cache, cpu); if (nc) { |
97654dfa2
|
955 |
free_block(cachep, nc->entry, nc->avail, node, &list); |
bf0dea23a
|
956 957 |
nc->avail = 0; } |
fbf1e473b
|
958 |
|
58463c1fe
|
959 |
if (!cpumask_empty(mask)) { |
ce8eb6c42
|
960 |
spin_unlock_irq(&n->list_lock); |
bf0dea23a
|
961 |
goto free_slab; |
fbf1e473b
|
962 |
} |
ce8eb6c42
|
963 |
shared = n->shared; |
fbf1e473b
|
964 965 |
if (shared) { free_block(cachep, shared->entry, |
97654dfa2
|
966 |
shared->avail, node, &list); |
ce8eb6c42
|
967 |
n->shared = NULL; |
fbf1e473b
|
968 |
} |
ce8eb6c42
|
969 970 |
alien = n->alien; n->alien = NULL; |
fbf1e473b
|
971 |
|
ce8eb6c42
|
972 |
spin_unlock_irq(&n->list_lock); |
fbf1e473b
|
973 974 975 976 977 978 |
kfree(shared); if (alien) { drain_alien_cache(cachep, alien); free_alien_cache(alien); } |
bf0dea23a
|
979 980 |
free_slab: |
97654dfa2
|
981 |
slabs_destroy(cachep, &list); |
fbf1e473b
|
982 983 984 985 986 987 |
} /* * 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. */ |
18004c5d4
|
988 |
list_for_each_entry(cachep, &slab_caches, list) { |
18bf85411
|
989 |
n = get_node(cachep, node); |
ce8eb6c42
|
990 |
if (!n) |
fbf1e473b
|
991 |
continue; |
a5aa63a5f
|
992 |
drain_freelist(cachep, n, INT_MAX); |
fbf1e473b
|
993 994 |
} } |
0db0628d9
|
995 |
static int cpuup_prepare(long cpu) |
1da177e4c
|
996 |
{ |
343e0d7a9
|
997 |
struct kmem_cache *cachep; |
7d6e6d09d
|
998 |
int node = cpu_to_mem(cpu); |
8f9f8d9e8
|
999 |
int err; |
1da177e4c
|
1000 |
|
fbf1e473b
|
1001 1002 1003 1004 |
/* * 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 |
ce8eb6c42
|
1005 |
* kmem_cache_node and not this cpu's kmem_cache_node |
fbf1e473b
|
1006 |
*/ |
6a67368c3
|
1007 |
err = init_cache_node_node(node); |
8f9f8d9e8
|
1008 1009 |
if (err < 0) goto bad; |
fbf1e473b
|
1010 1011 1012 1013 1014 |
/* * Now we can go ahead with allocating the shared arrays and * array caches */ |
18004c5d4
|
1015 |
list_for_each_entry(cachep, &slab_caches, list) { |
c3d332b6b
|
1016 1017 1018 |
err = setup_kmem_cache_node(cachep, node, GFP_KERNEL, false); if (err) goto bad; |
fbf1e473b
|
1019 |
} |
ce79ddc8e
|
1020 |
|
fbf1e473b
|
1021 1022 |
return 0; bad: |
12d00f6a1
|
1023 |
cpuup_canceled(cpu); |
fbf1e473b
|
1024 1025 |
return -ENOMEM; } |
6731d4f12
|
1026 |
int slab_prepare_cpu(unsigned int cpu) |
fbf1e473b
|
1027 |
{ |
6731d4f12
|
1028 |
int err; |
fbf1e473b
|
1029 |
|
6731d4f12
|
1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 |
mutex_lock(&slab_mutex); err = cpuup_prepare(cpu); mutex_unlock(&slab_mutex); return err; } /* * This is called for a failed online attempt and for a successful * offline. * * 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(). */ int slab_dead_cpu(unsigned int cpu) { mutex_lock(&slab_mutex); cpuup_canceled(cpu); mutex_unlock(&slab_mutex); return 0; } |
8f5be20bf
|
1053 |
#endif |
6731d4f12
|
1054 1055 1056 1057 1058 |
static int slab_online_cpu(unsigned int cpu) { start_cpu_timer(cpu); return 0; |
1da177e4c
|
1059 |
} |
6731d4f12
|
1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 |
static int slab_offline_cpu(unsigned int cpu) { /* * Shutdown cache reaper. Note that the slab_mutex is held so * that if cache_reap() is invoked it cannot do anything * expensive but will only modify reap_work and reschedule the * timer. */ cancel_delayed_work_sync(&per_cpu(slab_reap_work, cpu)); /* Now the cache_reaper is guaranteed to be not running. */ per_cpu(slab_reap_work, cpu).work.func = NULL; return 0; } |
1da177e4c
|
1073 |
|
8f9f8d9e8
|
1074 1075 1076 1077 1078 1079 |
#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. * |
18004c5d4
|
1080 |
* Must hold slab_mutex. |
8f9f8d9e8
|
1081 |
*/ |
6a67368c3
|
1082 |
static int __meminit drain_cache_node_node(int node) |
8f9f8d9e8
|
1083 1084 1085 |
{ struct kmem_cache *cachep; int ret = 0; |
18004c5d4
|
1086 |
list_for_each_entry(cachep, &slab_caches, list) { |
ce8eb6c42
|
1087 |
struct kmem_cache_node *n; |
8f9f8d9e8
|
1088 |
|
18bf85411
|
1089 |
n = get_node(cachep, node); |
ce8eb6c42
|
1090 |
if (!n) |
8f9f8d9e8
|
1091 |
continue; |
a5aa63a5f
|
1092 |
drain_freelist(cachep, n, INT_MAX); |
8f9f8d9e8
|
1093 |
|
ce8eb6c42
|
1094 1095 |
if (!list_empty(&n->slabs_full) || !list_empty(&n->slabs_partial)) { |
8f9f8d9e8
|
1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 |
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: |
18004c5d4
|
1116 |
mutex_lock(&slab_mutex); |
6a67368c3
|
1117 |
ret = init_cache_node_node(nid); |
18004c5d4
|
1118 |
mutex_unlock(&slab_mutex); |
8f9f8d9e8
|
1119 1120 |
break; case MEM_GOING_OFFLINE: |
18004c5d4
|
1121 |
mutex_lock(&slab_mutex); |
6a67368c3
|
1122 |
ret = drain_cache_node_node(nid); |
18004c5d4
|
1123 |
mutex_unlock(&slab_mutex); |
8f9f8d9e8
|
1124 1125 1126 1127 1128 1129 1130 1131 |
break; case MEM_ONLINE: case MEM_OFFLINE: case MEM_CANCEL_ONLINE: case MEM_CANCEL_OFFLINE: break; } out: |
5fda1bd5b
|
1132 |
return notifier_from_errno(ret); |
8f9f8d9e8
|
1133 1134 |
} #endif /* CONFIG_NUMA && CONFIG_MEMORY_HOTPLUG */ |
e498be7da
|
1135 |
/* |
ce8eb6c42
|
1136 |
* swap the static kmem_cache_node with kmalloced memory |
e498be7da
|
1137 |
*/ |
6744f087b
|
1138 |
static void __init init_list(struct kmem_cache *cachep, struct kmem_cache_node *list, |
8f9f8d9e8
|
1139 |
int nodeid) |
e498be7da
|
1140 |
{ |
6744f087b
|
1141 |
struct kmem_cache_node *ptr; |
e498be7da
|
1142 |
|
6744f087b
|
1143 |
ptr = kmalloc_node(sizeof(struct kmem_cache_node), GFP_NOWAIT, nodeid); |
e498be7da
|
1144 |
BUG_ON(!ptr); |
6744f087b
|
1145 |
memcpy(ptr, list, sizeof(struct kmem_cache_node)); |
2b2d5493e
|
1146 1147 1148 1149 |
/* * Do not assume that spinlocks can be initialized via memcpy: */ spin_lock_init(&ptr->list_lock); |
e498be7da
|
1150 |
MAKE_ALL_LISTS(cachep, ptr, nodeid); |
6a67368c3
|
1151 |
cachep->node[nodeid] = ptr; |
e498be7da
|
1152 |
} |
a737b3e2f
|
1153 |
/* |
ce8eb6c42
|
1154 1155 |
* For setting up all the kmem_cache_node for cache whose buffer_size is same as * size of kmem_cache_node. |
556a169da
|
1156 |
*/ |
ce8eb6c42
|
1157 |
static void __init set_up_node(struct kmem_cache *cachep, int index) |
556a169da
|
1158 1159 1160 1161 |
{ int node; for_each_online_node(node) { |
ce8eb6c42
|
1162 |
cachep->node[node] = &init_kmem_cache_node[index + node]; |
6a67368c3
|
1163 |
cachep->node[node]->next_reap = jiffies + |
5f0985bb1
|
1164 1165 |
REAPTIMEOUT_NODE + ((unsigned long)cachep) % REAPTIMEOUT_NODE; |
556a169da
|
1166 1167 1168 1169 |
} } /* |
a737b3e2f
|
1170 1171 |
* Initialisation. Called after the page allocator have been initialised and * before smp_init(). |
1da177e4c
|
1172 1173 1174 |
*/ void __init kmem_cache_init(void) { |
e498be7da
|
1175 |
int i; |
68126702b
|
1176 1177 |
BUILD_BUG_ON(sizeof(((struct page *)NULL)->lru) < sizeof(struct rcu_head)); |
9b030cb86
|
1178 |
kmem_cache = &kmem_cache_boot; |
8888177ea
|
1179 |
if (!IS_ENABLED(CONFIG_NUMA) || num_possible_nodes() == 1) |
62918a036
|
1180 |
use_alien_caches = 0; |
3c5834652
|
1181 |
for (i = 0; i < NUM_INIT_LISTS; i++) |
ce8eb6c42
|
1182 |
kmem_cache_node_init(&init_kmem_cache_node[i]); |
3c5834652
|
1183 |
|
1da177e4c
|
1184 1185 |
/* * Fragmentation resistance on low memory - only use bigger |
3df1cccdf
|
1186 1187 |
* page orders on machines with more than 32MB of memory if * not overridden on the command line. |
1da177e4c
|
1188 |
*/ |
3df1cccdf
|
1189 |
if (!slab_max_order_set && totalram_pages > (32 << 20) >> PAGE_SHIFT) |
543585cc5
|
1190 |
slab_max_order = SLAB_MAX_ORDER_HI; |
1da177e4c
|
1191 |
|
1da177e4c
|
1192 1193 |
/* Bootstrap is tricky, because several objects are allocated * from caches that do not exist yet: |
9b030cb86
|
1194 1195 1196 |
* 1) initialize the kmem_cache cache: it contains the struct * kmem_cache structures of all caches, except kmem_cache itself: * kmem_cache is statically allocated. |
e498be7da
|
1197 |
* Initially an __init data area is used for the head array and the |
ce8eb6c42
|
1198 |
* kmem_cache_node structures, it's replaced with a kmalloc allocated |
e498be7da
|
1199 |
* array at the end of the bootstrap. |
1da177e4c
|
1200 |
* 2) Create the first kmalloc cache. |
343e0d7a9
|
1201 |
* The struct kmem_cache for the new cache is allocated normally. |
e498be7da
|
1202 1203 1204 |
* An __init data area is used for the head array. * 3) Create the remaining kmalloc caches, with minimally sized * head arrays. |
9b030cb86
|
1205 |
* 4) Replace the __init data head arrays for kmem_cache and the first |
1da177e4c
|
1206 |
* kmalloc cache with kmalloc allocated arrays. |
ce8eb6c42
|
1207 |
* 5) Replace the __init data for kmem_cache_node for kmem_cache and |
e498be7da
|
1208 1209 |
* the other cache's with kmalloc allocated memory. * 6) Resize the head arrays of the kmalloc caches to their final sizes. |
1da177e4c
|
1210 |
*/ |
9b030cb86
|
1211 |
/* 1) create the kmem_cache */ |
1da177e4c
|
1212 |
|
8da3430d8
|
1213 |
/* |
b56efcf0a
|
1214 |
* struct kmem_cache size depends on nr_node_ids & nr_cpu_ids |
8da3430d8
|
1215 |
*/ |
2f9baa9fc
|
1216 |
create_boot_cache(kmem_cache, "kmem_cache", |
bf0dea23a
|
1217 |
offsetof(struct kmem_cache, node) + |
6744f087b
|
1218 |
nr_node_ids * sizeof(struct kmem_cache_node *), |
2f9baa9fc
|
1219 1220 |
SLAB_HWCACHE_ALIGN); list_add(&kmem_cache->list, &slab_caches); |
da9ec481d
|
1221 |
memcg_link_cache(kmem_cache); |
bf0dea23a
|
1222 |
slab_state = PARTIAL; |
1da177e4c
|
1223 |
|
a737b3e2f
|
1224 |
/* |
bf0dea23a
|
1225 1226 |
* Initialize the caches that provide memory for the kmem_cache_node * structures first. Without this, further allocations will bug. |
e498be7da
|
1227 |
*/ |
af3b5f876
|
1228 1229 |
kmalloc_caches[INDEX_NODE] = create_kmalloc_cache( kmalloc_info[INDEX_NODE].name, |
ce8eb6c42
|
1230 |
kmalloc_size(INDEX_NODE), ARCH_KMALLOC_FLAGS); |
bf0dea23a
|
1231 |
slab_state = PARTIAL_NODE; |
34cc6990d
|
1232 |
setup_kmalloc_cache_index_table(); |
e498be7da
|
1233 |
|
e0a427267
|
1234 |
slab_early_init = 0; |
ce8eb6c42
|
1235 |
/* 5) Replace the bootstrap kmem_cache_node */ |
e498be7da
|
1236 |
{ |
1ca4cb241
|
1237 |
int nid; |
9c09a95cf
|
1238 |
for_each_online_node(nid) { |
ce8eb6c42
|
1239 |
init_list(kmem_cache, &init_kmem_cache_node[CACHE_CACHE + nid], nid); |
556a169da
|
1240 |
|
bf0dea23a
|
1241 |
init_list(kmalloc_caches[INDEX_NODE], |
ce8eb6c42
|
1242 |
&init_kmem_cache_node[SIZE_NODE + nid], nid); |
e498be7da
|
1243 1244 |
} } |
1da177e4c
|
1245 |
|
f97d5f634
|
1246 |
create_kmalloc_caches(ARCH_KMALLOC_FLAGS); |
8429db5c6
|
1247 1248 1249 1250 1251 |
} void __init kmem_cache_init_late(void) { struct kmem_cache *cachep; |
97d066091
|
1252 |
slab_state = UP; |
52cef1891
|
1253 |
|
8429db5c6
|
1254 |
/* 6) resize the head arrays to their final sizes */ |
18004c5d4
|
1255 1256 |
mutex_lock(&slab_mutex); list_for_each_entry(cachep, &slab_caches, list) |
8429db5c6
|
1257 1258 |
if (enable_cpucache(cachep, GFP_NOWAIT)) BUG(); |
18004c5d4
|
1259 |
mutex_unlock(&slab_mutex); |
056c62418
|
1260 |
|
97d066091
|
1261 1262 |
/* Done! */ slab_state = FULL; |
8f9f8d9e8
|
1263 1264 1265 |
#ifdef CONFIG_NUMA /* * Register a memory hotplug callback that initializes and frees |
6a67368c3
|
1266 |
* node. |
8f9f8d9e8
|
1267 1268 1269 |
*/ hotplug_memory_notifier(slab_memory_callback, SLAB_CALLBACK_PRI); #endif |
a737b3e2f
|
1270 1271 1272 |
/* * The reap timers are started later, with a module init call: That part * of the kernel is not yet operational. |
1da177e4c
|
1273 1274 1275 1276 1277 |
*/ } static int __init cpucache_init(void) { |
6731d4f12
|
1278 |
int ret; |
1da177e4c
|
1279 |
|
a737b3e2f
|
1280 1281 |
/* * Register the timers that return unneeded pages to the page allocator |
1da177e4c
|
1282 |
*/ |
6731d4f12
|
1283 1284 1285 |
ret = cpuhp_setup_state(CPUHP_AP_ONLINE_DYN, "SLAB online", slab_online_cpu, slab_offline_cpu); WARN_ON(ret < 0); |
a164f8962
|
1286 1287 |
/* Done! */ |
97d066091
|
1288 |
slab_state = FULL; |
1da177e4c
|
1289 1290 |
return 0; } |
1da177e4c
|
1291 |
__initcall(cpucache_init); |
8bdec192b
|
1292 1293 1294 |
static noinline void slab_out_of_memory(struct kmem_cache *cachep, gfp_t gfpflags, int nodeid) { |
9a02d6999
|
1295 |
#if DEBUG |
ce8eb6c42
|
1296 |
struct kmem_cache_node *n; |
8bdec192b
|
1297 1298 |
unsigned long flags; int node; |
9a02d6999
|
1299 1300 1301 1302 1303 |
static DEFINE_RATELIMIT_STATE(slab_oom_rs, DEFAULT_RATELIMIT_INTERVAL, DEFAULT_RATELIMIT_BURST); if ((gfpflags & __GFP_NOWARN) || !__ratelimit(&slab_oom_rs)) return; |
8bdec192b
|
1304 |
|
5b3810e5c
|
1305 1306 1307 1308 1309 |
pr_warn("SLAB: Unable to allocate memory on node %d, gfp=%#x(%pGg) ", nodeid, gfpflags, &gfpflags); pr_warn(" cache: %s, object size: %d, order: %d ", |
3b0efdfa1
|
1310 |
cachep->name, cachep->size, cachep->gfporder); |
8bdec192b
|
1311 |
|
18bf85411
|
1312 |
for_each_kmem_cache_node(cachep, node, n) { |
bf00bd345
|
1313 |
unsigned long total_slabs, free_slabs, free_objs; |
8bdec192b
|
1314 |
|
ce8eb6c42
|
1315 |
spin_lock_irqsave(&n->list_lock, flags); |
bf00bd345
|
1316 1317 1318 |
total_slabs = n->total_slabs; free_slabs = n->free_slabs; free_objs = n->free_objects; |
ce8eb6c42
|
1319 |
spin_unlock_irqrestore(&n->list_lock, flags); |
8bdec192b
|
1320 |
|
bf00bd345
|
1321 1322 1323 1324 1325 |
pr_warn(" node %d: slabs: %ld/%ld, objs: %ld/%ld ", node, total_slabs - free_slabs, total_slabs, (total_slabs * cachep->num) - free_objs, total_slabs * cachep->num); |
8bdec192b
|
1326 |
} |
9a02d6999
|
1327 |
#endif |
8bdec192b
|
1328 |
} |
1da177e4c
|
1329 |
/* |
8a7d9b430
|
1330 1331 |
* Interface to system's page allocator. No need to hold the * kmem_cache_node ->list_lock. |
1da177e4c
|
1332 1333 1334 1335 1336 |
* * 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. */ |
0c3aa83e0
|
1337 1338 |
static struct page *kmem_getpages(struct kmem_cache *cachep, gfp_t flags, int nodeid) |
1da177e4c
|
1339 1340 |
{ struct page *page; |
e1b6aa6f1
|
1341 |
int nr_pages; |
765c4507a
|
1342 |
|
a618e89f1
|
1343 |
flags |= cachep->allocflags; |
e12ba74d8
|
1344 1345 |
if (cachep->flags & SLAB_RECLAIM_ACCOUNT) flags |= __GFP_RECLAIMABLE; |
e1b6aa6f1
|
1346 |
|
ae63fd26b
|
1347 |
page = __alloc_pages_node(nodeid, flags, cachep->gfporder); |
8bdec192b
|
1348 |
if (!page) { |
9a02d6999
|
1349 |
slab_out_of_memory(cachep, flags, nodeid); |
1da177e4c
|
1350 |
return NULL; |
8bdec192b
|
1351 |
} |
1da177e4c
|
1352 |
|
f3ccb2c42
|
1353 1354 1355 1356 |
if (memcg_charge_slab(page, flags, cachep->gfporder, cachep)) { __free_pages(page, cachep->gfporder); return NULL; } |
e1b6aa6f1
|
1357 |
nr_pages = (1 << cachep->gfporder); |
1da177e4c
|
1358 |
if (cachep->flags & SLAB_RECLAIM_ACCOUNT) |
7779f2123
|
1359 |
mod_lruvec_page_state(page, NR_SLAB_RECLAIMABLE, nr_pages); |
972d1a7b1
|
1360 |
else |
7779f2123
|
1361 |
mod_lruvec_page_state(page, NR_SLAB_UNRECLAIMABLE, nr_pages); |
f68f8dddb
|
1362 |
|
a57a49887
|
1363 |
__SetPageSlab(page); |
f68f8dddb
|
1364 1365 |
/* Record if ALLOC_NO_WATERMARKS was set when allocating the slab */ if (sk_memalloc_socks() && page_is_pfmemalloc(page)) |
a57a49887
|
1366 |
SetPageSlabPfmemalloc(page); |
072bb0aa5
|
1367 |
|
0c3aa83e0
|
1368 |
return page; |
1da177e4c
|
1369 1370 1371 1372 1373 |
} /* * Interface to system's page release. */ |
0c3aa83e0
|
1374 |
static void kmem_freepages(struct kmem_cache *cachep, struct page *page) |
1da177e4c
|
1375 |
{ |
27ee57c93
|
1376 1377 |
int order = cachep->gfporder; unsigned long nr_freed = (1 << order); |
1da177e4c
|
1378 |
|
972d1a7b1
|
1379 |
if (cachep->flags & SLAB_RECLAIM_ACCOUNT) |
7779f2123
|
1380 |
mod_lruvec_page_state(page, NR_SLAB_RECLAIMABLE, -nr_freed); |
972d1a7b1
|
1381 |
else |
7779f2123
|
1382 |
mod_lruvec_page_state(page, NR_SLAB_UNRECLAIMABLE, -nr_freed); |
73293c2f9
|
1383 |
|
a57a49887
|
1384 |
BUG_ON(!PageSlab(page)); |
73293c2f9
|
1385 |
__ClearPageSlabPfmemalloc(page); |
a57a49887
|
1386 |
__ClearPageSlab(page); |
8456a648c
|
1387 1388 |
page_mapcount_reset(page); page->mapping = NULL; |
1f458cbf1
|
1389 |
|
1da177e4c
|
1390 1391 |
if (current->reclaim_state) current->reclaim_state->reclaimed_slab += nr_freed; |
27ee57c93
|
1392 1393 |
memcg_uncharge_slab(page, order, cachep); __free_pages(page, order); |
1da177e4c
|
1394 1395 1396 1397 |
} static void kmem_rcu_free(struct rcu_head *head) { |
68126702b
|
1398 1399 |
struct kmem_cache *cachep; struct page *page; |
1da177e4c
|
1400 |
|
68126702b
|
1401 1402 1403 1404 |
page = container_of(head, struct page, rcu_head); cachep = page->slab_cache; kmem_freepages(cachep, page); |
1da177e4c
|
1405 1406 1407 |
} #if DEBUG |
40b441379
|
1408 1409 1410 1411 1412 1413 1414 1415 |
static bool is_debug_pagealloc_cache(struct kmem_cache *cachep) { if (debug_pagealloc_enabled() && OFF_SLAB(cachep) && (cachep->size % PAGE_SIZE) == 0) return true; return false; } |
1da177e4c
|
1416 1417 |
#ifdef CONFIG_DEBUG_PAGEALLOC |
343e0d7a9
|
1418 |
static void store_stackinfo(struct kmem_cache *cachep, unsigned long *addr, |
b28a02de8
|
1419 |
unsigned long caller) |
1da177e4c
|
1420 |
{ |
8c138bc00
|
1421 |
int size = cachep->object_size; |
1da177e4c
|
1422 |
|
3dafccf22
|
1423 |
addr = (unsigned long *)&((char *)addr)[obj_offset(cachep)]; |
1da177e4c
|
1424 |
|
b28a02de8
|
1425 |
if (size < 5 * sizeof(unsigned long)) |
1da177e4c
|
1426 |
return; |
b28a02de8
|
1427 1428 1429 1430 |
*addr++ = 0x12345678; *addr++ = caller; *addr++ = smp_processor_id(); size -= 3 * sizeof(unsigned long); |
1da177e4c
|
1431 1432 1433 1434 1435 1436 1437 |
{ unsigned long *sptr = &caller; unsigned long svalue; while (!kstack_end(sptr)) { svalue = *sptr++; if (kernel_text_address(svalue)) { |
b28a02de8
|
1438 |
*addr++ = svalue; |
1da177e4c
|
1439 1440 1441 1442 1443 1444 1445 |
size -= sizeof(unsigned long); if (size <= sizeof(unsigned long)) break; } } } |
b28a02de8
|
1446 |
*addr++ = 0x87654321; |
1da177e4c
|
1447 |
} |
40b441379
|
1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 |
static void slab_kernel_map(struct kmem_cache *cachep, void *objp, int map, unsigned long caller) { if (!is_debug_pagealloc_cache(cachep)) return; if (caller) store_stackinfo(cachep, objp, caller); kernel_map_pages(virt_to_page(objp), cachep->size / PAGE_SIZE, map); } #else static inline void slab_kernel_map(struct kmem_cache *cachep, void *objp, int map, unsigned long caller) {} |
1da177e4c
|
1464 |
#endif |
343e0d7a9
|
1465 |
static void poison_obj(struct kmem_cache *cachep, void *addr, unsigned char val) |
1da177e4c
|
1466 |
{ |
8c138bc00
|
1467 |
int size = cachep->object_size; |
3dafccf22
|
1468 |
addr = &((char *)addr)[obj_offset(cachep)]; |
1da177e4c
|
1469 1470 |
memset(addr, val, size); |
b28a02de8
|
1471 |
*(unsigned char *)(addr + size - 1) = POISON_END; |
1da177e4c
|
1472 1473 1474 1475 1476 |
} static void dump_line(char *data, int offset, int limit) { int i; |
aa83aa40e
|
1477 1478 |
unsigned char error = 0; int bad_count = 0; |
1170532bb
|
1479 |
pr_err("%03x: ", offset); |
aa83aa40e
|
1480 1481 1482 1483 1484 |
for (i = 0; i < limit; i++) { if (data[offset + i] != POISON_FREE) { error = data[offset + i]; bad_count++; } |
aa83aa40e
|
1485 |
} |
fdde6abb3
|
1486 1487 |
print_hex_dump(KERN_CONT, "", 0, 16, 1, &data[offset], limit, 1); |
aa83aa40e
|
1488 1489 1490 1491 |
if (bad_count == 1) { error ^= POISON_FREE; if (!(error & (error - 1))) { |
1170532bb
|
1492 1493 |
pr_err("Single bit error detected. Probably bad RAM. "); |
aa83aa40e
|
1494 |
#ifdef CONFIG_X86 |
1170532bb
|
1495 1496 |
pr_err("Run memtest86+ or a similar memory test tool. "); |
aa83aa40e
|
1497 |
#else |
1170532bb
|
1498 1499 |
pr_err("Run a memory test tool. "); |
aa83aa40e
|
1500 1501 1502 |
#endif } } |
1da177e4c
|
1503 1504 1505 1506 |
} #endif #if DEBUG |
343e0d7a9
|
1507 |
static void print_objinfo(struct kmem_cache *cachep, void *objp, int lines) |
1da177e4c
|
1508 1509 1510 1511 1512 |
{ int i, size; char *realobj; if (cachep->flags & SLAB_RED_ZONE) { |
1170532bb
|
1513 1514 1515 1516 |
pr_err("Redzone: 0x%llx/0x%llx ", *dbg_redzone1(cachep, objp), *dbg_redzone2(cachep, objp)); |
1da177e4c
|
1517 1518 1519 |
} if (cachep->flags & SLAB_STORE_USER) { |
1170532bb
|
1520 1521 |
pr_err("Last user: [<%p>](%pSR) ", |
071361d34
|
1522 1523 |
*dbg_userword(cachep, objp), *dbg_userword(cachep, objp)); |
1da177e4c
|
1524 |
} |
3dafccf22
|
1525 |
realobj = (char *)objp + obj_offset(cachep); |
8c138bc00
|
1526 |
size = cachep->object_size; |
b28a02de8
|
1527 |
for (i = 0; i < size && lines; i += 16, lines--) { |
1da177e4c
|
1528 1529 |
int limit; limit = 16; |
b28a02de8
|
1530 1531 |
if (i + limit > size) limit = size - i; |
1da177e4c
|
1532 1533 1534 |
dump_line(realobj, i, limit); } } |
343e0d7a9
|
1535 |
static void check_poison_obj(struct kmem_cache *cachep, void *objp) |
1da177e4c
|
1536 1537 1538 1539 |
{ char *realobj; int size, i; int lines = 0; |
40b441379
|
1540 1541 |
if (is_debug_pagealloc_cache(cachep)) return; |
3dafccf22
|
1542 |
realobj = (char *)objp + obj_offset(cachep); |
8c138bc00
|
1543 |
size = cachep->object_size; |
1da177e4c
|
1544 |
|
b28a02de8
|
1545 |
for (i = 0; i < size; i++) { |
1da177e4c
|
1546 |
char exp = POISON_FREE; |
b28a02de8
|
1547 |
if (i == size - 1) |
1da177e4c
|
1548 1549 1550 1551 1552 1553 |
exp = POISON_END; if (realobj[i] != exp) { int limit; /* Mismatch ! */ /* Print header */ if (lines == 0) { |
1170532bb
|
1554 1555 1556 1557 |
pr_err("Slab corruption (%s): %s start=%p, len=%d ", print_tainted(), cachep->name, realobj, size); |
1da177e4c
|
1558 1559 1560 |
print_objinfo(cachep, objp, 0); } /* Hexdump the affected line */ |
b28a02de8
|
1561 |
i = (i / 16) * 16; |
1da177e4c
|
1562 |
limit = 16; |
b28a02de8
|
1563 1564 |
if (i + limit > size) limit = size - i; |
1da177e4c
|
1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 |
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: */ |
8456a648c
|
1577 |
struct page *page = virt_to_head_page(objp); |
8fea4e96a
|
1578 |
unsigned int objnr; |
1da177e4c
|
1579 |
|
8456a648c
|
1580 |
objnr = obj_to_index(cachep, page, objp); |
1da177e4c
|
1581 |
if (objnr) { |
8456a648c
|
1582 |
objp = index_to_obj(cachep, page, objnr - 1); |
3dafccf22
|
1583 |
realobj = (char *)objp + obj_offset(cachep); |
1170532bb
|
1584 1585 |
pr_err("Prev obj: start=%p, len=%d ", realobj, size); |
1da177e4c
|
1586 1587 |
print_objinfo(cachep, objp, 2); } |
b28a02de8
|
1588 |
if (objnr + 1 < cachep->num) { |
8456a648c
|
1589 |
objp = index_to_obj(cachep, page, objnr + 1); |
3dafccf22
|
1590 |
realobj = (char *)objp + obj_offset(cachep); |
1170532bb
|
1591 1592 |
pr_err("Next obj: start=%p, len=%d ", realobj, size); |
1da177e4c
|
1593 1594 1595 1596 1597 |
print_objinfo(cachep, objp, 2); } } } #endif |
12dd36fae
|
1598 |
#if DEBUG |
8456a648c
|
1599 1600 |
static void slab_destroy_debugcheck(struct kmem_cache *cachep, struct page *page) |
1da177e4c
|
1601 |
{ |
1da177e4c
|
1602 |
int i; |
b03a017be
|
1603 1604 1605 1606 1607 |
if (OBJFREELIST_SLAB(cachep) && cachep->flags & SLAB_POISON) { poison_obj(cachep, page->freelist - obj_offset(cachep), POISON_FREE); } |
1da177e4c
|
1608 |
for (i = 0; i < cachep->num; i++) { |
8456a648c
|
1609 |
void *objp = index_to_obj(cachep, page, i); |
1da177e4c
|
1610 1611 |
if (cachep->flags & SLAB_POISON) { |
1da177e4c
|
1612 |
check_poison_obj(cachep, objp); |
40b441379
|
1613 |
slab_kernel_map(cachep, objp, 1, 0); |
1da177e4c
|
1614 1615 1616 |
} if (cachep->flags & SLAB_RED_ZONE) { if (*dbg_redzone1(cachep, objp) != RED_INACTIVE) |
756a025f0
|
1617 |
slab_error(cachep, "start of a freed object was overwritten"); |
1da177e4c
|
1618 |
if (*dbg_redzone2(cachep, objp) != RED_INACTIVE) |
756a025f0
|
1619 |
slab_error(cachep, "end of a freed object was overwritten"); |
1da177e4c
|
1620 |
} |
1da177e4c
|
1621 |
} |
12dd36fae
|
1622 |
} |
1da177e4c
|
1623 |
#else |
8456a648c
|
1624 1625 |
static void slab_destroy_debugcheck(struct kmem_cache *cachep, struct page *page) |
12dd36fae
|
1626 |
{ |
12dd36fae
|
1627 |
} |
1da177e4c
|
1628 |
#endif |
911851e6e
|
1629 1630 1631 |
/** * slab_destroy - destroy and release all objects in a slab * @cachep: cache pointer being destroyed |
cb8ee1a3d
|
1632 |
* @page: page pointer being destroyed |
911851e6e
|
1633 |
* |
8a7d9b430
|
1634 1635 1636 |
* Destroy all the objs in a slab page, and release the mem back to the system. * Before calling the slab page must have been unlinked from the cache. The * kmem_cache_node ->list_lock is not held/needed. |
12dd36fae
|
1637 |
*/ |
8456a648c
|
1638 |
static void slab_destroy(struct kmem_cache *cachep, struct page *page) |
12dd36fae
|
1639 |
{ |
7e0073552
|
1640 |
void *freelist; |
12dd36fae
|
1641 |
|
8456a648c
|
1642 1643 |
freelist = page->freelist; slab_destroy_debugcheck(cachep, page); |
5f0d5a3ae
|
1644 |
if (unlikely(cachep->flags & SLAB_TYPESAFE_BY_RCU)) |
bc4f610d5
|
1645 1646 |
call_rcu(&page->rcu_head, kmem_rcu_free); else |
0c3aa83e0
|
1647 |
kmem_freepages(cachep, page); |
68126702b
|
1648 1649 |
/* |
8456a648c
|
1650 |
* From now on, we don't use freelist |
68126702b
|
1651 1652 1653 |
* although actual page can be freed in rcu context */ if (OFF_SLAB(cachep)) |
8456a648c
|
1654 |
kmem_cache_free(cachep->freelist_cache, freelist); |
1da177e4c
|
1655 |
} |
97654dfa2
|
1656 1657 1658 1659 1660 1661 1662 1663 1664 |
static void slabs_destroy(struct kmem_cache *cachep, struct list_head *list) { struct page *page, *n; list_for_each_entry_safe(page, n, list, lru) { list_del(&page->lru); slab_destroy(cachep, page); } } |
1da177e4c
|
1665 |
/** |
a70773ddb
|
1666 1667 1668 |
* 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. |
a70773ddb
|
1669 1670 1671 |
* @flags: slab allocation flags * * Also calculates the number of objects per slab. |
4d268eba1
|
1672 1673 1674 1675 1676 |
* * 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
|
1677 |
static size_t calculate_slab_order(struct kmem_cache *cachep, |
2e6b36021
|
1678 |
size_t size, unsigned long flags) |
4d268eba1
|
1679 1680 |
{ size_t left_over = 0; |
9888e6fa7
|
1681 |
int gfporder; |
4d268eba1
|
1682 |
|
0aa817f07
|
1683 |
for (gfporder = 0; gfporder <= KMALLOC_MAX_ORDER; gfporder++) { |
4d268eba1
|
1684 1685 |
unsigned int num; size_t remainder; |
70f75067b
|
1686 |
num = cache_estimate(gfporder, size, flags, &remainder); |
4d268eba1
|
1687 1688 |
if (!num) continue; |
9888e6fa7
|
1689 |
|
f315e3fa1
|
1690 1691 1692 |
/* Can't handle number of objects more than SLAB_OBJ_MAX_NUM */ if (num > SLAB_OBJ_MAX_NUM) break; |
b1ab41c49
|
1693 |
if (flags & CFLGS_OFF_SLAB) { |
3217fd9bd
|
1694 1695 1696 1697 1698 1699 1700 |
struct kmem_cache *freelist_cache; size_t freelist_size; freelist_size = num * sizeof(freelist_idx_t); freelist_cache = kmalloc_slab(freelist_size, 0u); if (!freelist_cache) continue; |
b1ab41c49
|
1701 |
/* |
3217fd9bd
|
1702 |
* Needed to avoid possible looping condition |
76b342bdc
|
1703 |
* in cache_grow_begin() |
b1ab41c49
|
1704 |
*/ |
3217fd9bd
|
1705 1706 |
if (OFF_SLAB(freelist_cache)) continue; |
b1ab41c49
|
1707 |
|
3217fd9bd
|
1708 1709 1710 |
/* check if off slab has enough benefit */ if (freelist_cache->size > cachep->size / 2) continue; |
b1ab41c49
|
1711 |
} |
4d268eba1
|
1712 |
|
9888e6fa7
|
1713 |
/* Found something acceptable - save it away */ |
4d268eba1
|
1714 |
cachep->num = num; |
9888e6fa7
|
1715 |
cachep->gfporder = gfporder; |
4d268eba1
|
1716 1717 1718 |
left_over = remainder; /* |
f78bb8ad4
|
1719 1720 1721 1722 1723 1724 1725 1726 |
* 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
|
1727 1728 1729 |
* Large number of objects is good, but very large slabs are * currently bad for the gfp()s. */ |
543585cc5
|
1730 |
if (gfporder >= slab_max_order) |
4d268eba1
|
1731 |
break; |
9888e6fa7
|
1732 1733 1734 |
/* * Acceptable internal fragmentation? */ |
a737b3e2f
|
1735 |
if (left_over * 8 <= (PAGE_SIZE << gfporder)) |
4d268eba1
|
1736 1737 1738 1739 |
break; } return left_over; } |
bf0dea23a
|
1740 1741 1742 1743 1744 1745 1746 1747 |
static struct array_cache __percpu *alloc_kmem_cache_cpus( struct kmem_cache *cachep, int entries, int batchcount) { int cpu; size_t size; struct array_cache __percpu *cpu_cache; size = sizeof(void *) * entries + sizeof(struct array_cache); |
85c9f4b04
|
1748 |
cpu_cache = __alloc_percpu(size, sizeof(void *)); |
bf0dea23a
|
1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 |
if (!cpu_cache) return NULL; for_each_possible_cpu(cpu) { init_arraycache(per_cpu_ptr(cpu_cache, cpu), entries, batchcount); } return cpu_cache; } |
bd721ea73
|
1760 |
static int __ref setup_cpu_cache(struct kmem_cache *cachep, gfp_t gfp) |
f30cf7d13
|
1761 |
{ |
97d066091
|
1762 |
if (slab_state >= FULL) |
83b519e8b
|
1763 |
return enable_cpucache(cachep, gfp); |
2ed3a4ef9
|
1764 |
|
bf0dea23a
|
1765 1766 1767 |
cachep->cpu_cache = alloc_kmem_cache_cpus(cachep, 1, 1); if (!cachep->cpu_cache) return 1; |
97d066091
|
1768 |
if (slab_state == DOWN) { |
bf0dea23a
|
1769 1770 |
/* Creation of first cache (kmem_cache). */ set_up_node(kmem_cache, CACHE_CACHE); |
2f9baa9fc
|
1771 |
} else if (slab_state == PARTIAL) { |
bf0dea23a
|
1772 1773 |
/* For kmem_cache_node */ set_up_node(cachep, SIZE_NODE); |
f30cf7d13
|
1774 |
} else { |
bf0dea23a
|
1775 |
int node; |
f30cf7d13
|
1776 |
|
bf0dea23a
|
1777 1778 1779 1780 1781 |
for_each_online_node(node) { cachep->node[node] = kmalloc_node( sizeof(struct kmem_cache_node), gfp, node); BUG_ON(!cachep->node[node]); kmem_cache_node_init(cachep->node[node]); |
f30cf7d13
|
1782 1783 |
} } |
bf0dea23a
|
1784 |
|
6a67368c3
|
1785 |
cachep->node[numa_mem_id()]->next_reap = |
5f0985bb1
|
1786 1787 |
jiffies + REAPTIMEOUT_NODE + ((unsigned long)cachep) % REAPTIMEOUT_NODE; |
f30cf7d13
|
1788 1789 1790 1791 1792 1793 1794 |
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
|
1795 |
return 0; |
f30cf7d13
|
1796 |
} |
12220dea0
|
1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 |
unsigned long kmem_cache_flags(unsigned long object_size, unsigned long flags, const char *name, void (*ctor)(void *)) { return flags; } struct kmem_cache * __kmem_cache_alias(const char *name, size_t size, size_t align, unsigned long flags, void (*ctor)(void *)) { struct kmem_cache *cachep; cachep = find_mergeable(size, align, flags, name, ctor); if (cachep) { cachep->refcount++; /* * Adjust the object sizes so that we clear * the complete object on kzalloc. */ cachep->object_size = max_t(int, cachep->object_size, size); } return cachep; } |
b03a017be
|
1822 1823 1824 1825 1826 1827 |
static bool set_objfreelist_slab_cache(struct kmem_cache *cachep, size_t size, unsigned long flags) { size_t left; cachep->num = 0; |
5f0d5a3ae
|
1828 |
if (cachep->ctor || flags & SLAB_TYPESAFE_BY_RCU) |
b03a017be
|
1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 |
return false; left = calculate_slab_order(cachep, size, flags | CFLGS_OBJFREELIST_SLAB); if (!cachep->num) return false; if (cachep->num * sizeof(freelist_idx_t) > cachep->object_size) return false; cachep->colour = left / cachep->colour_off; return true; } |
158e319bb
|
1843 1844 1845 1846 1847 1848 1849 1850 |
static bool set_off_slab_cache(struct kmem_cache *cachep, size_t size, unsigned long flags) { size_t left; cachep->num = 0; /* |
3217fd9bd
|
1851 1852 |
* Always use on-slab management when SLAB_NOLEAKTRACE * to avoid recursive calls into kmemleak. |
158e319bb
|
1853 |
*/ |
158e319bb
|
1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 |
if (flags & SLAB_NOLEAKTRACE) return false; /* * Size is large, assume best to place the slab management obj * off-slab (should allow better packing of objs). */ left = calculate_slab_order(cachep, size, flags | CFLGS_OFF_SLAB); if (!cachep->num) return false; /* * 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 (left >= cachep->num * sizeof(freelist_idx_t)) return false; cachep->colour = left / cachep->colour_off; return true; } static bool set_on_slab_cache(struct kmem_cache *cachep, size_t size, unsigned long flags) { size_t left; cachep->num = 0; left = calculate_slab_order(cachep, size, flags); if (!cachep->num) return false; cachep->colour = left / cachep->colour_off; return true; } |
4d268eba1
|
1892 |
/** |
039363f38
|
1893 |
* __kmem_cache_create - Create a cache. |
a755b76ab
|
1894 |
* @cachep: cache management descriptor |
1da177e4c
|
1895 |
* @flags: SLAB flags |
1da177e4c
|
1896 1897 1898 |
* * Returns a ptr to the cache on success, NULL on failure. * Cannot be called within a int, but can be interrupted. |
20c2df83d
|
1899 |
* The @ctor is run when new pages are allocated by the cache. |
1da177e4c
|
1900 |
* |
1da177e4c
|
1901 1902 1903 1904 1905 1906 1907 1908 |
* 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
|
1909 1910 1911 1912 |
* %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. */ |
278b1bb13
|
1913 |
int |
8a13a4cc8
|
1914 |
__kmem_cache_create (struct kmem_cache *cachep, unsigned long flags) |
1da177e4c
|
1915 |
{ |
d4a5fca59
|
1916 |
size_t ralign = BYTES_PER_WORD; |
83b519e8b
|
1917 |
gfp_t gfp; |
278b1bb13
|
1918 |
int err; |
8a13a4cc8
|
1919 |
size_t size = cachep->size; |
1da177e4c
|
1920 |
|
1da177e4c
|
1921 |
#if DEBUG |
1da177e4c
|
1922 1923 1924 1925 1926 1927 1928 |
#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
|
1929 1930 |
if (size < 4096 || fls(size - 1) == fls(size-1 + REDZONE_ALIGN + 2 * sizeof(unsigned long long))) |
b28a02de8
|
1931 |
flags |= SLAB_RED_ZONE | SLAB_STORE_USER; |
5f0d5a3ae
|
1932 |
if (!(flags & SLAB_TYPESAFE_BY_RCU)) |
1da177e4c
|
1933 1934 |
flags |= SLAB_POISON; #endif |
1da177e4c
|
1935 |
#endif |
1da177e4c
|
1936 |
|
a737b3e2f
|
1937 1938 |
/* * Check that size is in terms of words. This is needed to avoid |
1da177e4c
|
1939 1940 1941 |
* unaligned accesses for some archs when redzoning is used, and makes * sure any on-slab bufctl's are also correctly aligned. */ |
e07719502
|
1942 |
size = ALIGN(size, BYTES_PER_WORD); |
1da177e4c
|
1943 |
|
87a927c71
|
1944 1945 1946 1947 |
if (flags & SLAB_RED_ZONE) { ralign = REDZONE_ALIGN; /* If redzoning, ensure that the second redzone is suitably * aligned, by adjusting the object size accordingly. */ |
e07719502
|
1948 |
size = ALIGN(size, REDZONE_ALIGN); |
87a927c71
|
1949 |
} |
ca5f9703d
|
1950 |
|
a44b56d35
|
1951 |
/* 3) caller mandated alignment */ |
8a13a4cc8
|
1952 1953 |
if (ralign < cachep->align) { ralign = cachep->align; |
1da177e4c
|
1954 |
} |
3ff84a7f3
|
1955 1956 |
/* disable debug if necessary */ if (ralign > __alignof__(unsigned long long)) |
a44b56d35
|
1957 |
flags &= ~(SLAB_RED_ZONE | SLAB_STORE_USER); |
a737b3e2f
|
1958 |
/* |
ca5f9703d
|
1959 |
* 4) Store it. |
1da177e4c
|
1960 |
*/ |
8a13a4cc8
|
1961 |
cachep->align = ralign; |
158e319bb
|
1962 1963 1964 1965 |
cachep->colour_off = cache_line_size(); /* Offset must be a multiple of the alignment. */ if (cachep->colour_off < cachep->align) cachep->colour_off = cachep->align; |
1da177e4c
|
1966 |
|
83b519e8b
|
1967 1968 1969 1970 |
if (slab_is_available()) gfp = GFP_KERNEL; else gfp = GFP_NOWAIT; |
1da177e4c
|
1971 |
#if DEBUG |
1da177e4c
|
1972 |
|
ca5f9703d
|
1973 1974 1975 1976 |
/* * Both debugging options require word-alignment which is calculated * into align above. */ |
1da177e4c
|
1977 |
if (flags & SLAB_RED_ZONE) { |
1da177e4c
|
1978 |
/* add space for red zone words */ |
3ff84a7f3
|
1979 1980 |
cachep->obj_offset += sizeof(unsigned long long); size += 2 * sizeof(unsigned long long); |
1da177e4c
|
1981 1982 |
} if (flags & SLAB_STORE_USER) { |
ca5f9703d
|
1983 |
/* user store requires one word storage behind the end of |
87a927c71
|
1984 1985 |
* the real object. But if the second red zone needs to be * aligned to 64 bits, we must allow that much space. |
1da177e4c
|
1986 |
*/ |
87a927c71
|
1987 1988 1989 1990 |
if (flags & SLAB_RED_ZONE) size += REDZONE_ALIGN; else size += BYTES_PER_WORD; |
1da177e4c
|
1991 |
} |
832a15d20
|
1992 |
#endif |
7ed2f9e66
|
1993 |
kasan_cache_create(cachep, &size, &flags); |
832a15d20
|
1994 1995 1996 1997 1998 1999 2000 2001 2002 |
size = ALIGN(size, cachep->align); /* * We should restrict the number of objects in a slab to implement * byte sized index. Refer comment on SLAB_OBJ_MIN_SIZE definition. */ if (FREELIST_BYTE_INDEX && size < SLAB_OBJ_MIN_SIZE) size = ALIGN(SLAB_OBJ_MIN_SIZE, cachep->align); #if DEBUG |
03a2d2a3e
|
2003 2004 2005 2006 2007 2008 2009 |
/* * To activate debug pagealloc, off-slab management is necessary * requirement. In early phase of initialization, small sized slab * doesn't get initialized so it would not be possible. So, we need * to check size >= 256. It guarantees that all necessary small * sized slab is initialized in current slab initialization sequence. */ |
40323278b
|
2010 |
if (debug_pagealloc_enabled() && (flags & SLAB_POISON) && |
f3a3c320d
|
2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 |
size >= 256 && cachep->object_size > cache_line_size()) { if (size < PAGE_SIZE || size % PAGE_SIZE == 0) { size_t tmp_size = ALIGN(size, PAGE_SIZE); if (set_off_slab_cache(cachep, tmp_size, flags)) { flags |= CFLGS_OFF_SLAB; cachep->obj_offset += tmp_size - size; size = tmp_size; goto done; } } |
1da177e4c
|
2022 2023 |
} #endif |
1da177e4c
|
2024 |
|
b03a017be
|
2025 2026 2027 2028 |
if (set_objfreelist_slab_cache(cachep, size, flags)) { flags |= CFLGS_OBJFREELIST_SLAB; goto done; } |
158e319bb
|
2029 |
if (set_off_slab_cache(cachep, size, flags)) { |
1da177e4c
|
2030 |
flags |= CFLGS_OFF_SLAB; |
158e319bb
|
2031 |
goto done; |
832a15d20
|
2032 |
} |
1da177e4c
|
2033 |
|
158e319bb
|
2034 2035 |
if (set_on_slab_cache(cachep, size, flags)) goto done; |
1da177e4c
|
2036 |
|
158e319bb
|
2037 |
return -E2BIG; |
1da177e4c
|
2038 |
|
158e319bb
|
2039 2040 |
done: cachep->freelist_size = cachep->num * sizeof(freelist_idx_t); |
1da177e4c
|
2041 |
cachep->flags = flags; |
a57a49887
|
2042 |
cachep->allocflags = __GFP_COMP; |
a3187e438
|
2043 |
if (flags & SLAB_CACHE_DMA) |
a618e89f1
|
2044 |
cachep->allocflags |= GFP_DMA; |
3b0efdfa1
|
2045 |
cachep->size = size; |
6a2d7a955
|
2046 |
cachep->reciprocal_buffer_size = reciprocal_value(size); |
1da177e4c
|
2047 |
|
40b441379
|
2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 |
#if DEBUG /* * 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 (IS_ENABLED(CONFIG_PAGE_POISONING) && (cachep->flags & SLAB_POISON) && is_debug_pagealloc_cache(cachep)) cachep->flags &= ~(SLAB_RED_ZONE | SLAB_STORE_USER); #endif if (OFF_SLAB(cachep)) { |
158e319bb
|
2061 2062 |
cachep->freelist_cache = kmalloc_slab(cachep->freelist_size, 0u); |
e5ac9c5ae
|
2063 |
} |
1da177e4c
|
2064 |
|
278b1bb13
|
2065 2066 |
err = setup_cpu_cache(cachep, gfp); if (err) { |
52b4b950b
|
2067 |
__kmem_cache_release(cachep); |
278b1bb13
|
2068 |
return err; |
2ed3a4ef9
|
2069 |
} |
1da177e4c
|
2070 |
|
278b1bb13
|
2071 |
return 0; |
1da177e4c
|
2072 |
} |
1da177e4c
|
2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 |
#if DEBUG static void check_irq_off(void) { BUG_ON(!irqs_disabled()); } static void check_irq_on(void) { BUG_ON(irqs_disabled()); } |
18726ca8b
|
2084 2085 2086 2087 |
static void check_mutex_acquired(void) { BUG_ON(!mutex_is_locked(&slab_mutex)); } |
343e0d7a9
|
2088 |
static void check_spinlock_acquired(struct kmem_cache *cachep) |
1da177e4c
|
2089 2090 2091 |
{ #ifdef CONFIG_SMP check_irq_off(); |
18bf85411
|
2092 |
assert_spin_locked(&get_node(cachep, numa_mem_id())->list_lock); |
1da177e4c
|
2093 2094 |
#endif } |
e498be7da
|
2095 |
|
343e0d7a9
|
2096 |
static void check_spinlock_acquired_node(struct kmem_cache *cachep, int node) |
e498be7da
|
2097 2098 2099 |
{ #ifdef CONFIG_SMP check_irq_off(); |
18bf85411
|
2100 |
assert_spin_locked(&get_node(cachep, node)->list_lock); |
e498be7da
|
2101 2102 |
#endif } |
1da177e4c
|
2103 2104 2105 |
#else #define check_irq_off() do { } while(0) #define check_irq_on() do { } while(0) |
18726ca8b
|
2106 |
#define check_mutex_acquired() do { } while(0) |
1da177e4c
|
2107 |
#define check_spinlock_acquired(x) do { } while(0) |
e498be7da
|
2108 |
#define check_spinlock_acquired_node(x, y) do { } while(0) |
1da177e4c
|
2109 |
#endif |
18726ca8b
|
2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 |
static void drain_array_locked(struct kmem_cache *cachep, struct array_cache *ac, int node, bool free_all, struct list_head *list) { int tofree; if (!ac || !ac->avail) return; tofree = free_all ? ac->avail : (ac->limit + 4) / 5; if (tofree > ac->avail) tofree = (ac->avail + 1) / 2; free_block(cachep, ac->entry, tofree, node, list); ac->avail -= tofree; memmove(ac->entry, &(ac->entry[tofree]), sizeof(void *) * ac->avail); } |
aab2207cf
|
2126 |
|
1da177e4c
|
2127 2128 |
static void do_drain(void *arg) { |
a737b3e2f
|
2129 |
struct kmem_cache *cachep = arg; |
1da177e4c
|
2130 |
struct array_cache *ac; |
7d6e6d09d
|
2131 |
int node = numa_mem_id(); |
18bf85411
|
2132 |
struct kmem_cache_node *n; |
97654dfa2
|
2133 |
LIST_HEAD(list); |
1da177e4c
|
2134 2135 |
check_irq_off(); |
9a2dba4b4
|
2136 |
ac = cpu_cache_get(cachep); |
18bf85411
|
2137 2138 |
n = get_node(cachep, node); spin_lock(&n->list_lock); |
97654dfa2
|
2139 |
free_block(cachep, ac->entry, ac->avail, node, &list); |
18bf85411
|
2140 |
spin_unlock(&n->list_lock); |
97654dfa2
|
2141 |
slabs_destroy(cachep, &list); |
1da177e4c
|
2142 2143 |
ac->avail = 0; } |
343e0d7a9
|
2144 |
static void drain_cpu_caches(struct kmem_cache *cachep) |
1da177e4c
|
2145 |
{ |
ce8eb6c42
|
2146 |
struct kmem_cache_node *n; |
e498be7da
|
2147 |
int node; |
18726ca8b
|
2148 |
LIST_HEAD(list); |
e498be7da
|
2149 |
|
15c8b6c1a
|
2150 |
on_each_cpu(do_drain, cachep, 1); |
1da177e4c
|
2151 |
check_irq_on(); |
18bf85411
|
2152 2153 |
for_each_kmem_cache_node(cachep, node, n) if (n->alien) |
ce8eb6c42
|
2154 |
drain_alien_cache(cachep, n->alien); |
a4523a8b3
|
2155 |
|
18726ca8b
|
2156 2157 2158 2159 2160 2161 2162 |
for_each_kmem_cache_node(cachep, node, n) { spin_lock_irq(&n->list_lock); drain_array_locked(cachep, n->shared, node, true, &list); spin_unlock_irq(&n->list_lock); slabs_destroy(cachep, &list); } |
1da177e4c
|
2163 |
} |
ed11d9eb2
|
2164 2165 2166 2167 2168 2169 2170 |
/* * 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, |
ce8eb6c42
|
2171 |
struct kmem_cache_node *n, int tofree) |
1da177e4c
|
2172 |
{ |
ed11d9eb2
|
2173 2174 |
struct list_head *p; int nr_freed; |
8456a648c
|
2175 |
struct page *page; |
1da177e4c
|
2176 |
|
ed11d9eb2
|
2177 |
nr_freed = 0; |
ce8eb6c42
|
2178 |
while (nr_freed < tofree && !list_empty(&n->slabs_free)) { |
1da177e4c
|
2179 |
|
ce8eb6c42
|
2180 2181 2182 2183 |
spin_lock_irq(&n->list_lock); p = n->slabs_free.prev; if (p == &n->slabs_free) { spin_unlock_irq(&n->list_lock); |
ed11d9eb2
|
2184 2185 |
goto out; } |
1da177e4c
|
2186 |
|
8456a648c
|
2187 |
page = list_entry(p, struct page, lru); |
8456a648c
|
2188 |
list_del(&page->lru); |
f728b0a5d
|
2189 |
n->free_slabs--; |
bf00bd345
|
2190 |
n->total_slabs--; |
ed11d9eb2
|
2191 2192 2193 2194 |
/* * Safe to drop the lock. The slab is no longer linked * to the cache. */ |
ce8eb6c42
|
2195 2196 |
n->free_objects -= cache->num; spin_unlock_irq(&n->list_lock); |
8456a648c
|
2197 |
slab_destroy(cache, page); |
ed11d9eb2
|
2198 |
nr_freed++; |
1da177e4c
|
2199 |
} |
ed11d9eb2
|
2200 2201 |
out: return nr_freed; |
1da177e4c
|
2202 |
} |
c9fc58640
|
2203 |
int __kmem_cache_shrink(struct kmem_cache *cachep) |
e498be7da
|
2204 |
{ |
18bf85411
|
2205 2206 |
int ret = 0; int node; |
ce8eb6c42
|
2207 |
struct kmem_cache_node *n; |
e498be7da
|
2208 2209 2210 2211 |
drain_cpu_caches(cachep); check_irq_on(); |
18bf85411
|
2212 |
for_each_kmem_cache_node(cachep, node, n) { |
a5aa63a5f
|
2213 |
drain_freelist(cachep, n, INT_MAX); |
ed11d9eb2
|
2214 |
|
ce8eb6c42
|
2215 2216 |
ret += !list_empty(&n->slabs_full) || !list_empty(&n->slabs_partial); |
e498be7da
|
2217 2218 2219 |
} return (ret ? 1 : 0); } |
c9fc58640
|
2220 2221 2222 2223 2224 2225 |
#ifdef CONFIG_MEMCG void __kmemcg_cache_deactivate(struct kmem_cache *cachep) { __kmem_cache_shrink(cachep); } #endif |
945cf2b61
|
2226 |
int __kmem_cache_shutdown(struct kmem_cache *cachep) |
1da177e4c
|
2227 |
{ |
c9fc58640
|
2228 |
return __kmem_cache_shrink(cachep); |
52b4b950b
|
2229 2230 2231 2232 |
} void __kmem_cache_release(struct kmem_cache *cachep) { |
12c3667fb
|
2233 |
int i; |
ce8eb6c42
|
2234 |
struct kmem_cache_node *n; |
1da177e4c
|
2235 |
|
c7ce4f60a
|
2236 |
cache_random_seq_destroy(cachep); |
bf0dea23a
|
2237 |
free_percpu(cachep->cpu_cache); |
1da177e4c
|
2238 |
|
ce8eb6c42
|
2239 |
/* NUMA: free the node structures */ |
18bf85411
|
2240 2241 2242 2243 2244 |
for_each_kmem_cache_node(cachep, i, n) { kfree(n->shared); free_alien_cache(n->alien); kfree(n); cachep->node[i] = NULL; |
12c3667fb
|
2245 |
} |
1da177e4c
|
2246 |
} |
1da177e4c
|
2247 |
|
e5ac9c5ae
|
2248 2249 |
/* * Get the memory for a slab management obj. |
5f0985bb1
|
2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 |
* * For a slab cache when the slab descriptor is off-slab, the * slab descriptor can't come from the same cache which is being created, * Because if it is the case, that means we defer the creation of * the kmalloc_{dma,}_cache of size sizeof(slab descriptor) to this point. * And we eventually call down to __kmem_cache_create(), which * in turn looks up in the kmalloc_{dma,}_caches for the disired-size one. * This is a "chicken-and-egg" problem. * * So the off-slab slab descriptor shall come from the kmalloc_{dma,}_caches, * which are all initialized during kmem_cache_init(). |
e5ac9c5ae
|
2261 |
*/ |
7e0073552
|
2262 |
static void *alloc_slabmgmt(struct kmem_cache *cachep, |
0c3aa83e0
|
2263 2264 |
struct page *page, int colour_off, gfp_t local_flags, int nodeid) |
1da177e4c
|
2265 |
{ |
7e0073552
|
2266 |
void *freelist; |
0c3aa83e0
|
2267 |
void *addr = page_address(page); |
b28a02de8
|
2268 |
|
2e6b36021
|
2269 2270 |
page->s_mem = addr + colour_off; page->active = 0; |
b03a017be
|
2271 2272 2273 |
if (OBJFREELIST_SLAB(cachep)) freelist = NULL; else if (OFF_SLAB(cachep)) { |
1da177e4c
|
2274 |
/* Slab management obj is off-slab. */ |
8456a648c
|
2275 |
freelist = kmem_cache_alloc_node(cachep->freelist_cache, |
8759ec50a
|
2276 |
local_flags, nodeid); |
8456a648c
|
2277 |
if (!freelist) |
1da177e4c
|
2278 2279 |
return NULL; } else { |
2e6b36021
|
2280 2281 2282 |
/* We will use last bytes at the slab for freelist */ freelist = addr + (PAGE_SIZE << cachep->gfporder) - cachep->freelist_size; |
1da177e4c
|
2283 |
} |
2e6b36021
|
2284 |
|
8456a648c
|
2285 |
return freelist; |
1da177e4c
|
2286 |
} |
7cc68973c
|
2287 |
static inline freelist_idx_t get_free_obj(struct page *page, unsigned int idx) |
1da177e4c
|
2288 |
{ |
a41adfaa2
|
2289 |
return ((freelist_idx_t *)page->freelist)[idx]; |
e5c58dfdc
|
2290 2291 2292 |
} static inline void set_free_obj(struct page *page, |
7cc68973c
|
2293 |
unsigned int idx, freelist_idx_t val) |
e5c58dfdc
|
2294 |
{ |
a41adfaa2
|
2295 |
((freelist_idx_t *)(page->freelist))[idx] = val; |
1da177e4c
|
2296 |
} |
10b2e9e8e
|
2297 |
static void cache_init_objs_debug(struct kmem_cache *cachep, struct page *page) |
1da177e4c
|
2298 |
{ |
10b2e9e8e
|
2299 |
#if DEBUG |
1da177e4c
|
2300 2301 2302 |
int i; for (i = 0; i < cachep->num; i++) { |
8456a648c
|
2303 |
void *objp = index_to_obj(cachep, page, i); |
10b2e9e8e
|
2304 |
|
1da177e4c
|
2305 2306 2307 2308 2309 2310 2311 2312 |
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
|
2313 2314 2315 |
* 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
|
2316 |
*/ |
7ed2f9e66
|
2317 2318 2319 |
if (cachep->ctor && !(cachep->flags & SLAB_POISON)) { kasan_unpoison_object_data(cachep, objp + obj_offset(cachep)); |
51cc50685
|
2320 |
cachep->ctor(objp + obj_offset(cachep)); |
7ed2f9e66
|
2321 2322 2323 |
kasan_poison_object_data( cachep, objp + obj_offset(cachep)); } |
1da177e4c
|
2324 2325 2326 |
if (cachep->flags & SLAB_RED_ZONE) { if (*dbg_redzone2(cachep, objp) != RED_INACTIVE) |
756a025f0
|
2327 |
slab_error(cachep, "constructor overwrote the end of an object"); |
1da177e4c
|
2328 |
if (*dbg_redzone1(cachep, objp) != RED_INACTIVE) |
756a025f0
|
2329 |
slab_error(cachep, "constructor overwrote the start of an object"); |
1da177e4c
|
2330 |
} |
40b441379
|
2331 2332 2333 2334 2335 |
/* need to poison the objs? */ if (cachep->flags & SLAB_POISON) { poison_obj(cachep, objp, POISON_FREE); slab_kernel_map(cachep, objp, 0, 0); } |
10b2e9e8e
|
2336 |
} |
1da177e4c
|
2337 |
#endif |
10b2e9e8e
|
2338 |
} |
c7ce4f60a
|
2339 2340 2341 2342 2343 |
#ifdef CONFIG_SLAB_FREELIST_RANDOM /* Hold information during a freelist initialization */ union freelist_init_state { struct { unsigned int pos; |
7c00fce98
|
2344 |
unsigned int *list; |
c7ce4f60a
|
2345 |
unsigned int count; |
c7ce4f60a
|
2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 |
}; struct rnd_state rnd_state; }; /* * Initialize the state based on the randomization methode available. * return true if the pre-computed list is available, false otherwize. */ static bool freelist_state_initialize(union freelist_init_state *state, struct kmem_cache *cachep, unsigned int count) { bool ret; unsigned int rand; /* Use best entropy available to define a random shift */ |
7c00fce98
|
2362 |
rand = get_random_int(); |
c7ce4f60a
|
2363 2364 2365 2366 2367 2368 2369 2370 |
/* Use a random state if the pre-computed list is not available */ if (!cachep->random_seq) { prandom_seed_state(&state->rnd_state, rand); ret = false; } else { state->list = cachep->random_seq; state->count = count; |
c4e490cf1
|
2371 |
state->pos = rand % count; |
c7ce4f60a
|
2372 2373 2374 2375 2376 2377 2378 2379 |
ret = true; } return ret; } /* Get the next entry on the list and randomize it using a random shift */ static freelist_idx_t next_random_slot(union freelist_init_state *state) { |
c4e490cf1
|
2380 2381 2382 |
if (state->pos >= state->count) state->pos = 0; return state->list[state->pos++]; |
c7ce4f60a
|
2383 |
} |
7c00fce98
|
2384 2385 2386 2387 2388 2389 |
/* Swap two freelist entries */ static void swap_free_obj(struct page *page, unsigned int a, unsigned int b) { swap(((freelist_idx_t *)page->freelist)[a], ((freelist_idx_t *)page->freelist)[b]); } |
c7ce4f60a
|
2390 2391 2392 2393 2394 2395 |
/* * Shuffle the freelist initialization state based on pre-computed lists. * return true if the list was successfully shuffled, false otherwise. */ static bool shuffle_freelist(struct kmem_cache *cachep, struct page *page) { |
7c00fce98
|
2396 |
unsigned int objfreelist = 0, i, rand, count = cachep->num; |
c7ce4f60a
|
2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 |
union freelist_init_state state; bool precomputed; if (count < 2) return false; precomputed = freelist_state_initialize(&state, cachep, count); /* Take a random entry as the objfreelist */ if (OBJFREELIST_SLAB(cachep)) { if (!precomputed) objfreelist = count - 1; else objfreelist = next_random_slot(&state); page->freelist = index_to_obj(cachep, page, objfreelist) + obj_offset(cachep); count--; } /* * On early boot, generate the list dynamically. * Later use a pre-computed list for speed. */ if (!precomputed) { |
7c00fce98
|
2421 2422 2423 2424 2425 2426 2427 2428 2429 |
for (i = 0; i < count; i++) set_free_obj(page, i, i); /* Fisher-Yates shuffle */ for (i = count - 1; i > 0; i--) { rand = prandom_u32_state(&state.rnd_state); rand %= (i + 1); swap_free_obj(page, i, rand); } |
c7ce4f60a
|
2430 2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 |
} else { for (i = 0; i < count; i++) set_free_obj(page, i, next_random_slot(&state)); } if (OBJFREELIST_SLAB(cachep)) set_free_obj(page, cachep->num - 1, objfreelist); return true; } #else static inline bool shuffle_freelist(struct kmem_cache *cachep, struct page *page) { return false; } #endif /* CONFIG_SLAB_FREELIST_RANDOM */ |
10b2e9e8e
|
2447 2448 2449 2450 |
static void cache_init_objs(struct kmem_cache *cachep, struct page *page) { int i; |
7ed2f9e66
|
2451 |
void *objp; |
c7ce4f60a
|
2452 |
bool shuffled; |
10b2e9e8e
|
2453 2454 |
cache_init_objs_debug(cachep, page); |
c7ce4f60a
|
2455 2456 2457 2458 |
/* Try to randomize the freelist if enabled */ shuffled = shuffle_freelist(cachep, page); if (!shuffled && OBJFREELIST_SLAB(cachep)) { |
b03a017be
|
2459 2460 2461 |
page->freelist = index_to_obj(cachep, page, cachep->num - 1) + obj_offset(cachep); } |
10b2e9e8e
|
2462 |
for (i = 0; i < cachep->num; i++) { |
b3cbd9bf7
|
2463 2464 |
objp = index_to_obj(cachep, page, i); kasan_init_slab_obj(cachep, objp); |
10b2e9e8e
|
2465 |
/* constructor could break poison info */ |
7ed2f9e66
|
2466 |
if (DEBUG == 0 && cachep->ctor) { |
7ed2f9e66
|
2467 2468 2469 2470 |
kasan_unpoison_object_data(cachep, objp); cachep->ctor(objp); kasan_poison_object_data(cachep, objp); } |
10b2e9e8e
|
2471 |
|
c7ce4f60a
|
2472 2473 |
if (!shuffled) set_free_obj(page, i, i); |
1da177e4c
|
2474 |
} |
1da177e4c
|
2475 |
} |
260b61dd4
|
2476 |
static void *slab_get_obj(struct kmem_cache *cachep, struct page *page) |
78d382d77
|
2477 |
{ |
b1cb0982b
|
2478 |
void *objp; |
78d382d77
|
2479 |
|
e5c58dfdc
|
2480 |
objp = index_to_obj(cachep, page, get_free_obj(page, page->active)); |
8456a648c
|
2481 |
page->active++; |
78d382d77
|
2482 |
|
d31676dfd
|
2483 2484 2485 2486 |
#if DEBUG if (cachep->flags & SLAB_STORE_USER) set_store_user_dirty(cachep); #endif |
78d382d77
|
2487 2488 |
return objp; } |
260b61dd4
|
2489 2490 |
static void slab_put_obj(struct kmem_cache *cachep, struct page *page, void *objp) |
78d382d77
|
2491 |
{ |
8456a648c
|
2492 |
unsigned int objnr = obj_to_index(cachep, page, objp); |
78d382d77
|
2493 |
#if DEBUG |
16025177e
|
2494 |
unsigned int i; |
b1cb0982b
|
2495 |
|
b1cb0982b
|
2496 |
/* Verify double free bug */ |
8456a648c
|
2497 |
for (i = page->active; i < cachep->num; i++) { |
e5c58dfdc
|
2498 |
if (get_free_obj(page, i) == objnr) { |
1170532bb
|
2499 2500 |
pr_err("slab: double free detected in cache '%s', objp %p ", |
756a025f0
|
2501 |
cachep->name, objp); |
b1cb0982b
|
2502 2503 |
BUG(); } |
78d382d77
|
2504 2505 |
} #endif |
8456a648c
|
2506 |
page->active--; |
b03a017be
|
2507 2508 |
if (!page->freelist) page->freelist = objp + obj_offset(cachep); |
e5c58dfdc
|
2509 |
set_free_obj(page, page->active, objnr); |
78d382d77
|
2510 |
} |
4776874ff
|
2511 2512 2513 |
/* * 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
|
2514 |
* virtual address for kfree, ksize, and slab debugging. |
4776874ff
|
2515 |
*/ |
8456a648c
|
2516 |
static void slab_map_pages(struct kmem_cache *cache, struct page *page, |
7e0073552
|
2517 |
void *freelist) |
1da177e4c
|
2518 |
{ |
a57a49887
|
2519 |
page->slab_cache = cache; |
8456a648c
|
2520 |
page->freelist = freelist; |
1da177e4c
|
2521 2522 2523 2524 2525 2526 |
} /* * 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. */ |
76b342bdc
|
2527 2528 |
static struct page *cache_grow_begin(struct kmem_cache *cachep, gfp_t flags, int nodeid) |
1da177e4c
|
2529 |
{ |
7e0073552
|
2530 |
void *freelist; |
b28a02de8
|
2531 2532 |
size_t offset; gfp_t local_flags; |
511e3a058
|
2533 |
int page_node; |
ce8eb6c42
|
2534 |
struct kmem_cache_node *n; |
511e3a058
|
2535 |
struct page *page; |
1da177e4c
|
2536 |
|
a737b3e2f
|
2537 2538 2539 |
/* * Be lazy and only check for valid flags here, keeping it out of the * critical path in kmem_cache_alloc(). |
1da177e4c
|
2540 |
*/ |
c871ac4e9
|
2541 |
if (unlikely(flags & GFP_SLAB_BUG_MASK)) { |
bacdcb346
|
2542 |
gfp_t invalid_mask = flags & GFP_SLAB_BUG_MASK; |
72baeef0c
|
2543 2544 2545 2546 2547 |
flags &= ~GFP_SLAB_BUG_MASK; pr_warn("Unexpected gfp: %#x (%pGg). Fixing up to gfp: %#x (%pGg). Fix your code! ", invalid_mask, &invalid_mask, flags, &flags); dump_stack(); |
c871ac4e9
|
2548 |
} |
6cb062296
|
2549 |
local_flags = flags & (GFP_CONSTRAINT_MASK|GFP_RECLAIM_MASK); |
1da177e4c
|
2550 |
|
1da177e4c
|
2551 |
check_irq_off(); |
d0164adc8
|
2552 |
if (gfpflags_allow_blocking(local_flags)) |
1da177e4c
|
2553 2554 2555 |
local_irq_enable(); /* |
a737b3e2f
|
2556 2557 |
* Get mem for the objs. Attempt to allocate a physical page from * 'nodeid'. |
e498be7da
|
2558 |
*/ |
511e3a058
|
2559 |
page = kmem_getpages(cachep, local_flags, nodeid); |
0c3aa83e0
|
2560 |
if (!page) |
1da177e4c
|
2561 |
goto failed; |
511e3a058
|
2562 2563 |
page_node = page_to_nid(page); n = get_node(cachep, page_node); |
03d1d43a1
|
2564 2565 2566 2567 2568 2569 2570 2571 2572 2573 2574 |
/* Get colour for the slab, and cal the next value. */ n->colour_next++; if (n->colour_next >= cachep->colour) n->colour_next = 0; offset = n->colour_next; if (offset >= cachep->colour) offset = 0; offset *= cachep->colour_off; |
1da177e4c
|
2575 |
/* Get slab management. */ |
8456a648c
|
2576 |
freelist = alloc_slabmgmt(cachep, page, offset, |
511e3a058
|
2577 |
local_flags & ~GFP_CONSTRAINT_MASK, page_node); |
b03a017be
|
2578 |
if (OFF_SLAB(cachep) && !freelist) |
1da177e4c
|
2579 |
goto opps1; |
8456a648c
|
2580 |
slab_map_pages(cachep, page, freelist); |
1da177e4c
|
2581 |
|
7ed2f9e66
|
2582 |
kasan_poison_slab(page); |
8456a648c
|
2583 |
cache_init_objs(cachep, page); |
1da177e4c
|
2584 |
|
d0164adc8
|
2585 |
if (gfpflags_allow_blocking(local_flags)) |
1da177e4c
|
2586 |
local_irq_disable(); |
1da177e4c
|
2587 |
|
76b342bdc
|
2588 |
return page; |
a737b3e2f
|
2589 |
opps1: |
0c3aa83e0
|
2590 |
kmem_freepages(cachep, page); |
a737b3e2f
|
2591 |
failed: |
d0164adc8
|
2592 |
if (gfpflags_allow_blocking(local_flags)) |
1da177e4c
|
2593 |
local_irq_disable(); |
76b342bdc
|
2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 |
return NULL; } static void cache_grow_end(struct kmem_cache *cachep, struct page *page) { struct kmem_cache_node *n; void *list = NULL; check_irq_off(); if (!page) return; INIT_LIST_HEAD(&page->lru); n = get_node(cachep, page_to_nid(page)); spin_lock(&n->list_lock); |
bf00bd345
|
2611 |
n->total_slabs++; |
f728b0a5d
|
2612 |
if (!page->active) { |
76b342bdc
|
2613 |
list_add_tail(&page->lru, &(n->slabs_free)); |
f728b0a5d
|
2614 |
n->free_slabs++; |
bf00bd345
|
2615 |
} else |
76b342bdc
|
2616 |
fixup_slab_list(cachep, n, page, &list); |
07a63c41f
|
2617 |
|
76b342bdc
|
2618 2619 2620 2621 2622 |
STATS_INC_GROWN(cachep); n->free_objects += cachep->num - page->active; spin_unlock(&n->list_lock); fixup_objfreelist_debug(cachep, &list); |
1da177e4c
|
2623 2624 2625 2626 2627 2628 2629 2630 |
} #if DEBUG /* * Perform extra freeing checks: * - detect bad pointers. * - POISON/RED_ZONE checking |
1da177e4c
|
2631 2632 2633 |
*/ static void kfree_debugcheck(const void *objp) { |
1da177e4c
|
2634 |
if (!virt_addr_valid(objp)) { |
1170532bb
|
2635 2636 |
pr_err("kfree_debugcheck: out of range ptr %lxh ", |
b28a02de8
|
2637 2638 |
(unsigned long)objp); BUG(); |
1da177e4c
|
2639 |
} |
1da177e4c
|
2640 |
} |
58ce1fd58
|
2641 2642 |
static inline void verify_redzone_free(struct kmem_cache *cache, void *obj) { |
b46b8f19c
|
2643 |
unsigned long long redzone1, redzone2; |
58ce1fd58
|
2644 2645 2646 2647 2648 2649 2650 2651 2652 2653 2654 2655 2656 2657 |
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"); |
1170532bb
|
2658 2659 2660 |
pr_err("%p: redzone 1:0x%llx, redzone 2:0x%llx ", obj, redzone1, redzone2); |
58ce1fd58
|
2661 |
} |
343e0d7a9
|
2662 |
static void *cache_free_debugcheck(struct kmem_cache *cachep, void *objp, |
7c0cb9c64
|
2663 |
unsigned long caller) |
1da177e4c
|
2664 |
{ |
1da177e4c
|
2665 |
unsigned int objnr; |
8456a648c
|
2666 |
struct page *page; |
1da177e4c
|
2667 |
|
80cbd911c
|
2668 |
BUG_ON(virt_to_cache(objp) != cachep); |
3dafccf22
|
2669 |
objp -= obj_offset(cachep); |
1da177e4c
|
2670 |
kfree_debugcheck(objp); |
b49af68ff
|
2671 |
page = virt_to_head_page(objp); |
1da177e4c
|
2672 |
|
1da177e4c
|
2673 |
if (cachep->flags & SLAB_RED_ZONE) { |
58ce1fd58
|
2674 |
verify_redzone_free(cachep, objp); |
1da177e4c
|
2675 2676 2677 |
*dbg_redzone1(cachep, objp) = RED_INACTIVE; *dbg_redzone2(cachep, objp) = RED_INACTIVE; } |
d31676dfd
|
2678 2679 |
if (cachep->flags & SLAB_STORE_USER) { set_store_user_dirty(cachep); |
7c0cb9c64
|
2680 |
*dbg_userword(cachep, objp) = (void *)caller; |
d31676dfd
|
2681 |
} |
1da177e4c
|
2682 |
|
8456a648c
|
2683 |
objnr = obj_to_index(cachep, page, objp); |
1da177e4c
|
2684 2685 |
BUG_ON(objnr >= cachep->num); |
8456a648c
|
2686 |
BUG_ON(objp != index_to_obj(cachep, page, objnr)); |
1da177e4c
|
2687 |
|
1da177e4c
|
2688 |
if (cachep->flags & SLAB_POISON) { |
1da177e4c
|
2689 |
poison_obj(cachep, objp, POISON_FREE); |
40b441379
|
2690 |
slab_kernel_map(cachep, objp, 0, caller); |
1da177e4c
|
2691 2692 2693 |
} return objp; } |
1da177e4c
|
2694 2695 2696 |
#else #define kfree_debugcheck(x) do { } while(0) #define cache_free_debugcheck(x,objp,z) (objp) |
1da177e4c
|
2697 |
#endif |
b03a017be
|
2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 |
static inline void fixup_objfreelist_debug(struct kmem_cache *cachep, void **list) { #if DEBUG void *next = *list; void *objp; while (next) { objp = next - obj_offset(cachep); next = *(void **)next; poison_obj(cachep, objp, POISON_FREE); } #endif } |
d8410234d
|
2712 |
static inline void fixup_slab_list(struct kmem_cache *cachep, |
b03a017be
|
2713 2714 |
struct kmem_cache_node *n, struct page *page, void **list) |
d8410234d
|
2715 2716 2717 |
{ /* move slabp to correct slabp list: */ list_del(&page->lru); |
b03a017be
|
2718 |
if (page->active == cachep->num) { |
d8410234d
|
2719 |
list_add(&page->lru, &n->slabs_full); |
b03a017be
|
2720 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 |
if (OBJFREELIST_SLAB(cachep)) { #if DEBUG /* Poisoning will be done without holding the lock */ if (cachep->flags & SLAB_POISON) { void **objp = page->freelist; *objp = *list; *list = objp; } #endif page->freelist = NULL; } } else |
d8410234d
|
2733 2734 |
list_add(&page->lru, &n->slabs_partial); } |
f68f8dddb
|
2735 2736 |
/* Try to find non-pfmemalloc slab if needed */ static noinline struct page *get_valid_first_slab(struct kmem_cache_node *n, |
bf00bd345
|
2737 |
struct page *page, bool pfmemalloc) |
f68f8dddb
|
2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 2752 2753 2754 2755 |
{ if (!page) return NULL; if (pfmemalloc) return page; if (!PageSlabPfmemalloc(page)) return page; /* No need to keep pfmemalloc slab if we have enough free objects */ if (n->free_objects > n->free_limit) { ClearPageSlabPfmemalloc(page); return page; } /* Move pfmemalloc slab to the end of list to speed up next search */ list_del(&page->lru); |
bf00bd345
|
2756 |
if (!page->active) { |
f68f8dddb
|
2757 |
list_add_tail(&page->lru, &n->slabs_free); |
bf00bd345
|
2758 |
n->free_slabs++; |
f728b0a5d
|
2759 |
} else |
f68f8dddb
|
2760 2761 2762 2763 2764 2765 |
list_add_tail(&page->lru, &n->slabs_partial); list_for_each_entry(page, &n->slabs_partial, lru) { if (!PageSlabPfmemalloc(page)) return page; } |
f728b0a5d
|
2766 |
n->free_touched = 1; |
f68f8dddb
|
2767 |
list_for_each_entry(page, &n->slabs_free, lru) { |
f728b0a5d
|
2768 |
if (!PageSlabPfmemalloc(page)) { |
bf00bd345
|
2769 |
n->free_slabs--; |
f68f8dddb
|
2770 |
return page; |
f728b0a5d
|
2771 |
} |
f68f8dddb
|
2772 2773 2774 2775 2776 2777 |
} return NULL; } static struct page *get_first_slab(struct kmem_cache_node *n, bool pfmemalloc) |
7aa0d2278
|
2778 2779 |
{ struct page *page; |
f728b0a5d
|
2780 |
assert_spin_locked(&n->list_lock); |
bf00bd345
|
2781 |
page = list_first_entry_or_null(&n->slabs_partial, struct page, lru); |
7aa0d2278
|
2782 2783 |
if (!page) { n->free_touched = 1; |
bf00bd345
|
2784 2785 |
page = list_first_entry_or_null(&n->slabs_free, struct page, lru); |
f728b0a5d
|
2786 |
if (page) |
bf00bd345
|
2787 |
n->free_slabs--; |
7aa0d2278
|
2788 |
} |
f68f8dddb
|
2789 |
if (sk_memalloc_socks()) |
bf00bd345
|
2790 |
page = get_valid_first_slab(n, page, pfmemalloc); |
f68f8dddb
|
2791 |
|
7aa0d2278
|
2792 2793 |
return page; } |
f68f8dddb
|
2794 2795 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812 2813 2814 2815 2816 2817 2818 2819 2820 |
static noinline void *cache_alloc_pfmemalloc(struct kmem_cache *cachep, struct kmem_cache_node *n, gfp_t flags) { struct page *page; void *obj; void *list = NULL; if (!gfp_pfmemalloc_allowed(flags)) return NULL; spin_lock(&n->list_lock); page = get_first_slab(n, true); if (!page) { spin_unlock(&n->list_lock); return NULL; } obj = slab_get_obj(cachep, page); n->free_objects--; fixup_slab_list(cachep, n, page, &list); spin_unlock(&n->list_lock); fixup_objfreelist_debug(cachep, &list); return obj; } |
213b46958
|
2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842 2843 |
/* * Slab list should be fixed up by fixup_slab_list() for existing slab * or cache_grow_end() for new slab */ static __always_inline int alloc_block(struct kmem_cache *cachep, struct array_cache *ac, struct page *page, int batchcount) { /* * There must be at least one object available for * allocation. */ BUG_ON(page->active >= cachep->num); while (page->active < cachep->num && batchcount--) { STATS_INC_ALLOCED(cachep); STATS_INC_ACTIVE(cachep); STATS_SET_HIGH(cachep); ac->entry[ac->avail++] = slab_get_obj(cachep, page); } return batchcount; } |
f68f8dddb
|
2844 |
static void *cache_alloc_refill(struct kmem_cache *cachep, gfp_t flags) |
1da177e4c
|
2845 2846 |
{ int batchcount; |
ce8eb6c42
|
2847 |
struct kmem_cache_node *n; |
801faf0db
|
2848 |
struct array_cache *ac, *shared; |
1ca4cb241
|
2849 |
int node; |
b03a017be
|
2850 |
void *list = NULL; |
76b342bdc
|
2851 |
struct page *page; |
1ca4cb241
|
2852 |
|
1da177e4c
|
2853 |
check_irq_off(); |
7d6e6d09d
|
2854 |
node = numa_mem_id(); |
f68f8dddb
|
2855 |
|
9a2dba4b4
|
2856 |
ac = cpu_cache_get(cachep); |
1da177e4c
|
2857 2858 |
batchcount = ac->batchcount; if (!ac->touched && batchcount > BATCHREFILL_LIMIT) { |
a737b3e2f
|
2859 2860 2861 2862 |
/* * If there was little recent activity on this cache, then * perform only a partial refill. Otherwise we could generate * refill bouncing. |
1da177e4c
|
2863 2864 2865 |
*/ batchcount = BATCHREFILL_LIMIT; } |
18bf85411
|
2866 |
n = get_node(cachep, node); |
e498be7da
|
2867 |
|
ce8eb6c42
|
2868 |
BUG_ON(ac->avail > 0 || !n); |
801faf0db
|
2869 2870 2871 |
shared = READ_ONCE(n->shared); if (!n->free_objects && (!shared || !shared->avail)) goto direct_grow; |
ce8eb6c42
|
2872 |
spin_lock(&n->list_lock); |
801faf0db
|
2873 |
shared = READ_ONCE(n->shared); |
1da177e4c
|
2874 |
|
3ded175a4
|
2875 |
/* See if we can refill from the shared array */ |
801faf0db
|
2876 2877 |
if (shared && transfer_objects(ac, shared, batchcount)) { shared->touched = 1; |
3ded175a4
|
2878 |
goto alloc_done; |
44b57f1cc
|
2879 |
} |
3ded175a4
|
2880 |
|
1da177e4c
|
2881 |
while (batchcount > 0) { |
1da177e4c
|
2882 |
/* Get slab alloc is to come from. */ |
f68f8dddb
|
2883 |
page = get_first_slab(n, false); |
7aa0d2278
|
2884 2885 |
if (!page) goto must_grow; |
1da177e4c
|
2886 |
|
1da177e4c
|
2887 |
check_spinlock_acquired(cachep); |
714b8171a
|
2888 |
|
213b46958
|
2889 |
batchcount = alloc_block(cachep, ac, page, batchcount); |
b03a017be
|
2890 |
fixup_slab_list(cachep, n, page, &list); |
1da177e4c
|
2891 |
} |
a737b3e2f
|
2892 |
must_grow: |
ce8eb6c42
|
2893 |
n->free_objects -= ac->avail; |
a737b3e2f
|
2894 |
alloc_done: |
ce8eb6c42
|
2895 |
spin_unlock(&n->list_lock); |
b03a017be
|
2896 |
fixup_objfreelist_debug(cachep, &list); |
1da177e4c
|
2897 |
|
801faf0db
|
2898 |
direct_grow: |
1da177e4c
|
2899 |
if (unlikely(!ac->avail)) { |
f68f8dddb
|
2900 2901 2902 2903 2904 2905 2906 |
/* Check if we can use obj in pfmemalloc slab */ if (sk_memalloc_socks()) { void *obj = cache_alloc_pfmemalloc(cachep, n, flags); if (obj) return obj; } |
76b342bdc
|
2907 |
page = cache_grow_begin(cachep, gfp_exact_node(flags), node); |
e498be7da
|
2908 |
|
76b342bdc
|
2909 2910 2911 2912 |
/* * cache_grow_begin() can reenable interrupts, * then ac could change. */ |
9a2dba4b4
|
2913 |
ac = cpu_cache_get(cachep); |
213b46958
|
2914 2915 2916 |
if (!ac->avail && page) alloc_block(cachep, ac, page, batchcount); cache_grow_end(cachep, page); |
072bb0aa5
|
2917 |
|
213b46958
|
2918 |
if (!ac->avail) |
1da177e4c
|
2919 |
return NULL; |
1da177e4c
|
2920 2921 |
} ac->touched = 1; |
072bb0aa5
|
2922 |
|
f68f8dddb
|
2923 |
return ac->entry[--ac->avail]; |
1da177e4c
|
2924 |
} |
a737b3e2f
|
2925 2926 |
static inline void cache_alloc_debugcheck_before(struct kmem_cache *cachep, gfp_t flags) |
1da177e4c
|
2927 |
{ |
d0164adc8
|
2928 |
might_sleep_if(gfpflags_allow_blocking(flags)); |
1da177e4c
|
2929 2930 2931 |
} #if DEBUG |
a737b3e2f
|
2932 |
static void *cache_alloc_debugcheck_after(struct kmem_cache *cachep, |
7c0cb9c64
|
2933 |
gfp_t flags, void *objp, unsigned long caller) |
1da177e4c
|
2934 |
{ |
b28a02de8
|
2935 |
if (!objp) |
1da177e4c
|
2936 |
return objp; |
b28a02de8
|
2937 |
if (cachep->flags & SLAB_POISON) { |
1da177e4c
|
2938 |
check_poison_obj(cachep, objp); |
40b441379
|
2939 |
slab_kernel_map(cachep, objp, 1, 0); |
1da177e4c
|
2940 2941 2942 |
poison_obj(cachep, objp, POISON_INUSE); } if (cachep->flags & SLAB_STORE_USER) |
7c0cb9c64
|
2943 |
*dbg_userword(cachep, objp) = (void *)caller; |
1da177e4c
|
2944 2945 |
if (cachep->flags & SLAB_RED_ZONE) { |
a737b3e2f
|
2946 2947 |
if (*dbg_redzone1(cachep, objp) != RED_INACTIVE || *dbg_redzone2(cachep, objp) != RED_INACTIVE) { |
756a025f0
|
2948 |
slab_error(cachep, "double free, or memory outside object was overwritten"); |
1170532bb
|
2949 2950 2951 2952 |
pr_err("%p: redzone 1:0x%llx, redzone 2:0x%llx ", objp, *dbg_redzone1(cachep, objp), *dbg_redzone2(cachep, objp)); |
1da177e4c
|
2953 2954 2955 2956 |
} *dbg_redzone1(cachep, objp) = RED_ACTIVE; *dbg_redzone2(cachep, objp) = RED_ACTIVE; } |
037873014
|
2957 |
|
3dafccf22
|
2958 |
objp += obj_offset(cachep); |
4f1049345
|
2959 |
if (cachep->ctor && cachep->flags & SLAB_POISON) |
51cc50685
|
2960 |
cachep->ctor(objp); |
7ea466f22
|
2961 2962 |
if (ARCH_SLAB_MINALIGN && ((unsigned long)objp & (ARCH_SLAB_MINALIGN-1))) { |
1170532bb
|
2963 2964 |
pr_err("0x%p: not aligned to ARCH_SLAB_MINALIGN=%d ", |
c225150b8
|
2965 |
objp, (int)ARCH_SLAB_MINALIGN); |
a44b56d35
|
2966 |
} |
1da177e4c
|
2967 2968 2969 2970 2971 |
return objp; } #else #define cache_alloc_debugcheck_after(a,b,objp,d) (objp) #endif |
343e0d7a9
|
2972 |
static inline void *____cache_alloc(struct kmem_cache *cachep, gfp_t flags) |
1da177e4c
|
2973 |
{ |
b28a02de8
|
2974 |
void *objp; |
1da177e4c
|
2975 |
struct array_cache *ac; |
5c3823008
|
2976 |
check_irq_off(); |
8a8b6502f
|
2977 |
|
9a2dba4b4
|
2978 |
ac = cpu_cache_get(cachep); |
1da177e4c
|
2979 |
if (likely(ac->avail)) { |
1da177e4c
|
2980 |
ac->touched = 1; |
f68f8dddb
|
2981 |
objp = ac->entry[--ac->avail]; |
072bb0aa5
|
2982 |
|
f68f8dddb
|
2983 2984 |
STATS_INC_ALLOCHIT(cachep); goto out; |
1da177e4c
|
2985 |
} |
072bb0aa5
|
2986 2987 |
STATS_INC_ALLOCMISS(cachep); |
f68f8dddb
|
2988 |
objp = cache_alloc_refill(cachep, flags); |
072bb0aa5
|
2989 2990 2991 2992 2993 2994 2995 |
/* * the 'ac' may be updated by cache_alloc_refill(), * and kmemleak_erase() requires its correct value. */ ac = cpu_cache_get(cachep); out: |
d5cff6352
|
2996 2997 2998 2999 3000 |
/* * 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
|
3001 3002 |
if (objp) kmemleak_erase(&ac->entry[ac->avail]); |
5c3823008
|
3003 3004 |
return objp; } |
e498be7da
|
3005 3006 |
#ifdef CONFIG_NUMA /* |
2ad654bc5
|
3007 |
* Try allocating on another node if PFA_SPREAD_SLAB is a mempolicy is set. |
c61afb181
|
3008 3009 3010 3011 3012 3013 3014 |
* * 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
|
3015 |
if (in_interrupt() || (flags & __GFP_THISNODE)) |
c61afb181
|
3016 |
return NULL; |
7d6e6d09d
|
3017 |
nid_alloc = nid_here = numa_mem_id(); |
c61afb181
|
3018 |
if (cpuset_do_slab_mem_spread() && (cachep->flags & SLAB_MEM_SPREAD)) |
6adef3ebe
|
3019 |
nid_alloc = cpuset_slab_spread_node(); |
c61afb181
|
3020 |
else if (current->mempolicy) |
2a389610a
|
3021 |
nid_alloc = mempolicy_slab_node(); |
c61afb181
|
3022 |
if (nid_alloc != nid_here) |
8b98c1699
|
3023 |
return ____cache_alloc_node(cachep, flags, nid_alloc); |
c61afb181
|
3024 3025 3026 3027 |
return NULL; } /* |
765c4507a
|
3028 |
* Fallback function if there was no memory available and no objects on a |
3c517a613
|
3029 |
* certain node and fall back is permitted. First we scan all the |
6a67368c3
|
3030 |
* available node for available objects. If that fails then we |
3c517a613
|
3031 3032 3033 |
* 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
|
3034 |
*/ |
8c8cc2c10
|
3035 |
static void *fallback_alloc(struct kmem_cache *cache, gfp_t flags) |
765c4507a
|
3036 |
{ |
8c8cc2c10
|
3037 |
struct zonelist *zonelist; |
dd1a239f6
|
3038 |
struct zoneref *z; |
54a6eb5c4
|
3039 3040 |
struct zone *zone; enum zone_type high_zoneidx = gfp_zone(flags); |
765c4507a
|
3041 |
void *obj = NULL; |
76b342bdc
|
3042 |
struct page *page; |
3c517a613
|
3043 |
int nid; |
cc9a6c877
|
3044 |
unsigned int cpuset_mems_cookie; |
8c8cc2c10
|
3045 3046 3047 |
if (flags & __GFP_THISNODE) return NULL; |
cc9a6c877
|
3048 |
retry_cpuset: |
d26914d11
|
3049 |
cpuset_mems_cookie = read_mems_allowed_begin(); |
2a389610a
|
3050 |
zonelist = node_zonelist(mempolicy_slab_node(), flags); |
cc9a6c877
|
3051 |
|
3c517a613
|
3052 3053 3054 3055 3056 |
retry: /* * Look through allowed nodes for objects available * from existing per node queues. */ |
54a6eb5c4
|
3057 3058 |
for_each_zone_zonelist(zone, z, zonelist, high_zoneidx) { nid = zone_to_nid(zone); |
aedb0eb10
|
3059 |
|
061d7074e
|
3060 |
if (cpuset_zone_allowed(zone, flags) && |
18bf85411
|
3061 3062 |
get_node(cache, nid) && get_node(cache, nid)->free_objects) { |
3c517a613
|
3063 |
obj = ____cache_alloc_node(cache, |
4167e9b2c
|
3064 |
gfp_exact_node(flags), nid); |
481c5346d
|
3065 3066 3067 |
if (obj) break; } |
3c517a613
|
3068 |
} |
cfce66047
|
3069 |
if (!obj) { |
3c517a613
|
3070 3071 3072 3073 3074 3075 |
/* * 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. */ |
76b342bdc
|
3076 3077 3078 3079 |
page = cache_grow_begin(cache, flags, numa_mem_id()); cache_grow_end(cache, page); if (page) { nid = page_to_nid(page); |
511e3a058
|
3080 3081 |
obj = ____cache_alloc_node(cache, gfp_exact_node(flags), nid); |
0c3aa83e0
|
3082 |
|
3c517a613
|
3083 |
/* |
511e3a058
|
3084 3085 |
* Another processor may allocate the objects in * the slab since we are not holding any locks. |
3c517a613
|
3086 |
*/ |
511e3a058
|
3087 3088 |
if (!obj) goto retry; |
3c517a613
|
3089 |
} |
aedb0eb10
|
3090 |
} |
cc9a6c877
|
3091 |
|
d26914d11
|
3092 |
if (unlikely(!obj && read_mems_allowed_retry(cpuset_mems_cookie))) |
cc9a6c877
|
3093 |
goto retry_cpuset; |
765c4507a
|
3094 3095 3096 3097 |
return obj; } /* |
e498be7da
|
3098 |
* A interface to enable slab creation on nodeid |
1da177e4c
|
3099 |
*/ |
8b98c1699
|
3100 |
static void *____cache_alloc_node(struct kmem_cache *cachep, gfp_t flags, |
a737b3e2f
|
3101 |
int nodeid) |
e498be7da
|
3102 |
{ |
8456a648c
|
3103 |
struct page *page; |
ce8eb6c42
|
3104 |
struct kmem_cache_node *n; |
213b46958
|
3105 |
void *obj = NULL; |
b03a017be
|
3106 |
void *list = NULL; |
b28a02de8
|
3107 |
|
7c3fbbdd0
|
3108 |
VM_BUG_ON(nodeid < 0 || nodeid >= MAX_NUMNODES); |
18bf85411
|
3109 |
n = get_node(cachep, nodeid); |
ce8eb6c42
|
3110 |
BUG_ON(!n); |
b28a02de8
|
3111 |
|
ca3b9b917
|
3112 |
check_irq_off(); |
ce8eb6c42
|
3113 |
spin_lock(&n->list_lock); |
f68f8dddb
|
3114 |
page = get_first_slab(n, false); |
7aa0d2278
|
3115 3116 |
if (!page) goto must_grow; |
b28a02de8
|
3117 |
|
b28a02de8
|
3118 |
check_spinlock_acquired_node(cachep, nodeid); |
b28a02de8
|
3119 3120 3121 3122 |
STATS_INC_NODEALLOCS(cachep); STATS_INC_ACTIVE(cachep); STATS_SET_HIGH(cachep); |
8456a648c
|
3123 |
BUG_ON(page->active == cachep->num); |
b28a02de8
|
3124 |
|
260b61dd4
|
3125 |
obj = slab_get_obj(cachep, page); |
ce8eb6c42
|
3126 |
n->free_objects--; |
b28a02de8
|
3127 |
|
b03a017be
|
3128 |
fixup_slab_list(cachep, n, page, &list); |
e498be7da
|
3129 |
|
ce8eb6c42
|
3130 |
spin_unlock(&n->list_lock); |
b03a017be
|
3131 |
fixup_objfreelist_debug(cachep, &list); |
213b46958
|
3132 |
return obj; |
e498be7da
|
3133 |
|
a737b3e2f
|
3134 |
must_grow: |
ce8eb6c42
|
3135 |
spin_unlock(&n->list_lock); |
76b342bdc
|
3136 |
page = cache_grow_begin(cachep, gfp_exact_node(flags), nodeid); |
213b46958
|
3137 3138 3139 3140 |
if (page) { /* This slab isn't counted yet so don't update free_objects */ obj = slab_get_obj(cachep, page); } |
76b342bdc
|
3141 |
cache_grow_end(cachep, page); |
1da177e4c
|
3142 |
|
213b46958
|
3143 |
return obj ? obj : fallback_alloc(cachep, flags); |
e498be7da
|
3144 |
} |
8c8cc2c10
|
3145 |
|
8c8cc2c10
|
3146 |
static __always_inline void * |
48356303f
|
3147 |
slab_alloc_node(struct kmem_cache *cachep, gfp_t flags, int nodeid, |
7c0cb9c64
|
3148 |
unsigned long caller) |
8c8cc2c10
|
3149 3150 3151 |
{ unsigned long save_flags; void *ptr; |
7d6e6d09d
|
3152 |
int slab_node = numa_mem_id(); |
8c8cc2c10
|
3153 |
|
dcce284a2
|
3154 |
flags &= gfp_allowed_mask; |
011eceaf0
|
3155 3156 |
cachep = slab_pre_alloc_hook(cachep, flags); if (unlikely(!cachep)) |
824ebef12
|
3157 |
return NULL; |
8c8cc2c10
|
3158 3159 |
cache_alloc_debugcheck_before(cachep, flags); local_irq_save(save_flags); |
eacbbae38
|
3160 |
if (nodeid == NUMA_NO_NODE) |
7d6e6d09d
|
3161 |
nodeid = slab_node; |
8c8cc2c10
|
3162 |
|
18bf85411
|
3163 |
if (unlikely(!get_node(cachep, nodeid))) { |
8c8cc2c10
|
3164 3165 3166 3167 |
/* Node not bootstrapped yet */ ptr = fallback_alloc(cachep, flags); goto out; } |
7d6e6d09d
|
3168 |
if (nodeid == slab_node) { |
8c8cc2c10
|
3169 3170 3171 3172 3173 3174 3175 3176 3177 3178 3179 3180 3181 3182 3183 |
/* * 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); |
d5e3ed66d
|
3184 3185 |
if (unlikely(flags & __GFP_ZERO) && ptr) memset(ptr, 0, cachep->object_size); |
d07dbea46
|
3186 |
|
d5e3ed66d
|
3187 |
slab_post_alloc_hook(cachep, flags, 1, &ptr); |
8c8cc2c10
|
3188 3189 3190 3191 3192 3193 3194 |
return ptr; } static __always_inline void * __do_cache_alloc(struct kmem_cache *cache, gfp_t flags) { void *objp; |
2ad654bc5
|
3195 |
if (current->mempolicy || cpuset_do_slab_mem_spread()) { |
8c8cc2c10
|
3196 3197 3198 3199 3200 3201 3202 3203 3204 3205 |
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
|
3206 3207 |
if (!objp) objp = ____cache_alloc_node(cache, flags, numa_mem_id()); |
8c8cc2c10
|
3208 3209 3210 3211 3212 3213 3214 3215 3216 3217 3218 3219 3220 3221 3222 |
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 * |
48356303f
|
3223 |
slab_alloc(struct kmem_cache *cachep, gfp_t flags, unsigned long caller) |
8c8cc2c10
|
3224 3225 3226 |
{ unsigned long save_flags; void *objp; |
dcce284a2
|
3227 |
flags &= gfp_allowed_mask; |
011eceaf0
|
3228 3229 |
cachep = slab_pre_alloc_hook(cachep, flags); if (unlikely(!cachep)) |
824ebef12
|
3230 |
return NULL; |
8c8cc2c10
|
3231 3232 3233 3234 3235 3236 |
cache_alloc_debugcheck_before(cachep, flags); local_irq_save(save_flags); objp = __do_cache_alloc(cachep, flags); local_irq_restore(save_flags); objp = cache_alloc_debugcheck_after(cachep, flags, objp, caller); prefetchw(objp); |
d5e3ed66d
|
3237 3238 |
if (unlikely(flags & __GFP_ZERO) && objp) memset(objp, 0, cachep->object_size); |
d07dbea46
|
3239 |
|
d5e3ed66d
|
3240 |
slab_post_alloc_hook(cachep, flags, 1, &objp); |
8c8cc2c10
|
3241 3242 |
return objp; } |
e498be7da
|
3243 3244 |
/* |
5f0985bb1
|
3245 |
* Caller needs to acquire correct kmem_cache_node's list_lock |
97654dfa2
|
3246 |
* @list: List of detached free slabs should be freed by caller |
e498be7da
|
3247 |
*/ |
97654dfa2
|
3248 3249 |
static void free_block(struct kmem_cache *cachep, void **objpp, int nr_objects, int node, struct list_head *list) |
1da177e4c
|
3250 3251 |
{ int i; |
25c063fbd
|
3252 |
struct kmem_cache_node *n = get_node(cachep, node); |
6052b7880
|
3253 3254 3255 |
struct page *page; n->free_objects += nr_objects; |
1da177e4c
|
3256 3257 |
for (i = 0; i < nr_objects; i++) { |
072bb0aa5
|
3258 |
void *objp; |
8456a648c
|
3259 |
struct page *page; |
1da177e4c
|
3260 |
|
072bb0aa5
|
3261 |
objp = objpp[i]; |
8456a648c
|
3262 |
page = virt_to_head_page(objp); |
8456a648c
|
3263 |
list_del(&page->lru); |
ff69416e6
|
3264 |
check_spinlock_acquired_node(cachep, node); |
260b61dd4
|
3265 |
slab_put_obj(cachep, page, objp); |
1da177e4c
|
3266 |
STATS_DEC_ACTIVE(cachep); |
1da177e4c
|
3267 3268 |
/* fixup slab chains */ |
f728b0a5d
|
3269 |
if (page->active == 0) { |
6052b7880
|
3270 |
list_add(&page->lru, &n->slabs_free); |
f728b0a5d
|
3271 |
n->free_slabs++; |
f728b0a5d
|
3272 |
} else { |
1da177e4c
|
3273 3274 3275 3276 |
/* Unconditionally move a slab to the end of the * partial list on free - maximum time for the * other objects to be freed, too. */ |
8456a648c
|
3277 |
list_add_tail(&page->lru, &n->slabs_partial); |
1da177e4c
|
3278 3279 |
} } |
6052b7880
|
3280 3281 3282 3283 3284 |
while (n->free_objects > n->free_limit && !list_empty(&n->slabs_free)) { n->free_objects -= cachep->num; page = list_last_entry(&n->slabs_free, struct page, lru); |
de24baecd
|
3285 |
list_move(&page->lru, list); |
f728b0a5d
|
3286 |
n->free_slabs--; |
bf00bd345
|
3287 |
n->total_slabs--; |
6052b7880
|
3288 |
} |
1da177e4c
|
3289 |
} |
343e0d7a9
|
3290 |
static void cache_flusharray(struct kmem_cache *cachep, struct array_cache *ac) |
1da177e4c
|
3291 3292 |
{ int batchcount; |
ce8eb6c42
|
3293 |
struct kmem_cache_node *n; |
7d6e6d09d
|
3294 |
int node = numa_mem_id(); |
97654dfa2
|
3295 |
LIST_HEAD(list); |
1da177e4c
|
3296 3297 |
batchcount = ac->batchcount; |
260b61dd4
|
3298 |
|
1da177e4c
|
3299 |
check_irq_off(); |
18bf85411
|
3300 |
n = get_node(cachep, node); |
ce8eb6c42
|
3301 3302 3303 |
spin_lock(&n->list_lock); if (n->shared) { struct array_cache *shared_array = n->shared; |
b28a02de8
|
3304 |
int max = shared_array->limit - shared_array->avail; |
1da177e4c
|
3305 3306 3307 |
if (max) { if (batchcount > max) batchcount = max; |
e498be7da
|
3308 |
memcpy(&(shared_array->entry[shared_array->avail]), |
b28a02de8
|
3309 |
ac->entry, sizeof(void *) * batchcount); |
1da177e4c
|
3310 3311 3312 3313 |
shared_array->avail += batchcount; goto free_done; } } |
97654dfa2
|
3314 |
free_block(cachep, ac->entry, batchcount, node, &list); |
a737b3e2f
|
3315 |
free_done: |
1da177e4c
|
3316 3317 3318 |
#if STATS { int i = 0; |
73c0219d8
|
3319 |
struct page *page; |
1da177e4c
|
3320 |
|
73c0219d8
|
3321 |
list_for_each_entry(page, &n->slabs_free, lru) { |
8456a648c
|
3322 |
BUG_ON(page->active); |
1da177e4c
|
3323 3324 |
i++; |
1da177e4c
|
3325 3326 3327 3328 |
} STATS_SET_FREEABLE(cachep, i); } #endif |
ce8eb6c42
|
3329 |
spin_unlock(&n->list_lock); |
97654dfa2
|
3330 |
slabs_destroy(cachep, &list); |
1da177e4c
|
3331 |
ac->avail -= batchcount; |
a737b3e2f
|
3332 |
memmove(ac->entry, &(ac->entry[batchcount]), sizeof(void *)*ac->avail); |
1da177e4c
|
3333 3334 3335 |
} /* |
a737b3e2f
|
3336 3337 |
* 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
|
3338 |
*/ |
a947eb95e
|
3339 |
static inline void __cache_free(struct kmem_cache *cachep, void *objp, |
7c0cb9c64
|
3340 |
unsigned long caller) |
1da177e4c
|
3341 |
{ |
55834c590
|
3342 3343 3344 3345 3346 3347 |
/* Put the object into the quarantine, don't touch it for now. */ if (kasan_slab_free(cachep, objp)) return; ___cache_free(cachep, objp, caller); } |
1da177e4c
|
3348 |
|
55834c590
|
3349 3350 3351 3352 |
void ___cache_free(struct kmem_cache *cachep, void *objp, unsigned long caller) { struct array_cache *ac = cpu_cache_get(cachep); |
7ed2f9e66
|
3353 |
|
1da177e4c
|
3354 |
check_irq_off(); |
d5cff6352
|
3355 |
kmemleak_free_recursive(objp, cachep->flags); |
a947eb95e
|
3356 |
objp = cache_free_debugcheck(cachep, objp, caller); |
1da177e4c
|
3357 |
|
1807a1aaf
|
3358 3359 3360 3361 3362 3363 3364 |
/* * 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
|
3365 |
if (nr_online_nodes > 1 && cache_free_alien(cachep, objp)) |
729bd0b74
|
3366 |
return; |
3d8801940
|
3367 |
if (ac->avail < ac->limit) { |
1da177e4c
|
3368 |
STATS_INC_FREEHIT(cachep); |
1da177e4c
|
3369 3370 3371 |
} else { STATS_INC_FREEMISS(cachep); cache_flusharray(cachep, ac); |
1da177e4c
|
3372 |
} |
42c8c99cd
|
3373 |
|
f68f8dddb
|
3374 3375 3376 3377 3378 3379 3380 3381 3382 3383 |
if (sk_memalloc_socks()) { struct page *page = virt_to_head_page(objp); if (unlikely(PageSlabPfmemalloc(page))) { cache_free_pfmemalloc(cachep, page, objp); return; } } ac->entry[ac->avail++] = objp; |
1da177e4c
|
3384 3385 3386 3387 3388 3389 3390 3391 3392 3393 |
} /** * 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
|
3394 |
void *kmem_cache_alloc(struct kmem_cache *cachep, gfp_t flags) |
1da177e4c
|
3395 |
{ |
48356303f
|
3396 |
void *ret = slab_alloc(cachep, flags, _RET_IP_); |
36555751c
|
3397 |
|
505f5dcb1
|
3398 |
kasan_slab_alloc(cachep, ret, flags); |
ca2b84cb3
|
3399 |
trace_kmem_cache_alloc(_RET_IP_, ret, |
8c138bc00
|
3400 |
cachep->object_size, cachep->size, flags); |
36555751c
|
3401 3402 |
return ret; |
1da177e4c
|
3403 3404 |
} EXPORT_SYMBOL(kmem_cache_alloc); |
7b0501dd6
|
3405 3406 3407 3408 3409 3410 3411 3412 3413 |
static __always_inline void cache_alloc_debugcheck_after_bulk(struct kmem_cache *s, gfp_t flags, size_t size, void **p, unsigned long caller) { size_t i; for (i = 0; i < size; i++) p[i] = cache_alloc_debugcheck_after(s, flags, p[i], caller); } |
865762a81
|
3414 |
int kmem_cache_alloc_bulk(struct kmem_cache *s, gfp_t flags, size_t size, |
2a777eac1
|
3415 |
void **p) |
484748f0b
|
3416 |
{ |
2a777eac1
|
3417 3418 3419 3420 3421 3422 3423 3424 3425 3426 3427 |
size_t i; s = slab_pre_alloc_hook(s, flags); if (!s) return 0; cache_alloc_debugcheck_before(s, flags); local_irq_disable(); for (i = 0; i < size; i++) { void *objp = __do_cache_alloc(s, flags); |
2a777eac1
|
3428 3429 3430 3431 3432 |
if (unlikely(!objp)) goto error; p[i] = objp; } local_irq_enable(); |
7b0501dd6
|
3433 |
cache_alloc_debugcheck_after_bulk(s, flags, size, p, _RET_IP_); |
2a777eac1
|
3434 3435 3436 3437 3438 3439 3440 3441 3442 3443 |
/* Clear memory outside IRQ disabled section */ if (unlikely(flags & __GFP_ZERO)) for (i = 0; i < size; i++) memset(p[i], 0, s->object_size); slab_post_alloc_hook(s, flags, size, p); /* FIXME: Trace call missing. Christoph would like a bulk variant */ return size; error: local_irq_enable(); |
7b0501dd6
|
3444 |
cache_alloc_debugcheck_after_bulk(s, flags, i, p, _RET_IP_); |
2a777eac1
|
3445 3446 3447 |
slab_post_alloc_hook(s, flags, i, p); __kmem_cache_free_bulk(s, i, p); return 0; |
484748f0b
|
3448 3449 |
} EXPORT_SYMBOL(kmem_cache_alloc_bulk); |
0f24f1287
|
3450 |
#ifdef CONFIG_TRACING |
85beb5869
|
3451 |
void * |
4052147c0
|
3452 |
kmem_cache_alloc_trace(struct kmem_cache *cachep, gfp_t flags, size_t size) |
36555751c
|
3453 |
{ |
85beb5869
|
3454 |
void *ret; |
48356303f
|
3455 |
ret = slab_alloc(cachep, flags, _RET_IP_); |
85beb5869
|
3456 |
|
505f5dcb1
|
3457 |
kasan_kmalloc(cachep, ret, size, flags); |
85beb5869
|
3458 |
trace_kmalloc(_RET_IP_, ret, |
ff4fcd01e
|
3459 |
size, cachep->size, flags); |
85beb5869
|
3460 |
return ret; |
36555751c
|
3461 |
} |
85beb5869
|
3462 |
EXPORT_SYMBOL(kmem_cache_alloc_trace); |
36555751c
|
3463 |
#endif |
1da177e4c
|
3464 |
#ifdef CONFIG_NUMA |
d0d04b78f
|
3465 3466 3467 3468 3469 3470 3471 3472 3473 3474 3475 |
/** * 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. * * 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. */ |
8b98c1699
|
3476 3477 |
void *kmem_cache_alloc_node(struct kmem_cache *cachep, gfp_t flags, int nodeid) { |
48356303f
|
3478 |
void *ret = slab_alloc_node(cachep, flags, nodeid, _RET_IP_); |
36555751c
|
3479 |
|
505f5dcb1
|
3480 |
kasan_slab_alloc(cachep, ret, flags); |
ca2b84cb3
|
3481 |
trace_kmem_cache_alloc_node(_RET_IP_, ret, |
8c138bc00
|
3482 |
cachep->object_size, cachep->size, |
ca2b84cb3
|
3483 |
flags, nodeid); |
36555751c
|
3484 3485 |
return ret; |
8b98c1699
|
3486 |
} |
1da177e4c
|
3487 |
EXPORT_SYMBOL(kmem_cache_alloc_node); |
0f24f1287
|
3488 |
#ifdef CONFIG_TRACING |
4052147c0
|
3489 |
void *kmem_cache_alloc_node_trace(struct kmem_cache *cachep, |
85beb5869
|
3490 |
gfp_t flags, |
4052147c0
|
3491 3492 |
int nodeid, size_t size) |
36555751c
|
3493 |
{ |
85beb5869
|
3494 |
void *ret; |
592f41450
|
3495 |
ret = slab_alloc_node(cachep, flags, nodeid, _RET_IP_); |
505f5dcb1
|
3496 3497 |
kasan_kmalloc(cachep, ret, size, flags); |
85beb5869
|
3498 |
trace_kmalloc_node(_RET_IP_, ret, |
ff4fcd01e
|
3499 |
size, cachep->size, |
85beb5869
|
3500 3501 |
flags, nodeid); return ret; |
36555751c
|
3502 |
} |
85beb5869
|
3503 |
EXPORT_SYMBOL(kmem_cache_alloc_node_trace); |
36555751c
|
3504 |
#endif |
8b98c1699
|
3505 |
static __always_inline void * |
7c0cb9c64
|
3506 |
__do_kmalloc_node(size_t size, gfp_t flags, int node, unsigned long caller) |
97e2bde47
|
3507 |
{ |
343e0d7a9
|
3508 |
struct kmem_cache *cachep; |
7ed2f9e66
|
3509 |
void *ret; |
97e2bde47
|
3510 |
|
2c59dd654
|
3511 |
cachep = kmalloc_slab(size, flags); |
6cb8f9132
|
3512 3513 |
if (unlikely(ZERO_OR_NULL_PTR(cachep))) return cachep; |
7ed2f9e66
|
3514 |
ret = kmem_cache_alloc_node_trace(cachep, flags, node, size); |
505f5dcb1
|
3515 |
kasan_kmalloc(cachep, ret, size, flags); |
7ed2f9e66
|
3516 3517 |
return ret; |
97e2bde47
|
3518 |
} |
8b98c1699
|
3519 |
|
8b98c1699
|
3520 3521 |
void *__kmalloc_node(size_t size, gfp_t flags, int node) { |
7c0cb9c64
|
3522 |
return __do_kmalloc_node(size, flags, node, _RET_IP_); |
8b98c1699
|
3523 |
} |
dbe5e69d2
|
3524 |
EXPORT_SYMBOL(__kmalloc_node); |
8b98c1699
|
3525 3526 |
void *__kmalloc_node_track_caller(size_t size, gfp_t flags, |
ce71e27c6
|
3527 |
int node, unsigned long caller) |
8b98c1699
|
3528 |
{ |
7c0cb9c64
|
3529 |
return __do_kmalloc_node(size, flags, node, caller); |
8b98c1699
|
3530 3531 |
} EXPORT_SYMBOL(__kmalloc_node_track_caller); |
8b98c1699
|
3532 |
#endif /* CONFIG_NUMA */ |
1da177e4c
|
3533 3534 |
/** |
800590f52
|
3535 |
* __do_kmalloc - allocate memory |
1da177e4c
|
3536 |
* @size: how many bytes of memory are required. |
800590f52
|
3537 |
* @flags: the type of memory to allocate (see kmalloc). |
911851e6e
|
3538 |
* @caller: function caller for debug tracking of the caller |
1da177e4c
|
3539 |
*/ |
7fd6b1413
|
3540 |
static __always_inline void *__do_kmalloc(size_t size, gfp_t flags, |
7c0cb9c64
|
3541 |
unsigned long caller) |
1da177e4c
|
3542 |
{ |
343e0d7a9
|
3543 |
struct kmem_cache *cachep; |
36555751c
|
3544 |
void *ret; |
1da177e4c
|
3545 |
|
2c59dd654
|
3546 |
cachep = kmalloc_slab(size, flags); |
a5c96d8a1
|
3547 3548 |
if (unlikely(ZERO_OR_NULL_PTR(cachep))) return cachep; |
48356303f
|
3549 |
ret = slab_alloc(cachep, flags, caller); |
36555751c
|
3550 |
|
505f5dcb1
|
3551 |
kasan_kmalloc(cachep, ret, size, flags); |
7c0cb9c64
|
3552 |
trace_kmalloc(caller, ret, |
3b0efdfa1
|
3553 |
size, cachep->size, flags); |
36555751c
|
3554 3555 |
return ret; |
7fd6b1413
|
3556 |
} |
7fd6b1413
|
3557 3558 |
void *__kmalloc(size_t size, gfp_t flags) { |
7c0cb9c64
|
3559 |
return __do_kmalloc(size, flags, _RET_IP_); |
1da177e4c
|
3560 3561 |
} EXPORT_SYMBOL(__kmalloc); |
ce71e27c6
|
3562 |
void *__kmalloc_track_caller(size_t size, gfp_t flags, unsigned long caller) |
7fd6b1413
|
3563 |
{ |
7c0cb9c64
|
3564 |
return __do_kmalloc(size, flags, caller); |
7fd6b1413
|
3565 3566 |
} EXPORT_SYMBOL(__kmalloc_track_caller); |
1d2c8eea6
|
3567 |
|
1da177e4c
|
3568 3569 3570 3571 3572 3573 3574 3575 |
/** * 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
|
3576 |
void kmem_cache_free(struct kmem_cache *cachep, void *objp) |
1da177e4c
|
3577 3578 |
{ unsigned long flags; |
b9ce5ef49
|
3579 3580 3581 |
cachep = cache_from_obj(cachep, objp); if (!cachep) return; |
1da177e4c
|
3582 3583 |
local_irq_save(flags); |
d97d476b1
|
3584 |
debug_check_no_locks_freed(objp, cachep->object_size); |
3ac7fe5a4
|
3585 |
if (!(cachep->flags & SLAB_DEBUG_OBJECTS)) |
8c138bc00
|
3586 |
debug_check_no_obj_freed(objp, cachep->object_size); |
7c0cb9c64
|
3587 |
__cache_free(cachep, objp, _RET_IP_); |
1da177e4c
|
3588 |
local_irq_restore(flags); |
36555751c
|
3589 |
|
ca2b84cb3
|
3590 |
trace_kmem_cache_free(_RET_IP_, objp); |
1da177e4c
|
3591 3592 |
} EXPORT_SYMBOL(kmem_cache_free); |
e6cdb58d1
|
3593 3594 3595 3596 3597 3598 3599 3600 |
void kmem_cache_free_bulk(struct kmem_cache *orig_s, size_t size, void **p) { struct kmem_cache *s; size_t i; local_irq_disable(); for (i = 0; i < size; i++) { void *objp = p[i]; |
ca2571955
|
3601 3602 3603 3604 |
if (!orig_s) /* called via kfree_bulk */ s = virt_to_cache(objp); else s = cache_from_obj(orig_s, objp); |
e6cdb58d1
|
3605 3606 3607 3608 3609 3610 3611 3612 3613 3614 3615 3616 |
debug_check_no_locks_freed(objp, s->object_size); if (!(s->flags & SLAB_DEBUG_OBJECTS)) debug_check_no_obj_freed(objp, s->object_size); __cache_free(s, objp, _RET_IP_); } local_irq_enable(); /* FIXME: add tracing */ } EXPORT_SYMBOL(kmem_cache_free_bulk); |
1da177e4c
|
3617 |
/** |
1da177e4c
|
3618 3619 3620 |
* kfree - free previously allocated memory * @objp: pointer returned by kmalloc. * |
80e93effc
|
3621 3622 |
* If @objp is NULL, no operation is performed. * |
1da177e4c
|
3623 3624 3625 3626 3627 |
* Don't free memory not originally allocated by kmalloc() * or you will run into trouble. */ void kfree(const void *objp) { |
343e0d7a9
|
3628 |
struct kmem_cache *c; |
1da177e4c
|
3629 |
unsigned long flags; |
2121db74b
|
3630 |
trace_kfree(_RET_IP_, objp); |
6cb8f9132
|
3631 |
if (unlikely(ZERO_OR_NULL_PTR(objp))) |
1da177e4c
|
3632 3633 3634 |
return; local_irq_save(flags); kfree_debugcheck(objp); |
6ed5eb221
|
3635 |
c = virt_to_cache(objp); |
8c138bc00
|
3636 3637 3638 |
debug_check_no_locks_freed(objp, c->object_size); debug_check_no_obj_freed(objp, c->object_size); |
7c0cb9c64
|
3639 |
__cache_free(c, (void *)objp, _RET_IP_); |
1da177e4c
|
3640 3641 3642 |
local_irq_restore(flags); } EXPORT_SYMBOL(kfree); |
e498be7da
|
3643 |
/* |
ce8eb6c42
|
3644 |
* This initializes kmem_cache_node or resizes various caches for all nodes. |
e498be7da
|
3645 |
*/ |
c3d332b6b
|
3646 |
static int setup_kmem_cache_nodes(struct kmem_cache *cachep, gfp_t gfp) |
e498be7da
|
3647 |
{ |
c3d332b6b
|
3648 |
int ret; |
e498be7da
|
3649 |
int node; |
ce8eb6c42
|
3650 |
struct kmem_cache_node *n; |
e498be7da
|
3651 |
|
9c09a95cf
|
3652 |
for_each_online_node(node) { |
c3d332b6b
|
3653 3654 |
ret = setup_kmem_cache_node(cachep, node, gfp, true); if (ret) |
e498be7da
|
3655 |
goto fail; |
e498be7da
|
3656 |
} |
c3d332b6b
|
3657 |
|
cafeb02e0
|
3658 |
return 0; |
0718dc2a8
|
3659 |
|
a737b3e2f
|
3660 |
fail: |
3b0efdfa1
|
3661 |
if (!cachep->list.next) { |
0718dc2a8
|
3662 3663 3664 |
/* Cache is not active yet. Roll back what we did */ node--; while (node >= 0) { |
18bf85411
|
3665 3666 |
n = get_node(cachep, node); if (n) { |
ce8eb6c42
|
3667 3668 3669 |
kfree(n->shared); free_alien_cache(n->alien); kfree(n); |
6a67368c3
|
3670 |
cachep->node[node] = NULL; |
0718dc2a8
|
3671 3672 3673 3674 |
} node--; } } |
cafeb02e0
|
3675 |
return -ENOMEM; |
e498be7da
|
3676 |
} |
18004c5d4
|
3677 |
/* Always called with the slab_mutex held */ |
943a451a8
|
3678 |
static int __do_tune_cpucache(struct kmem_cache *cachep, int limit, |
83b519e8b
|
3679 |
int batchcount, int shared, gfp_t gfp) |
1da177e4c
|
3680 |
{ |
bf0dea23a
|
3681 3682 |
struct array_cache __percpu *cpu_cache, *prev; int cpu; |
1da177e4c
|
3683 |
|
bf0dea23a
|
3684 3685 |
cpu_cache = alloc_kmem_cache_cpus(cachep, limit, batchcount); if (!cpu_cache) |
d2e7b7d0a
|
3686 |
return -ENOMEM; |
bf0dea23a
|
3687 3688 |
prev = cachep->cpu_cache; cachep->cpu_cache = cpu_cache; |
a87c75fbc
|
3689 3690 3691 3692 3693 3694 |
/* * Without a previous cpu_cache there's no need to synchronize remote * cpus, so skip the IPIs. */ if (prev) kick_all_cpus_sync(); |
e498be7da
|
3695 |
|
1da177e4c
|
3696 |
check_irq_on(); |
1da177e4c
|
3697 3698 |
cachep->batchcount = batchcount; cachep->limit = limit; |
e498be7da
|
3699 |
cachep->shared = shared; |
1da177e4c
|
3700 |
|
bf0dea23a
|
3701 |
if (!prev) |
c3d332b6b
|
3702 |
goto setup_node; |
bf0dea23a
|
3703 3704 |
for_each_online_cpu(cpu) { |
97654dfa2
|
3705 |
LIST_HEAD(list); |
18bf85411
|
3706 3707 |
int node; struct kmem_cache_node *n; |
bf0dea23a
|
3708 |
struct array_cache *ac = per_cpu_ptr(prev, cpu); |
18bf85411
|
3709 |
|
bf0dea23a
|
3710 |
node = cpu_to_mem(cpu); |
18bf85411
|
3711 3712 |
n = get_node(cachep, node); spin_lock_irq(&n->list_lock); |
bf0dea23a
|
3713 |
free_block(cachep, ac->entry, ac->avail, node, &list); |
18bf85411
|
3714 |
spin_unlock_irq(&n->list_lock); |
97654dfa2
|
3715 |
slabs_destroy(cachep, &list); |
1da177e4c
|
3716 |
} |
bf0dea23a
|
3717 |
free_percpu(prev); |
c3d332b6b
|
3718 3719 |
setup_node: return setup_kmem_cache_nodes(cachep, gfp); |
1da177e4c
|
3720 |
} |
943a451a8
|
3721 3722 3723 3724 |
static int do_tune_cpucache(struct kmem_cache *cachep, int limit, int batchcount, int shared, gfp_t gfp) { int ret; |
426589f57
|
3725 |
struct kmem_cache *c; |
943a451a8
|
3726 3727 3728 3729 3730 3731 3732 3733 |
ret = __do_tune_cpucache(cachep, limit, batchcount, shared, gfp); if (slab_state < FULL) return ret; if ((ret < 0) || !is_root_cache(cachep)) return ret; |
426589f57
|
3734 3735 3736 3737 |
lockdep_assert_held(&slab_mutex); for_each_memcg_cache(c, cachep) { /* return value determined by the root cache only */ __do_tune_cpucache(c, limit, batchcount, shared, gfp); |
943a451a8
|
3738 3739 3740 3741 |
} return ret; } |
18004c5d4
|
3742 |
/* Called with slab_mutex held always */ |
83b519e8b
|
3743 |
static int enable_cpucache(struct kmem_cache *cachep, gfp_t gfp) |
1da177e4c
|
3744 3745 |
{ int err; |
943a451a8
|
3746 3747 3748 |
int limit = 0; int shared = 0; int batchcount = 0; |
7c00fce98
|
3749 |
err = cache_random_seq_create(cachep, cachep->num, gfp); |
c7ce4f60a
|
3750 3751 |
if (err) goto end; |
943a451a8
|
3752 3753 3754 3755 3756 3757 |
if (!is_root_cache(cachep)) { struct kmem_cache *root = memcg_root_cache(cachep); limit = root->limit; shared = root->shared; batchcount = root->batchcount; } |
1da177e4c
|
3758 |
|
943a451a8
|
3759 3760 |
if (limit && shared && batchcount) goto skip_setup; |
a737b3e2f
|
3761 3762 |
/* * The head array serves three purposes: |
1da177e4c
|
3763 3764 |
* - create a LIFO ordering, i.e. return objects that are cache-warm * - reduce the number of spinlock operations. |
a737b3e2f
|
3765 |
* - reduce the number of linked list operations on the slab and |
1da177e4c
|
3766 3767 3768 3769 |
* bufctl chains: array operations are cheaper. * The numbers are guessed, we should auto-tune as described by * Bonwick. */ |
3b0efdfa1
|
3770 |
if (cachep->size > 131072) |
1da177e4c
|
3771 |
limit = 1; |
3b0efdfa1
|
3772 |
else if (cachep->size > PAGE_SIZE) |
1da177e4c
|
3773 |
limit = 8; |
3b0efdfa1
|
3774 |
else if (cachep->size > 1024) |
1da177e4c
|
3775 |
limit = 24; |
3b0efdfa1
|
3776 |
else if (cachep->size > 256) |
1da177e4c
|
3777 3778 3779 |
limit = 54; else limit = 120; |
a737b3e2f
|
3780 3781 |
/* * CPU bound tasks (e.g. network routing) can exhibit cpu bound |
1da177e4c
|
3782 3783 3784 3785 3786 3787 3788 3789 |
* 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; |
3b0efdfa1
|
3790 |
if (cachep->size <= PAGE_SIZE && num_possible_cpus() > 1) |
1da177e4c
|
3791 |
shared = 8; |
1da177e4c
|
3792 3793 |
#if DEBUG |
a737b3e2f
|
3794 3795 3796 |
/* * With debugging enabled, large batchcount lead to excessively long * periods with disabled local interrupts. Limit the batchcount |
1da177e4c
|
3797 3798 3799 3800 |
*/ if (limit > 32) limit = 32; #endif |
943a451a8
|
3801 3802 3803 |
batchcount = (limit + 1) / 2; skip_setup: err = do_tune_cpucache(cachep, limit, batchcount, shared, gfp); |
c7ce4f60a
|
3804 |
end: |
1da177e4c
|
3805 |
if (err) |
1170532bb
|
3806 3807 |
pr_err("enable_cpucache failed for %s, error %d ", |
b28a02de8
|
3808 |
cachep->name, -err); |
2ed3a4ef9
|
3809 |
return err; |
1da177e4c
|
3810 |
} |
1b55253a7
|
3811 |
/* |
ce8eb6c42
|
3812 3813 |
* Drain an array if it contains any elements taking the node lock only if * necessary. Note that the node listlock also protects the array_cache |
b18e7e654
|
3814 |
* if drain_array() is used on the shared array. |
1b55253a7
|
3815 |
*/ |
ce8eb6c42
|
3816 |
static void drain_array(struct kmem_cache *cachep, struct kmem_cache_node *n, |
18726ca8b
|
3817 |
struct array_cache *ac, int node) |
1da177e4c
|
3818 |
{ |
97654dfa2
|
3819 |
LIST_HEAD(list); |
18726ca8b
|
3820 3821 3822 |
/* ac from n->shared can be freed if we don't hold the slab_mutex. */ check_mutex_acquired(); |
1da177e4c
|
3823 |
|
1b55253a7
|
3824 3825 |
if (!ac || !ac->avail) return; |
18726ca8b
|
3826 3827 |
if (ac->touched) { |
1da177e4c
|
3828 |
ac->touched = 0; |
18726ca8b
|
3829 |
return; |
1da177e4c
|
3830 |
} |
18726ca8b
|
3831 3832 3833 3834 3835 3836 |
spin_lock_irq(&n->list_lock); drain_array_locked(cachep, ac, node, false, &list); spin_unlock_irq(&n->list_lock); slabs_destroy(cachep, &list); |
1da177e4c
|
3837 3838 3839 3840 |
} /** * cache_reap - Reclaim memory from caches. |
05fb6bf0b
|
3841 |
* @w: work descriptor |
1da177e4c
|
3842 3843 3844 3845 3846 3847 |
* * 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
|
3848 3849 |
* If we cannot acquire the cache chain mutex then just give up - we'll try * again on the next iteration. |
1da177e4c
|
3850 |
*/ |
7c5cae368
|
3851 |
static void cache_reap(struct work_struct *w) |
1da177e4c
|
3852 |
{ |
7a7c381d2
|
3853 |
struct kmem_cache *searchp; |
ce8eb6c42
|
3854 |
struct kmem_cache_node *n; |
7d6e6d09d
|
3855 |
int node = numa_mem_id(); |
bf6aede71
|
3856 |
struct delayed_work *work = to_delayed_work(w); |
1da177e4c
|
3857 |
|
18004c5d4
|
3858 |
if (!mutex_trylock(&slab_mutex)) |
1da177e4c
|
3859 |
/* Give up. Setup the next iteration. */ |
7c5cae368
|
3860 |
goto out; |
1da177e4c
|
3861 |
|
18004c5d4
|
3862 |
list_for_each_entry(searchp, &slab_caches, list) { |
1da177e4c
|
3863 |
check_irq_on(); |
35386e3b0
|
3864 |
/* |
ce8eb6c42
|
3865 |
* We only take the node lock if absolutely necessary and we |
35386e3b0
|
3866 3867 3868 |
* have established with reasonable certainty that * we can do some work if the lock was obtained. */ |
18bf85411
|
3869 |
n = get_node(searchp, node); |
35386e3b0
|
3870 |
|
ce8eb6c42
|
3871 |
reap_alien(searchp, n); |
1da177e4c
|
3872 |
|
18726ca8b
|
3873 |
drain_array(searchp, n, cpu_cache_get(searchp), node); |
1da177e4c
|
3874 |
|
35386e3b0
|
3875 3876 3877 3878 |
/* * These are racy checks but it does not matter * if we skip one check or scan twice. */ |
ce8eb6c42
|
3879 |
if (time_after(n->next_reap, jiffies)) |
35386e3b0
|
3880 |
goto next; |
1da177e4c
|
3881 |
|
5f0985bb1
|
3882 |
n->next_reap = jiffies + REAPTIMEOUT_NODE; |
1da177e4c
|
3883 |
|
18726ca8b
|
3884 |
drain_array(searchp, n, n->shared, node); |
1da177e4c
|
3885 |
|
ce8eb6c42
|
3886 3887 |
if (n->free_touched) n->free_touched = 0; |
ed11d9eb2
|
3888 3889 |
else { int freed; |
1da177e4c
|
3890 |
|
ce8eb6c42
|
3891 |
freed = drain_freelist(searchp, n, (n->free_limit + |
ed11d9eb2
|
3892 3893 3894 |
5 * searchp->num - 1) / (5 * searchp->num)); STATS_ADD_REAPED(searchp, freed); } |
35386e3b0
|
3895 |
next: |
1da177e4c
|
3896 3897 3898 |
cond_resched(); } check_irq_on(); |
18004c5d4
|
3899 |
mutex_unlock(&slab_mutex); |
8fce4d8e3
|
3900 |
next_reap_node(); |
7c5cae368
|
3901 |
out: |
a737b3e2f
|
3902 |
/* Set up the next iteration */ |
20eaa393f
|
3903 3904 |
schedule_delayed_work_on(smp_processor_id(), work, round_jiffies_relative(REAPTIMEOUT_AC)); |
1da177e4c
|
3905 |
} |
158a96242
|
3906 |
#ifdef CONFIG_SLABINFO |
0d7561c61
|
3907 |
void get_slabinfo(struct kmem_cache *cachep, struct slabinfo *sinfo) |
1da177e4c
|
3908 |
{ |
f728b0a5d
|
3909 |
unsigned long active_objs, num_objs, active_slabs; |
bf00bd345
|
3910 3911 |
unsigned long total_slabs = 0, free_objs = 0, shared_avail = 0; unsigned long free_slabs = 0; |
e498be7da
|
3912 |
int node; |
ce8eb6c42
|
3913 |
struct kmem_cache_node *n; |
1da177e4c
|
3914 |
|
18bf85411
|
3915 |
for_each_kmem_cache_node(cachep, node, n) { |
ca3b9b917
|
3916 |
check_irq_on(); |
ce8eb6c42
|
3917 |
spin_lock_irq(&n->list_lock); |
e498be7da
|
3918 |
|
bf00bd345
|
3919 3920 |
total_slabs += n->total_slabs; free_slabs += n->free_slabs; |
f728b0a5d
|
3921 |
free_objs += n->free_objects; |
07a63c41f
|
3922 |
|
ce8eb6c42
|
3923 3924 |
if (n->shared) shared_avail += n->shared->avail; |
e498be7da
|
3925 |
|
ce8eb6c42
|
3926 |
spin_unlock_irq(&n->list_lock); |
1da177e4c
|
3927 |
} |
bf00bd345
|
3928 3929 |
num_objs = total_slabs * cachep->num; active_slabs = total_slabs - free_slabs; |
f728b0a5d
|
3930 |
active_objs = num_objs - free_objs; |
1da177e4c
|
3931 |
|
0d7561c61
|
3932 3933 3934 |
sinfo->active_objs = active_objs; sinfo->num_objs = num_objs; sinfo->active_slabs = active_slabs; |
bf00bd345
|
3935 |
sinfo->num_slabs = total_slabs; |
0d7561c61
|
3936 3937 3938 3939 3940 3941 3942 3943 3944 3945 |
sinfo->shared_avail = shared_avail; sinfo->limit = cachep->limit; sinfo->batchcount = cachep->batchcount; sinfo->shared = cachep->shared; sinfo->objects_per_slab = cachep->num; sinfo->cache_order = cachep->gfporder; } void slabinfo_show_stats(struct seq_file *m, struct kmem_cache *cachep) { |
1da177e4c
|
3946 |
#if STATS |
ce8eb6c42
|
3947 |
{ /* node stats */ |
1da177e4c
|
3948 3949 3950 3951 3952 3953 |
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
|
3954 |
unsigned long node_allocs = cachep->node_allocs; |
e498be7da
|
3955 |
unsigned long node_frees = cachep->node_frees; |
fb7faf331
|
3956 |
unsigned long overflows = cachep->node_overflow; |
1da177e4c
|
3957 |
|
756a025f0
|
3958 |
seq_printf(m, " : globalstat %7lu %6lu %5lu %4lu %4lu %4lu %4lu %4lu %4lu", |
e92dd4fd1
|
3959 3960 3961 |
allocs, high, grown, reaped, errors, max_freeable, node_allocs, node_frees, overflows); |
1da177e4c
|
3962 3963 3964 3965 3966 3967 3968 3969 3970 |
} /* 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
|
3971 |
allochit, allocmiss, freehit, freemiss); |
1da177e4c
|
3972 3973 |
} #endif |
1da177e4c
|
3974 |
} |
1da177e4c
|
3975 3976 3977 3978 3979 3980 3981 3982 |
#define MAX_SLABINFO_WRITE 128 /** * slabinfo_write - Tuning for the slab allocator * @file: unused * @buffer: user buffer * @count: data length * @ppos: unused */ |
b7454ad3c
|
3983 |
ssize_t slabinfo_write(struct file *file, const char __user *buffer, |
b28a02de8
|
3984 |
size_t count, loff_t *ppos) |
1da177e4c
|
3985 |
{ |
b28a02de8
|
3986 |
char kbuf[MAX_SLABINFO_WRITE + 1], *tmp; |
1da177e4c
|
3987 |
int limit, batchcount, shared, res; |
7a7c381d2
|
3988 |
struct kmem_cache *cachep; |
b28a02de8
|
3989 |
|
1da177e4c
|
3990 3991 3992 3993 |
if (count > MAX_SLABINFO_WRITE) return -EINVAL; if (copy_from_user(&kbuf, buffer, count)) return -EFAULT; |
b28a02de8
|
3994 |
kbuf[MAX_SLABINFO_WRITE] = '\0'; |
1da177e4c
|
3995 3996 3997 3998 3999 4000 4001 4002 4003 4004 |
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. */ |
18004c5d4
|
4005 |
mutex_lock(&slab_mutex); |
1da177e4c
|
4006 |
res = -EINVAL; |
18004c5d4
|
4007 |
list_for_each_entry(cachep, &slab_caches, list) { |
1da177e4c
|
4008 |
if (!strcmp(cachep->name, kbuf)) { |
a737b3e2f
|
4009 4010 |
if (limit < 1 || batchcount < 1 || batchcount > limit || shared < 0) { |
e498be7da
|
4011 |
res = 0; |
1da177e4c
|
4012 |
} else { |
e498be7da
|
4013 |
res = do_tune_cpucache(cachep, limit, |
83b519e8b
|
4014 4015 |
batchcount, shared, GFP_KERNEL); |
1da177e4c
|
4016 4017 4018 4019 |
} break; } } |
18004c5d4
|
4020 |
mutex_unlock(&slab_mutex); |
1da177e4c
|
4021 4022 4023 4024 |
if (res >= 0) res = count; return res; } |
871751e25
|
4025 4026 |
#ifdef CONFIG_DEBUG_SLAB_LEAK |
871751e25
|
4027 4028 4029 4030 4031 4032 4033 4034 4035 4036 4037 4038 4039 4040 4041 4042 4043 4044 4045 4046 4047 4048 4049 4050 4051 4052 4053 4054 4055 |
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; } |
8456a648c
|
4056 4057 |
static void handle_slab(unsigned long *n, struct kmem_cache *c, struct page *page) |
871751e25
|
4058 4059 |
{ void *p; |
d31676dfd
|
4060 4061 |
int i, j; unsigned long v; |
b1cb0982b
|
4062 |
|
871751e25
|
4063 4064 |
if (n[0] == n[1]) return; |
8456a648c
|
4065 |
for (i = 0, p = page->s_mem; i < c->num; i++, p += c->size) { |
d31676dfd
|
4066 4067 4068 4069 4070 4071 4072 4073 4074 4075 |
bool active = true; for (j = page->active; j < c->num; j++) { if (get_free_obj(page, j) == i) { active = false; break; } } if (!active) |
871751e25
|
4076 |
continue; |
b1cb0982b
|
4077 |
|
d31676dfd
|
4078 4079 4080 4081 4082 4083 4084 4085 4086 4087 |
/* * probe_kernel_read() is used for DEBUG_PAGEALLOC. page table * mapping is established when actual object allocation and * we could mistakenly access the unmapped object in the cpu * cache. */ if (probe_kernel_read(&v, dbg_userword(c, p), sizeof(v))) continue; if (!add_caller(n, v)) |
871751e25
|
4088 4089 4090 4091 4092 4093 4094 |
return; } } static void show_symbol(struct seq_file *m, unsigned long address) { #ifdef CONFIG_KALLSYMS |
871751e25
|
4095 |
unsigned long offset, size; |
9281acea6
|
4096 |
char modname[MODULE_NAME_LEN], name[KSYM_NAME_LEN]; |
871751e25
|
4097 |
|
a5c43dae7
|
4098 |
if (lookup_symbol_attrs(address, &size, &offset, modname, name) == 0) { |
871751e25
|
4099 |
seq_printf(m, "%s+%#lx/%#lx", name, offset, size); |
a5c43dae7
|
4100 |
if (modname[0]) |
871751e25
|
4101 4102 4103 4104 4105 4106 4107 4108 4109 |
seq_printf(m, " [%s]", modname); return; } #endif seq_printf(m, "%p", (void *)address); } static int leaks_show(struct seq_file *m, void *p) { |
0672aa7c2
|
4110 |
struct kmem_cache *cachep = list_entry(p, struct kmem_cache, list); |
8456a648c
|
4111 |
struct page *page; |
ce8eb6c42
|
4112 |
struct kmem_cache_node *n; |
871751e25
|
4113 |
const char *name; |
db8450673
|
4114 |
unsigned long *x = m->private; |
871751e25
|
4115 4116 4117 4118 4119 4120 4121 |
int node; int i; if (!(cachep->flags & SLAB_STORE_USER)) return 0; if (!(cachep->flags & SLAB_RED_ZONE)) return 0; |
d31676dfd
|
4122 4123 4124 4125 4126 4127 4128 4129 4130 4131 4132 |
/* * Set store_user_clean and start to grab stored user information * for all objects on this cache. If some alloc/free requests comes * during the processing, information would be wrong so restart * whole processing. */ do { set_store_user_clean(cachep); drain_cpu_caches(cachep); x[1] = 0; |
871751e25
|
4133 |
|
d31676dfd
|
4134 |
for_each_kmem_cache_node(cachep, node, n) { |
871751e25
|
4135 |
|
d31676dfd
|
4136 4137 |
check_irq_on(); spin_lock_irq(&n->list_lock); |
871751e25
|
4138 |
|
d31676dfd
|
4139 4140 4141 4142 4143 4144 4145 |
list_for_each_entry(page, &n->slabs_full, lru) handle_slab(x, cachep, page); list_for_each_entry(page, &n->slabs_partial, lru) handle_slab(x, cachep, page); spin_unlock_irq(&n->list_lock); } } while (!is_store_user_clean(cachep)); |
871751e25
|
4146 |
|
871751e25
|
4147 |
name = cachep->name; |
db8450673
|
4148 |
if (x[0] == x[1]) { |
871751e25
|
4149 |
/* Increase the buffer size */ |
18004c5d4
|
4150 |
mutex_unlock(&slab_mutex); |
db8450673
|
4151 |
m->private = kzalloc(x[0] * 4 * sizeof(unsigned long), GFP_KERNEL); |
871751e25
|
4152 4153 |
if (!m->private) { /* Too bad, we are really out */ |
db8450673
|
4154 |
m->private = x; |
18004c5d4
|
4155 |
mutex_lock(&slab_mutex); |
871751e25
|
4156 4157 |
return -ENOMEM; } |
db8450673
|
4158 4159 |
*(unsigned long *)m->private = x[0] * 2; kfree(x); |
18004c5d4
|
4160 |
mutex_lock(&slab_mutex); |
871751e25
|
4161 4162 4163 4164 |
/* Now make sure this entry will be retried */ m->count = m->size; return 0; } |
db8450673
|
4165 4166 4167 |
for (i = 0; i < x[1]; i++) { seq_printf(m, "%s: %lu ", name, x[2*i+3]); show_symbol(m, x[2*i+2]); |
871751e25
|
4168 4169 4170 |
seq_putc(m, ' '); } |
d2e7b7d0a
|
4171 |
|
871751e25
|
4172 4173 |
return 0; } |
a0ec95a8e
|
4174 |
static const struct seq_operations slabstats_op = { |
1df3b26f2
|
4175 |
.start = slab_start, |
276a2439c
|
4176 4177 |
.next = slab_next, .stop = slab_stop, |
871751e25
|
4178 4179 |
.show = leaks_show, }; |
a0ec95a8e
|
4180 4181 4182 |
static int slabstats_open(struct inode *inode, struct file *file) { |
b208ce329
|
4183 4184 4185 4186 4187 4188 4189 4190 4191 |
unsigned long *n; n = __seq_open_private(file, &slabstats_op, PAGE_SIZE); if (!n) return -ENOMEM; *n = PAGE_SIZE / (2 * sizeof(unsigned long)); return 0; |
a0ec95a8e
|
4192 4193 4194 4195 4196 4197 4198 4199 4200 4201 4202 4203 4204 4205 |
} 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) { #ifdef CONFIG_DEBUG_SLAB_LEAK proc_create("slab_allocators", 0, NULL, &proc_slabstats_operations); |
871751e25
|
4206 |
#endif |
a0ec95a8e
|
4207 4208 4209 |
return 0; } module_init(slab_proc_init); |
1da177e4c
|
4210 |
#endif |
04385fc5e
|
4211 4212 4213 4214 4215 4216 4217 4218 4219 4220 4221 4222 4223 4224 4225 4226 4227 4228 4229 4230 4231 4232 4233 4234 4235 4236 4237 4238 4239 |
#ifdef CONFIG_HARDENED_USERCOPY /* * Rejects objects that are incorrectly sized. * * Returns NULL if check passes, otherwise const char * to name of cache * to indicate an error. */ const char *__check_heap_object(const void *ptr, unsigned long n, struct page *page) { struct kmem_cache *cachep; unsigned int objnr; unsigned long offset; /* Find and validate object. */ cachep = page->slab_cache; objnr = obj_to_index(cachep, page, (void *)ptr); BUG_ON(objnr >= cachep->num); /* Find offset within object. */ offset = ptr - index_to_obj(cachep, page, objnr) - obj_offset(cachep); /* Allow address range falling entirely within object size. */ if (offset <= cachep->object_size && n <= cachep->object_size - offset) return NULL; return cachep->name; } #endif /* CONFIG_HARDENED_USERCOPY */ |
00e145b6d
|
4240 4241 4242 4243 4244 4245 4246 4247 4248 4249 4250 4251 |
/** * 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
|
4252 |
size_t ksize(const void *objp) |
1da177e4c
|
4253 |
{ |
7ed2f9e66
|
4254 |
size_t size; |
ef8b4520b
|
4255 4256 |
BUG_ON(!objp); if (unlikely(objp == ZERO_SIZE_PTR)) |
00e145b6d
|
4257 |
return 0; |
1da177e4c
|
4258 |
|
7ed2f9e66
|
4259 4260 4261 4262 |
size = virt_to_cache(objp)->object_size; /* We assume that ksize callers could use the whole allocated area, * so we need to unpoison this area. */ |
4ebb31a42
|
4263 |
kasan_unpoison_shadow(objp, size); |
7ed2f9e66
|
4264 4265 |
return size; |
1da177e4c
|
4266 |
} |
b1aabecd5
|
4267 |
EXPORT_SYMBOL(ksize); |