]>
Commit | Line | Data |
---|---|---|
8cdea7c0 BS |
1 | /* memcontrol.h - Memory Controller |
2 | * | |
3 | * Copyright IBM Corporation, 2007 | |
4 | * Author Balbir Singh <balbir@linux.vnet.ibm.com> | |
5 | * | |
78fb7466 PE |
6 | * Copyright 2007 OpenVZ SWsoft Inc |
7 | * Author: Pavel Emelianov <xemul@openvz.org> | |
8 | * | |
8cdea7c0 BS |
9 | * This program is free software; you can redistribute it and/or modify |
10 | * it under the terms of the GNU General Public License as published by | |
11 | * the Free Software Foundation; either version 2 of the License, or | |
12 | * (at your option) any later version. | |
13 | * | |
14 | * This program is distributed in the hope that it will be useful, | |
15 | * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
16 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
17 | * GNU General Public License for more details. | |
18 | */ | |
19 | ||
20 | #ifndef _LINUX_MEMCONTROL_H | |
21 | #define _LINUX_MEMCONTROL_H | |
f8d66542 | 22 | #include <linux/cgroup.h> |
456f998e | 23 | #include <linux/vm_event_item.h> |
7ae1e1d0 | 24 | #include <linux/hardirq.h> |
a8964b9b | 25 | #include <linux/jump_label.h> |
33398cf2 MH |
26 | #include <linux/page_counter.h> |
27 | #include <linux/vmpressure.h> | |
28 | #include <linux/eventfd.h> | |
29 | #include <linux/mmzone.h> | |
30 | #include <linux/writeback.h> | |
fdf1cdb9 | 31 | #include <linux/page-flags.h> |
456f998e | 32 | |
78fb7466 | 33 | struct mem_cgroup; |
8697d331 BS |
34 | struct page; |
35 | struct mm_struct; | |
2633d7a0 | 36 | struct kmem_cache; |
78fb7466 | 37 | |
71cd3113 JW |
38 | /* Cgroup-specific page state, on top of universal node page state */ |
39 | enum memcg_stat_item { | |
40 | MEMCG_CACHE = NR_VM_NODE_STAT_ITEMS, | |
41 | MEMCG_RSS, | |
42 | MEMCG_RSS_HUGE, | |
43 | MEMCG_SWAP, | |
44 | MEMCG_SOCK, | |
45 | /* XXX: why are these zone and not node counters? */ | |
46 | MEMCG_KERNEL_STACK_KB, | |
27ee57c9 VD |
47 | MEMCG_SLAB_RECLAIMABLE, |
48 | MEMCG_SLAB_UNRECLAIMABLE, | |
b2807f07 | 49 | MEMCG_NR_STAT, |
2a7106f2 GT |
50 | }; |
51 | ||
71cd3113 JW |
52 | /* Cgroup-specific events, on top of universal VM events */ |
53 | enum memcg_event_item { | |
54 | MEMCG_LOW = NR_VM_EVENT_ITEMS, | |
55 | MEMCG_HIGH, | |
56 | MEMCG_MAX, | |
57 | MEMCG_OOM, | |
58 | MEMCG_NR_EVENTS, | |
59 | }; | |
60 | ||
5660048c | 61 | struct mem_cgroup_reclaim_cookie { |
ef8f2327 | 62 | pg_data_t *pgdat; |
5660048c JW |
63 | int priority; |
64 | unsigned int generation; | |
65 | }; | |
66 | ||
71cd3113 JW |
67 | #ifdef CONFIG_MEMCG |
68 | ||
69 | #define MEM_CGROUP_ID_SHIFT 16 | |
70 | #define MEM_CGROUP_ID_MAX USHRT_MAX | |
71 | ||
72 | struct mem_cgroup_id { | |
73 | int id; | |
74 | atomic_t ref; | |
75 | }; | |
76 | ||
33398cf2 MH |
77 | /* |
78 | * Per memcg event counter is incremented at every pagein/pageout. With THP, | |
79 | * it will be incremated by the number of pages. This counter is used for | |
80 | * for trigger some periodic events. This is straightforward and better | |
81 | * than using jiffies etc. to handle periodic memcg event. | |
82 | */ | |
83 | enum mem_cgroup_events_target { | |
84 | MEM_CGROUP_TARGET_THRESH, | |
85 | MEM_CGROUP_TARGET_SOFTLIMIT, | |
86 | MEM_CGROUP_TARGET_NUMAINFO, | |
87 | MEM_CGROUP_NTARGETS, | |
88 | }; | |
89 | ||
33398cf2 | 90 | struct mem_cgroup_stat_cpu { |
b2807f07 | 91 | long count[MEMCG_NR_STAT]; |
33398cf2 MH |
92 | unsigned long events[MEMCG_NR_EVENTS]; |
93 | unsigned long nr_page_events; | |
94 | unsigned long targets[MEM_CGROUP_NTARGETS]; | |
95 | }; | |
96 | ||
97 | struct mem_cgroup_reclaim_iter { | |
98 | struct mem_cgroup *position; | |
99 | /* scan generation, increased every round-trip */ | |
100 | unsigned int generation; | |
101 | }; | |
102 | ||
103 | /* | |
104 | * per-zone information in memory controller. | |
105 | */ | |
ef8f2327 | 106 | struct mem_cgroup_per_node { |
33398cf2 | 107 | struct lruvec lruvec; |
b4536f0c | 108 | unsigned long lru_zone_size[MAX_NR_ZONES][NR_LRU_LISTS]; |
33398cf2 MH |
109 | |
110 | struct mem_cgroup_reclaim_iter iter[DEF_PRIORITY + 1]; | |
111 | ||
112 | struct rb_node tree_node; /* RB tree node */ | |
113 | unsigned long usage_in_excess;/* Set to the value by which */ | |
114 | /* the soft limit is exceeded*/ | |
115 | bool on_tree; | |
116 | struct mem_cgroup *memcg; /* Back pointer, we cannot */ | |
117 | /* use container_of */ | |
118 | }; | |
119 | ||
33398cf2 MH |
120 | struct mem_cgroup_threshold { |
121 | struct eventfd_ctx *eventfd; | |
122 | unsigned long threshold; | |
123 | }; | |
124 | ||
125 | /* For threshold */ | |
126 | struct mem_cgroup_threshold_ary { | |
127 | /* An array index points to threshold just below or equal to usage. */ | |
128 | int current_threshold; | |
129 | /* Size of entries[] */ | |
130 | unsigned int size; | |
131 | /* Array of thresholds */ | |
132 | struct mem_cgroup_threshold entries[0]; | |
133 | }; | |
134 | ||
135 | struct mem_cgroup_thresholds { | |
136 | /* Primary thresholds array */ | |
137 | struct mem_cgroup_threshold_ary *primary; | |
138 | /* | |
139 | * Spare threshold array. | |
140 | * This is needed to make mem_cgroup_unregister_event() "never fail". | |
141 | * It must be able to store at least primary->size - 1 entries. | |
142 | */ | |
143 | struct mem_cgroup_threshold_ary *spare; | |
144 | }; | |
145 | ||
567e9ab2 JW |
146 | enum memcg_kmem_state { |
147 | KMEM_NONE, | |
148 | KMEM_ALLOCATED, | |
149 | KMEM_ONLINE, | |
150 | }; | |
151 | ||
33398cf2 MH |
152 | /* |
153 | * The memory controller data structure. The memory controller controls both | |
154 | * page cache and RSS per cgroup. We would eventually like to provide | |
155 | * statistics based on the statistics developed by Rik Van Riel for clock-pro, | |
156 | * to help the administrator determine what knobs to tune. | |
157 | */ | |
158 | struct mem_cgroup { | |
159 | struct cgroup_subsys_state css; | |
160 | ||
73f576c0 JW |
161 | /* Private memcg ID. Used to ID objects that outlive the cgroup */ |
162 | struct mem_cgroup_id id; | |
163 | ||
33398cf2 MH |
164 | /* Accounted resources */ |
165 | struct page_counter memory; | |
37e84351 | 166 | struct page_counter swap; |
0db15298 JW |
167 | |
168 | /* Legacy consumer-oriented counters */ | |
33398cf2 MH |
169 | struct page_counter memsw; |
170 | struct page_counter kmem; | |
0db15298 | 171 | struct page_counter tcpmem; |
33398cf2 MH |
172 | |
173 | /* Normal memory consumption range */ | |
174 | unsigned long low; | |
175 | unsigned long high; | |
176 | ||
f7e1cb6e JW |
177 | /* Range enforcement for interrupt charges */ |
178 | struct work_struct high_work; | |
179 | ||
33398cf2 MH |
180 | unsigned long soft_limit; |
181 | ||
182 | /* vmpressure notifications */ | |
183 | struct vmpressure vmpressure; | |
184 | ||
33398cf2 MH |
185 | /* |
186 | * Should the accounting and control be hierarchical, per subtree? | |
187 | */ | |
188 | bool use_hierarchy; | |
189 | ||
190 | /* protected by memcg_oom_lock */ | |
191 | bool oom_lock; | |
192 | int under_oom; | |
193 | ||
194 | int swappiness; | |
195 | /* OOM-Killer disable */ | |
196 | int oom_kill_disable; | |
197 | ||
472912a2 TH |
198 | /* handle for "memory.events" */ |
199 | struct cgroup_file events_file; | |
200 | ||
33398cf2 MH |
201 | /* protect arrays of thresholds */ |
202 | struct mutex thresholds_lock; | |
203 | ||
204 | /* thresholds for memory usage. RCU-protected */ | |
205 | struct mem_cgroup_thresholds thresholds; | |
206 | ||
207 | /* thresholds for mem+swap usage. RCU-protected */ | |
208 | struct mem_cgroup_thresholds memsw_thresholds; | |
209 | ||
210 | /* For oom notifier event fd */ | |
211 | struct list_head oom_notify; | |
212 | ||
213 | /* | |
214 | * Should we move charges of a task when a task is moved into this | |
215 | * mem_cgroup ? And what type of charges should we move ? | |
216 | */ | |
217 | unsigned long move_charge_at_immigrate; | |
218 | /* | |
219 | * set > 0 if pages under this cgroup are moving to other cgroup. | |
220 | */ | |
221 | atomic_t moving_account; | |
222 | /* taken only while moving_account > 0 */ | |
223 | spinlock_t move_lock; | |
224 | struct task_struct *move_lock_task; | |
225 | unsigned long move_lock_flags; | |
226 | /* | |
227 | * percpu counter. | |
228 | */ | |
229 | struct mem_cgroup_stat_cpu __percpu *stat; | |
33398cf2 | 230 | |
d886f4e4 JW |
231 | unsigned long socket_pressure; |
232 | ||
233 | /* Legacy tcp memory accounting */ | |
0db15298 JW |
234 | bool tcpmem_active; |
235 | int tcpmem_pressure; | |
d886f4e4 | 236 | |
127424c8 | 237 | #ifndef CONFIG_SLOB |
33398cf2 MH |
238 | /* Index in the kmem_cache->memcg_params.memcg_caches array */ |
239 | int kmemcg_id; | |
567e9ab2 | 240 | enum memcg_kmem_state kmem_state; |
bc2791f8 | 241 | struct list_head kmem_caches; |
33398cf2 MH |
242 | #endif |
243 | ||
244 | int last_scanned_node; | |
245 | #if MAX_NUMNODES > 1 | |
246 | nodemask_t scan_nodes; | |
247 | atomic_t numainfo_events; | |
248 | atomic_t numainfo_updating; | |
249 | #endif | |
250 | ||
251 | #ifdef CONFIG_CGROUP_WRITEBACK | |
252 | struct list_head cgwb_list; | |
253 | struct wb_domain cgwb_domain; | |
254 | #endif | |
255 | ||
256 | /* List of events which userspace want to receive */ | |
257 | struct list_head event_list; | |
258 | spinlock_t event_list_lock; | |
259 | ||
260 | struct mem_cgroup_per_node *nodeinfo[0]; | |
261 | /* WARNING: nodeinfo must be the last member here */ | |
262 | }; | |
7d828602 JW |
263 | |
264 | extern struct mem_cgroup *root_mem_cgroup; | |
56161634 | 265 | |
23047a96 JW |
266 | static inline bool mem_cgroup_disabled(void) |
267 | { | |
268 | return !cgroup_subsys_enabled(memory_cgrp_subsys); | |
269 | } | |
270 | ||
31176c78 | 271 | static inline void mem_cgroup_event(struct mem_cgroup *memcg, |
df0e53d0 | 272 | enum memcg_event_item event) |
33398cf2 | 273 | { |
df0e53d0 | 274 | this_cpu_inc(memcg->stat->events[event]); |
472912a2 | 275 | cgroup_file_notify(&memcg->events_file); |
33398cf2 | 276 | } |
241994ed JW |
277 | |
278 | bool mem_cgroup_low(struct mem_cgroup *root, struct mem_cgroup *memcg); | |
279 | ||
00501b53 | 280 | int mem_cgroup_try_charge(struct page *page, struct mm_struct *mm, |
f627c2f5 KS |
281 | gfp_t gfp_mask, struct mem_cgroup **memcgp, |
282 | bool compound); | |
00501b53 | 283 | void mem_cgroup_commit_charge(struct page *page, struct mem_cgroup *memcg, |
f627c2f5 KS |
284 | bool lrucare, bool compound); |
285 | void mem_cgroup_cancel_charge(struct page *page, struct mem_cgroup *memcg, | |
286 | bool compound); | |
0a31bc97 | 287 | void mem_cgroup_uncharge(struct page *page); |
747db954 | 288 | void mem_cgroup_uncharge_list(struct list_head *page_list); |
569b846d | 289 | |
6a93ca8f | 290 | void mem_cgroup_migrate(struct page *oldpage, struct page *newpage); |
569b846d | 291 | |
ef8f2327 MG |
292 | static struct mem_cgroup_per_node * |
293 | mem_cgroup_nodeinfo(struct mem_cgroup *memcg, int nid) | |
55779ec7 | 294 | { |
ef8f2327 | 295 | return memcg->nodeinfo[nid]; |
55779ec7 JW |
296 | } |
297 | ||
298 | /** | |
a9dd0a83 MG |
299 | * mem_cgroup_lruvec - get the lru list vector for a node or a memcg zone |
300 | * @node: node of the wanted lruvec | |
55779ec7 JW |
301 | * @memcg: memcg of the wanted lruvec |
302 | * | |
a9dd0a83 MG |
303 | * Returns the lru list vector holding pages for a given @node or a given |
304 | * @memcg and @zone. This can be the node lruvec, if the memory controller | |
55779ec7 JW |
305 | * is disabled. |
306 | */ | |
a9dd0a83 | 307 | static inline struct lruvec *mem_cgroup_lruvec(struct pglist_data *pgdat, |
ef8f2327 | 308 | struct mem_cgroup *memcg) |
55779ec7 | 309 | { |
ef8f2327 | 310 | struct mem_cgroup_per_node *mz; |
55779ec7 JW |
311 | struct lruvec *lruvec; |
312 | ||
313 | if (mem_cgroup_disabled()) { | |
a9dd0a83 | 314 | lruvec = node_lruvec(pgdat); |
55779ec7 JW |
315 | goto out; |
316 | } | |
317 | ||
ef8f2327 | 318 | mz = mem_cgroup_nodeinfo(memcg, pgdat->node_id); |
55779ec7 JW |
319 | lruvec = &mz->lruvec; |
320 | out: | |
321 | /* | |
322 | * Since a node can be onlined after the mem_cgroup was created, | |
599d0c95 | 323 | * we have to be prepared to initialize lruvec->pgdat here; |
55779ec7 JW |
324 | * and if offlined then reonlined, we need to reinitialize it. |
325 | */ | |
ef8f2327 MG |
326 | if (unlikely(lruvec->pgdat != pgdat)) |
327 | lruvec->pgdat = pgdat; | |
55779ec7 JW |
328 | return lruvec; |
329 | } | |
330 | ||
599d0c95 | 331 | struct lruvec *mem_cgroup_page_lruvec(struct page *, struct pglist_data *); |
c9b0ed51 | 332 | |
2314b42d | 333 | bool task_in_mem_cgroup(struct task_struct *task, struct mem_cgroup *memcg); |
64219994 | 334 | struct mem_cgroup *mem_cgroup_from_task(struct task_struct *p); |
e993d905 | 335 | |
33398cf2 MH |
336 | static inline |
337 | struct mem_cgroup *mem_cgroup_from_css(struct cgroup_subsys_state *css){ | |
338 | return css ? container_of(css, struct mem_cgroup, css) : NULL; | |
339 | } | |
340 | ||
8e8ae645 JW |
341 | #define mem_cgroup_from_counter(counter, member) \ |
342 | container_of(counter, struct mem_cgroup, member) | |
343 | ||
33398cf2 MH |
344 | struct mem_cgroup *mem_cgroup_iter(struct mem_cgroup *, |
345 | struct mem_cgroup *, | |
346 | struct mem_cgroup_reclaim_cookie *); | |
347 | void mem_cgroup_iter_break(struct mem_cgroup *, struct mem_cgroup *); | |
7c5f64f8 VD |
348 | int mem_cgroup_scan_tasks(struct mem_cgroup *, |
349 | int (*)(struct task_struct *, void *), void *); | |
33398cf2 | 350 | |
23047a96 JW |
351 | static inline unsigned short mem_cgroup_id(struct mem_cgroup *memcg) |
352 | { | |
353 | if (mem_cgroup_disabled()) | |
354 | return 0; | |
355 | ||
73f576c0 | 356 | return memcg->id.id; |
23047a96 | 357 | } |
73f576c0 | 358 | struct mem_cgroup *mem_cgroup_from_id(unsigned short id); |
23047a96 | 359 | |
2262185c RG |
360 | static inline struct mem_cgroup *lruvec_memcg(struct lruvec *lruvec) |
361 | { | |
362 | struct mem_cgroup_per_node *mz; | |
363 | ||
364 | if (mem_cgroup_disabled()) | |
365 | return NULL; | |
366 | ||
367 | mz = container_of(lruvec, struct mem_cgroup_per_node, lruvec); | |
368 | return mz->memcg; | |
369 | } | |
370 | ||
8e8ae645 JW |
371 | /** |
372 | * parent_mem_cgroup - find the accounting parent of a memcg | |
373 | * @memcg: memcg whose parent to find | |
374 | * | |
375 | * Returns the parent memcg, or NULL if this is the root or the memory | |
376 | * controller is in legacy no-hierarchy mode. | |
377 | */ | |
378 | static inline struct mem_cgroup *parent_mem_cgroup(struct mem_cgroup *memcg) | |
379 | { | |
380 | if (!memcg->memory.parent) | |
381 | return NULL; | |
382 | return mem_cgroup_from_counter(memcg->memory.parent, memory); | |
383 | } | |
384 | ||
33398cf2 MH |
385 | static inline bool mem_cgroup_is_descendant(struct mem_cgroup *memcg, |
386 | struct mem_cgroup *root) | |
387 | { | |
388 | if (root == memcg) | |
389 | return true; | |
390 | if (!root->use_hierarchy) | |
391 | return false; | |
392 | return cgroup_is_descendant(memcg->css.cgroup, root->css.cgroup); | |
393 | } | |
e1aab161 | 394 | |
2314b42d JW |
395 | static inline bool mm_match_cgroup(struct mm_struct *mm, |
396 | struct mem_cgroup *memcg) | |
2e4d4091 | 397 | { |
587af308 | 398 | struct mem_cgroup *task_memcg; |
413918bb | 399 | bool match = false; |
c3ac9a8a | 400 | |
2e4d4091 | 401 | rcu_read_lock(); |
587af308 | 402 | task_memcg = mem_cgroup_from_task(rcu_dereference(mm->owner)); |
413918bb | 403 | if (task_memcg) |
2314b42d | 404 | match = mem_cgroup_is_descendant(task_memcg, memcg); |
2e4d4091 | 405 | rcu_read_unlock(); |
c3ac9a8a | 406 | return match; |
2e4d4091 | 407 | } |
8a9f3ccd | 408 | |
64219994 | 409 | struct cgroup_subsys_state *mem_cgroup_css_from_page(struct page *page); |
2fc04524 | 410 | ino_t page_cgroup_ino(struct page *page); |
d324236b | 411 | |
eb01aaab VD |
412 | static inline bool mem_cgroup_online(struct mem_cgroup *memcg) |
413 | { | |
414 | if (mem_cgroup_disabled()) | |
415 | return true; | |
416 | return !!(memcg->css.flags & CSS_ONLINE); | |
417 | } | |
418 | ||
58ae83db KH |
419 | /* |
420 | * For memory reclaim. | |
421 | */ | |
889976db | 422 | int mem_cgroup_select_victim_node(struct mem_cgroup *memcg); |
33398cf2 MH |
423 | |
424 | void mem_cgroup_update_lru_size(struct lruvec *lruvec, enum lru_list lru, | |
b4536f0c | 425 | int zid, int nr_pages); |
33398cf2 | 426 | |
0a6b76dd VD |
427 | unsigned long mem_cgroup_node_nr_lru_pages(struct mem_cgroup *memcg, |
428 | int nid, unsigned int lru_mask); | |
429 | ||
33398cf2 MH |
430 | static inline |
431 | unsigned long mem_cgroup_get_lru_size(struct lruvec *lruvec, enum lru_list lru) | |
432 | { | |
ef8f2327 | 433 | struct mem_cgroup_per_node *mz; |
b4536f0c MH |
434 | unsigned long nr_pages = 0; |
435 | int zid; | |
33398cf2 | 436 | |
ef8f2327 | 437 | mz = container_of(lruvec, struct mem_cgroup_per_node, lruvec); |
b4536f0c MH |
438 | for (zid = 0; zid < MAX_NR_ZONES; zid++) |
439 | nr_pages += mz->lru_zone_size[zid][lru]; | |
440 | return nr_pages; | |
441 | } | |
442 | ||
443 | static inline | |
444 | unsigned long mem_cgroup_get_zone_lru_size(struct lruvec *lruvec, | |
445 | enum lru_list lru, int zone_idx) | |
446 | { | |
447 | struct mem_cgroup_per_node *mz; | |
448 | ||
449 | mz = container_of(lruvec, struct mem_cgroup_per_node, lruvec); | |
450 | return mz->lru_zone_size[zone_idx][lru]; | |
33398cf2 MH |
451 | } |
452 | ||
b23afb93 TH |
453 | void mem_cgroup_handle_over_high(void); |
454 | ||
7c5f64f8 VD |
455 | unsigned long mem_cgroup_get_limit(struct mem_cgroup *memcg); |
456 | ||
64219994 MH |
457 | void mem_cgroup_print_oom_info(struct mem_cgroup *memcg, |
458 | struct task_struct *p); | |
58ae83db | 459 | |
49426420 | 460 | static inline void mem_cgroup_oom_enable(void) |
519e5247 | 461 | { |
626ebc41 TH |
462 | WARN_ON(current->memcg_may_oom); |
463 | current->memcg_may_oom = 1; | |
519e5247 JW |
464 | } |
465 | ||
49426420 | 466 | static inline void mem_cgroup_oom_disable(void) |
519e5247 | 467 | { |
626ebc41 TH |
468 | WARN_ON(!current->memcg_may_oom); |
469 | current->memcg_may_oom = 0; | |
519e5247 JW |
470 | } |
471 | ||
3812c8c8 JW |
472 | static inline bool task_in_memcg_oom(struct task_struct *p) |
473 | { | |
626ebc41 | 474 | return p->memcg_in_oom; |
3812c8c8 JW |
475 | } |
476 | ||
49426420 | 477 | bool mem_cgroup_oom_synchronize(bool wait); |
3812c8c8 | 478 | |
c255a458 | 479 | #ifdef CONFIG_MEMCG_SWAP |
c077719b KH |
480 | extern int do_swap_account; |
481 | #endif | |
f8d66542 | 482 | |
62cccb8c JW |
483 | void lock_page_memcg(struct page *page); |
484 | void unlock_page_memcg(struct page *page); | |
d7365e78 | 485 | |
ccda7f43 JW |
486 | static inline unsigned long memcg_page_state(struct mem_cgroup *memcg, |
487 | enum memcg_stat_item idx) | |
2a2e4885 JW |
488 | { |
489 | long val = 0; | |
490 | int cpu; | |
491 | ||
492 | for_each_possible_cpu(cpu) | |
493 | val += per_cpu(memcg->stat->count[idx], cpu); | |
494 | ||
495 | if (val < 0) | |
496 | val = 0; | |
497 | ||
498 | return val; | |
499 | } | |
500 | ||
ccda7f43 JW |
501 | static inline void mod_memcg_state(struct mem_cgroup *memcg, |
502 | enum memcg_stat_item idx, int val) | |
2a2e4885 JW |
503 | { |
504 | if (!mem_cgroup_disabled()) | |
505 | this_cpu_add(memcg->stat->count[idx], val); | |
506 | } | |
507 | ||
ccda7f43 JW |
508 | static inline void inc_memcg_state(struct mem_cgroup *memcg, |
509 | enum memcg_stat_item idx) | |
2a2e4885 | 510 | { |
ccda7f43 | 511 | mod_memcg_state(memcg, idx, 1); |
2a2e4885 JW |
512 | } |
513 | ||
ccda7f43 JW |
514 | static inline void dec_memcg_state(struct mem_cgroup *memcg, |
515 | enum memcg_stat_item idx) | |
2a2e4885 | 516 | { |
ccda7f43 | 517 | mod_memcg_state(memcg, idx, -1); |
2a2e4885 JW |
518 | } |
519 | ||
33398cf2 | 520 | /** |
ccda7f43 | 521 | * mod_memcg_page_state - update page state statistics |
62cccb8c | 522 | * @page: the page |
33398cf2 MH |
523 | * @idx: page state item to account |
524 | * @val: number of pages (positive or negative) | |
525 | * | |
fdf1cdb9 JW |
526 | * The @page must be locked or the caller must use lock_page_memcg() |
527 | * to prevent double accounting when the page is concurrently being | |
528 | * moved to another memcg: | |
81f8c3a4 | 529 | * |
fdf1cdb9 | 530 | * lock_page(page) or lock_page_memcg(page) |
81f8c3a4 | 531 | * if (TestClearPageState(page)) |
ccda7f43 | 532 | * mod_memcg_page_state(page, state, -1); |
fdf1cdb9 | 533 | * unlock_page(page) or unlock_page_memcg(page) |
2a2e4885 JW |
534 | * |
535 | * Kernel pages are an exception to this, since they'll never move. | |
33398cf2 | 536 | */ |
ccda7f43 JW |
537 | static inline void mod_memcg_page_state(struct page *page, |
538 | enum memcg_stat_item idx, int val) | |
33398cf2 | 539 | { |
62cccb8c | 540 | if (page->mem_cgroup) |
ccda7f43 | 541 | mod_memcg_state(page->mem_cgroup, idx, val); |
33398cf2 MH |
542 | } |
543 | ||
ccda7f43 JW |
544 | static inline void inc_memcg_page_state(struct page *page, |
545 | enum memcg_stat_item idx) | |
2a7106f2 | 546 | { |
ccda7f43 | 547 | mod_memcg_page_state(page, idx, 1); |
2a7106f2 GT |
548 | } |
549 | ||
ccda7f43 JW |
550 | static inline void dec_memcg_page_state(struct page *page, |
551 | enum memcg_stat_item idx) | |
2a7106f2 | 552 | { |
ccda7f43 | 553 | mod_memcg_page_state(page, idx, -1); |
2a7106f2 GT |
554 | } |
555 | ||
ef8f2327 | 556 | unsigned long mem_cgroup_soft_limit_reclaim(pg_data_t *pgdat, int order, |
0608f43d AM |
557 | gfp_t gfp_mask, |
558 | unsigned long *total_scanned); | |
a63d83f4 | 559 | |
2262185c RG |
560 | static inline void count_memcg_events(struct mem_cgroup *memcg, |
561 | enum vm_event_item idx, | |
562 | unsigned long count) | |
563 | { | |
564 | if (!mem_cgroup_disabled()) | |
565 | this_cpu_add(memcg->stat->events[idx], count); | |
566 | } | |
567 | ||
568 | static inline void count_memcg_page_event(struct page *page, | |
569 | enum memcg_stat_item idx) | |
570 | { | |
571 | if (page->mem_cgroup) | |
572 | count_memcg_events(page->mem_cgroup, idx, 1); | |
573 | } | |
574 | ||
575 | static inline void count_memcg_event_mm(struct mm_struct *mm, | |
576 | enum vm_event_item idx) | |
68ae564b | 577 | { |
33398cf2 MH |
578 | struct mem_cgroup *memcg; |
579 | ||
68ae564b DR |
580 | if (mem_cgroup_disabled()) |
581 | return; | |
33398cf2 MH |
582 | |
583 | rcu_read_lock(); | |
584 | memcg = mem_cgroup_from_task(rcu_dereference(mm->owner)); | |
df0e53d0 JW |
585 | if (likely(memcg)) |
586 | this_cpu_inc(memcg->stat->events[idx]); | |
33398cf2 | 587 | rcu_read_unlock(); |
68ae564b | 588 | } |
ca3e0214 | 589 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE |
e94c8a9c | 590 | void mem_cgroup_split_huge_fixup(struct page *head); |
ca3e0214 KH |
591 | #endif |
592 | ||
c255a458 | 593 | #else /* CONFIG_MEMCG */ |
23047a96 JW |
594 | |
595 | #define MEM_CGROUP_ID_SHIFT 0 | |
596 | #define MEM_CGROUP_ID_MAX 0 | |
597 | ||
7a81b88c KH |
598 | struct mem_cgroup; |
599 | ||
23047a96 JW |
600 | static inline bool mem_cgroup_disabled(void) |
601 | { | |
602 | return true; | |
603 | } | |
604 | ||
31176c78 | 605 | static inline void mem_cgroup_event(struct mem_cgroup *memcg, |
df0e53d0 | 606 | enum memcg_event_item event) |
241994ed JW |
607 | { |
608 | } | |
609 | ||
610 | static inline bool mem_cgroup_low(struct mem_cgroup *root, | |
611 | struct mem_cgroup *memcg) | |
612 | { | |
613 | return false; | |
614 | } | |
615 | ||
00501b53 JW |
616 | static inline int mem_cgroup_try_charge(struct page *page, struct mm_struct *mm, |
617 | gfp_t gfp_mask, | |
f627c2f5 KS |
618 | struct mem_cgroup **memcgp, |
619 | bool compound) | |
7a81b88c | 620 | { |
00501b53 | 621 | *memcgp = NULL; |
7a81b88c KH |
622 | return 0; |
623 | } | |
624 | ||
00501b53 JW |
625 | static inline void mem_cgroup_commit_charge(struct page *page, |
626 | struct mem_cgroup *memcg, | |
f627c2f5 | 627 | bool lrucare, bool compound) |
7a81b88c KH |
628 | { |
629 | } | |
630 | ||
00501b53 | 631 | static inline void mem_cgroup_cancel_charge(struct page *page, |
f627c2f5 KS |
632 | struct mem_cgroup *memcg, |
633 | bool compound) | |
7a81b88c KH |
634 | { |
635 | } | |
636 | ||
0a31bc97 | 637 | static inline void mem_cgroup_uncharge(struct page *page) |
569b846d KH |
638 | { |
639 | } | |
640 | ||
747db954 | 641 | static inline void mem_cgroup_uncharge_list(struct list_head *page_list) |
8a9f3ccd BS |
642 | { |
643 | } | |
644 | ||
6a93ca8f | 645 | static inline void mem_cgroup_migrate(struct page *old, struct page *new) |
69029cd5 KH |
646 | { |
647 | } | |
648 | ||
a9dd0a83 | 649 | static inline struct lruvec *mem_cgroup_lruvec(struct pglist_data *pgdat, |
ef8f2327 | 650 | struct mem_cgroup *memcg) |
08e552c6 | 651 | { |
a9dd0a83 | 652 | return node_lruvec(pgdat); |
08e552c6 KH |
653 | } |
654 | ||
fa9add64 | 655 | static inline struct lruvec *mem_cgroup_page_lruvec(struct page *page, |
599d0c95 | 656 | struct pglist_data *pgdat) |
66e1707b | 657 | { |
599d0c95 | 658 | return &pgdat->lruvec; |
66e1707b BS |
659 | } |
660 | ||
587af308 | 661 | static inline bool mm_match_cgroup(struct mm_struct *mm, |
c0ff4b85 | 662 | struct mem_cgroup *memcg) |
bed7161a | 663 | { |
587af308 | 664 | return true; |
bed7161a BS |
665 | } |
666 | ||
ffbdccf5 DR |
667 | static inline bool task_in_mem_cgroup(struct task_struct *task, |
668 | const struct mem_cgroup *memcg) | |
4c4a2214 | 669 | { |
ffbdccf5 | 670 | return true; |
4c4a2214 DR |
671 | } |
672 | ||
5660048c JW |
673 | static inline struct mem_cgroup * |
674 | mem_cgroup_iter(struct mem_cgroup *root, | |
675 | struct mem_cgroup *prev, | |
676 | struct mem_cgroup_reclaim_cookie *reclaim) | |
677 | { | |
678 | return NULL; | |
679 | } | |
680 | ||
681 | static inline void mem_cgroup_iter_break(struct mem_cgroup *root, | |
682 | struct mem_cgroup *prev) | |
683 | { | |
684 | } | |
685 | ||
7c5f64f8 VD |
686 | static inline int mem_cgroup_scan_tasks(struct mem_cgroup *memcg, |
687 | int (*fn)(struct task_struct *, void *), void *arg) | |
688 | { | |
689 | return 0; | |
690 | } | |
691 | ||
23047a96 | 692 | static inline unsigned short mem_cgroup_id(struct mem_cgroup *memcg) |
f8d66542 | 693 | { |
23047a96 JW |
694 | return 0; |
695 | } | |
696 | ||
697 | static inline struct mem_cgroup *mem_cgroup_from_id(unsigned short id) | |
698 | { | |
699 | WARN_ON_ONCE(id); | |
700 | /* XXX: This should always return root_mem_cgroup */ | |
701 | return NULL; | |
f8d66542 | 702 | } |
a636b327 | 703 | |
2262185c RG |
704 | static inline struct mem_cgroup *lruvec_memcg(struct lruvec *lruvec) |
705 | { | |
706 | return NULL; | |
707 | } | |
708 | ||
eb01aaab | 709 | static inline bool mem_cgroup_online(struct mem_cgroup *memcg) |
14797e23 | 710 | { |
13308ca9 | 711 | return true; |
14797e23 KM |
712 | } |
713 | ||
a3d8e054 | 714 | static inline unsigned long |
4d7dcca2 | 715 | mem_cgroup_get_lru_size(struct lruvec *lruvec, enum lru_list lru) |
a3d8e054 KM |
716 | { |
717 | return 0; | |
718 | } | |
b4536f0c MH |
719 | static inline |
720 | unsigned long mem_cgroup_get_zone_lru_size(struct lruvec *lruvec, | |
721 | enum lru_list lru, int zone_idx) | |
722 | { | |
723 | return 0; | |
724 | } | |
a3d8e054 | 725 | |
0a6b76dd VD |
726 | static inline unsigned long |
727 | mem_cgroup_node_nr_lru_pages(struct mem_cgroup *memcg, | |
728 | int nid, unsigned int lru_mask) | |
729 | { | |
730 | return 0; | |
731 | } | |
732 | ||
7c5f64f8 VD |
733 | static inline unsigned long mem_cgroup_get_limit(struct mem_cgroup *memcg) |
734 | { | |
735 | return 0; | |
736 | } | |
737 | ||
e222432b BS |
738 | static inline void |
739 | mem_cgroup_print_oom_info(struct mem_cgroup *memcg, struct task_struct *p) | |
740 | { | |
741 | } | |
742 | ||
62cccb8c | 743 | static inline void lock_page_memcg(struct page *page) |
89c06bd5 KH |
744 | { |
745 | } | |
746 | ||
62cccb8c | 747 | static inline void unlock_page_memcg(struct page *page) |
89c06bd5 KH |
748 | { |
749 | } | |
750 | ||
b23afb93 TH |
751 | static inline void mem_cgroup_handle_over_high(void) |
752 | { | |
753 | } | |
754 | ||
49426420 | 755 | static inline void mem_cgroup_oom_enable(void) |
519e5247 JW |
756 | { |
757 | } | |
758 | ||
49426420 | 759 | static inline void mem_cgroup_oom_disable(void) |
519e5247 JW |
760 | { |
761 | } | |
762 | ||
3812c8c8 JW |
763 | static inline bool task_in_memcg_oom(struct task_struct *p) |
764 | { | |
765 | return false; | |
766 | } | |
767 | ||
49426420 | 768 | static inline bool mem_cgroup_oom_synchronize(bool wait) |
3812c8c8 JW |
769 | { |
770 | return false; | |
771 | } | |
772 | ||
ccda7f43 JW |
773 | static inline unsigned long memcg_page_state(struct mem_cgroup *memcg, |
774 | enum memcg_stat_item idx) | |
2a2e4885 JW |
775 | { |
776 | return 0; | |
777 | } | |
778 | ||
ccda7f43 JW |
779 | static inline void mod_memcg_state(struct mem_cgroup *memcg, |
780 | enum memcg_stat_item idx, | |
781 | int nr) | |
2a2e4885 JW |
782 | { |
783 | } | |
784 | ||
ccda7f43 JW |
785 | static inline void inc_memcg_state(struct mem_cgroup *memcg, |
786 | enum memcg_stat_item idx) | |
2a2e4885 JW |
787 | { |
788 | } | |
789 | ||
ccda7f43 JW |
790 | static inline void dec_memcg_state(struct mem_cgroup *memcg, |
791 | enum memcg_stat_item idx) | |
2a2e4885 JW |
792 | { |
793 | } | |
794 | ||
ccda7f43 JW |
795 | static inline void mod_memcg_page_state(struct page *page, |
796 | enum memcg_stat_item idx, | |
797 | int nr) | |
553af430 JW |
798 | { |
799 | } | |
800 | ||
ccda7f43 JW |
801 | static inline void inc_memcg_page_state(struct page *page, |
802 | enum memcg_stat_item idx) | |
2a7106f2 GT |
803 | { |
804 | } | |
805 | ||
ccda7f43 JW |
806 | static inline void dec_memcg_page_state(struct page *page, |
807 | enum memcg_stat_item idx) | |
d69b042f BS |
808 | { |
809 | } | |
810 | ||
4e416953 | 811 | static inline |
ef8f2327 | 812 | unsigned long mem_cgroup_soft_limit_reclaim(pg_data_t *pgdat, int order, |
0608f43d AM |
813 | gfp_t gfp_mask, |
814 | unsigned long *total_scanned) | |
4e416953 | 815 | { |
0608f43d | 816 | return 0; |
4e416953 BS |
817 | } |
818 | ||
e94c8a9c | 819 | static inline void mem_cgroup_split_huge_fixup(struct page *head) |
ca3e0214 KH |
820 | { |
821 | } | |
822 | ||
2262185c RG |
823 | static inline void count_memcg_events(struct mem_cgroup *memcg, |
824 | enum vm_event_item idx, | |
825 | unsigned long count) | |
826 | { | |
827 | } | |
828 | ||
829 | static inline void count_memcg_page_event(struct page *page, | |
830 | enum memcg_stat_item idx) | |
831 | { | |
832 | } | |
833 | ||
456f998e | 834 | static inline |
2262185c | 835 | void count_memcg_event_mm(struct mm_struct *mm, enum vm_event_item idx) |
456f998e YH |
836 | { |
837 | } | |
c255a458 | 838 | #endif /* CONFIG_MEMCG */ |
78fb7466 | 839 | |
52ebea74 | 840 | #ifdef CONFIG_CGROUP_WRITEBACK |
841710aa | 841 | |
52ebea74 | 842 | struct list_head *mem_cgroup_cgwb_list(struct mem_cgroup *memcg); |
841710aa | 843 | struct wb_domain *mem_cgroup_wb_domain(struct bdi_writeback *wb); |
c5edf9cd TH |
844 | void mem_cgroup_wb_stats(struct bdi_writeback *wb, unsigned long *pfilepages, |
845 | unsigned long *pheadroom, unsigned long *pdirty, | |
846 | unsigned long *pwriteback); | |
841710aa TH |
847 | |
848 | #else /* CONFIG_CGROUP_WRITEBACK */ | |
849 | ||
850 | static inline struct wb_domain *mem_cgroup_wb_domain(struct bdi_writeback *wb) | |
851 | { | |
852 | return NULL; | |
853 | } | |
854 | ||
c2aa723a | 855 | static inline void mem_cgroup_wb_stats(struct bdi_writeback *wb, |
c5edf9cd TH |
856 | unsigned long *pfilepages, |
857 | unsigned long *pheadroom, | |
c2aa723a TH |
858 | unsigned long *pdirty, |
859 | unsigned long *pwriteback) | |
860 | { | |
861 | } | |
862 | ||
841710aa | 863 | #endif /* CONFIG_CGROUP_WRITEBACK */ |
52ebea74 | 864 | |
e1aab161 | 865 | struct sock; |
baac50bb JW |
866 | bool mem_cgroup_charge_skmem(struct mem_cgroup *memcg, unsigned int nr_pages); |
867 | void mem_cgroup_uncharge_skmem(struct mem_cgroup *memcg, unsigned int nr_pages); | |
d886f4e4 | 868 | #ifdef CONFIG_MEMCG |
ef12947c JW |
869 | extern struct static_key_false memcg_sockets_enabled_key; |
870 | #define mem_cgroup_sockets_enabled static_branch_unlikely(&memcg_sockets_enabled_key) | |
2d758073 JW |
871 | void mem_cgroup_sk_alloc(struct sock *sk); |
872 | void mem_cgroup_sk_free(struct sock *sk); | |
baac50bb | 873 | static inline bool mem_cgroup_under_socket_pressure(struct mem_cgroup *memcg) |
e805605c | 874 | { |
0db15298 | 875 | if (!cgroup_subsys_on_dfl(memory_cgrp_subsys) && memcg->tcpmem_pressure) |
8e8ae645 | 876 | return true; |
8e8ae645 JW |
877 | do { |
878 | if (time_before(jiffies, memcg->socket_pressure)) | |
879 | return true; | |
880 | } while ((memcg = parent_mem_cgroup(memcg))); | |
881 | return false; | |
e805605c JW |
882 | } |
883 | #else | |
80e95fe0 | 884 | #define mem_cgroup_sockets_enabled 0 |
2d758073 JW |
885 | static inline void mem_cgroup_sk_alloc(struct sock *sk) { }; |
886 | static inline void mem_cgroup_sk_free(struct sock *sk) { }; | |
baac50bb | 887 | static inline bool mem_cgroup_under_socket_pressure(struct mem_cgroup *memcg) |
e805605c JW |
888 | { |
889 | return false; | |
890 | } | |
891 | #endif | |
7ae1e1d0 | 892 | |
45264778 VD |
893 | struct kmem_cache *memcg_kmem_get_cache(struct kmem_cache *cachep); |
894 | void memcg_kmem_put_cache(struct kmem_cache *cachep); | |
895 | int memcg_kmem_charge_memcg(struct page *page, gfp_t gfp, int order, | |
896 | struct mem_cgroup *memcg); | |
897 | int memcg_kmem_charge(struct page *page, gfp_t gfp, int order); | |
898 | void memcg_kmem_uncharge(struct page *page, int order); | |
899 | ||
127424c8 | 900 | #if defined(CONFIG_MEMCG) && !defined(CONFIG_SLOB) |
ef12947c | 901 | extern struct static_key_false memcg_kmem_enabled_key; |
17cc4dfe | 902 | extern struct workqueue_struct *memcg_kmem_cache_wq; |
749c5415 | 903 | |
dbcf73e2 | 904 | extern int memcg_nr_cache_ids; |
64219994 MH |
905 | void memcg_get_cache_ids(void); |
906 | void memcg_put_cache_ids(void); | |
ebe945c2 GC |
907 | |
908 | /* | |
909 | * Helper macro to loop through all memcg-specific caches. Callers must still | |
910 | * check if the cache is valid (it is either valid or NULL). | |
911 | * the slab_mutex must be held when looping through those caches | |
912 | */ | |
749c5415 | 913 | #define for_each_memcg_cache_index(_idx) \ |
dbcf73e2 | 914 | for ((_idx) = 0; (_idx) < memcg_nr_cache_ids; (_idx)++) |
749c5415 | 915 | |
7ae1e1d0 GC |
916 | static inline bool memcg_kmem_enabled(void) |
917 | { | |
ef12947c | 918 | return static_branch_unlikely(&memcg_kmem_enabled_key); |
7ae1e1d0 GC |
919 | } |
920 | ||
33398cf2 | 921 | /* |
9f706d68 | 922 | * helper for accessing a memcg's index. It will be used as an index in the |
33398cf2 MH |
923 | * child cache array in kmem_cache, and also to derive its name. This function |
924 | * will return -1 when this is not a kmem-limited memcg. | |
925 | */ | |
926 | static inline int memcg_cache_id(struct mem_cgroup *memcg) | |
927 | { | |
928 | return memcg ? memcg->kmemcg_id : -1; | |
929 | } | |
5722d094 | 930 | |
27ee57c9 VD |
931 | /** |
932 | * memcg_kmem_update_page_stat - update kmem page state statistics | |
933 | * @page: the page | |
934 | * @idx: page state item to account | |
935 | * @val: number of pages (positive or negative) | |
936 | */ | |
937 | static inline void memcg_kmem_update_page_stat(struct page *page, | |
71cd3113 | 938 | enum memcg_stat_item idx, int val) |
27ee57c9 VD |
939 | { |
940 | if (memcg_kmem_enabled() && page->mem_cgroup) | |
941 | this_cpu_add(page->mem_cgroup->stat->count[idx], val); | |
942 | } | |
943 | ||
7ae1e1d0 | 944 | #else |
749c5415 GC |
945 | #define for_each_memcg_cache_index(_idx) \ |
946 | for (; NULL; ) | |
947 | ||
b9ce5ef4 GC |
948 | static inline bool memcg_kmem_enabled(void) |
949 | { | |
950 | return false; | |
951 | } | |
952 | ||
2633d7a0 GC |
953 | static inline int memcg_cache_id(struct mem_cgroup *memcg) |
954 | { | |
955 | return -1; | |
956 | } | |
957 | ||
05257a1a VD |
958 | static inline void memcg_get_cache_ids(void) |
959 | { | |
960 | } | |
961 | ||
962 | static inline void memcg_put_cache_ids(void) | |
963 | { | |
964 | } | |
965 | ||
27ee57c9 | 966 | static inline void memcg_kmem_update_page_stat(struct page *page, |
71cd3113 | 967 | enum memcg_stat_item idx, int val) |
27ee57c9 VD |
968 | { |
969 | } | |
127424c8 JW |
970 | #endif /* CONFIG_MEMCG && !CONFIG_SLOB */ |
971 | ||
8cdea7c0 | 972 | #endif /* _LINUX_MEMCONTROL_H */ |