]>
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, | |
b2807f07 | 47 | MEMCG_NR_STAT, |
2a7106f2 GT |
48 | }; |
49 | ||
71cd3113 JW |
50 | /* Cgroup-specific events, on top of universal VM events */ |
51 | enum memcg_event_item { | |
52 | MEMCG_LOW = NR_VM_EVENT_ITEMS, | |
53 | MEMCG_HIGH, | |
54 | MEMCG_MAX, | |
55 | MEMCG_OOM, | |
56 | MEMCG_NR_EVENTS, | |
57 | }; | |
58 | ||
5660048c | 59 | struct mem_cgroup_reclaim_cookie { |
ef8f2327 | 60 | pg_data_t *pgdat; |
5660048c JW |
61 | int priority; |
62 | unsigned int generation; | |
63 | }; | |
64 | ||
71cd3113 JW |
65 | #ifdef CONFIG_MEMCG |
66 | ||
67 | #define MEM_CGROUP_ID_SHIFT 16 | |
68 | #define MEM_CGROUP_ID_MAX USHRT_MAX | |
69 | ||
70 | struct mem_cgroup_id { | |
71 | int id; | |
72 | atomic_t ref; | |
73 | }; | |
74 | ||
33398cf2 MH |
75 | /* |
76 | * Per memcg event counter is incremented at every pagein/pageout. With THP, | |
77 | * it will be incremated by the number of pages. This counter is used for | |
78 | * for trigger some periodic events. This is straightforward and better | |
79 | * than using jiffies etc. to handle periodic memcg event. | |
80 | */ | |
81 | enum mem_cgroup_events_target { | |
82 | MEM_CGROUP_TARGET_THRESH, | |
83 | MEM_CGROUP_TARGET_SOFTLIMIT, | |
84 | MEM_CGROUP_TARGET_NUMAINFO, | |
85 | MEM_CGROUP_NTARGETS, | |
86 | }; | |
87 | ||
33398cf2 | 88 | struct mem_cgroup_stat_cpu { |
b2807f07 | 89 | long count[MEMCG_NR_STAT]; |
33398cf2 MH |
90 | unsigned long events[MEMCG_NR_EVENTS]; |
91 | unsigned long nr_page_events; | |
92 | unsigned long targets[MEM_CGROUP_NTARGETS]; | |
93 | }; | |
94 | ||
95 | struct mem_cgroup_reclaim_iter { | |
96 | struct mem_cgroup *position; | |
97 | /* scan generation, increased every round-trip */ | |
98 | unsigned int generation; | |
99 | }; | |
100 | ||
101 | /* | |
102 | * per-zone information in memory controller. | |
103 | */ | |
ef8f2327 | 104 | struct mem_cgroup_per_node { |
33398cf2 | 105 | struct lruvec lruvec; |
b4536f0c | 106 | unsigned long lru_zone_size[MAX_NR_ZONES][NR_LRU_LISTS]; |
33398cf2 MH |
107 | |
108 | struct mem_cgroup_reclaim_iter iter[DEF_PRIORITY + 1]; | |
109 | ||
110 | struct rb_node tree_node; /* RB tree node */ | |
111 | unsigned long usage_in_excess;/* Set to the value by which */ | |
112 | /* the soft limit is exceeded*/ | |
113 | bool on_tree; | |
114 | struct mem_cgroup *memcg; /* Back pointer, we cannot */ | |
115 | /* use container_of */ | |
116 | }; | |
117 | ||
33398cf2 MH |
118 | struct mem_cgroup_threshold { |
119 | struct eventfd_ctx *eventfd; | |
120 | unsigned long threshold; | |
121 | }; | |
122 | ||
123 | /* For threshold */ | |
124 | struct mem_cgroup_threshold_ary { | |
125 | /* An array index points to threshold just below or equal to usage. */ | |
126 | int current_threshold; | |
127 | /* Size of entries[] */ | |
128 | unsigned int size; | |
129 | /* Array of thresholds */ | |
130 | struct mem_cgroup_threshold entries[0]; | |
131 | }; | |
132 | ||
133 | struct mem_cgroup_thresholds { | |
134 | /* Primary thresholds array */ | |
135 | struct mem_cgroup_threshold_ary *primary; | |
136 | /* | |
137 | * Spare threshold array. | |
138 | * This is needed to make mem_cgroup_unregister_event() "never fail". | |
139 | * It must be able to store at least primary->size - 1 entries. | |
140 | */ | |
141 | struct mem_cgroup_threshold_ary *spare; | |
142 | }; | |
143 | ||
567e9ab2 JW |
144 | enum memcg_kmem_state { |
145 | KMEM_NONE, | |
146 | KMEM_ALLOCATED, | |
147 | KMEM_ONLINE, | |
148 | }; | |
149 | ||
33398cf2 MH |
150 | /* |
151 | * The memory controller data structure. The memory controller controls both | |
152 | * page cache and RSS per cgroup. We would eventually like to provide | |
153 | * statistics based on the statistics developed by Rik Van Riel for clock-pro, | |
154 | * to help the administrator determine what knobs to tune. | |
155 | */ | |
156 | struct mem_cgroup { | |
157 | struct cgroup_subsys_state css; | |
158 | ||
73f576c0 JW |
159 | /* Private memcg ID. Used to ID objects that outlive the cgroup */ |
160 | struct mem_cgroup_id id; | |
161 | ||
33398cf2 MH |
162 | /* Accounted resources */ |
163 | struct page_counter memory; | |
37e84351 | 164 | struct page_counter swap; |
0db15298 JW |
165 | |
166 | /* Legacy consumer-oriented counters */ | |
33398cf2 MH |
167 | struct page_counter memsw; |
168 | struct page_counter kmem; | |
0db15298 | 169 | struct page_counter tcpmem; |
33398cf2 MH |
170 | |
171 | /* Normal memory consumption range */ | |
172 | unsigned long low; | |
173 | unsigned long high; | |
174 | ||
f7e1cb6e JW |
175 | /* Range enforcement for interrupt charges */ |
176 | struct work_struct high_work; | |
177 | ||
33398cf2 MH |
178 | unsigned long soft_limit; |
179 | ||
180 | /* vmpressure notifications */ | |
181 | struct vmpressure vmpressure; | |
182 | ||
33398cf2 MH |
183 | /* |
184 | * Should the accounting and control be hierarchical, per subtree? | |
185 | */ | |
186 | bool use_hierarchy; | |
187 | ||
188 | /* protected by memcg_oom_lock */ | |
189 | bool oom_lock; | |
190 | int under_oom; | |
191 | ||
192 | int swappiness; | |
193 | /* OOM-Killer disable */ | |
194 | int oom_kill_disable; | |
195 | ||
472912a2 TH |
196 | /* handle for "memory.events" */ |
197 | struct cgroup_file events_file; | |
198 | ||
33398cf2 MH |
199 | /* protect arrays of thresholds */ |
200 | struct mutex thresholds_lock; | |
201 | ||
202 | /* thresholds for memory usage. RCU-protected */ | |
203 | struct mem_cgroup_thresholds thresholds; | |
204 | ||
205 | /* thresholds for mem+swap usage. RCU-protected */ | |
206 | struct mem_cgroup_thresholds memsw_thresholds; | |
207 | ||
208 | /* For oom notifier event fd */ | |
209 | struct list_head oom_notify; | |
210 | ||
211 | /* | |
212 | * Should we move charges of a task when a task is moved into this | |
213 | * mem_cgroup ? And what type of charges should we move ? | |
214 | */ | |
215 | unsigned long move_charge_at_immigrate; | |
216 | /* | |
217 | * set > 0 if pages under this cgroup are moving to other cgroup. | |
218 | */ | |
219 | atomic_t moving_account; | |
220 | /* taken only while moving_account > 0 */ | |
221 | spinlock_t move_lock; | |
222 | struct task_struct *move_lock_task; | |
223 | unsigned long move_lock_flags; | |
224 | /* | |
225 | * percpu counter. | |
226 | */ | |
227 | struct mem_cgroup_stat_cpu __percpu *stat; | |
33398cf2 | 228 | |
d886f4e4 JW |
229 | unsigned long socket_pressure; |
230 | ||
231 | /* Legacy tcp memory accounting */ | |
0db15298 JW |
232 | bool tcpmem_active; |
233 | int tcpmem_pressure; | |
d886f4e4 | 234 | |
127424c8 | 235 | #ifndef CONFIG_SLOB |
33398cf2 MH |
236 | /* Index in the kmem_cache->memcg_params.memcg_caches array */ |
237 | int kmemcg_id; | |
567e9ab2 | 238 | enum memcg_kmem_state kmem_state; |
bc2791f8 | 239 | struct list_head kmem_caches; |
33398cf2 MH |
240 | #endif |
241 | ||
242 | int last_scanned_node; | |
243 | #if MAX_NUMNODES > 1 | |
244 | nodemask_t scan_nodes; | |
245 | atomic_t numainfo_events; | |
246 | atomic_t numainfo_updating; | |
247 | #endif | |
248 | ||
249 | #ifdef CONFIG_CGROUP_WRITEBACK | |
250 | struct list_head cgwb_list; | |
251 | struct wb_domain cgwb_domain; | |
252 | #endif | |
253 | ||
254 | /* List of events which userspace want to receive */ | |
255 | struct list_head event_list; | |
256 | spinlock_t event_list_lock; | |
257 | ||
258 | struct mem_cgroup_per_node *nodeinfo[0]; | |
259 | /* WARNING: nodeinfo must be the last member here */ | |
260 | }; | |
7d828602 JW |
261 | |
262 | extern struct mem_cgroup *root_mem_cgroup; | |
56161634 | 263 | |
23047a96 JW |
264 | static inline bool mem_cgroup_disabled(void) |
265 | { | |
266 | return !cgroup_subsys_enabled(memory_cgrp_subsys); | |
267 | } | |
268 | ||
31176c78 | 269 | static inline void mem_cgroup_event(struct mem_cgroup *memcg, |
df0e53d0 | 270 | enum memcg_event_item event) |
33398cf2 | 271 | { |
df0e53d0 | 272 | this_cpu_inc(memcg->stat->events[event]); |
472912a2 | 273 | cgroup_file_notify(&memcg->events_file); |
33398cf2 | 274 | } |
241994ed JW |
275 | |
276 | bool mem_cgroup_low(struct mem_cgroup *root, struct mem_cgroup *memcg); | |
277 | ||
00501b53 | 278 | int mem_cgroup_try_charge(struct page *page, struct mm_struct *mm, |
f627c2f5 KS |
279 | gfp_t gfp_mask, struct mem_cgroup **memcgp, |
280 | bool compound); | |
00501b53 | 281 | void mem_cgroup_commit_charge(struct page *page, struct mem_cgroup *memcg, |
f627c2f5 KS |
282 | bool lrucare, bool compound); |
283 | void mem_cgroup_cancel_charge(struct page *page, struct mem_cgroup *memcg, | |
284 | bool compound); | |
0a31bc97 | 285 | void mem_cgroup_uncharge(struct page *page); |
747db954 | 286 | void mem_cgroup_uncharge_list(struct list_head *page_list); |
569b846d | 287 | |
6a93ca8f | 288 | void mem_cgroup_migrate(struct page *oldpage, struct page *newpage); |
569b846d | 289 | |
ef8f2327 MG |
290 | static struct mem_cgroup_per_node * |
291 | mem_cgroup_nodeinfo(struct mem_cgroup *memcg, int nid) | |
55779ec7 | 292 | { |
ef8f2327 | 293 | return memcg->nodeinfo[nid]; |
55779ec7 JW |
294 | } |
295 | ||
296 | /** | |
a9dd0a83 MG |
297 | * mem_cgroup_lruvec - get the lru list vector for a node or a memcg zone |
298 | * @node: node of the wanted lruvec | |
55779ec7 JW |
299 | * @memcg: memcg of the wanted lruvec |
300 | * | |
a9dd0a83 MG |
301 | * Returns the lru list vector holding pages for a given @node or a given |
302 | * @memcg and @zone. This can be the node lruvec, if the memory controller | |
55779ec7 JW |
303 | * is disabled. |
304 | */ | |
a9dd0a83 | 305 | static inline struct lruvec *mem_cgroup_lruvec(struct pglist_data *pgdat, |
ef8f2327 | 306 | struct mem_cgroup *memcg) |
55779ec7 | 307 | { |
ef8f2327 | 308 | struct mem_cgroup_per_node *mz; |
55779ec7 JW |
309 | struct lruvec *lruvec; |
310 | ||
311 | if (mem_cgroup_disabled()) { | |
a9dd0a83 | 312 | lruvec = node_lruvec(pgdat); |
55779ec7 JW |
313 | goto out; |
314 | } | |
315 | ||
ef8f2327 | 316 | mz = mem_cgroup_nodeinfo(memcg, pgdat->node_id); |
55779ec7 JW |
317 | lruvec = &mz->lruvec; |
318 | out: | |
319 | /* | |
320 | * Since a node can be onlined after the mem_cgroup was created, | |
599d0c95 | 321 | * we have to be prepared to initialize lruvec->pgdat here; |
55779ec7 JW |
322 | * and if offlined then reonlined, we need to reinitialize it. |
323 | */ | |
ef8f2327 MG |
324 | if (unlikely(lruvec->pgdat != pgdat)) |
325 | lruvec->pgdat = pgdat; | |
55779ec7 JW |
326 | return lruvec; |
327 | } | |
328 | ||
599d0c95 | 329 | struct lruvec *mem_cgroup_page_lruvec(struct page *, struct pglist_data *); |
c9b0ed51 | 330 | |
2314b42d | 331 | bool task_in_mem_cgroup(struct task_struct *task, struct mem_cgroup *memcg); |
64219994 | 332 | struct mem_cgroup *mem_cgroup_from_task(struct task_struct *p); |
e993d905 | 333 | |
33398cf2 MH |
334 | static inline |
335 | struct mem_cgroup *mem_cgroup_from_css(struct cgroup_subsys_state *css){ | |
336 | return css ? container_of(css, struct mem_cgroup, css) : NULL; | |
337 | } | |
338 | ||
8e8ae645 JW |
339 | #define mem_cgroup_from_counter(counter, member) \ |
340 | container_of(counter, struct mem_cgroup, member) | |
341 | ||
33398cf2 MH |
342 | struct mem_cgroup *mem_cgroup_iter(struct mem_cgroup *, |
343 | struct mem_cgroup *, | |
344 | struct mem_cgroup_reclaim_cookie *); | |
345 | void mem_cgroup_iter_break(struct mem_cgroup *, struct mem_cgroup *); | |
7c5f64f8 VD |
346 | int mem_cgroup_scan_tasks(struct mem_cgroup *, |
347 | int (*)(struct task_struct *, void *), void *); | |
33398cf2 | 348 | |
23047a96 JW |
349 | static inline unsigned short mem_cgroup_id(struct mem_cgroup *memcg) |
350 | { | |
351 | if (mem_cgroup_disabled()) | |
352 | return 0; | |
353 | ||
73f576c0 | 354 | return memcg->id.id; |
23047a96 | 355 | } |
73f576c0 | 356 | struct mem_cgroup *mem_cgroup_from_id(unsigned short id); |
23047a96 | 357 | |
2262185c RG |
358 | static inline struct mem_cgroup *lruvec_memcg(struct lruvec *lruvec) |
359 | { | |
360 | struct mem_cgroup_per_node *mz; | |
361 | ||
362 | if (mem_cgroup_disabled()) | |
363 | return NULL; | |
364 | ||
365 | mz = container_of(lruvec, struct mem_cgroup_per_node, lruvec); | |
366 | return mz->memcg; | |
367 | } | |
368 | ||
8e8ae645 JW |
369 | /** |
370 | * parent_mem_cgroup - find the accounting parent of a memcg | |
371 | * @memcg: memcg whose parent to find | |
372 | * | |
373 | * Returns the parent memcg, or NULL if this is the root or the memory | |
374 | * controller is in legacy no-hierarchy mode. | |
375 | */ | |
376 | static inline struct mem_cgroup *parent_mem_cgroup(struct mem_cgroup *memcg) | |
377 | { | |
378 | if (!memcg->memory.parent) | |
379 | return NULL; | |
380 | return mem_cgroup_from_counter(memcg->memory.parent, memory); | |
381 | } | |
382 | ||
33398cf2 MH |
383 | static inline bool mem_cgroup_is_descendant(struct mem_cgroup *memcg, |
384 | struct mem_cgroup *root) | |
385 | { | |
386 | if (root == memcg) | |
387 | return true; | |
388 | if (!root->use_hierarchy) | |
389 | return false; | |
390 | return cgroup_is_descendant(memcg->css.cgroup, root->css.cgroup); | |
391 | } | |
e1aab161 | 392 | |
2314b42d JW |
393 | static inline bool mm_match_cgroup(struct mm_struct *mm, |
394 | struct mem_cgroup *memcg) | |
2e4d4091 | 395 | { |
587af308 | 396 | struct mem_cgroup *task_memcg; |
413918bb | 397 | bool match = false; |
c3ac9a8a | 398 | |
2e4d4091 | 399 | rcu_read_lock(); |
587af308 | 400 | task_memcg = mem_cgroup_from_task(rcu_dereference(mm->owner)); |
413918bb | 401 | if (task_memcg) |
2314b42d | 402 | match = mem_cgroup_is_descendant(task_memcg, memcg); |
2e4d4091 | 403 | rcu_read_unlock(); |
c3ac9a8a | 404 | return match; |
2e4d4091 | 405 | } |
8a9f3ccd | 406 | |
64219994 | 407 | struct cgroup_subsys_state *mem_cgroup_css_from_page(struct page *page); |
2fc04524 | 408 | ino_t page_cgroup_ino(struct page *page); |
d324236b | 409 | |
eb01aaab VD |
410 | static inline bool mem_cgroup_online(struct mem_cgroup *memcg) |
411 | { | |
412 | if (mem_cgroup_disabled()) | |
413 | return true; | |
414 | return !!(memcg->css.flags & CSS_ONLINE); | |
415 | } | |
416 | ||
58ae83db KH |
417 | /* |
418 | * For memory reclaim. | |
419 | */ | |
889976db | 420 | int mem_cgroup_select_victim_node(struct mem_cgroup *memcg); |
33398cf2 MH |
421 | |
422 | void mem_cgroup_update_lru_size(struct lruvec *lruvec, enum lru_list lru, | |
b4536f0c | 423 | int zid, int nr_pages); |
33398cf2 | 424 | |
0a6b76dd VD |
425 | unsigned long mem_cgroup_node_nr_lru_pages(struct mem_cgroup *memcg, |
426 | int nid, unsigned int lru_mask); | |
427 | ||
33398cf2 MH |
428 | static inline |
429 | unsigned long mem_cgroup_get_lru_size(struct lruvec *lruvec, enum lru_list lru) | |
430 | { | |
ef8f2327 | 431 | struct mem_cgroup_per_node *mz; |
b4536f0c MH |
432 | unsigned long nr_pages = 0; |
433 | int zid; | |
33398cf2 | 434 | |
ef8f2327 | 435 | mz = container_of(lruvec, struct mem_cgroup_per_node, lruvec); |
b4536f0c MH |
436 | for (zid = 0; zid < MAX_NR_ZONES; zid++) |
437 | nr_pages += mz->lru_zone_size[zid][lru]; | |
438 | return nr_pages; | |
439 | } | |
440 | ||
441 | static inline | |
442 | unsigned long mem_cgroup_get_zone_lru_size(struct lruvec *lruvec, | |
443 | enum lru_list lru, int zone_idx) | |
444 | { | |
445 | struct mem_cgroup_per_node *mz; | |
446 | ||
447 | mz = container_of(lruvec, struct mem_cgroup_per_node, lruvec); | |
448 | return mz->lru_zone_size[zone_idx][lru]; | |
33398cf2 MH |
449 | } |
450 | ||
b23afb93 TH |
451 | void mem_cgroup_handle_over_high(void); |
452 | ||
7c5f64f8 VD |
453 | unsigned long mem_cgroup_get_limit(struct mem_cgroup *memcg); |
454 | ||
64219994 MH |
455 | void mem_cgroup_print_oom_info(struct mem_cgroup *memcg, |
456 | struct task_struct *p); | |
58ae83db | 457 | |
49426420 | 458 | static inline void mem_cgroup_oom_enable(void) |
519e5247 | 459 | { |
626ebc41 TH |
460 | WARN_ON(current->memcg_may_oom); |
461 | current->memcg_may_oom = 1; | |
519e5247 JW |
462 | } |
463 | ||
49426420 | 464 | static inline void mem_cgroup_oom_disable(void) |
519e5247 | 465 | { |
626ebc41 TH |
466 | WARN_ON(!current->memcg_may_oom); |
467 | current->memcg_may_oom = 0; | |
519e5247 JW |
468 | } |
469 | ||
3812c8c8 JW |
470 | static inline bool task_in_memcg_oom(struct task_struct *p) |
471 | { | |
626ebc41 | 472 | return p->memcg_in_oom; |
3812c8c8 JW |
473 | } |
474 | ||
49426420 | 475 | bool mem_cgroup_oom_synchronize(bool wait); |
3812c8c8 | 476 | |
c255a458 | 477 | #ifdef CONFIG_MEMCG_SWAP |
c077719b KH |
478 | extern int do_swap_account; |
479 | #endif | |
f8d66542 | 480 | |
62cccb8c JW |
481 | void lock_page_memcg(struct page *page); |
482 | void unlock_page_memcg(struct page *page); | |
d7365e78 | 483 | |
ccda7f43 JW |
484 | static inline unsigned long memcg_page_state(struct mem_cgroup *memcg, |
485 | enum memcg_stat_item idx) | |
2a2e4885 JW |
486 | { |
487 | long val = 0; | |
488 | int cpu; | |
489 | ||
490 | for_each_possible_cpu(cpu) | |
491 | val += per_cpu(memcg->stat->count[idx], cpu); | |
492 | ||
493 | if (val < 0) | |
494 | val = 0; | |
495 | ||
496 | return val; | |
497 | } | |
498 | ||
ccda7f43 JW |
499 | static inline void mod_memcg_state(struct mem_cgroup *memcg, |
500 | enum memcg_stat_item idx, int val) | |
2a2e4885 JW |
501 | { |
502 | if (!mem_cgroup_disabled()) | |
503 | this_cpu_add(memcg->stat->count[idx], val); | |
504 | } | |
505 | ||
ccda7f43 JW |
506 | static inline void inc_memcg_state(struct mem_cgroup *memcg, |
507 | enum memcg_stat_item idx) | |
2a2e4885 | 508 | { |
ccda7f43 | 509 | mod_memcg_state(memcg, idx, 1); |
2a2e4885 JW |
510 | } |
511 | ||
ccda7f43 JW |
512 | static inline void dec_memcg_state(struct mem_cgroup *memcg, |
513 | enum memcg_stat_item idx) | |
2a2e4885 | 514 | { |
ccda7f43 | 515 | mod_memcg_state(memcg, idx, -1); |
2a2e4885 JW |
516 | } |
517 | ||
33398cf2 | 518 | /** |
ccda7f43 | 519 | * mod_memcg_page_state - update page state statistics |
62cccb8c | 520 | * @page: the page |
33398cf2 MH |
521 | * @idx: page state item to account |
522 | * @val: number of pages (positive or negative) | |
523 | * | |
fdf1cdb9 JW |
524 | * The @page must be locked or the caller must use lock_page_memcg() |
525 | * to prevent double accounting when the page is concurrently being | |
526 | * moved to another memcg: | |
81f8c3a4 | 527 | * |
fdf1cdb9 | 528 | * lock_page(page) or lock_page_memcg(page) |
81f8c3a4 | 529 | * if (TestClearPageState(page)) |
ccda7f43 | 530 | * mod_memcg_page_state(page, state, -1); |
fdf1cdb9 | 531 | * unlock_page(page) or unlock_page_memcg(page) |
2a2e4885 JW |
532 | * |
533 | * Kernel pages are an exception to this, since they'll never move. | |
33398cf2 | 534 | */ |
ccda7f43 JW |
535 | static inline void mod_memcg_page_state(struct page *page, |
536 | enum memcg_stat_item idx, int val) | |
33398cf2 | 537 | { |
62cccb8c | 538 | if (page->mem_cgroup) |
ccda7f43 | 539 | mod_memcg_state(page->mem_cgroup, idx, val); |
33398cf2 MH |
540 | } |
541 | ||
ccda7f43 JW |
542 | static inline void inc_memcg_page_state(struct page *page, |
543 | enum memcg_stat_item idx) | |
2a7106f2 | 544 | { |
ccda7f43 | 545 | mod_memcg_page_state(page, idx, 1); |
2a7106f2 GT |
546 | } |
547 | ||
ccda7f43 JW |
548 | static inline void dec_memcg_page_state(struct page *page, |
549 | enum memcg_stat_item idx) | |
2a7106f2 | 550 | { |
ccda7f43 | 551 | mod_memcg_page_state(page, idx, -1); |
2a7106f2 GT |
552 | } |
553 | ||
ef8f2327 | 554 | unsigned long mem_cgroup_soft_limit_reclaim(pg_data_t *pgdat, int order, |
0608f43d AM |
555 | gfp_t gfp_mask, |
556 | unsigned long *total_scanned); | |
a63d83f4 | 557 | |
2262185c RG |
558 | static inline void count_memcg_events(struct mem_cgroup *memcg, |
559 | enum vm_event_item idx, | |
560 | unsigned long count) | |
561 | { | |
562 | if (!mem_cgroup_disabled()) | |
563 | this_cpu_add(memcg->stat->events[idx], count); | |
564 | } | |
565 | ||
566 | static inline void count_memcg_page_event(struct page *page, | |
567 | enum memcg_stat_item idx) | |
568 | { | |
569 | if (page->mem_cgroup) | |
570 | count_memcg_events(page->mem_cgroup, idx, 1); | |
571 | } | |
572 | ||
573 | static inline void count_memcg_event_mm(struct mm_struct *mm, | |
574 | enum vm_event_item idx) | |
68ae564b | 575 | { |
33398cf2 MH |
576 | struct mem_cgroup *memcg; |
577 | ||
68ae564b DR |
578 | if (mem_cgroup_disabled()) |
579 | return; | |
33398cf2 MH |
580 | |
581 | rcu_read_lock(); | |
582 | memcg = mem_cgroup_from_task(rcu_dereference(mm->owner)); | |
8e675f7a | 583 | if (likely(memcg)) { |
df0e53d0 | 584 | this_cpu_inc(memcg->stat->events[idx]); |
8e675f7a KK |
585 | if (idx == OOM_KILL) |
586 | cgroup_file_notify(&memcg->events_file); | |
587 | } | |
33398cf2 | 588 | rcu_read_unlock(); |
68ae564b | 589 | } |
ca3e0214 | 590 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE |
e94c8a9c | 591 | void mem_cgroup_split_huge_fixup(struct page *head); |
ca3e0214 KH |
592 | #endif |
593 | ||
c255a458 | 594 | #else /* CONFIG_MEMCG */ |
23047a96 JW |
595 | |
596 | #define MEM_CGROUP_ID_SHIFT 0 | |
597 | #define MEM_CGROUP_ID_MAX 0 | |
598 | ||
7a81b88c KH |
599 | struct mem_cgroup; |
600 | ||
23047a96 JW |
601 | static inline bool mem_cgroup_disabled(void) |
602 | { | |
603 | return true; | |
604 | } | |
605 | ||
31176c78 | 606 | static inline void mem_cgroup_event(struct mem_cgroup *memcg, |
df0e53d0 | 607 | enum memcg_event_item event) |
241994ed JW |
608 | { |
609 | } | |
610 | ||
611 | static inline bool mem_cgroup_low(struct mem_cgroup *root, | |
612 | struct mem_cgroup *memcg) | |
613 | { | |
614 | return false; | |
615 | } | |
616 | ||
00501b53 JW |
617 | static inline int mem_cgroup_try_charge(struct page *page, struct mm_struct *mm, |
618 | gfp_t gfp_mask, | |
f627c2f5 KS |
619 | struct mem_cgroup **memcgp, |
620 | bool compound) | |
7a81b88c | 621 | { |
00501b53 | 622 | *memcgp = NULL; |
7a81b88c KH |
623 | return 0; |
624 | } | |
625 | ||
00501b53 JW |
626 | static inline void mem_cgroup_commit_charge(struct page *page, |
627 | struct mem_cgroup *memcg, | |
f627c2f5 | 628 | bool lrucare, bool compound) |
7a81b88c KH |
629 | { |
630 | } | |
631 | ||
00501b53 | 632 | static inline void mem_cgroup_cancel_charge(struct page *page, |
f627c2f5 KS |
633 | struct mem_cgroup *memcg, |
634 | bool compound) | |
7a81b88c KH |
635 | { |
636 | } | |
637 | ||
0a31bc97 | 638 | static inline void mem_cgroup_uncharge(struct page *page) |
569b846d KH |
639 | { |
640 | } | |
641 | ||
747db954 | 642 | static inline void mem_cgroup_uncharge_list(struct list_head *page_list) |
8a9f3ccd BS |
643 | { |
644 | } | |
645 | ||
6a93ca8f | 646 | static inline void mem_cgroup_migrate(struct page *old, struct page *new) |
69029cd5 KH |
647 | { |
648 | } | |
649 | ||
a9dd0a83 | 650 | static inline struct lruvec *mem_cgroup_lruvec(struct pglist_data *pgdat, |
ef8f2327 | 651 | struct mem_cgroup *memcg) |
08e552c6 | 652 | { |
a9dd0a83 | 653 | return node_lruvec(pgdat); |
08e552c6 KH |
654 | } |
655 | ||
fa9add64 | 656 | static inline struct lruvec *mem_cgroup_page_lruvec(struct page *page, |
599d0c95 | 657 | struct pglist_data *pgdat) |
66e1707b | 658 | { |
599d0c95 | 659 | return &pgdat->lruvec; |
66e1707b BS |
660 | } |
661 | ||
587af308 | 662 | static inline bool mm_match_cgroup(struct mm_struct *mm, |
c0ff4b85 | 663 | struct mem_cgroup *memcg) |
bed7161a | 664 | { |
587af308 | 665 | return true; |
bed7161a BS |
666 | } |
667 | ||
ffbdccf5 DR |
668 | static inline bool task_in_mem_cgroup(struct task_struct *task, |
669 | const struct mem_cgroup *memcg) | |
4c4a2214 | 670 | { |
ffbdccf5 | 671 | return true; |
4c4a2214 DR |
672 | } |
673 | ||
5660048c JW |
674 | static inline struct mem_cgroup * |
675 | mem_cgroup_iter(struct mem_cgroup *root, | |
676 | struct mem_cgroup *prev, | |
677 | struct mem_cgroup_reclaim_cookie *reclaim) | |
678 | { | |
679 | return NULL; | |
680 | } | |
681 | ||
682 | static inline void mem_cgroup_iter_break(struct mem_cgroup *root, | |
683 | struct mem_cgroup *prev) | |
684 | { | |
685 | } | |
686 | ||
7c5f64f8 VD |
687 | static inline int mem_cgroup_scan_tasks(struct mem_cgroup *memcg, |
688 | int (*fn)(struct task_struct *, void *), void *arg) | |
689 | { | |
690 | return 0; | |
691 | } | |
692 | ||
23047a96 | 693 | static inline unsigned short mem_cgroup_id(struct mem_cgroup *memcg) |
f8d66542 | 694 | { |
23047a96 JW |
695 | return 0; |
696 | } | |
697 | ||
698 | static inline struct mem_cgroup *mem_cgroup_from_id(unsigned short id) | |
699 | { | |
700 | WARN_ON_ONCE(id); | |
701 | /* XXX: This should always return root_mem_cgroup */ | |
702 | return NULL; | |
f8d66542 | 703 | } |
a636b327 | 704 | |
2262185c RG |
705 | static inline struct mem_cgroup *lruvec_memcg(struct lruvec *lruvec) |
706 | { | |
707 | return NULL; | |
708 | } | |
709 | ||
eb01aaab | 710 | static inline bool mem_cgroup_online(struct mem_cgroup *memcg) |
14797e23 | 711 | { |
13308ca9 | 712 | return true; |
14797e23 KM |
713 | } |
714 | ||
a3d8e054 | 715 | static inline unsigned long |
4d7dcca2 | 716 | mem_cgroup_get_lru_size(struct lruvec *lruvec, enum lru_list lru) |
a3d8e054 KM |
717 | { |
718 | return 0; | |
719 | } | |
b4536f0c MH |
720 | static inline |
721 | unsigned long mem_cgroup_get_zone_lru_size(struct lruvec *lruvec, | |
722 | enum lru_list lru, int zone_idx) | |
723 | { | |
724 | return 0; | |
725 | } | |
a3d8e054 | 726 | |
0a6b76dd VD |
727 | static inline unsigned long |
728 | mem_cgroup_node_nr_lru_pages(struct mem_cgroup *memcg, | |
729 | int nid, unsigned int lru_mask) | |
730 | { | |
731 | return 0; | |
732 | } | |
733 | ||
7c5f64f8 VD |
734 | static inline unsigned long mem_cgroup_get_limit(struct mem_cgroup *memcg) |
735 | { | |
736 | return 0; | |
737 | } | |
738 | ||
e222432b BS |
739 | static inline void |
740 | mem_cgroup_print_oom_info(struct mem_cgroup *memcg, struct task_struct *p) | |
741 | { | |
742 | } | |
743 | ||
62cccb8c | 744 | static inline void lock_page_memcg(struct page *page) |
89c06bd5 KH |
745 | { |
746 | } | |
747 | ||
62cccb8c | 748 | static inline void unlock_page_memcg(struct page *page) |
89c06bd5 KH |
749 | { |
750 | } | |
751 | ||
b23afb93 TH |
752 | static inline void mem_cgroup_handle_over_high(void) |
753 | { | |
754 | } | |
755 | ||
49426420 | 756 | static inline void mem_cgroup_oom_enable(void) |
519e5247 JW |
757 | { |
758 | } | |
759 | ||
49426420 | 760 | static inline void mem_cgroup_oom_disable(void) |
519e5247 JW |
761 | { |
762 | } | |
763 | ||
3812c8c8 JW |
764 | static inline bool task_in_memcg_oom(struct task_struct *p) |
765 | { | |
766 | return false; | |
767 | } | |
768 | ||
49426420 | 769 | static inline bool mem_cgroup_oom_synchronize(bool wait) |
3812c8c8 JW |
770 | { |
771 | return false; | |
772 | } | |
773 | ||
ccda7f43 JW |
774 | static inline unsigned long memcg_page_state(struct mem_cgroup *memcg, |
775 | enum memcg_stat_item idx) | |
2a2e4885 JW |
776 | { |
777 | return 0; | |
778 | } | |
779 | ||
ccda7f43 JW |
780 | static inline void mod_memcg_state(struct mem_cgroup *memcg, |
781 | enum memcg_stat_item idx, | |
782 | int nr) | |
2a2e4885 JW |
783 | { |
784 | } | |
785 | ||
ccda7f43 JW |
786 | static inline void inc_memcg_state(struct mem_cgroup *memcg, |
787 | enum memcg_stat_item idx) | |
2a2e4885 JW |
788 | { |
789 | } | |
790 | ||
ccda7f43 JW |
791 | static inline void dec_memcg_state(struct mem_cgroup *memcg, |
792 | enum memcg_stat_item idx) | |
2a2e4885 JW |
793 | { |
794 | } | |
795 | ||
ccda7f43 JW |
796 | static inline void mod_memcg_page_state(struct page *page, |
797 | enum memcg_stat_item idx, | |
798 | int nr) | |
553af430 JW |
799 | { |
800 | } | |
801 | ||
ccda7f43 JW |
802 | static inline void inc_memcg_page_state(struct page *page, |
803 | enum memcg_stat_item idx) | |
2a7106f2 GT |
804 | { |
805 | } | |
806 | ||
ccda7f43 JW |
807 | static inline void dec_memcg_page_state(struct page *page, |
808 | enum memcg_stat_item idx) | |
d69b042f BS |
809 | { |
810 | } | |
811 | ||
4e416953 | 812 | static inline |
ef8f2327 | 813 | unsigned long mem_cgroup_soft_limit_reclaim(pg_data_t *pgdat, int order, |
0608f43d AM |
814 | gfp_t gfp_mask, |
815 | unsigned long *total_scanned) | |
4e416953 | 816 | { |
0608f43d | 817 | return 0; |
4e416953 BS |
818 | } |
819 | ||
e94c8a9c | 820 | static inline void mem_cgroup_split_huge_fixup(struct page *head) |
ca3e0214 KH |
821 | { |
822 | } | |
823 | ||
2262185c RG |
824 | static inline void count_memcg_events(struct mem_cgroup *memcg, |
825 | enum vm_event_item idx, | |
826 | unsigned long count) | |
827 | { | |
828 | } | |
829 | ||
830 | static inline void count_memcg_page_event(struct page *page, | |
831 | enum memcg_stat_item idx) | |
832 | { | |
833 | } | |
834 | ||
456f998e | 835 | static inline |
2262185c | 836 | void count_memcg_event_mm(struct mm_struct *mm, enum vm_event_item idx) |
456f998e YH |
837 | { |
838 | } | |
c255a458 | 839 | #endif /* CONFIG_MEMCG */ |
78fb7466 | 840 | |
52ebea74 | 841 | #ifdef CONFIG_CGROUP_WRITEBACK |
841710aa | 842 | |
52ebea74 | 843 | struct list_head *mem_cgroup_cgwb_list(struct mem_cgroup *memcg); |
841710aa | 844 | struct wb_domain *mem_cgroup_wb_domain(struct bdi_writeback *wb); |
c5edf9cd TH |
845 | void mem_cgroup_wb_stats(struct bdi_writeback *wb, unsigned long *pfilepages, |
846 | unsigned long *pheadroom, unsigned long *pdirty, | |
847 | unsigned long *pwriteback); | |
841710aa TH |
848 | |
849 | #else /* CONFIG_CGROUP_WRITEBACK */ | |
850 | ||
851 | static inline struct wb_domain *mem_cgroup_wb_domain(struct bdi_writeback *wb) | |
852 | { | |
853 | return NULL; | |
854 | } | |
855 | ||
c2aa723a | 856 | static inline void mem_cgroup_wb_stats(struct bdi_writeback *wb, |
c5edf9cd TH |
857 | unsigned long *pfilepages, |
858 | unsigned long *pheadroom, | |
c2aa723a TH |
859 | unsigned long *pdirty, |
860 | unsigned long *pwriteback) | |
861 | { | |
862 | } | |
863 | ||
841710aa | 864 | #endif /* CONFIG_CGROUP_WRITEBACK */ |
52ebea74 | 865 | |
e1aab161 | 866 | struct sock; |
baac50bb JW |
867 | bool mem_cgroup_charge_skmem(struct mem_cgroup *memcg, unsigned int nr_pages); |
868 | void mem_cgroup_uncharge_skmem(struct mem_cgroup *memcg, unsigned int nr_pages); | |
d886f4e4 | 869 | #ifdef CONFIG_MEMCG |
ef12947c JW |
870 | extern struct static_key_false memcg_sockets_enabled_key; |
871 | #define mem_cgroup_sockets_enabled static_branch_unlikely(&memcg_sockets_enabled_key) | |
2d758073 JW |
872 | void mem_cgroup_sk_alloc(struct sock *sk); |
873 | void mem_cgroup_sk_free(struct sock *sk); | |
baac50bb | 874 | static inline bool mem_cgroup_under_socket_pressure(struct mem_cgroup *memcg) |
e805605c | 875 | { |
0db15298 | 876 | if (!cgroup_subsys_on_dfl(memory_cgrp_subsys) && memcg->tcpmem_pressure) |
8e8ae645 | 877 | return true; |
8e8ae645 JW |
878 | do { |
879 | if (time_before(jiffies, memcg->socket_pressure)) | |
880 | return true; | |
881 | } while ((memcg = parent_mem_cgroup(memcg))); | |
882 | return false; | |
e805605c JW |
883 | } |
884 | #else | |
80e95fe0 | 885 | #define mem_cgroup_sockets_enabled 0 |
2d758073 JW |
886 | static inline void mem_cgroup_sk_alloc(struct sock *sk) { }; |
887 | static inline void mem_cgroup_sk_free(struct sock *sk) { }; | |
baac50bb | 888 | static inline bool mem_cgroup_under_socket_pressure(struct mem_cgroup *memcg) |
e805605c JW |
889 | { |
890 | return false; | |
891 | } | |
892 | #endif | |
7ae1e1d0 | 893 | |
45264778 VD |
894 | struct kmem_cache *memcg_kmem_get_cache(struct kmem_cache *cachep); |
895 | void memcg_kmem_put_cache(struct kmem_cache *cachep); | |
896 | int memcg_kmem_charge_memcg(struct page *page, gfp_t gfp, int order, | |
897 | struct mem_cgroup *memcg); | |
898 | int memcg_kmem_charge(struct page *page, gfp_t gfp, int order); | |
899 | void memcg_kmem_uncharge(struct page *page, int order); | |
900 | ||
127424c8 | 901 | #if defined(CONFIG_MEMCG) && !defined(CONFIG_SLOB) |
ef12947c | 902 | extern struct static_key_false memcg_kmem_enabled_key; |
17cc4dfe | 903 | extern struct workqueue_struct *memcg_kmem_cache_wq; |
749c5415 | 904 | |
dbcf73e2 | 905 | extern int memcg_nr_cache_ids; |
64219994 MH |
906 | void memcg_get_cache_ids(void); |
907 | void memcg_put_cache_ids(void); | |
ebe945c2 GC |
908 | |
909 | /* | |
910 | * Helper macro to loop through all memcg-specific caches. Callers must still | |
911 | * check if the cache is valid (it is either valid or NULL). | |
912 | * the slab_mutex must be held when looping through those caches | |
913 | */ | |
749c5415 | 914 | #define for_each_memcg_cache_index(_idx) \ |
dbcf73e2 | 915 | for ((_idx) = 0; (_idx) < memcg_nr_cache_ids; (_idx)++) |
749c5415 | 916 | |
7ae1e1d0 GC |
917 | static inline bool memcg_kmem_enabled(void) |
918 | { | |
ef12947c | 919 | return static_branch_unlikely(&memcg_kmem_enabled_key); |
7ae1e1d0 GC |
920 | } |
921 | ||
33398cf2 | 922 | /* |
9f706d68 | 923 | * helper for accessing a memcg's index. It will be used as an index in the |
33398cf2 MH |
924 | * child cache array in kmem_cache, and also to derive its name. This function |
925 | * will return -1 when this is not a kmem-limited memcg. | |
926 | */ | |
927 | static inline int memcg_cache_id(struct mem_cgroup *memcg) | |
928 | { | |
929 | return memcg ? memcg->kmemcg_id : -1; | |
930 | } | |
5722d094 | 931 | |
7ae1e1d0 | 932 | #else |
749c5415 GC |
933 | #define for_each_memcg_cache_index(_idx) \ |
934 | for (; NULL; ) | |
935 | ||
b9ce5ef4 GC |
936 | static inline bool memcg_kmem_enabled(void) |
937 | { | |
938 | return false; | |
939 | } | |
940 | ||
2633d7a0 GC |
941 | static inline int memcg_cache_id(struct mem_cgroup *memcg) |
942 | { | |
943 | return -1; | |
944 | } | |
945 | ||
05257a1a VD |
946 | static inline void memcg_get_cache_ids(void) |
947 | { | |
948 | } | |
949 | ||
950 | static inline void memcg_put_cache_ids(void) | |
951 | { | |
952 | } | |
953 | ||
127424c8 JW |
954 | #endif /* CONFIG_MEMCG && !CONFIG_SLOB */ |
955 | ||
8cdea7c0 | 956 | #endif /* _LINUX_MEMCONTROL_H */ |