]>
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> | |
456f998e | 31 | |
78fb7466 | 32 | struct mem_cgroup; |
8697d331 BS |
33 | struct page; |
34 | struct mm_struct; | |
2633d7a0 | 35 | struct kmem_cache; |
78fb7466 | 36 | |
68b4876d SZ |
37 | /* |
38 | * The corresponding mem_cgroup_stat_names is defined in mm/memcontrol.c, | |
39 | * These two lists should keep in accord with each other. | |
40 | */ | |
41 | enum mem_cgroup_stat_index { | |
42 | /* | |
43 | * For MEM_CONTAINER_TYPE_ALL, usage = pagecache + rss. | |
44 | */ | |
45 | MEM_CGROUP_STAT_CACHE, /* # of pages charged as cache */ | |
46 | MEM_CGROUP_STAT_RSS, /* # of pages charged as anon rss */ | |
47 | MEM_CGROUP_STAT_RSS_HUGE, /* # of pages charged as anon huge */ | |
48 | MEM_CGROUP_STAT_FILE_MAPPED, /* # of pages charged as file rss */ | |
c4843a75 | 49 | MEM_CGROUP_STAT_DIRTY, /* # of dirty pages in page cache */ |
3ea67d06 | 50 | MEM_CGROUP_STAT_WRITEBACK, /* # of pages under writeback */ |
68b4876d SZ |
51 | MEM_CGROUP_STAT_SWAP, /* # of pages, swapped out */ |
52 | MEM_CGROUP_STAT_NSTATS, | |
2a7106f2 GT |
53 | }; |
54 | ||
5660048c JW |
55 | struct mem_cgroup_reclaim_cookie { |
56 | struct zone *zone; | |
57 | int priority; | |
58 | unsigned int generation; | |
59 | }; | |
60 | ||
241994ed JW |
61 | enum mem_cgroup_events_index { |
62 | MEM_CGROUP_EVENTS_PGPGIN, /* # of pages paged in */ | |
63 | MEM_CGROUP_EVENTS_PGPGOUT, /* # of pages paged out */ | |
64 | MEM_CGROUP_EVENTS_PGFAULT, /* # of page-faults */ | |
65 | MEM_CGROUP_EVENTS_PGMAJFAULT, /* # of major page-faults */ | |
66 | MEM_CGROUP_EVENTS_NSTATS, | |
67 | /* default hierarchy events */ | |
68 | MEMCG_LOW = MEM_CGROUP_EVENTS_NSTATS, | |
69 | MEMCG_HIGH, | |
70 | MEMCG_MAX, | |
71 | MEMCG_OOM, | |
72 | MEMCG_NR_EVENTS, | |
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 | ||
88 | /* | |
89 | * Bits in struct cg_proto.flags | |
90 | */ | |
91 | enum cg_proto_flags { | |
92 | /* Currently active and new sockets should be assigned to cgroups */ | |
93 | MEMCG_SOCK_ACTIVE, | |
94 | /* It was ever activated; we must disarm static keys on destruction */ | |
95 | MEMCG_SOCK_ACTIVATED, | |
96 | }; | |
97 | ||
98 | struct cg_proto { | |
99 | struct page_counter memory_allocated; /* Current allocated memory. */ | |
100 | struct percpu_counter sockets_allocated; /* Current number of sockets. */ | |
101 | int memory_pressure; | |
102 | long sysctl_mem[3]; | |
103 | unsigned long flags; | |
104 | /* | |
105 | * memcg field is used to find which memcg we belong directly | |
106 | * Each memcg struct can hold more than one cg_proto, so container_of | |
107 | * won't really cut. | |
108 | * | |
109 | * The elegant solution would be having an inverse function to | |
110 | * proto_cgroup in struct proto, but that means polluting the structure | |
111 | * for everybody, instead of just for memcg users. | |
112 | */ | |
113 | struct mem_cgroup *memcg; | |
114 | }; | |
115 | ||
c255a458 | 116 | #ifdef CONFIG_MEMCG |
33398cf2 MH |
117 | struct mem_cgroup_stat_cpu { |
118 | long count[MEM_CGROUP_STAT_NSTATS]; | |
119 | unsigned long events[MEMCG_NR_EVENTS]; | |
120 | unsigned long nr_page_events; | |
121 | unsigned long targets[MEM_CGROUP_NTARGETS]; | |
122 | }; | |
123 | ||
124 | struct mem_cgroup_reclaim_iter { | |
125 | struct mem_cgroup *position; | |
126 | /* scan generation, increased every round-trip */ | |
127 | unsigned int generation; | |
128 | }; | |
129 | ||
130 | /* | |
131 | * per-zone information in memory controller. | |
132 | */ | |
133 | struct mem_cgroup_per_zone { | |
134 | struct lruvec lruvec; | |
135 | unsigned long lru_size[NR_LRU_LISTS]; | |
136 | ||
137 | struct mem_cgroup_reclaim_iter iter[DEF_PRIORITY + 1]; | |
138 | ||
139 | struct rb_node tree_node; /* RB tree node */ | |
140 | unsigned long usage_in_excess;/* Set to the value by which */ | |
141 | /* the soft limit is exceeded*/ | |
142 | bool on_tree; | |
143 | struct mem_cgroup *memcg; /* Back pointer, we cannot */ | |
144 | /* use container_of */ | |
145 | }; | |
146 | ||
147 | struct mem_cgroup_per_node { | |
148 | struct mem_cgroup_per_zone zoneinfo[MAX_NR_ZONES]; | |
149 | }; | |
150 | ||
151 | struct mem_cgroup_threshold { | |
152 | struct eventfd_ctx *eventfd; | |
153 | unsigned long threshold; | |
154 | }; | |
155 | ||
156 | /* For threshold */ | |
157 | struct mem_cgroup_threshold_ary { | |
158 | /* An array index points to threshold just below or equal to usage. */ | |
159 | int current_threshold; | |
160 | /* Size of entries[] */ | |
161 | unsigned int size; | |
162 | /* Array of thresholds */ | |
163 | struct mem_cgroup_threshold entries[0]; | |
164 | }; | |
165 | ||
166 | struct mem_cgroup_thresholds { | |
167 | /* Primary thresholds array */ | |
168 | struct mem_cgroup_threshold_ary *primary; | |
169 | /* | |
170 | * Spare threshold array. | |
171 | * This is needed to make mem_cgroup_unregister_event() "never fail". | |
172 | * It must be able to store at least primary->size - 1 entries. | |
173 | */ | |
174 | struct mem_cgroup_threshold_ary *spare; | |
175 | }; | |
176 | ||
177 | /* | |
178 | * The memory controller data structure. The memory controller controls both | |
179 | * page cache and RSS per cgroup. We would eventually like to provide | |
180 | * statistics based on the statistics developed by Rik Van Riel for clock-pro, | |
181 | * to help the administrator determine what knobs to tune. | |
182 | */ | |
183 | struct mem_cgroup { | |
184 | struct cgroup_subsys_state css; | |
185 | ||
186 | /* Accounted resources */ | |
187 | struct page_counter memory; | |
188 | struct page_counter memsw; | |
189 | struct page_counter kmem; | |
190 | ||
191 | /* Normal memory consumption range */ | |
192 | unsigned long low; | |
193 | unsigned long high; | |
194 | ||
195 | unsigned long soft_limit; | |
196 | ||
197 | /* vmpressure notifications */ | |
198 | struct vmpressure vmpressure; | |
199 | ||
200 | /* css_online() has been completed */ | |
201 | int initialized; | |
202 | ||
203 | /* | |
204 | * Should the accounting and control be hierarchical, per subtree? | |
205 | */ | |
206 | bool use_hierarchy; | |
207 | ||
208 | /* protected by memcg_oom_lock */ | |
209 | bool oom_lock; | |
210 | int under_oom; | |
211 | ||
212 | int swappiness; | |
213 | /* OOM-Killer disable */ | |
214 | int oom_kill_disable; | |
215 | ||
216 | /* protect arrays of thresholds */ | |
217 | struct mutex thresholds_lock; | |
218 | ||
219 | /* thresholds for memory usage. RCU-protected */ | |
220 | struct mem_cgroup_thresholds thresholds; | |
221 | ||
222 | /* thresholds for mem+swap usage. RCU-protected */ | |
223 | struct mem_cgroup_thresholds memsw_thresholds; | |
224 | ||
225 | /* For oom notifier event fd */ | |
226 | struct list_head oom_notify; | |
227 | ||
228 | /* | |
229 | * Should we move charges of a task when a task is moved into this | |
230 | * mem_cgroup ? And what type of charges should we move ? | |
231 | */ | |
232 | unsigned long move_charge_at_immigrate; | |
233 | /* | |
234 | * set > 0 if pages under this cgroup are moving to other cgroup. | |
235 | */ | |
236 | atomic_t moving_account; | |
237 | /* taken only while moving_account > 0 */ | |
238 | spinlock_t move_lock; | |
239 | struct task_struct *move_lock_task; | |
240 | unsigned long move_lock_flags; | |
241 | /* | |
242 | * percpu counter. | |
243 | */ | |
244 | struct mem_cgroup_stat_cpu __percpu *stat; | |
245 | spinlock_t pcp_counter_lock; | |
246 | ||
247 | #if defined(CONFIG_MEMCG_KMEM) && defined(CONFIG_INET) | |
248 | struct cg_proto tcp_mem; | |
249 | #endif | |
250 | #if defined(CONFIG_MEMCG_KMEM) | |
251 | /* Index in the kmem_cache->memcg_params.memcg_caches array */ | |
252 | int kmemcg_id; | |
253 | bool kmem_acct_activated; | |
254 | bool kmem_acct_active; | |
255 | #endif | |
256 | ||
257 | int last_scanned_node; | |
258 | #if MAX_NUMNODES > 1 | |
259 | nodemask_t scan_nodes; | |
260 | atomic_t numainfo_events; | |
261 | atomic_t numainfo_updating; | |
262 | #endif | |
263 | ||
264 | #ifdef CONFIG_CGROUP_WRITEBACK | |
265 | struct list_head cgwb_list; | |
266 | struct wb_domain cgwb_domain; | |
267 | #endif | |
268 | ||
269 | /* List of events which userspace want to receive */ | |
270 | struct list_head event_list; | |
271 | spinlock_t event_list_lock; | |
272 | ||
273 | struct mem_cgroup_per_node *nodeinfo[0]; | |
274 | /* WARNING: nodeinfo must be the last member here */ | |
275 | }; | |
56161634 TH |
276 | extern struct cgroup_subsys_state *mem_cgroup_root_css; |
277 | ||
33398cf2 MH |
278 | /** |
279 | * mem_cgroup_events - count memory events against a cgroup | |
280 | * @memcg: the memory cgroup | |
281 | * @idx: the event index | |
282 | * @nr: the number of events to account for | |
283 | */ | |
284 | static inline void mem_cgroup_events(struct mem_cgroup *memcg, | |
241994ed | 285 | enum mem_cgroup_events_index idx, |
33398cf2 MH |
286 | unsigned int nr) |
287 | { | |
288 | this_cpu_add(memcg->stat->events[idx], nr); | |
289 | } | |
241994ed JW |
290 | |
291 | bool mem_cgroup_low(struct mem_cgroup *root, struct mem_cgroup *memcg); | |
292 | ||
00501b53 JW |
293 | int mem_cgroup_try_charge(struct page *page, struct mm_struct *mm, |
294 | gfp_t gfp_mask, struct mem_cgroup **memcgp); | |
295 | void mem_cgroup_commit_charge(struct page *page, struct mem_cgroup *memcg, | |
296 | bool lrucare); | |
297 | void mem_cgroup_cancel_charge(struct page *page, struct mem_cgroup *memcg); | |
0a31bc97 | 298 | void mem_cgroup_uncharge(struct page *page); |
747db954 | 299 | void mem_cgroup_uncharge_list(struct list_head *page_list); |
569b846d | 300 | |
0a31bc97 JW |
301 | void mem_cgroup_migrate(struct page *oldpage, struct page *newpage, |
302 | bool lrucare); | |
569b846d | 303 | |
0a31bc97 JW |
304 | struct lruvec *mem_cgroup_zone_lruvec(struct zone *, struct mem_cgroup *); |
305 | struct lruvec *mem_cgroup_page_lruvec(struct page *, struct zone *); | |
c9b0ed51 | 306 | |
2314b42d | 307 | bool task_in_mem_cgroup(struct task_struct *task, struct mem_cgroup *memcg); |
3062fc67 | 308 | |
e42d9d5d | 309 | extern struct mem_cgroup *try_get_mem_cgroup_from_page(struct page *page); |
cf475ad2 BS |
310 | extern struct mem_cgroup *mem_cgroup_from_task(struct task_struct *p); |
311 | ||
e1aab161 | 312 | extern struct mem_cgroup *parent_mem_cgroup(struct mem_cgroup *memcg); |
33398cf2 MH |
313 | static inline |
314 | struct mem_cgroup *mem_cgroup_from_css(struct cgroup_subsys_state *css){ | |
315 | return css ? container_of(css, struct mem_cgroup, css) : NULL; | |
316 | } | |
317 | ||
318 | struct mem_cgroup *mem_cgroup_iter(struct mem_cgroup *, | |
319 | struct mem_cgroup *, | |
320 | struct mem_cgroup_reclaim_cookie *); | |
321 | void mem_cgroup_iter_break(struct mem_cgroup *, struct mem_cgroup *); | |
322 | ||
323 | static inline bool mem_cgroup_is_descendant(struct mem_cgroup *memcg, | |
324 | struct mem_cgroup *root) | |
325 | { | |
326 | if (root == memcg) | |
327 | return true; | |
328 | if (!root->use_hierarchy) | |
329 | return false; | |
330 | return cgroup_is_descendant(memcg->css.cgroup, root->css.cgroup); | |
331 | } | |
e1aab161 | 332 | |
2314b42d JW |
333 | static inline bool mm_match_cgroup(struct mm_struct *mm, |
334 | struct mem_cgroup *memcg) | |
2e4d4091 | 335 | { |
587af308 | 336 | struct mem_cgroup *task_memcg; |
413918bb | 337 | bool match = false; |
c3ac9a8a | 338 | |
2e4d4091 | 339 | rcu_read_lock(); |
587af308 | 340 | task_memcg = mem_cgroup_from_task(rcu_dereference(mm->owner)); |
413918bb | 341 | if (task_memcg) |
2314b42d | 342 | match = mem_cgroup_is_descendant(task_memcg, memcg); |
2e4d4091 | 343 | rcu_read_unlock(); |
c3ac9a8a | 344 | return match; |
2e4d4091 | 345 | } |
8a9f3ccd | 346 | |
ad7fa852 | 347 | extern struct cgroup_subsys_state *mem_cgroup_css_from_page(struct page *page); |
d324236b | 348 | |
33398cf2 MH |
349 | static inline bool mem_cgroup_disabled(void) |
350 | { | |
351 | if (memory_cgrp_subsys.disabled) | |
352 | return true; | |
353 | return false; | |
354 | } | |
5660048c | 355 | |
58ae83db KH |
356 | /* |
357 | * For memory reclaim. | |
358 | */ | |
889976db | 359 | int mem_cgroup_select_victim_node(struct mem_cgroup *memcg); |
33398cf2 MH |
360 | |
361 | void mem_cgroup_update_lru_size(struct lruvec *lruvec, enum lru_list lru, | |
362 | int nr_pages); | |
363 | ||
364 | static inline bool mem_cgroup_lruvec_online(struct lruvec *lruvec) | |
365 | { | |
366 | struct mem_cgroup_per_zone *mz; | |
367 | struct mem_cgroup *memcg; | |
368 | ||
369 | if (mem_cgroup_disabled()) | |
370 | return true; | |
371 | ||
372 | mz = container_of(lruvec, struct mem_cgroup_per_zone, lruvec); | |
373 | memcg = mz->memcg; | |
374 | ||
375 | return !!(memcg->css.flags & CSS_ONLINE); | |
376 | } | |
377 | ||
378 | static inline | |
379 | unsigned long mem_cgroup_get_lru_size(struct lruvec *lruvec, enum lru_list lru) | |
380 | { | |
381 | struct mem_cgroup_per_zone *mz; | |
382 | ||
383 | mz = container_of(lruvec, struct mem_cgroup_per_zone, lruvec); | |
384 | return mz->lru_size[lru]; | |
385 | } | |
386 | ||
387 | static inline int mem_cgroup_inactive_anon_is_low(struct lruvec *lruvec) | |
388 | { | |
389 | unsigned long inactive_ratio; | |
390 | unsigned long inactive; | |
391 | unsigned long active; | |
392 | unsigned long gb; | |
393 | ||
394 | inactive = mem_cgroup_get_lru_size(lruvec, LRU_INACTIVE_ANON); | |
395 | active = mem_cgroup_get_lru_size(lruvec, LRU_ACTIVE_ANON); | |
396 | ||
397 | gb = (inactive + active) >> (30 - PAGE_SHIFT); | |
398 | if (gb) | |
399 | inactive_ratio = int_sqrt(10 * gb); | |
400 | else | |
401 | inactive_ratio = 1; | |
402 | ||
403 | return inactive * inactive_ratio < active; | |
404 | } | |
405 | ||
e222432b BS |
406 | extern void mem_cgroup_print_oom_info(struct mem_cgroup *memcg, |
407 | struct task_struct *p); | |
58ae83db | 408 | |
49426420 | 409 | static inline void mem_cgroup_oom_enable(void) |
519e5247 | 410 | { |
49426420 JW |
411 | WARN_ON(current->memcg_oom.may_oom); |
412 | current->memcg_oom.may_oom = 1; | |
519e5247 JW |
413 | } |
414 | ||
49426420 | 415 | static inline void mem_cgroup_oom_disable(void) |
519e5247 | 416 | { |
49426420 JW |
417 | WARN_ON(!current->memcg_oom.may_oom); |
418 | current->memcg_oom.may_oom = 0; | |
519e5247 JW |
419 | } |
420 | ||
3812c8c8 JW |
421 | static inline bool task_in_memcg_oom(struct task_struct *p) |
422 | { | |
49426420 | 423 | return p->memcg_oom.memcg; |
3812c8c8 JW |
424 | } |
425 | ||
49426420 | 426 | bool mem_cgroup_oom_synchronize(bool wait); |
3812c8c8 | 427 | |
c255a458 | 428 | #ifdef CONFIG_MEMCG_SWAP |
c077719b KH |
429 | extern int do_swap_account; |
430 | #endif | |
f8d66542 | 431 | |
6de22619 | 432 | struct mem_cgroup *mem_cgroup_begin_page_stat(struct page *page); |
6de22619 | 433 | void mem_cgroup_end_page_stat(struct mem_cgroup *memcg); |
d7365e78 | 434 | |
33398cf2 MH |
435 | /** |
436 | * mem_cgroup_update_page_stat - update page state statistics | |
437 | * @memcg: memcg to account against | |
438 | * @idx: page state item to account | |
439 | * @val: number of pages (positive or negative) | |
440 | * | |
441 | * See mem_cgroup_begin_page_stat() for locking requirements. | |
442 | */ | |
443 | static inline void mem_cgroup_update_page_stat(struct mem_cgroup *memcg, | |
444 | enum mem_cgroup_stat_index idx, int val) | |
445 | { | |
446 | VM_BUG_ON(!rcu_read_lock_held()); | |
447 | ||
448 | if (memcg) | |
449 | this_cpu_add(memcg->stat->count[idx], val); | |
450 | } | |
451 | ||
d7365e78 | 452 | static inline void mem_cgroup_inc_page_stat(struct mem_cgroup *memcg, |
68b4876d | 453 | enum mem_cgroup_stat_index idx) |
2a7106f2 | 454 | { |
d7365e78 | 455 | mem_cgroup_update_page_stat(memcg, idx, 1); |
2a7106f2 GT |
456 | } |
457 | ||
d7365e78 | 458 | static inline void mem_cgroup_dec_page_stat(struct mem_cgroup *memcg, |
68b4876d | 459 | enum mem_cgroup_stat_index idx) |
2a7106f2 | 460 | { |
d7365e78 | 461 | mem_cgroup_update_page_stat(memcg, idx, -1); |
2a7106f2 GT |
462 | } |
463 | ||
0608f43d AM |
464 | unsigned long mem_cgroup_soft_limit_reclaim(struct zone *zone, int order, |
465 | gfp_t gfp_mask, | |
466 | unsigned long *total_scanned); | |
a63d83f4 | 467 | |
68ae564b DR |
468 | static inline void mem_cgroup_count_vm_event(struct mm_struct *mm, |
469 | enum vm_event_item idx) | |
470 | { | |
33398cf2 MH |
471 | struct mem_cgroup *memcg; |
472 | ||
68ae564b DR |
473 | if (mem_cgroup_disabled()) |
474 | return; | |
33398cf2 MH |
475 | |
476 | rcu_read_lock(); | |
477 | memcg = mem_cgroup_from_task(rcu_dereference(mm->owner)); | |
478 | if (unlikely(!memcg)) | |
479 | goto out; | |
480 | ||
481 | switch (idx) { | |
482 | case PGFAULT: | |
483 | this_cpu_inc(memcg->stat->events[MEM_CGROUP_EVENTS_PGFAULT]); | |
484 | break; | |
485 | case PGMAJFAULT: | |
486 | this_cpu_inc(memcg->stat->events[MEM_CGROUP_EVENTS_PGMAJFAULT]); | |
487 | break; | |
488 | default: | |
489 | BUG(); | |
490 | } | |
491 | out: | |
492 | rcu_read_unlock(); | |
68ae564b | 493 | } |
ca3e0214 | 494 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE |
e94c8a9c | 495 | void mem_cgroup_split_huge_fixup(struct page *head); |
ca3e0214 KH |
496 | #endif |
497 | ||
c255a458 | 498 | #else /* CONFIG_MEMCG */ |
7a81b88c KH |
499 | struct mem_cgroup; |
500 | ||
56161634 TH |
501 | #define mem_cgroup_root_css ((struct cgroup_subsys_state *)ERR_PTR(-EINVAL)) |
502 | ||
241994ed JW |
503 | static inline void mem_cgroup_events(struct mem_cgroup *memcg, |
504 | enum mem_cgroup_events_index idx, | |
505 | unsigned int nr) | |
506 | { | |
507 | } | |
508 | ||
509 | static inline bool mem_cgroup_low(struct mem_cgroup *root, | |
510 | struct mem_cgroup *memcg) | |
511 | { | |
512 | return false; | |
513 | } | |
514 | ||
00501b53 JW |
515 | static inline int mem_cgroup_try_charge(struct page *page, struct mm_struct *mm, |
516 | gfp_t gfp_mask, | |
517 | struct mem_cgroup **memcgp) | |
7a81b88c | 518 | { |
00501b53 | 519 | *memcgp = NULL; |
7a81b88c KH |
520 | return 0; |
521 | } | |
522 | ||
00501b53 JW |
523 | static inline void mem_cgroup_commit_charge(struct page *page, |
524 | struct mem_cgroup *memcg, | |
525 | bool lrucare) | |
7a81b88c KH |
526 | { |
527 | } | |
528 | ||
00501b53 JW |
529 | static inline void mem_cgroup_cancel_charge(struct page *page, |
530 | struct mem_cgroup *memcg) | |
7a81b88c KH |
531 | { |
532 | } | |
533 | ||
0a31bc97 | 534 | static inline void mem_cgroup_uncharge(struct page *page) |
569b846d KH |
535 | { |
536 | } | |
537 | ||
747db954 | 538 | static inline void mem_cgroup_uncharge_list(struct list_head *page_list) |
8a9f3ccd BS |
539 | { |
540 | } | |
541 | ||
0a31bc97 JW |
542 | static inline void mem_cgroup_migrate(struct page *oldpage, |
543 | struct page *newpage, | |
544 | bool lrucare) | |
69029cd5 KH |
545 | { |
546 | } | |
547 | ||
925b7673 JW |
548 | static inline struct lruvec *mem_cgroup_zone_lruvec(struct zone *zone, |
549 | struct mem_cgroup *memcg) | |
08e552c6 | 550 | { |
925b7673 | 551 | return &zone->lruvec; |
08e552c6 KH |
552 | } |
553 | ||
fa9add64 HD |
554 | static inline struct lruvec *mem_cgroup_page_lruvec(struct page *page, |
555 | struct zone *zone) | |
66e1707b | 556 | { |
925b7673 | 557 | return &zone->lruvec; |
66e1707b BS |
558 | } |
559 | ||
e42d9d5d WF |
560 | static inline struct mem_cgroup *try_get_mem_cgroup_from_page(struct page *page) |
561 | { | |
562 | return NULL; | |
563 | } | |
564 | ||
587af308 | 565 | static inline bool mm_match_cgroup(struct mm_struct *mm, |
c0ff4b85 | 566 | struct mem_cgroup *memcg) |
bed7161a | 567 | { |
587af308 | 568 | return true; |
bed7161a BS |
569 | } |
570 | ||
ffbdccf5 DR |
571 | static inline bool task_in_mem_cgroup(struct task_struct *task, |
572 | const struct mem_cgroup *memcg) | |
4c4a2214 | 573 | { |
ffbdccf5 | 574 | return true; |
4c4a2214 DR |
575 | } |
576 | ||
5660048c JW |
577 | static inline struct mem_cgroup * |
578 | mem_cgroup_iter(struct mem_cgroup *root, | |
579 | struct mem_cgroup *prev, | |
580 | struct mem_cgroup_reclaim_cookie *reclaim) | |
581 | { | |
582 | return NULL; | |
583 | } | |
584 | ||
585 | static inline void mem_cgroup_iter_break(struct mem_cgroup *root, | |
586 | struct mem_cgroup *prev) | |
587 | { | |
588 | } | |
589 | ||
f8d66542 HT |
590 | static inline bool mem_cgroup_disabled(void) |
591 | { | |
592 | return true; | |
593 | } | |
a636b327 | 594 | |
14797e23 | 595 | static inline int |
c56d5c7d | 596 | mem_cgroup_inactive_anon_is_low(struct lruvec *lruvec) |
14797e23 KM |
597 | { |
598 | return 1; | |
599 | } | |
600 | ||
90cbc250 VD |
601 | static inline bool mem_cgroup_lruvec_online(struct lruvec *lruvec) |
602 | { | |
603 | return true; | |
604 | } | |
605 | ||
a3d8e054 | 606 | static inline unsigned long |
4d7dcca2 | 607 | mem_cgroup_get_lru_size(struct lruvec *lruvec, enum lru_list lru) |
a3d8e054 KM |
608 | { |
609 | return 0; | |
610 | } | |
611 | ||
fa9add64 HD |
612 | static inline void |
613 | mem_cgroup_update_lru_size(struct lruvec *lruvec, enum lru_list lru, | |
614 | int increment) | |
3e2f41f1 | 615 | { |
3e2f41f1 KM |
616 | } |
617 | ||
e222432b BS |
618 | static inline void |
619 | mem_cgroup_print_oom_info(struct mem_cgroup *memcg, struct task_struct *p) | |
620 | { | |
621 | } | |
622 | ||
6de22619 | 623 | static inline struct mem_cgroup *mem_cgroup_begin_page_stat(struct page *page) |
89c06bd5 | 624 | { |
d7365e78 | 625 | return NULL; |
89c06bd5 KH |
626 | } |
627 | ||
6de22619 | 628 | static inline void mem_cgroup_end_page_stat(struct mem_cgroup *memcg) |
89c06bd5 KH |
629 | { |
630 | } | |
631 | ||
49426420 | 632 | static inline void mem_cgroup_oom_enable(void) |
519e5247 JW |
633 | { |
634 | } | |
635 | ||
49426420 | 636 | static inline void mem_cgroup_oom_disable(void) |
519e5247 JW |
637 | { |
638 | } | |
639 | ||
3812c8c8 JW |
640 | static inline bool task_in_memcg_oom(struct task_struct *p) |
641 | { | |
642 | return false; | |
643 | } | |
644 | ||
49426420 | 645 | static inline bool mem_cgroup_oom_synchronize(bool wait) |
3812c8c8 JW |
646 | { |
647 | return false; | |
648 | } | |
649 | ||
d7365e78 | 650 | static inline void mem_cgroup_inc_page_stat(struct mem_cgroup *memcg, |
68b4876d | 651 | enum mem_cgroup_stat_index idx) |
2a7106f2 GT |
652 | { |
653 | } | |
654 | ||
d7365e78 | 655 | static inline void mem_cgroup_dec_page_stat(struct mem_cgroup *memcg, |
68b4876d | 656 | enum mem_cgroup_stat_index idx) |
d69b042f BS |
657 | { |
658 | } | |
659 | ||
4e416953 | 660 | static inline |
0608f43d AM |
661 | unsigned long mem_cgroup_soft_limit_reclaim(struct zone *zone, int order, |
662 | gfp_t gfp_mask, | |
663 | unsigned long *total_scanned) | |
4e416953 | 664 | { |
0608f43d | 665 | return 0; |
4e416953 BS |
666 | } |
667 | ||
e94c8a9c | 668 | static inline void mem_cgroup_split_huge_fixup(struct page *head) |
ca3e0214 KH |
669 | { |
670 | } | |
671 | ||
456f998e YH |
672 | static inline |
673 | void mem_cgroup_count_vm_event(struct mm_struct *mm, enum vm_event_item idx) | |
674 | { | |
675 | } | |
c255a458 | 676 | #endif /* CONFIG_MEMCG */ |
78fb7466 | 677 | |
e1aab161 GC |
678 | enum { |
679 | UNDER_LIMIT, | |
680 | SOFT_LIMIT, | |
681 | OVER_LIMIT, | |
682 | }; | |
683 | ||
52ebea74 | 684 | #ifdef CONFIG_CGROUP_WRITEBACK |
841710aa | 685 | |
52ebea74 | 686 | struct list_head *mem_cgroup_cgwb_list(struct mem_cgroup *memcg); |
841710aa | 687 | struct wb_domain *mem_cgroup_wb_domain(struct bdi_writeback *wb); |
c2aa723a TH |
688 | void mem_cgroup_wb_stats(struct bdi_writeback *wb, unsigned long *pavail, |
689 | unsigned long *pdirty, unsigned long *pwriteback); | |
841710aa TH |
690 | |
691 | #else /* CONFIG_CGROUP_WRITEBACK */ | |
692 | ||
693 | static inline struct wb_domain *mem_cgroup_wb_domain(struct bdi_writeback *wb) | |
694 | { | |
695 | return NULL; | |
696 | } | |
697 | ||
c2aa723a TH |
698 | static inline void mem_cgroup_wb_stats(struct bdi_writeback *wb, |
699 | unsigned long *pavail, | |
700 | unsigned long *pdirty, | |
701 | unsigned long *pwriteback) | |
702 | { | |
703 | } | |
704 | ||
841710aa | 705 | #endif /* CONFIG_CGROUP_WRITEBACK */ |
52ebea74 | 706 | |
e1aab161 | 707 | struct sock; |
cd59085a | 708 | #if defined(CONFIG_INET) && defined(CONFIG_MEMCG_KMEM) |
e1aab161 GC |
709 | void sock_update_memcg(struct sock *sk); |
710 | void sock_release_memcg(struct sock *sk); | |
711 | #else | |
712 | static inline void sock_update_memcg(struct sock *sk) | |
713 | { | |
714 | } | |
715 | static inline void sock_release_memcg(struct sock *sk) | |
716 | { | |
717 | } | |
cd59085a | 718 | #endif /* CONFIG_INET && CONFIG_MEMCG_KMEM */ |
7ae1e1d0 GC |
719 | |
720 | #ifdef CONFIG_MEMCG_KMEM | |
a8964b9b | 721 | extern struct static_key memcg_kmem_enabled_key; |
749c5415 | 722 | |
dbcf73e2 | 723 | extern int memcg_nr_cache_ids; |
05257a1a VD |
724 | extern void memcg_get_cache_ids(void); |
725 | extern void memcg_put_cache_ids(void); | |
ebe945c2 GC |
726 | |
727 | /* | |
728 | * Helper macro to loop through all memcg-specific caches. Callers must still | |
729 | * check if the cache is valid (it is either valid or NULL). | |
730 | * the slab_mutex must be held when looping through those caches | |
731 | */ | |
749c5415 | 732 | #define for_each_memcg_cache_index(_idx) \ |
dbcf73e2 | 733 | for ((_idx) = 0; (_idx) < memcg_nr_cache_ids; (_idx)++) |
749c5415 | 734 | |
7ae1e1d0 GC |
735 | static inline bool memcg_kmem_enabled(void) |
736 | { | |
a8964b9b | 737 | return static_key_false(&memcg_kmem_enabled_key); |
7ae1e1d0 GC |
738 | } |
739 | ||
33398cf2 MH |
740 | static inline bool memcg_kmem_is_active(struct mem_cgroup *memcg) |
741 | { | |
742 | return memcg->kmem_acct_active; | |
743 | } | |
cb731d6c | 744 | |
7ae1e1d0 GC |
745 | /* |
746 | * In general, we'll do everything in our power to not incur in any overhead | |
747 | * for non-memcg users for the kmem functions. Not even a function call, if we | |
748 | * can avoid it. | |
749 | * | |
750 | * Therefore, we'll inline all those functions so that in the best case, we'll | |
751 | * see that kmemcg is off for everybody and proceed quickly. If it is on, | |
752 | * we'll still do most of the flag checking inline. We check a lot of | |
753 | * conditions, but because they are pretty simple, they are expected to be | |
754 | * fast. | |
755 | */ | |
756 | bool __memcg_kmem_newpage_charge(gfp_t gfp, struct mem_cgroup **memcg, | |
757 | int order); | |
758 | void __memcg_kmem_commit_charge(struct page *page, | |
759 | struct mem_cgroup *memcg, int order); | |
760 | void __memcg_kmem_uncharge_pages(struct page *page, int order); | |
761 | ||
33398cf2 MH |
762 | /* |
763 | * helper for acessing a memcg's index. It will be used as an index in the | |
764 | * child cache array in kmem_cache, and also to derive its name. This function | |
765 | * will return -1 when this is not a kmem-limited memcg. | |
766 | */ | |
767 | static inline int memcg_cache_id(struct mem_cgroup *memcg) | |
768 | { | |
769 | return memcg ? memcg->kmemcg_id : -1; | |
770 | } | |
5722d094 | 771 | |
8135be5a VD |
772 | struct kmem_cache *__memcg_kmem_get_cache(struct kmem_cache *cachep); |
773 | void __memcg_kmem_put_cache(struct kmem_cache *cachep); | |
d7f25f8a | 774 | |
60d3fd32 VD |
775 | struct mem_cgroup *__mem_cgroup_from_kmem(void *ptr); |
776 | ||
dbf22eb6 VD |
777 | int memcg_charge_kmem(struct mem_cgroup *memcg, gfp_t gfp, |
778 | unsigned long nr_pages); | |
779 | void memcg_uncharge_kmem(struct mem_cgroup *memcg, unsigned long nr_pages); | |
5dfb4175 | 780 | |
7ae1e1d0 GC |
781 | /** |
782 | * memcg_kmem_newpage_charge: verify if a new kmem allocation is allowed. | |
783 | * @gfp: the gfp allocation flags. | |
784 | * @memcg: a pointer to the memcg this was charged against. | |
785 | * @order: allocation order. | |
786 | * | |
787 | * returns true if the memcg where the current task belongs can hold this | |
788 | * allocation. | |
789 | * | |
790 | * We return true automatically if this allocation is not to be accounted to | |
791 | * any memcg. | |
792 | */ | |
793 | static inline bool | |
794 | memcg_kmem_newpage_charge(gfp_t gfp, struct mem_cgroup **memcg, int order) | |
795 | { | |
796 | if (!memcg_kmem_enabled()) | |
797 | return true; | |
798 | ||
8f4fc071 VD |
799 | if (gfp & __GFP_NOACCOUNT) |
800 | return true; | |
7ae1e1d0 GC |
801 | /* |
802 | * __GFP_NOFAIL allocations will move on even if charging is not | |
803 | * possible. Therefore we don't even try, and have this allocation | |
3e32cb2e JW |
804 | * unaccounted. We could in theory charge it forcibly, but we hope |
805 | * those allocations are rare, and won't be worth the trouble. | |
7ae1e1d0 | 806 | */ |
52383431 | 807 | if (gfp & __GFP_NOFAIL) |
7ae1e1d0 GC |
808 | return true; |
809 | if (in_interrupt() || (!current->mm) || (current->flags & PF_KTHREAD)) | |
810 | return true; | |
811 | ||
812 | /* If the test is dying, just let it go. */ | |
813 | if (unlikely(fatal_signal_pending(current))) | |
814 | return true; | |
815 | ||
816 | return __memcg_kmem_newpage_charge(gfp, memcg, order); | |
817 | } | |
818 | ||
819 | /** | |
820 | * memcg_kmem_uncharge_pages: uncharge pages from memcg | |
821 | * @page: pointer to struct page being freed | |
822 | * @order: allocation order. | |
7ae1e1d0 GC |
823 | */ |
824 | static inline void | |
825 | memcg_kmem_uncharge_pages(struct page *page, int order) | |
826 | { | |
827 | if (memcg_kmem_enabled()) | |
828 | __memcg_kmem_uncharge_pages(page, order); | |
829 | } | |
830 | ||
831 | /** | |
832 | * memcg_kmem_commit_charge: embeds correct memcg in a page | |
833 | * @page: pointer to struct page recently allocated | |
834 | * @memcg: the memcg structure we charged against | |
835 | * @order: allocation order. | |
836 | * | |
837 | * Needs to be called after memcg_kmem_newpage_charge, regardless of success or | |
838 | * failure of the allocation. if @page is NULL, this function will revert the | |
1306a85a | 839 | * charges. Otherwise, it will commit @page to @memcg. |
7ae1e1d0 GC |
840 | */ |
841 | static inline void | |
842 | memcg_kmem_commit_charge(struct page *page, struct mem_cgroup *memcg, int order) | |
843 | { | |
844 | if (memcg_kmem_enabled() && memcg) | |
845 | __memcg_kmem_commit_charge(page, memcg, order); | |
846 | } | |
847 | ||
d7f25f8a GC |
848 | /** |
849 | * memcg_kmem_get_cache: selects the correct per-memcg cache for allocation | |
850 | * @cachep: the original global kmem cache | |
851 | * @gfp: allocation flags. | |
852 | * | |
5dfb4175 | 853 | * All memory allocated from a per-memcg cache is charged to the owner memcg. |
d7f25f8a GC |
854 | */ |
855 | static __always_inline struct kmem_cache * | |
856 | memcg_kmem_get_cache(struct kmem_cache *cachep, gfp_t gfp) | |
857 | { | |
858 | if (!memcg_kmem_enabled()) | |
859 | return cachep; | |
8f4fc071 VD |
860 | if (gfp & __GFP_NOACCOUNT) |
861 | return cachep; | |
d7f25f8a GC |
862 | if (gfp & __GFP_NOFAIL) |
863 | return cachep; | |
864 | if (in_interrupt() || (!current->mm) || (current->flags & PF_KTHREAD)) | |
865 | return cachep; | |
866 | if (unlikely(fatal_signal_pending(current))) | |
867 | return cachep; | |
868 | ||
056b7cce | 869 | return __memcg_kmem_get_cache(cachep); |
d7f25f8a | 870 | } |
8135be5a VD |
871 | |
872 | static __always_inline void memcg_kmem_put_cache(struct kmem_cache *cachep) | |
873 | { | |
874 | if (memcg_kmem_enabled()) | |
875 | __memcg_kmem_put_cache(cachep); | |
876 | } | |
60d3fd32 VD |
877 | |
878 | static __always_inline struct mem_cgroup *mem_cgroup_from_kmem(void *ptr) | |
879 | { | |
880 | if (!memcg_kmem_enabled()) | |
881 | return NULL; | |
882 | return __mem_cgroup_from_kmem(ptr); | |
883 | } | |
7ae1e1d0 | 884 | #else |
749c5415 GC |
885 | #define for_each_memcg_cache_index(_idx) \ |
886 | for (; NULL; ) | |
887 | ||
b9ce5ef4 GC |
888 | static inline bool memcg_kmem_enabled(void) |
889 | { | |
890 | return false; | |
891 | } | |
892 | ||
cb731d6c VD |
893 | static inline bool memcg_kmem_is_active(struct mem_cgroup *memcg) |
894 | { | |
895 | return false; | |
896 | } | |
897 | ||
7ae1e1d0 GC |
898 | static inline bool |
899 | memcg_kmem_newpage_charge(gfp_t gfp, struct mem_cgroup **memcg, int order) | |
900 | { | |
901 | return true; | |
902 | } | |
903 | ||
904 | static inline void memcg_kmem_uncharge_pages(struct page *page, int order) | |
905 | { | |
906 | } | |
907 | ||
908 | static inline void | |
909 | memcg_kmem_commit_charge(struct page *page, struct mem_cgroup *memcg, int order) | |
910 | { | |
911 | } | |
2633d7a0 GC |
912 | |
913 | static inline int memcg_cache_id(struct mem_cgroup *memcg) | |
914 | { | |
915 | return -1; | |
916 | } | |
917 | ||
05257a1a VD |
918 | static inline void memcg_get_cache_ids(void) |
919 | { | |
920 | } | |
921 | ||
922 | static inline void memcg_put_cache_ids(void) | |
923 | { | |
924 | } | |
925 | ||
d7f25f8a GC |
926 | static inline struct kmem_cache * |
927 | memcg_kmem_get_cache(struct kmem_cache *cachep, gfp_t gfp) | |
928 | { | |
929 | return cachep; | |
930 | } | |
8135be5a VD |
931 | |
932 | static inline void memcg_kmem_put_cache(struct kmem_cache *cachep) | |
933 | { | |
934 | } | |
60d3fd32 VD |
935 | |
936 | static inline struct mem_cgroup *mem_cgroup_from_kmem(void *ptr) | |
937 | { | |
938 | return NULL; | |
939 | } | |
7ae1e1d0 | 940 | #endif /* CONFIG_MEMCG_KMEM */ |
8cdea7c0 BS |
941 | #endif /* _LINUX_MEMCONTROL_H */ |
942 |