Blame view
mm/memcontrol.c
145 KB
8cdea7c05
|
1 2 3 4 5 |
/* memcontrol.c - Memory Controller * * Copyright IBM Corporation, 2007 * Author Balbir Singh <balbir@linux.vnet.ibm.com> * |
78fb74669
|
6 7 8 |
* Copyright 2007 OpenVZ SWsoft Inc * Author: Pavel Emelianov <xemul@openvz.org> * |
2e72b6347
|
9 10 11 12 |
* Memory thresholds * Copyright (C) 2009 Nokia Corporation * Author: Kirill A. Shutemov * |
8cdea7c05
|
13 14 15 16 17 18 19 20 21 22 23 24 25 26 |
* This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by * the Free Software Foundation; either version 2 of the License, or * (at your option) any later version. * * This program is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the * GNU General Public License for more details. */ #include <linux/res_counter.h> #include <linux/memcontrol.h> #include <linux/cgroup.h> |
78fb74669
|
27 |
#include <linux/mm.h> |
4ffef5fef
|
28 |
#include <linux/hugetlb.h> |
d13d14430
|
29 |
#include <linux/pagemap.h> |
d52aa412d
|
30 |
#include <linux/smp.h> |
8a9f3ccd2
|
31 |
#include <linux/page-flags.h> |
66e1707bc
|
32 |
#include <linux/backing-dev.h> |
8a9f3ccd2
|
33 34 |
#include <linux/bit_spinlock.h> #include <linux/rcupdate.h> |
e222432bf
|
35 |
#include <linux/limits.h> |
b9e15bafd
|
36 |
#include <linux/export.h> |
8c7c6e34a
|
37 |
#include <linux/mutex.h> |
f64c3f549
|
38 |
#include <linux/rbtree.h> |
b6ac57d50
|
39 |
#include <linux/slab.h> |
66e1707bc
|
40 |
#include <linux/swap.h> |
024914477
|
41 |
#include <linux/swapops.h> |
66e1707bc
|
42 |
#include <linux/spinlock.h> |
2e72b6347
|
43 44 |
#include <linux/eventfd.h> #include <linux/sort.h> |
66e1707bc
|
45 |
#include <linux/fs.h> |
d2ceb9b7d
|
46 |
#include <linux/seq_file.h> |
333279487
|
47 |
#include <linux/vmalloc.h> |
b69408e88
|
48 |
#include <linux/mm_inline.h> |
52d4b9ac0
|
49 |
#include <linux/page_cgroup.h> |
cdec2e426
|
50 |
#include <linux/cpu.h> |
158e0a2d1
|
51 |
#include <linux/oom.h> |
08e552c69
|
52 |
#include "internal.h" |
d1a4c0b37
|
53 54 |
#include <net/sock.h> #include <net/tcp_memcontrol.h> |
8cdea7c05
|
55 |
|
8697d3319
|
56 |
#include <asm/uaccess.h> |
cc8e970c3
|
57 |
#include <trace/events/vmscan.h> |
a181b0e88
|
58 |
struct cgroup_subsys mem_cgroup_subsys __read_mostly; |
a181b0e88
|
59 |
#define MEM_CGROUP_RECLAIM_RETRIES 5 |
6bbda35ce
|
60 |
static struct mem_cgroup *root_mem_cgroup __read_mostly; |
8cdea7c05
|
61 |
|
c255a4580
|
62 |
#ifdef CONFIG_MEMCG_SWAP |
338c84310
|
63 |
/* Turned on only when memory cgroup is enabled && really_do_swap_account = 1 */ |
c077719be
|
64 |
int do_swap_account __read_mostly; |
a42c390cf
|
65 66 |
/* for remember boot option*/ |
c255a4580
|
67 |
#ifdef CONFIG_MEMCG_SWAP_ENABLED |
a42c390cf
|
68 69 70 71 |
static int really_do_swap_account __initdata = 1; #else static int really_do_swap_account __initdata = 0; #endif |
c077719be
|
72 |
#else |
a0db00fcf
|
73 |
#define do_swap_account 0 |
c077719be
|
74 |
#endif |
8cdea7c05
|
75 |
/* |
d52aa412d
|
76 77 78 79 80 81 82 |
* Statistics for memory cgroup. */ enum mem_cgroup_stat_index { /* * For MEM_CONTAINER_TYPE_ALL, usage = pagecache + rss. */ MEM_CGROUP_STAT_CACHE, /* # of pages charged as cache */ |
d69b042f3
|
83 |
MEM_CGROUP_STAT_RSS, /* # of pages charged as anon rss */ |
d8046582d
|
84 |
MEM_CGROUP_STAT_FILE_MAPPED, /* # of pages charged as file rss */ |
bff6bb83f
|
85 |
MEM_CGROUP_STAT_SWAP, /* # of pages, swapped out */ |
d52aa412d
|
86 87 |
MEM_CGROUP_STAT_NSTATS, }; |
af7c4b0ec
|
88 89 90 91 92 93 |
static const char * const mem_cgroup_stat_names[] = { "cache", "rss", "mapped_file", "swap", }; |
e9f8974f2
|
94 95 96 |
enum mem_cgroup_events_index { MEM_CGROUP_EVENTS_PGPGIN, /* # of pages paged in */ MEM_CGROUP_EVENTS_PGPGOUT, /* # of pages paged out */ |
456f998ec
|
97 98 |
MEM_CGROUP_EVENTS_PGFAULT, /* # of page-faults */ MEM_CGROUP_EVENTS_PGMAJFAULT, /* # of major page-faults */ |
e9f8974f2
|
99 100 |
MEM_CGROUP_EVENTS_NSTATS, }; |
af7c4b0ec
|
101 102 103 104 105 106 107 |
static const char * const mem_cgroup_events_names[] = { "pgpgin", "pgpgout", "pgfault", "pgmajfault", }; |
7a159cc9d
|
108 109 110 111 112 113 114 115 116 |
/* * Per memcg event counter is incremented at every pagein/pageout. With THP, * it will be incremated by the number of pages. This counter is used for * for trigger some periodic events. This is straightforward and better * than using jiffies etc. to handle periodic memcg event. */ enum mem_cgroup_events_target { MEM_CGROUP_TARGET_THRESH, MEM_CGROUP_TARGET_SOFTLIMIT, |
453a9bf34
|
117 |
MEM_CGROUP_TARGET_NUMAINFO, |
7a159cc9d
|
118 119 |
MEM_CGROUP_NTARGETS, }; |
a0db00fcf
|
120 121 122 |
#define THRESHOLDS_EVENTS_TARGET 128 #define SOFTLIMIT_EVENTS_TARGET 1024 #define NUMAINFO_EVENTS_TARGET 1024 |
e9f8974f2
|
123 |
|
d52aa412d
|
124 |
struct mem_cgroup_stat_cpu { |
7a159cc9d
|
125 |
long count[MEM_CGROUP_STAT_NSTATS]; |
e9f8974f2
|
126 |
unsigned long events[MEM_CGROUP_EVENTS_NSTATS]; |
13114716c
|
127 |
unsigned long nr_page_events; |
7a159cc9d
|
128 |
unsigned long targets[MEM_CGROUP_NTARGETS]; |
d52aa412d
|
129 |
}; |
527a5ec9a
|
130 131 132 133 134 135 |
struct mem_cgroup_reclaim_iter { /* css_id of the last scanned hierarchy member */ int position; /* scan generation, increased every round-trip */ unsigned int generation; }; |
d52aa412d
|
136 |
/* |
6d12e2d8d
|
137 138 |
* per-zone information in memory controller. */ |
6d12e2d8d
|
139 |
struct mem_cgroup_per_zone { |
6290df545
|
140 |
struct lruvec lruvec; |
1eb492725
|
141 |
unsigned long lru_size[NR_LRU_LISTS]; |
3e2f41f1f
|
142 |
|
527a5ec9a
|
143 |
struct mem_cgroup_reclaim_iter reclaim_iter[DEF_PRIORITY + 1]; |
f64c3f549
|
144 145 146 147 |
struct rb_node tree_node; /* RB tree node */ unsigned long long usage_in_excess;/* Set to the value by which */ /* the soft limit is exceeded*/ bool on_tree; |
d79154bb5
|
148 |
struct mem_cgroup *memcg; /* Back pointer, we cannot */ |
4e4169535
|
149 |
/* use container_of */ |
6d12e2d8d
|
150 |
}; |
6d12e2d8d
|
151 152 153 154 155 156 157 158 159 160 |
struct mem_cgroup_per_node { struct mem_cgroup_per_zone zoneinfo[MAX_NR_ZONES]; }; struct mem_cgroup_lru_info { struct mem_cgroup_per_node *nodeinfo[MAX_NUMNODES]; }; /* |
f64c3f549
|
161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 |
* Cgroups above their limits are maintained in a RB-Tree, independent of * their hierarchy representation */ struct mem_cgroup_tree_per_zone { struct rb_root rb_root; spinlock_t lock; }; struct mem_cgroup_tree_per_node { struct mem_cgroup_tree_per_zone rb_tree_per_zone[MAX_NR_ZONES]; }; struct mem_cgroup_tree { struct mem_cgroup_tree_per_node *rb_tree_per_node[MAX_NUMNODES]; }; static struct mem_cgroup_tree soft_limit_tree __read_mostly; |
2e72b6347
|
179 180 181 182 |
struct mem_cgroup_threshold { struct eventfd_ctx *eventfd; u64 threshold; }; |
9490ff275
|
183 |
/* For threshold */ |
2e72b6347
|
184 |
struct mem_cgroup_threshold_ary { |
748dad36d
|
185 |
/* An array index points to threshold just below or equal to usage. */ |
5407a5625
|
186 |
int current_threshold; |
2e72b6347
|
187 188 189 190 191 |
/* Size of entries[] */ unsigned int size; /* Array of thresholds */ struct mem_cgroup_threshold entries[0]; }; |
2c488db27
|
192 193 194 195 196 197 198 199 200 201 202 |
struct mem_cgroup_thresholds { /* Primary thresholds array */ struct mem_cgroup_threshold_ary *primary; /* * Spare threshold array. * This is needed to make mem_cgroup_unregister_event() "never fail". * It must be able to store at least primary->size - 1 entries. */ struct mem_cgroup_threshold_ary *spare; }; |
9490ff275
|
203 204 205 206 207 |
/* for OOM */ struct mem_cgroup_eventfd_list { struct list_head list; struct eventfd_ctx *eventfd; }; |
2e72b6347
|
208 |
|
c0ff4b854
|
209 210 |
static void mem_cgroup_threshold(struct mem_cgroup *memcg); static void mem_cgroup_oom_notify(struct mem_cgroup *memcg); |
2e72b6347
|
211 |
|
f64c3f549
|
212 |
/* |
8cdea7c05
|
213 214 215 216 217 218 |
* The memory controller data structure. The memory controller controls both * page cache and RSS per cgroup. We would eventually like to provide * statistics based on the statistics developed by Rik Van Riel for clock-pro, * to help the administrator determine what knobs to tune. * * TODO: Add a water mark for the memory controller. Reclaim will begin when |
8a9f3ccd2
|
219 220 221 |
* we hit the water mark. May be even add a low water mark, such that * no reclaim occurs from a cgroup at it's low water mark, this is * a feature that will be implemented much later in the future. |
8cdea7c05
|
222 223 224 225 226 227 228 |
*/ struct mem_cgroup { struct cgroup_subsys_state css; /* * the counter to account for memory usage */ struct res_counter res; |
59927fb98
|
229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 |
union { /* * the counter to account for mem+swap usage. */ struct res_counter memsw; /* * rcu_freeing is used only when freeing struct mem_cgroup, * so put it into a union to avoid wasting more memory. * It must be disjoint from the css field. It could be * in a union with the res field, but res plays a much * larger part in mem_cgroup life than memsw, and might * be of interest, even at time of free, when debugging. * So share rcu_head with the less interesting memsw. */ struct rcu_head rcu_freeing; /* |
3afe36b1f
|
247 248 |
* We also need some space for a worker in deferred freeing. * By the time we call it, rcu_freeing is no longer in use. |
59927fb98
|
249 250 251 |
*/ struct work_struct work_freeing; }; |
8c7c6e34a
|
252 |
/* |
78fb74669
|
253 254 |
* Per cgroup active and inactive list, similar to the * per zone LRU lists. |
78fb74669
|
255 |
*/ |
6d12e2d8d
|
256 |
struct mem_cgroup_lru_info info; |
889976dbc
|
257 258 259 |
int last_scanned_node; #if MAX_NUMNODES > 1 nodemask_t scan_nodes; |
453a9bf34
|
260 261 |
atomic_t numainfo_events; atomic_t numainfo_updating; |
889976dbc
|
262 |
#endif |
18f59ea7d
|
263 264 265 266 |
/* * Should the accounting and control be hierarchical, per subtree? */ bool use_hierarchy; |
79dfdaccd
|
267 268 269 |
bool oom_lock; atomic_t under_oom; |
8c7c6e34a
|
270 |
atomic_t refcnt; |
14797e236
|
271 |
|
1f4c025b5
|
272 |
int swappiness; |
3c11ecf44
|
273 274 |
/* OOM-Killer disable */ int oom_kill_disable; |
a7885eb8a
|
275 |
|
22a668d7c
|
276 277 |
/* set when res.limit == memsw.limit */ bool memsw_is_minimum; |
2e72b6347
|
278 279 280 281 |
/* protect arrays of thresholds */ struct mutex thresholds_lock; /* thresholds for memory usage. RCU-protected */ |
2c488db27
|
282 |
struct mem_cgroup_thresholds thresholds; |
907860ed3
|
283 |
|
2e72b6347
|
284 |
/* thresholds for mem+swap usage. RCU-protected */ |
2c488db27
|
285 |
struct mem_cgroup_thresholds memsw_thresholds; |
907860ed3
|
286 |
|
9490ff275
|
287 288 |
/* For oom notifier event fd */ struct list_head oom_notify; |
185efc0f9
|
289 |
|
d52aa412d
|
290 |
/* |
7dc74be03
|
291 292 293 294 |
* Should we move charges of a task when a task is moved into this * mem_cgroup ? And what type of charges should we move ? */ unsigned long move_charge_at_immigrate; |
7dc74be03
|
295 |
/* |
619d094b5
|
296 297 298 |
* set > 0 if pages under this cgroup are moving to other cgroup. */ atomic_t moving_account; |
312734c04
|
299 300 |
/* taken only while moving_account > 0 */ spinlock_t move_lock; |
619d094b5
|
301 |
/* |
c62b1a3b3
|
302 |
* percpu counter. |
d52aa412d
|
303 |
*/ |
3a7951b4c
|
304 |
struct mem_cgroup_stat_cpu __percpu *stat; |
711d3d2c9
|
305 306 307 308 309 310 |
/* * used when a cpu is offlined or other synchronizations * See mem_cgroup_read_stat(). */ struct mem_cgroup_stat_cpu nocpu_base; spinlock_t pcp_counter_lock; |
d1a4c0b37
|
311 312 313 314 |
#ifdef CONFIG_INET struct tcp_memcontrol tcp_mem; #endif |
8cdea7c05
|
315 |
}; |
7dc74be03
|
316 317 318 319 320 321 |
/* Stuffs for move charges at task migration. */ /* * Types of charges to be moved. "move_charge_at_immitgrate" is treated as a * left-shifted bitmap of these types. */ enum move_type { |
4ffef5fef
|
322 |
MOVE_CHARGE_TYPE_ANON, /* private anonymous page and swap of it */ |
87946a722
|
323 |
MOVE_CHARGE_TYPE_FILE, /* file page(including tmpfs) and swap of it */ |
7dc74be03
|
324 325 |
NR_MOVE_TYPE, }; |
4ffef5fef
|
326 327 |
/* "mc" and its members are protected by cgroup_mutex */ static struct move_charge_struct { |
b1dd693e5
|
328 |
spinlock_t lock; /* for from, to */ |
4ffef5fef
|
329 330 331 |
struct mem_cgroup *from; struct mem_cgroup *to; unsigned long precharge; |
854ffa8d1
|
332 |
unsigned long moved_charge; |
483c30b51
|
333 |
unsigned long moved_swap; |
8033b97c9
|
334 335 336 |
struct task_struct *moving_task; /* a task moving charges */ wait_queue_head_t waitq; /* a waitq for other context */ } mc = { |
2bd9bb206
|
337 |
.lock = __SPIN_LOCK_UNLOCKED(mc.lock), |
8033b97c9
|
338 339 |
.waitq = __WAIT_QUEUE_HEAD_INITIALIZER(mc.waitq), }; |
4ffef5fef
|
340 |
|
90254a658
|
341 342 343 344 345 |
static bool move_anon(void) { return test_bit(MOVE_CHARGE_TYPE_ANON, &mc.to->move_charge_at_immigrate); } |
87946a722
|
346 347 348 349 350 |
static bool move_file(void) { return test_bit(MOVE_CHARGE_TYPE_FILE, &mc.to->move_charge_at_immigrate); } |
4e4169535
|
351 352 353 354 |
/* * Maximum loops in mem_cgroup_hierarchical_reclaim(), used for soft * limit reclaim to prevent infinite loops, if they ever occur. */ |
a0db00fcf
|
355 356 |
#define MEM_CGROUP_MAX_RECLAIM_LOOPS 100 #define MEM_CGROUP_MAX_SOFT_LIMIT_RECLAIM_LOOPS 2 |
4e4169535
|
357 |
|
217bc3194
|
358 359 |
enum charge_type { MEM_CGROUP_CHARGE_TYPE_CACHE = 0, |
41326c17f
|
360 |
MEM_CGROUP_CHARGE_TYPE_ANON, |
d13d14430
|
361 |
MEM_CGROUP_CHARGE_TYPE_SWAPOUT, /* for accounting swapcache */ |
8a9478ca7
|
362 |
MEM_CGROUP_CHARGE_TYPE_DROP, /* a page was unused swap cache */ |
c05555b57
|
363 364 |
NR_CHARGE_TYPE, }; |
8c7c6e34a
|
365 |
/* for encoding cft->private value on file */ |
65c64ce8e
|
366 367 368 |
#define _MEM (0) #define _MEMSWAP (1) #define _OOM_TYPE (2) |
a0db00fcf
|
369 370 |
#define MEMFILE_PRIVATE(x, val) ((x) << 16 | (val)) #define MEMFILE_TYPE(val) ((val) >> 16 & 0xffff) |
8c7c6e34a
|
371 |
#define MEMFILE_ATTR(val) ((val) & 0xffff) |
9490ff275
|
372 373 |
/* Used for OOM nofiier */ #define OOM_CONTROL (0) |
8c7c6e34a
|
374 |
|
75822b449
|
375 376 377 378 379 380 381 |
/* * Reclaim flags for mem_cgroup_hierarchical_reclaim */ #define MEM_CGROUP_RECLAIM_NOSWAP_BIT 0x0 #define MEM_CGROUP_RECLAIM_NOSWAP (1 << MEM_CGROUP_RECLAIM_NOSWAP_BIT) #define MEM_CGROUP_RECLAIM_SHRINK_BIT 0x1 #define MEM_CGROUP_RECLAIM_SHRINK (1 << MEM_CGROUP_RECLAIM_SHRINK_BIT) |
c0ff4b854
|
382 383 |
static void mem_cgroup_get(struct mem_cgroup *memcg); static void mem_cgroup_put(struct mem_cgroup *memcg); |
e1aab161e
|
384 |
|
b21451459
|
385 386 387 388 389 |
static inline struct mem_cgroup *mem_cgroup_from_css(struct cgroup_subsys_state *s) { return container_of(s, struct mem_cgroup, css); } |
e1aab161e
|
390 |
/* Writing them here to avoid exposing memcg's inner layout */ |
c255a4580
|
391 |
#ifdef CONFIG_MEMCG_KMEM |
e1aab161e
|
392 |
#include <net/sock.h> |
d1a4c0b37
|
393 |
#include <net/ip.h> |
e1aab161e
|
394 395 396 397 |
static bool mem_cgroup_is_root(struct mem_cgroup *memcg); void sock_update_memcg(struct sock *sk) { |
376be5ff8
|
398 |
if (mem_cgroup_sockets_enabled) { |
e1aab161e
|
399 |
struct mem_cgroup *memcg; |
3f1346193
|
400 |
struct cg_proto *cg_proto; |
e1aab161e
|
401 402 |
BUG_ON(!sk->sk_prot->proto_cgroup); |
f3f511e1c
|
403 404 405 406 407 408 409 410 411 412 413 414 415 |
/* Socket cloning can throw us here with sk_cgrp already * filled. It won't however, necessarily happen from * process context. So the test for root memcg given * the current task's memcg won't help us in this case. * * Respecting the original socket's memcg is a better * decision in this case. */ if (sk->sk_cgrp) { BUG_ON(mem_cgroup_is_root(sk->sk_cgrp->memcg)); mem_cgroup_get(sk->sk_cgrp->memcg); return; } |
e1aab161e
|
416 417 |
rcu_read_lock(); memcg = mem_cgroup_from_task(current); |
3f1346193
|
418 419 |
cg_proto = sk->sk_prot->proto_cgroup(memcg); if (!mem_cgroup_is_root(memcg) && memcg_proto_active(cg_proto)) { |
e1aab161e
|
420 |
mem_cgroup_get(memcg); |
3f1346193
|
421 |
sk->sk_cgrp = cg_proto; |
e1aab161e
|
422 423 424 425 426 427 428 429 |
} rcu_read_unlock(); } } EXPORT_SYMBOL(sock_update_memcg); void sock_release_memcg(struct sock *sk) { |
376be5ff8
|
430 |
if (mem_cgroup_sockets_enabled && sk->sk_cgrp) { |
e1aab161e
|
431 432 433 434 435 436 |
struct mem_cgroup *memcg; WARN_ON(!sk->sk_cgrp->memcg); memcg = sk->sk_cgrp->memcg; mem_cgroup_put(memcg); } } |
d1a4c0b37
|
437 |
|
319d3b9c9
|
438 |
#ifdef CONFIG_INET |
d1a4c0b37
|
439 440 441 442 443 444 445 446 |
struct cg_proto *tcp_proto_cgroup(struct mem_cgroup *memcg) { if (!memcg || mem_cgroup_is_root(memcg)) return NULL; return &memcg->tcp_mem.cg_proto; } EXPORT_SYMBOL(tcp_proto_cgroup); |
e1aab161e
|
447 |
#endif /* CONFIG_INET */ |
c255a4580
|
448 |
#endif /* CONFIG_MEMCG_KMEM */ |
e1aab161e
|
449 |
|
c255a4580
|
450 |
#if defined(CONFIG_INET) && defined(CONFIG_MEMCG_KMEM) |
3f1346193
|
451 452 453 454 455 456 457 458 459 460 461 |
static void disarm_sock_keys(struct mem_cgroup *memcg) { if (!memcg_proto_activated(&memcg->tcp_mem.cg_proto)) return; static_key_slow_dec(&memcg_socket_limit_enabled); } #else static void disarm_sock_keys(struct mem_cgroup *memcg) { } #endif |
c0ff4b854
|
462 |
static void drain_all_stock_async(struct mem_cgroup *memcg); |
8c7c6e34a
|
463 |
|
f64c3f549
|
464 |
static struct mem_cgroup_per_zone * |
c0ff4b854
|
465 |
mem_cgroup_zoneinfo(struct mem_cgroup *memcg, int nid, int zid) |
f64c3f549
|
466 |
{ |
c0ff4b854
|
467 |
return &memcg->info.nodeinfo[nid]->zoneinfo[zid]; |
f64c3f549
|
468 |
} |
c0ff4b854
|
469 |
struct cgroup_subsys_state *mem_cgroup_css(struct mem_cgroup *memcg) |
d324236b3
|
470 |
{ |
c0ff4b854
|
471 |
return &memcg->css; |
d324236b3
|
472 |
} |
f64c3f549
|
473 |
static struct mem_cgroup_per_zone * |
c0ff4b854
|
474 |
page_cgroup_zoneinfo(struct mem_cgroup *memcg, struct page *page) |
f64c3f549
|
475 |
{ |
97a6c37b3
|
476 477 |
int nid = page_to_nid(page); int zid = page_zonenum(page); |
f64c3f549
|
478 |
|
c0ff4b854
|
479 |
return mem_cgroup_zoneinfo(memcg, nid, zid); |
f64c3f549
|
480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 |
} static struct mem_cgroup_tree_per_zone * soft_limit_tree_node_zone(int nid, int zid) { return &soft_limit_tree.rb_tree_per_node[nid]->rb_tree_per_zone[zid]; } static struct mem_cgroup_tree_per_zone * soft_limit_tree_from_page(struct page *page) { int nid = page_to_nid(page); int zid = page_zonenum(page); return &soft_limit_tree.rb_tree_per_node[nid]->rb_tree_per_zone[zid]; } static void |
c0ff4b854
|
498 |
__mem_cgroup_insert_exceeded(struct mem_cgroup *memcg, |
f64c3f549
|
499 |
struct mem_cgroup_per_zone *mz, |
ef8745c1e
|
500 501 |
struct mem_cgroup_tree_per_zone *mctz, unsigned long long new_usage_in_excess) |
f64c3f549
|
502 503 504 505 506 507 508 |
{ struct rb_node **p = &mctz->rb_root.rb_node; struct rb_node *parent = NULL; struct mem_cgroup_per_zone *mz_node; if (mz->on_tree) return; |
ef8745c1e
|
509 510 511 |
mz->usage_in_excess = new_usage_in_excess; if (!mz->usage_in_excess) return; |
f64c3f549
|
512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 |
while (*p) { parent = *p; mz_node = rb_entry(parent, struct mem_cgroup_per_zone, tree_node); if (mz->usage_in_excess < mz_node->usage_in_excess) p = &(*p)->rb_left; /* * We can't avoid mem cgroups that are over their soft * limit by the same amount */ else if (mz->usage_in_excess >= mz_node->usage_in_excess) p = &(*p)->rb_right; } rb_link_node(&mz->tree_node, parent, p); rb_insert_color(&mz->tree_node, &mctz->rb_root); mz->on_tree = true; |
4e4169535
|
528 529 530 |
} static void |
c0ff4b854
|
531 |
__mem_cgroup_remove_exceeded(struct mem_cgroup *memcg, |
4e4169535
|
532 533 534 535 536 537 538 539 540 541 |
struct mem_cgroup_per_zone *mz, struct mem_cgroup_tree_per_zone *mctz) { if (!mz->on_tree) return; rb_erase(&mz->tree_node, &mctz->rb_root); mz->on_tree = false; } static void |
c0ff4b854
|
542 |
mem_cgroup_remove_exceeded(struct mem_cgroup *memcg, |
f64c3f549
|
543 544 545 546 |
struct mem_cgroup_per_zone *mz, struct mem_cgroup_tree_per_zone *mctz) { spin_lock(&mctz->lock); |
c0ff4b854
|
547 |
__mem_cgroup_remove_exceeded(memcg, mz, mctz); |
f64c3f549
|
548 549 |
spin_unlock(&mctz->lock); } |
f64c3f549
|
550 |
|
c0ff4b854
|
551 |
static void mem_cgroup_update_tree(struct mem_cgroup *memcg, struct page *page) |
f64c3f549
|
552 |
{ |
ef8745c1e
|
553 |
unsigned long long excess; |
f64c3f549
|
554 555 |
struct mem_cgroup_per_zone *mz; struct mem_cgroup_tree_per_zone *mctz; |
4e649152c
|
556 557 |
int nid = page_to_nid(page); int zid = page_zonenum(page); |
f64c3f549
|
558 559 560 |
mctz = soft_limit_tree_from_page(page); /* |
4e649152c
|
561 562 |
* Necessary to update all ancestors when hierarchy is used. * because their event counter is not touched. |
f64c3f549
|
563 |
*/ |
c0ff4b854
|
564 565 566 |
for (; memcg; memcg = parent_mem_cgroup(memcg)) { mz = mem_cgroup_zoneinfo(memcg, nid, zid); excess = res_counter_soft_limit_excess(&memcg->res); |
4e649152c
|
567 568 569 570 |
/* * We have to update the tree if mz is on RB-tree or * mem is over its softlimit. */ |
ef8745c1e
|
571 |
if (excess || mz->on_tree) { |
4e649152c
|
572 573 574 |
spin_lock(&mctz->lock); /* if on-tree, remove it */ if (mz->on_tree) |
c0ff4b854
|
575 |
__mem_cgroup_remove_exceeded(memcg, mz, mctz); |
4e649152c
|
576 |
/* |
ef8745c1e
|
577 578 |
* Insert again. mz->usage_in_excess will be updated. * If excess is 0, no tree ops. |
4e649152c
|
579 |
*/ |
c0ff4b854
|
580 |
__mem_cgroup_insert_exceeded(memcg, mz, mctz, excess); |
4e649152c
|
581 582 |
spin_unlock(&mctz->lock); } |
f64c3f549
|
583 584 |
} } |
c0ff4b854
|
585 |
static void mem_cgroup_remove_from_trees(struct mem_cgroup *memcg) |
f64c3f549
|
586 587 588 589 |
{ int node, zone; struct mem_cgroup_per_zone *mz; struct mem_cgroup_tree_per_zone *mctz; |
3ed28fa10
|
590 |
for_each_node(node) { |
f64c3f549
|
591 |
for (zone = 0; zone < MAX_NR_ZONES; zone++) { |
c0ff4b854
|
592 |
mz = mem_cgroup_zoneinfo(memcg, node, zone); |
f64c3f549
|
593 |
mctz = soft_limit_tree_node_zone(node, zone); |
c0ff4b854
|
594 |
mem_cgroup_remove_exceeded(memcg, mz, mctz); |
f64c3f549
|
595 596 597 |
} } } |
4e4169535
|
598 599 600 601 |
static struct mem_cgroup_per_zone * __mem_cgroup_largest_soft_limit_node(struct mem_cgroup_tree_per_zone *mctz) { struct rb_node *rightmost = NULL; |
26251eaf9
|
602 |
struct mem_cgroup_per_zone *mz; |
4e4169535
|
603 604 |
retry: |
26251eaf9
|
605 |
mz = NULL; |
4e4169535
|
606 607 608 609 610 611 612 613 614 615 |
rightmost = rb_last(&mctz->rb_root); if (!rightmost) goto done; /* Nothing to reclaim from */ mz = rb_entry(rightmost, struct mem_cgroup_per_zone, tree_node); /* * Remove the node now but someone else can add it back, * we will to add it back at the end of reclaim to its correct * position in the tree. */ |
d79154bb5
|
616 617 618 |
__mem_cgroup_remove_exceeded(mz->memcg, mz, mctz); if (!res_counter_soft_limit_excess(&mz->memcg->res) || !css_tryget(&mz->memcg->css)) |
4e4169535
|
619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 |
goto retry; done: return mz; } static struct mem_cgroup_per_zone * mem_cgroup_largest_soft_limit_node(struct mem_cgroup_tree_per_zone *mctz) { struct mem_cgroup_per_zone *mz; spin_lock(&mctz->lock); mz = __mem_cgroup_largest_soft_limit_node(mctz); spin_unlock(&mctz->lock); return mz; } |
711d3d2c9
|
634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 |
/* * Implementation Note: reading percpu statistics for memcg. * * Both of vmstat[] and percpu_counter has threshold and do periodic * synchronization to implement "quick" read. There are trade-off between * reading cost and precision of value. Then, we may have a chance to implement * a periodic synchronizion of counter in memcg's counter. * * But this _read() function is used for user interface now. The user accounts * memory usage by memory cgroup and he _always_ requires exact value because * he accounts memory. Even if we provide quick-and-fuzzy read, we always * have to visit all online cpus and make sum. So, for now, unnecessary * synchronization is not implemented. (just implemented for cpu hotplug) * * If there are kernel internal actions which can make use of some not-exact * value, and reading all cpu value can be performance bottleneck in some * common workload, threashold and synchonization as vmstat[] should be * implemented. */ |
c0ff4b854
|
653 |
static long mem_cgroup_read_stat(struct mem_cgroup *memcg, |
7a159cc9d
|
654 |
enum mem_cgroup_stat_index idx) |
c62b1a3b3
|
655 |
{ |
7a159cc9d
|
656 |
long val = 0; |
c62b1a3b3
|
657 |
int cpu; |
c62b1a3b3
|
658 |
|
711d3d2c9
|
659 660 |
get_online_cpus(); for_each_online_cpu(cpu) |
c0ff4b854
|
661 |
val += per_cpu(memcg->stat->count[idx], cpu); |
711d3d2c9
|
662 |
#ifdef CONFIG_HOTPLUG_CPU |
c0ff4b854
|
663 664 665 |
spin_lock(&memcg->pcp_counter_lock); val += memcg->nocpu_base.count[idx]; spin_unlock(&memcg->pcp_counter_lock); |
711d3d2c9
|
666 667 |
#endif put_online_cpus(); |
c62b1a3b3
|
668 669 |
return val; } |
c0ff4b854
|
670 |
static void mem_cgroup_swap_statistics(struct mem_cgroup *memcg, |
0c3e73e84
|
671 672 673 |
bool charge) { int val = (charge) ? 1 : -1; |
bff6bb83f
|
674 |
this_cpu_add(memcg->stat->count[MEM_CGROUP_STAT_SWAP], val); |
0c3e73e84
|
675 |
} |
c0ff4b854
|
676 |
static unsigned long mem_cgroup_read_events(struct mem_cgroup *memcg, |
e9f8974f2
|
677 678 679 680 681 682 |
enum mem_cgroup_events_index idx) { unsigned long val = 0; int cpu; for_each_online_cpu(cpu) |
c0ff4b854
|
683 |
val += per_cpu(memcg->stat->events[idx], cpu); |
e9f8974f2
|
684 |
#ifdef CONFIG_HOTPLUG_CPU |
c0ff4b854
|
685 686 687 |
spin_lock(&memcg->pcp_counter_lock); val += memcg->nocpu_base.events[idx]; spin_unlock(&memcg->pcp_counter_lock); |
e9f8974f2
|
688 689 690 |
#endif return val; } |
c0ff4b854
|
691 |
static void mem_cgroup_charge_statistics(struct mem_cgroup *memcg, |
b24028572
|
692 |
bool anon, int nr_pages) |
d52aa412d
|
693 |
{ |
c62b1a3b3
|
694 |
preempt_disable(); |
b24028572
|
695 696 697 698 699 700 |
/* * Here, RSS means 'mapped anon' and anon's SwapCache. Shmem/tmpfs is * counted as CACHE even if it's on ANON LRU. */ if (anon) __this_cpu_add(memcg->stat->count[MEM_CGROUP_STAT_RSS], |
c0ff4b854
|
701 |
nr_pages); |
d52aa412d
|
702 |
else |
b24028572
|
703 |
__this_cpu_add(memcg->stat->count[MEM_CGROUP_STAT_CACHE], |
c0ff4b854
|
704 |
nr_pages); |
55e462b05
|
705 |
|
e401f1761
|
706 707 |
/* pagein of a big page is an event. So, ignore page size */ if (nr_pages > 0) |
c0ff4b854
|
708 |
__this_cpu_inc(memcg->stat->events[MEM_CGROUP_EVENTS_PGPGIN]); |
3751d6043
|
709 |
else { |
c0ff4b854
|
710 |
__this_cpu_inc(memcg->stat->events[MEM_CGROUP_EVENTS_PGPGOUT]); |
3751d6043
|
711 712 |
nr_pages = -nr_pages; /* for event */ } |
e401f1761
|
713 |
|
13114716c
|
714 |
__this_cpu_add(memcg->stat->nr_page_events, nr_pages); |
2e72b6347
|
715 |
|
c62b1a3b3
|
716 |
preempt_enable(); |
6d12e2d8d
|
717 |
} |
bb2a0de92
|
718 |
unsigned long |
4d7dcca21
|
719 |
mem_cgroup_get_lru_size(struct lruvec *lruvec, enum lru_list lru) |
074291fea
|
720 721 722 723 724 725 726 727 |
{ struct mem_cgroup_per_zone *mz; mz = container_of(lruvec, struct mem_cgroup_per_zone, lruvec); return mz->lru_size[lru]; } static unsigned long |
c0ff4b854
|
728 |
mem_cgroup_zone_nr_lru_pages(struct mem_cgroup *memcg, int nid, int zid, |
bb2a0de92
|
729 |
unsigned int lru_mask) |
889976dbc
|
730 731 |
{ struct mem_cgroup_per_zone *mz; |
f156ab933
|
732 |
enum lru_list lru; |
bb2a0de92
|
733 |
unsigned long ret = 0; |
c0ff4b854
|
734 |
mz = mem_cgroup_zoneinfo(memcg, nid, zid); |
bb2a0de92
|
735 |
|
f156ab933
|
736 737 738 |
for_each_lru(lru) { if (BIT(lru) & lru_mask) ret += mz->lru_size[lru]; |
bb2a0de92
|
739 740 741 742 743 |
} return ret; } static unsigned long |
c0ff4b854
|
744 |
mem_cgroup_node_nr_lru_pages(struct mem_cgroup *memcg, |
bb2a0de92
|
745 746 |
int nid, unsigned int lru_mask) { |
889976dbc
|
747 748 |
u64 total = 0; int zid; |
bb2a0de92
|
749 |
for (zid = 0; zid < MAX_NR_ZONES; zid++) |
c0ff4b854
|
750 751 |
total += mem_cgroup_zone_nr_lru_pages(memcg, nid, zid, lru_mask); |
bb2a0de92
|
752 |
|
889976dbc
|
753 754 |
return total; } |
bb2a0de92
|
755 |
|
c0ff4b854
|
756 |
static unsigned long mem_cgroup_nr_lru_pages(struct mem_cgroup *memcg, |
bb2a0de92
|
757 |
unsigned int lru_mask) |
6d12e2d8d
|
758 |
{ |
889976dbc
|
759 |
int nid; |
6d12e2d8d
|
760 |
u64 total = 0; |
bb2a0de92
|
761 |
for_each_node_state(nid, N_HIGH_MEMORY) |
c0ff4b854
|
762 |
total += mem_cgroup_node_nr_lru_pages(memcg, nid, lru_mask); |
6d12e2d8d
|
763 |
return total; |
d52aa412d
|
764 |
} |
f53d7ce32
|
765 766 |
static bool mem_cgroup_event_ratelimit(struct mem_cgroup *memcg, enum mem_cgroup_events_target target) |
7a159cc9d
|
767 768 |
{ unsigned long val, next; |
13114716c
|
769 |
val = __this_cpu_read(memcg->stat->nr_page_events); |
4799401fe
|
770 |
next = __this_cpu_read(memcg->stat->targets[target]); |
7a159cc9d
|
771 |
/* from time_after() in jiffies.h */ |
f53d7ce32
|
772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 |
if ((long)next - (long)val < 0) { switch (target) { case MEM_CGROUP_TARGET_THRESH: next = val + THRESHOLDS_EVENTS_TARGET; break; case MEM_CGROUP_TARGET_SOFTLIMIT: next = val + SOFTLIMIT_EVENTS_TARGET; break; case MEM_CGROUP_TARGET_NUMAINFO: next = val + NUMAINFO_EVENTS_TARGET; break; default: break; } __this_cpu_write(memcg->stat->targets[target], next); return true; |
7a159cc9d
|
788 |
} |
f53d7ce32
|
789 |
return false; |
d2265e6fa
|
790 791 792 793 794 795 |
} /* * Check events in order. * */ |
c0ff4b854
|
796 |
static void memcg_check_events(struct mem_cgroup *memcg, struct page *page) |
d2265e6fa
|
797 |
{ |
4799401fe
|
798 |
preempt_disable(); |
d2265e6fa
|
799 |
/* threshold event is triggered in finer grain than soft limit */ |
f53d7ce32
|
800 801 |
if (unlikely(mem_cgroup_event_ratelimit(memcg, MEM_CGROUP_TARGET_THRESH))) { |
82b3f2a71
|
802 803 |
bool do_softlimit; bool do_numainfo __maybe_unused; |
f53d7ce32
|
804 805 806 807 808 809 810 811 |
do_softlimit = mem_cgroup_event_ratelimit(memcg, MEM_CGROUP_TARGET_SOFTLIMIT); #if MAX_NUMNODES > 1 do_numainfo = mem_cgroup_event_ratelimit(memcg, MEM_CGROUP_TARGET_NUMAINFO); #endif preempt_enable(); |
c0ff4b854
|
812 |
mem_cgroup_threshold(memcg); |
f53d7ce32
|
813 |
if (unlikely(do_softlimit)) |
c0ff4b854
|
814 |
mem_cgroup_update_tree(memcg, page); |
453a9bf34
|
815 |
#if MAX_NUMNODES > 1 |
f53d7ce32
|
816 |
if (unlikely(do_numainfo)) |
c0ff4b854
|
817 |
atomic_inc(&memcg->numainfo_events); |
453a9bf34
|
818 |
#endif |
f53d7ce32
|
819 820 |
} else preempt_enable(); |
d2265e6fa
|
821 |
} |
d1a4c0b37
|
822 |
struct mem_cgroup *mem_cgroup_from_cont(struct cgroup *cont) |
8cdea7c05
|
823 |
{ |
b21451459
|
824 825 |
return mem_cgroup_from_css( cgroup_subsys_state(cont, mem_cgroup_subsys_id)); |
8cdea7c05
|
826 |
} |
cf475ad28
|
827 |
struct mem_cgroup *mem_cgroup_from_task(struct task_struct *p) |
78fb74669
|
828 |
{ |
31a78f23b
|
829 830 831 832 833 834 835 |
/* * mm_update_next_owner() may clear mm->owner to NULL * if it races with swapoff, page migration, etc. * So this can be called with p == NULL. */ if (unlikely(!p)) return NULL; |
b21451459
|
836 |
return mem_cgroup_from_css(task_subsys_state(p, mem_cgroup_subsys_id)); |
78fb74669
|
837 |
} |
a433658c3
|
838 |
struct mem_cgroup *try_get_mem_cgroup_from_mm(struct mm_struct *mm) |
54595fe26
|
839 |
{ |
c0ff4b854
|
840 |
struct mem_cgroup *memcg = NULL; |
0b7f569e4
|
841 842 843 |
if (!mm) return NULL; |
54595fe26
|
844 845 846 847 848 849 850 |
/* * Because we have no locks, mm->owner's may be being moved to other * cgroup. We use css_tryget() here even if this looks * pessimistic (rather than adding locks here). */ rcu_read_lock(); do { |
c0ff4b854
|
851 852 |
memcg = mem_cgroup_from_task(rcu_dereference(mm->owner)); if (unlikely(!memcg)) |
54595fe26
|
853 |
break; |
c0ff4b854
|
854 |
} while (!css_tryget(&memcg->css)); |
54595fe26
|
855 |
rcu_read_unlock(); |
c0ff4b854
|
856 |
return memcg; |
54595fe26
|
857 |
} |
5660048cc
|
858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 |
/** * mem_cgroup_iter - iterate over memory cgroup hierarchy * @root: hierarchy root * @prev: previously returned memcg, NULL on first invocation * @reclaim: cookie for shared reclaim walks, NULL for full walks * * Returns references to children of the hierarchy below @root, or * @root itself, or %NULL after a full round-trip. * * Caller must pass the return value in @prev on subsequent * invocations for reference counting, or use mem_cgroup_iter_break() * to cancel a hierarchy walk before the round-trip is complete. * * Reclaimers can specify a zone and a priority level in @reclaim to * divide up the memcgs in the hierarchy among all concurrent * reclaimers operating on the same zone and priority. */ struct mem_cgroup *mem_cgroup_iter(struct mem_cgroup *root, struct mem_cgroup *prev, struct mem_cgroup_reclaim_cookie *reclaim) |
14067bb3e
|
878 |
{ |
9f3a0d093
|
879 880 |
struct mem_cgroup *memcg = NULL; int id = 0; |
711d3d2c9
|
881 |
|
5660048cc
|
882 883 |
if (mem_cgroup_disabled()) return NULL; |
9f3a0d093
|
884 885 |
if (!root) root = root_mem_cgroup; |
7d74b06f2
|
886 |
|
9f3a0d093
|
887 888 |
if (prev && !reclaim) id = css_id(&prev->css); |
14067bb3e
|
889 |
|
9f3a0d093
|
890 891 |
if (prev && prev != root) css_put(&prev->css); |
14067bb3e
|
892 |
|
9f3a0d093
|
893 894 895 896 897 |
if (!root->use_hierarchy && root != root_mem_cgroup) { if (prev) return NULL; return root; } |
14067bb3e
|
898 |
|
9f3a0d093
|
899 |
while (!memcg) { |
527a5ec9a
|
900 |
struct mem_cgroup_reclaim_iter *uninitialized_var(iter); |
9f3a0d093
|
901 |
struct cgroup_subsys_state *css; |
711d3d2c9
|
902 |
|
527a5ec9a
|
903 904 905 906 907 908 909 910 911 912 913 |
if (reclaim) { int nid = zone_to_nid(reclaim->zone); int zid = zone_idx(reclaim->zone); struct mem_cgroup_per_zone *mz; mz = mem_cgroup_zoneinfo(root, nid, zid); iter = &mz->reclaim_iter[reclaim->priority]; if (prev && reclaim->generation != iter->generation) return NULL; id = iter->position; } |
7d74b06f2
|
914 |
|
9f3a0d093
|
915 916 917 918 |
rcu_read_lock(); css = css_get_next(&mem_cgroup_subsys, id + 1, &root->css, &id); if (css) { if (css == &root->css || css_tryget(css)) |
b21451459
|
919 |
memcg = mem_cgroup_from_css(css); |
9f3a0d093
|
920 921 |
} else id = 0; |
14067bb3e
|
922 |
rcu_read_unlock(); |
14067bb3e
|
923 |
|
527a5ec9a
|
924 925 926 927 928 929 930 |
if (reclaim) { iter->position = id; if (!css) iter->generation++; else if (!prev && memcg) reclaim->generation = iter->generation; } |
9f3a0d093
|
931 932 933 934 935 |
if (prev && !css) return NULL; } return memcg; |
14067bb3e
|
936 |
} |
7d74b06f2
|
937 |
|
5660048cc
|
938 939 940 941 942 943 944 |
/** * mem_cgroup_iter_break - abort a hierarchy walk prematurely * @root: hierarchy root * @prev: last visited hierarchy member as returned by mem_cgroup_iter() */ void mem_cgroup_iter_break(struct mem_cgroup *root, struct mem_cgroup *prev) |
9f3a0d093
|
945 946 947 948 949 950 |
{ if (!root) root = root_mem_cgroup; if (prev && prev != root) css_put(&prev->css); } |
7d74b06f2
|
951 |
|
9f3a0d093
|
952 953 954 955 956 957 |
/* * Iteration constructs for visiting all cgroups (under a tree). If * loops are exited prematurely (break), mem_cgroup_iter_break() must * be used for reference counting. */ #define for_each_mem_cgroup_tree(iter, root) \ |
527a5ec9a
|
958 |
for (iter = mem_cgroup_iter(root, NULL, NULL); \ |
9f3a0d093
|
959 |
iter != NULL; \ |
527a5ec9a
|
960 |
iter = mem_cgroup_iter(root, iter, NULL)) |
711d3d2c9
|
961 |
|
9f3a0d093
|
962 |
#define for_each_mem_cgroup(iter) \ |
527a5ec9a
|
963 |
for (iter = mem_cgroup_iter(NULL, NULL, NULL); \ |
9f3a0d093
|
964 |
iter != NULL; \ |
527a5ec9a
|
965 |
iter = mem_cgroup_iter(NULL, iter, NULL)) |
14067bb3e
|
966 |
|
c0ff4b854
|
967 |
static inline bool mem_cgroup_is_root(struct mem_cgroup *memcg) |
4b3bde4c9
|
968 |
{ |
c0ff4b854
|
969 |
return (memcg == root_mem_cgroup); |
4b3bde4c9
|
970 |
} |
456f998ec
|
971 972 |
void mem_cgroup_count_vm_event(struct mm_struct *mm, enum vm_event_item idx) { |
c0ff4b854
|
973 |
struct mem_cgroup *memcg; |
456f998ec
|
974 975 976 977 978 |
if (!mm) return; rcu_read_lock(); |
c0ff4b854
|
979 980 |
memcg = mem_cgroup_from_task(rcu_dereference(mm->owner)); if (unlikely(!memcg)) |
456f998ec
|
981 982 983 |
goto out; switch (idx) { |
456f998ec
|
984 |
case PGFAULT: |
0e574a932
|
985 986 987 988 |
this_cpu_inc(memcg->stat->events[MEM_CGROUP_EVENTS_PGFAULT]); break; case PGMAJFAULT: this_cpu_inc(memcg->stat->events[MEM_CGROUP_EVENTS_PGMAJFAULT]); |
456f998ec
|
989 990 991 992 993 994 995 996 |
break; default: BUG(); } out: rcu_read_unlock(); } EXPORT_SYMBOL(mem_cgroup_count_vm_event); |
925b7673c
|
997 998 999 |
/** * mem_cgroup_zone_lruvec - get the lru list vector for a zone and memcg * @zone: zone of the wanted lruvec |
fa9add641
|
1000 |
* @memcg: memcg of the wanted lruvec |
925b7673c
|
1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 |
* * Returns the lru list vector holding pages for the given @zone and * @mem. This can be the global zone lruvec, if the memory controller * is disabled. */ struct lruvec *mem_cgroup_zone_lruvec(struct zone *zone, struct mem_cgroup *memcg) { struct mem_cgroup_per_zone *mz; if (mem_cgroup_disabled()) return &zone->lruvec; mz = mem_cgroup_zoneinfo(memcg, zone_to_nid(zone), zone_idx(zone)); return &mz->lruvec; } |
08e552c69
|
1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 |
/* * Following LRU functions are allowed to be used without PCG_LOCK. * Operations are called by routine of global LRU independently from memcg. * What we have to take care of here is validness of pc->mem_cgroup. * * Changes to pc->mem_cgroup happens when * 1. charge * 2. moving account * In typical case, "charge" is done before add-to-lru. Exception is SwapCache. * It is added to LRU before charge. * If PCG_USED bit is not set, page_cgroup is not added to this private LRU. * When moving account, the page is not on LRU. It's isolated. */ |
4f98a2fee
|
1030 |
|
925b7673c
|
1031 |
/** |
fa9add641
|
1032 |
* mem_cgroup_page_lruvec - return lruvec for adding an lru page |
925b7673c
|
1033 |
* @page: the page |
fa9add641
|
1034 |
* @zone: zone of the page |
925b7673c
|
1035 |
*/ |
fa9add641
|
1036 |
struct lruvec *mem_cgroup_page_lruvec(struct page *page, struct zone *zone) |
08e552c69
|
1037 |
{ |
08e552c69
|
1038 |
struct mem_cgroup_per_zone *mz; |
925b7673c
|
1039 1040 |
struct mem_cgroup *memcg; struct page_cgroup *pc; |
6d12e2d8d
|
1041 |
|
f8d665422
|
1042 |
if (mem_cgroup_disabled()) |
925b7673c
|
1043 |
return &zone->lruvec; |
08e552c69
|
1044 |
pc = lookup_page_cgroup(page); |
38c5d72f3
|
1045 |
memcg = pc->mem_cgroup; |
7512102cf
|
1046 1047 |
/* |
fa9add641
|
1048 |
* Surreptitiously switch any uncharged offlist page to root: |
7512102cf
|
1049 1050 1051 1052 1053 1054 1055 |
* an uncharged page off lru does nothing to secure * its former mem_cgroup from sudden removal. * * Our caller holds lru_lock, and PageCgroupUsed is updated * under page_cgroup lock: between them, they make all uses * of pc->mem_cgroup safe. */ |
fa9add641
|
1056 |
if (!PageLRU(page) && !PageCgroupUsed(pc) && memcg != root_mem_cgroup) |
7512102cf
|
1057 |
pc->mem_cgroup = memcg = root_mem_cgroup; |
925b7673c
|
1058 |
mz = page_cgroup_zoneinfo(memcg, page); |
925b7673c
|
1059 |
return &mz->lruvec; |
08e552c69
|
1060 |
} |
b69408e88
|
1061 |
|
925b7673c
|
1062 |
/** |
fa9add641
|
1063 1064 1065 1066 |
* mem_cgroup_update_lru_size - account for adding or removing an lru page * @lruvec: mem_cgroup per zone lru vector * @lru: index of lru list the page is sitting on * @nr_pages: positive when adding or negative when removing |
925b7673c
|
1067 |
* |
fa9add641
|
1068 1069 |
* This function must be called when a page is added to or removed from an * lru list. |
3f58a8294
|
1070 |
*/ |
fa9add641
|
1071 1072 |
void mem_cgroup_update_lru_size(struct lruvec *lruvec, enum lru_list lru, int nr_pages) |
3f58a8294
|
1073 1074 |
{ struct mem_cgroup_per_zone *mz; |
fa9add641
|
1075 |
unsigned long *lru_size; |
3f58a8294
|
1076 1077 1078 |
if (mem_cgroup_disabled()) return; |
fa9add641
|
1079 1080 1081 1082 |
mz = container_of(lruvec, struct mem_cgroup_per_zone, lruvec); lru_size = mz->lru_size + lru; *lru_size += nr_pages; VM_BUG_ON((long)(*lru_size) < 0); |
08e552c69
|
1083 |
} |
544122e5e
|
1084 |
|
08e552c69
|
1085 |
/* |
c0ff4b854
|
1086 |
* Checks whether given mem is same or in the root_mem_cgroup's |
3e92041d6
|
1087 1088 |
* hierarchy subtree */ |
c3ac9a8ad
|
1089 1090 |
bool __mem_cgroup_same_or_subtree(const struct mem_cgroup *root_memcg, struct mem_cgroup *memcg) |
3e92041d6
|
1091 |
{ |
91c63734f
|
1092 1093 |
if (root_memcg == memcg) return true; |
3a981f482
|
1094 |
if (!root_memcg->use_hierarchy || !memcg) |
91c63734f
|
1095 |
return false; |
c3ac9a8ad
|
1096 1097 1098 1099 1100 1101 1102 |
return css_is_ancestor(&memcg->css, &root_memcg->css); } static bool mem_cgroup_same_or_subtree(const struct mem_cgroup *root_memcg, struct mem_cgroup *memcg) { bool ret; |
91c63734f
|
1103 |
rcu_read_lock(); |
c3ac9a8ad
|
1104 |
ret = __mem_cgroup_same_or_subtree(root_memcg, memcg); |
91c63734f
|
1105 1106 |
rcu_read_unlock(); return ret; |
3e92041d6
|
1107 |
} |
c0ff4b854
|
1108 |
int task_in_mem_cgroup(struct task_struct *task, const struct mem_cgroup *memcg) |
4c4a22148
|
1109 1110 |
{ int ret; |
0b7f569e4
|
1111 |
struct mem_cgroup *curr = NULL; |
158e0a2d1
|
1112 |
struct task_struct *p; |
4c4a22148
|
1113 |
|
158e0a2d1
|
1114 |
p = find_lock_task_mm(task); |
de077d222
|
1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 |
if (p) { curr = try_get_mem_cgroup_from_mm(p->mm); task_unlock(p); } else { /* * All threads may have already detached their mm's, but the oom * killer still needs to detect if they have already been oom * killed to prevent needlessly killing additional tasks. */ task_lock(task); curr = mem_cgroup_from_task(task); if (curr) css_get(&curr->css); task_unlock(task); } |
0b7f569e4
|
1130 1131 |
if (!curr) return 0; |
d31f56dbf
|
1132 |
/* |
c0ff4b854
|
1133 |
* We should check use_hierarchy of "memcg" not "curr". Because checking |
d31f56dbf
|
1134 |
* use_hierarchy of "curr" here make this function true if hierarchy is |
c0ff4b854
|
1135 1136 |
* enabled in "curr" and "curr" is a child of "memcg" in *cgroup* * hierarchy(even if use_hierarchy is disabled in "memcg"). |
d31f56dbf
|
1137 |
*/ |
c0ff4b854
|
1138 |
ret = mem_cgroup_same_or_subtree(memcg, curr); |
0b7f569e4
|
1139 |
css_put(&curr->css); |
4c4a22148
|
1140 1141 |
return ret; } |
c56d5c7df
|
1142 |
int mem_cgroup_inactive_anon_is_low(struct lruvec *lruvec) |
14797e236
|
1143 |
{ |
9b272977e
|
1144 |
unsigned long inactive_ratio; |
14797e236
|
1145 |
unsigned long inactive; |
9b272977e
|
1146 |
unsigned long active; |
c772be939
|
1147 |
unsigned long gb; |
14797e236
|
1148 |
|
4d7dcca21
|
1149 1150 |
inactive = mem_cgroup_get_lru_size(lruvec, LRU_INACTIVE_ANON); active = mem_cgroup_get_lru_size(lruvec, LRU_ACTIVE_ANON); |
14797e236
|
1151 |
|
c772be939
|
1152 1153 1154 1155 1156 |
gb = (inactive + active) >> (30 - PAGE_SHIFT); if (gb) inactive_ratio = int_sqrt(10 * gb); else inactive_ratio = 1; |
9b272977e
|
1157 |
return inactive * inactive_ratio < active; |
14797e236
|
1158 |
} |
c56d5c7df
|
1159 |
int mem_cgroup_inactive_file_is_low(struct lruvec *lruvec) |
56e49d218
|
1160 1161 1162 |
{ unsigned long active; unsigned long inactive; |
4d7dcca21
|
1163 1164 |
inactive = mem_cgroup_get_lru_size(lruvec, LRU_INACTIVE_FILE); active = mem_cgroup_get_lru_size(lruvec, LRU_ACTIVE_FILE); |
56e49d218
|
1165 1166 1167 |
return (active > inactive); } |
6d61ef409
|
1168 1169 |
#define mem_cgroup_from_res_counter(counter, member) \ container_of(counter, struct mem_cgroup, member) |
19942822d
|
1170 |
/** |
9d11ea9f1
|
1171 |
* mem_cgroup_margin - calculate chargeable space of a memory cgroup |
dad7557eb
|
1172 |
* @memcg: the memory cgroup |
19942822d
|
1173 |
* |
9d11ea9f1
|
1174 |
* Returns the maximum amount of memory @mem can be charged with, in |
7ec99d621
|
1175 |
* pages. |
19942822d
|
1176 |
*/ |
c0ff4b854
|
1177 |
static unsigned long mem_cgroup_margin(struct mem_cgroup *memcg) |
19942822d
|
1178 |
{ |
9d11ea9f1
|
1179 |
unsigned long long margin; |
c0ff4b854
|
1180 |
margin = res_counter_margin(&memcg->res); |
9d11ea9f1
|
1181 |
if (do_swap_account) |
c0ff4b854
|
1182 |
margin = min(margin, res_counter_margin(&memcg->memsw)); |
7ec99d621
|
1183 |
return margin >> PAGE_SHIFT; |
19942822d
|
1184 |
} |
1f4c025b5
|
1185 |
int mem_cgroup_swappiness(struct mem_cgroup *memcg) |
a7885eb8a
|
1186 1187 |
{ struct cgroup *cgrp = memcg->css.cgroup; |
a7885eb8a
|
1188 1189 1190 1191 |
/* root ? */ if (cgrp->parent == NULL) return vm_swappiness; |
bf1ff2635
|
1192 |
return memcg->swappiness; |
a7885eb8a
|
1193 |
} |
619d094b5
|
1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 |
/* * memcg->moving_account is used for checking possibility that some thread is * calling move_account(). When a thread on CPU-A starts moving pages under * a memcg, other threads should check memcg->moving_account under * rcu_read_lock(), like this: * * CPU-A CPU-B * rcu_read_lock() * memcg->moving_account+1 if (memcg->mocing_account) * take heavy locks. * synchronize_rcu() update something. * rcu_read_unlock() * start move here. */ |
4331f7d33
|
1208 1209 1210 |
/* for quick checking without looking up memcg */ atomic_t memcg_moving __read_mostly; |
c0ff4b854
|
1211 |
static void mem_cgroup_start_move(struct mem_cgroup *memcg) |
32047e2a8
|
1212 |
{ |
4331f7d33
|
1213 |
atomic_inc(&memcg_moving); |
619d094b5
|
1214 |
atomic_inc(&memcg->moving_account); |
32047e2a8
|
1215 1216 |
synchronize_rcu(); } |
c0ff4b854
|
1217 |
static void mem_cgroup_end_move(struct mem_cgroup *memcg) |
32047e2a8
|
1218 |
{ |
619d094b5
|
1219 1220 1221 1222 |
/* * Now, mem_cgroup_clear_mc() may call this function with NULL. * We check NULL in callee rather than caller. */ |
4331f7d33
|
1223 1224 |
if (memcg) { atomic_dec(&memcg_moving); |
619d094b5
|
1225 |
atomic_dec(&memcg->moving_account); |
4331f7d33
|
1226 |
} |
32047e2a8
|
1227 |
} |
619d094b5
|
1228 |
|
32047e2a8
|
1229 1230 1231 |
/* * 2 routines for checking "mem" is under move_account() or not. * |
13fd1dd9d
|
1232 1233 |
* mem_cgroup_stolen() - checking whether a cgroup is mc.from or not. This * is used for avoiding races in accounting. If true, |
32047e2a8
|
1234 1235 1236 1237 1238 1239 |
* pc->mem_cgroup may be overwritten. * * mem_cgroup_under_move() - checking a cgroup is mc.from or mc.to or * under hierarchy of moving cgroups. This is for * waiting at hith-memory prressure caused by "move". */ |
13fd1dd9d
|
1240 |
static bool mem_cgroup_stolen(struct mem_cgroup *memcg) |
32047e2a8
|
1241 1242 |
{ VM_BUG_ON(!rcu_read_lock_held()); |
619d094b5
|
1243 |
return atomic_read(&memcg->moving_account) > 0; |
32047e2a8
|
1244 |
} |
4b5343346
|
1245 |
|
c0ff4b854
|
1246 |
static bool mem_cgroup_under_move(struct mem_cgroup *memcg) |
4b5343346
|
1247 |
{ |
2bd9bb206
|
1248 1249 |
struct mem_cgroup *from; struct mem_cgroup *to; |
4b5343346
|
1250 |
bool ret = false; |
2bd9bb206
|
1251 1252 1253 1254 1255 1256 1257 1258 1259 |
/* * Unlike task_move routines, we access mc.to, mc.from not under * mutual exclusion by cgroup_mutex. Here, we take spinlock instead. */ spin_lock(&mc.lock); from = mc.from; to = mc.to; if (!from) goto unlock; |
3e92041d6
|
1260 |
|
c0ff4b854
|
1261 1262 |
ret = mem_cgroup_same_or_subtree(memcg, from) || mem_cgroup_same_or_subtree(memcg, to); |
2bd9bb206
|
1263 1264 |
unlock: spin_unlock(&mc.lock); |
4b5343346
|
1265 1266 |
return ret; } |
c0ff4b854
|
1267 |
static bool mem_cgroup_wait_acct_move(struct mem_cgroup *memcg) |
4b5343346
|
1268 1269 |
{ if (mc.moving_task && current != mc.moving_task) { |
c0ff4b854
|
1270 |
if (mem_cgroup_under_move(memcg)) { |
4b5343346
|
1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 |
DEFINE_WAIT(wait); prepare_to_wait(&mc.waitq, &wait, TASK_INTERRUPTIBLE); /* moving charge context might have finished. */ if (mc.moving_task) schedule(); finish_wait(&mc.waitq, &wait); return true; } } return false; } |
312734c04
|
1282 1283 1284 1285 |
/* * Take this lock when * - a code tries to modify page's memcg while it's USED. * - a code tries to modify page state accounting in a memcg. |
13fd1dd9d
|
1286 |
* see mem_cgroup_stolen(), too. |
312734c04
|
1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 |
*/ static void move_lock_mem_cgroup(struct mem_cgroup *memcg, unsigned long *flags) { spin_lock_irqsave(&memcg->move_lock, *flags); } static void move_unlock_mem_cgroup(struct mem_cgroup *memcg, unsigned long *flags) { spin_unlock_irqrestore(&memcg->move_lock, *flags); } |
e222432bf
|
1299 |
/** |
6a6135b64
|
1300 |
* mem_cgroup_print_oom_info: Called from OOM with tasklist_lock held in read mode. |
e222432bf
|
1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 |
* @memcg: The memory cgroup that went over limit * @p: Task that is going to be killed * * NOTE: @memcg and @p's mem_cgroup can be different when hierarchy is * enabled */ void mem_cgroup_print_oom_info(struct mem_cgroup *memcg, struct task_struct *p) { struct cgroup *task_cgrp; struct cgroup *mem_cgrp; /* * Need a buffer in BSS, can't rely on allocations. The code relies * on the assumption that OOM is serialized for memory controller. * If this assumption is broken, revisit this code. */ static char memcg_name[PATH_MAX]; int ret; |
d31f56dbf
|
1318 |
if (!memcg || !p) |
e222432bf
|
1319 |
return; |
e222432bf
|
1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 |
rcu_read_lock(); mem_cgrp = memcg->css.cgroup; task_cgrp = task_cgroup(p, mem_cgroup_subsys_id); ret = cgroup_path(task_cgrp, memcg_name, PATH_MAX); if (ret < 0) { /* * Unfortunately, we are unable to convert to a useful name * But we'll still print out the usage information */ rcu_read_unlock(); goto done; } rcu_read_unlock(); printk(KERN_INFO "Task in %s killed", memcg_name); rcu_read_lock(); ret = cgroup_path(mem_cgrp, memcg_name, PATH_MAX); if (ret < 0) { rcu_read_unlock(); goto done; } rcu_read_unlock(); /* * Continues from above, so we don't need an KERN_ level */ printk(KERN_CONT " as a result of limit of %s ", memcg_name); done: printk(KERN_INFO "memory: usage %llukB, limit %llukB, failcnt %llu ", res_counter_read_u64(&memcg->res, RES_USAGE) >> 10, res_counter_read_u64(&memcg->res, RES_LIMIT) >> 10, res_counter_read_u64(&memcg->res, RES_FAILCNT)); printk(KERN_INFO "memory+swap: usage %llukB, limit %llukB, " "failcnt %llu ", res_counter_read_u64(&memcg->memsw, RES_USAGE) >> 10, res_counter_read_u64(&memcg->memsw, RES_LIMIT) >> 10, res_counter_read_u64(&memcg->memsw, RES_FAILCNT)); } |
81d39c20f
|
1365 1366 1367 1368 |
/* * This function returns the number of memcg under hierarchy tree. Returns * 1(self count) if no children. */ |
c0ff4b854
|
1369 |
static int mem_cgroup_count_children(struct mem_cgroup *memcg) |
81d39c20f
|
1370 1371 |
{ int num = 0; |
7d74b06f2
|
1372 |
struct mem_cgroup *iter; |
c0ff4b854
|
1373 |
for_each_mem_cgroup_tree(iter, memcg) |
7d74b06f2
|
1374 |
num++; |
81d39c20f
|
1375 1376 |
return num; } |
6d61ef409
|
1377 |
/* |
a63d83f42
|
1378 1379 |
* Return the memory (and swap, if configured) limit for a memcg. */ |
9cbb78bb3
|
1380 |
static u64 mem_cgroup_get_limit(struct mem_cgroup *memcg) |
a63d83f42
|
1381 1382 1383 |
{ u64 limit; u64 memsw; |
f3e8eb70b
|
1384 1385 |
limit = res_counter_read_u64(&memcg->res, RES_LIMIT); limit += total_swap_pages << PAGE_SHIFT; |
a63d83f42
|
1386 1387 1388 1389 1390 1391 1392 |
memsw = res_counter_read_u64(&memcg->memsw, RES_LIMIT); /* * If memsw is finite and limits the amount of swap space available * to this memcg, return that limit. */ return min(limit, memsw); } |
876aafbfd
|
1393 1394 |
void mem_cgroup_out_of_memory(struct mem_cgroup *memcg, gfp_t gfp_mask, int order) |
9cbb78bb3
|
1395 1396 1397 1398 1399 1400 |
{ struct mem_cgroup *iter; unsigned long chosen_points = 0; unsigned long totalpages; unsigned int points = 0; struct task_struct *chosen = NULL; |
876aafbfd
|
1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 |
/* * If current has a pending SIGKILL, then automatically select it. The * goal is to allow it to allocate so that it may quickly exit and free * its memory. */ if (fatal_signal_pending(current)) { set_thread_flag(TIF_MEMDIE); return; } check_panic_on_oom(CONSTRAINT_MEMCG, gfp_mask, order, NULL); |
9cbb78bb3
|
1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 |
totalpages = mem_cgroup_get_limit(memcg) >> PAGE_SHIFT ? : 1; for_each_mem_cgroup_tree(iter, memcg) { struct cgroup *cgroup = iter->css.cgroup; struct cgroup_iter it; struct task_struct *task; cgroup_iter_start(cgroup, &it); while ((task = cgroup_iter_next(cgroup, &it))) { switch (oom_scan_process_thread(task, totalpages, NULL, false)) { case OOM_SCAN_SELECT: if (chosen) put_task_struct(chosen); chosen = task; chosen_points = ULONG_MAX; get_task_struct(chosen); /* fall through */ case OOM_SCAN_CONTINUE: continue; case OOM_SCAN_ABORT: cgroup_iter_end(cgroup, &it); mem_cgroup_iter_break(memcg, iter); if (chosen) put_task_struct(chosen); return; case OOM_SCAN_OK: break; }; points = oom_badness(task, memcg, NULL, totalpages); if (points > chosen_points) { if (chosen) put_task_struct(chosen); chosen = task; chosen_points = points; get_task_struct(chosen); } } cgroup_iter_end(cgroup, &it); } if (!chosen) return; points = chosen_points * 1000 / totalpages; |
9cbb78bb3
|
1455 1456 |
oom_kill_process(chosen, gfp_mask, order, points, totalpages, memcg, NULL, "Memory cgroup out of memory"); |
9cbb78bb3
|
1457 |
} |
5660048cc
|
1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 |
static unsigned long mem_cgroup_reclaim(struct mem_cgroup *memcg, gfp_t gfp_mask, unsigned long flags) { unsigned long total = 0; bool noswap = false; int loop; if (flags & MEM_CGROUP_RECLAIM_NOSWAP) noswap = true; if (!(flags & MEM_CGROUP_RECLAIM_SHRINK) && memcg->memsw_is_minimum) noswap = true; for (loop = 0; loop < MEM_CGROUP_MAX_RECLAIM_LOOPS; loop++) { if (loop) drain_all_stock_async(memcg); total += try_to_free_mem_cgroup_pages(memcg, gfp_mask, noswap); /* * Allow limit shrinkers, which are triggered directly * by userspace, to catch signals and stop reclaim * after minimal progress, regardless of the margin. */ if (total && (flags & MEM_CGROUP_RECLAIM_SHRINK)) break; if (mem_cgroup_margin(memcg)) break; /* * If nothing was reclaimed after two attempts, there * may be no reclaimable pages in this hierarchy. */ if (loop && !total) break; } return total; } |
4d0c066d2
|
1493 1494 |
/** * test_mem_cgroup_node_reclaimable |
dad7557eb
|
1495 |
* @memcg: the target memcg |
4d0c066d2
|
1496 1497 1498 1499 1500 1501 1502 |
* @nid: the node ID to be checked. * @noswap : specify true here if the user wants flle only information. * * This function returns whether the specified memcg contains any * reclaimable pages on a node. Returns true if there are any reclaimable * pages in the node. */ |
c0ff4b854
|
1503 |
static bool test_mem_cgroup_node_reclaimable(struct mem_cgroup *memcg, |
4d0c066d2
|
1504 1505 |
int nid, bool noswap) { |
c0ff4b854
|
1506 |
if (mem_cgroup_node_nr_lru_pages(memcg, nid, LRU_ALL_FILE)) |
4d0c066d2
|
1507 1508 1509 |
return true; if (noswap || !total_swap_pages) return false; |
c0ff4b854
|
1510 |
if (mem_cgroup_node_nr_lru_pages(memcg, nid, LRU_ALL_ANON)) |
4d0c066d2
|
1511 1512 1513 1514 |
return true; return false; } |
889976dbc
|
1515 1516 1517 1518 1519 1520 1521 1522 |
#if MAX_NUMNODES > 1 /* * Always updating the nodemask is not very good - even if we have an empty * list or the wrong list here, we can start from some node and traverse all * nodes based on the zonelist. So update the list loosely once per 10 secs. * */ |
c0ff4b854
|
1523 |
static void mem_cgroup_may_update_nodemask(struct mem_cgroup *memcg) |
889976dbc
|
1524 1525 |
{ int nid; |
453a9bf34
|
1526 1527 1528 1529 |
/* * numainfo_events > 0 means there was at least NUMAINFO_EVENTS_TARGET * pagein/pageout changes since the last update. */ |
c0ff4b854
|
1530 |
if (!atomic_read(&memcg->numainfo_events)) |
453a9bf34
|
1531 |
return; |
c0ff4b854
|
1532 |
if (atomic_inc_return(&memcg->numainfo_updating) > 1) |
889976dbc
|
1533 |
return; |
889976dbc
|
1534 |
/* make a nodemask where this memcg uses memory from */ |
c0ff4b854
|
1535 |
memcg->scan_nodes = node_states[N_HIGH_MEMORY]; |
889976dbc
|
1536 1537 |
for_each_node_mask(nid, node_states[N_HIGH_MEMORY]) { |
c0ff4b854
|
1538 1539 |
if (!test_mem_cgroup_node_reclaimable(memcg, nid, false)) node_clear(nid, memcg->scan_nodes); |
889976dbc
|
1540 |
} |
453a9bf34
|
1541 |
|
c0ff4b854
|
1542 1543 |
atomic_set(&memcg->numainfo_events, 0); atomic_set(&memcg->numainfo_updating, 0); |
889976dbc
|
1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 |
} /* * Selecting a node where we start reclaim from. Because what we need is just * reducing usage counter, start from anywhere is O,K. Considering * memory reclaim from current node, there are pros. and cons. * * Freeing memory from current node means freeing memory from a node which * we'll use or we've used. So, it may make LRU bad. And if several threads * hit limits, it will see a contention on a node. But freeing from remote * node means more costs for memory reclaim because of memory latency. * * Now, we use round-robin. Better algorithm is welcomed. */ |
c0ff4b854
|
1558 |
int mem_cgroup_select_victim_node(struct mem_cgroup *memcg) |
889976dbc
|
1559 1560 |
{ int node; |
c0ff4b854
|
1561 1562 |
mem_cgroup_may_update_nodemask(memcg); node = memcg->last_scanned_node; |
889976dbc
|
1563 |
|
c0ff4b854
|
1564 |
node = next_node(node, memcg->scan_nodes); |
889976dbc
|
1565 |
if (node == MAX_NUMNODES) |
c0ff4b854
|
1566 |
node = first_node(memcg->scan_nodes); |
889976dbc
|
1567 1568 1569 1570 1571 1572 1573 1574 |
/* * We call this when we hit limit, not when pages are added to LRU. * No LRU may hold pages because all pages are UNEVICTABLE or * memcg is too small and all pages are not on LRU. In that case, * we use curret node. */ if (unlikely(node == MAX_NUMNODES)) node = numa_node_id(); |
c0ff4b854
|
1575 |
memcg->last_scanned_node = node; |
889976dbc
|
1576 1577 |
return node; } |
4d0c066d2
|
1578 1579 1580 1581 1582 1583 |
/* * Check all nodes whether it contains reclaimable pages or not. * For quick scan, we make use of scan_nodes. This will allow us to skip * unused nodes. But scan_nodes is lazily updated and may not cotain * enough new information. We need to do double check. */ |
6bbda35ce
|
1584 |
static bool mem_cgroup_reclaimable(struct mem_cgroup *memcg, bool noswap) |
4d0c066d2
|
1585 1586 1587 1588 1589 1590 1591 |
{ int nid; /* * quick check...making use of scan_node. * We can skip unused nodes. */ |
c0ff4b854
|
1592 1593 |
if (!nodes_empty(memcg->scan_nodes)) { for (nid = first_node(memcg->scan_nodes); |
4d0c066d2
|
1594 |
nid < MAX_NUMNODES; |
c0ff4b854
|
1595 |
nid = next_node(nid, memcg->scan_nodes)) { |
4d0c066d2
|
1596 |
|
c0ff4b854
|
1597 |
if (test_mem_cgroup_node_reclaimable(memcg, nid, noswap)) |
4d0c066d2
|
1598 1599 1600 1601 1602 1603 1604 |
return true; } } /* * Check rest of nodes. */ for_each_node_state(nid, N_HIGH_MEMORY) { |
c0ff4b854
|
1605 |
if (node_isset(nid, memcg->scan_nodes)) |
4d0c066d2
|
1606 |
continue; |
c0ff4b854
|
1607 |
if (test_mem_cgroup_node_reclaimable(memcg, nid, noswap)) |
4d0c066d2
|
1608 1609 1610 1611 |
return true; } return false; } |
889976dbc
|
1612 |
#else |
c0ff4b854
|
1613 |
int mem_cgroup_select_victim_node(struct mem_cgroup *memcg) |
889976dbc
|
1614 1615 1616 |
{ return 0; } |
4d0c066d2
|
1617 |
|
6bbda35ce
|
1618 |
static bool mem_cgroup_reclaimable(struct mem_cgroup *memcg, bool noswap) |
4d0c066d2
|
1619 |
{ |
c0ff4b854
|
1620 |
return test_mem_cgroup_node_reclaimable(memcg, 0, noswap); |
4d0c066d2
|
1621 |
} |
889976dbc
|
1622 |
#endif |
5660048cc
|
1623 1624 1625 1626 |
static int mem_cgroup_soft_reclaim(struct mem_cgroup *root_memcg, struct zone *zone, gfp_t gfp_mask, unsigned long *total_scanned) |
6d61ef409
|
1627 |
{ |
9f3a0d093
|
1628 |
struct mem_cgroup *victim = NULL; |
5660048cc
|
1629 |
int total = 0; |
04046e1a0
|
1630 |
int loop = 0; |
9d11ea9f1
|
1631 |
unsigned long excess; |
185efc0f9
|
1632 |
unsigned long nr_scanned; |
527a5ec9a
|
1633 1634 1635 1636 |
struct mem_cgroup_reclaim_cookie reclaim = { .zone = zone, .priority = 0, }; |
9d11ea9f1
|
1637 |
|
c0ff4b854
|
1638 |
excess = res_counter_soft_limit_excess(&root_memcg->res) >> PAGE_SHIFT; |
04046e1a0
|
1639 |
|
4e4169535
|
1640 |
while (1) { |
527a5ec9a
|
1641 |
victim = mem_cgroup_iter(root_memcg, victim, &reclaim); |
9f3a0d093
|
1642 |
if (!victim) { |
04046e1a0
|
1643 |
loop++; |
4e4169535
|
1644 1645 1646 1647 1648 1649 |
if (loop >= 2) { /* * If we have not been able to reclaim * anything, it might because there are * no reclaimable pages under this hierarchy */ |
5660048cc
|
1650 |
if (!total) |
4e4169535
|
1651 |
break; |
4e4169535
|
1652 |
/* |
25985edce
|
1653 |
* We want to do more targeted reclaim. |
4e4169535
|
1654 1655 1656 1657 1658 |
* excess >> 2 is not to excessive so as to * reclaim too much, nor too less that we keep * coming back to reclaim from this cgroup */ if (total >= (excess >> 2) || |
9f3a0d093
|
1659 |
(loop > MEM_CGROUP_MAX_RECLAIM_LOOPS)) |
4e4169535
|
1660 |
break; |
4e4169535
|
1661 |
} |
9f3a0d093
|
1662 |
continue; |
4e4169535
|
1663 |
} |
5660048cc
|
1664 |
if (!mem_cgroup_reclaimable(victim, false)) |
6d61ef409
|
1665 |
continue; |
5660048cc
|
1666 1667 1668 1669 |
total += mem_cgroup_shrink_node_zone(victim, gfp_mask, false, zone, &nr_scanned); *total_scanned += nr_scanned; if (!res_counter_soft_limit_excess(&root_memcg->res)) |
9f3a0d093
|
1670 |
break; |
6d61ef409
|
1671 |
} |
9f3a0d093
|
1672 |
mem_cgroup_iter_break(root_memcg, victim); |
04046e1a0
|
1673 |
return total; |
6d61ef409
|
1674 |
} |
867578cbc
|
1675 1676 1677 |
/* * Check OOM-Killer is already running under our hierarchy. * If someone is running, return false. |
1af8efe96
|
1678 |
* Has to be called with memcg_oom_lock |
867578cbc
|
1679 |
*/ |
c0ff4b854
|
1680 |
static bool mem_cgroup_oom_lock(struct mem_cgroup *memcg) |
867578cbc
|
1681 |
{ |
79dfdaccd
|
1682 |
struct mem_cgroup *iter, *failed = NULL; |
a636b327f
|
1683 |
|
9f3a0d093
|
1684 |
for_each_mem_cgroup_tree(iter, memcg) { |
23751be00
|
1685 |
if (iter->oom_lock) { |
79dfdaccd
|
1686 1687 1688 1689 |
/* * this subtree of our hierarchy is already locked * so we cannot give a lock. */ |
79dfdaccd
|
1690 |
failed = iter; |
9f3a0d093
|
1691 1692 |
mem_cgroup_iter_break(memcg, iter); break; |
23751be00
|
1693 1694 |
} else iter->oom_lock = true; |
7d74b06f2
|
1695 |
} |
867578cbc
|
1696 |
|
79dfdaccd
|
1697 |
if (!failed) |
23751be00
|
1698 |
return true; |
79dfdaccd
|
1699 1700 1701 1702 1703 |
/* * OK, we failed to lock the whole subtree so we have to clean up * what we set up to the failing subtree */ |
9f3a0d093
|
1704 |
for_each_mem_cgroup_tree(iter, memcg) { |
79dfdaccd
|
1705 |
if (iter == failed) { |
9f3a0d093
|
1706 1707 |
mem_cgroup_iter_break(memcg, iter); break; |
79dfdaccd
|
1708 1709 1710 |
} iter->oom_lock = false; } |
23751be00
|
1711 |
return false; |
a636b327f
|
1712 |
} |
0b7f569e4
|
1713 |
|
79dfdaccd
|
1714 |
/* |
1af8efe96
|
1715 |
* Has to be called with memcg_oom_lock |
79dfdaccd
|
1716 |
*/ |
c0ff4b854
|
1717 |
static int mem_cgroup_oom_unlock(struct mem_cgroup *memcg) |
0b7f569e4
|
1718 |
{ |
7d74b06f2
|
1719 |
struct mem_cgroup *iter; |
c0ff4b854
|
1720 |
for_each_mem_cgroup_tree(iter, memcg) |
79dfdaccd
|
1721 1722 1723 |
iter->oom_lock = false; return 0; } |
c0ff4b854
|
1724 |
static void mem_cgroup_mark_under_oom(struct mem_cgroup *memcg) |
79dfdaccd
|
1725 1726 |
{ struct mem_cgroup *iter; |
c0ff4b854
|
1727 |
for_each_mem_cgroup_tree(iter, memcg) |
79dfdaccd
|
1728 1729 |
atomic_inc(&iter->under_oom); } |
c0ff4b854
|
1730 |
static void mem_cgroup_unmark_under_oom(struct mem_cgroup *memcg) |
79dfdaccd
|
1731 1732 |
{ struct mem_cgroup *iter; |
867578cbc
|
1733 1734 1735 1736 1737 |
/* * When a new child is created while the hierarchy is under oom, * mem_cgroup_oom_lock() may not be called. We have to use * atomic_add_unless() here. */ |
c0ff4b854
|
1738 |
for_each_mem_cgroup_tree(iter, memcg) |
79dfdaccd
|
1739 |
atomic_add_unless(&iter->under_oom, -1, 0); |
0b7f569e4
|
1740 |
} |
1af8efe96
|
1741 |
static DEFINE_SPINLOCK(memcg_oom_lock); |
867578cbc
|
1742 |
static DECLARE_WAIT_QUEUE_HEAD(memcg_oom_waitq); |
dc98df5a1
|
1743 |
struct oom_wait_info { |
d79154bb5
|
1744 |
struct mem_cgroup *memcg; |
dc98df5a1
|
1745 1746 1747 1748 1749 1750 |
wait_queue_t wait; }; static int memcg_oom_wake_function(wait_queue_t *wait, unsigned mode, int sync, void *arg) { |
d79154bb5
|
1751 1752 |
struct mem_cgroup *wake_memcg = (struct mem_cgroup *)arg; struct mem_cgroup *oom_wait_memcg; |
dc98df5a1
|
1753 1754 1755 |
struct oom_wait_info *oom_wait_info; oom_wait_info = container_of(wait, struct oom_wait_info, wait); |
d79154bb5
|
1756 |
oom_wait_memcg = oom_wait_info->memcg; |
dc98df5a1
|
1757 |
|
dc98df5a1
|
1758 |
/* |
d79154bb5
|
1759 |
* Both of oom_wait_info->memcg and wake_memcg are stable under us. |
dc98df5a1
|
1760 1761 |
* Then we can use css_is_ancestor without taking care of RCU. */ |
c0ff4b854
|
1762 1763 |
if (!mem_cgroup_same_or_subtree(oom_wait_memcg, wake_memcg) && !mem_cgroup_same_or_subtree(wake_memcg, oom_wait_memcg)) |
dc98df5a1
|
1764 |
return 0; |
dc98df5a1
|
1765 1766 |
return autoremove_wake_function(wait, mode, sync, arg); } |
c0ff4b854
|
1767 |
static void memcg_wakeup_oom(struct mem_cgroup *memcg) |
dc98df5a1
|
1768 |
{ |
c0ff4b854
|
1769 1770 |
/* for filtering, pass "memcg" as argument. */ __wake_up(&memcg_oom_waitq, TASK_NORMAL, 0, memcg); |
dc98df5a1
|
1771 |
} |
c0ff4b854
|
1772 |
static void memcg_oom_recover(struct mem_cgroup *memcg) |
3c11ecf44
|
1773 |
{ |
c0ff4b854
|
1774 1775 |
if (memcg && atomic_read(&memcg->under_oom)) memcg_wakeup_oom(memcg); |
3c11ecf44
|
1776 |
} |
867578cbc
|
1777 1778 1779 |
/* * try to call OOM killer. returns false if we should exit memory-reclaim loop. */ |
6bbda35ce
|
1780 1781 |
static bool mem_cgroup_handle_oom(struct mem_cgroup *memcg, gfp_t mask, int order) |
0b7f569e4
|
1782 |
{ |
dc98df5a1
|
1783 |
struct oom_wait_info owait; |
3c11ecf44
|
1784 |
bool locked, need_to_kill; |
867578cbc
|
1785 |
|
d79154bb5
|
1786 |
owait.memcg = memcg; |
dc98df5a1
|
1787 1788 1789 1790 |
owait.wait.flags = 0; owait.wait.func = memcg_oom_wake_function; owait.wait.private = current; INIT_LIST_HEAD(&owait.wait.task_list); |
3c11ecf44
|
1791 |
need_to_kill = true; |
c0ff4b854
|
1792 |
mem_cgroup_mark_under_oom(memcg); |
79dfdaccd
|
1793 |
|
c0ff4b854
|
1794 |
/* At first, try to OOM lock hierarchy under memcg.*/ |
1af8efe96
|
1795 |
spin_lock(&memcg_oom_lock); |
c0ff4b854
|
1796 |
locked = mem_cgroup_oom_lock(memcg); |
867578cbc
|
1797 1798 1799 1800 1801 |
/* * Even if signal_pending(), we can't quit charge() loop without * accounting. So, UNINTERRUPTIBLE is appropriate. But SIGKILL * under OOM is always welcomed, use TASK_KILLABLE here. */ |
3c11ecf44
|
1802 |
prepare_to_wait(&memcg_oom_waitq, &owait.wait, TASK_KILLABLE); |
c0ff4b854
|
1803 |
if (!locked || memcg->oom_kill_disable) |
3c11ecf44
|
1804 1805 |
need_to_kill = false; if (locked) |
c0ff4b854
|
1806 |
mem_cgroup_oom_notify(memcg); |
1af8efe96
|
1807 |
spin_unlock(&memcg_oom_lock); |
867578cbc
|
1808 |
|
3c11ecf44
|
1809 1810 |
if (need_to_kill) { finish_wait(&memcg_oom_waitq, &owait.wait); |
e845e1993
|
1811 |
mem_cgroup_out_of_memory(memcg, mask, order); |
3c11ecf44
|
1812 |
} else { |
867578cbc
|
1813 |
schedule(); |
dc98df5a1
|
1814 |
finish_wait(&memcg_oom_waitq, &owait.wait); |
867578cbc
|
1815 |
} |
1af8efe96
|
1816 |
spin_lock(&memcg_oom_lock); |
79dfdaccd
|
1817 |
if (locked) |
c0ff4b854
|
1818 1819 |
mem_cgroup_oom_unlock(memcg); memcg_wakeup_oom(memcg); |
1af8efe96
|
1820 |
spin_unlock(&memcg_oom_lock); |
867578cbc
|
1821 |
|
c0ff4b854
|
1822 |
mem_cgroup_unmark_under_oom(memcg); |
79dfdaccd
|
1823 |
|
867578cbc
|
1824 1825 1826 |
if (test_thread_flag(TIF_MEMDIE) || fatal_signal_pending(current)) return false; /* Give chance to dying process */ |
715a5ee82
|
1827 |
schedule_timeout_uninterruptible(1); |
867578cbc
|
1828 |
return true; |
0b7f569e4
|
1829 |
} |
d69b042f3
|
1830 1831 1832 |
/* * Currently used to update mapped file statistics, but the routine can be * generalized to update other statistics as well. |
32047e2a8
|
1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 |
* * Notes: Race condition * * We usually use page_cgroup_lock() for accessing page_cgroup member but * it tends to be costly. But considering some conditions, we doesn't need * to do so _always_. * * Considering "charge", lock_page_cgroup() is not required because all * file-stat operations happen after a page is attached to radix-tree. There * are no race with "charge". * * Considering "uncharge", we know that memcg doesn't clear pc->mem_cgroup * at "uncharge" intentionally. So, we always see valid pc->mem_cgroup even * if there are race with "uncharge". Statistics itself is properly handled * by flags. * * Considering "move", this is an only case we see a race. To make the race |
619d094b5
|
1850 1851 |
* small, we check mm->moving_account and detect there are possibility of race * If there is, we take a lock. |
d69b042f3
|
1852 |
*/ |
26174efd4
|
1853 |
|
89c06bd52
|
1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 |
void __mem_cgroup_begin_update_page_stat(struct page *page, bool *locked, unsigned long *flags) { struct mem_cgroup *memcg; struct page_cgroup *pc; pc = lookup_page_cgroup(page); again: memcg = pc->mem_cgroup; if (unlikely(!memcg || !PageCgroupUsed(pc))) return; /* * If this memory cgroup is not under account moving, we don't |
da92c47d0
|
1867 |
* need to take move_lock_mem_cgroup(). Because we already hold |
89c06bd52
|
1868 |
* rcu_read_lock(), any calls to move_account will be delayed until |
13fd1dd9d
|
1869 |
* rcu_read_unlock() if mem_cgroup_stolen() == true. |
89c06bd52
|
1870 |
*/ |
13fd1dd9d
|
1871 |
if (!mem_cgroup_stolen(memcg)) |
89c06bd52
|
1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 |
return; move_lock_mem_cgroup(memcg, flags); if (memcg != pc->mem_cgroup || !PageCgroupUsed(pc)) { move_unlock_mem_cgroup(memcg, flags); goto again; } *locked = true; } void __mem_cgroup_end_update_page_stat(struct page *page, unsigned long *flags) { struct page_cgroup *pc = lookup_page_cgroup(page); /* * It's guaranteed that pc->mem_cgroup never changes while * lock is held because a routine modifies pc->mem_cgroup |
da92c47d0
|
1889 |
* should take move_lock_mem_cgroup(). |
89c06bd52
|
1890 1891 1892 |
*/ move_unlock_mem_cgroup(pc->mem_cgroup, flags); } |
2a7106f2c
|
1893 1894 |
void mem_cgroup_update_page_stat(struct page *page, enum mem_cgroup_page_stat_item idx, int val) |
d69b042f3
|
1895 |
{ |
c0ff4b854
|
1896 |
struct mem_cgroup *memcg; |
32047e2a8
|
1897 |
struct page_cgroup *pc = lookup_page_cgroup(page); |
dbd4ea78f
|
1898 |
unsigned long uninitialized_var(flags); |
d69b042f3
|
1899 |
|
cfa449461
|
1900 |
if (mem_cgroup_disabled()) |
d69b042f3
|
1901 |
return; |
89c06bd52
|
1902 |
|
c0ff4b854
|
1903 1904 |
memcg = pc->mem_cgroup; if (unlikely(!memcg || !PageCgroupUsed(pc))) |
89c06bd52
|
1905 |
return; |
26174efd4
|
1906 |
|
26174efd4
|
1907 |
switch (idx) { |
2a7106f2c
|
1908 |
case MEMCG_NR_FILE_MAPPED: |
2a7106f2c
|
1909 |
idx = MEM_CGROUP_STAT_FILE_MAPPED; |
26174efd4
|
1910 1911 1912 |
break; default: BUG(); |
8725d5416
|
1913 |
} |
d69b042f3
|
1914 |
|
c0ff4b854
|
1915 |
this_cpu_add(memcg->stat->count[idx], val); |
d69b042f3
|
1916 |
} |
26174efd4
|
1917 |
|
f817ed485
|
1918 |
/* |
cdec2e426
|
1919 1920 1921 |
* size of first charge trial. "32" comes from vmscan.c's magic value. * TODO: maybe necessary to use big numbers in big irons. */ |
7ec99d621
|
1922 |
#define CHARGE_BATCH 32U |
cdec2e426
|
1923 1924 |
struct memcg_stock_pcp { struct mem_cgroup *cached; /* this never be root cgroup */ |
11c9ea4e8
|
1925 |
unsigned int nr_pages; |
cdec2e426
|
1926 |
struct work_struct work; |
26fe61684
|
1927 |
unsigned long flags; |
a0db00fcf
|
1928 |
#define FLUSHING_CACHED_CHARGE 0 |
cdec2e426
|
1929 1930 |
}; static DEFINE_PER_CPU(struct memcg_stock_pcp, memcg_stock); |
9f50fad65
|
1931 |
static DEFINE_MUTEX(percpu_charge_mutex); |
cdec2e426
|
1932 1933 |
/* |
11c9ea4e8
|
1934 |
* Try to consume stocked charge on this cpu. If success, one page is consumed |
cdec2e426
|
1935 1936 1937 1938 |
* from local stock and true is returned. If the stock is 0 or charges from a * cgroup which is not current target, returns false. This stock will be * refilled. */ |
c0ff4b854
|
1939 |
static bool consume_stock(struct mem_cgroup *memcg) |
cdec2e426
|
1940 1941 1942 1943 1944 |
{ struct memcg_stock_pcp *stock; bool ret = true; stock = &get_cpu_var(memcg_stock); |
c0ff4b854
|
1945 |
if (memcg == stock->cached && stock->nr_pages) |
11c9ea4e8
|
1946 |
stock->nr_pages--; |
cdec2e426
|
1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 |
else /* need to call res_counter_charge */ ret = false; put_cpu_var(memcg_stock); return ret; } /* * Returns stocks cached in percpu to res_counter and reset cached information. */ static void drain_stock(struct memcg_stock_pcp *stock) { struct mem_cgroup *old = stock->cached; |
11c9ea4e8
|
1959 1960 1961 1962 |
if (stock->nr_pages) { unsigned long bytes = stock->nr_pages * PAGE_SIZE; res_counter_uncharge(&old->res, bytes); |
cdec2e426
|
1963 |
if (do_swap_account) |
11c9ea4e8
|
1964 1965 |
res_counter_uncharge(&old->memsw, bytes); stock->nr_pages = 0; |
cdec2e426
|
1966 1967 |
} stock->cached = NULL; |
cdec2e426
|
1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 |
} /* * This must be called under preempt disabled or must be called by * a thread which is pinned to local cpu. */ static void drain_local_stock(struct work_struct *dummy) { struct memcg_stock_pcp *stock = &__get_cpu_var(memcg_stock); drain_stock(stock); |
26fe61684
|
1978 |
clear_bit(FLUSHING_CACHED_CHARGE, &stock->flags); |
cdec2e426
|
1979 1980 1981 1982 |
} /* * Cache charges(val) which is from res_counter, to local per_cpu area. |
320cc51d9
|
1983 |
* This will be consumed by consume_stock() function, later. |
cdec2e426
|
1984 |
*/ |
c0ff4b854
|
1985 |
static void refill_stock(struct mem_cgroup *memcg, unsigned int nr_pages) |
cdec2e426
|
1986 1987 |
{ struct memcg_stock_pcp *stock = &get_cpu_var(memcg_stock); |
c0ff4b854
|
1988 |
if (stock->cached != memcg) { /* reset if necessary */ |
cdec2e426
|
1989 |
drain_stock(stock); |
c0ff4b854
|
1990 |
stock->cached = memcg; |
cdec2e426
|
1991 |
} |
11c9ea4e8
|
1992 |
stock->nr_pages += nr_pages; |
cdec2e426
|
1993 1994 1995 1996 |
put_cpu_var(memcg_stock); } /* |
c0ff4b854
|
1997 |
* Drains all per-CPU charge caches for given root_memcg resp. subtree |
d38144b7a
|
1998 1999 |
* of the hierarchy under it. sync flag says whether we should block * until the work is done. |
cdec2e426
|
2000 |
*/ |
c0ff4b854
|
2001 |
static void drain_all_stock(struct mem_cgroup *root_memcg, bool sync) |
cdec2e426
|
2002 |
{ |
26fe61684
|
2003 |
int cpu, curcpu; |
d38144b7a
|
2004 |
|
cdec2e426
|
2005 |
/* Notify other cpus that system-wide "drain" is running */ |
cdec2e426
|
2006 |
get_online_cpus(); |
5af12d0ef
|
2007 |
curcpu = get_cpu(); |
cdec2e426
|
2008 2009 |
for_each_online_cpu(cpu) { struct memcg_stock_pcp *stock = &per_cpu(memcg_stock, cpu); |
c0ff4b854
|
2010 |
struct mem_cgroup *memcg; |
26fe61684
|
2011 |
|
c0ff4b854
|
2012 2013 |
memcg = stock->cached; if (!memcg || !stock->nr_pages) |
26fe61684
|
2014 |
continue; |
c0ff4b854
|
2015 |
if (!mem_cgroup_same_or_subtree(root_memcg, memcg)) |
3e92041d6
|
2016 |
continue; |
d1a05b697
|
2017 2018 2019 2020 2021 2022 |
if (!test_and_set_bit(FLUSHING_CACHED_CHARGE, &stock->flags)) { if (cpu == curcpu) drain_local_stock(&stock->work); else schedule_work_on(cpu, &stock->work); } |
cdec2e426
|
2023 |
} |
5af12d0ef
|
2024 |
put_cpu(); |
d38144b7a
|
2025 2026 2027 2028 2029 2030 |
if (!sync) goto out; for_each_online_cpu(cpu) { struct memcg_stock_pcp *stock = &per_cpu(memcg_stock, cpu); |
9f50fad65
|
2031 |
if (test_bit(FLUSHING_CACHED_CHARGE, &stock->flags)) |
d38144b7a
|
2032 2033 2034 |
flush_work(&stock->work); } out: |
cdec2e426
|
2035 |
put_online_cpus(); |
d38144b7a
|
2036 2037 2038 2039 2040 2041 2042 2043 |
} /* * Tries to drain stocked charges in other cpus. This function is asynchronous * and just put a work per cpu for draining localy on each cpu. Caller can * expects some charges will be back to res_counter later but cannot wait for * it. */ |
c0ff4b854
|
2044 |
static void drain_all_stock_async(struct mem_cgroup *root_memcg) |
d38144b7a
|
2045 |
{ |
9f50fad65
|
2046 2047 2048 2049 2050 |
/* * If someone calls draining, avoid adding more kworker runs. */ if (!mutex_trylock(&percpu_charge_mutex)) return; |
c0ff4b854
|
2051 |
drain_all_stock(root_memcg, false); |
9f50fad65
|
2052 |
mutex_unlock(&percpu_charge_mutex); |
cdec2e426
|
2053 2054 2055 |
} /* This is a synchronous drain interface. */ |
c0ff4b854
|
2056 |
static void drain_all_stock_sync(struct mem_cgroup *root_memcg) |
cdec2e426
|
2057 2058 |
{ /* called when force_empty is called */ |
9f50fad65
|
2059 |
mutex_lock(&percpu_charge_mutex); |
c0ff4b854
|
2060 |
drain_all_stock(root_memcg, true); |
9f50fad65
|
2061 |
mutex_unlock(&percpu_charge_mutex); |
cdec2e426
|
2062 |
} |
711d3d2c9
|
2063 2064 2065 2066 |
/* * This function drains percpu counter value from DEAD cpu and * move it to local cpu. Note that this function can be preempted. */ |
c0ff4b854
|
2067 |
static void mem_cgroup_drain_pcp_counter(struct mem_cgroup *memcg, int cpu) |
711d3d2c9
|
2068 2069 |
{ int i; |
c0ff4b854
|
2070 |
spin_lock(&memcg->pcp_counter_lock); |
6104621de
|
2071 |
for (i = 0; i < MEM_CGROUP_STAT_NSTATS; i++) { |
c0ff4b854
|
2072 |
long x = per_cpu(memcg->stat->count[i], cpu); |
711d3d2c9
|
2073 |
|
c0ff4b854
|
2074 2075 |
per_cpu(memcg->stat->count[i], cpu) = 0; memcg->nocpu_base.count[i] += x; |
711d3d2c9
|
2076 |
} |
e9f8974f2
|
2077 |
for (i = 0; i < MEM_CGROUP_EVENTS_NSTATS; i++) { |
c0ff4b854
|
2078 |
unsigned long x = per_cpu(memcg->stat->events[i], cpu); |
e9f8974f2
|
2079 |
|
c0ff4b854
|
2080 2081 |
per_cpu(memcg->stat->events[i], cpu) = 0; memcg->nocpu_base.events[i] += x; |
e9f8974f2
|
2082 |
} |
c0ff4b854
|
2083 |
spin_unlock(&memcg->pcp_counter_lock); |
711d3d2c9
|
2084 2085 2086 |
} static int __cpuinit memcg_cpu_hotplug_callback(struct notifier_block *nb, |
cdec2e426
|
2087 2088 2089 2090 2091 |
unsigned long action, void *hcpu) { int cpu = (unsigned long)hcpu; struct memcg_stock_pcp *stock; |
711d3d2c9
|
2092 |
struct mem_cgroup *iter; |
cdec2e426
|
2093 |
|
619d094b5
|
2094 |
if (action == CPU_ONLINE) |
1489ebad8
|
2095 |
return NOTIFY_OK; |
1489ebad8
|
2096 |
|
d833049bd
|
2097 |
if (action != CPU_DEAD && action != CPU_DEAD_FROZEN) |
cdec2e426
|
2098 |
return NOTIFY_OK; |
711d3d2c9
|
2099 |
|
9f3a0d093
|
2100 |
for_each_mem_cgroup(iter) |
711d3d2c9
|
2101 |
mem_cgroup_drain_pcp_counter(iter, cpu); |
cdec2e426
|
2102 2103 2104 2105 |
stock = &per_cpu(memcg_stock, cpu); drain_stock(stock); return NOTIFY_OK; } |
4b5343346
|
2106 2107 2108 2109 2110 2111 2112 2113 2114 |
/* See __mem_cgroup_try_charge() for details */ enum { CHARGE_OK, /* success */ CHARGE_RETRY, /* need to retry but retry is not bad */ CHARGE_NOMEM, /* we can't do more. return -ENOMEM */ CHARGE_WOULDBLOCK, /* GFP_WAIT wasn't set and no enough res. */ CHARGE_OOM_DIE, /* the current is killed because of OOM */ }; |
c0ff4b854
|
2115 |
static int mem_cgroup_do_charge(struct mem_cgroup *memcg, gfp_t gfp_mask, |
7ec99d621
|
2116 |
unsigned int nr_pages, bool oom_check) |
4b5343346
|
2117 |
{ |
7ec99d621
|
2118 |
unsigned long csize = nr_pages * PAGE_SIZE; |
4b5343346
|
2119 2120 2121 2122 |
struct mem_cgroup *mem_over_limit; struct res_counter *fail_res; unsigned long flags = 0; int ret; |
c0ff4b854
|
2123 |
ret = res_counter_charge(&memcg->res, csize, &fail_res); |
4b5343346
|
2124 2125 2126 2127 |
if (likely(!ret)) { if (!do_swap_account) return CHARGE_OK; |
c0ff4b854
|
2128 |
ret = res_counter_charge(&memcg->memsw, csize, &fail_res); |
4b5343346
|
2129 2130 |
if (likely(!ret)) return CHARGE_OK; |
c0ff4b854
|
2131 |
res_counter_uncharge(&memcg->res, csize); |
4b5343346
|
2132 2133 2134 2135 |
mem_over_limit = mem_cgroup_from_res_counter(fail_res, memsw); flags |= MEM_CGROUP_RECLAIM_NOSWAP; } else mem_over_limit = mem_cgroup_from_res_counter(fail_res, res); |
9221edb71
|
2136 |
/* |
7ec99d621
|
2137 2138 |
* nr_pages can be either a huge page (HPAGE_PMD_NR), a batch * of regular pages (CHARGE_BATCH), or a single regular page (1). |
9221edb71
|
2139 2140 2141 2142 |
* * Never reclaim on behalf of optional batching, retry with a * single page instead. */ |
7ec99d621
|
2143 |
if (nr_pages == CHARGE_BATCH) |
4b5343346
|
2144 2145 2146 2147 |
return CHARGE_RETRY; if (!(gfp_mask & __GFP_WAIT)) return CHARGE_WOULDBLOCK; |
5660048cc
|
2148 |
ret = mem_cgroup_reclaim(mem_over_limit, gfp_mask, flags); |
7ec99d621
|
2149 |
if (mem_cgroup_margin(mem_over_limit) >= nr_pages) |
19942822d
|
2150 |
return CHARGE_RETRY; |
4b5343346
|
2151 |
/* |
19942822d
|
2152 2153 2154 2155 2156 2157 2158 |
* Even though the limit is exceeded at this point, reclaim * may have been able to free some pages. Retry the charge * before killing the task. * * Only for regular pages, though: huge pages are rather * unlikely to succeed so close to the limit, and we fall back * to regular pages anyway in case of failure. |
4b5343346
|
2159 |
*/ |
7ec99d621
|
2160 |
if (nr_pages == 1 && ret) |
4b5343346
|
2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 |
return CHARGE_RETRY; /* * At task move, charge accounts can be doubly counted. So, it's * better to wait until the end of task_move if something is going on. */ if (mem_cgroup_wait_acct_move(mem_over_limit)) return CHARGE_RETRY; /* If we don't need to call oom-killer at el, return immediately */ if (!oom_check) return CHARGE_NOMEM; /* check OOM */ |
e845e1993
|
2174 |
if (!mem_cgroup_handle_oom(mem_over_limit, gfp_mask, get_order(csize))) |
4b5343346
|
2175 2176 2177 2178 |
return CHARGE_OOM_DIE; return CHARGE_RETRY; } |
cdec2e426
|
2179 |
/* |
38c5d72f3
|
2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 |
* __mem_cgroup_try_charge() does * 1. detect memcg to be charged against from passed *mm and *ptr, * 2. update res_counter * 3. call memory reclaim if necessary. * * In some special case, if the task is fatal, fatal_signal_pending() or * has TIF_MEMDIE, this function returns -EINTR while writing root_mem_cgroup * to *ptr. There are two reasons for this. 1: fatal threads should quit as soon * as possible without any hazards. 2: all pages should have a valid * pc->mem_cgroup. If mm is NULL and the caller doesn't pass a valid memcg * pointer, that is treated as a charge to root_mem_cgroup. * * So __mem_cgroup_try_charge() will return * 0 ... on success, filling *ptr with a valid memcg pointer. * -ENOMEM ... charge failure because of resource limits. * -EINTR ... if thread is fatal. *ptr is filled with root_mem_cgroup. * * Unlike the exported interface, an "oom" parameter is added. if oom==true, * the oom-killer can be invoked. |
8a9f3ccd2
|
2199 |
*/ |
f817ed485
|
2200 |
static int __mem_cgroup_try_charge(struct mm_struct *mm, |
ec1685109
|
2201 |
gfp_t gfp_mask, |
7ec99d621
|
2202 |
unsigned int nr_pages, |
c0ff4b854
|
2203 |
struct mem_cgroup **ptr, |
7ec99d621
|
2204 |
bool oom) |
8a9f3ccd2
|
2205 |
{ |
7ec99d621
|
2206 |
unsigned int batch = max(CHARGE_BATCH, nr_pages); |
4b5343346
|
2207 |
int nr_oom_retries = MEM_CGROUP_RECLAIM_RETRIES; |
c0ff4b854
|
2208 |
struct mem_cgroup *memcg = NULL; |
4b5343346
|
2209 |
int ret; |
a636b327f
|
2210 |
|
867578cbc
|
2211 2212 2213 2214 2215 2216 2217 2218 |
/* * Unlike gloval-vm's OOM-kill, we're not in memory shortage * in system level. So, allow to go ahead dying process in addition to * MEMDIE process. */ if (unlikely(test_thread_flag(TIF_MEMDIE) || fatal_signal_pending(current))) goto bypass; |
a636b327f
|
2219 |
|
8a9f3ccd2
|
2220 |
/* |
3be91277e
|
2221 2222 |
* We always charge the cgroup the mm_struct belongs to. * The mm_struct's mem_cgroup changes on task migration if the |
8a9f3ccd2
|
2223 |
* thread group leader migrates. It's possible that mm is not |
24467cacc
|
2224 |
* set, if so charge the root memcg (happens for pagecache usage). |
8a9f3ccd2
|
2225 |
*/ |
c0ff4b854
|
2226 |
if (!*ptr && !mm) |
38c5d72f3
|
2227 |
*ptr = root_mem_cgroup; |
f75ca9620
|
2228 |
again: |
c0ff4b854
|
2229 2230 2231 2232 |
if (*ptr) { /* css should be a valid one */ memcg = *ptr; VM_BUG_ON(css_is_removed(&memcg->css)); if (mem_cgroup_is_root(memcg)) |
f75ca9620
|
2233 |
goto done; |
c0ff4b854
|
2234 |
if (nr_pages == 1 && consume_stock(memcg)) |
f75ca9620
|
2235 |
goto done; |
c0ff4b854
|
2236 |
css_get(&memcg->css); |
4b5343346
|
2237 |
} else { |
f75ca9620
|
2238 |
struct task_struct *p; |
54595fe26
|
2239 |
|
f75ca9620
|
2240 2241 |
rcu_read_lock(); p = rcu_dereference(mm->owner); |
f75ca9620
|
2242 |
/* |
ebb76ce16
|
2243 |
* Because we don't have task_lock(), "p" can exit. |
c0ff4b854
|
2244 |
* In that case, "memcg" can point to root or p can be NULL with |
ebb76ce16
|
2245 2246 2247 2248 2249 2250 |
* race with swapoff. Then, we have small risk of mis-accouning. * But such kind of mis-account by race always happens because * we don't have cgroup_mutex(). It's overkill and we allo that * small race, here. * (*) swapoff at el will charge against mm-struct not against * task-struct. So, mm->owner can be NULL. |
f75ca9620
|
2251 |
*/ |
c0ff4b854
|
2252 |
memcg = mem_cgroup_from_task(p); |
38c5d72f3
|
2253 2254 2255 |
if (!memcg) memcg = root_mem_cgroup; if (mem_cgroup_is_root(memcg)) { |
f75ca9620
|
2256 2257 2258 |
rcu_read_unlock(); goto done; } |
c0ff4b854
|
2259 |
if (nr_pages == 1 && consume_stock(memcg)) { |
f75ca9620
|
2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 |
/* * It seems dagerous to access memcg without css_get(). * But considering how consume_stok works, it's not * necessary. If consume_stock success, some charges * from this memcg are cached on this cpu. So, we * don't need to call css_get()/css_tryget() before * calling consume_stock(). */ rcu_read_unlock(); goto done; } /* after here, we may be blocked. we need to get refcnt */ |
c0ff4b854
|
2272 |
if (!css_tryget(&memcg->css)) { |
f75ca9620
|
2273 2274 2275 2276 2277 |
rcu_read_unlock(); goto again; } rcu_read_unlock(); } |
8a9f3ccd2
|
2278 |
|
4b5343346
|
2279 2280 |
do { bool oom_check; |
7a81b88cb
|
2281 |
|
4b5343346
|
2282 |
/* If killed, bypass charge */ |
f75ca9620
|
2283 |
if (fatal_signal_pending(current)) { |
c0ff4b854
|
2284 |
css_put(&memcg->css); |
4b5343346
|
2285 |
goto bypass; |
f75ca9620
|
2286 |
} |
6d61ef409
|
2287 |
|
4b5343346
|
2288 2289 2290 2291 |
oom_check = false; if (oom && !nr_oom_retries) { oom_check = true; nr_oom_retries = MEM_CGROUP_RECLAIM_RETRIES; |
cdec2e426
|
2292 |
} |
66e1707bc
|
2293 |
|
c0ff4b854
|
2294 |
ret = mem_cgroup_do_charge(memcg, gfp_mask, batch, oom_check); |
4b5343346
|
2295 2296 2297 2298 |
switch (ret) { case CHARGE_OK: break; case CHARGE_RETRY: /* not in OOM situation but retry */ |
7ec99d621
|
2299 |
batch = nr_pages; |
c0ff4b854
|
2300 2301 |
css_put(&memcg->css); memcg = NULL; |
f75ca9620
|
2302 |
goto again; |
4b5343346
|
2303 |
case CHARGE_WOULDBLOCK: /* !__GFP_WAIT */ |
c0ff4b854
|
2304 |
css_put(&memcg->css); |
4b5343346
|
2305 2306 |
goto nomem; case CHARGE_NOMEM: /* OOM routine works */ |
f75ca9620
|
2307 |
if (!oom) { |
c0ff4b854
|
2308 |
css_put(&memcg->css); |
867578cbc
|
2309 |
goto nomem; |
f75ca9620
|
2310 |
} |
4b5343346
|
2311 2312 2313 2314 |
/* If oom, we never return -ENOMEM */ nr_oom_retries--; break; case CHARGE_OOM_DIE: /* Killed by OOM Killer */ |
c0ff4b854
|
2315 |
css_put(&memcg->css); |
867578cbc
|
2316 |
goto bypass; |
66e1707bc
|
2317 |
} |
4b5343346
|
2318 |
} while (ret != CHARGE_OK); |
7ec99d621
|
2319 |
if (batch > nr_pages) |
c0ff4b854
|
2320 2321 |
refill_stock(memcg, batch - nr_pages); css_put(&memcg->css); |
0c3e73e84
|
2322 |
done: |
c0ff4b854
|
2323 |
*ptr = memcg; |
7a81b88cb
|
2324 2325 |
return 0; nomem: |
c0ff4b854
|
2326 |
*ptr = NULL; |
7a81b88cb
|
2327 |
return -ENOMEM; |
867578cbc
|
2328 |
bypass: |
38c5d72f3
|
2329 2330 |
*ptr = root_mem_cgroup; return -EINTR; |
7a81b88cb
|
2331 |
} |
8a9f3ccd2
|
2332 |
|
a3b2d6926
|
2333 |
/* |
a3032a2c1
|
2334 2335 2336 2337 |
* Somemtimes we have to undo a charge we got by try_charge(). * This function is for that and do uncharge, put css's refcnt. * gotten by try_charge(). */ |
c0ff4b854
|
2338 |
static void __mem_cgroup_cancel_charge(struct mem_cgroup *memcg, |
e7018b8d2
|
2339 |
unsigned int nr_pages) |
a3032a2c1
|
2340 |
{ |
c0ff4b854
|
2341 |
if (!mem_cgroup_is_root(memcg)) { |
e7018b8d2
|
2342 |
unsigned long bytes = nr_pages * PAGE_SIZE; |
c0ff4b854
|
2343 |
res_counter_uncharge(&memcg->res, bytes); |
a3032a2c1
|
2344 |
if (do_swap_account) |
c0ff4b854
|
2345 |
res_counter_uncharge(&memcg->memsw, bytes); |
a3032a2c1
|
2346 |
} |
854ffa8d1
|
2347 |
} |
a3032a2c1
|
2348 |
/* |
d01dd17f1
|
2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 |
* Cancel chrages in this cgroup....doesn't propagate to parent cgroup. * This is useful when moving usage to parent cgroup. */ static void __mem_cgroup_cancel_local_charge(struct mem_cgroup *memcg, unsigned int nr_pages) { unsigned long bytes = nr_pages * PAGE_SIZE; if (mem_cgroup_is_root(memcg)) return; res_counter_uncharge_until(&memcg->res, memcg->res.parent, bytes); if (do_swap_account) res_counter_uncharge_until(&memcg->memsw, memcg->memsw.parent, bytes); } /* |
a3b2d6926
|
2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 |
* A helper function to get mem_cgroup from ID. must be called under * rcu_read_lock(). The caller must check css_is_removed() or some if * it's concern. (dropping refcnt from swap can be called against removed * memcg.) */ static struct mem_cgroup *mem_cgroup_lookup(unsigned short id) { struct cgroup_subsys_state *css; /* ID 0 is unused ID */ if (!id) return NULL; css = css_lookup(&mem_cgroup_subsys, id); if (!css) return NULL; |
b21451459
|
2382 |
return mem_cgroup_from_css(css); |
a3b2d6926
|
2383 |
} |
e42d9d5d4
|
2384 |
struct mem_cgroup *try_get_mem_cgroup_from_page(struct page *page) |
b5a84319a
|
2385 |
{ |
c0ff4b854
|
2386 |
struct mem_cgroup *memcg = NULL; |
3c776e646
|
2387 |
struct page_cgroup *pc; |
a3b2d6926
|
2388 |
unsigned short id; |
b5a84319a
|
2389 |
swp_entry_t ent; |
3c776e646
|
2390 |
VM_BUG_ON(!PageLocked(page)); |
3c776e646
|
2391 |
pc = lookup_page_cgroup(page); |
c0bd3f63c
|
2392 |
lock_page_cgroup(pc); |
a3b2d6926
|
2393 |
if (PageCgroupUsed(pc)) { |
c0ff4b854
|
2394 2395 2396 |
memcg = pc->mem_cgroup; if (memcg && !css_tryget(&memcg->css)) memcg = NULL; |
e42d9d5d4
|
2397 |
} else if (PageSwapCache(page)) { |
3c776e646
|
2398 |
ent.val = page_private(page); |
9fb4b7cc0
|
2399 |
id = lookup_swap_cgroup_id(ent); |
a3b2d6926
|
2400 |
rcu_read_lock(); |
c0ff4b854
|
2401 2402 2403 |
memcg = mem_cgroup_lookup(id); if (memcg && !css_tryget(&memcg->css)) memcg = NULL; |
a3b2d6926
|
2404 |
rcu_read_unlock(); |
3c776e646
|
2405 |
} |
c0bd3f63c
|
2406 |
unlock_page_cgroup(pc); |
c0ff4b854
|
2407 |
return memcg; |
b5a84319a
|
2408 |
} |
c0ff4b854
|
2409 |
static void __mem_cgroup_commit_charge(struct mem_cgroup *memcg, |
5564e88ba
|
2410 |
struct page *page, |
7ec99d621
|
2411 |
unsigned int nr_pages, |
9ce70c024
|
2412 2413 |
enum charge_type ctype, bool lrucare) |
7a81b88cb
|
2414 |
{ |
ce587e65e
|
2415 |
struct page_cgroup *pc = lookup_page_cgroup(page); |
9ce70c024
|
2416 |
struct zone *uninitialized_var(zone); |
fa9add641
|
2417 |
struct lruvec *lruvec; |
9ce70c024
|
2418 |
bool was_on_lru = false; |
b24028572
|
2419 |
bool anon; |
9ce70c024
|
2420 |
|
ca3e02141
|
2421 |
lock_page_cgroup(pc); |
90deb7883
|
2422 |
VM_BUG_ON(PageCgroupUsed(pc)); |
ca3e02141
|
2423 2424 2425 2426 |
/* * we don't need page_cgroup_lock about tail pages, becase they are not * accessed by any other context at this point. */ |
9ce70c024
|
2427 2428 2429 2430 2431 2432 2433 2434 2435 |
/* * In some cases, SwapCache and FUSE(splice_buf->radixtree), the page * may already be on some other mem_cgroup's LRU. Take care of it. */ if (lrucare) { zone = page_zone(page); spin_lock_irq(&zone->lru_lock); if (PageLRU(page)) { |
fa9add641
|
2436 |
lruvec = mem_cgroup_zone_lruvec(zone, pc->mem_cgroup); |
9ce70c024
|
2437 |
ClearPageLRU(page); |
fa9add641
|
2438 |
del_page_from_lru_list(page, lruvec, page_lru(page)); |
9ce70c024
|
2439 2440 2441 |
was_on_lru = true; } } |
c0ff4b854
|
2442 |
pc->mem_cgroup = memcg; |
261fb61a8
|
2443 2444 2445 2446 2447 2448 2449 |
/* * We access a page_cgroup asynchronously without lock_page_cgroup(). * Especially when a page_cgroup is taken from a page, pc->mem_cgroup * is accessed after testing USED bit. To make pc->mem_cgroup visible * before USED bit, we need memory barrier here. * See mem_cgroup_add_lru_list(), etc. */ |
08e552c69
|
2450 |
smp_wmb(); |
b24028572
|
2451 |
SetPageCgroupUsed(pc); |
3be91277e
|
2452 |
|
9ce70c024
|
2453 2454 |
if (lrucare) { if (was_on_lru) { |
fa9add641
|
2455 |
lruvec = mem_cgroup_zone_lruvec(zone, pc->mem_cgroup); |
9ce70c024
|
2456 2457 |
VM_BUG_ON(PageLRU(page)); SetPageLRU(page); |
fa9add641
|
2458 |
add_page_to_lru_list(page, lruvec, page_lru(page)); |
9ce70c024
|
2459 2460 2461 |
} spin_unlock_irq(&zone->lru_lock); } |
41326c17f
|
2462 |
if (ctype == MEM_CGROUP_CHARGE_TYPE_ANON) |
b24028572
|
2463 2464 2465 2466 2467 |
anon = true; else anon = false; mem_cgroup_charge_statistics(memcg, anon, nr_pages); |
52d4b9ac0
|
2468 |
unlock_page_cgroup(pc); |
9ce70c024
|
2469 |
|
430e48631
|
2470 2471 2472 2473 2474 |
/* * "charge_statistics" updated event counter. Then, check it. * Insert ancestor (and ancestor's ancestors), to softlimit RB-tree. * if they exceeds softlimit. */ |
c0ff4b854
|
2475 |
memcg_check_events(memcg, page); |
7a81b88cb
|
2476 |
} |
66e1707bc
|
2477 |
|
ca3e02141
|
2478 |
#ifdef CONFIG_TRANSPARENT_HUGEPAGE |
a0db00fcf
|
2479 |
#define PCGF_NOCOPY_AT_SPLIT (1 << PCG_LOCK | 1 << PCG_MIGRATION) |
ca3e02141
|
2480 2481 |
/* * Because tail pages are not marked as "used", set it. We're under |
e94c8a9cb
|
2482 2483 2484 |
* zone->lru_lock, 'splitting on pmd' and compound_lock. * charge/uncharge will be never happen and move_account() is done under * compound_lock(), so we don't have to take care of races. |
ca3e02141
|
2485 |
*/ |
e94c8a9cb
|
2486 |
void mem_cgroup_split_huge_fixup(struct page *head) |
ca3e02141
|
2487 2488 |
{ struct page_cgroup *head_pc = lookup_page_cgroup(head); |
e94c8a9cb
|
2489 2490 |
struct page_cgroup *pc; int i; |
ca3e02141
|
2491 |
|
3d37c4a91
|
2492 2493 |
if (mem_cgroup_disabled()) return; |
e94c8a9cb
|
2494 2495 2496 2497 |
for (i = 1; i < HPAGE_PMD_NR; i++) { pc = head_pc + i; pc->mem_cgroup = head_pc->mem_cgroup; smp_wmb();/* see __commit_charge() */ |
e94c8a9cb
|
2498 2499 |
pc->flags = head_pc->flags & ~PCGF_NOCOPY_AT_SPLIT; } |
ca3e02141
|
2500 |
} |
12d271078
|
2501 |
#endif /* CONFIG_TRANSPARENT_HUGEPAGE */ |
ca3e02141
|
2502 |
|
f817ed485
|
2503 |
/** |
de3638d9c
|
2504 |
* mem_cgroup_move_account - move account of the page |
5564e88ba
|
2505 |
* @page: the page |
7ec99d621
|
2506 |
* @nr_pages: number of regular pages (>1 for huge pages) |
f817ed485
|
2507 2508 2509 2510 2511 |
* @pc: page_cgroup of the page. * @from: mem_cgroup which the page is moved from. * @to: mem_cgroup which the page is moved to. @from != @to. * * The caller must confirm following. |
08e552c69
|
2512 |
* - page is not on LRU (isolate_page() is useful.) |
7ec99d621
|
2513 |
* - compound_lock is held when nr_pages > 1 |
f817ed485
|
2514 |
* |
2f3479b14
|
2515 2516 |
* This function doesn't do "charge" to new cgroup and doesn't do "uncharge" * from old cgroup. |
f817ed485
|
2517 |
*/ |
7ec99d621
|
2518 2519 2520 2521 |
static int mem_cgroup_move_account(struct page *page, unsigned int nr_pages, struct page_cgroup *pc, struct mem_cgroup *from, |
2f3479b14
|
2522 |
struct mem_cgroup *to) |
f817ed485
|
2523 |
{ |
de3638d9c
|
2524 2525 |
unsigned long flags; int ret; |
b24028572
|
2526 |
bool anon = PageAnon(page); |
987eba66e
|
2527 |
|
f817ed485
|
2528 |
VM_BUG_ON(from == to); |
5564e88ba
|
2529 |
VM_BUG_ON(PageLRU(page)); |
de3638d9c
|
2530 2531 2532 2533 2534 2535 2536 |
/* * The page is isolated from LRU. So, collapse function * will not handle this page. But page splitting can happen. * Do this check under compound_page_lock(). The caller should * hold it. */ ret = -EBUSY; |
7ec99d621
|
2537 |
if (nr_pages > 1 && !PageTransHuge(page)) |
de3638d9c
|
2538 2539 2540 2541 2542 2543 2544 |
goto out; lock_page_cgroup(pc); ret = -EINVAL; if (!PageCgroupUsed(pc) || pc->mem_cgroup != from) goto unlock; |
312734c04
|
2545 |
move_lock_mem_cgroup(from, &flags); |
f817ed485
|
2546 |
|
2ff76f119
|
2547 |
if (!anon && page_mapped(page)) { |
c62b1a3b3
|
2548 2549 2550 2551 2552 |
/* Update mapped_file data for mem_cgroup */ preempt_disable(); __this_cpu_dec(from->stat->count[MEM_CGROUP_STAT_FILE_MAPPED]); __this_cpu_inc(to->stat->count[MEM_CGROUP_STAT_FILE_MAPPED]); preempt_enable(); |
d69b042f3
|
2553 |
} |
b24028572
|
2554 |
mem_cgroup_charge_statistics(from, anon, -nr_pages); |
d69b042f3
|
2555 |
|
854ffa8d1
|
2556 |
/* caller should have done css_get */ |
08e552c69
|
2557 |
pc->mem_cgroup = to; |
b24028572
|
2558 |
mem_cgroup_charge_statistics(to, anon, nr_pages); |
887032670
|
2559 2560 2561 |
/* * We charges against "to" which may not have any tasks. Then, "to" * can be under rmdir(). But in current implementation, caller of |
4ffef5fef
|
2562 |
* this function is just force_empty() and move charge, so it's |
25985edce
|
2563 |
* guaranteed that "to" is never removed. So, we don't check rmdir |
4ffef5fef
|
2564 |
* status here. |
887032670
|
2565 |
*/ |
312734c04
|
2566 |
move_unlock_mem_cgroup(from, &flags); |
de3638d9c
|
2567 2568 |
ret = 0; unlock: |
57f9fd7d2
|
2569 |
unlock_page_cgroup(pc); |
d2265e6fa
|
2570 2571 2572 |
/* * check events */ |
5564e88ba
|
2573 2574 |
memcg_check_events(to, page); memcg_check_events(from, page); |
de3638d9c
|
2575 |
out: |
f817ed485
|
2576 2577 2578 2579 2580 2581 |
return ret; } /* * move charges to its parent. */ |
5564e88ba
|
2582 2583 |
static int mem_cgroup_move_parent(struct page *page, struct page_cgroup *pc, |
6068bf010
|
2584 |
struct mem_cgroup *child) |
f817ed485
|
2585 |
{ |
f817ed485
|
2586 |
struct mem_cgroup *parent; |
7ec99d621
|
2587 |
unsigned int nr_pages; |
4be4489fe
|
2588 |
unsigned long uninitialized_var(flags); |
f817ed485
|
2589 2590 2591 |
int ret; /* Is ROOT ? */ |
cc926f784
|
2592 |
if (mem_cgroup_is_root(child)) |
f817ed485
|
2593 |
return -EINVAL; |
57f9fd7d2
|
2594 2595 2596 2597 2598 |
ret = -EBUSY; if (!get_page_unless_zero(page)) goto out; if (isolate_lru_page(page)) goto put; |
52dbb9050
|
2599 |
|
7ec99d621
|
2600 |
nr_pages = hpage_nr_pages(page); |
08e552c69
|
2601 |
|
cc926f784
|
2602 2603 2604 2605 2606 2607 |
parent = parent_mem_cgroup(child); /* * If no parent, move charges to root cgroup. */ if (!parent) parent = root_mem_cgroup; |
f817ed485
|
2608 |
|
7ec99d621
|
2609 |
if (nr_pages > 1) |
987eba66e
|
2610 |
flags = compound_lock_irqsave(page); |
cc926f784
|
2611 |
ret = mem_cgroup_move_account(page, nr_pages, |
2f3479b14
|
2612 |
pc, child, parent); |
cc926f784
|
2613 2614 |
if (!ret) __mem_cgroup_cancel_local_charge(child, nr_pages); |
8dba474f0
|
2615 |
|
7ec99d621
|
2616 |
if (nr_pages > 1) |
987eba66e
|
2617 |
compound_unlock_irqrestore(page, flags); |
08e552c69
|
2618 |
putback_lru_page(page); |
57f9fd7d2
|
2619 |
put: |
40d58138f
|
2620 |
put_page(page); |
57f9fd7d2
|
2621 |
out: |
f817ed485
|
2622 2623 |
return ret; } |
7a81b88cb
|
2624 2625 2626 2627 2628 2629 2630 |
/* * Charge the memory controller for page usage. * Return * 0 if the charge was successful * < 0 if the cgroup is over its limit */ static int mem_cgroup_charge_common(struct page *page, struct mm_struct *mm, |
73045c47b
|
2631 |
gfp_t gfp_mask, enum charge_type ctype) |
7a81b88cb
|
2632 |
{ |
c0ff4b854
|
2633 |
struct mem_cgroup *memcg = NULL; |
7ec99d621
|
2634 |
unsigned int nr_pages = 1; |
8493ae439
|
2635 |
bool oom = true; |
7a81b88cb
|
2636 |
int ret; |
ec1685109
|
2637 |
|
37c2ac787
|
2638 |
if (PageTransHuge(page)) { |
7ec99d621
|
2639 |
nr_pages <<= compound_order(page); |
37c2ac787
|
2640 |
VM_BUG_ON(!PageTransHuge(page)); |
8493ae439
|
2641 2642 2643 2644 2645 |
/* * Never OOM-kill a process for a huge page. The * fault handler will fall back to regular pages. */ oom = false; |
37c2ac787
|
2646 |
} |
7a81b88cb
|
2647 |
|
c0ff4b854
|
2648 |
ret = __mem_cgroup_try_charge(mm, gfp_mask, nr_pages, &memcg, oom); |
38c5d72f3
|
2649 |
if (ret == -ENOMEM) |
7a81b88cb
|
2650 |
return ret; |
ce587e65e
|
2651 |
__mem_cgroup_commit_charge(memcg, page, nr_pages, ctype, false); |
8a9f3ccd2
|
2652 |
return 0; |
8a9f3ccd2
|
2653 |
} |
7a81b88cb
|
2654 2655 |
int mem_cgroup_newpage_charge(struct page *page, struct mm_struct *mm, gfp_t gfp_mask) |
217bc3194
|
2656 |
{ |
f8d665422
|
2657 |
if (mem_cgroup_disabled()) |
cede86acd
|
2658 |
return 0; |
7a0524cfc
|
2659 2660 2661 |
VM_BUG_ON(page_mapped(page)); VM_BUG_ON(page->mapping && !PageAnon(page)); VM_BUG_ON(!mm); |
217bc3194
|
2662 |
return mem_cgroup_charge_common(page, mm, gfp_mask, |
41326c17f
|
2663 |
MEM_CGROUP_CHARGE_TYPE_ANON); |
217bc3194
|
2664 |
} |
54595fe26
|
2665 2666 2667 |
/* * While swap-in, try_charge -> commit or cancel, the page is locked. * And when try_charge() successfully returns, one refcnt to memcg without |
21ae2956c
|
2668 |
* struct page_cgroup is acquired. This refcnt will be consumed by |
54595fe26
|
2669 2670 |
* "commit()" or removed by "cancel()" */ |
0435a2fdc
|
2671 2672 2673 2674 |
static int __mem_cgroup_try_charge_swapin(struct mm_struct *mm, struct page *page, gfp_t mask, struct mem_cgroup **memcgp) |
8c7c6e34a
|
2675 |
{ |
c0ff4b854
|
2676 |
struct mem_cgroup *memcg; |
90deb7883
|
2677 |
struct page_cgroup *pc; |
54595fe26
|
2678 |
int ret; |
8c7c6e34a
|
2679 |
|
90deb7883
|
2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 |
pc = lookup_page_cgroup(page); /* * Every swap fault against a single page tries to charge the * page, bail as early as possible. shmem_unuse() encounters * already charged pages, too. The USED bit is protected by * the page lock, which serializes swap cache removal, which * in turn serializes uncharging. */ if (PageCgroupUsed(pc)) return 0; |
8c7c6e34a
|
2690 2691 |
if (!do_swap_account) goto charge_cur_mm; |
c0ff4b854
|
2692 2693 |
memcg = try_get_mem_cgroup_from_page(page); if (!memcg) |
54595fe26
|
2694 |
goto charge_cur_mm; |
72835c86c
|
2695 2696 |
*memcgp = memcg; ret = __mem_cgroup_try_charge(NULL, mask, 1, memcgp, true); |
c0ff4b854
|
2697 |
css_put(&memcg->css); |
38c5d72f3
|
2698 2699 |
if (ret == -EINTR) ret = 0; |
54595fe26
|
2700 |
return ret; |
8c7c6e34a
|
2701 |
charge_cur_mm: |
38c5d72f3
|
2702 2703 2704 2705 |
ret = __mem_cgroup_try_charge(mm, mask, 1, memcgp, true); if (ret == -EINTR) ret = 0; return ret; |
8c7c6e34a
|
2706 |
} |
0435a2fdc
|
2707 2708 2709 2710 2711 2712 |
int mem_cgroup_try_charge_swapin(struct mm_struct *mm, struct page *page, gfp_t gfp_mask, struct mem_cgroup **memcgp) { *memcgp = NULL; if (mem_cgroup_disabled()) return 0; |
bdf4f4d21
|
2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 |
/* * A racing thread's fault, or swapoff, may have already * updated the pte, and even removed page from swap cache: in * those cases unuse_pte()'s pte_same() test will fail; but * there's also a KSM case which does need to charge the page. */ if (!PageSwapCache(page)) { int ret; ret = __mem_cgroup_try_charge(mm, gfp_mask, 1, memcgp, true); if (ret == -EINTR) ret = 0; return ret; } |
0435a2fdc
|
2727 2728 |
return __mem_cgroup_try_charge_swapin(mm, page, gfp_mask, memcgp); } |
827a03d22
|
2729 2730 2731 2732 2733 2734 2735 2736 |
void mem_cgroup_cancel_charge_swapin(struct mem_cgroup *memcg) { if (mem_cgroup_disabled()) return; if (!memcg) return; __mem_cgroup_cancel_charge(memcg, 1); } |
83aae4c73
|
2737 |
static void |
72835c86c
|
2738 |
__mem_cgroup_commit_charge_swapin(struct page *page, struct mem_cgroup *memcg, |
83aae4c73
|
2739 |
enum charge_type ctype) |
7a81b88cb
|
2740 |
{ |
f8d665422
|
2741 |
if (mem_cgroup_disabled()) |
7a81b88cb
|
2742 |
return; |
72835c86c
|
2743 |
if (!memcg) |
7a81b88cb
|
2744 |
return; |
72835c86c
|
2745 |
cgroup_exclude_rmdir(&memcg->css); |
5a6475a4e
|
2746 |
|
ce587e65e
|
2747 |
__mem_cgroup_commit_charge(memcg, page, 1, ctype, true); |
8c7c6e34a
|
2748 2749 2750 |
/* * Now swap is on-memory. This means this page may be * counted both as mem and swap....double count. |
03f3c4336
|
2751 2752 2753 |
* Fix it by uncharging from memsw. Basically, this SwapCache is stable * under lock_page(). But in do_swap_page()::memory.c, reuse_swap_page() * may call delete_from_swap_cache() before reach here. |
8c7c6e34a
|
2754 |
*/ |
03f3c4336
|
2755 |
if (do_swap_account && PageSwapCache(page)) { |
8c7c6e34a
|
2756 |
swp_entry_t ent = {.val = page_private(page)}; |
86493009d
|
2757 |
mem_cgroup_uncharge_swap(ent); |
8c7c6e34a
|
2758 |
} |
887032670
|
2759 2760 2761 2762 2763 |
/* * At swapin, we may charge account against cgroup which has no tasks. * So, rmdir()->pre_destroy() can be called while we do this charge. * In that case, we need to call pre_destroy() again. check it here. */ |
72835c86c
|
2764 |
cgroup_release_and_wakeup_rmdir(&memcg->css); |
7a81b88cb
|
2765 |
} |
72835c86c
|
2766 2767 |
void mem_cgroup_commit_charge_swapin(struct page *page, struct mem_cgroup *memcg) |
83aae4c73
|
2768 |
{ |
72835c86c
|
2769 |
__mem_cgroup_commit_charge_swapin(page, memcg, |
41326c17f
|
2770 |
MEM_CGROUP_CHARGE_TYPE_ANON); |
83aae4c73
|
2771 |
} |
827a03d22
|
2772 2773 |
int mem_cgroup_cache_charge(struct page *page, struct mm_struct *mm, gfp_t gfp_mask) |
7a81b88cb
|
2774 |
{ |
827a03d22
|
2775 2776 2777 |
struct mem_cgroup *memcg = NULL; enum charge_type type = MEM_CGROUP_CHARGE_TYPE_CACHE; int ret; |
f8d665422
|
2778 |
if (mem_cgroup_disabled()) |
827a03d22
|
2779 2780 2781 |
return 0; if (PageCompound(page)) return 0; |
827a03d22
|
2782 2783 2784 |
if (!PageSwapCache(page)) ret = mem_cgroup_charge_common(page, mm, gfp_mask, type); else { /* page is swapcache/shmem */ |
0435a2fdc
|
2785 2786 |
ret = __mem_cgroup_try_charge_swapin(mm, page, gfp_mask, &memcg); |
827a03d22
|
2787 2788 2789 2790 |
if (!ret) __mem_cgroup_commit_charge_swapin(page, memcg, type); } return ret; |
7a81b88cb
|
2791 |
} |
c0ff4b854
|
2792 |
static void mem_cgroup_do_uncharge(struct mem_cgroup *memcg, |
7ec99d621
|
2793 2794 |
unsigned int nr_pages, const enum charge_type ctype) |
569b846df
|
2795 2796 2797 |
{ struct memcg_batch_info *batch = NULL; bool uncharge_memsw = true; |
7ec99d621
|
2798 |
|
569b846df
|
2799 2800 2801 |
/* If swapout, usage of swap doesn't decrease */ if (!do_swap_account || ctype == MEM_CGROUP_CHARGE_TYPE_SWAPOUT) uncharge_memsw = false; |
569b846df
|
2802 2803 2804 2805 2806 2807 2808 2809 |
batch = ¤t->memcg_batch; /* * In usual, we do css_get() when we remember memcg pointer. * But in this case, we keep res->usage until end of a series of * uncharges. Then, it's ok to ignore memcg's refcnt. */ if (!batch->memcg) |
c0ff4b854
|
2810 |
batch->memcg = memcg; |
569b846df
|
2811 |
/* |
3c11ecf44
|
2812 |
* do_batch > 0 when unmapping pages or inode invalidate/truncate. |
25985edce
|
2813 |
* In those cases, all pages freed continuously can be expected to be in |
3c11ecf44
|
2814 2815 2816 2817 2818 2819 2820 |
* the same cgroup and we have chance to coalesce uncharges. * But we do uncharge one by one if this is killed by OOM(TIF_MEMDIE) * because we want to do uncharge as soon as possible. */ if (!batch->do_batch || test_thread_flag(TIF_MEMDIE)) goto direct_uncharge; |
7ec99d621
|
2821 |
if (nr_pages > 1) |
ec1685109
|
2822 |
goto direct_uncharge; |
3c11ecf44
|
2823 |
/* |
569b846df
|
2824 2825 2826 2827 |
* In typical case, batch->memcg == mem. This means we can * merge a series of uncharges to an uncharge of res_counter. * If not, we uncharge res_counter ony by one. */ |
c0ff4b854
|
2828 |
if (batch->memcg != memcg) |
569b846df
|
2829 2830 |
goto direct_uncharge; /* remember freed charge and uncharge it later */ |
7ffd4ca7a
|
2831 |
batch->nr_pages++; |
569b846df
|
2832 |
if (uncharge_memsw) |
7ffd4ca7a
|
2833 |
batch->memsw_nr_pages++; |
569b846df
|
2834 2835 |
return; direct_uncharge: |
c0ff4b854
|
2836 |
res_counter_uncharge(&memcg->res, nr_pages * PAGE_SIZE); |
569b846df
|
2837 |
if (uncharge_memsw) |
c0ff4b854
|
2838 2839 2840 |
res_counter_uncharge(&memcg->memsw, nr_pages * PAGE_SIZE); if (unlikely(batch->memcg != memcg)) memcg_oom_recover(memcg); |
569b846df
|
2841 |
} |
7a81b88cb
|
2842 |
|
8697d3319
|
2843 |
/* |
69029cd55
|
2844 |
* uncharge if !page_mapped(page) |
8a9f3ccd2
|
2845 |
*/ |
8c7c6e34a
|
2846 |
static struct mem_cgroup * |
0030f535a
|
2847 2848 |
__mem_cgroup_uncharge_common(struct page *page, enum charge_type ctype, bool end_migration) |
8a9f3ccd2
|
2849 |
{ |
c0ff4b854
|
2850 |
struct mem_cgroup *memcg = NULL; |
7ec99d621
|
2851 2852 |
unsigned int nr_pages = 1; struct page_cgroup *pc; |
b24028572
|
2853 |
bool anon; |
8a9f3ccd2
|
2854 |
|
f8d665422
|
2855 |
if (mem_cgroup_disabled()) |
8c7c6e34a
|
2856 |
return NULL; |
4077960e2
|
2857 |
|
0c59b89c8
|
2858 |
VM_BUG_ON(PageSwapCache(page)); |
d13d14430
|
2859 |
|
37c2ac787
|
2860 |
if (PageTransHuge(page)) { |
7ec99d621
|
2861 |
nr_pages <<= compound_order(page); |
37c2ac787
|
2862 2863 |
VM_BUG_ON(!PageTransHuge(page)); } |
8697d3319
|
2864 |
/* |
3c541e14b
|
2865 |
* Check if our page_cgroup is valid |
8697d3319
|
2866 |
*/ |
52d4b9ac0
|
2867 |
pc = lookup_page_cgroup(page); |
cfa449461
|
2868 |
if (unlikely(!PageCgroupUsed(pc))) |
8c7c6e34a
|
2869 |
return NULL; |
b9c565d5a
|
2870 |
|
52d4b9ac0
|
2871 |
lock_page_cgroup(pc); |
d13d14430
|
2872 |
|
c0ff4b854
|
2873 |
memcg = pc->mem_cgroup; |
8c7c6e34a
|
2874 |
|
d13d14430
|
2875 2876 |
if (!PageCgroupUsed(pc)) goto unlock_out; |
b24028572
|
2877 |
anon = PageAnon(page); |
d13d14430
|
2878 |
switch (ctype) { |
41326c17f
|
2879 |
case MEM_CGROUP_CHARGE_TYPE_ANON: |
2ff76f119
|
2880 2881 2882 2883 2884 |
/* * Generally PageAnon tells if it's the anon statistics to be * updated; but sometimes e.g. mem_cgroup_uncharge_page() is * used before page reached the stage of being marked PageAnon. */ |
b24028572
|
2885 2886 |
anon = true; /* fallthrough */ |
8a9478ca7
|
2887 |
case MEM_CGROUP_CHARGE_TYPE_DROP: |
ac39cf8cb
|
2888 |
/* See mem_cgroup_prepare_migration() */ |
0030f535a
|
2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 |
if (page_mapped(page)) goto unlock_out; /* * Pages under migration may not be uncharged. But * end_migration() /must/ be the one uncharging the * unused post-migration page and so it has to call * here with the migration bit still set. See the * res_counter handling below. */ if (!end_migration && PageCgroupMigration(pc)) |
d13d14430
|
2899 2900 2901 2902 2903 2904 2905 2906 2907 2908 2909 |
goto unlock_out; break; case MEM_CGROUP_CHARGE_TYPE_SWAPOUT: if (!PageAnon(page)) { /* Shared memory */ if (page->mapping && !page_is_file_cache(page)) goto unlock_out; } else if (page_mapped(page)) /* Anon */ goto unlock_out; break; default: break; |
52d4b9ac0
|
2910 |
} |
d13d14430
|
2911 |
|
b24028572
|
2912 |
mem_cgroup_charge_statistics(memcg, anon, -nr_pages); |
04046e1a0
|
2913 |
|
52d4b9ac0
|
2914 |
ClearPageCgroupUsed(pc); |
544122e5e
|
2915 2916 2917 2918 2919 2920 |
/* * pc->mem_cgroup is not cleared here. It will be accessed when it's * freed from LRU. This is safe because uncharged page is expected not * to be reused (freed soon). Exception is SwapCache, it's handled by * special functions. */ |
b9c565d5a
|
2921 |
|
52d4b9ac0
|
2922 |
unlock_page_cgroup(pc); |
f75ca9620
|
2923 |
/* |
c0ff4b854
|
2924 |
* even after unlock, we have memcg->res.usage here and this memcg |
f75ca9620
|
2925 2926 |
* will never be freed. */ |
c0ff4b854
|
2927 |
memcg_check_events(memcg, page); |
f75ca9620
|
2928 |
if (do_swap_account && ctype == MEM_CGROUP_CHARGE_TYPE_SWAPOUT) { |
c0ff4b854
|
2929 2930 |
mem_cgroup_swap_statistics(memcg, true); mem_cgroup_get(memcg); |
f75ca9620
|
2931 |
} |
0030f535a
|
2932 2933 2934 2935 2936 2937 |
/* * Migration does not charge the res_counter for the * replacement page, so leave it alone when phasing out the * page that is unused after the migration. */ if (!end_migration && !mem_cgroup_is_root(memcg)) |
c0ff4b854
|
2938 |
mem_cgroup_do_uncharge(memcg, nr_pages, ctype); |
6d12e2d8d
|
2939 |
|
c0ff4b854
|
2940 |
return memcg; |
d13d14430
|
2941 2942 2943 |
unlock_out: unlock_page_cgroup(pc); |
8c7c6e34a
|
2944 |
return NULL; |
3c541e14b
|
2945 |
} |
69029cd55
|
2946 2947 |
void mem_cgroup_uncharge_page(struct page *page) { |
52d4b9ac0
|
2948 2949 2950 |
/* early check. */ if (page_mapped(page)) return; |
40f23a21a
|
2951 |
VM_BUG_ON(page->mapping && !PageAnon(page)); |
0c59b89c8
|
2952 2953 |
if (PageSwapCache(page)) return; |
0030f535a
|
2954 |
__mem_cgroup_uncharge_common(page, MEM_CGROUP_CHARGE_TYPE_ANON, false); |
69029cd55
|
2955 2956 2957 2958 2959 |
} void mem_cgroup_uncharge_cache_page(struct page *page) { VM_BUG_ON(page_mapped(page)); |
b7abea963
|
2960 |
VM_BUG_ON(page->mapping); |
0030f535a
|
2961 |
__mem_cgroup_uncharge_common(page, MEM_CGROUP_CHARGE_TYPE_CACHE, false); |
69029cd55
|
2962 |
} |
569b846df
|
2963 2964 2965 2966 2967 2968 2969 2970 2971 2972 2973 2974 2975 2976 |
/* * Batch_start/batch_end is called in unmap_page_range/invlidate/trucate. * In that cases, pages are freed continuously and we can expect pages * are in the same memcg. All these calls itself limits the number of * pages freed at once, then uncharge_start/end() is called properly. * This may be called prural(2) times in a context, */ void mem_cgroup_uncharge_start(void) { current->memcg_batch.do_batch++; /* We can do nest. */ if (current->memcg_batch.do_batch == 1) { current->memcg_batch.memcg = NULL; |
7ffd4ca7a
|
2977 2978 |
current->memcg_batch.nr_pages = 0; current->memcg_batch.memsw_nr_pages = 0; |
569b846df
|
2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991 2992 2993 2994 2995 2996 2997 2998 |
} } void mem_cgroup_uncharge_end(void) { struct memcg_batch_info *batch = ¤t->memcg_batch; if (!batch->do_batch) return; batch->do_batch--; if (batch->do_batch) /* If stacked, do nothing. */ return; if (!batch->memcg) return; /* * This "batch->memcg" is valid without any css_get/put etc... * bacause we hide charges behind us. */ |
7ffd4ca7a
|
2999 3000 3001 3002 3003 3004 |
if (batch->nr_pages) res_counter_uncharge(&batch->memcg->res, batch->nr_pages * PAGE_SIZE); if (batch->memsw_nr_pages) res_counter_uncharge(&batch->memcg->memsw, batch->memsw_nr_pages * PAGE_SIZE); |
3c11ecf44
|
3005 |
memcg_oom_recover(batch->memcg); |
569b846df
|
3006 3007 3008 |
/* forget this pointer (for sanity check) */ batch->memcg = NULL; } |
e767e0561
|
3009 |
#ifdef CONFIG_SWAP |
8c7c6e34a
|
3010 |
/* |
e767e0561
|
3011 |
* called after __delete_from_swap_cache() and drop "page" account. |
8c7c6e34a
|
3012 3013 |
* memcg information is recorded to swap_cgroup of "ent" */ |
8a9478ca7
|
3014 3015 |
void mem_cgroup_uncharge_swapcache(struct page *page, swp_entry_t ent, bool swapout) |
8c7c6e34a
|
3016 3017 |
{ struct mem_cgroup *memcg; |
8a9478ca7
|
3018 3019 3020 3021 |
int ctype = MEM_CGROUP_CHARGE_TYPE_SWAPOUT; if (!swapout) /* this was a swap cache but the swap is unused ! */ ctype = MEM_CGROUP_CHARGE_TYPE_DROP; |
0030f535a
|
3022 |
memcg = __mem_cgroup_uncharge_common(page, ctype, false); |
8c7c6e34a
|
3023 |
|
f75ca9620
|
3024 3025 3026 3027 3028 |
/* * record memcg information, if swapout && memcg != NULL, * mem_cgroup_get() was called in uncharge(). */ if (do_swap_account && swapout && memcg) |
a3b2d6926
|
3029 |
swap_cgroup_record(ent, css_id(&memcg->css)); |
8c7c6e34a
|
3030 |
} |
e767e0561
|
3031 |
#endif |
8c7c6e34a
|
3032 |
|
c255a4580
|
3033 |
#ifdef CONFIG_MEMCG_SWAP |
8c7c6e34a
|
3034 3035 3036 3037 3038 |
/* * called from swap_entry_free(). remove record in swap_cgroup and * uncharge "memsw" account. */ void mem_cgroup_uncharge_swap(swp_entry_t ent) |
d13d14430
|
3039 |
{ |
8c7c6e34a
|
3040 |
struct mem_cgroup *memcg; |
a3b2d6926
|
3041 |
unsigned short id; |
8c7c6e34a
|
3042 3043 3044 |
if (!do_swap_account) return; |
a3b2d6926
|
3045 3046 3047 |
id = swap_cgroup_record(ent, 0); rcu_read_lock(); memcg = mem_cgroup_lookup(id); |
8c7c6e34a
|
3048 |
if (memcg) { |
a3b2d6926
|
3049 3050 3051 3052 |
/* * We uncharge this because swap is freed. * This memcg can be obsolete one. We avoid calling css_tryget */ |
0c3e73e84
|
3053 |
if (!mem_cgroup_is_root(memcg)) |
4e649152c
|
3054 |
res_counter_uncharge(&memcg->memsw, PAGE_SIZE); |
0c3e73e84
|
3055 |
mem_cgroup_swap_statistics(memcg, false); |
8c7c6e34a
|
3056 3057 |
mem_cgroup_put(memcg); } |
a3b2d6926
|
3058 |
rcu_read_unlock(); |
d13d14430
|
3059 |
} |
024914477
|
3060 3061 3062 3063 3064 3065 3066 3067 3068 3069 3070 3071 3072 3073 3074 3075 |
/** * mem_cgroup_move_swap_account - move swap charge and swap_cgroup's record. * @entry: swap entry to be moved * @from: mem_cgroup which the entry is moved from * @to: mem_cgroup which the entry is moved to * * It succeeds only when the swap_cgroup's record for this entry is the same * as the mem_cgroup's id of @from. * * Returns 0 on success, -EINVAL on failure. * * The caller must have charged to @to, IOW, called res_counter_charge() about * both res and memsw, and called css_get(). */ static int mem_cgroup_move_swap_account(swp_entry_t entry, |
e91cbb425
|
3076 |
struct mem_cgroup *from, struct mem_cgroup *to) |
024914477
|
3077 3078 3079 3080 3081 3082 3083 |
{ unsigned short old_id, new_id; old_id = css_id(&from->css); new_id = css_id(&to->css); if (swap_cgroup_cmpxchg(entry, old_id, new_id) == old_id) { |
024914477
|
3084 |
mem_cgroup_swap_statistics(from, false); |
483c30b51
|
3085 |
mem_cgroup_swap_statistics(to, true); |
024914477
|
3086 |
/* |
483c30b51
|
3087 3088 3089 3090 3091 3092 |
* This function is only called from task migration context now. * It postpones res_counter and refcount handling till the end * of task migration(mem_cgroup_clear_mc()) for performance * improvement. But we cannot postpone mem_cgroup_get(to) * because if the process that has been moved to @to does * swap-in, the refcount of @to might be decreased to 0. |
024914477
|
3093 |
*/ |
024914477
|
3094 |
mem_cgroup_get(to); |
024914477
|
3095 3096 3097 3098 3099 3100 |
return 0; } return -EINVAL; } #else static inline int mem_cgroup_move_swap_account(swp_entry_t entry, |
e91cbb425
|
3101 |
struct mem_cgroup *from, struct mem_cgroup *to) |
024914477
|
3102 3103 3104 |
{ return -EINVAL; } |
8c7c6e34a
|
3105 |
#endif |
d13d14430
|
3106 |
|
ae41be374
|
3107 |
/* |
01b1ae63c
|
3108 3109 |
* Before starting migration, account PAGE_SIZE to mem_cgroup that the old * page belongs to. |
ae41be374
|
3110 |
*/ |
0030f535a
|
3111 3112 |
void mem_cgroup_prepare_migration(struct page *page, struct page *newpage, struct mem_cgroup **memcgp) |
ae41be374
|
3113 |
{ |
c0ff4b854
|
3114 |
struct mem_cgroup *memcg = NULL; |
7ec99d621
|
3115 |
struct page_cgroup *pc; |
ac39cf8cb
|
3116 |
enum charge_type ctype; |
8869b8f6e
|
3117 |
|
72835c86c
|
3118 |
*memcgp = NULL; |
56039efa1
|
3119 |
|
ec1685109
|
3120 |
VM_BUG_ON(PageTransHuge(page)); |
f8d665422
|
3121 |
if (mem_cgroup_disabled()) |
0030f535a
|
3122 |
return; |
4077960e2
|
3123 |
|
52d4b9ac0
|
3124 3125 3126 |
pc = lookup_page_cgroup(page); lock_page_cgroup(pc); if (PageCgroupUsed(pc)) { |
c0ff4b854
|
3127 3128 |
memcg = pc->mem_cgroup; css_get(&memcg->css); |
ac39cf8cb
|
3129 3130 3131 3132 3133 3134 3135 3136 3137 3138 3139 3140 3141 3142 3143 3144 3145 3146 3147 3148 3149 3150 3151 3152 3153 3154 3155 3156 3157 3158 3159 |
/* * At migrating an anonymous page, its mapcount goes down * to 0 and uncharge() will be called. But, even if it's fully * unmapped, migration may fail and this page has to be * charged again. We set MIGRATION flag here and delay uncharge * until end_migration() is called * * Corner Case Thinking * A) * When the old page was mapped as Anon and it's unmap-and-freed * while migration was ongoing. * If unmap finds the old page, uncharge() of it will be delayed * until end_migration(). If unmap finds a new page, it's * uncharged when it make mapcount to be 1->0. If unmap code * finds swap_migration_entry, the new page will not be mapped * and end_migration() will find it(mapcount==0). * * B) * When the old page was mapped but migraion fails, the kernel * remaps it. A charge for it is kept by MIGRATION flag even * if mapcount goes down to 0. We can do remap successfully * without charging it again. * * C) * The "old" page is under lock_page() until the end of * migration, so, the old page itself will not be swapped-out. * If the new page is swapped out before end_migraton, our * hook to usual swap-out path will catch the event. */ if (PageAnon(page)) SetPageCgroupMigration(pc); |
e8589cc18
|
3160 |
} |
52d4b9ac0
|
3161 |
unlock_page_cgroup(pc); |
ac39cf8cb
|
3162 3163 3164 3165 |
/* * If the page is not charged at this point, * we return here. */ |
c0ff4b854
|
3166 |
if (!memcg) |
0030f535a
|
3167 |
return; |
01b1ae63c
|
3168 |
|
72835c86c
|
3169 |
*memcgp = memcg; |
ac39cf8cb
|
3170 3171 3172 3173 3174 3175 |
/* * We charge new page before it's used/mapped. So, even if unlock_page() * is called before end_migration, we can catch all events on this new * page. In the case new page is migrated but not remapped, new page's * mapcount will be finally 0 and we call uncharge in end_migration(). */ |
ac39cf8cb
|
3176 |
if (PageAnon(page)) |
41326c17f
|
3177 |
ctype = MEM_CGROUP_CHARGE_TYPE_ANON; |
ac39cf8cb
|
3178 |
else |
62ba7442c
|
3179 |
ctype = MEM_CGROUP_CHARGE_TYPE_CACHE; |
0030f535a
|
3180 3181 3182 3183 3184 |
/* * The page is committed to the memcg, but it's not actually * charged to the res_counter since we plan on replacing the * old one and only one page is going to be left afterwards. */ |
ce587e65e
|
3185 |
__mem_cgroup_commit_charge(memcg, newpage, 1, ctype, false); |
ae41be374
|
3186 |
} |
8869b8f6e
|
3187 |
|
69029cd55
|
3188 |
/* remove redundant charge if migration failed*/ |
c0ff4b854
|
3189 |
void mem_cgroup_end_migration(struct mem_cgroup *memcg, |
50de1dd96
|
3190 |
struct page *oldpage, struct page *newpage, bool migration_ok) |
ae41be374
|
3191 |
{ |
ac39cf8cb
|
3192 |
struct page *used, *unused; |
01b1ae63c
|
3193 |
struct page_cgroup *pc; |
b24028572
|
3194 |
bool anon; |
01b1ae63c
|
3195 |
|
c0ff4b854
|
3196 |
if (!memcg) |
01b1ae63c
|
3197 |
return; |
ac39cf8cb
|
3198 |
/* blocks rmdir() */ |
c0ff4b854
|
3199 |
cgroup_exclude_rmdir(&memcg->css); |
50de1dd96
|
3200 |
if (!migration_ok) { |
ac39cf8cb
|
3201 3202 |
used = oldpage; unused = newpage; |
01b1ae63c
|
3203 |
} else { |
ac39cf8cb
|
3204 |
used = newpage; |
01b1ae63c
|
3205 3206 |
unused = oldpage; } |
0030f535a
|
3207 |
anon = PageAnon(used); |
7d188958b
|
3208 3209 3210 3211 |
__mem_cgroup_uncharge_common(unused, anon ? MEM_CGROUP_CHARGE_TYPE_ANON : MEM_CGROUP_CHARGE_TYPE_CACHE, true); |
0030f535a
|
3212 |
css_put(&memcg->css); |
69029cd55
|
3213 |
/* |
ac39cf8cb
|
3214 3215 3216 |
* We disallowed uncharge of pages under migration because mapcount * of the page goes down to zero, temporarly. * Clear the flag and check the page should be charged. |
01b1ae63c
|
3217 |
*/ |
ac39cf8cb
|
3218 3219 3220 3221 |
pc = lookup_page_cgroup(oldpage); lock_page_cgroup(pc); ClearPageCgroupMigration(pc); unlock_page_cgroup(pc); |
ac39cf8cb
|
3222 |
|
01b1ae63c
|
3223 |
/* |
ac39cf8cb
|
3224 3225 3226 3227 3228 3229 |
* If a page is a file cache, radix-tree replacement is very atomic * and we can skip this check. When it was an Anon page, its mapcount * goes down to 0. But because we added MIGRATION flage, it's not * uncharged yet. There are several case but page->mapcount check * and USED bit check in mem_cgroup_uncharge_page() will do enough * check. (see prepare_charge() also) |
69029cd55
|
3230 |
*/ |
b24028572
|
3231 |
if (anon) |
ac39cf8cb
|
3232 |
mem_cgroup_uncharge_page(used); |
887032670
|
3233 |
/* |
ac39cf8cb
|
3234 3235 |
* At migration, we may charge account against cgroup which has no * tasks. |
887032670
|
3236 3237 3238 |
* So, rmdir()->pre_destroy() can be called while we do this charge. * In that case, we need to call pre_destroy() again. check it here. */ |
c0ff4b854
|
3239 |
cgroup_release_and_wakeup_rmdir(&memcg->css); |
ae41be374
|
3240 |
} |
78fb74669
|
3241 |
|
ab936cbcd
|
3242 3243 3244 3245 3246 3247 3248 3249 |
/* * At replace page cache, newpage is not under any memcg but it's on * LRU. So, this function doesn't touch res_counter but handles LRU * in correct way. Both pages are locked so we cannot race with uncharge. */ void mem_cgroup_replace_page_cache(struct page *oldpage, struct page *newpage) { |
bde05d1cc
|
3250 |
struct mem_cgroup *memcg = NULL; |
ab936cbcd
|
3251 |
struct page_cgroup *pc; |
ab936cbcd
|
3252 |
enum charge_type type = MEM_CGROUP_CHARGE_TYPE_CACHE; |
ab936cbcd
|
3253 3254 3255 3256 3257 3258 3259 |
if (mem_cgroup_disabled()) return; pc = lookup_page_cgroup(oldpage); /* fix accounting on old pages */ lock_page_cgroup(pc); |
bde05d1cc
|
3260 3261 3262 3263 3264 |
if (PageCgroupUsed(pc)) { memcg = pc->mem_cgroup; mem_cgroup_charge_statistics(memcg, false, -1); ClearPageCgroupUsed(pc); } |
ab936cbcd
|
3265 |
unlock_page_cgroup(pc); |
bde05d1cc
|
3266 3267 3268 3269 3270 3271 |
/* * When called from shmem_replace_page(), in some cases the * oldpage has already been charged, and in some cases not. */ if (!memcg) return; |
ab936cbcd
|
3272 3273 3274 3275 3276 |
/* * Even if newpage->mapping was NULL before starting replacement, * the newpage may be on LRU(or pagevec for LRU) already. We lock * LRU while we overwrite pc->mem_cgroup. */ |
ce587e65e
|
3277 |
__mem_cgroup_commit_charge(memcg, newpage, 1, type, true); |
ab936cbcd
|
3278 |
} |
f212ad7cf
|
3279 3280 3281 3282 3283 3284 |
#ifdef CONFIG_DEBUG_VM static struct page_cgroup *lookup_page_cgroup_used(struct page *page) { struct page_cgroup *pc; pc = lookup_page_cgroup(page); |
cfa449461
|
3285 3286 3287 3288 3289 |
/* * Can be NULL while feeding pages into the page allocator for * the first time, i.e. during boot or memory hotplug; * or when mem_cgroup_disabled(). */ |
f212ad7cf
|
3290 3291 3292 3293 3294 3295 3296 3297 3298 3299 3300 3301 3302 3303 3304 3305 3306 3307 3308 |
if (likely(pc) && PageCgroupUsed(pc)) return pc; return NULL; } bool mem_cgroup_bad_page_check(struct page *page) { if (mem_cgroup_disabled()) return false; return lookup_page_cgroup_used(page) != NULL; } void mem_cgroup_print_bad_page(struct page *page) { struct page_cgroup *pc; pc = lookup_page_cgroup_used(page); if (pc) { |
90b3feaec
|
3309 3310 |
printk(KERN_ALERT "pc:%p pc->flags:%lx pc->mem_cgroup:%p ", |
f212ad7cf
|
3311 |
pc, pc->flags, pc->mem_cgroup); |
f212ad7cf
|
3312 3313 3314 |
} } #endif |
8c7c6e34a
|
3315 |
static DEFINE_MUTEX(set_limit_mutex); |
d38d2a758
|
3316 |
static int mem_cgroup_resize_limit(struct mem_cgroup *memcg, |
8c7c6e34a
|
3317 |
unsigned long long val) |
628f42355
|
3318 |
{ |
81d39c20f
|
3319 |
int retry_count; |
3c11ecf44
|
3320 |
u64 memswlimit, memlimit; |
628f42355
|
3321 |
int ret = 0; |
81d39c20f
|
3322 3323 |
int children = mem_cgroup_count_children(memcg); u64 curusage, oldusage; |
3c11ecf44
|
3324 |
int enlarge; |
81d39c20f
|
3325 3326 3327 3328 3329 3330 3331 3332 3333 |
/* * For keeping hierarchical_reclaim simple, how long we should retry * is depends on callers. We set our retry-count to be function * of # of children which we should visit in this loop. */ retry_count = MEM_CGROUP_RECLAIM_RETRIES * children; oldusage = res_counter_read_u64(&memcg->res, RES_USAGE); |
628f42355
|
3334 |
|
3c11ecf44
|
3335 |
enlarge = 0; |
8c7c6e34a
|
3336 |
while (retry_count) { |
628f42355
|
3337 3338 3339 3340 |
if (signal_pending(current)) { ret = -EINTR; break; } |
8c7c6e34a
|
3341 3342 3343 |
/* * Rather than hide all in some function, I do this in * open coded manner. You see what this really does. |
aaad153e3
|
3344 |
* We have to guarantee memcg->res.limit <= memcg->memsw.limit. |
8c7c6e34a
|
3345 3346 3347 3348 3349 3350 |
*/ mutex_lock(&set_limit_mutex); memswlimit = res_counter_read_u64(&memcg->memsw, RES_LIMIT); if (memswlimit < val) { ret = -EINVAL; mutex_unlock(&set_limit_mutex); |
628f42355
|
3351 3352 |
break; } |
3c11ecf44
|
3353 3354 3355 3356 |
memlimit = res_counter_read_u64(&memcg->res, RES_LIMIT); if (memlimit < val) enlarge = 1; |
8c7c6e34a
|
3357 |
ret = res_counter_set_limit(&memcg->res, val); |
22a668d7c
|
3358 3359 3360 3361 3362 3363 |
if (!ret) { if (memswlimit == val) memcg->memsw_is_minimum = true; else memcg->memsw_is_minimum = false; } |
8c7c6e34a
|
3364 3365 3366 3367 |
mutex_unlock(&set_limit_mutex); if (!ret) break; |
5660048cc
|
3368 3369 |
mem_cgroup_reclaim(memcg, GFP_KERNEL, MEM_CGROUP_RECLAIM_SHRINK); |
81d39c20f
|
3370 3371 3372 3373 3374 3375 |
curusage = res_counter_read_u64(&memcg->res, RES_USAGE); /* Usage is reduced ? */ if (curusage >= oldusage) retry_count--; else oldusage = curusage; |
8c7c6e34a
|
3376 |
} |
3c11ecf44
|
3377 3378 |
if (!ret && enlarge) memcg_oom_recover(memcg); |
14797e236
|
3379 |
|
8c7c6e34a
|
3380 3381 |
return ret; } |
338c84310
|
3382 3383 |
static int mem_cgroup_resize_memsw_limit(struct mem_cgroup *memcg, unsigned long long val) |
8c7c6e34a
|
3384 |
{ |
81d39c20f
|
3385 |
int retry_count; |
3c11ecf44
|
3386 |
u64 memlimit, memswlimit, oldusage, curusage; |
81d39c20f
|
3387 3388 |
int children = mem_cgroup_count_children(memcg); int ret = -EBUSY; |
3c11ecf44
|
3389 |
int enlarge = 0; |
8c7c6e34a
|
3390 |
|
81d39c20f
|
3391 3392 3393 |
/* see mem_cgroup_resize_res_limit */ retry_count = children * MEM_CGROUP_RECLAIM_RETRIES; oldusage = res_counter_read_u64(&memcg->memsw, RES_USAGE); |
8c7c6e34a
|
3394 3395 3396 3397 3398 3399 3400 3401 |
while (retry_count) { if (signal_pending(current)) { ret = -EINTR; break; } /* * Rather than hide all in some function, I do this in * open coded manner. You see what this really does. |
aaad153e3
|
3402 |
* We have to guarantee memcg->res.limit <= memcg->memsw.limit. |
8c7c6e34a
|
3403 3404 3405 3406 3407 3408 3409 3410 |
*/ mutex_lock(&set_limit_mutex); memlimit = res_counter_read_u64(&memcg->res, RES_LIMIT); if (memlimit > val) { ret = -EINVAL; mutex_unlock(&set_limit_mutex); break; } |
3c11ecf44
|
3411 3412 3413 |
memswlimit = res_counter_read_u64(&memcg->memsw, RES_LIMIT); if (memswlimit < val) enlarge = 1; |
8c7c6e34a
|
3414 |
ret = res_counter_set_limit(&memcg->memsw, val); |
22a668d7c
|
3415 3416 3417 3418 3419 3420 |
if (!ret) { if (memlimit == val) memcg->memsw_is_minimum = true; else memcg->memsw_is_minimum = false; } |
8c7c6e34a
|
3421 3422 3423 3424 |
mutex_unlock(&set_limit_mutex); if (!ret) break; |
5660048cc
|
3425 3426 3427 |
mem_cgroup_reclaim(memcg, GFP_KERNEL, MEM_CGROUP_RECLAIM_NOSWAP | MEM_CGROUP_RECLAIM_SHRINK); |
8c7c6e34a
|
3428 |
curusage = res_counter_read_u64(&memcg->memsw, RES_USAGE); |
81d39c20f
|
3429 |
/* Usage is reduced ? */ |
8c7c6e34a
|
3430 |
if (curusage >= oldusage) |
628f42355
|
3431 |
retry_count--; |
81d39c20f
|
3432 3433 |
else oldusage = curusage; |
628f42355
|
3434 |
} |
3c11ecf44
|
3435 3436 |
if (!ret && enlarge) memcg_oom_recover(memcg); |
628f42355
|
3437 3438 |
return ret; } |
4e4169535
|
3439 |
unsigned long mem_cgroup_soft_limit_reclaim(struct zone *zone, int order, |
0ae5e89c6
|
3440 3441 |
gfp_t gfp_mask, unsigned long *total_scanned) |
4e4169535
|
3442 3443 3444 3445 3446 3447 |
{ unsigned long nr_reclaimed = 0; struct mem_cgroup_per_zone *mz, *next_mz = NULL; unsigned long reclaimed; int loop = 0; struct mem_cgroup_tree_per_zone *mctz; |
ef8745c1e
|
3448 |
unsigned long long excess; |
0ae5e89c6
|
3449 |
unsigned long nr_scanned; |
4e4169535
|
3450 3451 3452 |
if (order > 0) return 0; |
00918b6ab
|
3453 |
mctz = soft_limit_tree_node_zone(zone_to_nid(zone), zone_idx(zone)); |
4e4169535
|
3454 3455 3456 3457 3458 3459 3460 3461 3462 3463 3464 3465 |
/* * This loop can run a while, specially if mem_cgroup's continuously * keep exceeding their soft limit and putting the system under * pressure */ do { if (next_mz) mz = next_mz; else mz = mem_cgroup_largest_soft_limit_node(mctz); if (!mz) break; |
0ae5e89c6
|
3466 |
nr_scanned = 0; |
d79154bb5
|
3467 |
reclaimed = mem_cgroup_soft_reclaim(mz->memcg, zone, |
5660048cc
|
3468 |
gfp_mask, &nr_scanned); |
4e4169535
|
3469 |
nr_reclaimed += reclaimed; |
0ae5e89c6
|
3470 |
*total_scanned += nr_scanned; |
4e4169535
|
3471 3472 3473 3474 3475 3476 3477 3478 3479 3480 3481 3482 3483 3484 3485 3486 3487 3488 3489 3490 3491 3492 |
spin_lock(&mctz->lock); /* * If we failed to reclaim anything from this memory cgroup * it is time to move on to the next cgroup */ next_mz = NULL; if (!reclaimed) { do { /* * Loop until we find yet another one. * * By the time we get the soft_limit lock * again, someone might have aded the * group back on the RB tree. Iterate to * make sure we get a different mem. * mem_cgroup_largest_soft_limit_node returns * NULL if no other cgroup is present on * the tree */ next_mz = __mem_cgroup_largest_soft_limit_node(mctz); |
39cc98f1f
|
3493 |
if (next_mz == mz) |
d79154bb5
|
3494 |
css_put(&next_mz->memcg->css); |
39cc98f1f
|
3495 |
else /* next_mz == NULL or other memcg */ |
4e4169535
|
3496 3497 3498 |
break; } while (1); } |
d79154bb5
|
3499 3500 |
__mem_cgroup_remove_exceeded(mz->memcg, mz, mctz); excess = res_counter_soft_limit_excess(&mz->memcg->res); |
4e4169535
|
3501 3502 3503 3504 3505 3506 3507 3508 |
/* * One school of thought says that we should not add * back the node to the tree if reclaim returns 0. * But our reclaim could return 0, simply because due * to priority we are exposing a smaller subset of * memory to reclaim from. Consider this as a longer * term TODO. */ |
ef8745c1e
|
3509 |
/* If excess == 0, no tree ops */ |
d79154bb5
|
3510 |
__mem_cgroup_insert_exceeded(mz->memcg, mz, mctz, excess); |
4e4169535
|
3511 |
spin_unlock(&mctz->lock); |
d79154bb5
|
3512 |
css_put(&mz->memcg->css); |
4e4169535
|
3513 3514 3515 3516 3517 3518 3519 3520 3521 3522 3523 3524 |
loop++; /* * Could not reclaim anything and there are no more * mem cgroups to try or we seem to be looping without * reclaiming anything. */ if (!nr_reclaimed && (next_mz == NULL || loop > MEM_CGROUP_MAX_SOFT_LIMIT_RECLAIM_LOOPS)) break; } while (!nr_reclaimed); if (next_mz) |
d79154bb5
|
3525 |
css_put(&next_mz->memcg->css); |
4e4169535
|
3526 3527 |
return nr_reclaimed; } |
c9b0ed514
|
3528 |
/* |
3c935d189
|
3529 3530 3531 3532 |
* Traverse a specified page_cgroup list and try to drop them all. This doesn't * reclaim the pages page themselves - it just removes the page_cgroups. * Returns true if some page_cgroups were not freed, indicating that the caller * must retry this operation. |
cc8475822
|
3533 |
*/ |
3c935d189
|
3534 |
static bool mem_cgroup_force_empty_list(struct mem_cgroup *memcg, |
08e552c69
|
3535 |
int node, int zid, enum lru_list lru) |
cc8475822
|
3536 |
{ |
08e552c69
|
3537 |
struct mem_cgroup_per_zone *mz; |
08e552c69
|
3538 |
unsigned long flags, loop; |
072c56c13
|
3539 |
struct list_head *list; |
925b7673c
|
3540 3541 |
struct page *busy; struct zone *zone; |
072c56c13
|
3542 |
|
08e552c69
|
3543 |
zone = &NODE_DATA(node)->node_zones[zid]; |
c0ff4b854
|
3544 |
mz = mem_cgroup_zoneinfo(memcg, node, zid); |
6290df545
|
3545 |
list = &mz->lruvec.lists[lru]; |
cc8475822
|
3546 |
|
1eb492725
|
3547 |
loop = mz->lru_size[lru]; |
f817ed485
|
3548 3549 3550 3551 |
/* give some margin against EBUSY etc...*/ loop += 256; busy = NULL; while (loop--) { |
925b7673c
|
3552 |
struct page_cgroup *pc; |
5564e88ba
|
3553 |
struct page *page; |
08e552c69
|
3554 |
spin_lock_irqsave(&zone->lru_lock, flags); |
f817ed485
|
3555 |
if (list_empty(list)) { |
08e552c69
|
3556 |
spin_unlock_irqrestore(&zone->lru_lock, flags); |
52d4b9ac0
|
3557 |
break; |
f817ed485
|
3558 |
} |
925b7673c
|
3559 3560 3561 |
page = list_entry(list->prev, struct page, lru); if (busy == page) { list_move(&page->lru, list); |
648bcc771
|
3562 |
busy = NULL; |
08e552c69
|
3563 |
spin_unlock_irqrestore(&zone->lru_lock, flags); |
f817ed485
|
3564 3565 |
continue; } |
08e552c69
|
3566 |
spin_unlock_irqrestore(&zone->lru_lock, flags); |
f817ed485
|
3567 |
|
925b7673c
|
3568 |
pc = lookup_page_cgroup(page); |
5564e88ba
|
3569 |
|
3c935d189
|
3570 |
if (mem_cgroup_move_parent(page, pc, memcg)) { |
f817ed485
|
3571 |
/* found lock contention or "pc" is obsolete. */ |
925b7673c
|
3572 |
busy = page; |
f817ed485
|
3573 3574 3575 |
cond_resched(); } else busy = NULL; |
cc8475822
|
3576 |
} |
3c935d189
|
3577 |
return !list_empty(list); |
cc8475822
|
3578 3579 3580 3581 3582 3583 |
} /* * make mem_cgroup's charge to be 0 if there is no task. * This enables deleting this mem_cgroup. */ |
c0ff4b854
|
3584 |
static int mem_cgroup_force_empty(struct mem_cgroup *memcg, bool free_all) |
cc8475822
|
3585 |
{ |
f817ed485
|
3586 3587 3588 |
int ret; int node, zid, shrink; int nr_retries = MEM_CGROUP_RECLAIM_RETRIES; |
c0ff4b854
|
3589 |
struct cgroup *cgrp = memcg->css.cgroup; |
8869b8f6e
|
3590 |
|
c0ff4b854
|
3591 |
css_get(&memcg->css); |
f817ed485
|
3592 3593 |
shrink = 0; |
c1e862c1f
|
3594 3595 3596 |
/* should free all ? */ if (free_all) goto try_to_free; |
f817ed485
|
3597 |
move_account: |
fce664775
|
3598 |
do { |
f817ed485
|
3599 |
ret = -EBUSY; |
c1e862c1f
|
3600 3601 |
if (cgroup_task_count(cgrp) || !list_empty(&cgrp->children)) goto out; |
52d4b9ac0
|
3602 3603 |
/* This is for making all *used* pages to be on LRU. */ lru_add_drain_all(); |
c0ff4b854
|
3604 |
drain_all_stock_sync(memcg); |
f817ed485
|
3605 |
ret = 0; |
c0ff4b854
|
3606 |
mem_cgroup_start_move(memcg); |
299b4eaa3
|
3607 |
for_each_node_state(node, N_HIGH_MEMORY) { |
f817ed485
|
3608 |
for (zid = 0; !ret && zid < MAX_NR_ZONES; zid++) { |
f156ab933
|
3609 3610 |
enum lru_list lru; for_each_lru(lru) { |
c0ff4b854
|
3611 |
ret = mem_cgroup_force_empty_list(memcg, |
f156ab933
|
3612 |
node, zid, lru); |
f817ed485
|
3613 3614 3615 |
if (ret) break; } |
1ecaab2bd
|
3616 |
} |
f817ed485
|
3617 3618 3619 |
if (ret) break; } |
c0ff4b854
|
3620 3621 |
mem_cgroup_end_move(memcg); memcg_oom_recover(memcg); |
52d4b9ac0
|
3622 |
cond_resched(); |
fce664775
|
3623 |
/* "ret" should also be checked to ensure all lists are empty. */ |
569530fb1
|
3624 |
} while (res_counter_read_u64(&memcg->res, RES_USAGE) > 0 || ret); |
cc8475822
|
3625 |
out: |
c0ff4b854
|
3626 |
css_put(&memcg->css); |
cc8475822
|
3627 |
return ret; |
f817ed485
|
3628 3629 |
try_to_free: |
c1e862c1f
|
3630 3631 |
/* returns EBUSY if there is a task or if we come here twice. */ if (cgroup_task_count(cgrp) || !list_empty(&cgrp->children) || shrink) { |
f817ed485
|
3632 3633 3634 |
ret = -EBUSY; goto out; } |
c1e862c1f
|
3635 3636 |
/* we call try-to-free pages for make this cgroup empty */ lru_add_drain_all(); |
f817ed485
|
3637 3638 |
/* try to free all pages in this cgroup */ shrink = 1; |
569530fb1
|
3639 |
while (nr_retries && res_counter_read_u64(&memcg->res, RES_USAGE) > 0) { |
f817ed485
|
3640 |
int progress; |
c1e862c1f
|
3641 3642 3643 3644 3645 |
if (signal_pending(current)) { ret = -EINTR; goto out; } |
c0ff4b854
|
3646 |
progress = try_to_free_mem_cgroup_pages(memcg, GFP_KERNEL, |
185efc0f9
|
3647 |
false); |
c1e862c1f
|
3648 |
if (!progress) { |
f817ed485
|
3649 |
nr_retries--; |
c1e862c1f
|
3650 |
/* maybe some writeback is necessary */ |
8aa7e847d
|
3651 |
congestion_wait(BLK_RW_ASYNC, HZ/10); |
c1e862c1f
|
3652 |
} |
f817ed485
|
3653 3654 |
} |
08e552c69
|
3655 |
lru_add_drain(); |
f817ed485
|
3656 |
/* try move_account...there may be some *locked* pages. */ |
fce664775
|
3657 |
goto move_account; |
cc8475822
|
3658 |
} |
6bbda35ce
|
3659 |
static int mem_cgroup_force_empty_write(struct cgroup *cont, unsigned int event) |
c1e862c1f
|
3660 3661 3662 |
{ return mem_cgroup_force_empty(mem_cgroup_from_cont(cont), true); } |
18f59ea7d
|
3663 3664 3665 3666 3667 3668 3669 3670 3671 |
static u64 mem_cgroup_hierarchy_read(struct cgroup *cont, struct cftype *cft) { return mem_cgroup_from_cont(cont)->use_hierarchy; } static int mem_cgroup_hierarchy_write(struct cgroup *cont, struct cftype *cft, u64 val) { int retval = 0; |
c0ff4b854
|
3672 |
struct mem_cgroup *memcg = mem_cgroup_from_cont(cont); |
18f59ea7d
|
3673 |
struct cgroup *parent = cont->parent; |
c0ff4b854
|
3674 |
struct mem_cgroup *parent_memcg = NULL; |
18f59ea7d
|
3675 3676 |
if (parent) |
c0ff4b854
|
3677 |
parent_memcg = mem_cgroup_from_cont(parent); |
18f59ea7d
|
3678 3679 |
cgroup_lock(); |
567fb435b
|
3680 3681 3682 |
if (memcg->use_hierarchy == val) goto out; |
18f59ea7d
|
3683 |
/* |
af901ca18
|
3684 |
* If parent's use_hierarchy is set, we can't make any modifications |
18f59ea7d
|
3685 3686 3687 3688 3689 3690 |
* in the child subtrees. If it is unset, then the change can * occur, provided the current cgroup has no children. * * For the root cgroup, parent_mem is NULL, we allow value to be * set if there are no children. */ |
c0ff4b854
|
3691 |
if ((!parent_memcg || !parent_memcg->use_hierarchy) && |
18f59ea7d
|
3692 3693 |
(val == 1 || val == 0)) { if (list_empty(&cont->children)) |
c0ff4b854
|
3694 |
memcg->use_hierarchy = val; |
18f59ea7d
|
3695 3696 3697 3698 |
else retval = -EBUSY; } else retval = -EINVAL; |
567fb435b
|
3699 3700 |
out: |
18f59ea7d
|
3701 3702 3703 3704 |
cgroup_unlock(); return retval; } |
0c3e73e84
|
3705 |
|
c0ff4b854
|
3706 |
static unsigned long mem_cgroup_recursive_stat(struct mem_cgroup *memcg, |
7a159cc9d
|
3707 |
enum mem_cgroup_stat_index idx) |
0c3e73e84
|
3708 |
{ |
7d74b06f2
|
3709 |
struct mem_cgroup *iter; |
7a159cc9d
|
3710 |
long val = 0; |
0c3e73e84
|
3711 |
|
7a159cc9d
|
3712 |
/* Per-cpu values can be negative, use a signed accumulator */ |
c0ff4b854
|
3713 |
for_each_mem_cgroup_tree(iter, memcg) |
7d74b06f2
|
3714 3715 3716 3717 3718 |
val += mem_cgroup_read_stat(iter, idx); if (val < 0) /* race ? */ val = 0; return val; |
0c3e73e84
|
3719 |
} |
c0ff4b854
|
3720 |
static inline u64 mem_cgroup_usage(struct mem_cgroup *memcg, bool swap) |
104f39284
|
3721 |
{ |
7d74b06f2
|
3722 |
u64 val; |
104f39284
|
3723 |
|
c0ff4b854
|
3724 |
if (!mem_cgroup_is_root(memcg)) { |
104f39284
|
3725 |
if (!swap) |
65c64ce8e
|
3726 |
return res_counter_read_u64(&memcg->res, RES_USAGE); |
104f39284
|
3727 |
else |
65c64ce8e
|
3728 |
return res_counter_read_u64(&memcg->memsw, RES_USAGE); |
104f39284
|
3729 |
} |
c0ff4b854
|
3730 3731 |
val = mem_cgroup_recursive_stat(memcg, MEM_CGROUP_STAT_CACHE); val += mem_cgroup_recursive_stat(memcg, MEM_CGROUP_STAT_RSS); |
104f39284
|
3732 |
|
7d74b06f2
|
3733 |
if (swap) |
bff6bb83f
|
3734 |
val += mem_cgroup_recursive_stat(memcg, MEM_CGROUP_STAT_SWAP); |
104f39284
|
3735 3736 3737 |
return val << PAGE_SHIFT; } |
af36f906c
|
3738 3739 3740 |
static ssize_t mem_cgroup_read(struct cgroup *cont, struct cftype *cft, struct file *file, char __user *buf, size_t nbytes, loff_t *ppos) |
8cdea7c05
|
3741 |
{ |
c0ff4b854
|
3742 |
struct mem_cgroup *memcg = mem_cgroup_from_cont(cont); |
af36f906c
|
3743 |
char str[64]; |
104f39284
|
3744 |
u64 val; |
af36f906c
|
3745 |
int type, name, len; |
8c7c6e34a
|
3746 3747 3748 |
type = MEMFILE_TYPE(cft->private); name = MEMFILE_ATTR(cft->private); |
af36f906c
|
3749 3750 3751 |
if (!do_swap_account && type == _MEMSWAP) return -EOPNOTSUPP; |
8c7c6e34a
|
3752 3753 |
switch (type) { case _MEM: |
104f39284
|
3754 |
if (name == RES_USAGE) |
c0ff4b854
|
3755 |
val = mem_cgroup_usage(memcg, false); |
104f39284
|
3756 |
else |
c0ff4b854
|
3757 |
val = res_counter_read_u64(&memcg->res, name); |
8c7c6e34a
|
3758 3759 |
break; case _MEMSWAP: |
104f39284
|
3760 |
if (name == RES_USAGE) |
c0ff4b854
|
3761 |
val = mem_cgroup_usage(memcg, true); |
104f39284
|
3762 |
else |
c0ff4b854
|
3763 |
val = res_counter_read_u64(&memcg->memsw, name); |
8c7c6e34a
|
3764 3765 3766 |
break; default: BUG(); |
8c7c6e34a
|
3767 |
} |
af36f906c
|
3768 3769 3770 3771 |
len = scnprintf(str, sizeof(str), "%llu ", (unsigned long long)val); return simple_read_from_buffer(buf, nbytes, ppos, str, len); |
8cdea7c05
|
3772 |
} |
628f42355
|
3773 3774 3775 3776 |
/* * The user of this function is... * RES_LIMIT. */ |
856c13aa1
|
3777 3778 |
static int mem_cgroup_write(struct cgroup *cont, struct cftype *cft, const char *buffer) |
8cdea7c05
|
3779 |
{ |
628f42355
|
3780 |
struct mem_cgroup *memcg = mem_cgroup_from_cont(cont); |
8c7c6e34a
|
3781 |
int type, name; |
628f42355
|
3782 3783 |
unsigned long long val; int ret; |
8c7c6e34a
|
3784 3785 |
type = MEMFILE_TYPE(cft->private); name = MEMFILE_ATTR(cft->private); |
af36f906c
|
3786 3787 3788 |
if (!do_swap_account && type == _MEMSWAP) return -EOPNOTSUPP; |
8c7c6e34a
|
3789 |
switch (name) { |
628f42355
|
3790 |
case RES_LIMIT: |
4b3bde4c9
|
3791 3792 3793 3794 |
if (mem_cgroup_is_root(memcg)) { /* Can't set limit on root */ ret = -EINVAL; break; } |
628f42355
|
3795 3796 |
/* This function does all necessary parse...reuse it */ ret = res_counter_memparse_write_strategy(buffer, &val); |
8c7c6e34a
|
3797 3798 3799 |
if (ret) break; if (type == _MEM) |
628f42355
|
3800 |
ret = mem_cgroup_resize_limit(memcg, val); |
8c7c6e34a
|
3801 3802 |
else ret = mem_cgroup_resize_memsw_limit(memcg, val); |
628f42355
|
3803 |
break; |
296c81d89
|
3804 3805 3806 3807 3808 3809 3810 3811 3812 3813 3814 3815 3816 3817 |
case RES_SOFT_LIMIT: ret = res_counter_memparse_write_strategy(buffer, &val); if (ret) break; /* * For memsw, soft limits are hard to implement in terms * of semantics, for now, we support soft limits for * control without swap */ if (type == _MEM) ret = res_counter_set_soft_limit(&memcg->res, val); else ret = -EINVAL; break; |
628f42355
|
3818 3819 3820 3821 3822 |
default: ret = -EINVAL; /* should be BUG() ? */ break; } return ret; |
8cdea7c05
|
3823 |
} |
fee7b548e
|
3824 3825 3826 3827 3828 3829 3830 3831 3832 3833 3834 3835 3836 3837 3838 3839 3840 3841 3842 3843 3844 3845 3846 3847 3848 |
static void memcg_get_hierarchical_limit(struct mem_cgroup *memcg, unsigned long long *mem_limit, unsigned long long *memsw_limit) { struct cgroup *cgroup; unsigned long long min_limit, min_memsw_limit, tmp; min_limit = res_counter_read_u64(&memcg->res, RES_LIMIT); min_memsw_limit = res_counter_read_u64(&memcg->memsw, RES_LIMIT); cgroup = memcg->css.cgroup; if (!memcg->use_hierarchy) goto out; while (cgroup->parent) { cgroup = cgroup->parent; memcg = mem_cgroup_from_cont(cgroup); if (!memcg->use_hierarchy) break; tmp = res_counter_read_u64(&memcg->res, RES_LIMIT); min_limit = min(min_limit, tmp); tmp = res_counter_read_u64(&memcg->memsw, RES_LIMIT); min_memsw_limit = min(min_memsw_limit, tmp); } out: *mem_limit = min_limit; *memsw_limit = min_memsw_limit; |
fee7b548e
|
3849 |
} |
29f2a4dac
|
3850 |
static int mem_cgroup_reset(struct cgroup *cont, unsigned int event) |
c84872e16
|
3851 |
{ |
af36f906c
|
3852 |
struct mem_cgroup *memcg = mem_cgroup_from_cont(cont); |
8c7c6e34a
|
3853 |
int type, name; |
c84872e16
|
3854 |
|
8c7c6e34a
|
3855 3856 |
type = MEMFILE_TYPE(event); name = MEMFILE_ATTR(event); |
af36f906c
|
3857 3858 3859 |
if (!do_swap_account && type == _MEMSWAP) return -EOPNOTSUPP; |
8c7c6e34a
|
3860 |
switch (name) { |
29f2a4dac
|
3861 |
case RES_MAX_USAGE: |
8c7c6e34a
|
3862 |
if (type == _MEM) |
c0ff4b854
|
3863 |
res_counter_reset_max(&memcg->res); |
8c7c6e34a
|
3864 |
else |
c0ff4b854
|
3865 |
res_counter_reset_max(&memcg->memsw); |
29f2a4dac
|
3866 3867 |
break; case RES_FAILCNT: |
8c7c6e34a
|
3868 |
if (type == _MEM) |
c0ff4b854
|
3869 |
res_counter_reset_failcnt(&memcg->res); |
8c7c6e34a
|
3870 |
else |
c0ff4b854
|
3871 |
res_counter_reset_failcnt(&memcg->memsw); |
29f2a4dac
|
3872 3873 |
break; } |
f64c3f549
|
3874 |
|
85cc59db1
|
3875 |
return 0; |
c84872e16
|
3876 |
} |
7dc74be03
|
3877 3878 3879 3880 3881 |
static u64 mem_cgroup_move_charge_read(struct cgroup *cgrp, struct cftype *cft) { return mem_cgroup_from_cont(cgrp)->move_charge_at_immigrate; } |
024914477
|
3882 |
#ifdef CONFIG_MMU |
7dc74be03
|
3883 3884 3885 |
static int mem_cgroup_move_charge_write(struct cgroup *cgrp, struct cftype *cft, u64 val) { |
c0ff4b854
|
3886 |
struct mem_cgroup *memcg = mem_cgroup_from_cont(cgrp); |
7dc74be03
|
3887 3888 3889 3890 3891 3892 3893 3894 3895 |
if (val >= (1 << NR_MOVE_TYPE)) return -EINVAL; /* * We check this value several times in both in can_attach() and * attach(), so we need cgroup lock to prevent this value from being * inconsistent. */ cgroup_lock(); |
c0ff4b854
|
3896 |
memcg->move_charge_at_immigrate = val; |
7dc74be03
|
3897 3898 3899 3900 |
cgroup_unlock(); return 0; } |
024914477
|
3901 3902 3903 3904 3905 3906 3907 |
#else static int mem_cgroup_move_charge_write(struct cgroup *cgrp, struct cftype *cft, u64 val) { return -ENOSYS; } #endif |
7dc74be03
|
3908 |
|
406eb0c9b
|
3909 |
#ifdef CONFIG_NUMA |
ab2158848
|
3910 |
static int memcg_numa_stat_show(struct cgroup *cont, struct cftype *cft, |
fada52ca0
|
3911 |
struct seq_file *m) |
406eb0c9b
|
3912 3913 3914 3915 |
{ int nid; unsigned long total_nr, file_nr, anon_nr, unevictable_nr; unsigned long node_nr; |
d79154bb5
|
3916 |
struct mem_cgroup *memcg = mem_cgroup_from_cont(cont); |
406eb0c9b
|
3917 |
|
d79154bb5
|
3918 |
total_nr = mem_cgroup_nr_lru_pages(memcg, LRU_ALL); |
406eb0c9b
|
3919 3920 |
seq_printf(m, "total=%lu", total_nr); for_each_node_state(nid, N_HIGH_MEMORY) { |
d79154bb5
|
3921 |
node_nr = mem_cgroup_node_nr_lru_pages(memcg, nid, LRU_ALL); |
406eb0c9b
|
3922 3923 3924 3925 |
seq_printf(m, " N%d=%lu", nid, node_nr); } seq_putc(m, ' '); |
d79154bb5
|
3926 |
file_nr = mem_cgroup_nr_lru_pages(memcg, LRU_ALL_FILE); |
406eb0c9b
|
3927 3928 |
seq_printf(m, "file=%lu", file_nr); for_each_node_state(nid, N_HIGH_MEMORY) { |
d79154bb5
|
3929 |
node_nr = mem_cgroup_node_nr_lru_pages(memcg, nid, |
bb2a0de92
|
3930 |
LRU_ALL_FILE); |
406eb0c9b
|
3931 3932 3933 3934 |
seq_printf(m, " N%d=%lu", nid, node_nr); } seq_putc(m, ' '); |
d79154bb5
|
3935 |
anon_nr = mem_cgroup_nr_lru_pages(memcg, LRU_ALL_ANON); |
406eb0c9b
|
3936 3937 |
seq_printf(m, "anon=%lu", anon_nr); for_each_node_state(nid, N_HIGH_MEMORY) { |
d79154bb5
|
3938 |
node_nr = mem_cgroup_node_nr_lru_pages(memcg, nid, |
bb2a0de92
|
3939 |
LRU_ALL_ANON); |
406eb0c9b
|
3940 3941 3942 3943 |
seq_printf(m, " N%d=%lu", nid, node_nr); } seq_putc(m, ' '); |
d79154bb5
|
3944 |
unevictable_nr = mem_cgroup_nr_lru_pages(memcg, BIT(LRU_UNEVICTABLE)); |
406eb0c9b
|
3945 3946 |
seq_printf(m, "unevictable=%lu", unevictable_nr); for_each_node_state(nid, N_HIGH_MEMORY) { |
d79154bb5
|
3947 |
node_nr = mem_cgroup_node_nr_lru_pages(memcg, nid, |
bb2a0de92
|
3948 |
BIT(LRU_UNEVICTABLE)); |
406eb0c9b
|
3949 3950 3951 3952 3953 3954 3955 |
seq_printf(m, " N%d=%lu", nid, node_nr); } seq_putc(m, ' '); return 0; } #endif /* CONFIG_NUMA */ |
af7c4b0ec
|
3956 3957 3958 3959 3960 3961 3962 3963 3964 3965 3966 3967 |
static const char * const mem_cgroup_lru_names[] = { "inactive_anon", "active_anon", "inactive_file", "active_file", "unevictable", }; static inline void mem_cgroup_lru_names_not_uptodate(void) { BUILD_BUG_ON(ARRAY_SIZE(mem_cgroup_lru_names) != NR_LRU_LISTS); } |
ab2158848
|
3968 |
static int memcg_stat_show(struct cgroup *cont, struct cftype *cft, |
78ccf5b5a
|
3969 |
struct seq_file *m) |
d2ceb9b7d
|
3970 |
{ |
d79154bb5
|
3971 |
struct mem_cgroup *memcg = mem_cgroup_from_cont(cont); |
af7c4b0ec
|
3972 3973 |
struct mem_cgroup *mi; unsigned int i; |
406eb0c9b
|
3974 |
|
af7c4b0ec
|
3975 |
for (i = 0; i < MEM_CGROUP_STAT_NSTATS; i++) { |
bff6bb83f
|
3976 |
if (i == MEM_CGROUP_STAT_SWAP && !do_swap_account) |
1dd3a2732
|
3977 |
continue; |
af7c4b0ec
|
3978 3979 3980 |
seq_printf(m, "%s %ld ", mem_cgroup_stat_names[i], mem_cgroup_read_stat(memcg, i) * PAGE_SIZE); |
1dd3a2732
|
3981 |
} |
7b854121e
|
3982 |
|
af7c4b0ec
|
3983 3984 3985 3986 3987 3988 3989 3990 3991 |
for (i = 0; i < MEM_CGROUP_EVENTS_NSTATS; i++) seq_printf(m, "%s %lu ", mem_cgroup_events_names[i], mem_cgroup_read_events(memcg, i)); for (i = 0; i < NR_LRU_LISTS; i++) seq_printf(m, "%s %lu ", mem_cgroup_lru_names[i], mem_cgroup_nr_lru_pages(memcg, BIT(i)) * PAGE_SIZE); |
14067bb3e
|
3992 |
/* Hierarchical information */ |
fee7b548e
|
3993 3994 |
{ unsigned long long limit, memsw_limit; |
d79154bb5
|
3995 |
memcg_get_hierarchical_limit(memcg, &limit, &memsw_limit); |
78ccf5b5a
|
3996 3997 |
seq_printf(m, "hierarchical_memory_limit %llu ", limit); |
fee7b548e
|
3998 |
if (do_swap_account) |
78ccf5b5a
|
3999 4000 4001 |
seq_printf(m, "hierarchical_memsw_limit %llu ", memsw_limit); |
fee7b548e
|
4002 |
} |
7f016ee8b
|
4003 |
|
af7c4b0ec
|
4004 4005 |
for (i = 0; i < MEM_CGROUP_STAT_NSTATS; i++) { long long val = 0; |
bff6bb83f
|
4006 |
if (i == MEM_CGROUP_STAT_SWAP && !do_swap_account) |
1dd3a2732
|
4007 |
continue; |
af7c4b0ec
|
4008 4009 4010 4011 4012 4013 4014 4015 4016 4017 4018 4019 4020 4021 4022 4023 4024 4025 4026 4027 4028 4029 4030 |
for_each_mem_cgroup_tree(mi, memcg) val += mem_cgroup_read_stat(mi, i) * PAGE_SIZE; seq_printf(m, "total_%s %lld ", mem_cgroup_stat_names[i], val); } for (i = 0; i < MEM_CGROUP_EVENTS_NSTATS; i++) { unsigned long long val = 0; for_each_mem_cgroup_tree(mi, memcg) val += mem_cgroup_read_events(mi, i); seq_printf(m, "total_%s %llu ", mem_cgroup_events_names[i], val); } for (i = 0; i < NR_LRU_LISTS; i++) { unsigned long long val = 0; for_each_mem_cgroup_tree(mi, memcg) val += mem_cgroup_nr_lru_pages(mi, BIT(i)) * PAGE_SIZE; seq_printf(m, "total_%s %llu ", mem_cgroup_lru_names[i], val); |
1dd3a2732
|
4031 |
} |
14067bb3e
|
4032 |
|
7f016ee8b
|
4033 |
#ifdef CONFIG_DEBUG_VM |
7f016ee8b
|
4034 4035 4036 |
{ int nid, zid; struct mem_cgroup_per_zone *mz; |
89abfab13
|
4037 |
struct zone_reclaim_stat *rstat; |
7f016ee8b
|
4038 4039 4040 4041 4042 |
unsigned long recent_rotated[2] = {0, 0}; unsigned long recent_scanned[2] = {0, 0}; for_each_online_node(nid) for (zid = 0; zid < MAX_NR_ZONES; zid++) { |
d79154bb5
|
4043 |
mz = mem_cgroup_zoneinfo(memcg, nid, zid); |
89abfab13
|
4044 |
rstat = &mz->lruvec.reclaim_stat; |
7f016ee8b
|
4045 |
|
89abfab13
|
4046 4047 4048 4049 |
recent_rotated[0] += rstat->recent_rotated[0]; recent_rotated[1] += rstat->recent_rotated[1]; recent_scanned[0] += rstat->recent_scanned[0]; recent_scanned[1] += rstat->recent_scanned[1]; |
7f016ee8b
|
4050 |
} |
78ccf5b5a
|
4051 4052 4053 4054 4055 4056 4057 4058 |
seq_printf(m, "recent_rotated_anon %lu ", recent_rotated[0]); seq_printf(m, "recent_rotated_file %lu ", recent_rotated[1]); seq_printf(m, "recent_scanned_anon %lu ", recent_scanned[0]); seq_printf(m, "recent_scanned_file %lu ", recent_scanned[1]); |
7f016ee8b
|
4059 4060 |
} #endif |
d2ceb9b7d
|
4061 4062 |
return 0; } |
a7885eb8a
|
4063 4064 4065 |
static u64 mem_cgroup_swappiness_read(struct cgroup *cgrp, struct cftype *cft) { struct mem_cgroup *memcg = mem_cgroup_from_cont(cgrp); |
1f4c025b5
|
4066 |
return mem_cgroup_swappiness(memcg); |
a7885eb8a
|
4067 4068 4069 4070 4071 4072 4073 |
} static int mem_cgroup_swappiness_write(struct cgroup *cgrp, struct cftype *cft, u64 val) { struct mem_cgroup *memcg = mem_cgroup_from_cont(cgrp); struct mem_cgroup *parent; |
068b38c1f
|
4074 |
|
a7885eb8a
|
4075 4076 4077 4078 4079 4080 4081 |
if (val > 100) return -EINVAL; if (cgrp->parent == NULL) return -EINVAL; parent = mem_cgroup_from_cont(cgrp->parent); |
068b38c1f
|
4082 4083 |
cgroup_lock(); |
a7885eb8a
|
4084 4085 |
/* If under hierarchy, only empty-root can set this value */ if ((parent->use_hierarchy) || |
068b38c1f
|
4086 4087 |
(memcg->use_hierarchy && !list_empty(&cgrp->children))) { cgroup_unlock(); |
a7885eb8a
|
4088 |
return -EINVAL; |
068b38c1f
|
4089 |
} |
a7885eb8a
|
4090 |
|
a7885eb8a
|
4091 |
memcg->swappiness = val; |
a7885eb8a
|
4092 |
|
068b38c1f
|
4093 |
cgroup_unlock(); |
a7885eb8a
|
4094 4095 |
return 0; } |
2e72b6347
|
4096 4097 4098 4099 4100 4101 4102 4103 |
static void __mem_cgroup_threshold(struct mem_cgroup *memcg, bool swap) { struct mem_cgroup_threshold_ary *t; u64 usage; int i; rcu_read_lock(); if (!swap) |
2c488db27
|
4104 |
t = rcu_dereference(memcg->thresholds.primary); |
2e72b6347
|
4105 |
else |
2c488db27
|
4106 |
t = rcu_dereference(memcg->memsw_thresholds.primary); |
2e72b6347
|
4107 4108 4109 4110 4111 4112 4113 |
if (!t) goto unlock; usage = mem_cgroup_usage(memcg, swap); /* |
748dad36d
|
4114 |
* current_threshold points to threshold just below or equal to usage. |
2e72b6347
|
4115 4116 4117 |
* If it's not true, a threshold was crossed after last * call of __mem_cgroup_threshold(). */ |
5407a5625
|
4118 |
i = t->current_threshold; |
2e72b6347
|
4119 4120 4121 4122 4123 4124 4125 4126 4127 4128 4129 4130 4131 4132 4133 4134 4135 4136 4137 4138 4139 4140 4141 |
/* * Iterate backward over array of thresholds starting from * current_threshold and check if a threshold is crossed. * If none of thresholds below usage is crossed, we read * only one element of the array here. */ for (; i >= 0 && unlikely(t->entries[i].threshold > usage); i--) eventfd_signal(t->entries[i].eventfd, 1); /* i = current_threshold + 1 */ i++; /* * Iterate forward over array of thresholds starting from * current_threshold+1 and check if a threshold is crossed. * If none of thresholds above usage is crossed, we read * only one element of the array here. */ for (; i < t->size && unlikely(t->entries[i].threshold <= usage); i++) eventfd_signal(t->entries[i].eventfd, 1); /* Update current_threshold */ |
5407a5625
|
4142 |
t->current_threshold = i - 1; |
2e72b6347
|
4143 4144 4145 4146 4147 4148 |
unlock: rcu_read_unlock(); } static void mem_cgroup_threshold(struct mem_cgroup *memcg) { |
ad4ca5f4b
|
4149 4150 4151 4152 4153 4154 4155 |
while (memcg) { __mem_cgroup_threshold(memcg, false); if (do_swap_account) __mem_cgroup_threshold(memcg, true); memcg = parent_mem_cgroup(memcg); } |
2e72b6347
|
4156 4157 4158 4159 4160 4161 4162 4163 4164 |
} static int compare_thresholds(const void *a, const void *b) { const struct mem_cgroup_threshold *_a = a; const struct mem_cgroup_threshold *_b = b; return _a->threshold - _b->threshold; } |
c0ff4b854
|
4165 |
static int mem_cgroup_oom_notify_cb(struct mem_cgroup *memcg) |
9490ff275
|
4166 4167 |
{ struct mem_cgroup_eventfd_list *ev; |
c0ff4b854
|
4168 |
list_for_each_entry(ev, &memcg->oom_notify, list) |
9490ff275
|
4169 4170 4171 |
eventfd_signal(ev->eventfd, 1); return 0; } |
c0ff4b854
|
4172 |
static void mem_cgroup_oom_notify(struct mem_cgroup *memcg) |
9490ff275
|
4173 |
{ |
7d74b06f2
|
4174 |
struct mem_cgroup *iter; |
c0ff4b854
|
4175 |
for_each_mem_cgroup_tree(iter, memcg) |
7d74b06f2
|
4176 |
mem_cgroup_oom_notify_cb(iter); |
9490ff275
|
4177 4178 4179 4180 |
} static int mem_cgroup_usage_register_event(struct cgroup *cgrp, struct cftype *cft, struct eventfd_ctx *eventfd, const char *args) |
2e72b6347
|
4181 4182 |
{ struct mem_cgroup *memcg = mem_cgroup_from_cont(cgrp); |
2c488db27
|
4183 4184 |
struct mem_cgroup_thresholds *thresholds; struct mem_cgroup_threshold_ary *new; |
2e72b6347
|
4185 4186 |
int type = MEMFILE_TYPE(cft->private); u64 threshold, usage; |
2c488db27
|
4187 |
int i, size, ret; |
2e72b6347
|
4188 4189 4190 4191 4192 4193 |
ret = res_counter_memparse_write_strategy(args, &threshold); if (ret) return ret; mutex_lock(&memcg->thresholds_lock); |
2c488db27
|
4194 |
|
2e72b6347
|
4195 |
if (type == _MEM) |
2c488db27
|
4196 |
thresholds = &memcg->thresholds; |
2e72b6347
|
4197 |
else if (type == _MEMSWAP) |
2c488db27
|
4198 |
thresholds = &memcg->memsw_thresholds; |
2e72b6347
|
4199 4200 4201 4202 4203 4204 |
else BUG(); usage = mem_cgroup_usage(memcg, type == _MEMSWAP); /* Check if a threshold crossed before adding a new one */ |
2c488db27
|
4205 |
if (thresholds->primary) |
2e72b6347
|
4206 |
__mem_cgroup_threshold(memcg, type == _MEMSWAP); |
2c488db27
|
4207 |
size = thresholds->primary ? thresholds->primary->size + 1 : 1; |
2e72b6347
|
4208 4209 |
/* Allocate memory for new array of thresholds */ |
2c488db27
|
4210 |
new = kmalloc(sizeof(*new) + size * sizeof(struct mem_cgroup_threshold), |
2e72b6347
|
4211 |
GFP_KERNEL); |
2c488db27
|
4212 |
if (!new) { |
2e72b6347
|
4213 4214 4215 |
ret = -ENOMEM; goto unlock; } |
2c488db27
|
4216 |
new->size = size; |
2e72b6347
|
4217 4218 |
/* Copy thresholds (if any) to new array */ |
2c488db27
|
4219 4220 |
if (thresholds->primary) { memcpy(new->entries, thresholds->primary->entries, (size - 1) * |
2e72b6347
|
4221 |
sizeof(struct mem_cgroup_threshold)); |
2c488db27
|
4222 |
} |
2e72b6347
|
4223 |
/* Add new threshold */ |
2c488db27
|
4224 4225 |
new->entries[size - 1].eventfd = eventfd; new->entries[size - 1].threshold = threshold; |
2e72b6347
|
4226 4227 |
/* Sort thresholds. Registering of new threshold isn't time-critical */ |
2c488db27
|
4228 |
sort(new->entries, size, sizeof(struct mem_cgroup_threshold), |
2e72b6347
|
4229 4230 4231 |
compare_thresholds, NULL); /* Find current threshold */ |
2c488db27
|
4232 |
new->current_threshold = -1; |
2e72b6347
|
4233 |
for (i = 0; i < size; i++) { |
748dad36d
|
4234 |
if (new->entries[i].threshold <= usage) { |
2e72b6347
|
4235 |
/* |
2c488db27
|
4236 4237 |
* new->current_threshold will not be used until * rcu_assign_pointer(), so it's safe to increment |
2e72b6347
|
4238 4239 |
* it here. */ |
2c488db27
|
4240 |
++new->current_threshold; |
748dad36d
|
4241 4242 |
} else break; |
2e72b6347
|
4243 |
} |
2c488db27
|
4244 4245 4246 4247 4248 |
/* Free old spare buffer and save old primary buffer as spare */ kfree(thresholds->spare); thresholds->spare = thresholds->primary; rcu_assign_pointer(thresholds->primary, new); |
2e72b6347
|
4249 |
|
907860ed3
|
4250 |
/* To be sure that nobody uses thresholds */ |
2e72b6347
|
4251 |
synchronize_rcu(); |
2e72b6347
|
4252 4253 4254 4255 4256 |
unlock: mutex_unlock(&memcg->thresholds_lock); return ret; } |
907860ed3
|
4257 |
static void mem_cgroup_usage_unregister_event(struct cgroup *cgrp, |
9490ff275
|
4258 |
struct cftype *cft, struct eventfd_ctx *eventfd) |
2e72b6347
|
4259 4260 |
{ struct mem_cgroup *memcg = mem_cgroup_from_cont(cgrp); |
2c488db27
|
4261 4262 |
struct mem_cgroup_thresholds *thresholds; struct mem_cgroup_threshold_ary *new; |
2e72b6347
|
4263 4264 |
int type = MEMFILE_TYPE(cft->private); u64 usage; |
2c488db27
|
4265 |
int i, j, size; |
2e72b6347
|
4266 4267 4268 |
mutex_lock(&memcg->thresholds_lock); if (type == _MEM) |
2c488db27
|
4269 |
thresholds = &memcg->thresholds; |
2e72b6347
|
4270 |
else if (type == _MEMSWAP) |
2c488db27
|
4271 |
thresholds = &memcg->memsw_thresholds; |
2e72b6347
|
4272 4273 |
else BUG(); |
371528cae
|
4274 4275 |
if (!thresholds->primary) goto unlock; |
2e72b6347
|
4276 4277 4278 4279 4280 4281 |
usage = mem_cgroup_usage(memcg, type == _MEMSWAP); /* Check if a threshold crossed before removing */ __mem_cgroup_threshold(memcg, type == _MEMSWAP); /* Calculate new number of threshold */ |
2c488db27
|
4282 4283 4284 |
size = 0; for (i = 0; i < thresholds->primary->size; i++) { if (thresholds->primary->entries[i].eventfd != eventfd) |
2e72b6347
|
4285 4286 |
size++; } |
2c488db27
|
4287 |
new = thresholds->spare; |
907860ed3
|
4288 |
|
2e72b6347
|
4289 4290 |
/* Set thresholds array to NULL if we don't have thresholds */ if (!size) { |
2c488db27
|
4291 4292 |
kfree(new); new = NULL; |
907860ed3
|
4293 |
goto swap_buffers; |
2e72b6347
|
4294 |
} |
2c488db27
|
4295 |
new->size = size; |
2e72b6347
|
4296 4297 |
/* Copy thresholds and find current threshold */ |
2c488db27
|
4298 4299 4300 |
new->current_threshold = -1; for (i = 0, j = 0; i < thresholds->primary->size; i++) { if (thresholds->primary->entries[i].eventfd == eventfd) |
2e72b6347
|
4301 |
continue; |
2c488db27
|
4302 |
new->entries[j] = thresholds->primary->entries[i]; |
748dad36d
|
4303 |
if (new->entries[j].threshold <= usage) { |
2e72b6347
|
4304 |
/* |
2c488db27
|
4305 |
* new->current_threshold will not be used |
2e72b6347
|
4306 4307 4308 |
* until rcu_assign_pointer(), so it's safe to increment * it here. */ |
2c488db27
|
4309 |
++new->current_threshold; |
2e72b6347
|
4310 4311 4312 |
} j++; } |
907860ed3
|
4313 |
swap_buffers: |
2c488db27
|
4314 4315 |
/* Swap primary and spare array */ thresholds->spare = thresholds->primary; |
8c7577637
|
4316 4317 4318 4319 4320 |
/* If all events are unregistered, free the spare array */ if (!new) { kfree(thresholds->spare); thresholds->spare = NULL; } |
2c488db27
|
4321 |
rcu_assign_pointer(thresholds->primary, new); |
2e72b6347
|
4322 |
|
907860ed3
|
4323 |
/* To be sure that nobody uses thresholds */ |
2e72b6347
|
4324 |
synchronize_rcu(); |
371528cae
|
4325 |
unlock: |
2e72b6347
|
4326 |
mutex_unlock(&memcg->thresholds_lock); |
2e72b6347
|
4327 |
} |
c1e862c1f
|
4328 |
|
9490ff275
|
4329 4330 4331 4332 4333 4334 4335 4336 4337 4338 4339 |
static int mem_cgroup_oom_register_event(struct cgroup *cgrp, struct cftype *cft, struct eventfd_ctx *eventfd, const char *args) { struct mem_cgroup *memcg = mem_cgroup_from_cont(cgrp); struct mem_cgroup_eventfd_list *event; int type = MEMFILE_TYPE(cft->private); BUG_ON(type != _OOM_TYPE); event = kmalloc(sizeof(*event), GFP_KERNEL); if (!event) return -ENOMEM; |
1af8efe96
|
4340 |
spin_lock(&memcg_oom_lock); |
9490ff275
|
4341 4342 4343 4344 4345 |
event->eventfd = eventfd; list_add(&event->list, &memcg->oom_notify); /* already in OOM ? */ |
79dfdaccd
|
4346 |
if (atomic_read(&memcg->under_oom)) |
9490ff275
|
4347 |
eventfd_signal(eventfd, 1); |
1af8efe96
|
4348 |
spin_unlock(&memcg_oom_lock); |
9490ff275
|
4349 4350 4351 |
return 0; } |
907860ed3
|
4352 |
static void mem_cgroup_oom_unregister_event(struct cgroup *cgrp, |
9490ff275
|
4353 4354 |
struct cftype *cft, struct eventfd_ctx *eventfd) { |
c0ff4b854
|
4355 |
struct mem_cgroup *memcg = mem_cgroup_from_cont(cgrp); |
9490ff275
|
4356 4357 4358 4359 |
struct mem_cgroup_eventfd_list *ev, *tmp; int type = MEMFILE_TYPE(cft->private); BUG_ON(type != _OOM_TYPE); |
1af8efe96
|
4360 |
spin_lock(&memcg_oom_lock); |
9490ff275
|
4361 |
|
c0ff4b854
|
4362 |
list_for_each_entry_safe(ev, tmp, &memcg->oom_notify, list) { |
9490ff275
|
4363 4364 4365 4366 4367 |
if (ev->eventfd == eventfd) { list_del(&ev->list); kfree(ev); } } |
1af8efe96
|
4368 |
spin_unlock(&memcg_oom_lock); |
9490ff275
|
4369 |
} |
3c11ecf44
|
4370 4371 4372 |
static int mem_cgroup_oom_control_read(struct cgroup *cgrp, struct cftype *cft, struct cgroup_map_cb *cb) { |
c0ff4b854
|
4373 |
struct mem_cgroup *memcg = mem_cgroup_from_cont(cgrp); |
3c11ecf44
|
4374 |
|
c0ff4b854
|
4375 |
cb->fill(cb, "oom_kill_disable", memcg->oom_kill_disable); |
3c11ecf44
|
4376 |
|
c0ff4b854
|
4377 |
if (atomic_read(&memcg->under_oom)) |
3c11ecf44
|
4378 4379 4380 4381 4382 |
cb->fill(cb, "under_oom", 1); else cb->fill(cb, "under_oom", 0); return 0; } |
3c11ecf44
|
4383 4384 4385 |
static int mem_cgroup_oom_control_write(struct cgroup *cgrp, struct cftype *cft, u64 val) { |
c0ff4b854
|
4386 |
struct mem_cgroup *memcg = mem_cgroup_from_cont(cgrp); |
3c11ecf44
|
4387 4388 4389 4390 4391 4392 4393 4394 4395 4396 4397 |
struct mem_cgroup *parent; /* cannot set to root cgroup and only 0 and 1 are allowed */ if (!cgrp->parent || !((val == 0) || (val == 1))) return -EINVAL; parent = mem_cgroup_from_cont(cgrp->parent); cgroup_lock(); /* oom-kill-disable is a flag for subhierarchy. */ if ((parent->use_hierarchy) || |
c0ff4b854
|
4398 |
(memcg->use_hierarchy && !list_empty(&cgrp->children))) { |
3c11ecf44
|
4399 4400 4401 |
cgroup_unlock(); return -EINVAL; } |
c0ff4b854
|
4402 |
memcg->oom_kill_disable = val; |
4d845ebf4
|
4403 |
if (!val) |
c0ff4b854
|
4404 |
memcg_oom_recover(memcg); |
3c11ecf44
|
4405 4406 4407 |
cgroup_unlock(); return 0; } |
c255a4580
|
4408 |
#ifdef CONFIG_MEMCG_KMEM |
cbe128e34
|
4409 |
static int memcg_init_kmem(struct mem_cgroup *memcg, struct cgroup_subsys *ss) |
e5671dfae
|
4410 |
{ |
1d62e4365
|
4411 |
return mem_cgroup_sockets_init(memcg, ss); |
e5671dfae
|
4412 |
}; |
1d62e4365
|
4413 |
static void kmem_cgroup_destroy(struct mem_cgroup *memcg) |
d1a4c0b37
|
4414 |
{ |
1d62e4365
|
4415 |
mem_cgroup_sockets_destroy(memcg); |
d1a4c0b37
|
4416 |
} |
e5671dfae
|
4417 |
#else |
cbe128e34
|
4418 |
static int memcg_init_kmem(struct mem_cgroup *memcg, struct cgroup_subsys *ss) |
e5671dfae
|
4419 4420 4421 |
{ return 0; } |
d1a4c0b37
|
4422 |
|
1d62e4365
|
4423 |
static void kmem_cgroup_destroy(struct mem_cgroup *memcg) |
d1a4c0b37
|
4424 4425 |
{ } |
e5671dfae
|
4426 |
#endif |
8cdea7c05
|
4427 4428 |
static struct cftype mem_cgroup_files[] = { { |
0eea10301
|
4429 |
.name = "usage_in_bytes", |
8c7c6e34a
|
4430 |
.private = MEMFILE_PRIVATE(_MEM, RES_USAGE), |
af36f906c
|
4431 |
.read = mem_cgroup_read, |
9490ff275
|
4432 4433 |
.register_event = mem_cgroup_usage_register_event, .unregister_event = mem_cgroup_usage_unregister_event, |
8cdea7c05
|
4434 4435 |
}, { |
c84872e16
|
4436 |
.name = "max_usage_in_bytes", |
8c7c6e34a
|
4437 |
.private = MEMFILE_PRIVATE(_MEM, RES_MAX_USAGE), |
29f2a4dac
|
4438 |
.trigger = mem_cgroup_reset, |
af36f906c
|
4439 |
.read = mem_cgroup_read, |
c84872e16
|
4440 4441 |
}, { |
0eea10301
|
4442 |
.name = "limit_in_bytes", |
8c7c6e34a
|
4443 |
.private = MEMFILE_PRIVATE(_MEM, RES_LIMIT), |
856c13aa1
|
4444 |
.write_string = mem_cgroup_write, |
af36f906c
|
4445 |
.read = mem_cgroup_read, |
8cdea7c05
|
4446 4447 |
}, { |
296c81d89
|
4448 4449 4450 |
.name = "soft_limit_in_bytes", .private = MEMFILE_PRIVATE(_MEM, RES_SOFT_LIMIT), .write_string = mem_cgroup_write, |
af36f906c
|
4451 |
.read = mem_cgroup_read, |
296c81d89
|
4452 4453 |
}, { |
8cdea7c05
|
4454 |
.name = "failcnt", |
8c7c6e34a
|
4455 |
.private = MEMFILE_PRIVATE(_MEM, RES_FAILCNT), |
29f2a4dac
|
4456 |
.trigger = mem_cgroup_reset, |
af36f906c
|
4457 |
.read = mem_cgroup_read, |
8cdea7c05
|
4458 |
}, |
8697d3319
|
4459 |
{ |
d2ceb9b7d
|
4460 |
.name = "stat", |
ab2158848
|
4461 |
.read_seq_string = memcg_stat_show, |
d2ceb9b7d
|
4462 |
}, |
c1e862c1f
|
4463 4464 4465 4466 |
{ .name = "force_empty", .trigger = mem_cgroup_force_empty_write, }, |
18f59ea7d
|
4467 4468 4469 4470 4471 |
{ .name = "use_hierarchy", .write_u64 = mem_cgroup_hierarchy_write, .read_u64 = mem_cgroup_hierarchy_read, }, |
a7885eb8a
|
4472 4473 4474 4475 4476 |
{ .name = "swappiness", .read_u64 = mem_cgroup_swappiness_read, .write_u64 = mem_cgroup_swappiness_write, }, |
7dc74be03
|
4477 4478 4479 4480 4481 |
{ .name = "move_charge_at_immigrate", .read_u64 = mem_cgroup_move_charge_read, .write_u64 = mem_cgroup_move_charge_write, }, |
9490ff275
|
4482 4483 |
{ .name = "oom_control", |
3c11ecf44
|
4484 4485 |
.read_map = mem_cgroup_oom_control_read, .write_u64 = mem_cgroup_oom_control_write, |
9490ff275
|
4486 4487 4488 4489 |
.register_event = mem_cgroup_oom_register_event, .unregister_event = mem_cgroup_oom_unregister_event, .private = MEMFILE_PRIVATE(_OOM_TYPE, OOM_CONTROL), }, |
406eb0c9b
|
4490 4491 4492 |
#ifdef CONFIG_NUMA { .name = "numa_stat", |
ab2158848
|
4493 |
.read_seq_string = memcg_numa_stat_show, |
406eb0c9b
|
4494 4495 |
}, #endif |
c255a4580
|
4496 |
#ifdef CONFIG_MEMCG_SWAP |
8c7c6e34a
|
4497 4498 4499 |
{ .name = "memsw.usage_in_bytes", .private = MEMFILE_PRIVATE(_MEMSWAP, RES_USAGE), |
af36f906c
|
4500 |
.read = mem_cgroup_read, |
9490ff275
|
4501 4502 |
.register_event = mem_cgroup_usage_register_event, .unregister_event = mem_cgroup_usage_unregister_event, |
8c7c6e34a
|
4503 4504 4505 4506 4507 |
}, { .name = "memsw.max_usage_in_bytes", .private = MEMFILE_PRIVATE(_MEMSWAP, RES_MAX_USAGE), .trigger = mem_cgroup_reset, |
af36f906c
|
4508 |
.read = mem_cgroup_read, |
8c7c6e34a
|
4509 4510 4511 4512 4513 |
}, { .name = "memsw.limit_in_bytes", .private = MEMFILE_PRIVATE(_MEMSWAP, RES_LIMIT), .write_string = mem_cgroup_write, |
af36f906c
|
4514 |
.read = mem_cgroup_read, |
8c7c6e34a
|
4515 4516 4517 4518 4519 |
}, { .name = "memsw.failcnt", .private = MEMFILE_PRIVATE(_MEMSWAP, RES_FAILCNT), .trigger = mem_cgroup_reset, |
af36f906c
|
4520 |
.read = mem_cgroup_read, |
8c7c6e34a
|
4521 |
}, |
8c7c6e34a
|
4522 |
#endif |
6bc103498
|
4523 |
{ }, /* terminate */ |
af36f906c
|
4524 |
}; |
8c7c6e34a
|
4525 |
|
c0ff4b854
|
4526 |
static int alloc_mem_cgroup_per_zone_info(struct mem_cgroup *memcg, int node) |
6d12e2d8d
|
4527 4528 |
{ struct mem_cgroup_per_node *pn; |
1ecaab2bd
|
4529 |
struct mem_cgroup_per_zone *mz; |
41e3355de
|
4530 |
int zone, tmp = node; |
1ecaab2bd
|
4531 4532 4533 4534 4535 4536 4537 4538 |
/* * This routine is called against possible nodes. * But it's BUG to call kmalloc() against offline node. * * TODO: this routine can waste much memory for nodes which will * never be onlined. It's better to use memory hotplug callback * function. */ |
41e3355de
|
4539 4540 |
if (!node_state(node, N_NORMAL_MEMORY)) tmp = -1; |
17295c88a
|
4541 |
pn = kzalloc_node(sizeof(*pn), GFP_KERNEL, tmp); |
6d12e2d8d
|
4542 4543 |
if (!pn) return 1; |
1ecaab2bd
|
4544 |
|
1ecaab2bd
|
4545 4546 |
for (zone = 0; zone < MAX_NR_ZONES; zone++) { mz = &pn->zoneinfo[zone]; |
7f5e86c2c
|
4547 |
lruvec_init(&mz->lruvec, &NODE_DATA(node)->node_zones[zone]); |
f64c3f549
|
4548 |
mz->usage_in_excess = 0; |
4e4169535
|
4549 |
mz->on_tree = false; |
d79154bb5
|
4550 |
mz->memcg = memcg; |
1ecaab2bd
|
4551 |
} |
0a619e587
|
4552 |
memcg->info.nodeinfo[node] = pn; |
6d12e2d8d
|
4553 4554 |
return 0; } |
c0ff4b854
|
4555 |
static void free_mem_cgroup_per_zone_info(struct mem_cgroup *memcg, int node) |
1ecaab2bd
|
4556 |
{ |
c0ff4b854
|
4557 |
kfree(memcg->info.nodeinfo[node]); |
1ecaab2bd
|
4558 |
} |
333279487
|
4559 4560 |
static struct mem_cgroup *mem_cgroup_alloc(void) { |
d79154bb5
|
4561 |
struct mem_cgroup *memcg; |
c62b1a3b3
|
4562 |
int size = sizeof(struct mem_cgroup); |
333279487
|
4563 |
|
c62b1a3b3
|
4564 |
/* Can be very big if MAX_NUMNODES is very big */ |
c8dad2bb6
|
4565 |
if (size < PAGE_SIZE) |
d79154bb5
|
4566 |
memcg = kzalloc(size, GFP_KERNEL); |
333279487
|
4567 |
else |
d79154bb5
|
4568 |
memcg = vzalloc(size); |
333279487
|
4569 |
|
d79154bb5
|
4570 |
if (!memcg) |
e7bbcdf37
|
4571 |
return NULL; |
d79154bb5
|
4572 4573 |
memcg->stat = alloc_percpu(struct mem_cgroup_stat_cpu); if (!memcg->stat) |
d2e61b8dc
|
4574 |
goto out_free; |
d79154bb5
|
4575 4576 |
spin_lock_init(&memcg->pcp_counter_lock); return memcg; |
d2e61b8dc
|
4577 4578 4579 |
out_free: if (size < PAGE_SIZE) |
d79154bb5
|
4580 |
kfree(memcg); |
d2e61b8dc
|
4581 |
else |
d79154bb5
|
4582 |
vfree(memcg); |
d2e61b8dc
|
4583 |
return NULL; |
333279487
|
4584 |
} |
8c7c6e34a
|
4585 |
/* |
3afe36b1f
|
4586 |
* Helpers for freeing a kmalloc()ed/vzalloc()ed mem_cgroup by RCU, |
59927fb98
|
4587 4588 4589 |
* but in process context. The work_freeing structure is overlaid * on the rcu_freeing structure, which itself is overlaid on memsw. */ |
3afe36b1f
|
4590 |
static void free_work(struct work_struct *work) |
59927fb98
|
4591 4592 |
{ struct mem_cgroup *memcg; |
3afe36b1f
|
4593 |
int size = sizeof(struct mem_cgroup); |
59927fb98
|
4594 4595 |
memcg = container_of(work, struct mem_cgroup, work_freeing); |
3f1346193
|
4596 4597 4598 4599 4600 4601 4602 4603 4604 4605 4606 4607 |
/* * We need to make sure that (at least for now), the jump label * destruction code runs outside of the cgroup lock. This is because * get_online_cpus(), which is called from the static_branch update, * can't be called inside the cgroup_lock. cpusets are the ones * enforcing this dependency, so if they ever change, we might as well. * * schedule_work() will guarantee this happens. Be careful if you need * to move this code around, and make sure it is outside * the cgroup_lock. */ disarm_sock_keys(memcg); |
3afe36b1f
|
4608 4609 4610 4611 |
if (size < PAGE_SIZE) kfree(memcg); else vfree(memcg); |
59927fb98
|
4612 |
} |
3afe36b1f
|
4613 4614 |
static void free_rcu(struct rcu_head *rcu_head) |
59927fb98
|
4615 4616 4617 4618 |
{ struct mem_cgroup *memcg; memcg = container_of(rcu_head, struct mem_cgroup, rcu_freeing); |
3afe36b1f
|
4619 |
INIT_WORK(&memcg->work_freeing, free_work); |
59927fb98
|
4620 4621 4622 4623 |
schedule_work(&memcg->work_freeing); } /* |
8c7c6e34a
|
4624 4625 4626 4627 4628 4629 4630 |
* At destroying mem_cgroup, references from swap_cgroup can remain. * (scanning all at force_empty is too costly...) * * Instead of clearing all references at force_empty, we remember * the number of reference from swap_cgroup and free mem_cgroup when * it goes down to 0. * |
8c7c6e34a
|
4631 4632 |
* Removal of cgroup itself succeeds regardless of refs from swap. */ |
c0ff4b854
|
4633 |
static void __mem_cgroup_free(struct mem_cgroup *memcg) |
333279487
|
4634 |
{ |
08e552c69
|
4635 |
int node; |
c0ff4b854
|
4636 4637 |
mem_cgroup_remove_from_trees(memcg); free_css_id(&mem_cgroup_subsys, &memcg->css); |
04046e1a0
|
4638 |
|
3ed28fa10
|
4639 |
for_each_node(node) |
c0ff4b854
|
4640 |
free_mem_cgroup_per_zone_info(memcg, node); |
08e552c69
|
4641 |
|
c0ff4b854
|
4642 |
free_percpu(memcg->stat); |
3afe36b1f
|
4643 |
call_rcu(&memcg->rcu_freeing, free_rcu); |
333279487
|
4644 |
} |
c0ff4b854
|
4645 |
static void mem_cgroup_get(struct mem_cgroup *memcg) |
8c7c6e34a
|
4646 |
{ |
c0ff4b854
|
4647 |
atomic_inc(&memcg->refcnt); |
8c7c6e34a
|
4648 |
} |
c0ff4b854
|
4649 |
static void __mem_cgroup_put(struct mem_cgroup *memcg, int count) |
8c7c6e34a
|
4650 |
{ |
c0ff4b854
|
4651 4652 4653 |
if (atomic_sub_and_test(count, &memcg->refcnt)) { struct mem_cgroup *parent = parent_mem_cgroup(memcg); __mem_cgroup_free(memcg); |
7bcc1bb12
|
4654 4655 4656 |
if (parent) mem_cgroup_put(parent); } |
8c7c6e34a
|
4657 |
} |
c0ff4b854
|
4658 |
static void mem_cgroup_put(struct mem_cgroup *memcg) |
483c30b51
|
4659 |
{ |
c0ff4b854
|
4660 |
__mem_cgroup_put(memcg, 1); |
483c30b51
|
4661 |
} |
7bcc1bb12
|
4662 4663 4664 |
/* * Returns the parent mem_cgroup in memcgroup hierarchy with hierarchy enabled. */ |
e1aab161e
|
4665 |
struct mem_cgroup *parent_mem_cgroup(struct mem_cgroup *memcg) |
7bcc1bb12
|
4666 |
{ |
c0ff4b854
|
4667 |
if (!memcg->res.parent) |
7bcc1bb12
|
4668 |
return NULL; |
c0ff4b854
|
4669 |
return mem_cgroup_from_res_counter(memcg->res.parent, res); |
7bcc1bb12
|
4670 |
} |
e1aab161e
|
4671 |
EXPORT_SYMBOL(parent_mem_cgroup); |
333279487
|
4672 |
|
c255a4580
|
4673 |
#ifdef CONFIG_MEMCG_SWAP |
c077719be
|
4674 4675 |
static void __init enable_swap_cgroup(void) { |
f8d665422
|
4676 |
if (!mem_cgroup_disabled() && really_do_swap_account) |
c077719be
|
4677 4678 4679 4680 4681 4682 4683 |
do_swap_account = 1; } #else static void __init enable_swap_cgroup(void) { } #endif |
f64c3f549
|
4684 4685 4686 4687 4688 |
static int mem_cgroup_soft_limit_tree_init(void) { struct mem_cgroup_tree_per_node *rtpn; struct mem_cgroup_tree_per_zone *rtpz; int tmp, node, zone; |
3ed28fa10
|
4689 |
for_each_node(node) { |
f64c3f549
|
4690 4691 4692 4693 4694 |
tmp = node; if (!node_state(node, N_NORMAL_MEMORY)) tmp = -1; rtpn = kzalloc_node(sizeof(*rtpn), GFP_KERNEL, tmp); if (!rtpn) |
c3cecc683
|
4695 |
goto err_cleanup; |
f64c3f549
|
4696 4697 4698 4699 4700 4701 4702 4703 4704 4705 |
soft_limit_tree.rb_tree_per_node[node] = rtpn; for (zone = 0; zone < MAX_NR_ZONES; zone++) { rtpz = &rtpn->rb_tree_per_zone[zone]; rtpz->rb_root = RB_ROOT; spin_lock_init(&rtpz->lock); } } return 0; |
c3cecc683
|
4706 4707 |
err_cleanup: |
3ed28fa10
|
4708 |
for_each_node(node) { |
c3cecc683
|
4709 4710 4711 4712 4713 4714 |
if (!soft_limit_tree.rb_tree_per_node[node]) break; kfree(soft_limit_tree.rb_tree_per_node[node]); soft_limit_tree.rb_tree_per_node[node] = NULL; } return 1; |
f64c3f549
|
4715 |
} |
0eb253e22
|
4716 |
static struct cgroup_subsys_state * __ref |
761b3ef50
|
4717 |
mem_cgroup_create(struct cgroup *cont) |
8cdea7c05
|
4718 |
{ |
c0ff4b854
|
4719 |
struct mem_cgroup *memcg, *parent; |
04046e1a0
|
4720 |
long error = -ENOMEM; |
6d12e2d8d
|
4721 |
int node; |
8cdea7c05
|
4722 |
|
c0ff4b854
|
4723 4724 |
memcg = mem_cgroup_alloc(); if (!memcg) |
04046e1a0
|
4725 |
return ERR_PTR(error); |
78fb74669
|
4726 |
|
3ed28fa10
|
4727 |
for_each_node(node) |
c0ff4b854
|
4728 |
if (alloc_mem_cgroup_per_zone_info(memcg, node)) |
6d12e2d8d
|
4729 |
goto free_out; |
f64c3f549
|
4730 |
|
c077719be
|
4731 |
/* root ? */ |
28dbc4b6a
|
4732 |
if (cont->parent == NULL) { |
cdec2e426
|
4733 |
int cpu; |
c077719be
|
4734 |
enable_swap_cgroup(); |
28dbc4b6a
|
4735 |
parent = NULL; |
f64c3f549
|
4736 4737 |
if (mem_cgroup_soft_limit_tree_init()) goto free_out; |
a41c58a66
|
4738 |
root_mem_cgroup = memcg; |
cdec2e426
|
4739 4740 4741 4742 4743 |
for_each_possible_cpu(cpu) { struct memcg_stock_pcp *stock = &per_cpu(memcg_stock, cpu); INIT_WORK(&stock->work, drain_local_stock); } |
711d3d2c9
|
4744 |
hotcpu_notifier(memcg_cpu_hotplug_callback, 0); |
18f59ea7d
|
4745 |
} else { |
28dbc4b6a
|
4746 |
parent = mem_cgroup_from_cont(cont->parent); |
c0ff4b854
|
4747 4748 |
memcg->use_hierarchy = parent->use_hierarchy; memcg->oom_kill_disable = parent->oom_kill_disable; |
18f59ea7d
|
4749 |
} |
28dbc4b6a
|
4750 |
|
18f59ea7d
|
4751 |
if (parent && parent->use_hierarchy) { |
c0ff4b854
|
4752 4753 |
res_counter_init(&memcg->res, &parent->res); res_counter_init(&memcg->memsw, &parent->memsw); |
7bcc1bb12
|
4754 4755 4756 4757 4758 4759 4760 |
/* * We increment refcnt of the parent to ensure that we can * safely access it on res_counter_charge/uncharge. * This refcnt will be decremented when freeing this * mem_cgroup(see mem_cgroup_put). */ mem_cgroup_get(parent); |
18f59ea7d
|
4761 |
} else { |
c0ff4b854
|
4762 4763 |
res_counter_init(&memcg->res, NULL); res_counter_init(&memcg->memsw, NULL); |
18f59ea7d
|
4764 |
} |
c0ff4b854
|
4765 4766 |
memcg->last_scanned_node = MAX_NUMNODES; INIT_LIST_HEAD(&memcg->oom_notify); |
6d61ef409
|
4767 |
|
a7885eb8a
|
4768 |
if (parent) |
c0ff4b854
|
4769 4770 4771 4772 |
memcg->swappiness = mem_cgroup_swappiness(parent); atomic_set(&memcg->refcnt, 1); memcg->move_charge_at_immigrate = 0; mutex_init(&memcg->thresholds_lock); |
312734c04
|
4773 |
spin_lock_init(&memcg->move_lock); |
cbe128e34
|
4774 4775 4776 4777 4778 4779 4780 4781 4782 4783 4784 |
error = memcg_init_kmem(memcg, &mem_cgroup_subsys); if (error) { /* * We call put now because our (and parent's) refcnts * are already in place. mem_cgroup_put() will internally * call __mem_cgroup_free, so return directly */ mem_cgroup_put(memcg); return ERR_PTR(error); } |
c0ff4b854
|
4785 |
return &memcg->css; |
6d12e2d8d
|
4786 |
free_out: |
c0ff4b854
|
4787 |
__mem_cgroup_free(memcg); |
04046e1a0
|
4788 |
return ERR_PTR(error); |
8cdea7c05
|
4789 |
} |
761b3ef50
|
4790 |
static int mem_cgroup_pre_destroy(struct cgroup *cont) |
df878fb04
|
4791 |
{ |
c0ff4b854
|
4792 |
struct mem_cgroup *memcg = mem_cgroup_from_cont(cont); |
ec64f5154
|
4793 |
|
c0ff4b854
|
4794 |
return mem_cgroup_force_empty(memcg, false); |
df878fb04
|
4795 |
} |
761b3ef50
|
4796 |
static void mem_cgroup_destroy(struct cgroup *cont) |
8cdea7c05
|
4797 |
{ |
c0ff4b854
|
4798 |
struct mem_cgroup *memcg = mem_cgroup_from_cont(cont); |
c268e9946
|
4799 |
|
1d62e4365
|
4800 |
kmem_cgroup_destroy(memcg); |
d1a4c0b37
|
4801 |
|
c0ff4b854
|
4802 |
mem_cgroup_put(memcg); |
8cdea7c05
|
4803 |
} |
024914477
|
4804 |
#ifdef CONFIG_MMU |
7dc74be03
|
4805 |
/* Handlers for move charge at task migration. */ |
854ffa8d1
|
4806 4807 |
#define PRECHARGE_COUNT_AT_ONCE 256 static int mem_cgroup_do_precharge(unsigned long count) |
7dc74be03
|
4808 |
{ |
854ffa8d1
|
4809 4810 |
int ret = 0; int batch_count = PRECHARGE_COUNT_AT_ONCE; |
c0ff4b854
|
4811 |
struct mem_cgroup *memcg = mc.to; |
4ffef5fef
|
4812 |
|
c0ff4b854
|
4813 |
if (mem_cgroup_is_root(memcg)) { |
854ffa8d1
|
4814 4815 4816 4817 4818 4819 4820 4821 |
mc.precharge += count; /* we don't need css_get for root */ return ret; } /* try to charge at once */ if (count > 1) { struct res_counter *dummy; /* |
c0ff4b854
|
4822 |
* "memcg" cannot be under rmdir() because we've already checked |
854ffa8d1
|
4823 4824 4825 4826 |
* by cgroup_lock_live_cgroup() that it is not removed and we * are still under the same cgroup_mutex. So we can postpone * css_get(). */ |
c0ff4b854
|
4827 |
if (res_counter_charge(&memcg->res, PAGE_SIZE * count, &dummy)) |
854ffa8d1
|
4828 |
goto one_by_one; |
c0ff4b854
|
4829 |
if (do_swap_account && res_counter_charge(&memcg->memsw, |
854ffa8d1
|
4830 |
PAGE_SIZE * count, &dummy)) { |
c0ff4b854
|
4831 |
res_counter_uncharge(&memcg->res, PAGE_SIZE * count); |
854ffa8d1
|
4832 4833 4834 |
goto one_by_one; } mc.precharge += count; |
854ffa8d1
|
4835 4836 4837 4838 4839 4840 4841 4842 4843 4844 4845 4846 4847 |
return ret; } one_by_one: /* fall back to one by one charge */ while (count--) { if (signal_pending(current)) { ret = -EINTR; break; } if (!batch_count--) { batch_count = PRECHARGE_COUNT_AT_ONCE; cond_resched(); } |
c0ff4b854
|
4848 4849 |
ret = __mem_cgroup_try_charge(NULL, GFP_KERNEL, 1, &memcg, false); |
38c5d72f3
|
4850 |
if (ret) |
854ffa8d1
|
4851 |
/* mem_cgroup_clear_mc() will do uncharge later */ |
38c5d72f3
|
4852 |
return ret; |
854ffa8d1
|
4853 4854 |
mc.precharge++; } |
4ffef5fef
|
4855 4856 4857 4858 |
return ret; } /** |
8d32ff844
|
4859 |
* get_mctgt_type - get target type of moving charge |
4ffef5fef
|
4860 4861 4862 |
* @vma: the vma the pte to be checked belongs * @addr: the address corresponding to the pte to be checked * @ptent: the pte to be checked |
024914477
|
4863 |
* @target: the pointer the target page or swap ent will be stored(can be NULL) |
4ffef5fef
|
4864 4865 4866 4867 4868 4869 |
* * Returns * 0(MC_TARGET_NONE): if the pte is not a target for move charge. * 1(MC_TARGET_PAGE): if the page corresponding to this pte is a target for * move charge. if @target is not NULL, the page is stored in target->page * with extra refcnt got(Callers should handle it). |
024914477
|
4870 4871 4872 |
* 2(MC_TARGET_SWAP): if the swap entry corresponding to this pte is a * target for charge migration. if @target is not NULL, the entry is stored * in target->ent. |
4ffef5fef
|
4873 4874 4875 |
* * Called with pte lock held. */ |
4ffef5fef
|
4876 4877 |
union mc_target { struct page *page; |
024914477
|
4878 |
swp_entry_t ent; |
4ffef5fef
|
4879 |
}; |
4ffef5fef
|
4880 |
enum mc_target_type { |
8d32ff844
|
4881 |
MC_TARGET_NONE = 0, |
4ffef5fef
|
4882 |
MC_TARGET_PAGE, |
024914477
|
4883 |
MC_TARGET_SWAP, |
4ffef5fef
|
4884 |
}; |
90254a658
|
4885 4886 |
static struct page *mc_handle_present_pte(struct vm_area_struct *vma, unsigned long addr, pte_t ptent) |
4ffef5fef
|
4887 |
{ |
90254a658
|
4888 |
struct page *page = vm_normal_page(vma, addr, ptent); |
4ffef5fef
|
4889 |
|
90254a658
|
4890 4891 4892 4893 |
if (!page || !page_mapped(page)) return NULL; if (PageAnon(page)) { /* we don't move shared anon */ |
4b91355e9
|
4894 |
if (!move_anon()) |
90254a658
|
4895 |
return NULL; |
87946a722
|
4896 4897 |
} else if (!move_file()) /* we ignore mapcount for file pages */ |
90254a658
|
4898 4899 4900 4901 4902 4903 |
return NULL; if (!get_page_unless_zero(page)) return NULL; return page; } |
4b91355e9
|
4904 |
#ifdef CONFIG_SWAP |
90254a658
|
4905 4906 4907 |
static struct page *mc_handle_swap_pte(struct vm_area_struct *vma, unsigned long addr, pte_t ptent, swp_entry_t *entry) { |
90254a658
|
4908 4909 4910 4911 4912 |
struct page *page = NULL; swp_entry_t ent = pte_to_swp_entry(ptent); if (!move_anon() || non_swap_entry(ent)) return NULL; |
4b91355e9
|
4913 4914 4915 4916 4917 |
/* * Because lookup_swap_cache() updates some statistics counter, * we call find_get_page() with swapper_space directly. */ page = find_get_page(&swapper_space, ent.val); |
90254a658
|
4918 4919 4920 4921 4922 |
if (do_swap_account) entry->val = ent.val; return page; } |
4b91355e9
|
4923 4924 4925 4926 4927 4928 4929 |
#else static struct page *mc_handle_swap_pte(struct vm_area_struct *vma, unsigned long addr, pte_t ptent, swp_entry_t *entry) { return NULL; } #endif |
90254a658
|
4930 |
|
87946a722
|
4931 4932 4933 4934 |
static struct page *mc_handle_file_pte(struct vm_area_struct *vma, unsigned long addr, pte_t ptent, swp_entry_t *entry) { struct page *page = NULL; |
87946a722
|
4935 4936 4937 4938 4939 4940 4941 |
struct address_space *mapping; pgoff_t pgoff; if (!vma->vm_file) /* anonymous vma */ return NULL; if (!move_file()) return NULL; |
87946a722
|
4942 4943 4944 4945 4946 4947 4948 |
mapping = vma->vm_file->f_mapping; if (pte_none(ptent)) pgoff = linear_page_index(vma, addr); else /* pte_file(ptent) is true */ pgoff = pte_to_pgoff(ptent); /* page is moved even if it's not RSS of this task(page-faulted). */ |
aa3b18955
|
4949 4950 4951 4952 4953 4954 |
page = find_get_page(mapping, pgoff); #ifdef CONFIG_SWAP /* shmem/tmpfs may report page out on swap: account for that too. */ if (radix_tree_exceptional_entry(page)) { swp_entry_t swap = radix_to_swp_entry(page); |
87946a722
|
4955 |
if (do_swap_account) |
aa3b18955
|
4956 4957 |
*entry = swap; page = find_get_page(&swapper_space, swap.val); |
87946a722
|
4958 |
} |
aa3b18955
|
4959 |
#endif |
87946a722
|
4960 4961 |
return page; } |
8d32ff844
|
4962 |
static enum mc_target_type get_mctgt_type(struct vm_area_struct *vma, |
90254a658
|
4963 4964 4965 4966 |
unsigned long addr, pte_t ptent, union mc_target *target) { struct page *page = NULL; struct page_cgroup *pc; |
8d32ff844
|
4967 |
enum mc_target_type ret = MC_TARGET_NONE; |
90254a658
|
4968 4969 4970 4971 4972 4973 |
swp_entry_t ent = { .val = 0 }; if (pte_present(ptent)) page = mc_handle_present_pte(vma, addr, ptent); else if (is_swap_pte(ptent)) page = mc_handle_swap_pte(vma, addr, ptent, &ent); |
87946a722
|
4974 4975 |
else if (pte_none(ptent) || pte_file(ptent)) page = mc_handle_file_pte(vma, addr, ptent, &ent); |
90254a658
|
4976 4977 |
if (!page && !ent.val) |
8d32ff844
|
4978 |
return ret; |
024914477
|
4979 4980 4981 4982 4983 4984 4985 4986 4987 4988 4989 4990 4991 4992 4993 |
if (page) { pc = lookup_page_cgroup(page); /* * Do only loose check w/o page_cgroup lock. * mem_cgroup_move_account() checks the pc is valid or not under * the lock. */ if (PageCgroupUsed(pc) && pc->mem_cgroup == mc.from) { ret = MC_TARGET_PAGE; if (target) target->page = page; } if (!ret || !target) put_page(page); } |
90254a658
|
4994 4995 |
/* There is a swap entry and a page doesn't exist or isn't charged */ if (ent.val && !ret && |
9fb4b7cc0
|
4996 |
css_id(&mc.from->css) == lookup_swap_cgroup_id(ent)) { |
7f0f15464
|
4997 4998 4999 |
ret = MC_TARGET_SWAP; if (target) target->ent = ent; |
4ffef5fef
|
5000 |
} |
4ffef5fef
|
5001 5002 |
return ret; } |
12724850e
|
5003 5004 5005 5006 5007 5008 5009 5010 5011 5012 5013 5014 5015 5016 5017 5018 5019 5020 5021 5022 5023 5024 5025 5026 5027 5028 5029 5030 5031 5032 5033 5034 5035 5036 |
#ifdef CONFIG_TRANSPARENT_HUGEPAGE /* * We don't consider swapping or file mapped pages because THP does not * support them for now. * Caller should make sure that pmd_trans_huge(pmd) is true. */ static enum mc_target_type get_mctgt_type_thp(struct vm_area_struct *vma, unsigned long addr, pmd_t pmd, union mc_target *target) { struct page *page = NULL; struct page_cgroup *pc; enum mc_target_type ret = MC_TARGET_NONE; page = pmd_page(pmd); VM_BUG_ON(!page || !PageHead(page)); if (!move_anon()) return ret; pc = lookup_page_cgroup(page); if (PageCgroupUsed(pc) && pc->mem_cgroup == mc.from) { ret = MC_TARGET_PAGE; if (target) { get_page(page); target->page = page; } } return ret; } #else static inline enum mc_target_type get_mctgt_type_thp(struct vm_area_struct *vma, unsigned long addr, pmd_t pmd, union mc_target *target) { return MC_TARGET_NONE; } #endif |
4ffef5fef
|
5037 5038 5039 5040 5041 5042 5043 |
static int mem_cgroup_count_precharge_pte_range(pmd_t *pmd, unsigned long addr, unsigned long end, struct mm_walk *walk) { struct vm_area_struct *vma = walk->private; pte_t *pte; spinlock_t *ptl; |
12724850e
|
5044 5045 5046 5047 |
if (pmd_trans_huge_lock(pmd, vma) == 1) { if (get_mctgt_type_thp(vma, addr, *pmd, NULL) == MC_TARGET_PAGE) mc.precharge += HPAGE_PMD_NR; spin_unlock(&vma->vm_mm->page_table_lock); |
1a5a9906d
|
5048 |
return 0; |
12724850e
|
5049 |
} |
033193275
|
5050 |
|
45f83cefe
|
5051 5052 |
if (pmd_trans_unstable(pmd)) return 0; |
4ffef5fef
|
5053 5054 |
pte = pte_offset_map_lock(vma->vm_mm, pmd, addr, &ptl); for (; addr != end; pte++, addr += PAGE_SIZE) |
8d32ff844
|
5055 |
if (get_mctgt_type(vma, addr, *pte, NULL)) |
4ffef5fef
|
5056 5057 5058 |
mc.precharge++; /* increment precharge temporarily */ pte_unmap_unlock(pte - 1, ptl); cond_resched(); |
7dc74be03
|
5059 5060 |
return 0; } |
4ffef5fef
|
5061 5062 5063 5064 |
static unsigned long mem_cgroup_count_precharge(struct mm_struct *mm) { unsigned long precharge; struct vm_area_struct *vma; |
dfe076b09
|
5065 |
down_read(&mm->mmap_sem); |
4ffef5fef
|
5066 5067 5068 5069 5070 5071 5072 5073 |
for (vma = mm->mmap; vma; vma = vma->vm_next) { struct mm_walk mem_cgroup_count_precharge_walk = { .pmd_entry = mem_cgroup_count_precharge_pte_range, .mm = mm, .private = vma, }; if (is_vm_hugetlb_page(vma)) continue; |
4ffef5fef
|
5074 5075 5076 |
walk_page_range(vma->vm_start, vma->vm_end, &mem_cgroup_count_precharge_walk); } |
dfe076b09
|
5077 |
up_read(&mm->mmap_sem); |
4ffef5fef
|
5078 5079 5080 5081 5082 5083 |
precharge = mc.precharge; mc.precharge = 0; return precharge; } |
4ffef5fef
|
5084 5085 |
static int mem_cgroup_precharge_mc(struct mm_struct *mm) { |
dfe076b09
|
5086 5087 5088 5089 5090 |
unsigned long precharge = mem_cgroup_count_precharge(mm); VM_BUG_ON(mc.moving_task); mc.moving_task = current; return mem_cgroup_do_precharge(precharge); |
4ffef5fef
|
5091 |
} |
dfe076b09
|
5092 5093 |
/* cancels all extra charges on mc.from and mc.to, and wakes up all waiters. */ static void __mem_cgroup_clear_mc(void) |
4ffef5fef
|
5094 |
{ |
2bd9bb206
|
5095 5096 |
struct mem_cgroup *from = mc.from; struct mem_cgroup *to = mc.to; |
4ffef5fef
|
5097 |
/* we must uncharge all the leftover precharges from mc.to */ |
854ffa8d1
|
5098 5099 5100 5101 5102 5103 5104 5105 5106 5107 5108 |
if (mc.precharge) { __mem_cgroup_cancel_charge(mc.to, mc.precharge); mc.precharge = 0; } /* * we didn't uncharge from mc.from at mem_cgroup_move_account(), so * we must uncharge here. */ if (mc.moved_charge) { __mem_cgroup_cancel_charge(mc.from, mc.moved_charge); mc.moved_charge = 0; |
4ffef5fef
|
5109 |
} |
483c30b51
|
5110 5111 |
/* we must fixup refcnts and charges */ if (mc.moved_swap) { |
483c30b51
|
5112 5113 5114 5115 5116 5117 5118 5119 5120 5121 5122 5123 5124 |
/* uncharge swap account from the old cgroup */ if (!mem_cgroup_is_root(mc.from)) res_counter_uncharge(&mc.from->memsw, PAGE_SIZE * mc.moved_swap); __mem_cgroup_put(mc.from, mc.moved_swap); if (!mem_cgroup_is_root(mc.to)) { /* * we charged both to->res and to->memsw, so we should * uncharge to->res. */ res_counter_uncharge(&mc.to->res, PAGE_SIZE * mc.moved_swap); |
483c30b51
|
5125 5126 |
} /* we've already done mem_cgroup_get(mc.to) */ |
483c30b51
|
5127 5128 |
mc.moved_swap = 0; } |
dfe076b09
|
5129 5130 5131 5132 5133 5134 5135 5136 5137 5138 5139 5140 5141 5142 5143 |
memcg_oom_recover(from); memcg_oom_recover(to); wake_up_all(&mc.waitq); } static void mem_cgroup_clear_mc(void) { struct mem_cgroup *from = mc.from; /* * we must clear moving_task before waking up waiters at the end of * task migration. */ mc.moving_task = NULL; __mem_cgroup_clear_mc(); |
2bd9bb206
|
5144 |
spin_lock(&mc.lock); |
4ffef5fef
|
5145 5146 |
mc.from = NULL; mc.to = NULL; |
2bd9bb206
|
5147 |
spin_unlock(&mc.lock); |
32047e2a8
|
5148 |
mem_cgroup_end_move(from); |
4ffef5fef
|
5149 |
} |
761b3ef50
|
5150 5151 |
static int mem_cgroup_can_attach(struct cgroup *cgroup, struct cgroup_taskset *tset) |
7dc74be03
|
5152 |
{ |
2f7ee5691
|
5153 |
struct task_struct *p = cgroup_taskset_first(tset); |
7dc74be03
|
5154 |
int ret = 0; |
c0ff4b854
|
5155 |
struct mem_cgroup *memcg = mem_cgroup_from_cont(cgroup); |
7dc74be03
|
5156 |
|
c0ff4b854
|
5157 |
if (memcg->move_charge_at_immigrate) { |
7dc74be03
|
5158 5159 |
struct mm_struct *mm; struct mem_cgroup *from = mem_cgroup_from_task(p); |
c0ff4b854
|
5160 |
VM_BUG_ON(from == memcg); |
7dc74be03
|
5161 5162 5163 5164 |
mm = get_task_mm(p); if (!mm) return 0; |
7dc74be03
|
5165 |
/* We move charges only when we move a owner of the mm */ |
4ffef5fef
|
5166 5167 5168 5169 |
if (mm->owner == p) { VM_BUG_ON(mc.from); VM_BUG_ON(mc.to); VM_BUG_ON(mc.precharge); |
854ffa8d1
|
5170 |
VM_BUG_ON(mc.moved_charge); |
483c30b51
|
5171 |
VM_BUG_ON(mc.moved_swap); |
32047e2a8
|
5172 |
mem_cgroup_start_move(from); |
2bd9bb206
|
5173 |
spin_lock(&mc.lock); |
4ffef5fef
|
5174 |
mc.from = from; |
c0ff4b854
|
5175 |
mc.to = memcg; |
2bd9bb206
|
5176 |
spin_unlock(&mc.lock); |
dfe076b09
|
5177 |
/* We set mc.moving_task later */ |
4ffef5fef
|
5178 5179 5180 5181 |
ret = mem_cgroup_precharge_mc(mm); if (ret) mem_cgroup_clear_mc(); |
dfe076b09
|
5182 5183 |
} mmput(mm); |
7dc74be03
|
5184 5185 5186 |
} return ret; } |
761b3ef50
|
5187 5188 |
static void mem_cgroup_cancel_attach(struct cgroup *cgroup, struct cgroup_taskset *tset) |
7dc74be03
|
5189 |
{ |
4ffef5fef
|
5190 |
mem_cgroup_clear_mc(); |
7dc74be03
|
5191 |
} |
4ffef5fef
|
5192 5193 5194 |
static int mem_cgroup_move_charge_pte_range(pmd_t *pmd, unsigned long addr, unsigned long end, struct mm_walk *walk) |
7dc74be03
|
5195 |
{ |
4ffef5fef
|
5196 5197 5198 5199 |
int ret = 0; struct vm_area_struct *vma = walk->private; pte_t *pte; spinlock_t *ptl; |
12724850e
|
5200 5201 5202 5203 |
enum mc_target_type target_type; union mc_target target; struct page *page; struct page_cgroup *pc; |
4ffef5fef
|
5204 |
|
12724850e
|
5205 5206 5207 5208 5209 5210 5211 5212 5213 5214 5215 |
/* * We don't take compound_lock() here but no race with splitting thp * happens because: * - if pmd_trans_huge_lock() returns 1, the relevant thp is not * under splitting, which means there's no concurrent thp split, * - if another thread runs into split_huge_page() just after we * entered this if-block, the thread must wait for page table lock * to be unlocked in __split_huge_page_splitting(), where the main * part of thp split is not executed yet. */ if (pmd_trans_huge_lock(pmd, vma) == 1) { |
62ade86ab
|
5216 |
if (mc.precharge < HPAGE_PMD_NR) { |
12724850e
|
5217 5218 5219 5220 5221 5222 5223 5224 5225 |
spin_unlock(&vma->vm_mm->page_table_lock); return 0; } target_type = get_mctgt_type_thp(vma, addr, *pmd, &target); if (target_type == MC_TARGET_PAGE) { page = target.page; if (!isolate_lru_page(page)) { pc = lookup_page_cgroup(page); if (!mem_cgroup_move_account(page, HPAGE_PMD_NR, |
2f3479b14
|
5226 |
pc, mc.from, mc.to)) { |
12724850e
|
5227 5228 5229 5230 5231 5232 5233 5234 |
mc.precharge -= HPAGE_PMD_NR; mc.moved_charge += HPAGE_PMD_NR; } putback_lru_page(page); } put_page(page); } spin_unlock(&vma->vm_mm->page_table_lock); |
1a5a9906d
|
5235 |
return 0; |
12724850e
|
5236 |
} |
45f83cefe
|
5237 5238 |
if (pmd_trans_unstable(pmd)) return 0; |
4ffef5fef
|
5239 5240 5241 5242 |
retry: pte = pte_offset_map_lock(vma->vm_mm, pmd, addr, &ptl); for (; addr != end; addr += PAGE_SIZE) { pte_t ptent = *(pte++); |
024914477
|
5243 |
swp_entry_t ent; |
4ffef5fef
|
5244 5245 5246 |
if (!mc.precharge) break; |
8d32ff844
|
5247 |
switch (get_mctgt_type(vma, addr, ptent, &target)) { |
4ffef5fef
|
5248 5249 5250 5251 5252 |
case MC_TARGET_PAGE: page = target.page; if (isolate_lru_page(page)) goto put; pc = lookup_page_cgroup(page); |
7ec99d621
|
5253 |
if (!mem_cgroup_move_account(page, 1, pc, |
2f3479b14
|
5254 |
mc.from, mc.to)) { |
4ffef5fef
|
5255 |
mc.precharge--; |
854ffa8d1
|
5256 5257 |
/* we uncharge from mc.from later. */ mc.moved_charge++; |
4ffef5fef
|
5258 5259 |
} putback_lru_page(page); |
8d32ff844
|
5260 |
put: /* get_mctgt_type() gets the page */ |
4ffef5fef
|
5261 5262 |
put_page(page); break; |
024914477
|
5263 5264 |
case MC_TARGET_SWAP: ent = target.ent; |
e91cbb425
|
5265 |
if (!mem_cgroup_move_swap_account(ent, mc.from, mc.to)) { |
024914477
|
5266 |
mc.precharge--; |
483c30b51
|
5267 5268 5269 |
/* we fixup refcnts and charges later. */ mc.moved_swap++; } |
024914477
|
5270 |
break; |
4ffef5fef
|
5271 5272 5273 5274 5275 5276 5277 5278 5279 5280 5281 5282 5283 5284 |
default: break; } } pte_unmap_unlock(pte - 1, ptl); cond_resched(); if (addr != end) { /* * We have consumed all precharges we got in can_attach(). * We try charge one by one, but don't do any additional * charges to mc.to if we have failed in charge once in attach() * phase. */ |
854ffa8d1
|
5285 |
ret = mem_cgroup_do_precharge(1); |
4ffef5fef
|
5286 5287 5288 5289 5290 5291 5292 5293 5294 5295 5296 5297 |
if (!ret) goto retry; } return ret; } static void mem_cgroup_move_charge(struct mm_struct *mm) { struct vm_area_struct *vma; lru_add_drain_all(); |
dfe076b09
|
5298 5299 5300 5301 5302 5303 5304 5305 5306 5307 5308 5309 5310 |
retry: if (unlikely(!down_read_trylock(&mm->mmap_sem))) { /* * Someone who are holding the mmap_sem might be waiting in * waitq. So we cancel all extra charges, wake up all waiters, * and retry. Because we cancel precharges, we might not be able * to move enough charges, but moving charge is a best-effort * feature anyway, so it wouldn't be a big problem. */ __mem_cgroup_clear_mc(); cond_resched(); goto retry; } |
4ffef5fef
|
5311 5312 5313 5314 5315 5316 5317 5318 5319 |
for (vma = mm->mmap; vma; vma = vma->vm_next) { int ret; struct mm_walk mem_cgroup_move_charge_walk = { .pmd_entry = mem_cgroup_move_charge_pte_range, .mm = mm, .private = vma, }; if (is_vm_hugetlb_page(vma)) continue; |
4ffef5fef
|
5320 5321 5322 5323 5324 5325 5326 5327 5328 |
ret = walk_page_range(vma->vm_start, vma->vm_end, &mem_cgroup_move_charge_walk); if (ret) /* * means we have consumed all precharges and failed in * doing additional charge. Just abandon here. */ break; } |
dfe076b09
|
5329 |
up_read(&mm->mmap_sem); |
7dc74be03
|
5330 |
} |
761b3ef50
|
5331 5332 |
static void mem_cgroup_move_task(struct cgroup *cont, struct cgroup_taskset *tset) |
67e465a77
|
5333 |
{ |
2f7ee5691
|
5334 |
struct task_struct *p = cgroup_taskset_first(tset); |
a433658c3
|
5335 |
struct mm_struct *mm = get_task_mm(p); |
dfe076b09
|
5336 |
|
dfe076b09
|
5337 |
if (mm) { |
a433658c3
|
5338 5339 |
if (mc.to) mem_cgroup_move_charge(mm); |
dfe076b09
|
5340 5341 |
mmput(mm); } |
a433658c3
|
5342 5343 |
if (mc.to) mem_cgroup_clear_mc(); |
67e465a77
|
5344 |
} |
5cfb80a73
|
5345 |
#else /* !CONFIG_MMU */ |
761b3ef50
|
5346 5347 |
static int mem_cgroup_can_attach(struct cgroup *cgroup, struct cgroup_taskset *tset) |
5cfb80a73
|
5348 5349 5350 |
{ return 0; } |
761b3ef50
|
5351 5352 |
static void mem_cgroup_cancel_attach(struct cgroup *cgroup, struct cgroup_taskset *tset) |
5cfb80a73
|
5353 5354 |
{ } |
761b3ef50
|
5355 5356 |
static void mem_cgroup_move_task(struct cgroup *cont, struct cgroup_taskset *tset) |
5cfb80a73
|
5357 5358 5359 |
{ } #endif |
67e465a77
|
5360 |
|
8cdea7c05
|
5361 5362 5363 5364 |
struct cgroup_subsys mem_cgroup_subsys = { .name = "memory", .subsys_id = mem_cgroup_subsys_id, .create = mem_cgroup_create, |
df878fb04
|
5365 |
.pre_destroy = mem_cgroup_pre_destroy, |
8cdea7c05
|
5366 |
.destroy = mem_cgroup_destroy, |
7dc74be03
|
5367 5368 |
.can_attach = mem_cgroup_can_attach, .cancel_attach = mem_cgroup_cancel_attach, |
67e465a77
|
5369 |
.attach = mem_cgroup_move_task, |
6bc103498
|
5370 |
.base_cftypes = mem_cgroup_files, |
6d12e2d8d
|
5371 |
.early_init = 0, |
04046e1a0
|
5372 |
.use_id = 1, |
48ddbe194
|
5373 |
.__DEPRECATED_clear_css_refs = true, |
8cdea7c05
|
5374 |
}; |
c077719be
|
5375 |
|
c255a4580
|
5376 |
#ifdef CONFIG_MEMCG_SWAP |
a42c390cf
|
5377 5378 5379 |
static int __init enable_swap_account(char *s) { /* consider enabled if no parameter or 1 is given */ |
a2c8990ae
|
5380 |
if (!strcmp(s, "1")) |
a42c390cf
|
5381 |
really_do_swap_account = 1; |
a2c8990ae
|
5382 |
else if (!strcmp(s, "0")) |
a42c390cf
|
5383 5384 5385 |
really_do_swap_account = 0; return 1; } |
a2c8990ae
|
5386 |
__setup("swapaccount=", enable_swap_account); |
c077719be
|
5387 |
|
c077719be
|
5388 |
#endif |