Blame view
mm/vmscan.c
126 KB
b24413180 License cleanup: ... |
1 |
// SPDX-License-Identifier: GPL-2.0 |
1da177e4c Linux-2.6.12-rc2 |
2 3 4 5 6 7 8 9 10 11 12 13 |
/* * linux/mm/vmscan.c * * Copyright (C) 1991, 1992, 1993, 1994 Linus Torvalds * * Swap reorganised 29.12.95, Stephen Tweedie. * kswapd added: 7.1.96 sct * Removed kswapd_ctl limits, and swap out as many pages as needed * to bring the system back to freepages.high: 2.4.97, Rik van Riel. * Zone aware kswapd started 02/00, Kanoj Sarcar (kanoj@sgi.com). * Multiqueue VM started 5.8.00, Rik van Riel. */ |
b1de0d139 mm: convert some ... |
14 |
#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt |
1da177e4c Linux-2.6.12-rc2 |
15 |
#include <linux/mm.h> |
5b3cc15af sched/headers: Pr... |
16 |
#include <linux/sched/mm.h> |
1da177e4c Linux-2.6.12-rc2 |
17 |
#include <linux/module.h> |
5a0e3ad6a include cleanup: ... |
18 |
#include <linux/gfp.h> |
1da177e4c Linux-2.6.12-rc2 |
19 20 21 22 23 |
#include <linux/kernel_stat.h> #include <linux/swap.h> #include <linux/pagemap.h> #include <linux/init.h> #include <linux/highmem.h> |
70ddf637e memcg: add memory... |
24 |
#include <linux/vmpressure.h> |
e129b5c23 [PATCH] vm: add p... |
25 |
#include <linux/vmstat.h> |
1da177e4c Linux-2.6.12-rc2 |
26 27 28 29 30 31 |
#include <linux/file.h> #include <linux/writeback.h> #include <linux/blkdev.h> #include <linux/buffer_head.h> /* for try_to_release_page(), buffer_heads_over_limit */ #include <linux/mm_inline.h> |
1da177e4c Linux-2.6.12-rc2 |
32 33 34 35 36 |
#include <linux/backing-dev.h> #include <linux/rmap.h> #include <linux/topology.h> #include <linux/cpu.h> #include <linux/cpuset.h> |
3e7d34497 mm: vmscan: recla... |
37 |
#include <linux/compaction.h> |
1da177e4c Linux-2.6.12-rc2 |
38 39 |
#include <linux/notifier.h> #include <linux/rwsem.h> |
248a0301e [PATCH] mm: make ... |
40 |
#include <linux/delay.h> |
3218ae14b [PATCH] pgdat all... |
41 |
#include <linux/kthread.h> |
7dfb71030 [PATCH] Add inclu... |
42 |
#include <linux/freezer.h> |
66e1707bc Memory controller... |
43 |
#include <linux/memcontrol.h> |
873b47717 per-task-delay-ac... |
44 |
#include <linux/delayacct.h> |
af936a160 vmscan: unevictab... |
45 |
#include <linux/sysctl.h> |
929bea7c7 vmscan: all_unrec... |
46 |
#include <linux/oom.h> |
64e3d12f7 mm, drm/i915: mar... |
47 |
#include <linux/pagevec.h> |
268bb0ce3 sanitize <linux/p... |
48 |
#include <linux/prefetch.h> |
b1de0d139 mm: convert some ... |
49 |
#include <linux/printk.h> |
f9fe48bec dax: support dirt... |
50 |
#include <linux/dax.h> |
eb414681d psi: pressure sta... |
51 |
#include <linux/psi.h> |
1da177e4c Linux-2.6.12-rc2 |
52 53 54 55 56 |
#include <asm/tlbflush.h> #include <asm/div64.h> #include <linux/swapops.h> |
117aad1e9 mm: avoid reinser... |
57 |
#include <linux/balloon_compaction.h> |
1da177e4c Linux-2.6.12-rc2 |
58 |
|
0f8053a50 [PATCH] mm: make ... |
59 |
#include "internal.h" |
33906bc5c vmscan: tracing: ... |
60 61 |
#define CREATE_TRACE_POINTS #include <trace/events/vmscan.h> |
1da177e4c Linux-2.6.12-rc2 |
62 |
struct scan_control { |
22fba3354 vmscan: separate ... |
63 64 |
/* How many pages shrink_list() should reclaim */ unsigned long nr_to_reclaim; |
ee814fe23 mm: vmscan: clean... |
65 66 67 68 69 |
/* * Nodemask of nodes allowed by the caller. If NULL, all nodes * are scanned. */ nodemask_t *nodemask; |
9e3b2f8cd mm/vmscan: store ... |
70 |
|
5f53e7629 vmscan: page_chec... |
71 |
/* |
f16015fbf mm: vmscan: disti... |
72 73 74 75 |
* The memory cgroup that hit its limit and as a result is the * primary target of this reclaim invocation. */ struct mem_cgroup *target_mem_cgroup; |
66e1707bc Memory controller... |
76 |
|
1276ad68e mm: vmscan: scan ... |
77 |
/* Writepage batching in laptop mode; RECLAIM_WRITE */ |
ee814fe23 mm: vmscan: clean... |
78 79 80 81 82 83 84 |
unsigned int may_writepage:1; /* Can mapped pages be reclaimed? */ unsigned int may_unmap:1; /* Can pages be swapped as part of reclaim? */ unsigned int may_swap:1; |
d6622f636 mm/vmscan: more r... |
85 86 87 88 89 90 91 |
/* * Cgroups are not reclaimed below their configured memory.low, * unless we threaten to OOM. If any cgroups are skipped due to * memory.low and nothing was reclaimed, go back for memory.low. */ unsigned int memcg_low_reclaim:1; unsigned int memcg_low_skipped:1; |
241994ed8 mm: memcontrol: d... |
92 |
|
ee814fe23 mm: vmscan: clean... |
93 94 95 96 |
unsigned int hibernation_mode:1; /* One of the zones is ready for compaction */ unsigned int compaction_ready:1; |
bb451fdf3 mm/vmscan.c: cond... |
97 98 99 100 101 102 103 104 105 106 107 |
/* Allocation order */ s8 order; /* Scan (total_size >> priority) pages at once */ s8 priority; /* The highest zone to isolate pages for reclaim from */ s8 reclaim_idx; /* This context's GFP mask */ gfp_t gfp_mask; |
ee814fe23 mm: vmscan: clean... |
108 109 110 111 112 |
/* Incremented by the number of inactive pages that were scanned */ unsigned long nr_scanned; /* Number of pages freed so far during a call to shrink_zones() */ unsigned long nr_reclaimed; |
d108c7721 mm/vmscan: don't ... |
113 114 115 116 117 118 119 120 121 122 |
struct { unsigned int dirty; unsigned int unqueued_dirty; unsigned int congested; unsigned int writeback; unsigned int immediate; unsigned int file_taken; unsigned int taken; } nr; |
e5ca8071f mm/vmscan.c: add ... |
123 124 125 |
/* for recording the reclaimed slab by now */ struct reclaim_state reclaim_state; |
1da177e4c Linux-2.6.12-rc2 |
126 |
}; |
1da177e4c Linux-2.6.12-rc2 |
127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 |
#ifdef ARCH_HAS_PREFETCH #define prefetch_prev_lru_page(_page, _base, _field) \ do { \ if ((_page)->lru.prev != _base) { \ struct page *prev; \ \ prev = lru_to_page(&(_page->lru)); \ prefetch(&prev->_field); \ } \ } while (0) #else #define prefetch_prev_lru_page(_page, _base, _field) do { } while (0) #endif #ifdef ARCH_HAS_PREFETCHW #define prefetchw_prev_lru_page(_page, _base, _field) \ do { \ if ((_page)->lru.prev != _base) { \ struct page *prev; \ \ prev = lru_to_page(&(_page->lru)); \ prefetchw(&prev->_field); \ } \ } while (0) #else #define prefetchw_prev_lru_page(_page, _base, _field) do { } while (0) #endif /* * From 0 .. 100. Higher means more swappy. */ int vm_swappiness = 60; |
d0480be44 mm: update the de... |
159 160 161 162 163 |
/* * The total number of pages which are beyond the high watermark within all * zones. */ unsigned long vm_total_pages; |
1da177e4c Linux-2.6.12-rc2 |
164 |
|
0a432dcbe mm: shrinker: mak... |
165 166 167 168 169 170 171 172 173 174 175 |
static void set_task_reclaim_state(struct task_struct *task, struct reclaim_state *rs) { /* Check for an overwrite */ WARN_ON_ONCE(rs && task->reclaim_state); /* Check for the nulling of an already-nulled member */ WARN_ON_ONCE(!rs && !task->reclaim_state); task->reclaim_state = rs; } |
1da177e4c Linux-2.6.12-rc2 |
176 177 |
static LIST_HEAD(shrinker_list); static DECLARE_RWSEM(shrinker_rwsem); |
0a432dcbe mm: shrinker: mak... |
178 |
#ifdef CONFIG_MEMCG |
7e010df53 mm: use special v... |
179 180 181 182 183 184 185 186 187 188 189 190 |
/* * We allow subsystems to populate their shrinker-related * LRU lists before register_shrinker_prepared() is called * for the shrinker, since we don't want to impose * restrictions on their internal registration order. * In this case shrink_slab_memcg() may find corresponding * bit is set in the shrinkers map. * * This value is used by the function to detect registering * shrinkers and to skip do_shrink_slab() calls for them. */ #define SHRINKER_REGISTERING ((struct shrinker *)~0UL) |
b4c2b231c mm: assign id to ... |
191 192 193 194 195 196 197 198 199 |
static DEFINE_IDR(shrinker_idr); static int shrinker_nr_max; static int prealloc_memcg_shrinker(struct shrinker *shrinker) { int id, ret = -ENOMEM; down_write(&shrinker_rwsem); /* This may call shrinker, so it must use down_read_trylock() */ |
7e010df53 mm: use special v... |
200 |
id = idr_alloc(&shrinker_idr, SHRINKER_REGISTERING, 0, 0, GFP_KERNEL); |
b4c2b231c mm: assign id to ... |
201 202 |
if (id < 0) goto unlock; |
0a4465d34 mm, memcg: assign... |
203 204 205 206 207 |
if (id >= shrinker_nr_max) { if (memcg_expand_shrinker_maps(id)) { idr_remove(&shrinker_idr, id); goto unlock; } |
b4c2b231c mm: assign id to ... |
208 |
shrinker_nr_max = id + 1; |
0a4465d34 mm, memcg: assign... |
209 |
} |
b4c2b231c mm: assign id to ... |
210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 |
shrinker->id = id; ret = 0; unlock: up_write(&shrinker_rwsem); return ret; } static void unregister_memcg_shrinker(struct shrinker *shrinker) { int id = shrinker->id; BUG_ON(id < 0); down_write(&shrinker_rwsem); idr_remove(&shrinker_idr, id); up_write(&shrinker_rwsem); } |
b4c2b231c mm: assign id to ... |
227 |
|
89b5fae53 mm: vmscan: disti... |
228 229 |
static bool global_reclaim(struct scan_control *sc) { |
f16015fbf mm: vmscan: disti... |
230 |
return !sc->target_mem_cgroup; |
89b5fae53 mm: vmscan: disti... |
231 |
} |
97c9341f7 mm: vmscan: disab... |
232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 |
/** * sane_reclaim - is the usual dirty throttling mechanism operational? * @sc: scan_control in question * * The normal page dirty throttling mechanism in balance_dirty_pages() is * completely broken with the legacy memcg and direct stalling in * shrink_page_list() is used for throttling instead, which lacks all the * niceties such as fairness, adaptive pausing, bandwidth proportional * allocation and configurability. * * This function tests whether the vmscan currently in progress can assume * that the normal dirty throttling mechanism is operational. */ static bool sane_reclaim(struct scan_control *sc) { struct mem_cgroup *memcg = sc->target_mem_cgroup; if (!memcg) return true; #ifdef CONFIG_CGROUP_WRITEBACK |
69234acee Merge branch 'for... |
253 |
if (cgroup_subsys_on_dfl(memory_cgrp_subsys)) |
97c9341f7 mm: vmscan: disab... |
254 255 256 257 |
return true; #endif return false; } |
e3c1ac586 mm/vmscan: don't ... |
258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 |
static void set_memcg_congestion(pg_data_t *pgdat, struct mem_cgroup *memcg, bool congested) { struct mem_cgroup_per_node *mn; if (!memcg) return; mn = mem_cgroup_nodeinfo(memcg, pgdat->node_id); WRITE_ONCE(mn->congested, congested); } static bool memcg_congested(pg_data_t *pgdat, struct mem_cgroup *memcg) { struct mem_cgroup_per_node *mn; mn = mem_cgroup_nodeinfo(memcg, pgdat->node_id); return READ_ONCE(mn->congested); } |
91a45470f per-zone and recl... |
281 |
#else |
0a432dcbe mm: shrinker: mak... |
282 283 284 285 286 287 288 289 |
static int prealloc_memcg_shrinker(struct shrinker *shrinker) { return 0; } static void unregister_memcg_shrinker(struct shrinker *shrinker) { } |
89b5fae53 mm: vmscan: disti... |
290 291 292 293 |
static bool global_reclaim(struct scan_control *sc) { return true; } |
97c9341f7 mm: vmscan: disab... |
294 295 296 297 298 |
static bool sane_reclaim(struct scan_control *sc) { return true; } |
e3c1ac586 mm/vmscan: don't ... |
299 300 301 302 303 304 305 306 307 308 309 310 |
static inline void set_memcg_congestion(struct pglist_data *pgdat, struct mem_cgroup *memcg, bool congested) { } static inline bool memcg_congested(struct pglist_data *pgdat, struct mem_cgroup *memcg) { return false; } |
91a45470f per-zone and recl... |
311 |
#endif |
5a1c84b40 mm: remove reclai... |
312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 |
/* * This misses isolated pages which are not accounted for to save counters. * As the data only determines if reclaim or compaction continues, it is * not expected that isolated pages will be a dominating factor. */ unsigned long zone_reclaimable_pages(struct zone *zone) { unsigned long nr; nr = zone_page_state_snapshot(zone, NR_ZONE_INACTIVE_FILE) + zone_page_state_snapshot(zone, NR_ZONE_ACTIVE_FILE); if (get_nr_swap_pages() > 0) nr += zone_page_state_snapshot(zone, NR_ZONE_INACTIVE_ANON) + zone_page_state_snapshot(zone, NR_ZONE_ACTIVE_ANON); return nr; } |
fd5388037 mm, vmscan: clean... |
329 330 331 332 333 334 335 |
/** * lruvec_lru_size - Returns the number of pages on the given LRU list. * @lruvec: lru vector * @lru: lru to use * @zone_idx: zones to consider (use MAX_NR_ZONES for the whole LRU list) */ unsigned long lruvec_lru_size(struct lruvec *lruvec, enum lru_list lru, int zone_idx) |
c9f299d98 mm: add zone nr_p... |
336 |
{ |
b11edebbc mm: memcg: get nu... |
337 |
unsigned long lru_size = 0; |
fd5388037 mm, vmscan: clean... |
338 |
int zid; |
b11edebbc mm: memcg: get nu... |
339 340 341 342 |
if (!mem_cgroup_disabled()) { for (zid = 0; zid < MAX_NR_ZONES; zid++) lru_size += mem_cgroup_get_zone_lru_size(lruvec, lru, zid); } else |
fd5388037 mm, vmscan: clean... |
343 |
lru_size = node_page_state(lruvec_pgdat(lruvec), NR_LRU_BASE + lru); |
a3d8e0549 memcg: add mem_cg... |
344 |
|
fd5388037 mm, vmscan: clean... |
345 346 347 |
for (zid = zone_idx + 1; zid < MAX_NR_ZONES; zid++) { struct zone *zone = &lruvec_pgdat(lruvec)->node_zones[zid]; unsigned long size; |
c9f299d98 mm: add zone nr_p... |
348 |
|
fd5388037 mm, vmscan: clean... |
349 350 351 352 353 354 355 356 357 358 359 360 |
if (!managed_zone(zone)) continue; if (!mem_cgroup_disabled()) size = mem_cgroup_get_zone_lru_size(lruvec, lru, zid); else size = zone_page_state(&lruvec_pgdat(lruvec)->node_zones[zid], NR_ZONE_LRU_BASE + lru); lru_size -= min(size, lru_size); } return lru_size; |
b4536f0c8 mm, memcg: fix th... |
361 |
|
b4536f0c8 mm, memcg: fix th... |
362 |
} |
1da177e4c Linux-2.6.12-rc2 |
363 |
/* |
1d3d4437e vmscan: per-node ... |
364 |
* Add a shrinker callback to be called from the vm. |
1da177e4c Linux-2.6.12-rc2 |
365 |
*/ |
8e04944f0 mm,vmscan: Allow ... |
366 |
int prealloc_shrinker(struct shrinker *shrinker) |
1da177e4c Linux-2.6.12-rc2 |
367 |
{ |
b9726c26d numa: make "nr_no... |
368 |
unsigned int size = sizeof(*shrinker->nr_deferred); |
1d3d4437e vmscan: per-node ... |
369 |
|
1d3d4437e vmscan: per-node ... |
370 371 372 373 374 375 |
if (shrinker->flags & SHRINKER_NUMA_AWARE) size *= nr_node_ids; shrinker->nr_deferred = kzalloc(size, GFP_KERNEL); if (!shrinker->nr_deferred) return -ENOMEM; |
b4c2b231c mm: assign id to ... |
376 377 378 379 380 |
if (shrinker->flags & SHRINKER_MEMCG_AWARE) { if (prealloc_memcg_shrinker(shrinker)) goto free_deferred; } |
8e04944f0 mm,vmscan: Allow ... |
381 |
return 0; |
b4c2b231c mm: assign id to ... |
382 383 384 385 386 |
free_deferred: kfree(shrinker->nr_deferred); shrinker->nr_deferred = NULL; return -ENOMEM; |
8e04944f0 mm,vmscan: Allow ... |
387 388 389 390 |
} void free_prealloced_shrinker(struct shrinker *shrinker) { |
b4c2b231c mm: assign id to ... |
391 392 393 394 395 |
if (!shrinker->nr_deferred) return; if (shrinker->flags & SHRINKER_MEMCG_AWARE) unregister_memcg_shrinker(shrinker); |
8e04944f0 mm,vmscan: Allow ... |
396 397 398 |
kfree(shrinker->nr_deferred); shrinker->nr_deferred = NULL; } |
1d3d4437e vmscan: per-node ... |
399 |
|
8e04944f0 mm,vmscan: Allow ... |
400 401 |
void register_shrinker_prepared(struct shrinker *shrinker) { |
8e1f936b7 mm: clean up and ... |
402 403 |
down_write(&shrinker_rwsem); list_add_tail(&shrinker->list, &shrinker_list); |
b58b8e933 mm: vmscan: prote... |
404 |
#ifdef CONFIG_MEMCG |
8df4a44cc mm: check shrinke... |
405 406 |
if (shrinker->flags & SHRINKER_MEMCG_AWARE) idr_replace(&shrinker_idr, shrinker, shrinker->id); |
7e010df53 mm: use special v... |
407 |
#endif |
8e1f936b7 mm: clean up and ... |
408 |
up_write(&shrinker_rwsem); |
8e04944f0 mm,vmscan: Allow ... |
409 410 411 412 413 414 415 416 417 |
} int register_shrinker(struct shrinker *shrinker) { int err = prealloc_shrinker(shrinker); if (err) return err; register_shrinker_prepared(shrinker); |
1d3d4437e vmscan: per-node ... |
418 |
return 0; |
1da177e4c Linux-2.6.12-rc2 |
419 |
} |
8e1f936b7 mm: clean up and ... |
420 |
EXPORT_SYMBOL(register_shrinker); |
1da177e4c Linux-2.6.12-rc2 |
421 422 423 424 |
/* * Remove one */ |
8e1f936b7 mm: clean up and ... |
425 |
void unregister_shrinker(struct shrinker *shrinker) |
1da177e4c Linux-2.6.12-rc2 |
426 |
{ |
bb422a738 mm,vmscan: Make u... |
427 428 |
if (!shrinker->nr_deferred) return; |
b4c2b231c mm: assign id to ... |
429 430 |
if (shrinker->flags & SHRINKER_MEMCG_AWARE) unregister_memcg_shrinker(shrinker); |
1da177e4c Linux-2.6.12-rc2 |
431 432 433 |
down_write(&shrinker_rwsem); list_del(&shrinker->list); up_write(&shrinker_rwsem); |
ae3933216 mm/vmscan.c: don'... |
434 |
kfree(shrinker->nr_deferred); |
bb422a738 mm,vmscan: Make u... |
435 |
shrinker->nr_deferred = NULL; |
1da177e4c Linux-2.6.12-rc2 |
436 |
} |
8e1f936b7 mm: clean up and ... |
437 |
EXPORT_SYMBOL(unregister_shrinker); |
1da177e4c Linux-2.6.12-rc2 |
438 439 |
#define SHRINK_BATCH 128 |
1d3d4437e vmscan: per-node ... |
440 |
|
cb731d6c6 vmscan: per memor... |
441 |
static unsigned long do_shrink_slab(struct shrink_control *shrinkctl, |
9092c71bb mm: use sc->prior... |
442 |
struct shrinker *shrinker, int priority) |
1d3d4437e vmscan: per-node ... |
443 444 445 446 |
{ unsigned long freed = 0; unsigned long long delta; long total_scan; |
d5bc5fd3f mm: vmscan: shrin... |
447 |
long freeable; |
1d3d4437e vmscan: per-node ... |
448 449 450 451 452 |
long nr; long new_nr; int nid = shrinkctl->nid; long batch_size = shrinker->batch ? shrinker->batch : SHRINK_BATCH; |
5f33a0803 mm/vmscan.c: set ... |
453 |
long scanned = 0, next_deferred; |
1d3d4437e vmscan: per-node ... |
454 |
|
ac7fb3ad2 mm/vmscan.c: move... |
455 456 |
if (!(shrinker->flags & SHRINKER_NUMA_AWARE)) nid = 0; |
d5bc5fd3f mm: vmscan: shrin... |
457 |
freeable = shrinker->count_objects(shrinker, shrinkctl); |
9b996468c mm: add SHRINK_EM... |
458 459 |
if (freeable == 0 || freeable == SHRINK_EMPTY) return freeable; |
1d3d4437e vmscan: per-node ... |
460 461 462 463 464 465 466 467 468 |
/* * copy the current shrinker scan count into a local variable * and zero it so that other concurrent shrinker invocations * don't also do this scanning work. */ nr = atomic_long_xchg(&shrinker->nr_deferred[nid], 0); total_scan = nr; |
4b85afbda mm: zero-seek shr... |
469 470 471 472 473 474 475 476 477 478 479 480 |
if (shrinker->seeks) { delta = freeable >> priority; delta *= 4; do_div(delta, shrinker->seeks); } else { /* * These objects don't require any IO to create. Trim * them aggressively under memory pressure to keep * them from causing refetches in the IO caches. */ delta = freeable / 2; } |
172b06c32 mm: slowly shrink... |
481 |
|
1d3d4437e vmscan: per-node ... |
482 483 |
total_scan += delta; if (total_scan < 0) { |
d75f773c8 treewide: Switch ... |
484 485 |
pr_err("shrink_slab: %pS negative objects to delete nr=%ld ", |
a0b02131c shrinker: Kill ol... |
486 |
shrinker->scan_objects, total_scan); |
d5bc5fd3f mm: vmscan: shrin... |
487 |
total_scan = freeable; |
5f33a0803 mm/vmscan.c: set ... |
488 489 490 |
next_deferred = nr; } else next_deferred = total_scan; |
1d3d4437e vmscan: per-node ... |
491 492 493 494 495 496 497 |
/* * We need to avoid excessive windup on filesystem shrinkers * due to large numbers of GFP_NOFS allocations causing the * shrinkers to return -1 all the time. This results in a large * nr being built up so when a shrink that can do some work * comes along it empties the entire cache due to nr >>> |
d5bc5fd3f mm: vmscan: shrin... |
498 |
* freeable. This is bad for sustaining a working set in |
1d3d4437e vmscan: per-node ... |
499 500 501 502 503 |
* memory. * * Hence only allow the shrinker to scan the entire cache when * a large delta change is calculated directly. */ |
d5bc5fd3f mm: vmscan: shrin... |
504 505 |
if (delta < freeable / 4) total_scan = min(total_scan, freeable / 2); |
1d3d4437e vmscan: per-node ... |
506 507 508 509 510 511 |
/* * Avoid risking looping forever due to too large nr value: * never try to free more than twice the estimate number of * freeable entries. */ |
d5bc5fd3f mm: vmscan: shrin... |
512 513 |
if (total_scan > freeable * 2) total_scan = freeable * 2; |
1d3d4437e vmscan: per-node ... |
514 515 |
trace_mm_shrink_slab_start(shrinker, shrinkctl, nr, |
9092c71bb mm: use sc->prior... |
516 |
freeable, delta, total_scan, priority); |
1d3d4437e vmscan: per-node ... |
517 |
|
0b1fb40a3 mm: vmscan: shrin... |
518 519 520 521 522 523 524 525 526 527 528 |
/* * Normally, we should not scan less than batch_size objects in one * pass to avoid too frequent shrinker calls, but if the slab has less * than batch_size objects in total and we are really tight on memory, * we will try to reclaim all available objects, otherwise we can end * up failing allocations although there are plenty of reclaimable * objects spread over several slabs with usage less than the * batch_size. * * We detect the "tight on memory" situations by looking at the total * number of objects we want to scan (total_scan). If it is greater |
d5bc5fd3f mm: vmscan: shrin... |
529 |
* than the total number of objects on slab (freeable), we must be |
0b1fb40a3 mm: vmscan: shrin... |
530 531 532 533 |
* scanning at high prio and therefore should try to reclaim as much as * possible. */ while (total_scan >= batch_size || |
d5bc5fd3f mm: vmscan: shrin... |
534 |
total_scan >= freeable) { |
a0b02131c shrinker: Kill ol... |
535 |
unsigned long ret; |
0b1fb40a3 mm: vmscan: shrin... |
536 |
unsigned long nr_to_scan = min(batch_size, total_scan); |
1d3d4437e vmscan: per-node ... |
537 |
|
0b1fb40a3 mm: vmscan: shrin... |
538 |
shrinkctl->nr_to_scan = nr_to_scan; |
d460acb5b mm: track actual ... |
539 |
shrinkctl->nr_scanned = nr_to_scan; |
a0b02131c shrinker: Kill ol... |
540 541 542 543 |
ret = shrinker->scan_objects(shrinker, shrinkctl); if (ret == SHRINK_STOP) break; freed += ret; |
1d3d4437e vmscan: per-node ... |
544 |
|
d460acb5b mm: track actual ... |
545 546 547 |
count_vm_events(SLABS_SCANNED, shrinkctl->nr_scanned); total_scan -= shrinkctl->nr_scanned; scanned += shrinkctl->nr_scanned; |
1d3d4437e vmscan: per-node ... |
548 549 550 |
cond_resched(); } |
5f33a0803 mm/vmscan.c: set ... |
551 552 553 554 |
if (next_deferred >= scanned) next_deferred -= scanned; else next_deferred = 0; |
1d3d4437e vmscan: per-node ... |
555 556 557 558 559 |
/* * move the unused scan count back into the shrinker in a * manner that handles concurrent updates. If we exhausted the * scan, there is no need to do an update. */ |
5f33a0803 mm/vmscan.c: set ... |
560 561 |
if (next_deferred > 0) new_nr = atomic_long_add_return(next_deferred, |
1d3d4437e vmscan: per-node ... |
562 563 564 |
&shrinker->nr_deferred[nid]); else new_nr = atomic_long_read(&shrinker->nr_deferred[nid]); |
df9024a8c mm: shrinker: add... |
565 |
trace_mm_shrink_slab_end(shrinker, nid, freed, nr, new_nr, total_scan); |
1d3d4437e vmscan: per-node ... |
566 |
return freed; |
1495f230f vmscan: change sh... |
567 |
} |
0a432dcbe mm: shrinker: mak... |
568 |
#ifdef CONFIG_MEMCG |
b0dedc49a mm/vmscan.c: iter... |
569 570 571 572 |
static unsigned long shrink_slab_memcg(gfp_t gfp_mask, int nid, struct mem_cgroup *memcg, int priority) { struct memcg_shrinker_map *map; |
b8e57efa2 mm/vmscan.c: fix ... |
573 574 |
unsigned long ret, freed = 0; int i; |
b0dedc49a mm/vmscan.c: iter... |
575 |
|
0a432dcbe mm: shrinker: mak... |
576 |
if (!mem_cgroup_online(memcg)) |
b0dedc49a mm/vmscan.c: iter... |
577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 |
return 0; if (!down_read_trylock(&shrinker_rwsem)) return 0; map = rcu_dereference_protected(memcg->nodeinfo[nid]->shrinker_map, true); if (unlikely(!map)) goto unlock; for_each_set_bit(i, map->map, shrinker_nr_max) { struct shrink_control sc = { .gfp_mask = gfp_mask, .nid = nid, .memcg = memcg, }; struct shrinker *shrinker; shrinker = idr_find(&shrinker_idr, i); |
7e010df53 mm: use special v... |
596 597 598 |
if (unlikely(!shrinker || shrinker == SHRINKER_REGISTERING)) { if (!shrinker) clear_bit(i, map->map); |
b0dedc49a mm/vmscan.c: iter... |
599 600 |
continue; } |
0a432dcbe mm: shrinker: mak... |
601 602 603 604 |
/* Call non-slab shrinkers even though kmem is disabled */ if (!memcg_kmem_enabled() && !(shrinker->flags & SHRINKER_NONSLAB)) continue; |
b0dedc49a mm/vmscan.c: iter... |
605 |
ret = do_shrink_slab(&sc, shrinker, priority); |
f90280d6b mm/vmscan.c: clea... |
606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 |
if (ret == SHRINK_EMPTY) { clear_bit(i, map->map); /* * After the shrinker reported that it had no objects to * free, but before we cleared the corresponding bit in * the memcg shrinker map, a new object might have been * added. To make sure, we have the bit set in this * case, we invoke the shrinker one more time and reset * the bit if it reports that it is not empty anymore. * The memory barrier here pairs with the barrier in * memcg_set_shrinker_bit(): * * list_lru_add() shrink_slab_memcg() * list_add_tail() clear_bit() * <MB> <MB> * set_bit() do_shrink_slab() */ smp_mb__after_atomic(); ret = do_shrink_slab(&sc, shrinker, priority); if (ret == SHRINK_EMPTY) ret = 0; else memcg_set_shrinker_bit(memcg, nid, i); } |
b0dedc49a mm/vmscan.c: iter... |
630 631 632 633 634 635 636 637 638 639 640 |
freed += ret; if (rwsem_is_contended(&shrinker_rwsem)) { freed = freed ? : 1; break; } } unlock: up_read(&shrinker_rwsem); return freed; } |
0a432dcbe mm: shrinker: mak... |
641 |
#else /* CONFIG_MEMCG */ |
b0dedc49a mm/vmscan.c: iter... |
642 643 644 645 646 |
static unsigned long shrink_slab_memcg(gfp_t gfp_mask, int nid, struct mem_cgroup *memcg, int priority) { return 0; } |
0a432dcbe mm: shrinker: mak... |
647 |
#endif /* CONFIG_MEMCG */ |
b0dedc49a mm/vmscan.c: iter... |
648 |
|
6b4f7799c mm: vmscan: invok... |
649 |
/** |
cb731d6c6 vmscan: per memor... |
650 |
* shrink_slab - shrink slab caches |
6b4f7799c mm: vmscan: invok... |
651 652 |
* @gfp_mask: allocation context * @nid: node whose slab caches to target |
cb731d6c6 vmscan: per memor... |
653 |
* @memcg: memory cgroup whose slab caches to target |
9092c71bb mm: use sc->prior... |
654 |
* @priority: the reclaim priority |
1da177e4c Linux-2.6.12-rc2 |
655 |
* |
6b4f7799c mm: vmscan: invok... |
656 |
* Call the shrink functions to age shrinkable caches. |
1da177e4c Linux-2.6.12-rc2 |
657 |
* |
6b4f7799c mm: vmscan: invok... |
658 659 |
* @nid is passed along to shrinkers with SHRINKER_NUMA_AWARE set, * unaware shrinkers will receive a node id of 0 instead. |
1da177e4c Linux-2.6.12-rc2 |
660 |
* |
aeed1d325 mm/vmscan.c: gene... |
661 662 |
* @memcg specifies the memory cgroup to target. Unaware shrinkers * are called only if it is the root cgroup. |
cb731d6c6 vmscan: per memor... |
663 |
* |
9092c71bb mm: use sc->prior... |
664 665 |
* @priority is sc->priority, we take the number of objects and >> by priority * in order to get the scan target. |
b15e0905f [PATCH] vmscan: n... |
666 |
* |
6b4f7799c mm: vmscan: invok... |
667 |
* Returns the number of reclaimed slab objects. |
1da177e4c Linux-2.6.12-rc2 |
668 |
*/ |
cb731d6c6 vmscan: per memor... |
669 670 |
static unsigned long shrink_slab(gfp_t gfp_mask, int nid, struct mem_cgroup *memcg, |
9092c71bb mm: use sc->prior... |
671 |
int priority) |
1da177e4c Linux-2.6.12-rc2 |
672 |
{ |
b8e57efa2 mm/vmscan.c: fix ... |
673 |
unsigned long ret, freed = 0; |
1da177e4c Linux-2.6.12-rc2 |
674 |
struct shrinker *shrinker; |
fa1e512fa mm: vmscan: check... |
675 676 677 678 679 680 681 682 |
/* * The root memcg might be allocated even though memcg is disabled * via "cgroup_disable=memory" boot parameter. This could make * mem_cgroup_is_root() return false, then just run memcg slab * shrink, but skip global shrink. This may result in premature * oom. */ if (!mem_cgroup_disabled() && !mem_cgroup_is_root(memcg)) |
b0dedc49a mm/vmscan.c: iter... |
683 |
return shrink_slab_memcg(gfp_mask, nid, memcg, priority); |
cb731d6c6 vmscan: per memor... |
684 |
|
e830c63a6 mm,vmscan: don't ... |
685 |
if (!down_read_trylock(&shrinker_rwsem)) |
f06590bd7 mm: vmscan: corre... |
686 |
goto out; |
1da177e4c Linux-2.6.12-rc2 |
687 688 |
list_for_each_entry(shrinker, &shrinker_list, list) { |
6b4f7799c mm: vmscan: invok... |
689 690 691 |
struct shrink_control sc = { .gfp_mask = gfp_mask, .nid = nid, |
cb731d6c6 vmscan: per memor... |
692 |
.memcg = memcg, |
6b4f7799c mm: vmscan: invok... |
693 |
}; |
ec97097bc mm: vmscan: call ... |
694 |
|
9b996468c mm: add SHRINK_EM... |
695 696 697 698 |
ret = do_shrink_slab(&sc, shrinker, priority); if (ret == SHRINK_EMPTY) ret = 0; freed += ret; |
e496612c5 mm: do not stall ... |
699 700 701 702 703 704 705 706 707 |
/* * Bail out if someone want to register a new shrinker to * prevent the regsitration from being stalled for long periods * by parallel ongoing shrinking. */ if (rwsem_is_contended(&shrinker_rwsem)) { freed = freed ? : 1; break; } |
1da177e4c Linux-2.6.12-rc2 |
708 |
} |
6b4f7799c mm: vmscan: invok... |
709 |
|
1da177e4c Linux-2.6.12-rc2 |
710 |
up_read(&shrinker_rwsem); |
f06590bd7 mm: vmscan: corre... |
711 712 |
out: cond_resched(); |
24f7c6b98 mm: new shrinker API |
713 |
return freed; |
1da177e4c Linux-2.6.12-rc2 |
714 |
} |
cb731d6c6 vmscan: per memor... |
715 716 717 718 719 720 721 722 |
void drop_slab_node(int nid) { unsigned long freed; do { struct mem_cgroup *memcg = NULL; freed = 0; |
aeed1d325 mm/vmscan.c: gene... |
723 |
memcg = mem_cgroup_iter(NULL, NULL, NULL); |
cb731d6c6 vmscan: per memor... |
724 |
do { |
9092c71bb mm: use sc->prior... |
725 |
freed += shrink_slab(GFP_KERNEL, nid, memcg, 0); |
cb731d6c6 vmscan: per memor... |
726 727 728 729 730 731 732 733 734 735 736 |
} while ((memcg = mem_cgroup_iter(NULL, memcg, NULL)) != NULL); } while (freed > 10); } void drop_slab(void) { int nid; for_each_online_node(nid) drop_slab_node(nid); } |
1da177e4c Linux-2.6.12-rc2 |
737 738 |
static inline int is_page_cache_freeable(struct page *page) { |
ceddc3a52 mm: document is_p... |
739 740 |
/* * A freeable page cache page is referenced only by the caller |
67891ffff mm: Convert is_pa... |
741 742 |
* that isolated the page, the page cache and optional buffer * heads at page->private. |
ceddc3a52 mm: document is_p... |
743 |
*/ |
67891ffff mm: Convert is_pa... |
744 |
int page_cache_pins = PageTransHuge(page) && PageSwapCache(page) ? |
bd4c82c22 mm, THP, swap: de... |
745 |
HPAGE_PMD_NR : 1; |
67891ffff mm: Convert is_pa... |
746 |
return page_count(page) - page_has_private(page) == 1 + page_cache_pins; |
1da177e4c Linux-2.6.12-rc2 |
747 |
} |
703c27088 writeback: implem... |
748 |
static int may_write_to_inode(struct inode *inode, struct scan_control *sc) |
1da177e4c Linux-2.6.12-rc2 |
749 |
{ |
930d91525 [PATCH] Swap Migr... |
750 |
if (current->flags & PF_SWAPWRITE) |
1da177e4c Linux-2.6.12-rc2 |
751 |
return 1; |
703c27088 writeback: implem... |
752 |
if (!inode_write_congested(inode)) |
1da177e4c Linux-2.6.12-rc2 |
753 |
return 1; |
703c27088 writeback: implem... |
754 |
if (inode_to_bdi(inode) == current->backing_dev_info) |
1da177e4c Linux-2.6.12-rc2 |
755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 |
return 1; return 0; } /* * We detected a synchronous write error writing a page out. Probably * -ENOSPC. We need to propagate that into the address_space for a subsequent * fsync(), msync() or close(). * * The tricky part is that after writepage we cannot touch the mapping: nothing * prevents it from being freed up. But we have a ref on the page and once * that page is locked, the mapping is pinned. * * We're allowed to run sleeping lock_page() here because we know the caller has * __GFP_FS. */ static void handle_write_error(struct address_space *mapping, struct page *page, int error) { |
7eaceacca block: remove per... |
774 |
lock_page(page); |
3e9f45bd1 Factor outstandin... |
775 776 |
if (page_mapping(page) == mapping) mapping_set_error(mapping, error); |
1da177e4c Linux-2.6.12-rc2 |
777 778 |
unlock_page(page); } |
04e62a29b [PATCH] More page... |
779 780 781 782 783 784 785 786 787 788 789 |
/* possible outcome of pageout() */ typedef enum { /* failed to write page out, page is locked */ PAGE_KEEP, /* move page to the active list, page is locked */ PAGE_ACTIVATE, /* page has been sent to the disk successfully, page is unlocked */ PAGE_SUCCESS, /* page is clean and locked */ PAGE_CLEAN, } pageout_t; |
1da177e4c Linux-2.6.12-rc2 |
790 |
/* |
1742f19fa [PATCH] vmscan: r... |
791 792 |
* pageout is called by shrink_page_list() for each dirty page. * Calls ->writepage(). |
1da177e4c Linux-2.6.12-rc2 |
793 |
*/ |
c661b078f synchronous lumpy... |
794 |
static pageout_t pageout(struct page *page, struct address_space *mapping, |
7d3579e8e vmscan: narrow th... |
795 |
struct scan_control *sc) |
1da177e4c Linux-2.6.12-rc2 |
796 797 798 799 800 801 802 803 |
{ /* * If the page is dirty, only perform writeback if that write * will be non-blocking. To prevent this allocation from being * stalled by pagecache activity. But note that there may be * stalls if we need to run get_block(). We could test * PagePrivate for that. * |
8174202b3 write_iter varian... |
804 |
* If this process is currently in __generic_file_write_iter() against |
1da177e4c Linux-2.6.12-rc2 |
805 806 807 808 809 810 811 |
* this page's queue, we can perform writeback even if that * will block. * * If the page is swapcache, write it back even if that would * block, for some throttling. This happens by accident, because * swap_backing_dev_info is bust: it doesn't reflect the * congestion state of the swapdevs. Easy to fix, if needed. |
1da177e4c Linux-2.6.12-rc2 |
812 813 814 815 816 817 818 819 |
*/ if (!is_page_cache_freeable(page)) return PAGE_KEEP; if (!mapping) { /* * Some data journaling orphaned pages can have * page->mapping == NULL while being dirty with clean buffers. */ |
266cf658e FS-Cache: Recruit... |
820 |
if (page_has_private(page)) { |
1da177e4c Linux-2.6.12-rc2 |
821 822 |
if (try_to_free_buffers(page)) { ClearPageDirty(page); |
b1de0d139 mm: convert some ... |
823 824 |
pr_info("%s: orphaned page ", __func__); |
1da177e4c Linux-2.6.12-rc2 |
825 826 827 828 829 830 831 |
return PAGE_CLEAN; } } return PAGE_KEEP; } if (mapping->a_ops->writepage == NULL) return PAGE_ACTIVATE; |
703c27088 writeback: implem... |
832 |
if (!may_write_to_inode(mapping->host, sc)) |
1da177e4c Linux-2.6.12-rc2 |
833 834 835 836 837 838 839 |
return PAGE_KEEP; if (clear_page_dirty_for_io(page)) { int res; struct writeback_control wbc = { .sync_mode = WB_SYNC_NONE, .nr_to_write = SWAP_CLUSTER_MAX, |
111ebb6e6 [PATCH] writeback... |
840 841 |
.range_start = 0, .range_end = LLONG_MAX, |
1da177e4c Linux-2.6.12-rc2 |
842 843 844 845 846 847 848 |
.for_reclaim = 1, }; SetPageReclaim(page); res = mapping->a_ops->writepage(page, &wbc); if (res < 0) handle_write_error(mapping, page, res); |
994fc28c7 [PATCH] add AOP_T... |
849 |
if (res == AOP_WRITEPAGE_ACTIVATE) { |
1da177e4c Linux-2.6.12-rc2 |
850 851 852 |
ClearPageReclaim(page); return PAGE_ACTIVATE; } |
c661b078f synchronous lumpy... |
853 |
|
1da177e4c Linux-2.6.12-rc2 |
854 855 856 857 |
if (!PageWriteback(page)) { /* synchronous write or broken a_ops? */ ClearPageReclaim(page); } |
3aa238511 mm/vmscan.c: chan... |
858 |
trace_mm_vmscan_writepage(page); |
c4a25635b mm: move vmscan w... |
859 |
inc_node_page_state(page, NR_VMSCAN_WRITE); |
1da177e4c Linux-2.6.12-rc2 |
860 861 862 863 864 |
return PAGE_SUCCESS; } return PAGE_CLEAN; } |
a649fd927 [PATCH] invalidat... |
865 |
/* |
e286781d5 mm: speculative p... |
866 867 |
* Same as remove_mapping, but if the page is removed from the mapping, it * gets returned with a refcount of 0. |
a649fd927 [PATCH] invalidat... |
868 |
*/ |
a528910e1 mm: thrash detect... |
869 870 |
static int __remove_mapping(struct address_space *mapping, struct page *page, bool reclaimed) |
49d2e9cc4 [PATCH] Swap Migr... |
871 |
{ |
c4843a759 memcg: add per cg... |
872 |
unsigned long flags; |
bd4c82c22 mm, THP, swap: de... |
873 |
int refcount; |
c4843a759 memcg: add per cg... |
874 |
|
28e4d965e [PATCH] mm: remov... |
875 876 |
BUG_ON(!PageLocked(page)); BUG_ON(mapping != page_mapping(page)); |
49d2e9cc4 [PATCH] Swap Migr... |
877 |
|
b93b01631 page cache: use x... |
878 |
xa_lock_irqsave(&mapping->i_pages, flags); |
49d2e9cc4 [PATCH] Swap Migr... |
879 |
/* |
0fd0e6b05 [PATCH] page inva... |
880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 |
* The non racy check for a busy page. * * Must be careful with the order of the tests. When someone has * a ref to the page, it may be possible that they dirty it then * drop the reference. So if PageDirty is tested before page_count * here, then the following race may occur: * * get_user_pages(&page); * [user mapping goes away] * write_to(page); * !PageDirty(page) [good] * SetPageDirty(page); * put_page(page); * !page_count(page) [good, discard it] * * [oops, our write_to data is lost] * * Reversing the order of the tests ensures such a situation cannot * escape unnoticed. The smp_rmb is needed to ensure the page->flags |
0139aa7b7 mm: rename _count... |
899 |
* load is not satisfied before that of page->_refcount. |
0fd0e6b05 [PATCH] page inva... |
900 901 |
* * Note that if SetPageDirty is always performed via set_page_dirty, |
b93b01631 page cache: use x... |
902 |
* and thus under the i_pages lock, then this ordering is not required. |
49d2e9cc4 [PATCH] Swap Migr... |
903 |
*/ |
906d278d7 mm/vmscan.c: supp... |
904 |
refcount = 1 + compound_nr(page); |
bd4c82c22 mm, THP, swap: de... |
905 |
if (!page_ref_freeze(page, refcount)) |
49d2e9cc4 [PATCH] Swap Migr... |
906 |
goto cannot_free; |
1c4c3b99c mm: fix page_free... |
907 |
/* note: atomic_cmpxchg in page_ref_freeze provides the smp_rmb */ |
e286781d5 mm: speculative p... |
908 |
if (unlikely(PageDirty(page))) { |
bd4c82c22 mm, THP, swap: de... |
909 |
page_ref_unfreeze(page, refcount); |
49d2e9cc4 [PATCH] Swap Migr... |
910 |
goto cannot_free; |
e286781d5 mm: speculative p... |
911 |
} |
49d2e9cc4 [PATCH] Swap Migr... |
912 913 914 |
if (PageSwapCache(page)) { swp_entry_t swap = { .val = page_private(page) }; |
0a31bc97c mm: memcontrol: r... |
915 |
mem_cgroup_swapout(page, swap); |
4e17ec250 mm: Convert delet... |
916 |
__delete_from_swap_cache(page, swap); |
b93b01631 page cache: use x... |
917 |
xa_unlock_irqrestore(&mapping->i_pages, flags); |
75f6d6d29 mm, THP, swap: un... |
918 |
put_swap_page(page, swap); |
e286781d5 mm: speculative p... |
919 |
} else { |
6072d13c4 Call the filesyst... |
920 |
void (*freepage)(struct page *); |
a528910e1 mm: thrash detect... |
921 |
void *shadow = NULL; |
6072d13c4 Call the filesyst... |
922 923 |
freepage = mapping->a_ops->freepage; |
a528910e1 mm: thrash detect... |
924 925 926 927 928 929 930 931 932 |
/* * Remember a shadow entry for reclaimed file cache in * order to detect refaults, thus thrashing, later on. * * But don't store shadows in an address space that is * already exiting. This is not just an optizimation, * inode reclaim needs to empty out the radix tree or * the nodes are lost. Don't plant shadows behind its * back. |
f9fe48bec dax: support dirt... |
933 934 935 936 937 |
* * We also don't store shadows for DAX mappings because the * only page cache pages found in these are zero pages * covering holes, and because we don't want to mix DAX * exceptional entries and shadow exceptional entries in the |
b93b01631 page cache: use x... |
938 |
* same address_space. |
a528910e1 mm: thrash detect... |
939 940 |
*/ if (reclaimed && page_is_file_cache(page) && |
f9fe48bec dax: support dirt... |
941 |
!mapping_exiting(mapping) && !dax_mapping(mapping)) |
a7ca12f9d mm/workingset: re... |
942 |
shadow = workingset_eviction(page); |
62cccb8c8 mm: simplify lock... |
943 |
__delete_from_page_cache(page, shadow); |
b93b01631 page cache: use x... |
944 |
xa_unlock_irqrestore(&mapping->i_pages, flags); |
6072d13c4 Call the filesyst... |
945 946 947 |
if (freepage != NULL) freepage(page); |
49d2e9cc4 [PATCH] Swap Migr... |
948 |
} |
49d2e9cc4 [PATCH] Swap Migr... |
949 950 951 |
return 1; cannot_free: |
b93b01631 page cache: use x... |
952 |
xa_unlock_irqrestore(&mapping->i_pages, flags); |
49d2e9cc4 [PATCH] Swap Migr... |
953 954 |
return 0; } |
1da177e4c Linux-2.6.12-rc2 |
955 |
/* |
e286781d5 mm: speculative p... |
956 957 958 959 960 961 962 |
* Attempt to detach a locked page from its ->mapping. If it is dirty or if * someone else has a ref on the page, abort and return 0. If it was * successfully detached, return 1. Assumes the caller has a single ref on * this page. */ int remove_mapping(struct address_space *mapping, struct page *page) { |
a528910e1 mm: thrash detect... |
963 |
if (__remove_mapping(mapping, page, false)) { |
e286781d5 mm: speculative p... |
964 965 966 967 968 |
/* * Unfreezing the refcount with 1 rather than 2 effectively * drops the pagecache ref for us without requiring another * atomic operation. */ |
fe896d187 mm: introduce pag... |
969 |
page_ref_unfreeze(page, 1); |
e286781d5 mm: speculative p... |
970 971 972 973 |
return 1; } return 0; } |
894bc3104 Unevictable LRU I... |
974 975 976 977 978 979 980 981 982 |
/** * putback_lru_page - put previously isolated page onto appropriate LRU list * @page: page to be put back to appropriate lru list * * Add previously isolated @page to appropriate LRU list. * Page may still be unevictable for other reasons. * * lru_lock must not be held, interrupts must be enabled. */ |
894bc3104 Unevictable LRU I... |
983 984 |
void putback_lru_page(struct page *page) { |
9c4e6b1a7 mm, mlock, vmscan... |
985 |
lru_cache_add(page); |
894bc3104 Unevictable LRU I... |
986 987 |
put_page(page); /* drop ref from isolate */ } |
dfc8d636c vmscan: factor ou... |
988 989 990 |
enum page_references { PAGEREF_RECLAIM, PAGEREF_RECLAIM_CLEAN, |
645747462 vmscan: detect ma... |
991 |
PAGEREF_KEEP, |
dfc8d636c vmscan: factor ou... |
992 993 994 995 996 997 |
PAGEREF_ACTIVATE, }; static enum page_references page_check_references(struct page *page, struct scan_control *sc) { |
645747462 vmscan: detect ma... |
998 |
int referenced_ptes, referenced_page; |
dfc8d636c vmscan: factor ou... |
999 |
unsigned long vm_flags; |
dfc8d636c vmscan: factor ou... |
1000 |
|
c3ac9a8ad mm: memcg: count ... |
1001 1002 |
referenced_ptes = page_referenced(page, 1, sc->target_mem_cgroup, &vm_flags); |
645747462 vmscan: detect ma... |
1003 |
referenced_page = TestClearPageReferenced(page); |
dfc8d636c vmscan: factor ou... |
1004 |
|
dfc8d636c vmscan: factor ou... |
1005 1006 1007 1008 1009 1010 |
/* * Mlock lost the isolation race with us. Let try_to_unmap() * move the page to the unevictable list. */ if (vm_flags & VM_LOCKED) return PAGEREF_RECLAIM; |
645747462 vmscan: detect ma... |
1011 |
if (referenced_ptes) { |
e48982734 mm: consider all ... |
1012 |
if (PageSwapBacked(page)) |
645747462 vmscan: detect ma... |
1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 |
return PAGEREF_ACTIVATE; /* * All mapped pages start out with page table * references from the instantiating fault, so we need * to look twice if a mapped file page is used more * than once. * * Mark it and spare it for another trip around the * inactive list. Another page table reference will * lead to its activation. * * Note: the mark is set for activated pages as well * so that recently deactivated but used pages are * quickly recovered. */ SetPageReferenced(page); |
34dbc67a6 vmscan: promote s... |
1029 |
if (referenced_page || referenced_ptes > 1) |
645747462 vmscan: detect ma... |
1030 |
return PAGEREF_ACTIVATE; |
c909e9936 vmscan: activate ... |
1031 1032 1033 1034 1035 |
/* * Activate file-backed executable pages after first usage. */ if (vm_flags & VM_EXEC) return PAGEREF_ACTIVATE; |
645747462 vmscan: detect ma... |
1036 1037 |
return PAGEREF_KEEP; } |
dfc8d636c vmscan: factor ou... |
1038 1039 |
/* Reclaim if clean, defer dirty pages to writeback */ |
2e30244a7 vmscan,tmpfs: tre... |
1040 |
if (referenced_page && !PageSwapBacked(page)) |
645747462 vmscan: detect ma... |
1041 1042 1043 |
return PAGEREF_RECLAIM_CLEAN; return PAGEREF_RECLAIM; |
dfc8d636c vmscan: factor ou... |
1044 |
} |
e2be15f6c mm: vmscan: stall... |
1045 1046 1047 1048 |
/* Check if a page is dirty or under writeback */ static void page_check_dirty_writeback(struct page *page, bool *dirty, bool *writeback) { |
b45972265 mm: vmscan: take ... |
1049 |
struct address_space *mapping; |
e2be15f6c mm: vmscan: stall... |
1050 1051 1052 1053 |
/* * Anonymous pages are not handled by flushers and must be written * from reclaim context. Do not stall reclaim based on them */ |
802a3a92a mm: reclaim MADV_... |
1054 1055 |
if (!page_is_file_cache(page) || (PageAnon(page) && !PageSwapBacked(page))) { |
e2be15f6c mm: vmscan: stall... |
1056 1057 1058 1059 1060 1061 1062 1063 |
*dirty = false; *writeback = false; return; } /* By default assume that the page flags are accurate */ *dirty = PageDirty(page); *writeback = PageWriteback(page); |
b45972265 mm: vmscan: take ... |
1064 1065 1066 1067 1068 1069 1070 1071 |
/* Verify dirty/writeback state if the filesystem supports it */ if (!page_has_private(page)) return; mapping = page_mapping(page); if (mapping && mapping->a_ops->is_dirty_writeback) mapping->a_ops->is_dirty_writeback(page, dirty, writeback); |
e2be15f6c mm: vmscan: stall... |
1072 |
} |
e286781d5 mm: speculative p... |
1073 |
/* |
1742f19fa [PATCH] vmscan: r... |
1074 |
* shrink_page_list() returns the number of reclaimed pages |
1da177e4c Linux-2.6.12-rc2 |
1075 |
*/ |
1742f19fa [PATCH] vmscan: r... |
1076 |
static unsigned long shrink_page_list(struct list_head *page_list, |
599d0c954 mm, vmscan: move ... |
1077 |
struct pglist_data *pgdat, |
f84f6e2b0 mm: vmscan: do no... |
1078 |
struct scan_control *sc, |
02c6de8d7 mm: cma: discard ... |
1079 |
enum ttu_flags ttu_flags, |
3c710c1ad mm, vmscan: extra... |
1080 |
struct reclaim_stat *stat, |
8940b34a4 mm: change PAGERE... |
1081 |
bool ignore_references) |
1da177e4c Linux-2.6.12-rc2 |
1082 1083 |
{ LIST_HEAD(ret_pages); |
abe4c3b50 vmscan: set up pa... |
1084 |
LIST_HEAD(free_pages); |
3c710c1ad mm, vmscan: extra... |
1085 |
unsigned nr_reclaimed = 0; |
886cf1901 mm: move recent_r... |
1086 |
unsigned pgactivate = 0; |
1da177e4c Linux-2.6.12-rc2 |
1087 |
|
060f005f0 mm/vmscan.c: do n... |
1088 |
memset(stat, 0, sizeof(*stat)); |
1da177e4c Linux-2.6.12-rc2 |
1089 |
cond_resched(); |
1da177e4c Linux-2.6.12-rc2 |
1090 1091 1092 1093 |
while (!list_empty(page_list)) { struct address_space *mapping; struct page *page; int may_enter_fs; |
8940b34a4 mm: change PAGERE... |
1094 |
enum page_references references = PAGEREF_RECLAIM; |
e2be15f6c mm: vmscan: stall... |
1095 |
bool dirty, writeback; |
98879b3b9 mm: vmscan: corre... |
1096 |
unsigned int nr_pages; |
1da177e4c Linux-2.6.12-rc2 |
1097 1098 1099 1100 1101 |
cond_resched(); page = lru_to_page(page_list); list_del(&page->lru); |
529ae9aaa mm: rename page t... |
1102 |
if (!trylock_page(page)) |
1da177e4c Linux-2.6.12-rc2 |
1103 |
goto keep; |
309381fea mm: dump page whe... |
1104 |
VM_BUG_ON_PAGE(PageActive(page), page); |
1da177e4c Linux-2.6.12-rc2 |
1105 |
|
d8c6546b1 mm: introduce com... |
1106 |
nr_pages = compound_nr(page); |
98879b3b9 mm: vmscan: corre... |
1107 1108 1109 |
/* Account the number of base pages even though THP */ sc->nr_scanned += nr_pages; |
80e434260 [PATCH] zone recl... |
1110 |
|
39b5f29ac mm: remove vma ar... |
1111 |
if (unlikely(!page_evictable(page))) |
ad6b67041 mm: remove SWAP_M... |
1112 |
goto activate_locked; |
894bc3104 Unevictable LRU I... |
1113 |
|
a6dc60f89 vmscan: rename sc... |
1114 |
if (!sc->may_unmap && page_mapped(page)) |
80e434260 [PATCH] zone recl... |
1115 |
goto keep_locked; |
c661b078f synchronous lumpy... |
1116 1117 |
may_enter_fs = (sc->gfp_mask & __GFP_FS) || (PageSwapCache(page) && (sc->gfp_mask & __GFP_IO)); |
283aba9f9 mm: vmscan: block... |
1118 |
/* |
894befec4 mm/vmscan: update... |
1119 |
* The number of dirty pages determines if a node is marked |
e2be15f6c mm: vmscan: stall... |
1120 1121 1122 1123 1124 1125 |
* reclaim_congested which affects wait_iff_congested. kswapd * will stall and start writing pages if the tail of the LRU * is all dirty unqueued pages. */ page_check_dirty_writeback(page, &dirty, &writeback); if (dirty || writeback) |
060f005f0 mm/vmscan.c: do n... |
1126 |
stat->nr_dirty++; |
e2be15f6c mm: vmscan: stall... |
1127 1128 |
if (dirty && !writeback) |
060f005f0 mm/vmscan.c: do n... |
1129 |
stat->nr_unqueued_dirty++; |
e2be15f6c mm: vmscan: stall... |
1130 |
|
d04e8acd0 mm: vmscan: treat... |
1131 1132 1133 1134 1135 1136 |
/* * Treat this page as congested if the underlying BDI is or if * pages are cycling through the LRU so quickly that the * pages marked for immediate reclaim are making it to the * end of the LRU a second time. */ |
e2be15f6c mm: vmscan: stall... |
1137 |
mapping = page_mapping(page); |
1da58ee2a mm: vmscan: count... |
1138 |
if (((dirty || writeback) && mapping && |
703c27088 writeback: implem... |
1139 |
inode_write_congested(mapping->host)) || |
d04e8acd0 mm: vmscan: treat... |
1140 |
(writeback && PageReclaim(page))) |
060f005f0 mm/vmscan.c: do n... |
1141 |
stat->nr_congested++; |
e2be15f6c mm: vmscan: stall... |
1142 1143 |
/* |
283aba9f9 mm: vmscan: block... |
1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 |
* If a page at the tail of the LRU is under writeback, there * are three cases to consider. * * 1) If reclaim is encountering an excessive number of pages * under writeback and this page is both under writeback and * PageReclaim then it indicates that pages are being queued * for IO but are being recycled through the LRU before the * IO can complete. Waiting on the page itself risks an * indefinite stall if it is impossible to writeback the * page due to IO error or disconnected storage so instead |
b1a6f21e3 mm: vmscan: stall... |
1154 1155 |
* note that the LRU is being scanned too quickly and the * caller can stall after page list has been processed. |
283aba9f9 mm: vmscan: block... |
1156 |
* |
97c9341f7 mm: vmscan: disab... |
1157 |
* 2) Global or new memcg reclaim encounters a page that is |
ecf5fc6e9 mm, vmscan: Do no... |
1158 1159 1160 |
* not marked for immediate reclaim, or the caller does not * have __GFP_FS (or __GFP_IO if it's simply going to swap, * not to fs). In this case mark the page for immediate |
97c9341f7 mm: vmscan: disab... |
1161 |
* reclaim and continue scanning. |
283aba9f9 mm: vmscan: block... |
1162 |
* |
ecf5fc6e9 mm, vmscan: Do no... |
1163 1164 |
* Require may_enter_fs because we would wait on fs, which * may not have submitted IO yet. And the loop driver might |
283aba9f9 mm: vmscan: block... |
1165 1166 1167 1168 1169 |
* enter reclaim, and deadlock if it waits on a page for * which it is needed to do the write (loop masks off * __GFP_IO|__GFP_FS for this reason); but more thought * would probably show more reasons. * |
7fadc8202 mm, vmscan: unloc... |
1170 |
* 3) Legacy memcg encounters a page that is already marked |
283aba9f9 mm: vmscan: block... |
1171 1172 1173 1174 |
* PageReclaim. memcg does not have any dirty pages * throttling so we could easily OOM just because too many * pages are in writeback and there is nothing else to * reclaim. Wait for the writeback to complete. |
c55e8d035 mm: vmscan: move ... |
1175 1176 1177 1178 1179 1180 1181 1182 1183 |
* * In cases 1) and 2) we activate the pages to get them out of * the way while we continue scanning for clean pages on the * inactive list and refilling from the active list. The * observation here is that waiting for disk writes is more * expensive than potentially causing reloads down the line. * Since they're marked for immediate reclaim, they won't put * memory pressure on the cache working set any longer than it * takes to write them to disk. |
283aba9f9 mm: vmscan: block... |
1184 |
*/ |
c661b078f synchronous lumpy... |
1185 |
if (PageWriteback(page)) { |
283aba9f9 mm: vmscan: block... |
1186 1187 1188 |
/* Case 1 above */ if (current_is_kswapd() && PageReclaim(page) && |
599d0c954 mm, vmscan: move ... |
1189 |
test_bit(PGDAT_WRITEBACK, &pgdat->flags)) { |
060f005f0 mm/vmscan.c: do n... |
1190 |
stat->nr_immediate++; |
c55e8d035 mm: vmscan: move ... |
1191 |
goto activate_locked; |
283aba9f9 mm: vmscan: block... |
1192 1193 |
/* Case 2 above */ |
97c9341f7 mm: vmscan: disab... |
1194 |
} else if (sane_reclaim(sc) || |
ecf5fc6e9 mm, vmscan: Do no... |
1195 |
!PageReclaim(page) || !may_enter_fs) { |
c3b94f44f memcg: further pr... |
1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 |
/* * This is slightly racy - end_page_writeback() * might have just cleared PageReclaim, then * setting PageReclaim here end up interpreted * as PageReadahead - but that does not matter * enough to care. What we do want is for this * page to have PageReclaim set next time memcg * reclaim reaches the tests above, so it will * then wait_on_page_writeback() to avoid OOM; * and it's also appropriate in global reclaim. */ SetPageReclaim(page); |
060f005f0 mm/vmscan.c: do n... |
1208 |
stat->nr_writeback++; |
c55e8d035 mm: vmscan: move ... |
1209 |
goto activate_locked; |
283aba9f9 mm: vmscan: block... |
1210 1211 1212 |
/* Case 3 above */ } else { |
7fadc8202 mm, vmscan: unloc... |
1213 |
unlock_page(page); |
283aba9f9 mm: vmscan: block... |
1214 |
wait_on_page_writeback(page); |
7fadc8202 mm, vmscan: unloc... |
1215 1216 1217 |
/* then go back and try same page again */ list_add_tail(&page->lru, page_list); continue; |
e62e384e9 memcg: prevent OO... |
1218 |
} |
c661b078f synchronous lumpy... |
1219 |
} |
1da177e4c Linux-2.6.12-rc2 |
1220 |
|
8940b34a4 mm: change PAGERE... |
1221 |
if (!ignore_references) |
02c6de8d7 mm: cma: discard ... |
1222 |
references = page_check_references(page, sc); |
dfc8d636c vmscan: factor ou... |
1223 1224 |
switch (references) { case PAGEREF_ACTIVATE: |
1da177e4c Linux-2.6.12-rc2 |
1225 |
goto activate_locked; |
645747462 vmscan: detect ma... |
1226 |
case PAGEREF_KEEP: |
98879b3b9 mm: vmscan: corre... |
1227 |
stat->nr_ref_keep += nr_pages; |
645747462 vmscan: detect ma... |
1228 |
goto keep_locked; |
dfc8d636c vmscan: factor ou... |
1229 1230 1231 1232 |
case PAGEREF_RECLAIM: case PAGEREF_RECLAIM_CLEAN: ; /* try to reclaim the page below */ } |
1da177e4c Linux-2.6.12-rc2 |
1233 |
|
1da177e4c Linux-2.6.12-rc2 |
1234 1235 1236 |
/* * Anonymous process memory has backing store? * Try to allocate it some swap space here. |
802a3a92a mm: reclaim MADV_... |
1237 |
* Lazyfree page could be freed directly |
1da177e4c Linux-2.6.12-rc2 |
1238 |
*/ |
bd4c82c22 mm, THP, swap: de... |
1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 |
if (PageAnon(page) && PageSwapBacked(page)) { if (!PageSwapCache(page)) { if (!(sc->gfp_mask & __GFP_IO)) goto keep_locked; if (PageTransHuge(page)) { /* cannot split THP, skip it */ if (!can_split_huge_page(page, NULL)) goto activate_locked; /* * Split pages without a PMD map right * away. Chances are some or all of the * tail pages can be freed without IO. */ if (!compound_mapcount(page) && split_huge_page_to_list(page, page_list)) goto activate_locked; } if (!add_to_swap(page)) { if (!PageTransHuge(page)) |
98879b3b9 mm: vmscan: corre... |
1259 |
goto activate_locked_split; |
bd4c82c22 mm, THP, swap: de... |
1260 1261 1262 1263 |
/* Fallback to swap normal pages */ if (split_huge_page_to_list(page, page_list)) goto activate_locked; |
fe490cc0f mm, THP, swap: ad... |
1264 1265 1266 |
#ifdef CONFIG_TRANSPARENT_HUGEPAGE count_vm_event(THP_SWPOUT_FALLBACK); #endif |
bd4c82c22 mm, THP, swap: de... |
1267 |
if (!add_to_swap(page)) |
98879b3b9 mm: vmscan: corre... |
1268 |
goto activate_locked_split; |
bd4c82c22 mm, THP, swap: de... |
1269 |
} |
0f0746589 mm, THP, swap: mo... |
1270 |
|
bd4c82c22 mm, THP, swap: de... |
1271 |
may_enter_fs = 1; |
1da177e4c Linux-2.6.12-rc2 |
1272 |
|
bd4c82c22 mm, THP, swap: de... |
1273 1274 1275 |
/* Adding to swap updated mapping */ mapping = page_mapping(page); } |
7751b2da6 vmscan: split fil... |
1276 1277 1278 1279 |
} else if (unlikely(PageTransHuge(page))) { /* Split file THP */ if (split_huge_page_to_list(page, page_list)) goto keep_locked; |
e2be15f6c mm: vmscan: stall... |
1280 |
} |
1da177e4c Linux-2.6.12-rc2 |
1281 1282 |
/* |
98879b3b9 mm: vmscan: corre... |
1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 |
* THP may get split above, need minus tail pages and update * nr_pages to avoid accounting tail pages twice. * * The tail pages that are added into swap cache successfully * reach here. */ if ((nr_pages > 1) && !PageTransHuge(page)) { sc->nr_scanned -= (nr_pages - 1); nr_pages = 1; } /* |
1da177e4c Linux-2.6.12-rc2 |
1295 1296 1297 |
* The page is mapped into the page tables of one or more * processes. Try to unmap it here. */ |
802a3a92a mm: reclaim MADV_... |
1298 |
if (page_mapped(page)) { |
bd4c82c22 mm, THP, swap: de... |
1299 1300 1301 1302 1303 |
enum ttu_flags flags = ttu_flags | TTU_BATCH_FLUSH; if (unlikely(PageTransHuge(page))) flags |= TTU_SPLIT_HUGE_PMD; if (!try_to_unmap(page, flags)) { |
98879b3b9 mm: vmscan: corre... |
1304 |
stat->nr_unmap_fail += nr_pages; |
1da177e4c Linux-2.6.12-rc2 |
1305 |
goto activate_locked; |
1da177e4c Linux-2.6.12-rc2 |
1306 1307 1308 1309 |
} } if (PageDirty(page)) { |
ee72886d8 mm: vmscan: do no... |
1310 |
/* |
4eda48235 mm: vmscan: only ... |
1311 1312 1313 1314 1315 1316 1317 1318 |
* Only kswapd can writeback filesystem pages * to avoid risk of stack overflow. But avoid * injecting inefficient single-page IO into * flusher writeback as much as possible: only * write pages when we've encountered many * dirty pages, and when we've already scanned * the rest of the LRU for clean pages and see * the same dirty pages again (PageReclaim). |
ee72886d8 mm: vmscan: do no... |
1319 |
*/ |
f84f6e2b0 mm: vmscan: do no... |
1320 |
if (page_is_file_cache(page) && |
4eda48235 mm: vmscan: only ... |
1321 1322 |
(!current_is_kswapd() || !PageReclaim(page) || !test_bit(PGDAT_DIRTY, &pgdat->flags))) { |
49ea7eb65 mm: vmscan: immed... |
1323 1324 1325 1326 1327 1328 |
/* * Immediately reclaim when written back. * Similar in principal to deactivate_page() * except we already have the page isolated * and know it's dirty */ |
c4a25635b mm: move vmscan w... |
1329 |
inc_node_page_state(page, NR_VMSCAN_IMMEDIATE); |
49ea7eb65 mm: vmscan: immed... |
1330 |
SetPageReclaim(page); |
c55e8d035 mm: vmscan: move ... |
1331 |
goto activate_locked; |
ee72886d8 mm: vmscan: do no... |
1332 |
} |
dfc8d636c vmscan: factor ou... |
1333 |
if (references == PAGEREF_RECLAIM_CLEAN) |
1da177e4c Linux-2.6.12-rc2 |
1334 |
goto keep_locked; |
4dd4b9202 revert "kswapd sh... |
1335 |
if (!may_enter_fs) |
1da177e4c Linux-2.6.12-rc2 |
1336 |
goto keep_locked; |
52a8363ea [PATCH] mm: impro... |
1337 |
if (!sc->may_writepage) |
1da177e4c Linux-2.6.12-rc2 |
1338 |
goto keep_locked; |
d950c9477 mm: defer flush o... |
1339 1340 1341 1342 1343 1344 |
/* * Page is dirty. Flush the TLB if a writable entry * potentially exists to avoid CPU writes after IO * starts and then write it out here. */ try_to_unmap_flush_dirty(); |
7d3579e8e vmscan: narrow th... |
1345 |
switch (pageout(page, mapping, sc)) { |
1da177e4c Linux-2.6.12-rc2 |
1346 1347 1348 1349 1350 |
case PAGE_KEEP: goto keep_locked; case PAGE_ACTIVATE: goto activate_locked; case PAGE_SUCCESS: |
7d3579e8e vmscan: narrow th... |
1351 |
if (PageWriteback(page)) |
41ac1999c mm: vmscan: do no... |
1352 |
goto keep; |
7d3579e8e vmscan: narrow th... |
1353 |
if (PageDirty(page)) |
1da177e4c Linux-2.6.12-rc2 |
1354 |
goto keep; |
7d3579e8e vmscan: narrow th... |
1355 |
|
1da177e4c Linux-2.6.12-rc2 |
1356 1357 1358 1359 |
/* * A synchronous write - probably a ramdisk. Go * ahead and try to reclaim the page. */ |
529ae9aaa mm: rename page t... |
1360 |
if (!trylock_page(page)) |
1da177e4c Linux-2.6.12-rc2 |
1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 |
goto keep; if (PageDirty(page) || PageWriteback(page)) goto keep_locked; mapping = page_mapping(page); case PAGE_CLEAN: ; /* try to free the page below */ } } /* * If the page has buffers, try to free the buffer mappings * associated with this page. If we succeed we try to free * the page as well. * * We do this even if the page is PageDirty(). * try_to_release_page() does not perform I/O, but it is * possible for a page to have PageDirty set, but it is actually * clean (all its buffers are clean). This happens if the * buffers were written out directly, with submit_bh(). ext3 |
894bc3104 Unevictable LRU I... |
1380 |
* will do this, as well as the blockdev mapping. |
1da177e4c Linux-2.6.12-rc2 |
1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 |
* try_to_release_page() will discover that cleanness and will * drop the buffers and mark the page clean - it can be freed. * * Rarely, pages can have buffers and no ->mapping. These are * the pages which were not successfully invalidated in * truncate_complete_page(). We try to drop those buffers here * and if that worked, and the page is no longer mapped into * process address space (page_count == 1) it can be freed. * Otherwise, leave the page on the LRU so it is swappable. */ |
266cf658e FS-Cache: Recruit... |
1391 |
if (page_has_private(page)) { |
1da177e4c Linux-2.6.12-rc2 |
1392 1393 |
if (!try_to_release_page(page, sc->gfp_mask)) goto activate_locked; |
e286781d5 mm: speculative p... |
1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 |
if (!mapping && page_count(page) == 1) { unlock_page(page); if (put_page_testzero(page)) goto free_it; else { /* * rare race with speculative reference. * the speculative reference will free * this page shortly, so we may * increment nr_reclaimed here (and * leave it off the LRU). */ nr_reclaimed++; continue; } } |
1da177e4c Linux-2.6.12-rc2 |
1410 |
} |
802a3a92a mm: reclaim MADV_... |
1411 1412 1413 1414 1415 1416 1417 1418 |
if (PageAnon(page) && !PageSwapBacked(page)) { /* follow __remove_mapping for reference */ if (!page_ref_freeze(page, 1)) goto keep_locked; if (PageDirty(page)) { page_ref_unfreeze(page, 1); goto keep_locked; } |
1da177e4c Linux-2.6.12-rc2 |
1419 |
|
802a3a92a mm: reclaim MADV_... |
1420 |
count_vm_event(PGLAZYFREED); |
2262185c5 mm: per-cgroup me... |
1421 |
count_memcg_page_event(page, PGLAZYFREED); |
802a3a92a mm: reclaim MADV_... |
1422 1423 |
} else if (!mapping || !__remove_mapping(mapping, page, true)) goto keep_locked; |
9a1ea439b mm: put_and_wait_... |
1424 1425 |
unlock_page(page); |
e286781d5 mm: speculative p... |
1426 |
free_it: |
98879b3b9 mm: vmscan: corre... |
1427 1428 1429 1430 1431 |
/* * THP may get swapped out in a whole, need account * all base pages. */ nr_reclaimed += nr_pages; |
abe4c3b50 vmscan: set up pa... |
1432 1433 1434 1435 1436 |
/* * Is there need to periodically free_page_list? It would * appear not as the counts should be low */ |
7ae88534c mm: move mem_cgro... |
1437 |
if (unlikely(PageTransHuge(page))) |
bd4c82c22 mm, THP, swap: de... |
1438 |
(*get_compound_page_dtor(page))(page); |
7ae88534c mm: move mem_cgro... |
1439 |
else |
bd4c82c22 mm, THP, swap: de... |
1440 |
list_add(&page->lru, &free_pages); |
1da177e4c Linux-2.6.12-rc2 |
1441 |
continue; |
98879b3b9 mm: vmscan: corre... |
1442 1443 1444 1445 1446 1447 1448 1449 1450 |
activate_locked_split: /* * The tail pages that are failed to add into swap cache * reach here. Fixup nr_scanned and nr_pages. */ if (nr_pages > 1) { sc->nr_scanned -= (nr_pages - 1); nr_pages = 1; } |
1da177e4c Linux-2.6.12-rc2 |
1451 |
activate_locked: |
68a22394c vmscan: free swap... |
1452 |
/* Not a candidate for swapping, so reclaim swap space. */ |
ad6b67041 mm: remove SWAP_M... |
1453 1454 |
if (PageSwapCache(page) && (mem_cgroup_swap_full(page) || PageMlocked(page))) |
a2c43eed8 mm: try_to_free_s... |
1455 |
try_to_free_swap(page); |
309381fea mm: dump page whe... |
1456 |
VM_BUG_ON_PAGE(PageActive(page), page); |
ad6b67041 mm: remove SWAP_M... |
1457 |
if (!PageMlocked(page)) { |
886cf1901 mm: move recent_r... |
1458 |
int type = page_is_file_cache(page); |
ad6b67041 mm: remove SWAP_M... |
1459 |
SetPageActive(page); |
98879b3b9 mm: vmscan: corre... |
1460 |
stat->nr_activate[type] += nr_pages; |
2262185c5 mm: per-cgroup me... |
1461 |
count_memcg_page_event(page, PGACTIVATE); |
ad6b67041 mm: remove SWAP_M... |
1462 |
} |
1da177e4c Linux-2.6.12-rc2 |
1463 1464 1465 1466 |
keep_locked: unlock_page(page); keep: list_add(&page->lru, &ret_pages); |
309381fea mm: dump page whe... |
1467 |
VM_BUG_ON_PAGE(PageLRU(page) || PageUnevictable(page), page); |
1da177e4c Linux-2.6.12-rc2 |
1468 |
} |
abe4c3b50 vmscan: set up pa... |
1469 |
|
98879b3b9 mm: vmscan: corre... |
1470 |
pgactivate = stat->nr_activate[0] + stat->nr_activate[1]; |
747db954c mm: memcontrol: u... |
1471 |
mem_cgroup_uncharge_list(&free_pages); |
72b252aed mm: send one IPI ... |
1472 |
try_to_unmap_flush(); |
2d4894b5d mm: remove cold p... |
1473 |
free_unref_page_list(&free_pages); |
abe4c3b50 vmscan: set up pa... |
1474 |
|
1da177e4c Linux-2.6.12-rc2 |
1475 |
list_splice(&ret_pages, page_list); |
886cf1901 mm: move recent_r... |
1476 |
count_vm_events(PGACTIVATE, pgactivate); |
060f005f0 mm/vmscan.c: do n... |
1477 |
|
05ff51376 [PATCH] vmscan re... |
1478 |
return nr_reclaimed; |
1da177e4c Linux-2.6.12-rc2 |
1479 |
} |
02c6de8d7 mm: cma: discard ... |
1480 1481 1482 1483 1484 1485 1486 1487 |
unsigned long reclaim_clean_pages_from_list(struct zone *zone, struct list_head *page_list) { struct scan_control sc = { .gfp_mask = GFP_KERNEL, .priority = DEF_PRIORITY, .may_unmap = 1, }; |
060f005f0 mm/vmscan.c: do n... |
1488 |
struct reclaim_stat dummy_stat; |
3c710c1ad mm, vmscan: extra... |
1489 |
unsigned long ret; |
02c6de8d7 mm: cma: discard ... |
1490 1491 1492 1493 |
struct page *page, *next; LIST_HEAD(clean_pages); list_for_each_entry_safe(page, next, page_list, lru) { |
117aad1e9 mm: avoid reinser... |
1494 |
if (page_is_file_cache(page) && !PageDirty(page) && |
a58f2cef2 mm/vmscan.c: fix ... |
1495 |
!__PageMovable(page) && !PageUnevictable(page)) { |
02c6de8d7 mm: cma: discard ... |
1496 1497 1498 1499 |
ClearPageActive(page); list_move(&page->lru, &clean_pages); } } |
599d0c954 mm, vmscan: move ... |
1500 |
ret = shrink_page_list(&clean_pages, zone->zone_pgdat, &sc, |
060f005f0 mm/vmscan.c: do n... |
1501 |
TTU_IGNORE_ACCESS, &dummy_stat, true); |
02c6de8d7 mm: cma: discard ... |
1502 |
list_splice(&clean_pages, page_list); |
599d0c954 mm, vmscan: move ... |
1503 |
mod_node_page_state(zone->zone_pgdat, NR_ISOLATED_FILE, -ret); |
02c6de8d7 mm: cma: discard ... |
1504 1505 |
return ret; } |
5ad333eb6 Lumpy Reclaim V4 |
1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 |
/* * Attempt to remove the specified page from its LRU. Only take this page * if it is of the appropriate PageActive status. Pages which are being * freed elsewhere are also ignored. * * page: page to consider * mode: one of the LRU isolation modes defined above * * returns 0 on success, -ve errno on failure. */ |
f3fd4a619 mm: remove lru ty... |
1516 |
int __isolate_lru_page(struct page *page, isolate_mode_t mode) |
5ad333eb6 Lumpy Reclaim V4 |
1517 1518 1519 1520 1521 1522 |
{ int ret = -EINVAL; /* Only take pages on the LRU. */ if (!PageLRU(page)) return ret; |
e46a28790 CMA: migrate mloc... |
1523 1524 |
/* Compaction should not handle unevictable pages but CMA can do so */ if (PageUnevictable(page) && !(mode & ISOLATE_UNEVICTABLE)) |
894bc3104 Unevictable LRU I... |
1525 |
return ret; |
5ad333eb6 Lumpy Reclaim V4 |
1526 |
ret = -EBUSY; |
08e552c69 memcg: synchroniz... |
1527 |
|
c82449352 mm: compaction: m... |
1528 1529 1530 1531 1532 |
/* * To minimise LRU disruption, the caller can indicate that it only * wants to isolate pages it will be able to operate on without * blocking - clean pages for the most part. * |
c82449352 mm: compaction: m... |
1533 1534 1535 |
* ISOLATE_ASYNC_MIGRATE is used to indicate that it only wants to pages * that it is possible to migrate without blocking */ |
1276ad68e mm: vmscan: scan ... |
1536 |
if (mode & ISOLATE_ASYNC_MIGRATE) { |
c82449352 mm: compaction: m... |
1537 1538 1539 1540 1541 1542 |
/* All the caller can do on PageWriteback is block */ if (PageWriteback(page)) return ret; if (PageDirty(page)) { struct address_space *mapping; |
69d763fc6 mm: pin address_s... |
1543 |
bool migrate_dirty; |
c82449352 mm: compaction: m... |
1544 |
|
c82449352 mm: compaction: m... |
1545 1546 1547 |
/* * Only pages without mappings or that have a * ->migratepage callback are possible to migrate |
69d763fc6 mm: pin address_s... |
1548 1549 1550 1551 1552 |
* without blocking. However, we can be racing with * truncation so it's necessary to lock the page * to stabilise the mapping as truncation holds * the page lock until after the page is removed * from the page cache. |
c82449352 mm: compaction: m... |
1553 |
*/ |
69d763fc6 mm: pin address_s... |
1554 1555 |
if (!trylock_page(page)) return ret; |
c82449352 mm: compaction: m... |
1556 |
mapping = page_mapping(page); |
145e1a71e mm: fix the NULL ... |
1557 |
migrate_dirty = !mapping || mapping->a_ops->migratepage; |
69d763fc6 mm: pin address_s... |
1558 1559 |
unlock_page(page); if (!migrate_dirty) |
c82449352 mm: compaction: m... |
1560 1561 1562 |
return ret; } } |
39deaf858 mm: compaction: m... |
1563 |
|
f80c06736 mm: zone_reclaim:... |
1564 1565 |
if ((mode & ISOLATE_UNMAPPED) && page_mapped(page)) return ret; |
5ad333eb6 Lumpy Reclaim V4 |
1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 |
if (likely(get_page_unless_zero(page))) { /* * Be careful not to clear PageLRU until after we're * sure the page is not being freed elsewhere -- the * page release code relies on it. */ ClearPageLRU(page); ret = 0; } return ret; } |
7ee36a14f mm, vmscan: Updat... |
1578 1579 1580 1581 1582 1583 |
/* * Update LRU sizes after isolating pages. The LRU size updates must * be complete before mem_cgroup_update_lru_size due to a santity check. */ static __always_inline void update_lru_sizes(struct lruvec *lruvec, |
b4536f0c8 mm, memcg: fix th... |
1584 |
enum lru_list lru, unsigned long *nr_zone_taken) |
7ee36a14f mm, vmscan: Updat... |
1585 |
{ |
7ee36a14f mm, vmscan: Updat... |
1586 |
int zid; |
7ee36a14f mm, vmscan: Updat... |
1587 1588 1589 1590 1591 |
for (zid = 0; zid < MAX_NR_ZONES; zid++) { if (!nr_zone_taken[zid]) continue; __update_lru_size(lruvec, lru, zid, -nr_zone_taken[zid]); |
7ee36a14f mm, vmscan: Updat... |
1592 |
#ifdef CONFIG_MEMCG |
b4536f0c8 mm, memcg: fix th... |
1593 |
mem_cgroup_update_lru_size(lruvec, lru, zid, -nr_zone_taken[zid]); |
7ee36a14f mm, vmscan: Updat... |
1594 |
#endif |
b4536f0c8 mm, memcg: fix th... |
1595 |
} |
7ee36a14f mm, vmscan: Updat... |
1596 |
} |
f4b7e272b mm: remove zone_l... |
1597 1598 |
/** * pgdat->lru_lock is heavily contended. Some of the functions that |
1da177e4c Linux-2.6.12-rc2 |
1599 1600 1601 1602 1603 1604 1605 1606 |
* shrink the lists perform better by taking out a batch of pages * and working on them outside the LRU lock. * * For pagecache intensive workloads, this function is the hottest * spot in the kernel (apart from copy_*_user functions). * * Appropriate locks must be held before calling this function. * |
791b48b64 mm: vmscan: scan ... |
1607 |
* @nr_to_scan: The number of eligible pages to look through on the list. |
5dc35979e mm/vmscan: push l... |
1608 |
* @lruvec: The LRU vector to pull pages from. |
1da177e4c Linux-2.6.12-rc2 |
1609 |
* @dst: The temp list to put pages on to. |
f626012db mm: remove isolat... |
1610 |
* @nr_scanned: The number of pages that were scanned. |
fe2c2a106 vmscan: reclaim a... |
1611 |
* @sc: The scan_control struct for this reclaim session |
5ad333eb6 Lumpy Reclaim V4 |
1612 |
* @mode: One of the LRU isolation modes |
3cb994517 mm: push lru inde... |
1613 |
* @lru: LRU list id for isolating |
1da177e4c Linux-2.6.12-rc2 |
1614 1615 1616 |
* * returns how many pages were moved onto *@dst. */ |
69e05944a [PATCH] vmscan: u... |
1617 |
static unsigned long isolate_lru_pages(unsigned long nr_to_scan, |
5dc35979e mm/vmscan: push l... |
1618 |
struct lruvec *lruvec, struct list_head *dst, |
fe2c2a106 vmscan: reclaim a... |
1619 |
unsigned long *nr_scanned, struct scan_control *sc, |
a9e7c39fa mm/vmscan.c: remo... |
1620 |
enum lru_list lru) |
1da177e4c Linux-2.6.12-rc2 |
1621 |
{ |
75b00af77 mm: trivial clean... |
1622 |
struct list_head *src = &lruvec->lists[lru]; |
69e05944a [PATCH] vmscan: u... |
1623 |
unsigned long nr_taken = 0; |
599d0c954 mm, vmscan: move ... |
1624 |
unsigned long nr_zone_taken[MAX_NR_ZONES] = { 0 }; |
7cc30fcfd mm: vmstat: accou... |
1625 |
unsigned long nr_skipped[MAX_NR_ZONES] = { 0, }; |
3db65812d Revert "mm, vmsca... |
1626 |
unsigned long skipped = 0; |
791b48b64 mm: vmscan: scan ... |
1627 |
unsigned long scan, total_scan, nr_pages; |
b2e18757f mm, vmscan: begin... |
1628 |
LIST_HEAD(pages_skipped); |
a9e7c39fa mm/vmscan.c: remo... |
1629 |
isolate_mode_t mode = (sc->may_unmap ? 0 : ISOLATE_UNMAPPED); |
1da177e4c Linux-2.6.12-rc2 |
1630 |
|
98879b3b9 mm: vmscan: corre... |
1631 |
total_scan = 0; |
791b48b64 mm: vmscan: scan ... |
1632 |
scan = 0; |
98879b3b9 mm: vmscan: corre... |
1633 |
while (scan < nr_to_scan && !list_empty(src)) { |
5ad333eb6 Lumpy Reclaim V4 |
1634 |
struct page *page; |
5ad333eb6 Lumpy Reclaim V4 |
1635 |
|
1da177e4c Linux-2.6.12-rc2 |
1636 1637 |
page = lru_to_page(src); prefetchw_prev_lru_page(page, src, flags); |
309381fea mm: dump page whe... |
1638 |
VM_BUG_ON_PAGE(!PageLRU(page), page); |
8d438f96d [PATCH] mm: PageL... |
1639 |
|
d8c6546b1 mm: introduce com... |
1640 |
nr_pages = compound_nr(page); |
98879b3b9 mm: vmscan: corre... |
1641 |
total_scan += nr_pages; |
b2e18757f mm, vmscan: begin... |
1642 1643 |
if (page_zonenum(page) > sc->reclaim_idx) { list_move(&page->lru, &pages_skipped); |
98879b3b9 mm: vmscan: corre... |
1644 |
nr_skipped[page_zonenum(page)] += nr_pages; |
b2e18757f mm, vmscan: begin... |
1645 1646 |
continue; } |
791b48b64 mm: vmscan: scan ... |
1647 1648 1649 1650 1651 |
/* * Do not count skipped pages because that makes the function * return with no isolated pages if the LRU mostly contains * ineligible pages. This causes the VM to not reclaim any * pages, triggering a premature OOM. |
98879b3b9 mm: vmscan: corre... |
1652 1653 1654 1655 |
* * Account all tail pages of THP. This would not cause * premature OOM since __isolate_lru_page() returns -EBUSY * only when the page is being freed somewhere else. |
791b48b64 mm: vmscan: scan ... |
1656 |
*/ |
98879b3b9 mm: vmscan: corre... |
1657 |
scan += nr_pages; |
f3fd4a619 mm: remove lru ty... |
1658 |
switch (__isolate_lru_page(page, mode)) { |
5ad333eb6 Lumpy Reclaim V4 |
1659 |
case 0: |
599d0c954 mm, vmscan: move ... |
1660 1661 |
nr_taken += nr_pages; nr_zone_taken[page_zonenum(page)] += nr_pages; |
5ad333eb6 Lumpy Reclaim V4 |
1662 |
list_move(&page->lru, dst); |
5ad333eb6 Lumpy Reclaim V4 |
1663 1664 1665 1666 1667 1668 |
break; case -EBUSY: /* else it is being freed elsewhere */ list_move(&page->lru, src); continue; |
46453a6e1 [PATCH] mm: never... |
1669 |
|
5ad333eb6 Lumpy Reclaim V4 |
1670 1671 1672 |
default: BUG(); } |
1da177e4c Linux-2.6.12-rc2 |
1673 |
} |
b2e18757f mm, vmscan: begin... |
1674 1675 1676 1677 1678 1679 1680 |
/* * Splice any skipped pages to the start of the LRU list. Note that * this disrupts the LRU order when reclaiming for lower zones but * we cannot splice to the tail. If we did then the SWAP_CLUSTER_MAX * scanning would soon rescan the same pages to skip and put the * system at risk of premature OOM. */ |
7cc30fcfd mm: vmstat: accou... |
1681 1682 |
if (!list_empty(&pages_skipped)) { int zid; |
3db65812d Revert "mm, vmsca... |
1683 |
list_splice(&pages_skipped, src); |
7cc30fcfd mm: vmstat: accou... |
1684 1685 1686 1687 1688 |
for (zid = 0; zid < MAX_NR_ZONES; zid++) { if (!nr_skipped[zid]) continue; __count_zid_vm_events(PGSCAN_SKIP, zid, nr_skipped[zid]); |
1265e3a69 mm, vmscan: show ... |
1689 |
skipped += nr_skipped[zid]; |
7cc30fcfd mm: vmstat: accou... |
1690 1691 |
} } |
791b48b64 mm: vmscan: scan ... |
1692 |
*nr_scanned = total_scan; |
1265e3a69 mm, vmscan: show ... |
1693 |
trace_mm_vmscan_lru_isolate(sc->reclaim_idx, sc->order, nr_to_scan, |
791b48b64 mm: vmscan: scan ... |
1694 |
total_scan, skipped, nr_taken, mode, lru); |
b4536f0c8 mm, memcg: fix th... |
1695 |
update_lru_sizes(lruvec, lru, nr_zone_taken); |
1da177e4c Linux-2.6.12-rc2 |
1696 1697 |
return nr_taken; } |
62695a84e vmscan: move isol... |
1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 |
/** * isolate_lru_page - tries to isolate a page from its LRU list * @page: page to isolate from its LRU list * * Isolates a @page from an LRU list, clears PageLRU and adjusts the * vmstat statistic corresponding to whatever LRU list the page was on. * * Returns 0 if the page was removed from an LRU list. * Returns -EBUSY if the page was not on an LRU list. * * The returned page will have PageLRU() cleared. If it was found on |
894bc3104 Unevictable LRU I... |
1709 1710 1711 |
* the active list, it will have PageActive set. If it was found on * the unevictable list, it will have the PageUnevictable bit set. That flag * may need to be cleared by the caller before letting the page go. |
62695a84e vmscan: move isol... |
1712 1713 1714 1715 1716 |
* * The vmstat statistic corresponding to the list on which the page was * found will be decremented. * * Restrictions: |
a5d09bed7 mm: docs: add bla... |
1717 |
* |
62695a84e vmscan: move isol... |
1718 1719 1720 1721 1722 1723 1724 1725 1726 |
* (1) Must be called with an elevated refcount on the page. This is a * fundamentnal difference from isolate_lru_pages (which is called * without a stable reference). * (2) the lru_lock must not be held. * (3) interrupts must be enabled. */ int isolate_lru_page(struct page *page) { int ret = -EBUSY; |
309381fea mm: dump page whe... |
1727 |
VM_BUG_ON_PAGE(!page_count(page), page); |
cf2a82ee4 mm: downgrade VM_... |
1728 |
WARN_RATELIMIT(PageTail(page), "trying to isolate tail page"); |
0c917313a mm: strictly requ... |
1729 |
|
62695a84e vmscan: move isol... |
1730 |
if (PageLRU(page)) { |
f4b7e272b mm: remove zone_l... |
1731 |
pg_data_t *pgdat = page_pgdat(page); |
fa9add641 mm/memcg: apply a... |
1732 |
struct lruvec *lruvec; |
62695a84e vmscan: move isol... |
1733 |
|
f4b7e272b mm: remove zone_l... |
1734 1735 |
spin_lock_irq(&pgdat->lru_lock); lruvec = mem_cgroup_page_lruvec(page, pgdat); |
0c917313a mm: strictly requ... |
1736 |
if (PageLRU(page)) { |
894bc3104 Unevictable LRU I... |
1737 |
int lru = page_lru(page); |
0c917313a mm: strictly requ... |
1738 |
get_page(page); |
62695a84e vmscan: move isol... |
1739 |
ClearPageLRU(page); |
fa9add641 mm/memcg: apply a... |
1740 1741 |
del_page_from_lru_list(page, lruvec, lru); ret = 0; |
62695a84e vmscan: move isol... |
1742 |
} |
f4b7e272b mm: remove zone_l... |
1743 |
spin_unlock_irq(&pgdat->lru_lock); |
62695a84e vmscan: move isol... |
1744 1745 1746 |
} return ret; } |
5ad333eb6 Lumpy Reclaim V4 |
1747 |
/* |
d37dd5dcb vmscan: comment t... |
1748 1749 1750 1751 1752 |
* A direct reclaimer may isolate SWAP_CLUSTER_MAX pages from the LRU list and * then get resheduled. When there are massive number of tasks doing page * allocation, such sleeping direct reclaimers may keep piling up on each CPU, * the LRU list will go small and be scanned faster than necessary, leading to * unnecessary swapping, thrashing and OOM. |
35cd78156 vmscan: throttle ... |
1753 |
*/ |
599d0c954 mm, vmscan: move ... |
1754 |
static int too_many_isolated(struct pglist_data *pgdat, int file, |
35cd78156 vmscan: throttle ... |
1755 1756 1757 1758 1759 1760 |
struct scan_control *sc) { unsigned long inactive, isolated; if (current_is_kswapd()) return 0; |
97c9341f7 mm: vmscan: disab... |
1761 |
if (!sane_reclaim(sc)) |
35cd78156 vmscan: throttle ... |
1762 1763 1764 |
return 0; if (file) { |
599d0c954 mm, vmscan: move ... |
1765 1766 |
inactive = node_page_state(pgdat, NR_INACTIVE_FILE); isolated = node_page_state(pgdat, NR_ISOLATED_FILE); |
35cd78156 vmscan: throttle ... |
1767 |
} else { |
599d0c954 mm, vmscan: move ... |
1768 1769 |
inactive = node_page_state(pgdat, NR_INACTIVE_ANON); isolated = node_page_state(pgdat, NR_ISOLATED_ANON); |
35cd78156 vmscan: throttle ... |
1770 |
} |
3cf23841b mm/vmscan.c: avoi... |
1771 1772 1773 1774 1775 |
/* * GFP_NOIO/GFP_NOFS callers are allowed to isolate more pages, so they * won't get blocked by normal direct-reclaimers, forming a circular * deadlock. */ |
d0164adc8 mm, page_alloc: d... |
1776 |
if ((sc->gfp_mask & (__GFP_IO | __GFP_FS)) == (__GFP_IO | __GFP_FS)) |
3cf23841b mm/vmscan.c: avoi... |
1777 |
inactive >>= 3; |
35cd78156 vmscan: throttle ... |
1778 1779 |
return isolated > inactive; } |
a222f3415 mm: generalize pu... |
1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 |
/* * This moves pages from @list to corresponding LRU list. * * We move them the other way if the page is referenced by one or more * processes, from rmap. * * If the pages are mostly unmapped, the processing is fast and it is * appropriate to hold zone_lru_lock across the whole operation. But if * the pages are mapped, the processing is slow (page_referenced()) so we * should drop zone_lru_lock around each page. It's impossible to balance * this, so instead we remove the pages from the LRU while processing them. * It is safe to rely on PG_active against the non-LRU pages in here because * nobody will play with that bit on a non-LRU page. * * The downside is that we have to touch page->_refcount against each page. * But we had to alter page->flags anyway. * * Returns the number of pages moved to the given lruvec. */ static unsigned noinline_for_stack move_pages_to_lru(struct lruvec *lruvec, struct list_head *list) |
666356297 vmscan: set up pa... |
1802 |
{ |
599d0c954 mm, vmscan: move ... |
1803 |
struct pglist_data *pgdat = lruvec_pgdat(lruvec); |
a222f3415 mm: generalize pu... |
1804 |
int nr_pages, nr_moved = 0; |
3f79768f2 mm: rearrange put... |
1805 |
LIST_HEAD(pages_to_free); |
a222f3415 mm: generalize pu... |
1806 1807 |
struct page *page; enum lru_list lru; |
666356297 vmscan: set up pa... |
1808 |
|
a222f3415 mm: generalize pu... |
1809 1810 |
while (!list_empty(list)) { page = lru_to_page(list); |
309381fea mm: dump page whe... |
1811 |
VM_BUG_ON_PAGE(PageLRU(page), page); |
39b5f29ac mm: remove vma ar... |
1812 |
if (unlikely(!page_evictable(page))) { |
a222f3415 mm: generalize pu... |
1813 |
list_del(&page->lru); |
599d0c954 mm, vmscan: move ... |
1814 |
spin_unlock_irq(&pgdat->lru_lock); |
666356297 vmscan: set up pa... |
1815 |
putback_lru_page(page); |
599d0c954 mm, vmscan: move ... |
1816 |
spin_lock_irq(&pgdat->lru_lock); |
666356297 vmscan: set up pa... |
1817 1818 |
continue; } |
599d0c954 mm, vmscan: move ... |
1819 |
lruvec = mem_cgroup_page_lruvec(page, pgdat); |
fa9add641 mm/memcg: apply a... |
1820 |
|
7a608572a Revert "mm: batch... |
1821 |
SetPageLRU(page); |
666356297 vmscan: set up pa... |
1822 |
lru = page_lru(page); |
a222f3415 mm: generalize pu... |
1823 1824 1825 1826 |
nr_pages = hpage_nr_pages(page); update_lru_size(lruvec, lru, page_zonenum(page), nr_pages); list_move(&page->lru, &lruvec->lists[lru]); |
fa9add641 mm/memcg: apply a... |
1827 |
|
2bcf88796 mm: take pagevecs... |
1828 1829 1830 |
if (put_page_testzero(page)) { __ClearPageLRU(page); __ClearPageActive(page); |
fa9add641 mm/memcg: apply a... |
1831 |
del_page_from_lru_list(page, lruvec, lru); |
2bcf88796 mm: take pagevecs... |
1832 1833 |
if (unlikely(PageCompound(page))) { |
599d0c954 mm, vmscan: move ... |
1834 |
spin_unlock_irq(&pgdat->lru_lock); |
2bcf88796 mm: take pagevecs... |
1835 |
(*get_compound_page_dtor(page))(page); |
599d0c954 mm, vmscan: move ... |
1836 |
spin_lock_irq(&pgdat->lru_lock); |
2bcf88796 mm: take pagevecs... |
1837 1838 |
} else list_add(&page->lru, &pages_to_free); |
a222f3415 mm: generalize pu... |
1839 1840 |
} else { nr_moved += nr_pages; |
666356297 vmscan: set up pa... |
1841 1842 |
} } |
666356297 vmscan: set up pa... |
1843 |
|
3f79768f2 mm: rearrange put... |
1844 1845 1846 |
/* * To save our caller's stack, now use input list for pages to free. */ |
a222f3415 mm: generalize pu... |
1847 1848 1849 |
list_splice(&pages_to_free, list); return nr_moved; |
666356297 vmscan: set up pa... |
1850 1851 1852 |
} /* |
399ba0b95 mm/vmscan.c: avoi... |
1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 |
* If a kernel thread (such as nfsd for loop-back mounts) services * a backing device by writing to the page cache it sets PF_LESS_THROTTLE. * In that case we should only throttle if the backing device it is * writing to is congested. In other cases it is safe to throttle. */ static int current_may_throttle(void) { return !(current->flags & PF_LESS_THROTTLE) || current->backing_dev_info == NULL || bdi_write_congested(current->backing_dev_info); } /* |
b2e18757f mm, vmscan: begin... |
1866 |
* shrink_inactive_list() is a helper for shrink_node(). It returns the number |
1742f19fa [PATCH] vmscan: r... |
1867 |
* of reclaimed pages |
1da177e4c Linux-2.6.12-rc2 |
1868 |
*/ |
666356297 vmscan: set up pa... |
1869 |
static noinline_for_stack unsigned long |
1a93be0e7 mm/vmscan: push l... |
1870 |
shrink_inactive_list(unsigned long nr_to_scan, struct lruvec *lruvec, |
9e3b2f8cd mm/vmscan: store ... |
1871 |
struct scan_control *sc, enum lru_list lru) |
1da177e4c Linux-2.6.12-rc2 |
1872 1873 |
{ LIST_HEAD(page_list); |
e247dbce5 vmscan: simplify ... |
1874 |
unsigned long nr_scanned; |
05ff51376 [PATCH] vmscan re... |
1875 |
unsigned long nr_reclaimed = 0; |
e247dbce5 vmscan: simplify ... |
1876 |
unsigned long nr_taken; |
060f005f0 mm/vmscan.c: do n... |
1877 |
struct reclaim_stat stat; |
3cb994517 mm: push lru inde... |
1878 |
int file = is_file_lru(lru); |
f46b79120 mm/vmscan.c: simp... |
1879 |
enum vm_event_item item; |
599d0c954 mm, vmscan: move ... |
1880 |
struct pglist_data *pgdat = lruvec_pgdat(lruvec); |
1a93be0e7 mm/vmscan: push l... |
1881 |
struct zone_reclaim_stat *reclaim_stat = &lruvec->reclaim_stat; |
db73ee0d4 mm, vmscan: do no... |
1882 |
bool stalled = false; |
78dc583d3 vmscan: low order... |
1883 |
|
599d0c954 mm, vmscan: move ... |
1884 |
while (unlikely(too_many_isolated(pgdat, file, sc))) { |
db73ee0d4 mm, vmscan: do no... |
1885 1886 1887 1888 1889 1890 |
if (stalled) return 0; /* wait a bit for the reclaimer. */ msleep(100); stalled = true; |
35cd78156 vmscan: throttle ... |
1891 1892 1893 1894 1895 |
/* We are about to die and free our memory. Return now. */ if (fatal_signal_pending(current)) return SWAP_CLUSTER_MAX; } |
1da177e4c Linux-2.6.12-rc2 |
1896 |
lru_add_drain(); |
f80c06736 mm: zone_reclaim:... |
1897 |
|
599d0c954 mm, vmscan: move ... |
1898 |
spin_lock_irq(&pgdat->lru_lock); |
b35ea17b7 mm: shrink_inacti... |
1899 |
|
5dc35979e mm/vmscan: push l... |
1900 |
nr_taken = isolate_lru_pages(nr_to_scan, lruvec, &page_list, |
a9e7c39fa mm/vmscan.c: remo... |
1901 |
&nr_scanned, sc, lru); |
95d918fc0 mm/vmscan: remove... |
1902 |
|
599d0c954 mm, vmscan: move ... |
1903 |
__mod_node_page_state(pgdat, NR_ISOLATED_ANON + file, nr_taken); |
9d5e6a9f2 mm: update_lru_si... |
1904 |
reclaim_stat->recent_scanned[file] += nr_taken; |
95d918fc0 mm/vmscan: remove... |
1905 |
|
f46b79120 mm/vmscan.c: simp... |
1906 1907 1908 1909 |
item = current_is_kswapd() ? PGSCAN_KSWAPD : PGSCAN_DIRECT; if (global_reclaim(sc)) __count_vm_events(item, nr_scanned); __count_memcg_events(lruvec_memcg(lruvec), item, nr_scanned); |
599d0c954 mm, vmscan: move ... |
1910 |
spin_unlock_irq(&pgdat->lru_lock); |
b35ea17b7 mm: shrink_inacti... |
1911 |
|
d563c0501 vmscan: handle is... |
1912 |
if (nr_taken == 0) |
666356297 vmscan: set up pa... |
1913 |
return 0; |
5ad333eb6 Lumpy Reclaim V4 |
1914 |
|
a128ca71f mm: delete unnece... |
1915 |
nr_reclaimed = shrink_page_list(&page_list, pgdat, sc, 0, |
3c710c1ad mm, vmscan: extra... |
1916 |
&stat, false); |
c661b078f synchronous lumpy... |
1917 |
|
599d0c954 mm, vmscan: move ... |
1918 |
spin_lock_irq(&pgdat->lru_lock); |
3f79768f2 mm: rearrange put... |
1919 |
|
f46b79120 mm/vmscan.c: simp... |
1920 1921 1922 1923 |
item = current_is_kswapd() ? PGSTEAL_KSWAPD : PGSTEAL_DIRECT; if (global_reclaim(sc)) __count_vm_events(item, nr_reclaimed); __count_memcg_events(lruvec_memcg(lruvec), item, nr_reclaimed); |
b17f18aff mm/vmscan.c: fix ... |
1924 1925 |
reclaim_stat->recent_rotated[0] += stat.nr_activate[0]; reclaim_stat->recent_rotated[1] += stat.nr_activate[1]; |
a74609faf [PATCH] mm: page_... |
1926 |
|
a222f3415 mm: generalize pu... |
1927 |
move_pages_to_lru(lruvec, &page_list); |
3f79768f2 mm: rearrange put... |
1928 |
|
599d0c954 mm, vmscan: move ... |
1929 |
__mod_node_page_state(pgdat, NR_ISOLATED_ANON + file, -nr_taken); |
3f79768f2 mm: rearrange put... |
1930 |
|
599d0c954 mm, vmscan: move ... |
1931 |
spin_unlock_irq(&pgdat->lru_lock); |
3f79768f2 mm: rearrange put... |
1932 |
|
747db954c mm: memcontrol: u... |
1933 |
mem_cgroup_uncharge_list(&page_list); |
2d4894b5d mm: remove cold p... |
1934 |
free_unref_page_list(&page_list); |
e11da5b4f tracing, vmscan: ... |
1935 |
|
92df3a723 mm: vmscan: throt... |
1936 |
/* |
1c610d5f9 mm/vmscan: wake u... |
1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 |
* If dirty pages are scanned that are not queued for IO, it * implies that flushers are not doing their job. This can * happen when memory pressure pushes dirty pages to the end of * the LRU before the dirty limits are breached and the dirty * data has expired. It can also happen when the proportion of * dirty pages grows not through writes but through memory * pressure reclaiming all the clean cache. And in some cases, * the flushers simply cannot keep up with the allocation * rate. Nudge the flusher threads in case they are asleep. */ if (stat.nr_unqueued_dirty == nr_taken) wakeup_flusher_threads(WB_REASON_VMSCAN); |
d108c7721 mm/vmscan: don't ... |
1949 1950 1951 1952 1953 1954 1955 1956 |
sc->nr.dirty += stat.nr_dirty; sc->nr.congested += stat.nr_congested; sc->nr.unqueued_dirty += stat.nr_unqueued_dirty; sc->nr.writeback += stat.nr_writeback; sc->nr.immediate += stat.nr_immediate; sc->nr.taken += nr_taken; if (file) sc->nr.file_taken += nr_taken; |
8e9502828 mm: vmscan: move ... |
1957 |
|
599d0c954 mm, vmscan: move ... |
1958 |
trace_mm_vmscan_lru_shrink_inactive(pgdat->node_id, |
d51d1e645 mm, vmscan, traci... |
1959 |
nr_scanned, nr_reclaimed, &stat, sc->priority, file); |
05ff51376 [PATCH] vmscan re... |
1960 |
return nr_reclaimed; |
1da177e4c Linux-2.6.12-rc2 |
1961 |
} |
f626012db mm: remove isolat... |
1962 |
static void shrink_active_list(unsigned long nr_to_scan, |
1a93be0e7 mm/vmscan: push l... |
1963 |
struct lruvec *lruvec, |
f16015fbf mm: vmscan: disti... |
1964 |
struct scan_control *sc, |
9e3b2f8cd mm/vmscan: store ... |
1965 |
enum lru_list lru) |
1da177e4c Linux-2.6.12-rc2 |
1966 |
{ |
44c241f16 mm: rename pgmove... |
1967 |
unsigned long nr_taken; |
f626012db mm: remove isolat... |
1968 |
unsigned long nr_scanned; |
6fe6b7e35 vmscan: report vm... |
1969 |
unsigned long vm_flags; |
1da177e4c Linux-2.6.12-rc2 |
1970 |
LIST_HEAD(l_hold); /* The pages which were snipped off */ |
8cab4754d vmscan: make mapp... |
1971 |
LIST_HEAD(l_active); |
b69408e88 vmscan: Use an in... |
1972 |
LIST_HEAD(l_inactive); |
1da177e4c Linux-2.6.12-rc2 |
1973 |
struct page *page; |
1a93be0e7 mm/vmscan: push l... |
1974 |
struct zone_reclaim_stat *reclaim_stat = &lruvec->reclaim_stat; |
9d998b4f1 mm, vmscan: add a... |
1975 1976 |
unsigned nr_deactivate, nr_activate; unsigned nr_rotated = 0; |
3cb994517 mm: push lru inde... |
1977 |
int file = is_file_lru(lru); |
599d0c954 mm, vmscan: move ... |
1978 |
struct pglist_data *pgdat = lruvec_pgdat(lruvec); |
1da177e4c Linux-2.6.12-rc2 |
1979 1980 |
lru_add_drain(); |
f80c06736 mm: zone_reclaim:... |
1981 |
|
599d0c954 mm, vmscan: move ... |
1982 |
spin_lock_irq(&pgdat->lru_lock); |
925b7673c mm: make per-memc... |
1983 |
|
5dc35979e mm/vmscan: push l... |
1984 |
nr_taken = isolate_lru_pages(nr_to_scan, lruvec, &l_hold, |
a9e7c39fa mm/vmscan.c: remo... |
1985 |
&nr_scanned, sc, lru); |
89b5fae53 mm: vmscan: disti... |
1986 |
|
599d0c954 mm, vmscan: move ... |
1987 |
__mod_node_page_state(pgdat, NR_ISOLATED_ANON + file, nr_taken); |
b7c46d151 mm: drop unneeded... |
1988 |
reclaim_stat->recent_scanned[file] += nr_taken; |
1cfb419b3 per-zone and recl... |
1989 |
|
599d0c954 mm, vmscan: move ... |
1990 |
__count_vm_events(PGREFILL, nr_scanned); |
2fa2690ca mm/vmscan.c: don'... |
1991 |
__count_memcg_events(lruvec_memcg(lruvec), PGREFILL, nr_scanned); |
9d5e6a9f2 mm: update_lru_si... |
1992 |
|
599d0c954 mm, vmscan: move ... |
1993 |
spin_unlock_irq(&pgdat->lru_lock); |
1da177e4c Linux-2.6.12-rc2 |
1994 |
|
1da177e4c Linux-2.6.12-rc2 |
1995 1996 1997 1998 |
while (!list_empty(&l_hold)) { cond_resched(); page = lru_to_page(&l_hold); list_del(&page->lru); |
7e9cd4842 vmscan: fix pagec... |
1999 |
|
39b5f29ac mm: remove vma ar... |
2000 |
if (unlikely(!page_evictable(page))) { |
894bc3104 Unevictable LRU I... |
2001 2002 2003 |
putback_lru_page(page); continue; } |
cc715d99e mm: vmscan: forci... |
2004 2005 2006 2007 2008 2009 2010 |
if (unlikely(buffer_heads_over_limit)) { if (page_has_private(page) && trylock_page(page)) { if (page_has_private(page)) try_to_release_page(page, 0); unlock_page(page); } } |
c3ac9a8ad mm: memcg: count ... |
2011 2012 |
if (page_referenced(page, 0, sc->target_mem_cgroup, &vm_flags)) { |
9992af102 thp: scale nr_rot... |
2013 |
nr_rotated += hpage_nr_pages(page); |
8cab4754d vmscan: make mapp... |
2014 2015 2016 2017 2018 2019 2020 2021 2022 |
/* * Identify referenced, file-backed active pages and * give them one more trip around the active list. So * that executable code get better chances to stay in * memory under moderate memory pressure. Anon pages * are not likely to be evicted by use-once streaming * IO, plus JVM can create lots of anon VM_EXEC pages, * so we ignore them here. */ |
41e20983f vmscan: limit VM_... |
2023 |
if ((vm_flags & VM_EXEC) && page_is_file_cache(page)) { |
8cab4754d vmscan: make mapp... |
2024 2025 2026 2027 |
list_add(&page->lru, &l_active); continue; } } |
7e9cd4842 vmscan: fix pagec... |
2028 |
|
5205e56ee vmscan: move Clea... |
2029 |
ClearPageActive(page); /* we are de-activating */ |
1899ad18c mm: workingset: t... |
2030 |
SetPageWorkingset(page); |
1da177e4c Linux-2.6.12-rc2 |
2031 2032 |
list_add(&page->lru, &l_inactive); } |
b555749aa vmscan: shrink_ac... |
2033 |
/* |
8cab4754d vmscan: make mapp... |
2034 |
* Move pages back to the lru list. |
b555749aa vmscan: shrink_ac... |
2035 |
*/ |
599d0c954 mm, vmscan: move ... |
2036 |
spin_lock_irq(&pgdat->lru_lock); |
4f98a2fee vmscan: split LRU... |
2037 |
/* |
8cab4754d vmscan: make mapp... |
2038 2039 2040 |
* Count referenced pages from currently used mappings as rotated, * even though only some of them are actually re-activated. This * helps balance scan pressure between file and anonymous pages in |
7c0db9e91 mm, vmscan: fix a... |
2041 |
* get_scan_count. |
7e9cd4842 vmscan: fix pagec... |
2042 |
*/ |
b7c46d151 mm: drop unneeded... |
2043 |
reclaim_stat->recent_rotated[file] += nr_rotated; |
556adecba vmscan: second ch... |
2044 |
|
a222f3415 mm: generalize pu... |
2045 2046 |
nr_activate = move_pages_to_lru(lruvec, &l_active); nr_deactivate = move_pages_to_lru(lruvec, &l_inactive); |
f372d89e5 mm: remove pages_... |
2047 2048 |
/* Keep all free pages in l_active list */ list_splice(&l_inactive, &l_active); |
9851ac135 mm: move nr_deact... |
2049 2050 2051 |
__count_vm_events(PGDEACTIVATE, nr_deactivate); __count_memcg_events(lruvec_memcg(lruvec), PGDEACTIVATE, nr_deactivate); |
599d0c954 mm, vmscan: move ... |
2052 2053 |
__mod_node_page_state(pgdat, NR_ISOLATED_ANON + file, -nr_taken); spin_unlock_irq(&pgdat->lru_lock); |
2bcf88796 mm: take pagevecs... |
2054 |
|
f372d89e5 mm: remove pages_... |
2055 2056 |
mem_cgroup_uncharge_list(&l_active); free_unref_page_list(&l_active); |
9d998b4f1 mm, vmscan: add a... |
2057 2058 |
trace_mm_vmscan_lru_shrink_active(pgdat->node_id, nr_taken, nr_activate, nr_deactivate, nr_rotated, sc->priority, file); |
1da177e4c Linux-2.6.12-rc2 |
2059 |
} |
1a4e58cce mm: introduce MAD... |
2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 |
unsigned long reclaim_pages(struct list_head *page_list) { int nid = -1; unsigned long nr_reclaimed = 0; LIST_HEAD(node_page_list); struct reclaim_stat dummy_stat; struct page *page; struct scan_control sc = { .gfp_mask = GFP_KERNEL, .priority = DEF_PRIORITY, .may_writepage = 1, .may_unmap = 1, .may_swap = 1, }; while (!list_empty(page_list)) { page = lru_to_page(page_list); if (nid == -1) { nid = page_to_nid(page); INIT_LIST_HEAD(&node_page_list); } if (nid == page_to_nid(page)) { ClearPageActive(page); list_move(&page->lru, &node_page_list); continue; } nr_reclaimed += shrink_page_list(&node_page_list, NODE_DATA(nid), &sc, 0, &dummy_stat, false); while (!list_empty(&node_page_list)) { page = lru_to_page(&node_page_list); list_del(&page->lru); putback_lru_page(page); } nid = -1; } if (!list_empty(&node_page_list)) { nr_reclaimed += shrink_page_list(&node_page_list, NODE_DATA(nid), &sc, 0, &dummy_stat, false); while (!list_empty(&node_page_list)) { page = lru_to_page(&node_page_list); list_del(&page->lru); putback_lru_page(page); } } return nr_reclaimed; } |
59dc76b0d mm: vmscan: reduc... |
2115 2116 2117 |
/* * The inactive anon list should be small enough that the VM never has * to do too much work. |
14797e236 memcg: add inacti... |
2118 |
* |
59dc76b0d mm: vmscan: reduc... |
2119 2120 2121 |
* The inactive file list should be small enough to leave most memory * to the established workingset on the scan-resistant active list, * but large enough to avoid thrashing the aggregate readahead window. |
56e49d218 vmscan: evict use... |
2122 |
* |
59dc76b0d mm: vmscan: reduc... |
2123 2124 |
* Both inactive lists should also be large enough that each inactive * page has a chance to be referenced again before it is reclaimed. |
56e49d218 vmscan: evict use... |
2125 |
* |
2a2e48854 mm: vmscan: fix I... |
2126 2127 |
* If that fails and refaulting is observed, the inactive list grows. * |
59dc76b0d mm: vmscan: reduc... |
2128 |
* The inactive_ratio is the target ratio of ACTIVE to INACTIVE pages |
3a50d14d0 mm: remove unused... |
2129 |
* on this LRU, maintained by the pageout code. An inactive_ratio |
59dc76b0d mm: vmscan: reduc... |
2130 |
* of 3 means 3:1 or 25% of the pages are kept on the inactive list. |
56e49d218 vmscan: evict use... |
2131 |
* |
59dc76b0d mm: vmscan: reduc... |
2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 |
* total target max * memory ratio inactive * ------------------------------------- * 10MB 1 5MB * 100MB 1 50MB * 1GB 3 250MB * 10GB 10 0.9GB * 100GB 31 3GB * 1TB 101 10GB * 10TB 320 32GB |
56e49d218 vmscan: evict use... |
2142 |
*/ |
f8d1a3116 mm: consider whet... |
2143 |
static bool inactive_list_is_low(struct lruvec *lruvec, bool file, |
2c012a4ad mm: vmscan: scan ... |
2144 |
struct scan_control *sc, bool trace) |
56e49d218 vmscan: evict use... |
2145 |
{ |
fd5388037 mm, vmscan: clean... |
2146 |
enum lru_list active_lru = file * LRU_FILE + LRU_ACTIVE; |
2a2e48854 mm: vmscan: fix I... |
2147 2148 2149 2150 2151 |
struct pglist_data *pgdat = lruvec_pgdat(lruvec); enum lru_list inactive_lru = file * LRU_FILE; unsigned long inactive, active; unsigned long inactive_ratio; unsigned long refaults; |
59dc76b0d mm: vmscan: reduc... |
2152 |
unsigned long gb; |
e3790144c mm: refactor inac... |
2153 |
|
59dc76b0d mm: vmscan: reduc... |
2154 2155 2156 2157 2158 2159 |
/* * If we don't have swap space, anonymous page deactivation * is pointless. */ if (!file && !total_swap_pages) return false; |
56e49d218 vmscan: evict use... |
2160 |
|
fd5388037 mm, vmscan: clean... |
2161 2162 |
inactive = lruvec_lru_size(lruvec, inactive_lru, sc->reclaim_idx); active = lruvec_lru_size(lruvec, active_lru, sc->reclaim_idx); |
f8d1a3116 mm: consider whet... |
2163 |
|
2a2e48854 mm: vmscan: fix I... |
2164 2165 2166 2167 2168 |
/* * When refaults are being observed, it means a new workingset * is being established. Disable active list protection to get * rid of the stale workingset quickly. */ |
205b20cc5 mm: memcontrol: m... |
2169 |
refaults = lruvec_page_state_local(lruvec, WORKINGSET_ACTIVATE); |
2c012a4ad mm: vmscan: scan ... |
2170 |
if (file && lruvec->refaults != refaults) { |
2a2e48854 mm: vmscan: fix I... |
2171 2172 2173 2174 2175 2176 2177 2178 |
inactive_ratio = 0; } else { gb = (inactive + active) >> (30 - PAGE_SHIFT); if (gb) inactive_ratio = int_sqrt(10 * gb); else inactive_ratio = 1; } |
59dc76b0d mm: vmscan: reduc... |
2179 |
|
2c012a4ad mm: vmscan: scan ... |
2180 |
if (trace) |
2a2e48854 mm: vmscan: fix I... |
2181 2182 2183 2184 |
trace_mm_vmscan_inactive_list_is_low(pgdat->node_id, sc->reclaim_idx, lruvec_lru_size(lruvec, inactive_lru, MAX_NR_ZONES), inactive, lruvec_lru_size(lruvec, active_lru, MAX_NR_ZONES), active, inactive_ratio, file); |
fd5388037 mm, vmscan: clean... |
2185 |
|
59dc76b0d mm: vmscan: reduc... |
2186 |
return inactive * inactive_ratio < active; |
b39415b27 vmscan: do not ev... |
2187 |
} |
4f98a2fee vmscan: split LRU... |
2188 |
static unsigned long shrink_list(enum lru_list lru, unsigned long nr_to_scan, |
3b991208b mm: fix inactive ... |
2189 |
struct lruvec *lruvec, struct scan_control *sc) |
b69408e88 vmscan: Use an in... |
2190 |
{ |
b39415b27 vmscan: do not ev... |
2191 |
if (is_active_lru(lru)) { |
3b991208b mm: fix inactive ... |
2192 |
if (inactive_list_is_low(lruvec, is_file_lru(lru), sc, true)) |
1a93be0e7 mm/vmscan: push l... |
2193 |
shrink_active_list(nr_to_scan, lruvec, sc, lru); |
556adecba vmscan: second ch... |
2194 2195 |
return 0; } |
1a93be0e7 mm/vmscan: push l... |
2196 |
return shrink_inactive_list(nr_to_scan, lruvec, sc, lru); |
4f98a2fee vmscan: split LRU... |
2197 |
} |
9a2651140 mm: vmscan: clean... |
2198 2199 2200 2201 2202 2203 |
enum scan_balance { SCAN_EQUAL, SCAN_FRACT, SCAN_ANON, SCAN_FILE, }; |
4f98a2fee vmscan: split LRU... |
2204 2205 2206 2207 2208 2209 |
/* * Determine how aggressively the anon and file LRU lists should be * scanned. The relative value of each set of LRU lists is determined * by looking at the fraction of the pages scanned we did rotate back * onto the active list instead of evict. * |
be7bd59db mm: fix page recl... |
2210 2211 |
* nr[0] = anon inactive pages to scan; nr[1] = anon active pages to scan * nr[2] = file inactive pages to scan; nr[3] = file active pages to scan |
4f98a2fee vmscan: split LRU... |
2212 |
*/ |
333776785 mm: vmscan: pass ... |
2213 |
static void get_scan_count(struct lruvec *lruvec, struct mem_cgroup *memcg, |
6b4f7799c mm: vmscan: invok... |
2214 2215 |
struct scan_control *sc, unsigned long *nr, unsigned long *lru_pages) |
4f98a2fee vmscan: split LRU... |
2216 |
{ |
333776785 mm: vmscan: pass ... |
2217 |
int swappiness = mem_cgroup_swappiness(memcg); |
9a2651140 mm: vmscan: clean... |
2218 2219 2220 |
struct zone_reclaim_stat *reclaim_stat = &lruvec->reclaim_stat; u64 fraction[2]; u64 denominator = 0; /* gcc */ |
599d0c954 mm, vmscan: move ... |
2221 |
struct pglist_data *pgdat = lruvec_pgdat(lruvec); |
4f98a2fee vmscan: split LRU... |
2222 |
unsigned long anon_prio, file_prio; |
9a2651140 mm: vmscan: clean... |
2223 |
enum scan_balance scan_balance; |
0bf1457f0 mm: vmscan: do no... |
2224 |
unsigned long anon, file; |
4f98a2fee vmscan: split LRU... |
2225 |
unsigned long ap, fp; |
4111304da mm: enum lru_list... |
2226 |
enum lru_list lru; |
76a33fc38 vmscan: prevent g... |
2227 2228 |
/* If we have no swap space, do not bother scanning anon pages. */ |
d8b38438a mm: vmscan: do no... |
2229 |
if (!sc->may_swap || mem_cgroup_get_nr_swap_pages(memcg) <= 0) { |
9a2651140 mm: vmscan: clean... |
2230 |
scan_balance = SCAN_FILE; |
76a33fc38 vmscan: prevent g... |
2231 2232 |
goto out; } |
4f98a2fee vmscan: split LRU... |
2233 |
|
10316b313 mm: vmscan: clari... |
2234 2235 2236 2237 2238 2239 2240 |
/* * Global reclaim will swap to prevent OOM even with no * swappiness, but memcg users want to use this knob to * disable swapping for individual groups completely when * using the memory controller's swap limit feature would be * too expensive. */ |
02695175c mm: vmscan: move ... |
2241 |
if (!global_reclaim(sc) && !swappiness) { |
9a2651140 mm: vmscan: clean... |
2242 |
scan_balance = SCAN_FILE; |
10316b313 mm: vmscan: clari... |
2243 2244 2245 2246 2247 2248 2249 2250 |
goto out; } /* * Do not apply any pressure balancing cleverness when the * system is close to OOM, scan both anon and file equally * (unless the swappiness setting disagrees with swapping). */ |
02695175c mm: vmscan: move ... |
2251 |
if (!sc->priority && swappiness) { |
9a2651140 mm: vmscan: clean... |
2252 |
scan_balance = SCAN_EQUAL; |
10316b313 mm: vmscan: clari... |
2253 2254 |
goto out; } |
11d16c25b mm: vmscan: impro... |
2255 |
/* |
623762517 revert "mm: vmsca... |
2256 2257 2258 2259 2260 2261 2262 2263 2264 |
* Prevent the reclaimer from falling into the cache trap: as * cache pages start out inactive, every cache fault will tip * the scan balance towards the file LRU. And as the file LRU * shrinks, so does the window for rotation from references. * This means we have a runaway feedback loop where a tiny * thrashing file LRU becomes infinitely more attractive than * anon pages. Try to detect this based on file LRU size. */ if (global_reclaim(sc)) { |
599d0c954 mm, vmscan: move ... |
2265 2266 2267 2268 |
unsigned long pgdatfile; unsigned long pgdatfree; int z; unsigned long total_high_wmark = 0; |
2ab051e11 memcg, vmscan: Fi... |
2269 |
|
599d0c954 mm, vmscan: move ... |
2270 2271 2272 2273 2274 2275 |
pgdatfree = sum_zone_node_page_state(pgdat->node_id, NR_FREE_PAGES); pgdatfile = node_page_state(pgdat, NR_ACTIVE_FILE) + node_page_state(pgdat, NR_INACTIVE_FILE); for (z = 0; z < MAX_NR_ZONES; z++) { struct zone *zone = &pgdat->node_zones[z]; |
6aa303def mm, vmscan: only ... |
2276 |
if (!managed_zone(zone)) |
599d0c954 mm, vmscan: move ... |
2277 2278 2279 2280 |
continue; total_high_wmark += high_wmark_pages(zone); } |
623762517 revert "mm: vmsca... |
2281 |
|
599d0c954 mm, vmscan: move ... |
2282 |
if (unlikely(pgdatfile + pgdatfree <= total_high_wmark)) { |
062262267 mm, vmscan: avoid... |
2283 2284 2285 2286 2287 |
/* * Force SCAN_ANON if there are enough inactive * anonymous pages on the LRU in eligible zones. * Otherwise, the small LRU gets thrashed. */ |
3b991208b mm: fix inactive ... |
2288 |
if (!inactive_list_is_low(lruvec, false, sc, false) && |
062262267 mm, vmscan: avoid... |
2289 2290 2291 2292 2293 |
lruvec_lru_size(lruvec, LRU_INACTIVE_ANON, sc->reclaim_idx) >> sc->priority) { scan_balance = SCAN_ANON; goto out; } |
623762517 revert "mm: vmsca... |
2294 2295 2296 2297 |
} } /* |
316bda0e6 vmscan: do not fo... |
2298 2299 2300 2301 2302 2303 2304 |
* If there is enough inactive page cache, i.e. if the size of the * inactive list is greater than that of the active list *and* the * inactive list actually has some pages to scan on this priority, we * do not reclaim anything from the anonymous working set right now. * Without the second condition we could end up never scanning an * lruvec even if it has plenty of old anonymous pages unless the * system is under heavy pressure. |
7c5bd705d mm: memcg: only e... |
2305 |
*/ |
3b991208b mm: fix inactive ... |
2306 |
if (!inactive_list_is_low(lruvec, true, sc, false) && |
71ab6cfe8 mm, vmscan: consi... |
2307 |
lruvec_lru_size(lruvec, LRU_INACTIVE_FILE, sc->reclaim_idx) >> sc->priority) { |
9a2651140 mm: vmscan: clean... |
2308 |
scan_balance = SCAN_FILE; |
7c5bd705d mm: memcg: only e... |
2309 2310 |
goto out; } |
9a2651140 mm: vmscan: clean... |
2311 |
scan_balance = SCAN_FRACT; |
7c5bd705d mm: memcg: only e... |
2312 |
/* |
58c37f6e0 vmscan: protect r... |
2313 2314 2315 |
* With swappiness at 100, anonymous and file have the same priority. * This scanning priority is essentially the inverse of IO cost. */ |
02695175c mm: vmscan: move ... |
2316 |
anon_prio = swappiness; |
75b00af77 mm: trivial clean... |
2317 |
file_prio = 200 - anon_prio; |
58c37f6e0 vmscan: protect r... |
2318 2319 |
/* |
4f98a2fee vmscan: split LRU... |
2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 |
* OK, so we have swap space and a fair amount of page cache * pages. We use the recently rotated / recently scanned * ratios to determine how valuable each cache is. * * Because workloads change over time (and to avoid overflow) * we keep these statistics as a floating average, which ends * up weighing recent references more than old ones. * * anon in [0], file in [1] */ |
2ab051e11 memcg, vmscan: Fi... |
2330 |
|
fd5388037 mm, vmscan: clean... |
2331 2332 2333 2334 |
anon = lruvec_lru_size(lruvec, LRU_ACTIVE_ANON, MAX_NR_ZONES) + lruvec_lru_size(lruvec, LRU_INACTIVE_ANON, MAX_NR_ZONES); file = lruvec_lru_size(lruvec, LRU_ACTIVE_FILE, MAX_NR_ZONES) + lruvec_lru_size(lruvec, LRU_INACTIVE_FILE, MAX_NR_ZONES); |
2ab051e11 memcg, vmscan: Fi... |
2335 |
|
599d0c954 mm, vmscan: move ... |
2336 |
spin_lock_irq(&pgdat->lru_lock); |
6e9015716 mm: introduce zon... |
2337 |
if (unlikely(reclaim_stat->recent_scanned[0] > anon / 4)) { |
6e9015716 mm: introduce zon... |
2338 2339 |
reclaim_stat->recent_scanned[0] /= 2; reclaim_stat->recent_rotated[0] /= 2; |
4f98a2fee vmscan: split LRU... |
2340 |
} |
6e9015716 mm: introduce zon... |
2341 |
if (unlikely(reclaim_stat->recent_scanned[1] > file / 4)) { |
6e9015716 mm: introduce zon... |
2342 2343 |
reclaim_stat->recent_scanned[1] /= 2; reclaim_stat->recent_rotated[1] /= 2; |
4f98a2fee vmscan: split LRU... |
2344 2345 2346 |
} /* |
00d8089c5 vmscan: fix get_s... |
2347 2348 2349 |
* The amount of pressure on anon vs file pages is inversely * proportional to the fraction of recently scanned pages on * each list that were recently referenced and in active use. |
4f98a2fee vmscan: split LRU... |
2350 |
*/ |
fe35004fb mm: avoid swappin... |
2351 |
ap = anon_prio * (reclaim_stat->recent_scanned[0] + 1); |
6e9015716 mm: introduce zon... |
2352 |
ap /= reclaim_stat->recent_rotated[0] + 1; |
4f98a2fee vmscan: split LRU... |
2353 |
|
fe35004fb mm: avoid swappin... |
2354 |
fp = file_prio * (reclaim_stat->recent_scanned[1] + 1); |
6e9015716 mm: introduce zon... |
2355 |
fp /= reclaim_stat->recent_rotated[1] + 1; |
599d0c954 mm, vmscan: move ... |
2356 |
spin_unlock_irq(&pgdat->lru_lock); |
4f98a2fee vmscan: split LRU... |
2357 |
|
76a33fc38 vmscan: prevent g... |
2358 2359 2360 2361 |
fraction[0] = ap; fraction[1] = fp; denominator = ap + fp + 1; out: |
688035f72 mm: don't avoid h... |
2362 2363 2364 |
*lru_pages = 0; for_each_evictable_lru(lru) { int file = is_file_lru(lru); |
9783aa991 mm, memcg: propor... |
2365 |
unsigned long lruvec_size; |
688035f72 mm: don't avoid h... |
2366 |
unsigned long scan; |
1bc63fb12 mm, memcg: make s... |
2367 |
unsigned long protection; |
9783aa991 mm, memcg: propor... |
2368 2369 |
lruvec_size = lruvec_lru_size(lruvec, lru, sc->reclaim_idx); |
1bc63fb12 mm, memcg: make s... |
2370 2371 |
protection = mem_cgroup_protection(memcg, sc->memcg_low_reclaim); |
9783aa991 mm, memcg: propor... |
2372 |
|
1bc63fb12 mm, memcg: make s... |
2373 |
if (protection) { |
9783aa991 mm, memcg: propor... |
2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 |
/* * Scale a cgroup's reclaim pressure by proportioning * its current usage to its memory.low or memory.min * setting. * * This is important, as otherwise scanning aggression * becomes extremely binary -- from nothing as we * approach the memory protection threshold, to totally * nominal as we exceed it. This results in requiring * setting extremely liberal protection thresholds. It * also means we simply get no protection at all if we * set it too low, which is not ideal. |
1bc63fb12 mm, memcg: make s... |
2386 2387 2388 2389 |
* * If there is any protection in place, we reduce scan * pressure by how much of the total memory used is * within protection thresholds. |
9783aa991 mm, memcg: propor... |
2390 |
* |
9de7ca46a mm, memcg: make m... |
2391 2392 2393 2394 2395 2396 2397 2398 |
* There is one special case: in the first reclaim pass, * we skip over all groups that are within their low * protection. If that fails to reclaim enough pages to * satisfy the reclaim goal, we come back and override * the best-effort low protection. However, we still * ideally want to honor how well-behaved groups are in * that case instead of simply punishing them all * equally. As such, we reclaim them based on how much |
1bc63fb12 mm, memcg: make s... |
2399 2400 2401 |
* memory they are using, reducing the scan pressure * again by how much of the total memory used is under * hard protection. |
9783aa991 mm, memcg: propor... |
2402 |
*/ |
1bc63fb12 mm, memcg: make s... |
2403 2404 2405 2406 2407 2408 2409 |
unsigned long cgroup_size = mem_cgroup_size(memcg); /* Avoid TOCTOU with earlier protection check */ cgroup_size = max(cgroup_size, protection); scan = lruvec_size - lruvec_size * protection / cgroup_size; |
9783aa991 mm, memcg: propor... |
2410 2411 |
/* |
1bc63fb12 mm, memcg: make s... |
2412 |
* Minimally target SWAP_CLUSTER_MAX pages to keep |
9de7ca46a mm, memcg: make m... |
2413 2414 |
* reclaim moving forwards, avoiding decremeting * sc->priority further than desirable. |
9783aa991 mm, memcg: propor... |
2415 |
*/ |
1bc63fb12 mm, memcg: make s... |
2416 |
scan = max(scan, SWAP_CLUSTER_MAX); |
9783aa991 mm, memcg: propor... |
2417 2418 2419 2420 2421 |
} else { scan = lruvec_size; } scan >>= sc->priority; |
6b4f7799c mm: vmscan: invok... |
2422 |
|
688035f72 mm: don't avoid h... |
2423 2424 2425 2426 2427 |
/* * If the cgroup's already been deleted, make sure to * scrape out the remaining cache. */ if (!scan && !mem_cgroup_online(memcg)) |
9783aa991 mm, memcg: propor... |
2428 |
scan = min(lruvec_size, SWAP_CLUSTER_MAX); |
6b4f7799c mm: vmscan: invok... |
2429 |
|
688035f72 mm: don't avoid h... |
2430 2431 2432 2433 2434 |
switch (scan_balance) { case SCAN_EQUAL: /* Scan lists relative to size */ break; case SCAN_FRACT: |
9a2651140 mm: vmscan: clean... |
2435 |
/* |
688035f72 mm: don't avoid h... |
2436 2437 |
* Scan types proportional to swappiness and * their relative recent reclaim efficiency. |
198f5aa0f mm/vmscan.c: don'... |
2438 2439 2440 |
* Make sure we don't miss the last page on * the offlined memory cgroups because of a * round-off error. |
9a2651140 mm: vmscan: clean... |
2441 |
*/ |
198f5aa0f mm/vmscan.c: don'... |
2442 2443 2444 |
scan = mem_cgroup_online(memcg) ? div64_u64(scan * fraction[file], denominator) : DIV64_U64_ROUND_UP(scan * fraction[file], |
68600f623 mm: don't miss th... |
2445 |
denominator); |
688035f72 mm: don't avoid h... |
2446 2447 2448 2449 2450 |
break; case SCAN_FILE: case SCAN_ANON: /* Scan one type exclusively */ if ((scan_balance == SCAN_FILE) != file) { |
9783aa991 mm, memcg: propor... |
2451 |
lruvec_size = 0; |
688035f72 mm: don't avoid h... |
2452 2453 2454 2455 2456 2457 |
scan = 0; } break; default: /* Look ma, no brain */ BUG(); |
9a2651140 mm: vmscan: clean... |
2458 |
} |
688035f72 mm: don't avoid h... |
2459 |
|
9783aa991 mm, memcg: propor... |
2460 |
*lru_pages += lruvec_size; |
688035f72 mm: don't avoid h... |
2461 |
nr[lru] = scan; |
76a33fc38 vmscan: prevent g... |
2462 |
} |
6e08a369e vmscan: cleanup t... |
2463 |
} |
4f98a2fee vmscan: split LRU... |
2464 |
|
9b4f98cda mm: vmscan: compa... |
2465 |
/* |
a9dd0a831 mm, vmscan: make ... |
2466 |
* This is a basic per-node page freer. Used by both kswapd and direct reclaim. |
9b4f98cda mm: vmscan: compa... |
2467 |
*/ |
a9dd0a831 mm, vmscan: make ... |
2468 |
static void shrink_node_memcg(struct pglist_data *pgdat, struct mem_cgroup *memcg, |
333776785 mm: vmscan: pass ... |
2469 |
struct scan_control *sc, unsigned long *lru_pages) |
9b4f98cda mm: vmscan: compa... |
2470 |
{ |
ef8f23279 mm, memcg: move m... |
2471 |
struct lruvec *lruvec = mem_cgroup_lruvec(pgdat, memcg); |
9b4f98cda mm: vmscan: compa... |
2472 |
unsigned long nr[NR_LRU_LISTS]; |
e82e0561d mm: vmscan: obey ... |
2473 |
unsigned long targets[NR_LRU_LISTS]; |
9b4f98cda mm: vmscan: compa... |
2474 2475 2476 2477 2478 |
unsigned long nr_to_scan; enum lru_list lru; unsigned long nr_reclaimed = 0; unsigned long nr_to_reclaim = sc->nr_to_reclaim; struct blk_plug plug; |
1a501907b mm: vmscan: use p... |
2479 |
bool scan_adjusted; |
9b4f98cda mm: vmscan: compa... |
2480 |
|
333776785 mm: vmscan: pass ... |
2481 |
get_scan_count(lruvec, memcg, sc, nr, lru_pages); |
9b4f98cda mm: vmscan: compa... |
2482 |
|
e82e0561d mm: vmscan: obey ... |
2483 2484 |
/* Record the original scan target for proportional adjustments later */ memcpy(targets, nr, sizeof(nr)); |
1a501907b mm: vmscan: use p... |
2485 2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 |
/* * Global reclaiming within direct reclaim at DEF_PRIORITY is a normal * event that can occur when there is little memory pressure e.g. * multiple streaming readers/writers. Hence, we do not abort scanning * when the requested number of pages are reclaimed when scanning at * DEF_PRIORITY on the assumption that the fact we are direct * reclaiming implies that kswapd is not keeping up and it is best to * do a batch of work at once. For memcg reclaim one check is made to * abort proportional reclaim if either the file or anon lru has already * dropped to zero at the first pass. */ scan_adjusted = (global_reclaim(sc) && !current_is_kswapd() && sc->priority == DEF_PRIORITY); |
9b4f98cda mm: vmscan: compa... |
2498 2499 2500 |
blk_start_plug(&plug); while (nr[LRU_INACTIVE_ANON] || nr[LRU_ACTIVE_FILE] || nr[LRU_INACTIVE_FILE]) { |
e82e0561d mm: vmscan: obey ... |
2501 2502 |
unsigned long nr_anon, nr_file, percentage; unsigned long nr_scanned; |
9b4f98cda mm: vmscan: compa... |
2503 2504 2505 2506 2507 2508 |
for_each_evictable_lru(lru) { if (nr[lru]) { nr_to_scan = min(nr[lru], SWAP_CLUSTER_MAX); nr[lru] -= nr_to_scan; nr_reclaimed += shrink_list(lru, nr_to_scan, |
3b991208b mm: fix inactive ... |
2509 |
lruvec, sc); |
9b4f98cda mm: vmscan: compa... |
2510 2511 |
} } |
e82e0561d mm: vmscan: obey ... |
2512 |
|
bd041733c mm, vmscan: add c... |
2513 |
cond_resched(); |
e82e0561d mm: vmscan: obey ... |
2514 2515 |
if (nr_reclaimed < nr_to_reclaim || scan_adjusted) continue; |
9b4f98cda mm: vmscan: compa... |
2516 |
/* |
e82e0561d mm: vmscan: obey ... |
2517 |
* For kswapd and memcg, reclaim at least the number of pages |
1a501907b mm: vmscan: use p... |
2518 |
* requested. Ensure that the anon and file LRUs are scanned |
e82e0561d mm: vmscan: obey ... |
2519 2520 2521 2522 2523 2524 |
* proportionally what was requested by get_scan_count(). We * stop reclaiming one LRU and reduce the amount scanning * proportional to the original scan target. */ nr_file = nr[LRU_INACTIVE_FILE] + nr[LRU_ACTIVE_FILE]; nr_anon = nr[LRU_INACTIVE_ANON] + nr[LRU_ACTIVE_ANON]; |
1a501907b mm: vmscan: use p... |
2525 2526 2527 2528 2529 2530 2531 2532 |
/* * It's just vindictive to attack the larger once the smaller * has gone to zero. And given the way we stop scanning the * smaller below, this makes sure that we only make one nudge * towards proportionality once we've got nr_to_reclaim. */ if (!nr_file || !nr_anon) break; |
e82e0561d mm: vmscan: obey ... |
2533 2534 2535 2536 2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 |
if (nr_file > nr_anon) { unsigned long scan_target = targets[LRU_INACTIVE_ANON] + targets[LRU_ACTIVE_ANON] + 1; lru = LRU_BASE; percentage = nr_anon * 100 / scan_target; } else { unsigned long scan_target = targets[LRU_INACTIVE_FILE] + targets[LRU_ACTIVE_FILE] + 1; lru = LRU_FILE; percentage = nr_file * 100 / scan_target; } /* Stop scanning the smaller of the LRU */ nr[lru] = 0; nr[lru + LRU_ACTIVE] = 0; /* * Recalculate the other LRU scan count based on its original * scan target and the percentage scanning already complete */ lru = (lru == LRU_FILE) ? LRU_BASE : LRU_FILE; nr_scanned = targets[lru] - nr[lru]; nr[lru] = targets[lru] * (100 - percentage) / 100; nr[lru] -= min(nr[lru], nr_scanned); lru += LRU_ACTIVE; nr_scanned = targets[lru] - nr[lru]; nr[lru] = targets[lru] * (100 - percentage) / 100; nr[lru] -= min(nr[lru], nr_scanned); scan_adjusted = true; |
9b4f98cda mm: vmscan: compa... |
2564 2565 2566 2567 2568 2569 2570 2571 |
} blk_finish_plug(&plug); sc->nr_reclaimed += nr_reclaimed; /* * Even if we did not try to evict anon pages at all, we want to * rebalance the anon lru active/inactive ratio. */ |
3b991208b mm: fix inactive ... |
2572 |
if (inactive_list_is_low(lruvec, false, sc, true)) |
9b4f98cda mm: vmscan: compa... |
2573 2574 |
shrink_active_list(SWAP_CLUSTER_MAX, lruvec, sc, LRU_ACTIVE_ANON); |
9b4f98cda mm: vmscan: compa... |
2575 |
} |
23b9da55c mm: vmscan: remov... |
2576 |
/* Use reclaim/compaction for costly allocs or under memory pressure */ |
9e3b2f8cd mm/vmscan: store ... |
2577 |
static bool in_reclaim_compaction(struct scan_control *sc) |
23b9da55c mm: vmscan: remov... |
2578 |
{ |
d84da3f9e mm: use IS_ENABLE... |
2579 |
if (IS_ENABLED(CONFIG_COMPACTION) && sc->order && |
23b9da55c mm: vmscan: remov... |
2580 |
(sc->order > PAGE_ALLOC_COSTLY_ORDER || |
9e3b2f8cd mm/vmscan: store ... |
2581 |
sc->priority < DEF_PRIORITY - 2)) |
23b9da55c mm: vmscan: remov... |
2582 2583 2584 2585 |
return true; return false; } |
1da177e4c Linux-2.6.12-rc2 |
2586 |
/* |
23b9da55c mm: vmscan: remov... |
2587 2588 2589 2590 2591 |
* Reclaim/compaction is used for high-order allocation requests. It reclaims * order-0 pages before compacting the zone. should_continue_reclaim() returns * true if more pages should be reclaimed such that when the page allocator * calls try_to_compact_zone() that it will have enough free pages to succeed. * It will give up earlier than that if there is difficulty reclaiming pages. |
3e7d34497 mm: vmscan: recla... |
2592 |
*/ |
a9dd0a831 mm, vmscan: make ... |
2593 |
static inline bool should_continue_reclaim(struct pglist_data *pgdat, |
3e7d34497 mm: vmscan: recla... |
2594 |
unsigned long nr_reclaimed, |
3e7d34497 mm: vmscan: recla... |
2595 2596 2597 2598 |
struct scan_control *sc) { unsigned long pages_for_compaction; unsigned long inactive_lru_pages; |
a9dd0a831 mm, vmscan: make ... |
2599 |
int z; |
3e7d34497 mm: vmscan: recla... |
2600 2601 |
/* If not in reclaim/compaction mode, stop */ |
9e3b2f8cd mm/vmscan: store ... |
2602 |
if (!in_reclaim_compaction(sc)) |
3e7d34497 mm: vmscan: recla... |
2603 |
return false; |
5ee04716c mm, reclaim: clea... |
2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 |
/* * Stop if we failed to reclaim any pages from the last SWAP_CLUSTER_MAX * number of pages that were scanned. This will return to the caller * with the risk reclaim/compaction and the resulting allocation attempt * fails. In the past we have tried harder for __GFP_RETRY_MAYFAIL * allocations through requiring that the full LRU list has been scanned * first, by assuming that zero delta of sc->nr_scanned means full LRU * scan, but that approximation was wrong, and there were corner cases * where always a non-zero amount of pages were scanned. */ if (!nr_reclaimed) return false; |
3e7d34497 mm: vmscan: recla... |
2616 |
|
3e7d34497 mm: vmscan: recla... |
2617 |
/* If compaction would go ahead or the allocation would succeed, stop */ |
a9dd0a831 mm, vmscan: make ... |
2618 2619 |
for (z = 0; z <= sc->reclaim_idx; z++) { struct zone *zone = &pgdat->node_zones[z]; |
6aa303def mm, vmscan: only ... |
2620 |
if (!managed_zone(zone)) |
a9dd0a831 mm, vmscan: make ... |
2621 2622 2623 |
continue; switch (compaction_suitable(zone, sc->order, 0, sc->reclaim_idx)) { |
cf378319d mm, compaction: r... |
2624 |
case COMPACT_SUCCESS: |
a9dd0a831 mm, vmscan: make ... |
2625 2626 2627 2628 2629 2630 |
case COMPACT_CONTINUE: return false; default: /* check next zone */ ; } |
3e7d34497 mm: vmscan: recla... |
2631 |
} |
1c6c15971 mm, reclaim: make... |
2632 2633 2634 2635 2636 2637 2638 2639 2640 |
/* * If we have not reclaimed enough pages for compaction and the * inactive lists are large enough, continue reclaiming */ pages_for_compaction = compact_gap(sc->order); inactive_lru_pages = node_page_state(pgdat, NR_INACTIVE_FILE); if (get_nr_swap_pages() > 0) inactive_lru_pages += node_page_state(pgdat, NR_INACTIVE_ANON); |
5ee04716c mm, reclaim: clea... |
2641 |
return inactive_lru_pages > pages_for_compaction; |
3e7d34497 mm: vmscan: recla... |
2642 |
} |
e3c1ac586 mm/vmscan: don't ... |
2643 2644 2645 2646 2647 |
static bool pgdat_memcg_congested(pg_data_t *pgdat, struct mem_cgroup *memcg) { return test_bit(PGDAT_CONGESTED, &pgdat->flags) || (memcg && memcg_congested(pgdat, memcg)); } |
970a39a36 mm, vmscan: avoid... |
2648 |
static bool shrink_node(pg_data_t *pgdat, struct scan_control *sc) |
1da177e4c Linux-2.6.12-rc2 |
2649 |
{ |
cb731d6c6 vmscan: per memor... |
2650 |
struct reclaim_state *reclaim_state = current->reclaim_state; |
f0fdc5e8e vmscan: fix zone ... |
2651 |
unsigned long nr_reclaimed, nr_scanned; |
2344d7e44 mm: vmscan: remov... |
2652 |
bool reclaimable = false; |
1da177e4c Linux-2.6.12-rc2 |
2653 |
|
9b4f98cda mm: vmscan: compa... |
2654 2655 |
do { struct mem_cgroup *root = sc->target_mem_cgroup; |
a9dd0a831 mm, vmscan: make ... |
2656 |
unsigned long node_lru_pages = 0; |
694fbc0fe revert "memcg: en... |
2657 |
struct mem_cgroup *memcg; |
3e7d34497 mm: vmscan: recla... |
2658 |
|
d108c7721 mm/vmscan: don't ... |
2659 |
memset(&sc->nr, 0, sizeof(sc->nr)); |
9b4f98cda mm: vmscan: compa... |
2660 2661 |
nr_reclaimed = sc->nr_reclaimed; nr_scanned = sc->nr_scanned; |
1da177e4c Linux-2.6.12-rc2 |
2662 |
|
1ba6fc9af mm: vmscan: do no... |
2663 |
memcg = mem_cgroup_iter(root, NULL, NULL); |
694fbc0fe revert "memcg: en... |
2664 |
do { |
6b4f7799c mm: vmscan: invok... |
2665 |
unsigned long lru_pages; |
8e8ae6452 mm: memcontrol: h... |
2666 |
unsigned long reclaimed; |
cb731d6c6 vmscan: per memor... |
2667 |
unsigned long scanned; |
5660048cc mm: move memcg hi... |
2668 |
|
a5bc1c7a9 mm: memcg: fix me... |
2669 2670 2671 2672 2673 2674 2675 |
/* * This loop can become CPU-bound when target memcgs * aren't eligible for reclaim - either because they * don't have any reclaimable pages, or because their * memory is explicitly protected. Avoid soft lockups. */ cond_resched(); |
bf8d5d52f memcg: introduce ... |
2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 |
switch (mem_cgroup_protected(root, memcg)) { case MEMCG_PROT_MIN: /* * Hard protection. * If there is no reclaimable memory, OOM. */ continue; case MEMCG_PROT_LOW: /* * Soft protection. * Respect the protection only as long as * there is an unprotected supply * of reclaimable memory from other cgroups. */ |
d6622f636 mm/vmscan: more r... |
2690 2691 |
if (!sc->memcg_low_reclaim) { sc->memcg_low_skipped = 1; |
241994ed8 mm: memcontrol: d... |
2692 |
continue; |
d6622f636 mm/vmscan: more r... |
2693 |
} |
e27be240d mm: memcg: make s... |
2694 |
memcg_memory_event(memcg, MEMCG_LOW); |
bf8d5d52f memcg: introduce ... |
2695 2696 |
break; case MEMCG_PROT_NONE: |
9783aa991 mm, memcg: propor... |
2697 2698 2699 2700 2701 2702 2703 |
/* * All protection thresholds breached. We may * still choose to vary the scan pressure * applied based on by how much the cgroup in * question has exceeded its protection * thresholds (see get_scan_count). */ |
bf8d5d52f memcg: introduce ... |
2704 |
break; |
241994ed8 mm: memcontrol: d... |
2705 |
} |
8e8ae6452 mm: memcontrol: h... |
2706 |
reclaimed = sc->nr_reclaimed; |
cb731d6c6 vmscan: per memor... |
2707 |
scanned = sc->nr_scanned; |
a9dd0a831 mm, vmscan: make ... |
2708 2709 |
shrink_node_memcg(pgdat, memcg, sc, &lru_pages); node_lru_pages += lru_pages; |
f16015fbf mm: vmscan: disti... |
2710 |
|
28360f398 mm, vmscan: do no... |
2711 2712 |
shrink_slab(sc->gfp_mask, pgdat->node_id, memcg, sc->priority); |
cb731d6c6 vmscan: per memor... |
2713 |
|
8e8ae6452 mm: memcontrol: h... |
2714 2715 2716 2717 |
/* Record the group's reclaim efficiency */ vmpressure(sc->gfp_mask, memcg, false, sc->nr_scanned - scanned, sc->nr_reclaimed - reclaimed); |
1ba6fc9af mm: vmscan: do no... |
2718 |
} while ((memcg = mem_cgroup_iter(root, memcg, NULL))); |
70ddf637e memcg: add memory... |
2719 |
|
cb731d6c6 vmscan: per memor... |
2720 2721 2722 |
if (reclaim_state) { sc->nr_reclaimed += reclaim_state->reclaimed_slab; reclaim_state->reclaimed_slab = 0; |
6b4f7799c mm: vmscan: invok... |
2723 |
} |
8e8ae6452 mm: memcontrol: h... |
2724 2725 |
/* Record the subtree's reclaim efficiency */ vmpressure(sc->gfp_mask, sc->target_mem_cgroup, true, |
70ddf637e memcg: add memory... |
2726 2727 |
sc->nr_scanned - nr_scanned, sc->nr_reclaimed - nr_reclaimed); |
2344d7e44 mm: vmscan: remov... |
2728 2729 |
if (sc->nr_reclaimed - nr_reclaimed) reclaimable = true; |
e3c1ac586 mm/vmscan: don't ... |
2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 |
if (current_is_kswapd()) { /* * If reclaim is isolating dirty pages under writeback, * it implies that the long-lived page allocation rate * is exceeding the page laundering rate. Either the * global limits are not being effective at throttling * processes due to the page distribution throughout * zones or there is heavy usage of a slow backing * device. The only option is to throttle from reclaim * context which is not ideal as there is no guarantee * the dirtying process is throttled in the same way * balance_dirty_pages() manages. * * Once a node is flagged PGDAT_WRITEBACK, kswapd will * count the number of pages under pages flagged for * immediate reclaim and stall if any are encountered * in the nr_immediate check below. */ if (sc->nr.writeback && sc->nr.writeback == sc->nr.taken) set_bit(PGDAT_WRITEBACK, &pgdat->flags); |
d108c7721 mm/vmscan: don't ... |
2750 |
|
d108c7721 mm/vmscan: don't ... |
2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773 |
/* * Tag a node as congested if all the dirty pages * scanned were backed by a congested BDI and * wait_iff_congested will stall. */ if (sc->nr.dirty && sc->nr.dirty == sc->nr.congested) set_bit(PGDAT_CONGESTED, &pgdat->flags); /* Allow kswapd to start writing pages during reclaim.*/ if (sc->nr.unqueued_dirty == sc->nr.file_taken) set_bit(PGDAT_DIRTY, &pgdat->flags); /* * If kswapd scans pages marked marked for immediate * reclaim and under writeback (nr_immediate), it * implies that pages are cycling through the LRU * faster than they are written so also forcibly stall. */ if (sc->nr.immediate) congestion_wait(BLK_RW_ASYNC, HZ/10); } /* |
e3c1ac586 mm/vmscan: don't ... |
2774 2775 2776 2777 2778 2779 2780 2781 |
* Legacy memcg will stall in page writeback so avoid forcibly * stalling in wait_iff_congested(). */ if (!global_reclaim(sc) && sane_reclaim(sc) && sc->nr.dirty && sc->nr.dirty == sc->nr.congested) set_memcg_congestion(pgdat, root, true); /* |
d108c7721 mm/vmscan: don't ... |
2782 2783 2784 2785 2786 2787 |
* Stall direct reclaim for IO completions if underlying BDIs * and node is congested. Allow kswapd to continue until it * starts encountering unqueued dirty pages or cycling through * the LRU too quickly. */ if (!sc->hibernation_mode && !current_is_kswapd() && |
e3c1ac586 mm/vmscan: don't ... |
2788 2789 |
current_may_throttle() && pgdat_memcg_congested(pgdat, root)) wait_iff_congested(BLK_RW_ASYNC, HZ/10); |
d108c7721 mm/vmscan: don't ... |
2790 |
|
a9dd0a831 mm, vmscan: make ... |
2791 |
} while (should_continue_reclaim(pgdat, sc->nr_reclaimed - nr_reclaimed, |
5ee04716c mm, reclaim: clea... |
2792 |
sc)); |
2344d7e44 mm: vmscan: remov... |
2793 |
|
c73322d09 mm: fix 100% CPU ... |
2794 2795 2796 2797 2798 2799 2800 2801 |
/* * Kswapd gives up on balancing particular nodes after too * many failures to reclaim anything from them and goes to * sleep. On reclaim progress, reset the failure counter. A * successful direct reclaim run will revive a dormant kswapd. */ if (reclaimable) pgdat->kswapd_failures = 0; |
2344d7e44 mm: vmscan: remov... |
2802 |
return reclaimable; |
f16015fbf mm: vmscan: disti... |
2803 |
} |
53853e2d2 mm, compaction: d... |
2804 |
/* |
fdd4c6149 mm, vmscan: make ... |
2805 2806 2807 |
* Returns true if compaction should go ahead for a costly-order request, or * the allocation would already succeed without compaction. Return false if we * should reclaim first. |
53853e2d2 mm, compaction: d... |
2808 |
*/ |
4f588331b mm, vmscan: avoid... |
2809 |
static inline bool compaction_ready(struct zone *zone, struct scan_control *sc) |
fe4b1b244 mm: vmscan: when ... |
2810 |
{ |
31483b6ad mm, vmscan: remov... |
2811 |
unsigned long watermark; |
fdd4c6149 mm, vmscan: make ... |
2812 |
enum compact_result suitable; |
fe4b1b244 mm: vmscan: when ... |
2813 |
|
fdd4c6149 mm, vmscan: make ... |
2814 2815 2816 2817 2818 2819 2820 |
suitable = compaction_suitable(zone, sc->order, 0, sc->reclaim_idx); if (suitable == COMPACT_SUCCESS) /* Allocation should succeed already. Don't reclaim. */ return true; if (suitable == COMPACT_SKIPPED) /* Compaction cannot yet proceed. Do reclaim. */ return false; |
fe4b1b244 mm: vmscan: when ... |
2821 |
|
53853e2d2 mm, compaction: d... |
2822 |
/* |
fdd4c6149 mm, vmscan: make ... |
2823 2824 2825 2826 2827 2828 2829 |
* Compaction is already possible, but it takes time to run and there * are potentially other callers using the pages just freed. So proceed * with reclaim to make a buffer of free pages available to give * compaction a reasonable chance of completing and allocating the page. * Note that we won't actually reclaim the whole buffer in one attempt * as the target watermark in should_continue_reclaim() is lower. But if * we are already above the high+gap watermark, don't reclaim at all. |
53853e2d2 mm, compaction: d... |
2830 |
*/ |
fdd4c6149 mm, vmscan: make ... |
2831 |
watermark = high_wmark_pages(zone) + compact_gap(sc->order); |
fe4b1b244 mm: vmscan: when ... |
2832 |
|
fdd4c6149 mm, vmscan: make ... |
2833 |
return zone_watermark_ok_safe(zone, 0, watermark, sc->reclaim_idx); |
fe4b1b244 mm: vmscan: when ... |
2834 |
} |
1da177e4c Linux-2.6.12-rc2 |
2835 2836 2837 2838 2839 |
/* * This is the direct reclaim path, for page-allocating processes. We only * try to reclaim pages from zones which will satisfy the caller's allocation * request. * |
1da177e4c Linux-2.6.12-rc2 |
2840 2841 2842 |
* If a zone is deemed to be full of pinned pages then just give it a light * scan then give up on it. */ |
0a0337e0d mm, oom: rework o... |
2843 |
static void shrink_zones(struct zonelist *zonelist, struct scan_control *sc) |
1da177e4c Linux-2.6.12-rc2 |
2844 |
{ |
dd1a239f6 mm: have zonelist... |
2845 |
struct zoneref *z; |
54a6eb5c4 mm: use two zonel... |
2846 |
struct zone *zone; |
0608f43da revert "memcg, vm... |
2847 2848 |
unsigned long nr_soft_reclaimed; unsigned long nr_soft_scanned; |
619d0d76c mm/vmscan: restor... |
2849 |
gfp_t orig_mask; |
79dafcdca mm, vmscan: by de... |
2850 |
pg_data_t *last_pgdat = NULL; |
1cfb419b3 per-zone and recl... |
2851 |
|
cc715d99e mm: vmscan: forci... |
2852 2853 2854 2855 2856 |
/* * If the number of buffer_heads in the machine exceeds the maximum * allowed level, force direct reclaim to scan the highmem zone as * highmem pages could be pinning lowmem pages storing buffer_heads */ |
619d0d76c mm/vmscan: restor... |
2857 |
orig_mask = sc->gfp_mask; |
b2e18757f mm, vmscan: begin... |
2858 |
if (buffer_heads_over_limit) { |
cc715d99e mm: vmscan: forci... |
2859 |
sc->gfp_mask |= __GFP_HIGHMEM; |
4f588331b mm, vmscan: avoid... |
2860 |
sc->reclaim_idx = gfp_zone(sc->gfp_mask); |
b2e18757f mm, vmscan: begin... |
2861 |
} |
cc715d99e mm: vmscan: forci... |
2862 |
|
d4debc66d vmscan: remove un... |
2863 |
for_each_zone_zonelist_nodemask(zone, z, zonelist, |
b2e18757f mm, vmscan: begin... |
2864 |
sc->reclaim_idx, sc->nodemask) { |
b2e18757f mm, vmscan: begin... |
2865 |
/* |
1cfb419b3 per-zone and recl... |
2866 2867 2868 |
* Take care memory controller reclaiming has small influence * to global LRU. */ |
89b5fae53 mm: vmscan: disti... |
2869 |
if (global_reclaim(sc)) { |
344736f29 cpuset: simplify ... |
2870 2871 |
if (!cpuset_zone_allowed(zone, GFP_KERNEL | __GFP_HARDWALL)) |
1cfb419b3 per-zone and recl... |
2872 |
continue; |
65ec02cb9 mm: vmscan: move ... |
2873 |
|
0b06496a3 mm: vmscan: rewor... |
2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 |
/* * If we already have plenty of memory free for * compaction in this zone, don't free any more. * Even though compaction is invoked for any * non-zero order, only frequent costly order * reclamation is disruptive enough to become a * noticeable problem, like transparent huge * page allocations. */ if (IS_ENABLED(CONFIG_COMPACTION) && sc->order > PAGE_ALLOC_COSTLY_ORDER && |
4f588331b mm, vmscan: avoid... |
2885 |
compaction_ready(zone, sc)) { |
0b06496a3 mm: vmscan: rewor... |
2886 2887 |
sc->compaction_ready = true; continue; |
e0887c19b vmscan: limit dir... |
2888 |
} |
0b06496a3 mm: vmscan: rewor... |
2889 |
|
0608f43da revert "memcg, vm... |
2890 |
/* |
79dafcdca mm, vmscan: by de... |
2891 2892 2893 2894 2895 2896 2897 2898 2899 |
* Shrink each node in the zonelist once. If the * zonelist is ordered by zone (not the default) then a * node may be shrunk multiple times but in that case * the user prefers lower zones being preserved. */ if (zone->zone_pgdat == last_pgdat) continue; /* |
0608f43da revert "memcg, vm... |
2900 2901 2902 2903 2904 2905 |
* This steals pages from memory cgroups over softlimit * and returns the number of reclaimed pages and * scanned pages. This works for global memory pressure * and balancing, not for a memcg's limit. */ nr_soft_scanned = 0; |
ef8f23279 mm, memcg: move m... |
2906 |
nr_soft_reclaimed = mem_cgroup_soft_limit_reclaim(zone->zone_pgdat, |
0608f43da revert "memcg, vm... |
2907 2908 2909 2910 |
sc->order, sc->gfp_mask, &nr_soft_scanned); sc->nr_reclaimed += nr_soft_reclaimed; sc->nr_scanned += nr_soft_scanned; |
ac34a1a3c memcg: fix direct... |
2911 |
/* need some check for avoid more shrink_zone() */ |
1cfb419b3 per-zone and recl... |
2912 |
} |
408d85441 [PATCH] oom: use ... |
2913 |
|
79dafcdca mm, vmscan: by de... |
2914 2915 2916 2917 |
/* See comment about same check for global reclaim above */ if (zone->zone_pgdat == last_pgdat) continue; last_pgdat = zone->zone_pgdat; |
970a39a36 mm, vmscan: avoid... |
2918 |
shrink_node(zone->zone_pgdat, sc); |
1da177e4c Linux-2.6.12-rc2 |
2919 |
} |
e0c23279c vmscan: abort rec... |
2920 |
|
65ec02cb9 mm: vmscan: move ... |
2921 |
/* |
619d0d76c mm/vmscan: restor... |
2922 2923 2924 2925 |
* Restore to original mask to avoid the impact on the caller if we * promoted it to __GFP_HIGHMEM. */ sc->gfp_mask = orig_mask; |
1da177e4c Linux-2.6.12-rc2 |
2926 |
} |
4f98a2fee vmscan: split LRU... |
2927 |
|
2a2e48854 mm: vmscan: fix I... |
2928 2929 2930 2931 2932 2933 2934 2935 |
static void snapshot_refaults(struct mem_cgroup *root_memcg, pg_data_t *pgdat) { struct mem_cgroup *memcg; memcg = mem_cgroup_iter(root_memcg, NULL, NULL); do { unsigned long refaults; struct lruvec *lruvec; |
2a2e48854 mm: vmscan: fix I... |
2936 |
lruvec = mem_cgroup_lruvec(pgdat, memcg); |
205b20cc5 mm: memcontrol: m... |
2937 |
refaults = lruvec_page_state_local(lruvec, WORKINGSET_ACTIVATE); |
2a2e48854 mm: vmscan: fix I... |
2938 2939 2940 |
lruvec->refaults = refaults; } while ((memcg = mem_cgroup_iter(root_memcg, memcg, NULL))); } |
1da177e4c Linux-2.6.12-rc2 |
2941 2942 2943 2944 2945 2946 2947 2948 |
/* * This is the main entry point to direct page reclaim. * * If a full scan of the inactive list fails to free enough memory then we * are "out of memory" and something needs to be killed. * * If the caller is !__GFP_FS then the probability of a failure is reasonably * high - the zone may be full of dirty or under-writeback pages, which this |
5b0830cb9 writeback: get ri... |
2949 2950 2951 2952 |
* caller can't do much about. We kick the writeback threads and take explicit * naps in the hope that some of these pages can be written. But if the * allocating task holds filesystem locks which prevent writeout this might not * work, and the allocation attempt will fail. |
a41f24ea9 page allocator: s... |
2953 2954 2955 |
* * returns: 0, if no pages reclaimed * else, the number of pages reclaimed |
1da177e4c Linux-2.6.12-rc2 |
2956 |
*/ |
dac1d27bc mm: use zonelists... |
2957 |
static unsigned long do_try_to_free_pages(struct zonelist *zonelist, |
3115cd914 mm: vmscan: remov... |
2958 |
struct scan_control *sc) |
1da177e4c Linux-2.6.12-rc2 |
2959 |
{ |
241994ed8 mm: memcontrol: d... |
2960 |
int initial_priority = sc->priority; |
2a2e48854 mm: vmscan: fix I... |
2961 2962 2963 |
pg_data_t *last_pgdat; struct zoneref *z; struct zone *zone; |
241994ed8 mm: memcontrol: d... |
2964 |
retry: |
873b47717 per-task-delay-ac... |
2965 |
delayacct_freepages_start(); |
89b5fae53 mm: vmscan: disti... |
2966 |
if (global_reclaim(sc)) |
7cc30fcfd mm: vmstat: accou... |
2967 |
__count_zid_vm_events(ALLOCSTALL, sc->reclaim_idx, 1); |
1da177e4c Linux-2.6.12-rc2 |
2968 |
|
9e3b2f8cd mm/vmscan: store ... |
2969 |
do { |
70ddf637e memcg: add memory... |
2970 2971 |
vmpressure_prio(sc->gfp_mask, sc->target_mem_cgroup, sc->priority); |
66e1707bc Memory controller... |
2972 |
sc->nr_scanned = 0; |
0a0337e0d mm, oom: rework o... |
2973 |
shrink_zones(zonelist, sc); |
c6a8a8c58 vmscan: recalcula... |
2974 |
|
bb21c7ce1 vmscan: fix do_tr... |
2975 |
if (sc->nr_reclaimed >= sc->nr_to_reclaim) |
0b06496a3 mm: vmscan: rewor... |
2976 2977 2978 2979 |
break; if (sc->compaction_ready) break; |
1da177e4c Linux-2.6.12-rc2 |
2980 2981 |
/* |
0e50ce3b5 mm: use up free s... |
2982 2983 2984 2985 2986 |
* If we're getting trouble reclaiming, start doing * writepage even in laptop mode. */ if (sc->priority < DEF_PRIORITY - 2) sc->may_writepage = 1; |
0b06496a3 mm: vmscan: rewor... |
2987 |
} while (--sc->priority >= 0); |
bb21c7ce1 vmscan: fix do_tr... |
2988 |
|
2a2e48854 mm: vmscan: fix I... |
2989 2990 2991 2992 2993 2994 2995 |
last_pgdat = NULL; for_each_zone_zonelist_nodemask(zone, z, zonelist, sc->reclaim_idx, sc->nodemask) { if (zone->zone_pgdat == last_pgdat) continue; last_pgdat = zone->zone_pgdat; snapshot_refaults(sc->target_mem_cgroup, zone->zone_pgdat); |
e3c1ac586 mm/vmscan: don't ... |
2996 |
set_memcg_congestion(last_pgdat, sc->target_mem_cgroup, false); |
2a2e48854 mm: vmscan: fix I... |
2997 |
} |
873b47717 per-task-delay-ac... |
2998 |
delayacct_freepages_end(); |
bb21c7ce1 vmscan: fix do_tr... |
2999 3000 |
if (sc->nr_reclaimed) return sc->nr_reclaimed; |
0cee34fd7 mm: vmscan: check... |
3001 |
/* Aborted reclaim to try compaction? don't OOM, then */ |
0b06496a3 mm: vmscan: rewor... |
3002 |
if (sc->compaction_ready) |
7335084d4 mm: vmscan: do no... |
3003 |
return 1; |
241994ed8 mm: memcontrol: d... |
3004 |
/* Untapped cgroup reserves? Don't OOM, retry. */ |
d6622f636 mm/vmscan: more r... |
3005 |
if (sc->memcg_low_skipped) { |
241994ed8 mm: memcontrol: d... |
3006 |
sc->priority = initial_priority; |
d6622f636 mm/vmscan: more r... |
3007 3008 |
sc->memcg_low_reclaim = 1; sc->memcg_low_skipped = 0; |
241994ed8 mm: memcontrol: d... |
3009 3010 |
goto retry; } |
bb21c7ce1 vmscan: fix do_tr... |
3011 |
return 0; |
1da177e4c Linux-2.6.12-rc2 |
3012 |
} |
c73322d09 mm: fix 100% CPU ... |
3013 |
static bool allow_direct_reclaim(pg_data_t *pgdat) |
5515061d2 mm: throttle dire... |
3014 3015 3016 3017 3018 3019 |
{ struct zone *zone; unsigned long pfmemalloc_reserve = 0; unsigned long free_pages = 0; int i; bool wmark_ok; |
c73322d09 mm: fix 100% CPU ... |
3020 3021 |
if (pgdat->kswapd_failures >= MAX_RECLAIM_RETRIES) return true; |
5515061d2 mm: throttle dire... |
3022 3023 |
for (i = 0; i <= ZONE_NORMAL; i++) { zone = &pgdat->node_zones[i]; |
d450abd81 mm: fix check for... |
3024 3025 3026 3027 |
if (!managed_zone(zone)) continue; if (!zone_reclaimable_pages(zone)) |
675becce1 mm: vmscan: do no... |
3028 |
continue; |
5515061d2 mm: throttle dire... |
3029 3030 3031 |
pfmemalloc_reserve += min_wmark_pages(zone); free_pages += zone_page_state(zone, NR_FREE_PAGES); } |
675becce1 mm: vmscan: do no... |
3032 3033 3034 |
/* If there are no reserves (unexpected config) then do not throttle */ if (!pfmemalloc_reserve) return true; |
5515061d2 mm: throttle dire... |
3035 3036 3037 3038 |
wmark_ok = free_pages > pfmemalloc_reserve / 2; /* kswapd must be awake if processes are being throttled */ if (!wmark_ok && waitqueue_active(&pgdat->kswapd_wait)) { |
19b9ff114 mm/vmscan.c: fix ... |
3039 3040 |
if (READ_ONCE(pgdat->kswapd_classzone_idx) > ZONE_NORMAL) WRITE_ONCE(pgdat->kswapd_classzone_idx, ZONE_NORMAL); |
5515061d2 mm: throttle dire... |
3041 3042 3043 3044 3045 3046 3047 3048 3049 3050 |
wake_up_interruptible(&pgdat->kswapd_wait); } return wmark_ok; } /* * Throttle direct reclaimers if backing storage is backed by the network * and the PFMEMALLOC reserve for the preferred node is getting dangerously * depleted. kswapd will continue to make progress and wake the processes |
50694c28f mm: vmscan: check... |
3051 3052 3053 3054 |
* when the low watermark is reached. * * Returns true if a fatal signal was delivered during throttling. If this * happens, the page allocator should not consider triggering the OOM killer. |
5515061d2 mm: throttle dire... |
3055 |
*/ |
50694c28f mm: vmscan: check... |
3056 |
static bool throttle_direct_reclaim(gfp_t gfp_mask, struct zonelist *zonelist, |
5515061d2 mm: throttle dire... |
3057 3058 |
nodemask_t *nodemask) { |
675becce1 mm: vmscan: do no... |
3059 |
struct zoneref *z; |
5515061d2 mm: throttle dire... |
3060 |
struct zone *zone; |
675becce1 mm: vmscan: do no... |
3061 |
pg_data_t *pgdat = NULL; |
5515061d2 mm: throttle dire... |
3062 3063 3064 3065 3066 3067 3068 3069 3070 |
/* * Kernel threads should not be throttled as they may be indirectly * responsible for cleaning pages necessary for reclaim to make forward * progress. kjournald for example may enter direct reclaim while * committing a transaction where throttling it could forcing other * processes to block on log_wait_commit(). */ if (current->flags & PF_KTHREAD) |
50694c28f mm: vmscan: check... |
3071 3072 3073 3074 3075 3076 3077 3078 |
goto out; /* * If a fatal signal is pending, this process should not throttle. * It should return quickly so it can exit and free its memory */ if (fatal_signal_pending(current)) goto out; |
5515061d2 mm: throttle dire... |
3079 |
|
675becce1 mm: vmscan: do no... |
3080 3081 3082 3083 3084 3085 3086 3087 3088 3089 3090 3091 3092 3093 3094 |
/* * Check if the pfmemalloc reserves are ok by finding the first node * with a usable ZONE_NORMAL or lower zone. The expectation is that * GFP_KERNEL will be required for allocating network buffers when * swapping over the network so ZONE_HIGHMEM is unusable. * * Throttling is based on the first usable node and throttled processes * wait on a queue until kswapd makes progress and wakes them. There * is an affinity then between processes waking up and where reclaim * progress has been made assuming the process wakes on the same node. * More importantly, processes running on remote nodes will not compete * for remote pfmemalloc reserves and processes on different nodes * should make reasonable progress. */ for_each_zone_zonelist_nodemask(zone, z, zonelist, |
17636faad mm/vmscan: fix hi... |
3095 |
gfp_zone(gfp_mask), nodemask) { |
675becce1 mm: vmscan: do no... |
3096 3097 3098 3099 3100 |
if (zone_idx(zone) > ZONE_NORMAL) continue; /* Throttle based on the first usable node */ pgdat = zone->zone_pgdat; |
c73322d09 mm: fix 100% CPU ... |
3101 |
if (allow_direct_reclaim(pgdat)) |
675becce1 mm: vmscan: do no... |
3102 3103 3104 3105 3106 3107 |
goto out; break; } /* If no zone was usable by the allocation flags then do not throttle */ if (!pgdat) |
50694c28f mm: vmscan: check... |
3108 |
goto out; |
5515061d2 mm: throttle dire... |
3109 |
|
68243e76e mm: account for t... |
3110 3111 |
/* Account for the throttling */ count_vm_event(PGSCAN_DIRECT_THROTTLE); |
5515061d2 mm: throttle dire... |
3112 3113 3114 3115 3116 3117 3118 3119 3120 3121 |
/* * If the caller cannot enter the filesystem, it's possible that it * is due to the caller holding an FS lock or performing a journal * transaction in the case of a filesystem like ext[3|4]. In this case, * it is not safe to block on pfmemalloc_wait as kswapd could be * blocked waiting on the same lock. Instead, throttle for up to a * second before continuing. */ if (!(gfp_mask & __GFP_FS)) { wait_event_interruptible_timeout(pgdat->pfmemalloc_wait, |
c73322d09 mm: fix 100% CPU ... |
3122 |
allow_direct_reclaim(pgdat), HZ); |
50694c28f mm: vmscan: check... |
3123 3124 |
goto check_pending; |
5515061d2 mm: throttle dire... |
3125 3126 3127 3128 |
} /* Throttle until kswapd wakes the process */ wait_event_killable(zone->zone_pgdat->pfmemalloc_wait, |
c73322d09 mm: fix 100% CPU ... |
3129 |
allow_direct_reclaim(pgdat)); |
50694c28f mm: vmscan: check... |
3130 3131 3132 3133 3134 3135 3136 |
check_pending: if (fatal_signal_pending(current)) return true; out: return false; |
5515061d2 mm: throttle dire... |
3137 |
} |
dac1d27bc mm: use zonelists... |
3138 |
unsigned long try_to_free_pages(struct zonelist *zonelist, int order, |
327c0e968 vmscan: fix it to... |
3139 |
gfp_t gfp_mask, nodemask_t *nodemask) |
66e1707bc Memory controller... |
3140 |
{ |
33906bc5c vmscan: tracing: ... |
3141 |
unsigned long nr_reclaimed; |
66e1707bc Memory controller... |
3142 |
struct scan_control sc = { |
ee814fe23 mm: vmscan: clean... |
3143 |
.nr_to_reclaim = SWAP_CLUSTER_MAX, |
f2f43e566 mm/vmscan.c: fix ... |
3144 |
.gfp_mask = current_gfp_context(gfp_mask), |
b2e18757f mm, vmscan: begin... |
3145 |
.reclaim_idx = gfp_zone(gfp_mask), |
ee814fe23 mm: vmscan: clean... |
3146 3147 3148 |
.order = order, .nodemask = nodemask, .priority = DEF_PRIORITY, |
66e1707bc Memory controller... |
3149 |
.may_writepage = !laptop_mode, |
a6dc60f89 vmscan: rename sc... |
3150 |
.may_unmap = 1, |
2e2e42598 vmscan,memcg: rei... |
3151 |
.may_swap = 1, |
66e1707bc Memory controller... |
3152 |
}; |
5515061d2 mm: throttle dire... |
3153 |
/* |
bb451fdf3 mm/vmscan.c: cond... |
3154 3155 3156 3157 3158 3159 3160 3161 |
* scan_control uses s8 fields for order, priority, and reclaim_idx. * Confirm they are large enough for max values. */ BUILD_BUG_ON(MAX_ORDER > S8_MAX); BUILD_BUG_ON(DEF_PRIORITY > S8_MAX); BUILD_BUG_ON(MAX_NR_ZONES > S8_MAX); /* |
50694c28f mm: vmscan: check... |
3162 3163 3164 |
* Do not enter reclaim if fatal signal was delivered while throttled. * 1 is returned so that the page allocator does not OOM kill at this * point. |
5515061d2 mm: throttle dire... |
3165 |
*/ |
f2f43e566 mm/vmscan.c: fix ... |
3166 |
if (throttle_direct_reclaim(sc.gfp_mask, zonelist, nodemask)) |
5515061d2 mm: throttle dire... |
3167 |
return 1; |
1732d2b01 mm/vmscan.c: add ... |
3168 |
set_task_reclaim_state(current, &sc.reclaim_state); |
3481c37ff mm/vmscan: drop m... |
3169 |
trace_mm_vmscan_direct_reclaim_begin(order, sc.gfp_mask); |
33906bc5c vmscan: tracing: ... |
3170 |
|
3115cd914 mm: vmscan: remov... |
3171 |
nr_reclaimed = do_try_to_free_pages(zonelist, &sc); |
33906bc5c vmscan: tracing: ... |
3172 3173 |
trace_mm_vmscan_direct_reclaim_end(nr_reclaimed); |
1732d2b01 mm/vmscan.c: add ... |
3174 |
set_task_reclaim_state(current, NULL); |
33906bc5c vmscan: tracing: ... |
3175 3176 |
return nr_reclaimed; |
66e1707bc Memory controller... |
3177 |
} |
c255a4580 memcg: rename con... |
3178 |
#ifdef CONFIG_MEMCG |
66e1707bc Memory controller... |
3179 |
|
d2e5fb927 mm, memcg: do not... |
3180 |
/* Only used by soft limit reclaim. Do not reuse for anything else. */ |
a9dd0a831 mm, vmscan: make ... |
3181 |
unsigned long mem_cgroup_shrink_node(struct mem_cgroup *memcg, |
4e4169535 memory controller... |
3182 |
gfp_t gfp_mask, bool noswap, |
ef8f23279 mm, memcg: move m... |
3183 |
pg_data_t *pgdat, |
0ae5e89c6 memcg: count the ... |
3184 |
unsigned long *nr_scanned) |
4e4169535 memory controller... |
3185 3186 |
{ struct scan_control sc = { |
b8f5c5664 memcg: sc.nr_to_r... |
3187 |
.nr_to_reclaim = SWAP_CLUSTER_MAX, |
ee814fe23 mm: vmscan: clean... |
3188 |
.target_mem_cgroup = memcg, |
4e4169535 memory controller... |
3189 3190 |
.may_writepage = !laptop_mode, .may_unmap = 1, |
b2e18757f mm, vmscan: begin... |
3191 |
.reclaim_idx = MAX_NR_ZONES - 1, |
4e4169535 memory controller... |
3192 |
.may_swap = !noswap, |
4e4169535 memory controller... |
3193 |
}; |
6b4f7799c mm: vmscan: invok... |
3194 |
unsigned long lru_pages; |
0ae5e89c6 memcg: count the ... |
3195 |
|
d2e5fb927 mm, memcg: do not... |
3196 |
WARN_ON_ONCE(!current->reclaim_state); |
4e4169535 memory controller... |
3197 3198 |
sc.gfp_mask = (gfp_mask & GFP_RECLAIM_MASK) | (GFP_HIGHUSER_MOVABLE & ~GFP_RECLAIM_MASK); |
bdce6d9eb memcg, vmscan: ad... |
3199 |
|
9e3b2f8cd mm/vmscan: store ... |
3200 |
trace_mm_vmscan_memcg_softlimit_reclaim_begin(sc.order, |
3481c37ff mm/vmscan: drop m... |
3201 |
sc.gfp_mask); |
bdce6d9eb memcg, vmscan: ad... |
3202 |
|
4e4169535 memory controller... |
3203 3204 3205 |
/* * NOTE: Although we can get the priority field, using it * here is not a good idea, since it limits the pages we can scan. |
a9dd0a831 mm, vmscan: make ... |
3206 |
* if we don't reclaim here, the shrink_node from balance_pgdat |
4e4169535 memory controller... |
3207 3208 3209 |
* will pick up pages from other mem cgroup's as well. We hack * the priority and make it zero. */ |
ef8f23279 mm, memcg: move m... |
3210 |
shrink_node_memcg(pgdat, memcg, &sc, &lru_pages); |
bdce6d9eb memcg, vmscan: ad... |
3211 3212 |
trace_mm_vmscan_memcg_softlimit_reclaim_end(sc.nr_reclaimed); |
0ae5e89c6 memcg: count the ... |
3213 |
*nr_scanned = sc.nr_scanned; |
0308f7cf1 mm/vmscan.c: calc... |
3214 |
|
4e4169535 memory controller... |
3215 3216 |
return sc.nr_reclaimed; } |
72835c86c mm: unify remaini... |
3217 |
unsigned long try_to_free_mem_cgroup_pages(struct mem_cgroup *memcg, |
b70a2a21d mm: memcontrol: f... |
3218 |
unsigned long nr_pages, |
a7885eb8a memcg: swappiness |
3219 |
gfp_t gfp_mask, |
b70a2a21d mm: memcontrol: f... |
3220 |
bool may_swap) |
66e1707bc Memory controller... |
3221 |
{ |
4e4169535 memory controller... |
3222 |
struct zonelist *zonelist; |
bdce6d9eb memcg, vmscan: ad... |
3223 |
unsigned long nr_reclaimed; |
eb414681d psi: pressure sta... |
3224 |
unsigned long pflags; |
889976dbc memcg: reclaim me... |
3225 |
int nid; |
499118e96 mm: introduce mem... |
3226 |
unsigned int noreclaim_flag; |
66e1707bc Memory controller... |
3227 |
struct scan_control sc = { |
b70a2a21d mm: memcontrol: f... |
3228 |
.nr_to_reclaim = max(nr_pages, SWAP_CLUSTER_MAX), |
7dea19f9e mm: introduce mem... |
3229 |
.gfp_mask = (current_gfp_context(gfp_mask) & GFP_RECLAIM_MASK) | |
a09ed5e00 vmscan: change sh... |
3230 |
(GFP_HIGHUSER_MOVABLE & ~GFP_RECLAIM_MASK), |
b2e18757f mm, vmscan: begin... |
3231 |
.reclaim_idx = MAX_NR_ZONES - 1, |
ee814fe23 mm: vmscan: clean... |
3232 3233 3234 3235 |
.target_mem_cgroup = memcg, .priority = DEF_PRIORITY, .may_writepage = !laptop_mode, .may_unmap = 1, |
b70a2a21d mm: memcontrol: f... |
3236 |
.may_swap = may_swap, |
a09ed5e00 vmscan: change sh... |
3237 |
}; |
66e1707bc Memory controller... |
3238 |
|
1732d2b01 mm/vmscan.c: add ... |
3239 |
set_task_reclaim_state(current, &sc.reclaim_state); |
889976dbc memcg: reclaim me... |
3240 3241 3242 3243 3244 |
/* * Unlike direct reclaim via alloc_pages(), memcg's reclaim doesn't * take care of from where we get pages. So the node where we start the * scan does not need to be the current node. */ |
72835c86c mm: unify remaini... |
3245 |
nid = mem_cgroup_select_victim_node(memcg); |
889976dbc memcg: reclaim me... |
3246 |
|
c9634cf01 mm: use zonelist ... |
3247 |
zonelist = &NODE_DATA(nid)->node_zonelists[ZONELIST_FALLBACK]; |
bdce6d9eb memcg, vmscan: ad... |
3248 |
|
3481c37ff mm/vmscan: drop m... |
3249 |
trace_mm_vmscan_memcg_reclaim_begin(0, sc.gfp_mask); |
bdce6d9eb memcg, vmscan: ad... |
3250 |
|
eb414681d psi: pressure sta... |
3251 |
psi_memstall_enter(&pflags); |
499118e96 mm: introduce mem... |
3252 |
noreclaim_flag = memalloc_noreclaim_save(); |
eb414681d psi: pressure sta... |
3253 |
|
3115cd914 mm: vmscan: remov... |
3254 |
nr_reclaimed = do_try_to_free_pages(zonelist, &sc); |
eb414681d psi: pressure sta... |
3255 |
|
499118e96 mm: introduce mem... |
3256 |
memalloc_noreclaim_restore(noreclaim_flag); |
eb414681d psi: pressure sta... |
3257 |
psi_memstall_leave(&pflags); |
bdce6d9eb memcg, vmscan: ad... |
3258 3259 |
trace_mm_vmscan_memcg_reclaim_end(nr_reclaimed); |
1732d2b01 mm/vmscan.c: add ... |
3260 |
set_task_reclaim_state(current, NULL); |
bdce6d9eb memcg, vmscan: ad... |
3261 3262 |
return nr_reclaimed; |
66e1707bc Memory controller... |
3263 3264 |
} #endif |
1d82de618 mm, vmscan: make ... |
3265 |
static void age_active_anon(struct pglist_data *pgdat, |
ef8f23279 mm, memcg: move m... |
3266 |
struct scan_control *sc) |
f16015fbf mm: vmscan: disti... |
3267 |
{ |
b95a2f2d4 mm: vmscan: conve... |
3268 |
struct mem_cgroup *memcg; |
f16015fbf mm: vmscan: disti... |
3269 |
|
b95a2f2d4 mm: vmscan: conve... |
3270 3271 3272 3273 3274 |
if (!total_swap_pages) return; memcg = mem_cgroup_iter(NULL, NULL, NULL); do { |
ef8f23279 mm, memcg: move m... |
3275 |
struct lruvec *lruvec = mem_cgroup_lruvec(pgdat, memcg); |
b95a2f2d4 mm: vmscan: conve... |
3276 |
|
3b991208b mm: fix inactive ... |
3277 |
if (inactive_list_is_low(lruvec, false, sc, true)) |
1a93be0e7 mm/vmscan: push l... |
3278 |
shrink_active_list(SWAP_CLUSTER_MAX, lruvec, |
9e3b2f8cd mm/vmscan: store ... |
3279 |
sc, LRU_ACTIVE_ANON); |
b95a2f2d4 mm: vmscan: conve... |
3280 3281 3282 |
memcg = mem_cgroup_iter(NULL, memcg, NULL); } while (memcg); |
f16015fbf mm: vmscan: disti... |
3283 |
} |
1c30844d2 mm: reclaim small... |
3284 3285 3286 3287 3288 3289 3290 3291 3292 3293 3294 3295 3296 3297 3298 3299 3300 3301 3302 3303 3304 3305 3306 |
static bool pgdat_watermark_boosted(pg_data_t *pgdat, int classzone_idx) { int i; struct zone *zone; /* * Check for watermark boosts top-down as the higher zones * are more likely to be boosted. Both watermarks and boosts * should not be checked at the time time as reclaim would * start prematurely when there is no boosting and a lower * zone is balanced. */ for (i = classzone_idx; i >= 0; i--) { zone = pgdat->node_zones + i; if (!managed_zone(zone)) continue; if (zone->watermark_boost) return true; } return false; } |
e716f2eb2 mm, vmscan: preve... |
3307 3308 3309 3310 3311 |
/* * Returns true if there is an eligible zone balanced for the request order * and classzone_idx */ static bool pgdat_balanced(pg_data_t *pgdat, int order, int classzone_idx) |
60cefed48 mm: vmscan: fix e... |
3312 |
{ |
e716f2eb2 mm, vmscan: preve... |
3313 3314 3315 |
int i; unsigned long mark = -1; struct zone *zone; |
60cefed48 mm: vmscan: fix e... |
3316 |
|
1c30844d2 mm: reclaim small... |
3317 3318 3319 3320 |
/* * Check watermarks bottom-up as lower zones are more likely to * meet watermarks. */ |
e716f2eb2 mm, vmscan: preve... |
3321 3322 |
for (i = 0; i <= classzone_idx; i++) { zone = pgdat->node_zones + i; |
6256c6b49 mm, vmscan: remov... |
3323 |
|
e716f2eb2 mm, vmscan: preve... |
3324 3325 3326 3327 3328 3329 3330 3331 3332 3333 3334 3335 3336 3337 3338 3339 3340 |
if (!managed_zone(zone)) continue; mark = high_wmark_pages(zone); if (zone_watermark_ok_safe(zone, order, mark, classzone_idx)) return true; } /* * If a node has no populated zone within classzone_idx, it does not * need balancing by definition. This can happen if a zone-restricted * allocation tries to wake a remote kswapd. */ if (mark == -1) return true; return false; |
60cefed48 mm: vmscan: fix e... |
3341 |
} |
631b6e083 mm, vmscan: only ... |
3342 3343 3344 3345 3346 3347 3348 |
/* Clear pgdat state for congested, dirty or under writeback. */ static void clear_pgdat_congested(pg_data_t *pgdat) { clear_bit(PGDAT_CONGESTED, &pgdat->flags); clear_bit(PGDAT_DIRTY, &pgdat->flags); clear_bit(PGDAT_WRITEBACK, &pgdat->flags); } |
1741c8775 mm: kswapd: keep ... |
3349 |
/* |
5515061d2 mm: throttle dire... |
3350 3351 3352 3353 3354 |
* Prepare kswapd for sleeping. This verifies that there are no processes * waiting in throttle_direct_reclaim() and that watermarks have been met. * * Returns true if kswapd is ready to sleep */ |
d9f21d426 mm, vmscan: avoid... |
3355 |
static bool prepare_kswapd_sleep(pg_data_t *pgdat, int order, int classzone_idx) |
f50de2d38 vmscan: have kswa... |
3356 |
{ |
5515061d2 mm: throttle dire... |
3357 |
/* |
9e5e36617 mm, vmscan: preve... |
3358 |
* The throttled processes are normally woken up in balance_pgdat() as |
c73322d09 mm: fix 100% CPU ... |
3359 |
* soon as allow_direct_reclaim() is true. But there is a potential |
9e5e36617 mm, vmscan: preve... |
3360 3361 3362 3363 3364 3365 3366 3367 3368 |
* race between when kswapd checks the watermarks and a process gets * throttled. There is also a potential race if processes get * throttled, kswapd wakes, a large process exits thereby balancing the * zones, which causes kswapd to exit balance_pgdat() before reaching * the wake up checks. If kswapd is going to sleep, no process should * be sleeping on pfmemalloc_wait, so wake them now if necessary. If * the wake up is premature, processes will wake kswapd and get * throttled again. The difference from wake ups in balance_pgdat() is * that here we are under prepare_to_wait(). |
5515061d2 mm: throttle dire... |
3369 |
*/ |
9e5e36617 mm, vmscan: preve... |
3370 3371 |
if (waitqueue_active(&pgdat->pfmemalloc_wait)) wake_up_all(&pgdat->pfmemalloc_wait); |
f50de2d38 vmscan: have kswa... |
3372 |
|
c73322d09 mm: fix 100% CPU ... |
3373 3374 3375 |
/* Hopeless node, leave it to direct reclaim */ if (pgdat->kswapd_failures >= MAX_RECLAIM_RETRIES) return true; |
e716f2eb2 mm, vmscan: preve... |
3376 3377 3378 |
if (pgdat_balanced(pgdat, order, classzone_idx)) { clear_pgdat_congested(pgdat); return true; |
1d82de618 mm, vmscan: make ... |
3379 |
} |
333b0a459 mm, vmscan: fix z... |
3380 |
return false; |
f50de2d38 vmscan: have kswa... |
3381 |
} |
1da177e4c Linux-2.6.12-rc2 |
3382 |
/* |
1d82de618 mm, vmscan: make ... |
3383 3384 |
* kswapd shrinks a node of pages that are at or below the highest usable * zone that is currently unbalanced. |
b8e83b942 mm: vmscan: flatt... |
3385 3386 |
* * Returns true if kswapd scanned at least the requested number of pages to |
283aba9f9 mm: vmscan: block... |
3387 3388 |
* reclaim or if the lack of progress was due to pages under writeback. * This is used to determine if the scanning priority needs to be raised. |
75485363c mm: vmscan: limit... |
3389 |
*/ |
1d82de618 mm, vmscan: make ... |
3390 |
static bool kswapd_shrink_node(pg_data_t *pgdat, |
accf62422 mm, kswapd: repla... |
3391 |
struct scan_control *sc) |
75485363c mm: vmscan: limit... |
3392 |
{ |
1d82de618 mm, vmscan: make ... |
3393 3394 |
struct zone *zone; int z; |
75485363c mm: vmscan: limit... |
3395 |
|
1d82de618 mm, vmscan: make ... |
3396 3397 |
/* Reclaim a number of pages proportional to the number of zones */ sc->nr_to_reclaim = 0; |
970a39a36 mm, vmscan: avoid... |
3398 |
for (z = 0; z <= sc->reclaim_idx; z++) { |
1d82de618 mm, vmscan: make ... |
3399 |
zone = pgdat->node_zones + z; |
6aa303def mm, vmscan: only ... |
3400 |
if (!managed_zone(zone)) |
1d82de618 mm, vmscan: make ... |
3401 |
continue; |
7c954f6de mm: vmscan: move ... |
3402 |
|
1d82de618 mm, vmscan: make ... |
3403 3404 |
sc->nr_to_reclaim += max(high_wmark_pages(zone), SWAP_CLUSTER_MAX); } |
7c954f6de mm: vmscan: move ... |
3405 3406 |
/* |
1d82de618 mm, vmscan: make ... |
3407 3408 |
* Historically care was taken to put equal pressure on all zones but * now pressure is applied based on node LRU order. |
7c954f6de mm: vmscan: move ... |
3409 |
*/ |
970a39a36 mm, vmscan: avoid... |
3410 |
shrink_node(pgdat, sc); |
283aba9f9 mm: vmscan: block... |
3411 |
|
7c954f6de mm: vmscan: move ... |
3412 |
/* |
1d82de618 mm, vmscan: make ... |
3413 3414 3415 3416 3417 |
* Fragmentation may mean that the system cannot be rebalanced for * high-order allocations. If twice the allocation size has been * reclaimed then recheck watermarks only at order-0 to prevent * excessive reclaim. Assume that a process requested a high-order * can direct reclaim/compact. |
7c954f6de mm: vmscan: move ... |
3418 |
*/ |
9861a62c3 mm, compaction: c... |
3419 |
if (sc->order && sc->nr_reclaimed >= compact_gap(sc->order)) |
1d82de618 mm, vmscan: make ... |
3420 |
sc->order = 0; |
7c954f6de mm: vmscan: move ... |
3421 |
|
b8e83b942 mm: vmscan: flatt... |
3422 |
return sc->nr_scanned >= sc->nr_to_reclaim; |
75485363c mm: vmscan: limit... |
3423 3424 3425 |
} /* |
1d82de618 mm, vmscan: make ... |
3426 3427 3428 |
* For kswapd, balance_pgdat() will reclaim pages across a node from zones * that are eligible for use by the caller until at least one zone is * balanced. |
1da177e4c Linux-2.6.12-rc2 |
3429 |
* |
1d82de618 mm, vmscan: make ... |
3430 |
* Returns the order kswapd finished reclaiming at. |
1da177e4c Linux-2.6.12-rc2 |
3431 3432 |
* * kswapd scans the zones in the highmem->normal->dma direction. It skips |
418589663 page allocator: u... |
3433 |
* zones which have free_pages > high_wmark_pages(zone), but once a zone is |
8bb4e7a2e mm: fix some typo... |
3434 |
* found to have free_pages <= high_wmark_pages(zone), any page in that zone |
1d82de618 mm, vmscan: make ... |
3435 3436 |
* or lower is eligible for reclaim until at least one usable zone is * balanced. |
1da177e4c Linux-2.6.12-rc2 |
3437 |
*/ |
accf62422 mm, kswapd: repla... |
3438 |
static int balance_pgdat(pg_data_t *pgdat, int order, int classzone_idx) |
1da177e4c Linux-2.6.12-rc2 |
3439 |
{ |
1da177e4c Linux-2.6.12-rc2 |
3440 |
int i; |
0608f43da revert "memcg, vm... |
3441 3442 |
unsigned long nr_soft_reclaimed; unsigned long nr_soft_scanned; |
eb414681d psi: pressure sta... |
3443 |
unsigned long pflags; |
1c30844d2 mm: reclaim small... |
3444 3445 3446 |
unsigned long nr_boost_reclaim; unsigned long zone_boosts[MAX_NR_ZONES] = { 0, }; bool boosted; |
1d82de618 mm, vmscan: make ... |
3447 |
struct zone *zone; |
179e96395 [PATCH] vmscan: s... |
3448 3449 |
struct scan_control sc = { .gfp_mask = GFP_KERNEL, |
ee814fe23 mm: vmscan: clean... |
3450 |
.order = order, |
a6dc60f89 vmscan: rename sc... |
3451 |
.may_unmap = 1, |
179e96395 [PATCH] vmscan: s... |
3452 |
}; |
93781325d lockdep: fix fs_r... |
3453 |
|
1732d2b01 mm/vmscan.c: add ... |
3454 |
set_task_reclaim_state(current, &sc.reclaim_state); |
eb414681d psi: pressure sta... |
3455 |
psi_memstall_enter(&pflags); |
93781325d lockdep: fix fs_r... |
3456 |
__fs_reclaim_acquire(); |
f8891e5e1 [PATCH] Light wei... |
3457 |
count_vm_event(PAGEOUTRUN); |
1da177e4c Linux-2.6.12-rc2 |
3458 |
|
1c30844d2 mm: reclaim small... |
3459 3460 3461 3462 3463 3464 3465 3466 3467 3468 3469 3470 3471 3472 3473 3474 3475 3476 |
/* * Account for the reclaim boost. Note that the zone boost is left in * place so that parallel allocations that are near the watermark will * stall or direct reclaim until kswapd is finished. */ nr_boost_reclaim = 0; for (i = 0; i <= classzone_idx; i++) { zone = pgdat->node_zones + i; if (!managed_zone(zone)) continue; nr_boost_reclaim += zone->watermark_boost; zone_boosts[i] = zone->watermark_boost; } boosted = nr_boost_reclaim; restart: sc.priority = DEF_PRIORITY; |
9e3b2f8cd mm/vmscan: store ... |
3477 |
do { |
c73322d09 mm: fix 100% CPU ... |
3478 |
unsigned long nr_reclaimed = sc.nr_reclaimed; |
b8e83b942 mm: vmscan: flatt... |
3479 |
bool raise_priority = true; |
1c30844d2 mm: reclaim small... |
3480 |
bool balanced; |
93781325d lockdep: fix fs_r... |
3481 |
bool ret; |
b8e83b942 mm: vmscan: flatt... |
3482 |
|
84c7a7771 mm, vmscan: Have ... |
3483 |
sc.reclaim_idx = classzone_idx; |
1da177e4c Linux-2.6.12-rc2 |
3484 |
|
86c79f6b5 mm: vmscan: do no... |
3485 |
/* |
84c7a7771 mm, vmscan: Have ... |
3486 3487 3488 3489 3490 3491 3492 3493 |
* If the number of buffer_heads exceeds the maximum allowed * then consider reclaiming from all zones. This has a dual * purpose -- on 64-bit systems it is expected that * buffer_heads are stripped during active rotation. On 32-bit * systems, highmem pages can pin lowmem memory and shrinking * buffers can relieve lowmem pressure. Reclaim may still not * go ahead if all eligible zones for the original allocation * request are balanced to avoid excessive reclaim from kswapd. |
86c79f6b5 mm: vmscan: do no... |
3494 3495 3496 3497 |
*/ if (buffer_heads_over_limit) { for (i = MAX_NR_ZONES - 1; i >= 0; i--) { zone = pgdat->node_zones + i; |
6aa303def mm, vmscan: only ... |
3498 |
if (!managed_zone(zone)) |
86c79f6b5 mm: vmscan: do no... |
3499 |
continue; |
cc715d99e mm: vmscan: forci... |
3500 |
|
970a39a36 mm, vmscan: avoid... |
3501 |
sc.reclaim_idx = i; |
e1dbeda60 [PATCH] balance_p... |
3502 |
break; |
1da177e4c Linux-2.6.12-rc2 |
3503 |
} |
1da177e4c Linux-2.6.12-rc2 |
3504 |
} |
dafcb73e3 mm: avoid calling... |
3505 |
|
86c79f6b5 mm: vmscan: do no... |
3506 |
/* |
1c30844d2 mm: reclaim small... |
3507 3508 3509 3510 3511 3512 3513 3514 3515 3516 3517 3518 3519 3520 3521 3522 |
* If the pgdat is imbalanced then ignore boosting and preserve * the watermarks for a later time and restart. Note that the * zone watermarks will be still reset at the end of balancing * on the grounds that the normal reclaim should be enough to * re-evaluate if boosting is required when kswapd next wakes. */ balanced = pgdat_balanced(pgdat, sc.order, classzone_idx); if (!balanced && nr_boost_reclaim) { nr_boost_reclaim = 0; goto restart; } /* * If boosting is not active then only reclaim if there are no * eligible zones. Note that sc.reclaim_idx is not used as * buffer_heads_over_limit may have adjusted it. |
86c79f6b5 mm: vmscan: do no... |
3523 |
*/ |
1c30844d2 mm: reclaim small... |
3524 |
if (!nr_boost_reclaim && balanced) |
e716f2eb2 mm, vmscan: preve... |
3525 |
goto out; |
e1dbeda60 [PATCH] balance_p... |
3526 |
|
1c30844d2 mm: reclaim small... |
3527 3528 3529 3530 3531 3532 3533 3534 3535 3536 3537 3538 |
/* Limit the priority of boosting to avoid reclaim writeback */ if (nr_boost_reclaim && sc.priority == DEF_PRIORITY - 2) raise_priority = false; /* * Do not writeback or swap pages for boosted reclaim. The * intent is to relieve pressure not issue sub-optimal IO * from reclaim context. If no pages are reclaimed, the * reclaim will be aborted. */ sc.may_writepage = !laptop_mode && !nr_boost_reclaim; sc.may_swap = !nr_boost_reclaim; |
1c30844d2 mm: reclaim small... |
3539 |
|
1da177e4c Linux-2.6.12-rc2 |
3540 |
/* |
1d82de618 mm, vmscan: make ... |
3541 3542 3543 3544 3545 |
* Do some background aging of the anon list, to give * pages a chance to be referenced before reclaiming. All * pages are rotated regardless of classzone as this is * about consistent aging. */ |
ef8f23279 mm, memcg: move m... |
3546 |
age_active_anon(pgdat, &sc); |
1d82de618 mm, vmscan: make ... |
3547 3548 |
/* |
b7ea3c417 mm: vmscan: check... |
3549 3550 3551 |
* If we're getting trouble reclaiming, start doing writepage * even in laptop mode. */ |
047d72c30 mm: remove seemin... |
3552 |
if (sc.priority < DEF_PRIORITY - 2) |
b7ea3c417 mm: vmscan: check... |
3553 |
sc.may_writepage = 1; |
1d82de618 mm, vmscan: make ... |
3554 3555 3556 |
/* Call soft limit reclaim before calling shrink_node. */ sc.nr_scanned = 0; nr_soft_scanned = 0; |
ef8f23279 mm, memcg: move m... |
3557 |
nr_soft_reclaimed = mem_cgroup_soft_limit_reclaim(pgdat, sc.order, |
1d82de618 mm, vmscan: make ... |
3558 3559 |
sc.gfp_mask, &nr_soft_scanned); sc.nr_reclaimed += nr_soft_reclaimed; |
b7ea3c417 mm: vmscan: check... |
3560 |
/* |
1d82de618 mm, vmscan: make ... |
3561 3562 3563 |
* There should be no need to raise the scanning priority if * enough pages are already being scanned that that high * watermark would be met at 100% efficiency. |
1da177e4c Linux-2.6.12-rc2 |
3564 |
*/ |
970a39a36 mm, vmscan: avoid... |
3565 |
if (kswapd_shrink_node(pgdat, &sc)) |
1d82de618 mm, vmscan: make ... |
3566 |
raise_priority = false; |
5515061d2 mm: throttle dire... |
3567 3568 3569 3570 3571 3572 3573 |
/* * If the low watermark is met there is no need for processes * to be throttled on pfmemalloc_wait as they should not be * able to safely make forward progress. Wake them */ if (waitqueue_active(&pgdat->pfmemalloc_wait) && |
c73322d09 mm: fix 100% CPU ... |
3574 |
allow_direct_reclaim(pgdat)) |
cfc511557 mm, vmscan: wake ... |
3575 |
wake_up_all(&pgdat->pfmemalloc_wait); |
5515061d2 mm: throttle dire... |
3576 |
|
b8e83b942 mm: vmscan: flatt... |
3577 |
/* Check if kswapd should be suspending */ |
93781325d lockdep: fix fs_r... |
3578 3579 3580 3581 |
__fs_reclaim_release(); ret = try_to_freeze(); __fs_reclaim_acquire(); if (ret || kthread_should_stop()) |
b8e83b942 mm: vmscan: flatt... |
3582 |
break; |
8357376d3 [PATCH] swsusp: I... |
3583 |
|
73ce02e96 mm: stop kswapd's... |
3584 |
/* |
b8e83b942 mm: vmscan: flatt... |
3585 3586 |
* Raise priority if scanning rate is too low or there was no * progress in reclaiming pages |
73ce02e96 mm: stop kswapd's... |
3587 |
*/ |
c73322d09 mm: fix 100% CPU ... |
3588 |
nr_reclaimed = sc.nr_reclaimed - nr_reclaimed; |
1c30844d2 mm: reclaim small... |
3589 3590 3591 3592 3593 3594 3595 3596 3597 |
nr_boost_reclaim -= min(nr_boost_reclaim, nr_reclaimed); /* * If reclaim made no progress for a boost, stop reclaim as * IO cannot be queued and it could be an infinite loop in * extreme circumstances. */ if (nr_boost_reclaim && !nr_reclaimed) break; |
c73322d09 mm: fix 100% CPU ... |
3598 |
if (raise_priority || !nr_reclaimed) |
b8e83b942 mm: vmscan: flatt... |
3599 |
sc.priority--; |
1d82de618 mm, vmscan: make ... |
3600 |
} while (sc.priority >= 1); |
1da177e4c Linux-2.6.12-rc2 |
3601 |
|
c73322d09 mm: fix 100% CPU ... |
3602 3603 |
if (!sc.nr_reclaimed) pgdat->kswapd_failures++; |
b8e83b942 mm: vmscan: flatt... |
3604 |
out: |
1c30844d2 mm: reclaim small... |
3605 3606 3607 3608 3609 3610 3611 3612 3613 3614 3615 3616 3617 3618 3619 3620 3621 3622 3623 3624 3625 |
/* If reclaim was boosted, account for the reclaim done in this pass */ if (boosted) { unsigned long flags; for (i = 0; i <= classzone_idx; i++) { if (!zone_boosts[i]) continue; /* Increments are under the zone lock */ zone = pgdat->node_zones + i; spin_lock_irqsave(&zone->lock, flags); zone->watermark_boost -= min(zone->watermark_boost, zone_boosts[i]); spin_unlock_irqrestore(&zone->lock, flags); } /* * As there is now likely space, wakeup kcompact to defragment * pageblocks. */ wakeup_kcompactd(pgdat, pageblock_order, classzone_idx); } |
2a2e48854 mm: vmscan: fix I... |
3626 |
snapshot_refaults(NULL, pgdat); |
93781325d lockdep: fix fs_r... |
3627 |
__fs_reclaim_release(); |
eb414681d psi: pressure sta... |
3628 |
psi_memstall_leave(&pflags); |
1732d2b01 mm/vmscan.c: add ... |
3629 |
set_task_reclaim_state(current, NULL); |
e5ca8071f mm/vmscan.c: add ... |
3630 |
|
0abdee2bd mm: kswapd: use t... |
3631 |
/* |
1d82de618 mm, vmscan: make ... |
3632 3633 3634 3635 |
* Return the order kswapd stopped reclaiming at as * prepare_kswapd_sleep() takes it into account. If another caller * entered the allocator slow path while kswapd was awake, order will * remain at the higher level. |
0abdee2bd mm: kswapd: use t... |
3636 |
*/ |
1d82de618 mm, vmscan: make ... |
3637 |
return sc.order; |
1da177e4c Linux-2.6.12-rc2 |
3638 |
} |
e716f2eb2 mm, vmscan: preve... |
3639 |
/* |
dffcac2cb mm/vmscan.c: prev... |
3640 3641 3642 3643 3644 |
* The pgdat->kswapd_classzone_idx is used to pass the highest zone index to be * reclaimed by kswapd from the waker. If the value is MAX_NR_ZONES which is not * a valid index then either kswapd runs for first time or kswapd couldn't sleep * after previous reclaim attempt (node is still unbalanced). In that case * return the zone index of the previous kswapd reclaim cycle. |
e716f2eb2 mm, vmscan: preve... |
3645 3646 |
*/ static enum zone_type kswapd_classzone_idx(pg_data_t *pgdat, |
dffcac2cb mm/vmscan.c: prev... |
3647 |
enum zone_type prev_classzone_idx) |
e716f2eb2 mm, vmscan: preve... |
3648 |
{ |
19b9ff114 mm/vmscan.c: fix ... |
3649 3650 3651 |
enum zone_type curr_idx = READ_ONCE(pgdat->kswapd_classzone_idx); return curr_idx == MAX_NR_ZONES ? prev_classzone_idx : curr_idx; |
e716f2eb2 mm, vmscan: preve... |
3652 |
} |
38087d9b0 mm, vmscan: simpl... |
3653 3654 |
static void kswapd_try_to_sleep(pg_data_t *pgdat, int alloc_order, int reclaim_order, unsigned int classzone_idx) |
f0bc0a60b vmscan: factor ou... |
3655 3656 3657 3658 3659 3660 3661 3662 |
{ long remaining = 0; DEFINE_WAIT(wait); if (freezing(current) || kthread_should_stop()) return; prepare_to_wait(&pgdat->kswapd_wait, &wait, TASK_INTERRUPTIBLE); |
333b0a459 mm, vmscan: fix z... |
3663 3664 3665 3666 3667 3668 3669 |
/* * Try to sleep for a short interval. Note that kcompactd will only be * woken if it is possible to sleep for a short interval. This is * deliberate on the assumption that if reclaim cannot keep an * eligible zone balanced that it's also unlikely that compaction will * succeed. */ |
d9f21d426 mm, vmscan: avoid... |
3670 |
if (prepare_kswapd_sleep(pgdat, reclaim_order, classzone_idx)) { |
fd901c953 mm: wake kcompact... |
3671 3672 3673 3674 3675 3676 3677 3678 3679 3680 3681 3682 |
/* * Compaction records what page blocks it recently failed to * isolate pages from and skips them in the future scanning. * When kswapd is going to sleep, it is reasonable to assume * that pages and compaction may succeed so reset the cache. */ reset_isolation_suitable(pgdat); /* * We have freed the memory, now we should compact it to make * allocation of the requested order possible. */ |
38087d9b0 mm, vmscan: simpl... |
3683 |
wakeup_kcompactd(pgdat, alloc_order, classzone_idx); |
fd901c953 mm: wake kcompact... |
3684 |
|
f0bc0a60b vmscan: factor ou... |
3685 |
remaining = schedule_timeout(HZ/10); |
38087d9b0 mm, vmscan: simpl... |
3686 3687 3688 3689 3690 3691 3692 |
/* * If woken prematurely then reset kswapd_classzone_idx and * order. The values will either be from a wakeup request or * the previous request that slept prematurely. */ if (remaining) { |
19b9ff114 mm/vmscan.c: fix ... |
3693 3694 3695 3696 3697 |
WRITE_ONCE(pgdat->kswapd_classzone_idx, kswapd_classzone_idx(pgdat, classzone_idx)); if (READ_ONCE(pgdat->kswapd_order) < reclaim_order) WRITE_ONCE(pgdat->kswapd_order, reclaim_order); |
38087d9b0 mm, vmscan: simpl... |
3698 |
} |
f0bc0a60b vmscan: factor ou... |
3699 3700 3701 3702 3703 3704 3705 3706 |
finish_wait(&pgdat->kswapd_wait, &wait); prepare_to_wait(&pgdat->kswapd_wait, &wait, TASK_INTERRUPTIBLE); } /* * After a short sleep, check if it was a premature sleep. If not, then * go fully to sleep until explicitly woken up. */ |
d9f21d426 mm, vmscan: avoid... |
3707 3708 |
if (!remaining && prepare_kswapd_sleep(pgdat, reclaim_order, classzone_idx)) { |
f0bc0a60b vmscan: factor ou... |
3709 3710 3711 3712 3713 3714 3715 3716 3717 3718 3719 |
trace_mm_vmscan_kswapd_sleep(pgdat->node_id); /* * vmstat counters are not perfectly accurate and the estimated * value for counters such as NR_FREE_PAGES can deviate from the * true value by nr_online_cpus * threshold. To avoid the zone * watermarks being breached while under pressure, we reduce the * per-cpu vmstat threshold while kswapd is awake and restore * them before going back to sleep. */ set_pgdat_percpu_threshold(pgdat, calculate_normal_threshold); |
1c7e7f6c0 mm: fix lost kswa... |
3720 3721 3722 |
if (!kthread_should_stop()) schedule(); |
f0bc0a60b vmscan: factor ou... |
3723 3724 3725 3726 3727 3728 3729 3730 3731 |
set_pgdat_percpu_threshold(pgdat, calculate_pressure_threshold); } else { if (remaining) count_vm_event(KSWAPD_LOW_WMARK_HIT_QUICKLY); else count_vm_event(KSWAPD_HIGH_WMARK_HIT_QUICKLY); } finish_wait(&pgdat->kswapd_wait, &wait); } |
1da177e4c Linux-2.6.12-rc2 |
3732 3733 |
/* * The background pageout daemon, started as a kernel thread |
4f98a2fee vmscan: split LRU... |
3734 |
* from the init process. |
1da177e4c Linux-2.6.12-rc2 |
3735 3736 3737 3738 3739 3740 3741 3742 3743 3744 3745 3746 |
* * This basically trickles out pages so that we have _some_ * free memory available even if there is no other activity * that frees anything up. This is needed for things like routing * etc, where we otherwise might have all activity going on in * asynchronous contexts that cannot page things out. * * If there are applications that are active memory-allocators * (most normal use), this basically shouldn't matter. */ static int kswapd(void *p) { |
e716f2eb2 mm, vmscan: preve... |
3747 3748 |
unsigned int alloc_order, reclaim_order; unsigned int classzone_idx = MAX_NR_ZONES - 1; |
1da177e4c Linux-2.6.12-rc2 |
3749 3750 |
pg_data_t *pgdat = (pg_data_t*)p; struct task_struct *tsk = current; |
a70f73028 cpumask: replace ... |
3751 |
const struct cpumask *cpumask = cpumask_of_node(pgdat->node_id); |
1da177e4c Linux-2.6.12-rc2 |
3752 |
|
174596a0b cpumask: convert mm/ |
3753 |
if (!cpumask_empty(cpumask)) |
c5f59f083 nodemask: use new... |
3754 |
set_cpus_allowed_ptr(tsk, cpumask); |
1da177e4c Linux-2.6.12-rc2 |
3755 3756 3757 3758 3759 3760 3761 3762 3763 3764 3765 3766 3767 |
/* * Tell the memory management that we're a "memory allocator", * and that if we need more memory we should get access to it * regardless (see "__alloc_pages()"). "kswapd" should * never get caught in the normal page freeing logic. * * (Kswapd normally doesn't need memory anyway, but sometimes * you need a small amount of memory in order to be able to * page out something else, and this flag essentially protects * us from recursively trying to free more memory as we're * trying to free the first piece of memory in the first place). */ |
930d91525 [PATCH] Swap Migr... |
3768 |
tsk->flags |= PF_MEMALLOC | PF_SWAPWRITE | PF_KSWAPD; |
831441862 Freezer: make ker... |
3769 |
set_freezable(); |
1da177e4c Linux-2.6.12-rc2 |
3770 |
|
19b9ff114 mm/vmscan.c: fix ... |
3771 3772 |
WRITE_ONCE(pgdat->kswapd_order, 0); WRITE_ONCE(pgdat->kswapd_classzone_idx, MAX_NR_ZONES); |
1da177e4c Linux-2.6.12-rc2 |
3773 |
for ( ; ; ) { |
6f6313d48 mm/vmscan.c: try_... |
3774 |
bool ret; |
3e1d1d28d [PATCH] Cleanup p... |
3775 |
|
19b9ff114 mm/vmscan.c: fix ... |
3776 |
alloc_order = reclaim_order = READ_ONCE(pgdat->kswapd_order); |
e716f2eb2 mm, vmscan: preve... |
3777 |
classzone_idx = kswapd_classzone_idx(pgdat, classzone_idx); |
38087d9b0 mm, vmscan: simpl... |
3778 3779 3780 |
kswapd_try_sleep: kswapd_try_to_sleep(pgdat, alloc_order, reclaim_order, classzone_idx); |
215ddd666 mm: vmscan: only ... |
3781 |
|
38087d9b0 mm, vmscan: simpl... |
3782 |
/* Read the new order and classzone_idx */ |
19b9ff114 mm/vmscan.c: fix ... |
3783 |
alloc_order = reclaim_order = READ_ONCE(pgdat->kswapd_order); |
dffcac2cb mm/vmscan.c: prev... |
3784 |
classzone_idx = kswapd_classzone_idx(pgdat, classzone_idx); |
19b9ff114 mm/vmscan.c: fix ... |
3785 3786 |
WRITE_ONCE(pgdat->kswapd_order, 0); WRITE_ONCE(pgdat->kswapd_classzone_idx, MAX_NR_ZONES); |
1da177e4c Linux-2.6.12-rc2 |
3787 |
|
8fe23e057 mm: clear node in... |
3788 3789 3790 3791 3792 3793 3794 3795 |
ret = try_to_freeze(); if (kthread_should_stop()) break; /* * We can speed up thawing tasks if we don't call balance_pgdat * after returning from the refrigerator */ |
38087d9b0 mm, vmscan: simpl... |
3796 3797 3798 3799 3800 3801 3802 3803 3804 3805 3806 |
if (ret) continue; /* * Reclaim begins at the requested order but if a high-order * reclaim fails then kswapd falls back to reclaiming for * order-0. If that happens, kswapd will consider sleeping * for the order it finished reclaiming at (reclaim_order) * but kcompactd is woken to compact for the original * request (alloc_order). */ |
e5146b12e mm, vmscan: add c... |
3807 3808 |
trace_mm_vmscan_kswapd_wake(pgdat->node_id, classzone_idx, alloc_order); |
38087d9b0 mm, vmscan: simpl... |
3809 3810 3811 |
reclaim_order = balance_pgdat(pgdat, alloc_order, classzone_idx); if (reclaim_order < alloc_order) goto kswapd_try_sleep; |
1da177e4c Linux-2.6.12-rc2 |
3812 |
} |
b0a8cc58e mm: bugfix: set c... |
3813 |
|
71abdc15a mm: vmscan: clear... |
3814 |
tsk->flags &= ~(PF_MEMALLOC | PF_SWAPWRITE | PF_KSWAPD); |
71abdc15a mm: vmscan: clear... |
3815 |
|
1da177e4c Linux-2.6.12-rc2 |
3816 3817 3818 3819 |
return 0; } /* |
5ecd9d403 mm, page_alloc: w... |
3820 3821 3822 3823 3824 |
* A zone is low on free memory or too fragmented for high-order memory. If * kswapd should reclaim (direct reclaim is deferred), wake it up for the zone's * pgdat. It will wake up kcompactd after reclaiming memory. If kswapd reclaim * has failed or is not needed, still wake up kcompactd if only compaction is * needed. |
1da177e4c Linux-2.6.12-rc2 |
3825 |
*/ |
5ecd9d403 mm, page_alloc: w... |
3826 3827 |
void wakeup_kswapd(struct zone *zone, gfp_t gfp_flags, int order, enum zone_type classzone_idx) |
1da177e4c Linux-2.6.12-rc2 |
3828 3829 |
{ pg_data_t *pgdat; |
19b9ff114 mm/vmscan.c: fix ... |
3830 |
enum zone_type curr_idx; |
1da177e4c Linux-2.6.12-rc2 |
3831 |
|
6aa303def mm, vmscan: only ... |
3832 |
if (!managed_zone(zone)) |
1da177e4c Linux-2.6.12-rc2 |
3833 |
return; |
5ecd9d403 mm, page_alloc: w... |
3834 |
if (!cpuset_zone_allowed(zone, gfp_flags)) |
1da177e4c Linux-2.6.12-rc2 |
3835 |
return; |
19b9ff114 mm/vmscan.c: fix ... |
3836 |
|
88f5acf88 mm: page allocato... |
3837 |
pgdat = zone->zone_pgdat; |
19b9ff114 mm/vmscan.c: fix ... |
3838 3839 3840 3841 3842 3843 3844 |
curr_idx = READ_ONCE(pgdat->kswapd_classzone_idx); if (curr_idx == MAX_NR_ZONES || curr_idx < classzone_idx) WRITE_ONCE(pgdat->kswapd_classzone_idx, classzone_idx); if (READ_ONCE(pgdat->kswapd_order) < order) WRITE_ONCE(pgdat->kswapd_order, order); |
dffcac2cb mm/vmscan.c: prev... |
3845 |
|
8d0986e28 [PATCH] vm: kswap... |
3846 |
if (!waitqueue_active(&pgdat->kswapd_wait)) |
1da177e4c Linux-2.6.12-rc2 |
3847 |
return; |
e1a556374 mm, vmscan: only ... |
3848 |
|
5ecd9d403 mm, page_alloc: w... |
3849 3850 |
/* Hopeless node, leave it to direct reclaim if possible */ if (pgdat->kswapd_failures >= MAX_RECLAIM_RETRIES || |
1c30844d2 mm: reclaim small... |
3851 3852 |
(pgdat_balanced(pgdat, order, classzone_idx) && !pgdat_watermark_boosted(pgdat, classzone_idx))) { |
5ecd9d403 mm, page_alloc: w... |
3853 3854 3855 3856 3857 3858 3859 3860 3861 |
/* * There may be plenty of free memory available, but it's too * fragmented for high-order allocations. Wake up kcompactd * and rely on compaction_suitable() to determine if it's * needed. If it fails, it will defer subsequent attempts to * ratelimit its work. */ if (!(gfp_flags & __GFP_DIRECT_RECLAIM)) wakeup_kcompactd(pgdat, order, classzone_idx); |
e716f2eb2 mm, vmscan: preve... |
3862 |
return; |
5ecd9d403 mm, page_alloc: w... |
3863 |
} |
88f5acf88 mm: page allocato... |
3864 |
|
5ecd9d403 mm, page_alloc: w... |
3865 3866 |
trace_mm_vmscan_wakeup_kswapd(pgdat->node_id, classzone_idx, order, gfp_flags); |
8d0986e28 [PATCH] vm: kswap... |
3867 |
wake_up_interruptible(&pgdat->kswapd_wait); |
1da177e4c Linux-2.6.12-rc2 |
3868 |
} |
c6f37f121 PM/Suspend: Do no... |
3869 |
#ifdef CONFIG_HIBERNATION |
1da177e4c Linux-2.6.12-rc2 |
3870 |
/* |
7b51755c3 vmscan: kill hibe... |
3871 |
* Try to free `nr_to_reclaim' of memory, system-wide, and return the number of |
d6277db4a [PATCH] swsusp: r... |
3872 3873 3874 3875 3876 |
* freed pages. * * Rather than trying to age LRUs the aim is to preserve the overall * LRU order by reclaiming preferentially * inactive > active > active referenced > active mapped |
1da177e4c Linux-2.6.12-rc2 |
3877 |
*/ |
7b51755c3 vmscan: kill hibe... |
3878 |
unsigned long shrink_all_memory(unsigned long nr_to_reclaim) |
1da177e4c Linux-2.6.12-rc2 |
3879 |
{ |
d6277db4a [PATCH] swsusp: r... |
3880 |
struct scan_control sc = { |
ee814fe23 mm: vmscan: clean... |
3881 |
.nr_to_reclaim = nr_to_reclaim, |
7b51755c3 vmscan: kill hibe... |
3882 |
.gfp_mask = GFP_HIGHUSER_MOVABLE, |
b2e18757f mm, vmscan: begin... |
3883 |
.reclaim_idx = MAX_NR_ZONES - 1, |
ee814fe23 mm: vmscan: clean... |
3884 |
.priority = DEF_PRIORITY, |
d6277db4a [PATCH] swsusp: r... |
3885 |
.may_writepage = 1, |
ee814fe23 mm: vmscan: clean... |
3886 3887 |
.may_unmap = 1, .may_swap = 1, |
7b51755c3 vmscan: kill hibe... |
3888 |
.hibernation_mode = 1, |
1da177e4c Linux-2.6.12-rc2 |
3889 |
}; |
a09ed5e00 vmscan: change sh... |
3890 |
struct zonelist *zonelist = node_zonelist(numa_node_id(), sc.gfp_mask); |
7b51755c3 vmscan: kill hibe... |
3891 |
unsigned long nr_reclaimed; |
499118e96 mm: introduce mem... |
3892 |
unsigned int noreclaim_flag; |
1da177e4c Linux-2.6.12-rc2 |
3893 |
|
d92a8cfcb locking/lockdep: ... |
3894 |
fs_reclaim_acquire(sc.gfp_mask); |
93781325d lockdep: fix fs_r... |
3895 |
noreclaim_flag = memalloc_noreclaim_save(); |
1732d2b01 mm/vmscan.c: add ... |
3896 |
set_task_reclaim_state(current, &sc.reclaim_state); |
d6277db4a [PATCH] swsusp: r... |
3897 |
|
3115cd914 mm: vmscan: remov... |
3898 |
nr_reclaimed = do_try_to_free_pages(zonelist, &sc); |
d979677c4 mm: shrink_all_me... |
3899 |
|
1732d2b01 mm/vmscan.c: add ... |
3900 |
set_task_reclaim_state(current, NULL); |
499118e96 mm: introduce mem... |
3901 |
memalloc_noreclaim_restore(noreclaim_flag); |
93781325d lockdep: fix fs_r... |
3902 |
fs_reclaim_release(sc.gfp_mask); |
d6277db4a [PATCH] swsusp: r... |
3903 |
|
7b51755c3 vmscan: kill hibe... |
3904 |
return nr_reclaimed; |
1da177e4c Linux-2.6.12-rc2 |
3905 |
} |
c6f37f121 PM/Suspend: Do no... |
3906 |
#endif /* CONFIG_HIBERNATION */ |
1da177e4c Linux-2.6.12-rc2 |
3907 |
|
1da177e4c Linux-2.6.12-rc2 |
3908 3909 3910 3911 |
/* It's optimal to keep kswapds on the same CPUs as their memory, but not required for correctness. So if the last cpu in a node goes away, we get changed to run anywhere: as the first one comes back, restore their cpu bindings. */ |
517bbed90 mm/vmscan: Conver... |
3912 |
static int kswapd_cpu_online(unsigned int cpu) |
1da177e4c Linux-2.6.12-rc2 |
3913 |
{ |
58c0a4a78 Fix panic of cpu ... |
3914 |
int nid; |
1da177e4c Linux-2.6.12-rc2 |
3915 |
|
517bbed90 mm/vmscan: Conver... |
3916 3917 3918 |
for_each_node_state(nid, N_MEMORY) { pg_data_t *pgdat = NODE_DATA(nid); const struct cpumask *mask; |
a70f73028 cpumask: replace ... |
3919 |
|
517bbed90 mm/vmscan: Conver... |
3920 |
mask = cpumask_of_node(pgdat->node_id); |
c5f59f083 nodemask: use new... |
3921 |
|
517bbed90 mm/vmscan: Conver... |
3922 3923 3924 |
if (cpumask_any_and(cpu_online_mask, mask) < nr_cpu_ids) /* One of our CPUs online: restore mask */ set_cpus_allowed_ptr(pgdat->kswapd, mask); |
1da177e4c Linux-2.6.12-rc2 |
3925 |
} |
517bbed90 mm/vmscan: Conver... |
3926 |
return 0; |
1da177e4c Linux-2.6.12-rc2 |
3927 |
} |
1da177e4c Linux-2.6.12-rc2 |
3928 |
|
3218ae14b [PATCH] pgdat all... |
3929 3930 3931 3932 3933 3934 3935 3936 3937 3938 3939 3940 3941 3942 3943 |
/* * This kswapd start function will be called by init and node-hot-add. * On node-hot-add, kswapd will moved to proper cpus if cpus are hot-added. */ int kswapd_run(int nid) { pg_data_t *pgdat = NODE_DATA(nid); int ret = 0; if (pgdat->kswapd) return 0; pgdat->kswapd = kthread_run(kswapd, pgdat, "kswapd%d", nid); if (IS_ERR(pgdat->kswapd)) { /* failure at boot is fatal */ |
c6202adf3 mm/vmscan: Adjust... |
3944 |
BUG_ON(system_state < SYSTEM_RUNNING); |
d5dc0ad92 mm/vmscan: fix er... |
3945 3946 3947 |
pr_err("Failed to start kswapd on node %d ", nid); ret = PTR_ERR(pgdat->kswapd); |
d72515b85 mm/vmscan: fix er... |
3948 |
pgdat->kswapd = NULL; |
3218ae14b [PATCH] pgdat all... |
3949 3950 3951 |
} return ret; } |
8fe23e057 mm: clear node in... |
3952 |
/* |
d8adde17e memory hotplug: f... |
3953 |
* Called by memory hotplug when all memory in a node is offlined. Caller must |
bfc8c9013 mem-hotplug: impl... |
3954 |
* hold mem_hotplug_begin/end(). |
8fe23e057 mm: clear node in... |
3955 3956 3957 3958 |
*/ void kswapd_stop(int nid) { struct task_struct *kswapd = NODE_DATA(nid)->kswapd; |
d8adde17e memory hotplug: f... |
3959 |
if (kswapd) { |
8fe23e057 mm: clear node in... |
3960 |
kthread_stop(kswapd); |
d8adde17e memory hotplug: f... |
3961 3962 |
NODE_DATA(nid)->kswapd = NULL; } |
8fe23e057 mm: clear node in... |
3963 |
} |
1da177e4c Linux-2.6.12-rc2 |
3964 3965 |
static int __init kswapd_init(void) { |
517bbed90 mm/vmscan: Conver... |
3966 |
int nid, ret; |
69e05944a [PATCH] vmscan: u... |
3967 |
|
1da177e4c Linux-2.6.12-rc2 |
3968 |
swap_setup(); |
48fb2e240 vmscan: use N_MEM... |
3969 |
for_each_node_state(nid, N_MEMORY) |
3218ae14b [PATCH] pgdat all... |
3970 |
kswapd_run(nid); |
517bbed90 mm/vmscan: Conver... |
3971 3972 3973 3974 |
ret = cpuhp_setup_state_nocalls(CPUHP_AP_ONLINE_DYN, "mm/vmscan:online", kswapd_cpu_online, NULL); WARN_ON(ret < 0); |
1da177e4c Linux-2.6.12-rc2 |
3975 3976 3977 3978 |
return 0; } module_init(kswapd_init) |
9eeff2395 [PATCH] Zone recl... |
3979 3980 3981 |
#ifdef CONFIG_NUMA /* |
a5f5f91da mm: convert zone_... |
3982 |
* Node reclaim mode |
9eeff2395 [PATCH] Zone recl... |
3983 |
* |
a5f5f91da mm: convert zone_... |
3984 |
* If non-zero call node_reclaim when the number of free pages falls below |
9eeff2395 [PATCH] Zone recl... |
3985 |
* the watermarks. |
9eeff2395 [PATCH] Zone recl... |
3986 |
*/ |
a5f5f91da mm: convert zone_... |
3987 |
int node_reclaim_mode __read_mostly; |
9eeff2395 [PATCH] Zone recl... |
3988 |
|
1b2ffb789 [PATCH] Zone recl... |
3989 |
#define RECLAIM_OFF 0 |
7d03431cf swapfile/vmscan: ... |
3990 |
#define RECLAIM_ZONE (1<<0) /* Run shrink_inactive_list on the zone */ |
1b2ffb789 [PATCH] Zone recl... |
3991 |
#define RECLAIM_WRITE (1<<1) /* Writeout pages during reclaim */ |
95bbc0c72 mm: rename RECLAI... |
3992 |
#define RECLAIM_UNMAP (1<<2) /* Unmap pages during reclaim */ |
1b2ffb789 [PATCH] Zone recl... |
3993 |
|
9eeff2395 [PATCH] Zone recl... |
3994 |
/* |
a5f5f91da mm: convert zone_... |
3995 |
* Priority for NODE_RECLAIM. This determines the fraction of pages |
a92f71263 [PATCH] zone_recl... |
3996 3997 3998 |
* of a node considered for each zone_reclaim. 4 scans 1/16th of * a zone. */ |
a5f5f91da mm: convert zone_... |
3999 |
#define NODE_RECLAIM_PRIORITY 4 |
a92f71263 [PATCH] zone_recl... |
4000 |
|
9eeff2395 [PATCH] Zone recl... |
4001 |
/* |
a5f5f91da mm: convert zone_... |
4002 |
* Percentage of pages in a zone that must be unmapped for node_reclaim to |
9614634fe [PATCH] ZVC/zone_... |
4003 4004 4005 4006 4007 |
* occur. */ int sysctl_min_unmapped_ratio = 1; /* |
0ff38490c [PATCH] zone_recl... |
4008 4009 4010 4011 |
* If the number of slab pages in a zone grows beyond this percentage then * slab reclaim needs to occur. */ int sysctl_min_slab_ratio = 5; |
11fb99898 mm: move most fil... |
4012 |
static inline unsigned long node_unmapped_file_pages(struct pglist_data *pgdat) |
90afa5de6 vmscan: properly ... |
4013 |
{ |
11fb99898 mm: move most fil... |
4014 4015 4016 |
unsigned long file_mapped = node_page_state(pgdat, NR_FILE_MAPPED); unsigned long file_lru = node_page_state(pgdat, NR_INACTIVE_FILE) + node_page_state(pgdat, NR_ACTIVE_FILE); |
90afa5de6 vmscan: properly ... |
4017 4018 4019 4020 4021 4022 4023 4024 4025 4026 |
/* * It's possible for there to be more file mapped pages than * accounted for by the pages on the file LRU lists because * tmpfs pages accounted for as ANON can also be FILE_MAPPED */ return (file_lru > file_mapped) ? (file_lru - file_mapped) : 0; } /* Work out how many page cache pages we can reclaim in this reclaim_mode */ |
a5f5f91da mm: convert zone_... |
4027 |
static unsigned long node_pagecache_reclaimable(struct pglist_data *pgdat) |
90afa5de6 vmscan: properly ... |
4028 |
{ |
d031a1579 mm/vmscan.c: fix ... |
4029 4030 |
unsigned long nr_pagecache_reclaimable; unsigned long delta = 0; |
90afa5de6 vmscan: properly ... |
4031 4032 |
/* |
95bbc0c72 mm: rename RECLAI... |
4033 |
* If RECLAIM_UNMAP is set, then all file pages are considered |
90afa5de6 vmscan: properly ... |
4034 |
* potentially reclaimable. Otherwise, we have to worry about |
11fb99898 mm: move most fil... |
4035 |
* pages like swapcache and node_unmapped_file_pages() provides |
90afa5de6 vmscan: properly ... |
4036 4037 |
* a better estimate */ |
a5f5f91da mm: convert zone_... |
4038 4039 |
if (node_reclaim_mode & RECLAIM_UNMAP) nr_pagecache_reclaimable = node_page_state(pgdat, NR_FILE_PAGES); |
90afa5de6 vmscan: properly ... |
4040 |
else |
a5f5f91da mm: convert zone_... |
4041 |
nr_pagecache_reclaimable = node_unmapped_file_pages(pgdat); |
90afa5de6 vmscan: properly ... |
4042 4043 |
/* If we can't clean pages, remove dirty pages from consideration */ |
a5f5f91da mm: convert zone_... |
4044 4045 |
if (!(node_reclaim_mode & RECLAIM_WRITE)) delta += node_page_state(pgdat, NR_FILE_DIRTY); |
90afa5de6 vmscan: properly ... |
4046 4047 4048 4049 4050 4051 4052 |
/* Watch for any possible underflows due to delta */ if (unlikely(delta > nr_pagecache_reclaimable)) delta = nr_pagecache_reclaimable; return nr_pagecache_reclaimable - delta; } |
0ff38490c [PATCH] zone_recl... |
4053 |
/* |
a5f5f91da mm: convert zone_... |
4054 |
* Try to free up some pages from this node through reclaim. |
9eeff2395 [PATCH] Zone recl... |
4055 |
*/ |
a5f5f91da mm: convert zone_... |
4056 |
static int __node_reclaim(struct pglist_data *pgdat, gfp_t gfp_mask, unsigned int order) |
9eeff2395 [PATCH] Zone recl... |
4057 |
{ |
7fb2d46d3 [PATCH] zone_recl... |
4058 |
/* Minimum pages needed in order to stay on node */ |
69e05944a [PATCH] vmscan: u... |
4059 |
const unsigned long nr_pages = 1 << order; |
9eeff2395 [PATCH] Zone recl... |
4060 |
struct task_struct *p = current; |
499118e96 mm: introduce mem... |
4061 |
unsigned int noreclaim_flag; |
179e96395 [PATCH] vmscan: s... |
4062 |
struct scan_control sc = { |
62b726c1b mm/vmscan.c:__zon... |
4063 |
.nr_to_reclaim = max(nr_pages, SWAP_CLUSTER_MAX), |
f2f43e566 mm/vmscan.c: fix ... |
4064 |
.gfp_mask = current_gfp_context(gfp_mask), |
bd2f6199c vmscan: respect h... |
4065 |
.order = order, |
a5f5f91da mm: convert zone_... |
4066 4067 4068 |
.priority = NODE_RECLAIM_PRIORITY, .may_writepage = !!(node_reclaim_mode & RECLAIM_WRITE), .may_unmap = !!(node_reclaim_mode & RECLAIM_UNMAP), |
ee814fe23 mm: vmscan: clean... |
4069 |
.may_swap = 1, |
f2f43e566 mm/vmscan.c: fix ... |
4070 |
.reclaim_idx = gfp_zone(gfp_mask), |
179e96395 [PATCH] vmscan: s... |
4071 |
}; |
9eeff2395 [PATCH] Zone recl... |
4072 |
|
132bb8cfc mm/vmscan: add tr... |
4073 4074 |
trace_mm_vmscan_node_reclaim_begin(pgdat->node_id, order, sc.gfp_mask); |
9eeff2395 [PATCH] Zone recl... |
4075 |
cond_resched(); |
93781325d lockdep: fix fs_r... |
4076 |
fs_reclaim_acquire(sc.gfp_mask); |
d4f7796e9 [PATCH] vmscan: f... |
4077 |
/* |
95bbc0c72 mm: rename RECLAI... |
4078 |
* We need to be able to allocate from the reserves for RECLAIM_UNMAP |
d4f7796e9 [PATCH] vmscan: f... |
4079 |
* and we also need to be able to write out pages for RECLAIM_WRITE |
95bbc0c72 mm: rename RECLAI... |
4080 |
* and RECLAIM_UNMAP. |
d4f7796e9 [PATCH] vmscan: f... |
4081 |
*/ |
499118e96 mm: introduce mem... |
4082 4083 |
noreclaim_flag = memalloc_noreclaim_save(); p->flags |= PF_SWAPWRITE; |
1732d2b01 mm/vmscan.c: add ... |
4084 |
set_task_reclaim_state(p, &sc.reclaim_state); |
c84db23c6 [PATCH] zone_recl... |
4085 |
|
a5f5f91da mm: convert zone_... |
4086 |
if (node_pagecache_reclaimable(pgdat) > pgdat->min_unmapped_pages) { |
0ff38490c [PATCH] zone_recl... |
4087 |
/* |
894befec4 mm/vmscan: update... |
4088 |
* Free memory by calling shrink node with increasing |
0ff38490c [PATCH] zone_recl... |
4089 4090 |
* priorities until we have enough memory freed. */ |
0ff38490c [PATCH] zone_recl... |
4091 |
do { |
970a39a36 mm, vmscan: avoid... |
4092 |
shrink_node(pgdat, &sc); |
9e3b2f8cd mm/vmscan: store ... |
4093 |
} while (sc.nr_reclaimed < nr_pages && --sc.priority >= 0); |
0ff38490c [PATCH] zone_recl... |
4094 |
} |
c84db23c6 [PATCH] zone_recl... |
4095 |
|
1732d2b01 mm/vmscan.c: add ... |
4096 |
set_task_reclaim_state(p, NULL); |
499118e96 mm: introduce mem... |
4097 4098 |
current->flags &= ~PF_SWAPWRITE; memalloc_noreclaim_restore(noreclaim_flag); |
93781325d lockdep: fix fs_r... |
4099 |
fs_reclaim_release(sc.gfp_mask); |
132bb8cfc mm/vmscan: add tr... |
4100 4101 |
trace_mm_vmscan_node_reclaim_end(sc.nr_reclaimed); |
a79311c14 vmscan: bail out ... |
4102 |
return sc.nr_reclaimed >= nr_pages; |
9eeff2395 [PATCH] Zone recl... |
4103 |
} |
179e96395 [PATCH] vmscan: s... |
4104 |
|
a5f5f91da mm: convert zone_... |
4105 |
int node_reclaim(struct pglist_data *pgdat, gfp_t gfp_mask, unsigned int order) |
179e96395 [PATCH] vmscan: s... |
4106 |
{ |
d773ed6b8 mm: test and set ... |
4107 |
int ret; |
179e96395 [PATCH] vmscan: s... |
4108 4109 |
/* |
a5f5f91da mm: convert zone_... |
4110 |
* Node reclaim reclaims unmapped file backed pages and |
0ff38490c [PATCH] zone_recl... |
4111 |
* slab pages if we are over the defined limits. |
34aa1330f [PATCH] zoned vm ... |
4112 |
* |
9614634fe [PATCH] ZVC/zone_... |
4113 4114 |
* A small portion of unmapped file backed pages is needed for * file I/O otherwise pages read by file I/O will be immediately |
a5f5f91da mm: convert zone_... |
4115 4116 |
* thrown out if the node is overallocated. So we do not reclaim * if less than a specified percentage of the node is used by |
9614634fe [PATCH] ZVC/zone_... |
4117 |
* unmapped file backed pages. |
179e96395 [PATCH] vmscan: s... |
4118 |
*/ |
a5f5f91da mm: convert zone_... |
4119 |
if (node_pagecache_reclaimable(pgdat) <= pgdat->min_unmapped_pages && |
385386cff mm: vmstat: move ... |
4120 |
node_page_state(pgdat, NR_SLAB_RECLAIMABLE) <= pgdat->min_slab_pages) |
a5f5f91da mm: convert zone_... |
4121 |
return NODE_RECLAIM_FULL; |
179e96395 [PATCH] vmscan: s... |
4122 4123 |
/* |
d773ed6b8 mm: test and set ... |
4124 |
* Do not scan if the allocation should not be delayed. |
179e96395 [PATCH] vmscan: s... |
4125 |
*/ |
d0164adc8 mm, page_alloc: d... |
4126 |
if (!gfpflags_allow_blocking(gfp_mask) || (current->flags & PF_MEMALLOC)) |
a5f5f91da mm: convert zone_... |
4127 |
return NODE_RECLAIM_NOSCAN; |
179e96395 [PATCH] vmscan: s... |
4128 4129 |
/* |
a5f5f91da mm: convert zone_... |
4130 |
* Only run node reclaim on the local node or on nodes that do not |
179e96395 [PATCH] vmscan: s... |
4131 4132 4133 4134 |
* have associated processors. This will favor the local processor * over remote processors and spread off node memory allocations * as wide as possible. */ |
a5f5f91da mm: convert zone_... |
4135 4136 |
if (node_state(pgdat->node_id, N_CPU) && pgdat->node_id != numa_node_id()) return NODE_RECLAIM_NOSCAN; |
d773ed6b8 mm: test and set ... |
4137 |
|
a5f5f91da mm: convert zone_... |
4138 4139 |
if (test_and_set_bit(PGDAT_RECLAIM_LOCKED, &pgdat->flags)) return NODE_RECLAIM_NOSCAN; |
fa5e084e4 vmscan: do not un... |
4140 |
|
a5f5f91da mm: convert zone_... |
4141 4142 |
ret = __node_reclaim(pgdat, gfp_mask, order); clear_bit(PGDAT_RECLAIM_LOCKED, &pgdat->flags); |
d773ed6b8 mm: test and set ... |
4143 |
|
24cf72518 vmscan: count the... |
4144 4145 |
if (!ret) count_vm_event(PGSCAN_ZONE_RECLAIM_FAILED); |
d773ed6b8 mm: test and set ... |
4146 |
return ret; |
179e96395 [PATCH] vmscan: s... |
4147 |
} |
9eeff2395 [PATCH] Zone recl... |
4148 |
#endif |
894bc3104 Unevictable LRU I... |
4149 |
|
894bc3104 Unevictable LRU I... |
4150 4151 4152 |
/* * page_evictable - test whether a page is evictable * @page: the page to test |
894bc3104 Unevictable LRU I... |
4153 4154 |
* * Test whether page is evictable--i.e., should be placed on active/inactive |
39b5f29ac mm: remove vma ar... |
4155 |
* lists vs unevictable list. |
894bc3104 Unevictable LRU I... |
4156 4157 |
* * Reasons page might not be evictable: |
ba9ddf493 Ramfs and Ram Dis... |
4158 |
* (1) page's mapping marked unevictable |
b291f0003 mlock: mlocked pa... |
4159 |
* (2) page is part of an mlocked VMA |
ba9ddf493 Ramfs and Ram Dis... |
4160 |
* |
894bc3104 Unevictable LRU I... |
4161 |
*/ |
39b5f29ac mm: remove vma ar... |
4162 |
int page_evictable(struct page *page) |
894bc3104 Unevictable LRU I... |
4163 |
{ |
e92bb4dd9 mm: fix races bet... |
4164 4165 4166 4167 4168 4169 4170 |
int ret; /* Prevent address_space of inode and swap cache from being freed */ rcu_read_lock(); ret = !mapping_unevictable(page_mapping(page)) && !PageMlocked(page); rcu_read_unlock(); return ret; |
894bc3104 Unevictable LRU I... |
4171 |
} |
89e004ea5 SHM_LOCKED pages ... |
4172 4173 |
/** |
64e3d12f7 mm, drm/i915: mar... |
4174 4175 4176 |
* check_move_unevictable_pages - check pages for evictability and move to * appropriate zone lru list * @pvec: pagevec with lru pages to check |
89e004ea5 SHM_LOCKED pages ... |
4177 |
* |
64e3d12f7 mm, drm/i915: mar... |
4178 4179 4180 |
* Checks pages for evictability, if an evictable page is in the unevictable * lru list, moves it to the appropriate evictable lru list. This function * should be only used for lru pages. |
89e004ea5 SHM_LOCKED pages ... |
4181 |
*/ |
64e3d12f7 mm, drm/i915: mar... |
4182 |
void check_move_unevictable_pages(struct pagevec *pvec) |
89e004ea5 SHM_LOCKED pages ... |
4183 |
{ |
925b7673c mm: make per-memc... |
4184 |
struct lruvec *lruvec; |
785b99feb mm, vmscan: relea... |
4185 |
struct pglist_data *pgdat = NULL; |
245132643 SHM_UNLOCK: fix U... |
4186 4187 4188 |
int pgscanned = 0; int pgrescued = 0; int i; |
89e004ea5 SHM_LOCKED pages ... |
4189 |
|
64e3d12f7 mm, drm/i915: mar... |
4190 4191 |
for (i = 0; i < pvec->nr; i++) { struct page *page = pvec->pages[i]; |
785b99feb mm, vmscan: relea... |
4192 |
struct pglist_data *pagepgdat = page_pgdat(page); |
89e004ea5 SHM_LOCKED pages ... |
4193 |
|
245132643 SHM_UNLOCK: fix U... |
4194 |
pgscanned++; |
785b99feb mm, vmscan: relea... |
4195 4196 4197 4198 4199 |
if (pagepgdat != pgdat) { if (pgdat) spin_unlock_irq(&pgdat->lru_lock); pgdat = pagepgdat; spin_lock_irq(&pgdat->lru_lock); |
245132643 SHM_UNLOCK: fix U... |
4200 |
} |
785b99feb mm, vmscan: relea... |
4201 |
lruvec = mem_cgroup_page_lruvec(page, pgdat); |
89e004ea5 SHM_LOCKED pages ... |
4202 |
|
245132643 SHM_UNLOCK: fix U... |
4203 4204 |
if (!PageLRU(page) || !PageUnevictable(page)) continue; |
89e004ea5 SHM_LOCKED pages ... |
4205 |
|
39b5f29ac mm: remove vma ar... |
4206 |
if (page_evictable(page)) { |
245132643 SHM_UNLOCK: fix U... |
4207 |
enum lru_list lru = page_lru_base_type(page); |
309381fea mm: dump page whe... |
4208 |
VM_BUG_ON_PAGE(PageActive(page), page); |
245132643 SHM_UNLOCK: fix U... |
4209 |
ClearPageUnevictable(page); |
fa9add641 mm/memcg: apply a... |
4210 4211 |
del_page_from_lru_list(page, lruvec, LRU_UNEVICTABLE); add_page_to_lru_list(page, lruvec, lru); |
245132643 SHM_UNLOCK: fix U... |
4212 |
pgrescued++; |
89e004ea5 SHM_LOCKED pages ... |
4213 |
} |
245132643 SHM_UNLOCK: fix U... |
4214 |
} |
89e004ea5 SHM_LOCKED pages ... |
4215 |
|
785b99feb mm, vmscan: relea... |
4216 |
if (pgdat) { |
245132643 SHM_UNLOCK: fix U... |
4217 4218 |
__count_vm_events(UNEVICTABLE_PGRESCUED, pgrescued); __count_vm_events(UNEVICTABLE_PGSCANNED, pgscanned); |
785b99feb mm, vmscan: relea... |
4219 |
spin_unlock_irq(&pgdat->lru_lock); |
89e004ea5 SHM_LOCKED pages ... |
4220 |
} |
89e004ea5 SHM_LOCKED pages ... |
4221 |
} |
64e3d12f7 mm, drm/i915: mar... |
4222 |
EXPORT_SYMBOL_GPL(check_move_unevictable_pages); |