Blame view
mm/vmscan.c
123 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 |
|
7cf111bc3 mm: vmscan: deter... |
77 78 79 80 81 |
/* * Scan pressure balancing between anon and file LRUs */ unsigned long anon_cost; unsigned long file_cost; |
b91ac3743 mm: vmscan: enfor... |
82 83 84 85 86 87 |
/* Can active pages be deactivated as part of reclaim? */ #define DEACTIVATE_ANON 1 #define DEACTIVATE_FILE 2 unsigned int may_deactivate:2; unsigned int force_deactivate:1; unsigned int skipped_deactivate:1; |
1276ad68e mm: vmscan: scan ... |
88 |
/* Writepage batching in laptop mode; RECLAIM_WRITE */ |
ee814fe23 mm: vmscan: clean... |
89 90 91 92 93 94 95 |
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... |
96 97 98 99 100 101 102 |
/* * 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... |
103 |
|
ee814fe23 mm: vmscan: clean... |
104 105 106 107 |
unsigned int hibernation_mode:1; /* One of the zones is ready for compaction */ unsigned int compaction_ready:1; |
b91ac3743 mm: vmscan: enfor... |
108 109 |
/* There is easily reclaimable cold cache in the current node */ unsigned int cache_trim_mode:1; |
53138cea7 mm: vmscan: move ... |
110 111 |
/* The file pages on the current node are dangerously low */ unsigned int file_is_tiny:1; |
bb451fdf3 mm/vmscan.c: cond... |
112 113 114 115 116 117 118 119 120 121 122 |
/* 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... |
123 124 125 126 127 |
/* 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 ... |
128 129 130 131 132 133 134 135 136 137 |
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 ... |
138 139 140 |
/* for recording the reclaimed slab by now */ struct reclaim_state reclaim_state; |
1da177e4c Linux-2.6.12-rc2 |
141 |
}; |
1da177e4c Linux-2.6.12-rc2 |
142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 |
#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 /* |
c843966c5 mm: allow swappin... |
157 |
* From 0 .. 200. Higher means more swappy. |
1da177e4c Linux-2.6.12-rc2 |
158 159 |
*/ int vm_swappiness = 60; |
1da177e4c Linux-2.6.12-rc2 |
160 |
|
0a432dcbe mm: shrinker: mak... |
161 162 163 164 165 166 167 168 169 170 171 |
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 |
172 173 |
static LIST_HEAD(shrinker_list); static DECLARE_RWSEM(shrinker_rwsem); |
0a432dcbe mm: shrinker: mak... |
174 |
#ifdef CONFIG_MEMCG |
7e010df53 mm: use special v... |
175 176 177 178 179 180 181 182 183 184 185 186 |
/* * 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 ... |
187 188 189 190 191 192 193 194 195 |
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... |
196 |
id = idr_alloc(&shrinker_idr, SHRINKER_REGISTERING, 0, 0, GFP_KERNEL); |
b4c2b231c mm: assign id to ... |
197 198 |
if (id < 0) goto unlock; |
0a4465d34 mm, memcg: assign... |
199 200 201 202 203 |
if (id >= shrinker_nr_max) { if (memcg_expand_shrinker_maps(id)) { idr_remove(&shrinker_idr, id); goto unlock; } |
b4c2b231c mm: assign id to ... |
204 |
shrinker_nr_max = id + 1; |
0a4465d34 mm, memcg: assign... |
205 |
} |
b4c2b231c mm: assign id to ... |
206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 |
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 ... |
223 |
|
b5ead35e7 mm: vmscan: namin... |
224 |
static bool cgroup_reclaim(struct scan_control *sc) |
89b5fae53 mm: vmscan: disti... |
225 |
{ |
b5ead35e7 mm: vmscan: namin... |
226 |
return sc->target_mem_cgroup; |
89b5fae53 mm: vmscan: disti... |
227 |
} |
97c9341f7 mm: vmscan: disab... |
228 229 |
/** |
b5ead35e7 mm: vmscan: namin... |
230 |
* writeback_throttling_sane - is the usual dirty throttling mechanism available? |
97c9341f7 mm: vmscan: disab... |
231 232 233 234 235 236 237 238 239 240 241 |
* @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. */ |
b5ead35e7 mm: vmscan: namin... |
242 |
static bool writeback_throttling_sane(struct scan_control *sc) |
97c9341f7 mm: vmscan: disab... |
243 |
{ |
b5ead35e7 mm: vmscan: namin... |
244 |
if (!cgroup_reclaim(sc)) |
97c9341f7 mm: vmscan: disab... |
245 246 |
return true; #ifdef CONFIG_CGROUP_WRITEBACK |
69234acee Merge branch 'for... |
247 |
if (cgroup_subsys_on_dfl(memory_cgrp_subsys)) |
97c9341f7 mm: vmscan: disab... |
248 249 250 251 |
return true; #endif return false; } |
91a45470f per-zone and recl... |
252 |
#else |
0a432dcbe mm: shrinker: mak... |
253 254 255 256 257 258 259 260 |
static int prealloc_memcg_shrinker(struct shrinker *shrinker) { return 0; } static void unregister_memcg_shrinker(struct shrinker *shrinker) { } |
b5ead35e7 mm: vmscan: namin... |
261 |
static bool cgroup_reclaim(struct scan_control *sc) |
89b5fae53 mm: vmscan: disti... |
262 |
{ |
b5ead35e7 mm: vmscan: namin... |
263 |
return false; |
89b5fae53 mm: vmscan: disti... |
264 |
} |
97c9341f7 mm: vmscan: disab... |
265 |
|
b5ead35e7 mm: vmscan: namin... |
266 |
static bool writeback_throttling_sane(struct scan_control *sc) |
97c9341f7 mm: vmscan: disab... |
267 268 269 |
{ return true; } |
91a45470f per-zone and recl... |
270 |
#endif |
5a1c84b40 mm: remove reclai... |
271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 |
/* * 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... |
288 289 290 291 292 293 294 |
/** * 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... |
295 |
{ |
de3b01506 mm: vmscan: simpl... |
296 |
unsigned long size = 0; |
fd5388037 mm, vmscan: clean... |
297 |
int zid; |
de3b01506 mm: vmscan: simpl... |
298 |
for (zid = 0; zid <= zone_idx && zid < MAX_NR_ZONES; zid++) { |
fd5388037 mm, vmscan: clean... |
299 |
struct zone *zone = &lruvec_pgdat(lruvec)->node_zones[zid]; |
c9f299d98 mm: add zone nr_p... |
300 |
|
fd5388037 mm, vmscan: clean... |
301 302 303 304 |
if (!managed_zone(zone)) continue; if (!mem_cgroup_disabled()) |
de3b01506 mm: vmscan: simpl... |
305 |
size += mem_cgroup_get_zone_lru_size(lruvec, lru, zid); |
fd5388037 mm, vmscan: clean... |
306 |
else |
de3b01506 mm: vmscan: simpl... |
307 |
size += zone_page_state(zone, NR_ZONE_LRU_BASE + lru); |
fd5388037 mm, vmscan: clean... |
308 |
} |
de3b01506 mm: vmscan: simpl... |
309 |
return size; |
b4536f0c8 mm, memcg: fix th... |
310 |
} |
1da177e4c Linux-2.6.12-rc2 |
311 |
/* |
1d3d4437e vmscan: per-node ... |
312 |
* Add a shrinker callback to be called from the vm. |
1da177e4c Linux-2.6.12-rc2 |
313 |
*/ |
8e04944f0 mm,vmscan: Allow ... |
314 |
int prealloc_shrinker(struct shrinker *shrinker) |
1da177e4c Linux-2.6.12-rc2 |
315 |
{ |
b9726c26d numa: make "nr_no... |
316 |
unsigned int size = sizeof(*shrinker->nr_deferred); |
1d3d4437e vmscan: per-node ... |
317 |
|
1d3d4437e vmscan: per-node ... |
318 319 320 321 322 323 |
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 ... |
324 325 326 327 328 |
if (shrinker->flags & SHRINKER_MEMCG_AWARE) { if (prealloc_memcg_shrinker(shrinker)) goto free_deferred; } |
8e04944f0 mm,vmscan: Allow ... |
329 |
return 0; |
b4c2b231c mm: assign id to ... |
330 331 332 333 334 |
free_deferred: kfree(shrinker->nr_deferred); shrinker->nr_deferred = NULL; return -ENOMEM; |
8e04944f0 mm,vmscan: Allow ... |
335 336 337 338 |
} void free_prealloced_shrinker(struct shrinker *shrinker) { |
b4c2b231c mm: assign id to ... |
339 340 341 342 343 |
if (!shrinker->nr_deferred) return; if (shrinker->flags & SHRINKER_MEMCG_AWARE) unregister_memcg_shrinker(shrinker); |
8e04944f0 mm,vmscan: Allow ... |
344 345 346 |
kfree(shrinker->nr_deferred); shrinker->nr_deferred = NULL; } |
1d3d4437e vmscan: per-node ... |
347 |
|
8e04944f0 mm,vmscan: Allow ... |
348 349 |
void register_shrinker_prepared(struct shrinker *shrinker) { |
8e1f936b7 mm: clean up and ... |
350 351 |
down_write(&shrinker_rwsem); list_add_tail(&shrinker->list, &shrinker_list); |
42a9a53bb mm: vmscan: prote... |
352 |
#ifdef CONFIG_MEMCG |
8df4a44cc mm: check shrinke... |
353 354 |
if (shrinker->flags & SHRINKER_MEMCG_AWARE) idr_replace(&shrinker_idr, shrinker, shrinker->id); |
7e010df53 mm: use special v... |
355 |
#endif |
8e1f936b7 mm: clean up and ... |
356 |
up_write(&shrinker_rwsem); |
8e04944f0 mm,vmscan: Allow ... |
357 358 359 360 361 362 363 364 365 |
} int register_shrinker(struct shrinker *shrinker) { int err = prealloc_shrinker(shrinker); if (err) return err; register_shrinker_prepared(shrinker); |
1d3d4437e vmscan: per-node ... |
366 |
return 0; |
1da177e4c Linux-2.6.12-rc2 |
367 |
} |
8e1f936b7 mm: clean up and ... |
368 |
EXPORT_SYMBOL(register_shrinker); |
1da177e4c Linux-2.6.12-rc2 |
369 370 371 372 |
/* * Remove one */ |
8e1f936b7 mm: clean up and ... |
373 |
void unregister_shrinker(struct shrinker *shrinker) |
1da177e4c Linux-2.6.12-rc2 |
374 |
{ |
bb422a738 mm,vmscan: Make u... |
375 376 |
if (!shrinker->nr_deferred) return; |
b4c2b231c mm: assign id to ... |
377 378 |
if (shrinker->flags & SHRINKER_MEMCG_AWARE) unregister_memcg_shrinker(shrinker); |
1da177e4c Linux-2.6.12-rc2 |
379 380 381 |
down_write(&shrinker_rwsem); list_del(&shrinker->list); up_write(&shrinker_rwsem); |
ae3933216 mm/vmscan.c: don'... |
382 |
kfree(shrinker->nr_deferred); |
bb422a738 mm,vmscan: Make u... |
383 |
shrinker->nr_deferred = NULL; |
1da177e4c Linux-2.6.12-rc2 |
384 |
} |
8e1f936b7 mm: clean up and ... |
385 |
EXPORT_SYMBOL(unregister_shrinker); |
1da177e4c Linux-2.6.12-rc2 |
386 387 |
#define SHRINK_BATCH 128 |
1d3d4437e vmscan: per-node ... |
388 |
|
cb731d6c6 vmscan: per memor... |
389 |
static unsigned long do_shrink_slab(struct shrink_control *shrinkctl, |
9092c71bb mm: use sc->prior... |
390 |
struct shrinker *shrinker, int priority) |
1d3d4437e vmscan: per-node ... |
391 392 393 394 |
{ unsigned long freed = 0; unsigned long long delta; long total_scan; |
d5bc5fd3f mm: vmscan: shrin... |
395 |
long freeable; |
1d3d4437e vmscan: per-node ... |
396 397 398 399 400 |
long nr; long new_nr; int nid = shrinkctl->nid; long batch_size = shrinker->batch ? shrinker->batch : SHRINK_BATCH; |
5f33a0803 mm/vmscan.c: set ... |
401 |
long scanned = 0, next_deferred; |
1d3d4437e vmscan: per-node ... |
402 |
|
ac7fb3ad2 mm/vmscan.c: move... |
403 404 |
if (!(shrinker->flags & SHRINKER_NUMA_AWARE)) nid = 0; |
d5bc5fd3f mm: vmscan: shrin... |
405 |
freeable = shrinker->count_objects(shrinker, shrinkctl); |
9b996468c mm: add SHRINK_EM... |
406 407 |
if (freeable == 0 || freeable == SHRINK_EMPTY) return freeable; |
1d3d4437e vmscan: per-node ... |
408 409 410 411 412 413 414 415 416 |
/* * 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... |
417 418 419 420 421 422 423 424 425 426 427 428 |
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... |
429 |
|
1d3d4437e vmscan: per-node ... |
430 431 |
total_scan += delta; if (total_scan < 0) { |
d75f773c8 treewide: Switch ... |
432 433 |
pr_err("shrink_slab: %pS negative objects to delete nr=%ld ", |
a0b02131c shrinker: Kill ol... |
434 |
shrinker->scan_objects, total_scan); |
d5bc5fd3f mm: vmscan: shrin... |
435 |
total_scan = freeable; |
5f33a0803 mm/vmscan.c: set ... |
436 437 438 |
next_deferred = nr; } else next_deferred = total_scan; |
1d3d4437e vmscan: per-node ... |
439 440 441 442 443 444 445 |
/* * 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... |
446 |
* freeable. This is bad for sustaining a working set in |
1d3d4437e vmscan: per-node ... |
447 448 449 450 451 |
* memory. * * Hence only allow the shrinker to scan the entire cache when * a large delta change is calculated directly. */ |
d5bc5fd3f mm: vmscan: shrin... |
452 453 |
if (delta < freeable / 4) total_scan = min(total_scan, freeable / 2); |
1d3d4437e vmscan: per-node ... |
454 455 456 457 458 459 |
/* * 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... |
460 461 |
if (total_scan > freeable * 2) total_scan = freeable * 2; |
1d3d4437e vmscan: per-node ... |
462 463 |
trace_mm_shrink_slab_start(shrinker, shrinkctl, nr, |
9092c71bb mm: use sc->prior... |
464 |
freeable, delta, total_scan, priority); |
1d3d4437e vmscan: per-node ... |
465 |
|
0b1fb40a3 mm: vmscan: shrin... |
466 467 468 469 470 471 472 473 474 475 476 |
/* * 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... |
477 |
* than the total number of objects on slab (freeable), we must be |
0b1fb40a3 mm: vmscan: shrin... |
478 479 480 481 |
* scanning at high prio and therefore should try to reclaim as much as * possible. */ while (total_scan >= batch_size || |
d5bc5fd3f mm: vmscan: shrin... |
482 |
total_scan >= freeable) { |
a0b02131c shrinker: Kill ol... |
483 |
unsigned long ret; |
0b1fb40a3 mm: vmscan: shrin... |
484 |
unsigned long nr_to_scan = min(batch_size, total_scan); |
1d3d4437e vmscan: per-node ... |
485 |
|
0b1fb40a3 mm: vmscan: shrin... |
486 |
shrinkctl->nr_to_scan = nr_to_scan; |
d460acb5b mm: track actual ... |
487 |
shrinkctl->nr_scanned = nr_to_scan; |
a0b02131c shrinker: Kill ol... |
488 489 490 491 |
ret = shrinker->scan_objects(shrinker, shrinkctl); if (ret == SHRINK_STOP) break; freed += ret; |
1d3d4437e vmscan: per-node ... |
492 |
|
d460acb5b mm: track actual ... |
493 494 495 |
count_vm_events(SLABS_SCANNED, shrinkctl->nr_scanned); total_scan -= shrinkctl->nr_scanned; scanned += shrinkctl->nr_scanned; |
1d3d4437e vmscan: per-node ... |
496 497 498 |
cond_resched(); } |
5f33a0803 mm/vmscan.c: set ... |
499 500 501 502 |
if (next_deferred >= scanned) next_deferred -= scanned; else next_deferred = 0; |
1d3d4437e vmscan: per-node ... |
503 504 505 506 507 |
/* * 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 ... |
508 509 |
if (next_deferred > 0) new_nr = atomic_long_add_return(next_deferred, |
1d3d4437e vmscan: per-node ... |
510 511 512 |
&shrinker->nr_deferred[nid]); else new_nr = atomic_long_read(&shrinker->nr_deferred[nid]); |
df9024a8c mm: shrinker: add... |
513 |
trace_mm_shrink_slab_end(shrinker, nid, freed, nr, new_nr, total_scan); |
1d3d4437e vmscan: per-node ... |
514 |
return freed; |
1495f230f vmscan: change sh... |
515 |
} |
0a432dcbe mm: shrinker: mak... |
516 |
#ifdef CONFIG_MEMCG |
b0dedc49a mm/vmscan.c: iter... |
517 518 519 520 |
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 ... |
521 522 |
unsigned long ret, freed = 0; int i; |
b0dedc49a mm/vmscan.c: iter... |
523 |
|
0a432dcbe mm: shrinker: mak... |
524 |
if (!mem_cgroup_online(memcg)) |
b0dedc49a mm/vmscan.c: iter... |
525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 |
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... |
544 545 546 |
if (unlikely(!shrinker || shrinker == SHRINKER_REGISTERING)) { if (!shrinker) clear_bit(i, map->map); |
b0dedc49a mm/vmscan.c: iter... |
547 548 |
continue; } |
0a432dcbe mm: shrinker: mak... |
549 550 551 552 |
/* Call non-slab shrinkers even though kmem is disabled */ if (!memcg_kmem_enabled() && !(shrinker->flags & SHRINKER_NONSLAB)) continue; |
b0dedc49a mm/vmscan.c: iter... |
553 |
ret = do_shrink_slab(&sc, shrinker, priority); |
f90280d6b mm/vmscan.c: clea... |
554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 |
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... |
578 579 580 581 582 583 584 585 586 587 588 |
freed += ret; if (rwsem_is_contended(&shrinker_rwsem)) { freed = freed ? : 1; break; } } unlock: up_read(&shrinker_rwsem); return freed; } |
0a432dcbe mm: shrinker: mak... |
589 |
#else /* CONFIG_MEMCG */ |
b0dedc49a mm/vmscan.c: iter... |
590 591 592 593 594 |
static unsigned long shrink_slab_memcg(gfp_t gfp_mask, int nid, struct mem_cgroup *memcg, int priority) { return 0; } |
0a432dcbe mm: shrinker: mak... |
595 |
#endif /* CONFIG_MEMCG */ |
b0dedc49a mm/vmscan.c: iter... |
596 |
|
6b4f7799c mm: vmscan: invok... |
597 |
/** |
cb731d6c6 vmscan: per memor... |
598 |
* shrink_slab - shrink slab caches |
6b4f7799c mm: vmscan: invok... |
599 600 |
* @gfp_mask: allocation context * @nid: node whose slab caches to target |
cb731d6c6 vmscan: per memor... |
601 |
* @memcg: memory cgroup whose slab caches to target |
9092c71bb mm: use sc->prior... |
602 |
* @priority: the reclaim priority |
1da177e4c Linux-2.6.12-rc2 |
603 |
* |
6b4f7799c mm: vmscan: invok... |
604 |
* Call the shrink functions to age shrinkable caches. |
1da177e4c Linux-2.6.12-rc2 |
605 |
* |
6b4f7799c mm: vmscan: invok... |
606 607 |
* @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 |
608 |
* |
aeed1d325 mm/vmscan.c: gene... |
609 610 |
* @memcg specifies the memory cgroup to target. Unaware shrinkers * are called only if it is the root cgroup. |
cb731d6c6 vmscan: per memor... |
611 |
* |
9092c71bb mm: use sc->prior... |
612 613 |
* @priority is sc->priority, we take the number of objects and >> by priority * in order to get the scan target. |
b15e0905f [PATCH] vmscan: n... |
614 |
* |
6b4f7799c mm: vmscan: invok... |
615 |
* Returns the number of reclaimed slab objects. |
1da177e4c Linux-2.6.12-rc2 |
616 |
*/ |
cb731d6c6 vmscan: per memor... |
617 618 |
static unsigned long shrink_slab(gfp_t gfp_mask, int nid, struct mem_cgroup *memcg, |
9092c71bb mm: use sc->prior... |
619 |
int priority) |
1da177e4c Linux-2.6.12-rc2 |
620 |
{ |
b8e57efa2 mm/vmscan.c: fix ... |
621 |
unsigned long ret, freed = 0; |
1da177e4c Linux-2.6.12-rc2 |
622 |
struct shrinker *shrinker; |
fa1e512fa mm: vmscan: check... |
623 624 625 626 627 628 629 630 |
/* * 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... |
631 |
return shrink_slab_memcg(gfp_mask, nid, memcg, priority); |
cb731d6c6 vmscan: per memor... |
632 |
|
e830c63a6 mm,vmscan: don't ... |
633 |
if (!down_read_trylock(&shrinker_rwsem)) |
f06590bd7 mm: vmscan: corre... |
634 |
goto out; |
1da177e4c Linux-2.6.12-rc2 |
635 636 |
list_for_each_entry(shrinker, &shrinker_list, list) { |
6b4f7799c mm: vmscan: invok... |
637 638 639 |
struct shrink_control sc = { .gfp_mask = gfp_mask, .nid = nid, |
cb731d6c6 vmscan: per memor... |
640 |
.memcg = memcg, |
6b4f7799c mm: vmscan: invok... |
641 |
}; |
ec97097bc mm: vmscan: call ... |
642 |
|
9b996468c mm: add SHRINK_EM... |
643 644 645 646 |
ret = do_shrink_slab(&sc, shrinker, priority); if (ret == SHRINK_EMPTY) ret = 0; freed += ret; |
e496612c5 mm: do not stall ... |
647 648 |
/* * Bail out if someone want to register a new shrinker to |
55b65a57c mm/vmsan: fix som... |
649 |
* prevent the registration from being stalled for long periods |
e496612c5 mm: do not stall ... |
650 651 652 653 654 655 |
* by parallel ongoing shrinking. */ if (rwsem_is_contended(&shrinker_rwsem)) { freed = freed ? : 1; break; } |
1da177e4c Linux-2.6.12-rc2 |
656 |
} |
6b4f7799c mm: vmscan: invok... |
657 |
|
1da177e4c Linux-2.6.12-rc2 |
658 |
up_read(&shrinker_rwsem); |
f06590bd7 mm: vmscan: corre... |
659 660 |
out: cond_resched(); |
24f7c6b98 mm: new shrinker API |
661 |
return freed; |
1da177e4c Linux-2.6.12-rc2 |
662 |
} |
cb731d6c6 vmscan: per memor... |
663 664 665 666 667 668 |
void drop_slab_node(int nid) { unsigned long freed; do { struct mem_cgroup *memcg = NULL; |
069c411de mm/vmscan: fix in... |
669 670 |
if (fatal_signal_pending(current)) return; |
cb731d6c6 vmscan: per memor... |
671 |
freed = 0; |
aeed1d325 mm/vmscan.c: gene... |
672 |
memcg = mem_cgroup_iter(NULL, NULL, NULL); |
cb731d6c6 vmscan: per memor... |
673 |
do { |
9092c71bb mm: use sc->prior... |
674 |
freed += shrink_slab(GFP_KERNEL, nid, memcg, 0); |
cb731d6c6 vmscan: per memor... |
675 676 677 678 679 680 681 682 683 684 685 |
} 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 |
686 687 |
static inline int is_page_cache_freeable(struct page *page) { |
ceddc3a52 mm: document is_p... |
688 689 |
/* * A freeable page cache page is referenced only by the caller |
67891ffff mm: Convert is_pa... |
690 691 |
* that isolated the page, the page cache and optional buffer * heads at page->private. |
ceddc3a52 mm: document is_p... |
692 |
*/ |
3efe62e46 mm/vmscan: allow ... |
693 |
int page_cache_pins = thp_nr_pages(page); |
67891ffff mm: Convert is_pa... |
694 |
return page_count(page) - page_has_private(page) == 1 + page_cache_pins; |
1da177e4c Linux-2.6.12-rc2 |
695 |
} |
cb16556d9 mm/vmscan.c: remo... |
696 |
static int may_write_to_inode(struct inode *inode) |
1da177e4c Linux-2.6.12-rc2 |
697 |
{ |
930d91525 [PATCH] Swap Migr... |
698 |
if (current->flags & PF_SWAPWRITE) |
1da177e4c Linux-2.6.12-rc2 |
699 |
return 1; |
703c27088 writeback: implem... |
700 |
if (!inode_write_congested(inode)) |
1da177e4c Linux-2.6.12-rc2 |
701 |
return 1; |
703c27088 writeback: implem... |
702 |
if (inode_to_bdi(inode) == current->backing_dev_info) |
1da177e4c Linux-2.6.12-rc2 |
703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 |
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... |
722 |
lock_page(page); |
3e9f45bd1 Factor outstandin... |
723 724 |
if (page_mapping(page) == mapping) mapping_set_error(mapping, error); |
1da177e4c Linux-2.6.12-rc2 |
725 726 |
unlock_page(page); } |
04e62a29b [PATCH] More page... |
727 728 729 730 731 732 733 734 735 736 737 |
/* 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 |
738 |
/* |
1742f19fa [PATCH] vmscan: r... |
739 740 |
* pageout is called by shrink_page_list() for each dirty page. * Calls ->writepage(). |
1da177e4c Linux-2.6.12-rc2 |
741 |
*/ |
cb16556d9 mm/vmscan.c: remo... |
742 |
static pageout_t pageout(struct page *page, struct address_space *mapping) |
1da177e4c Linux-2.6.12-rc2 |
743 744 745 746 747 748 749 750 |
{ /* * 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... |
751 |
* If this process is currently in __generic_file_write_iter() against |
1da177e4c Linux-2.6.12-rc2 |
752 753 754 755 756 757 758 |
* 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 |
759 760 761 762 763 764 765 766 |
*/ 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... |
767 |
if (page_has_private(page)) { |
1da177e4c Linux-2.6.12-rc2 |
768 769 |
if (try_to_free_buffers(page)) { ClearPageDirty(page); |
b1de0d139 mm: convert some ... |
770 771 |
pr_info("%s: orphaned page ", __func__); |
1da177e4c Linux-2.6.12-rc2 |
772 773 774 775 776 777 778 |
return PAGE_CLEAN; } } return PAGE_KEEP; } if (mapping->a_ops->writepage == NULL) return PAGE_ACTIVATE; |
cb16556d9 mm/vmscan.c: remo... |
779 |
if (!may_write_to_inode(mapping->host)) |
1da177e4c Linux-2.6.12-rc2 |
780 781 782 783 784 785 786 |
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... |
787 788 |
.range_start = 0, .range_end = LLONG_MAX, |
1da177e4c Linux-2.6.12-rc2 |
789 790 791 792 793 794 795 |
.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... |
796 |
if (res == AOP_WRITEPAGE_ACTIVATE) { |
1da177e4c Linux-2.6.12-rc2 |
797 798 799 |
ClearPageReclaim(page); return PAGE_ACTIVATE; } |
c661b078f synchronous lumpy... |
800 |
|
1da177e4c Linux-2.6.12-rc2 |
801 802 803 804 |
if (!PageWriteback(page)) { /* synchronous write or broken a_ops? */ ClearPageReclaim(page); } |
3aa238511 mm/vmscan.c: chan... |
805 |
trace_mm_vmscan_writepage(page); |
c4a25635b mm: move vmscan w... |
806 |
inc_node_page_state(page, NR_VMSCAN_WRITE); |
1da177e4c Linux-2.6.12-rc2 |
807 808 809 810 811 |
return PAGE_SUCCESS; } return PAGE_CLEAN; } |
a649fd927 [PATCH] invalidat... |
812 |
/* |
e286781d5 mm: speculative p... |
813 814 |
* Same as remove_mapping, but if the page is removed from the mapping, it * gets returned with a refcount of 0. |
a649fd927 [PATCH] invalidat... |
815 |
*/ |
a528910e1 mm: thrash detect... |
816 |
static int __remove_mapping(struct address_space *mapping, struct page *page, |
b910718a9 mm: vmscan: detec... |
817 |
bool reclaimed, struct mem_cgroup *target_memcg) |
49d2e9cc4 [PATCH] Swap Migr... |
818 |
{ |
c4843a759 memcg: add per cg... |
819 |
unsigned long flags; |
bd4c82c22 mm, THP, swap: de... |
820 |
int refcount; |
aae466b00 mm/swap: implemen... |
821 |
void *shadow = NULL; |
c4843a759 memcg: add per cg... |
822 |
|
28e4d965e [PATCH] mm: remov... |
823 824 |
BUG_ON(!PageLocked(page)); BUG_ON(mapping != page_mapping(page)); |
49d2e9cc4 [PATCH] Swap Migr... |
825 |
|
b93b01631 page cache: use x... |
826 |
xa_lock_irqsave(&mapping->i_pages, flags); |
49d2e9cc4 [PATCH] Swap Migr... |
827 |
/* |
0fd0e6b05 [PATCH] page inva... |
828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 |
* 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... |
847 |
* load is not satisfied before that of page->_refcount. |
0fd0e6b05 [PATCH] page inva... |
848 849 |
* * Note that if SetPageDirty is always performed via set_page_dirty, |
b93b01631 page cache: use x... |
850 |
* and thus under the i_pages lock, then this ordering is not required. |
49d2e9cc4 [PATCH] Swap Migr... |
851 |
*/ |
906d278d7 mm/vmscan.c: supp... |
852 |
refcount = 1 + compound_nr(page); |
bd4c82c22 mm, THP, swap: de... |
853 |
if (!page_ref_freeze(page, refcount)) |
49d2e9cc4 [PATCH] Swap Migr... |
854 |
goto cannot_free; |
1c4c3b99c mm: fix page_free... |
855 |
/* note: atomic_cmpxchg in page_ref_freeze provides the smp_rmb */ |
e286781d5 mm: speculative p... |
856 |
if (unlikely(PageDirty(page))) { |
bd4c82c22 mm, THP, swap: de... |
857 |
page_ref_unfreeze(page, refcount); |
49d2e9cc4 [PATCH] Swap Migr... |
858 |
goto cannot_free; |
e286781d5 mm: speculative p... |
859 |
} |
49d2e9cc4 [PATCH] Swap Migr... |
860 861 862 |
if (PageSwapCache(page)) { swp_entry_t swap = { .val = page_private(page) }; |
0a31bc97c mm: memcontrol: r... |
863 |
mem_cgroup_swapout(page, swap); |
aae466b00 mm/swap: implemen... |
864 865 866 |
if (reclaimed && !mapping_exiting(mapping)) shadow = workingset_eviction(page, target_memcg); __delete_from_swap_cache(page, swap, shadow); |
b93b01631 page cache: use x... |
867 |
xa_unlock_irqrestore(&mapping->i_pages, flags); |
75f6d6d29 mm, THP, swap: un... |
868 |
put_swap_page(page, swap); |
e286781d5 mm: speculative p... |
869 |
} else { |
6072d13c4 Call the filesyst... |
870 871 872 |
void (*freepage)(struct page *); freepage = mapping->a_ops->freepage; |
a528910e1 mm: thrash detect... |
873 874 875 876 877 |
/* * 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 |
238c30468 mm/vmscan.c: fix ... |
878 |
* already exiting. This is not just an optimization, |
a528910e1 mm: thrash detect... |
879 880 881 |
* 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... |
882 883 884 885 886 |
* * 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... |
887 |
* same address_space. |
a528910e1 mm: thrash detect... |
888 |
*/ |
9de4f22a6 mm: code cleanup ... |
889 |
if (reclaimed && page_is_file_lru(page) && |
f9fe48bec dax: support dirt... |
890 |
!mapping_exiting(mapping) && !dax_mapping(mapping)) |
b910718a9 mm: vmscan: detec... |
891 |
shadow = workingset_eviction(page, target_memcg); |
62cccb8c8 mm: simplify lock... |
892 |
__delete_from_page_cache(page, shadow); |
b93b01631 page cache: use x... |
893 |
xa_unlock_irqrestore(&mapping->i_pages, flags); |
6072d13c4 Call the filesyst... |
894 895 896 |
if (freepage != NULL) freepage(page); |
49d2e9cc4 [PATCH] Swap Migr... |
897 |
} |
49d2e9cc4 [PATCH] Swap Migr... |
898 899 900 |
return 1; cannot_free: |
b93b01631 page cache: use x... |
901 |
xa_unlock_irqrestore(&mapping->i_pages, flags); |
49d2e9cc4 [PATCH] Swap Migr... |
902 903 |
return 0; } |
1da177e4c Linux-2.6.12-rc2 |
904 |
/* |
e286781d5 mm: speculative p... |
905 906 907 908 909 910 911 |
* 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) { |
b910718a9 mm: vmscan: detec... |
912 |
if (__remove_mapping(mapping, page, false, NULL)) { |
e286781d5 mm: speculative p... |
913 914 915 916 917 |
/* * Unfreezing the refcount with 1 rather than 2 effectively * drops the pagecache ref for us without requiring another * atomic operation. */ |
fe896d187 mm: introduce pag... |
918 |
page_ref_unfreeze(page, 1); |
e286781d5 mm: speculative p... |
919 920 921 922 |
return 1; } return 0; } |
894bc3104 Unevictable LRU I... |
923 924 925 926 927 928 929 930 931 |
/** * 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... |
932 933 |
void putback_lru_page(struct page *page) { |
9c4e6b1a7 mm, mlock, vmscan... |
934 |
lru_cache_add(page); |
894bc3104 Unevictable LRU I... |
935 936 |
put_page(page); /* drop ref from isolate */ } |
dfc8d636c vmscan: factor ou... |
937 938 939 |
enum page_references { PAGEREF_RECLAIM, PAGEREF_RECLAIM_CLEAN, |
645747462 vmscan: detect ma... |
940 |
PAGEREF_KEEP, |
dfc8d636c vmscan: factor ou... |
941 942 943 944 945 946 |
PAGEREF_ACTIVATE, }; static enum page_references page_check_references(struct page *page, struct scan_control *sc) { |
645747462 vmscan: detect ma... |
947 |
int referenced_ptes, referenced_page; |
dfc8d636c vmscan: factor ou... |
948 |
unsigned long vm_flags; |
dfc8d636c vmscan: factor ou... |
949 |
|
c3ac9a8ad mm: memcg: count ... |
950 951 |
referenced_ptes = page_referenced(page, 1, sc->target_mem_cgroup, &vm_flags); |
645747462 vmscan: detect ma... |
952 |
referenced_page = TestClearPageReferenced(page); |
dfc8d636c vmscan: factor ou... |
953 |
|
dfc8d636c vmscan: factor ou... |
954 955 956 957 958 959 |
/* * 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... |
960 |
if (referenced_ptes) { |
645747462 vmscan: detect ma... |
961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 |
/* * 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... |
976 |
if (referenced_page || referenced_ptes > 1) |
645747462 vmscan: detect ma... |
977 |
return PAGEREF_ACTIVATE; |
c909e9936 vmscan: activate ... |
978 979 980 |
/* * Activate file-backed executable pages after first usage. */ |
b518154e5 mm/vmscan: protec... |
981 |
if ((vm_flags & VM_EXEC) && !PageSwapBacked(page)) |
c909e9936 vmscan: activate ... |
982 |
return PAGEREF_ACTIVATE; |
645747462 vmscan: detect ma... |
983 984 |
return PAGEREF_KEEP; } |
dfc8d636c vmscan: factor ou... |
985 986 |
/* Reclaim if clean, defer dirty pages to writeback */ |
2e30244a7 vmscan,tmpfs: tre... |
987 |
if (referenced_page && !PageSwapBacked(page)) |
645747462 vmscan: detect ma... |
988 989 990 |
return PAGEREF_RECLAIM_CLEAN; return PAGEREF_RECLAIM; |
dfc8d636c vmscan: factor ou... |
991 |
} |
e2be15f6c mm: vmscan: stall... |
992 993 994 995 |
/* 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 ... |
996 |
struct address_space *mapping; |
e2be15f6c mm: vmscan: stall... |
997 998 999 1000 |
/* * Anonymous pages are not handled by flushers and must be written * from reclaim context. Do not stall reclaim based on them */ |
9de4f22a6 mm: code cleanup ... |
1001 |
if (!page_is_file_lru(page) || |
802a3a92a mm: reclaim MADV_... |
1002 |
(PageAnon(page) && !PageSwapBacked(page))) { |
e2be15f6c mm: vmscan: stall... |
1003 1004 1005 1006 1007 1008 1009 1010 |
*dirty = false; *writeback = false; return; } /* By default assume that the page flags are accurate */ *dirty = PageDirty(page); *writeback = PageWriteback(page); |
b45972265 mm: vmscan: take ... |
1011 1012 1013 1014 1015 1016 1017 1018 |
/* 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... |
1019 |
} |
e286781d5 mm: speculative p... |
1020 |
/* |
1742f19fa [PATCH] vmscan: r... |
1021 |
* shrink_page_list() returns the number of reclaimed pages |
1da177e4c Linux-2.6.12-rc2 |
1022 |
*/ |
730ec8c01 mm/vmscan.c: chan... |
1023 1024 1025 |
static unsigned int shrink_page_list(struct list_head *page_list, struct pglist_data *pgdat, struct scan_control *sc, |
730ec8c01 mm/vmscan.c: chan... |
1026 1027 |
struct reclaim_stat *stat, bool ignore_references) |
1da177e4c Linux-2.6.12-rc2 |
1028 1029 |
{ LIST_HEAD(ret_pages); |
abe4c3b50 vmscan: set up pa... |
1030 |
LIST_HEAD(free_pages); |
730ec8c01 mm/vmscan.c: chan... |
1031 1032 |
unsigned int nr_reclaimed = 0; unsigned int pgactivate = 0; |
1da177e4c Linux-2.6.12-rc2 |
1033 |
|
060f005f0 mm/vmscan.c: do n... |
1034 |
memset(stat, 0, sizeof(*stat)); |
1da177e4c Linux-2.6.12-rc2 |
1035 |
cond_resched(); |
1da177e4c Linux-2.6.12-rc2 |
1036 1037 1038 |
while (!list_empty(page_list)) { struct address_space *mapping; struct page *page; |
8940b34a4 mm: change PAGERE... |
1039 |
enum page_references references = PAGEREF_RECLAIM; |
4b7930626 mm/vmscan.c: make... |
1040 |
bool dirty, writeback, may_enter_fs; |
98879b3b9 mm: vmscan: corre... |
1041 |
unsigned int nr_pages; |
1da177e4c Linux-2.6.12-rc2 |
1042 1043 1044 1045 1046 |
cond_resched(); page = lru_to_page(page_list); list_del(&page->lru); |
529ae9aaa mm: rename page t... |
1047 |
if (!trylock_page(page)) |
1da177e4c Linux-2.6.12-rc2 |
1048 |
goto keep; |
309381fea mm: dump page whe... |
1049 |
VM_BUG_ON_PAGE(PageActive(page), page); |
1da177e4c Linux-2.6.12-rc2 |
1050 |
|
d8c6546b1 mm: introduce com... |
1051 |
nr_pages = compound_nr(page); |
98879b3b9 mm: vmscan: corre... |
1052 1053 1054 |
/* Account the number of base pages even though THP */ sc->nr_scanned += nr_pages; |
80e434260 [PATCH] zone recl... |
1055 |
|
39b5f29ac mm: remove vma ar... |
1056 |
if (unlikely(!page_evictable(page))) |
ad6b67041 mm: remove SWAP_M... |
1057 |
goto activate_locked; |
894bc3104 Unevictable LRU I... |
1058 |
|
a6dc60f89 vmscan: rename sc... |
1059 |
if (!sc->may_unmap && page_mapped(page)) |
80e434260 [PATCH] zone recl... |
1060 |
goto keep_locked; |
c661b078f synchronous lumpy... |
1061 1062 |
may_enter_fs = (sc->gfp_mask & __GFP_FS) || (PageSwapCache(page) && (sc->gfp_mask & __GFP_IO)); |
283aba9f9 mm: vmscan: block... |
1063 |
/* |
894befec4 mm/vmscan: update... |
1064 |
* The number of dirty pages determines if a node is marked |
e2be15f6c mm: vmscan: stall... |
1065 1066 1067 1068 1069 1070 |
* 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... |
1071 |
stat->nr_dirty++; |
e2be15f6c mm: vmscan: stall... |
1072 1073 |
if (dirty && !writeback) |
060f005f0 mm/vmscan.c: do n... |
1074 |
stat->nr_unqueued_dirty++; |
e2be15f6c mm: vmscan: stall... |
1075 |
|
d04e8acd0 mm: vmscan: treat... |
1076 1077 1078 1079 1080 1081 |
/* * 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... |
1082 |
mapping = page_mapping(page); |
1da58ee2a mm: vmscan: count... |
1083 |
if (((dirty || writeback) && mapping && |
703c27088 writeback: implem... |
1084 |
inode_write_congested(mapping->host)) || |
d04e8acd0 mm: vmscan: treat... |
1085 |
(writeback && PageReclaim(page))) |
060f005f0 mm/vmscan.c: do n... |
1086 |
stat->nr_congested++; |
e2be15f6c mm: vmscan: stall... |
1087 1088 |
/* |
283aba9f9 mm: vmscan: block... |
1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 |
* 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... |
1099 1100 |
* note that the LRU is being scanned too quickly and the * caller can stall after page list has been processed. |
283aba9f9 mm: vmscan: block... |
1101 |
* |
97c9341f7 mm: vmscan: disab... |
1102 |
* 2) Global or new memcg reclaim encounters a page that is |
ecf5fc6e9 mm, vmscan: Do no... |
1103 1104 1105 |
* 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... |
1106 |
* reclaim and continue scanning. |
283aba9f9 mm: vmscan: block... |
1107 |
* |
ecf5fc6e9 mm, vmscan: Do no... |
1108 1109 |
* 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... |
1110 1111 1112 1113 1114 |
* 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... |
1115 |
* 3) Legacy memcg encounters a page that is already marked |
283aba9f9 mm: vmscan: block... |
1116 1117 1118 1119 |
* 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 ... |
1120 1121 1122 1123 1124 1125 1126 1127 1128 |
* * 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... |
1129 |
*/ |
c661b078f synchronous lumpy... |
1130 |
if (PageWriteback(page)) { |
283aba9f9 mm: vmscan: block... |
1131 1132 1133 |
/* Case 1 above */ if (current_is_kswapd() && PageReclaim(page) && |
599d0c954 mm, vmscan: move ... |
1134 |
test_bit(PGDAT_WRITEBACK, &pgdat->flags)) { |
060f005f0 mm/vmscan.c: do n... |
1135 |
stat->nr_immediate++; |
c55e8d035 mm: vmscan: move ... |
1136 |
goto activate_locked; |
283aba9f9 mm: vmscan: block... |
1137 1138 |
/* Case 2 above */ |
b5ead35e7 mm: vmscan: namin... |
1139 |
} else if (writeback_throttling_sane(sc) || |
ecf5fc6e9 mm, vmscan: Do no... |
1140 |
!PageReclaim(page) || !may_enter_fs) { |
c3b94f44f memcg: further pr... |
1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 |
/* * 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... |
1153 |
stat->nr_writeback++; |
c55e8d035 mm: vmscan: move ... |
1154 |
goto activate_locked; |
283aba9f9 mm: vmscan: block... |
1155 1156 1157 |
/* Case 3 above */ } else { |
7fadc8202 mm, vmscan: unloc... |
1158 |
unlock_page(page); |
283aba9f9 mm: vmscan: block... |
1159 |
wait_on_page_writeback(page); |
7fadc8202 mm, vmscan: unloc... |
1160 1161 1162 |
/* then go back and try same page again */ list_add_tail(&page->lru, page_list); continue; |
e62e384e9 memcg: prevent OO... |
1163 |
} |
c661b078f synchronous lumpy... |
1164 |
} |
1da177e4c Linux-2.6.12-rc2 |
1165 |
|
8940b34a4 mm: change PAGERE... |
1166 |
if (!ignore_references) |
02c6de8d7 mm: cma: discard ... |
1167 |
references = page_check_references(page, sc); |
dfc8d636c vmscan: factor ou... |
1168 1169 |
switch (references) { case PAGEREF_ACTIVATE: |
1da177e4c Linux-2.6.12-rc2 |
1170 |
goto activate_locked; |
645747462 vmscan: detect ma... |
1171 |
case PAGEREF_KEEP: |
98879b3b9 mm: vmscan: corre... |
1172 |
stat->nr_ref_keep += nr_pages; |
645747462 vmscan: detect ma... |
1173 |
goto keep_locked; |
dfc8d636c vmscan: factor ou... |
1174 1175 1176 1177 |
case PAGEREF_RECLAIM: case PAGEREF_RECLAIM_CLEAN: ; /* try to reclaim the page below */ } |
1da177e4c Linux-2.6.12-rc2 |
1178 |
|
1da177e4c Linux-2.6.12-rc2 |
1179 1180 1181 |
/* * Anonymous process memory has backing store? * Try to allocate it some swap space here. |
802a3a92a mm: reclaim MADV_... |
1182 |
* Lazyfree page could be freed directly |
1da177e4c Linux-2.6.12-rc2 |
1183 |
*/ |
bd4c82c22 mm, THP, swap: de... |
1184 1185 1186 1187 |
if (PageAnon(page) && PageSwapBacked(page)) { if (!PageSwapCache(page)) { if (!(sc->gfp_mask & __GFP_IO)) goto keep_locked; |
72c5ce894 mm: don't put pin... |
1188 1189 |
if (page_maybe_dma_pinned(page)) goto keep_locked; |
bd4c82c22 mm, THP, swap: de... |
1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 |
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... |
1206 |
goto activate_locked_split; |
bd4c82c22 mm, THP, swap: de... |
1207 1208 1209 1210 |
/* Fallback to swap normal pages */ if (split_huge_page_to_list(page, page_list)) goto activate_locked; |
fe490cc0f mm, THP, swap: ad... |
1211 1212 1213 |
#ifdef CONFIG_TRANSPARENT_HUGEPAGE count_vm_event(THP_SWPOUT_FALLBACK); #endif |
bd4c82c22 mm, THP, swap: de... |
1214 |
if (!add_to_swap(page)) |
98879b3b9 mm: vmscan: corre... |
1215 |
goto activate_locked_split; |
bd4c82c22 mm, THP, swap: de... |
1216 |
} |
0f0746589 mm, THP, swap: mo... |
1217 |
|
4b7930626 mm/vmscan.c: make... |
1218 |
may_enter_fs = true; |
1da177e4c Linux-2.6.12-rc2 |
1219 |
|
bd4c82c22 mm, THP, swap: de... |
1220 1221 1222 |
/* Adding to swap updated mapping */ mapping = page_mapping(page); } |
7751b2da6 vmscan: split fil... |
1223 1224 1225 1226 |
} else if (unlikely(PageTransHuge(page))) { /* Split file THP */ if (split_huge_page_to_list(page, page_list)) goto keep_locked; |
e2be15f6c mm: vmscan: stall... |
1227 |
} |
1da177e4c Linux-2.6.12-rc2 |
1228 1229 |
/* |
98879b3b9 mm: vmscan: corre... |
1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 |
* 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 |
1242 1243 1244 |
* The page is mapped into the page tables of one or more * processes. Try to unmap it here. */ |
802a3a92a mm: reclaim MADV_... |
1245 |
if (page_mapped(page)) { |
dd156e3fc mm/rmap: always d... |
1246 |
enum ttu_flags flags = TTU_BATCH_FLUSH; |
1f318a9b0 mm/vmscan: count ... |
1247 |
bool was_swapbacked = PageSwapBacked(page); |
bd4c82c22 mm, THP, swap: de... |
1248 1249 1250 |
if (unlikely(PageTransHuge(page))) flags |= TTU_SPLIT_HUGE_PMD; |
1f318a9b0 mm/vmscan: count ... |
1251 |
|
bd4c82c22 mm, THP, swap: de... |
1252 |
if (!try_to_unmap(page, flags)) { |
98879b3b9 mm: vmscan: corre... |
1253 |
stat->nr_unmap_fail += nr_pages; |
1f318a9b0 mm/vmscan: count ... |
1254 1255 |
if (!was_swapbacked && PageSwapBacked(page)) stat->nr_lazyfree_fail += nr_pages; |
1da177e4c Linux-2.6.12-rc2 |
1256 |
goto activate_locked; |
1da177e4c Linux-2.6.12-rc2 |
1257 1258 1259 1260 |
} } if (PageDirty(page)) { |
ee72886d8 mm: vmscan: do no... |
1261 |
/* |
4eda48235 mm: vmscan: only ... |
1262 1263 1264 1265 1266 1267 1268 1269 |
* 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... |
1270 |
*/ |
9de4f22a6 mm: code cleanup ... |
1271 |
if (page_is_file_lru(page) && |
4eda48235 mm: vmscan: only ... |
1272 1273 |
(!current_is_kswapd() || !PageReclaim(page) || !test_bit(PGDAT_DIRTY, &pgdat->flags))) { |
49ea7eb65 mm: vmscan: immed... |
1274 1275 1276 1277 1278 1279 |
/* * 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... |
1280 |
inc_node_page_state(page, NR_VMSCAN_IMMEDIATE); |
49ea7eb65 mm: vmscan: immed... |
1281 |
SetPageReclaim(page); |
c55e8d035 mm: vmscan: move ... |
1282 |
goto activate_locked; |
ee72886d8 mm: vmscan: do no... |
1283 |
} |
dfc8d636c vmscan: factor ou... |
1284 |
if (references == PAGEREF_RECLAIM_CLEAN) |
1da177e4c Linux-2.6.12-rc2 |
1285 |
goto keep_locked; |
4dd4b9202 revert "kswapd sh... |
1286 |
if (!may_enter_fs) |
1da177e4c Linux-2.6.12-rc2 |
1287 |
goto keep_locked; |
52a8363ea [PATCH] mm: impro... |
1288 |
if (!sc->may_writepage) |
1da177e4c Linux-2.6.12-rc2 |
1289 |
goto keep_locked; |
d950c9477 mm: defer flush o... |
1290 1291 1292 1293 1294 1295 |
/* * 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(); |
cb16556d9 mm/vmscan.c: remo... |
1296 |
switch (pageout(page, mapping)) { |
1da177e4c Linux-2.6.12-rc2 |
1297 1298 1299 1300 1301 |
case PAGE_KEEP: goto keep_locked; case PAGE_ACTIVATE: goto activate_locked; case PAGE_SUCCESS: |
6c357848b mm: replace hpage... |
1302 |
stat->nr_pageout += thp_nr_pages(page); |
96f8bf4fb mm: vmscan: recla... |
1303 |
|
7d3579e8e vmscan: narrow th... |
1304 |
if (PageWriteback(page)) |
41ac1999c mm: vmscan: do no... |
1305 |
goto keep; |
7d3579e8e vmscan: narrow th... |
1306 |
if (PageDirty(page)) |
1da177e4c Linux-2.6.12-rc2 |
1307 |
goto keep; |
7d3579e8e vmscan: narrow th... |
1308 |
|
1da177e4c Linux-2.6.12-rc2 |
1309 1310 1311 1312 |
/* * A synchronous write - probably a ramdisk. Go * ahead and try to reclaim the page. */ |
529ae9aaa mm: rename page t... |
1313 |
if (!trylock_page(page)) |
1da177e4c Linux-2.6.12-rc2 |
1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 |
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... |
1333 |
* will do this, as well as the blockdev mapping. |
1da177e4c Linux-2.6.12-rc2 |
1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 |
* 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... |
1344 |
if (page_has_private(page)) { |
1da177e4c Linux-2.6.12-rc2 |
1345 1346 |
if (!try_to_release_page(page, sc->gfp_mask)) goto activate_locked; |
e286781d5 mm: speculative p... |
1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 |
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 |
1363 |
} |
802a3a92a mm: reclaim MADV_... |
1364 1365 1366 1367 1368 1369 1370 1371 |
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 |
1372 |
|
802a3a92a mm: reclaim MADV_... |
1373 |
count_vm_event(PGLAZYFREED); |
2262185c5 mm: per-cgroup me... |
1374 |
count_memcg_page_event(page, PGLAZYFREED); |
b910718a9 mm: vmscan: detec... |
1375 1376 |
} else if (!mapping || !__remove_mapping(mapping, page, true, sc->target_mem_cgroup)) |
802a3a92a mm: reclaim MADV_... |
1377 |
goto keep_locked; |
9a1ea439b mm: put_and_wait_... |
1378 1379 |
unlock_page(page); |
e286781d5 mm: speculative p... |
1380 |
free_it: |
98879b3b9 mm: vmscan: corre... |
1381 1382 1383 1384 1385 |
/* * THP may get swapped out in a whole, need account * all base pages. */ nr_reclaimed += nr_pages; |
abe4c3b50 vmscan: set up pa... |
1386 1387 1388 1389 1390 |
/* * Is there need to periodically free_page_list? It would * appear not as the counts should be low */ |
7ae88534c mm: move mem_cgro... |
1391 |
if (unlikely(PageTransHuge(page))) |
ff45fc3ca mm: simplify call... |
1392 |
destroy_compound_page(page); |
7ae88534c mm: move mem_cgro... |
1393 |
else |
bd4c82c22 mm, THP, swap: de... |
1394 |
list_add(&page->lru, &free_pages); |
1da177e4c Linux-2.6.12-rc2 |
1395 |
continue; |
98879b3b9 mm: vmscan: corre... |
1396 1397 1398 1399 1400 1401 1402 1403 1404 |
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 |
1405 |
activate_locked: |
68a22394c vmscan: free swap... |
1406 |
/* Not a candidate for swapping, so reclaim swap space. */ |
ad6b67041 mm: remove SWAP_M... |
1407 1408 |
if (PageSwapCache(page) && (mem_cgroup_swap_full(page) || PageMlocked(page))) |
a2c43eed8 mm: try_to_free_s... |
1409 |
try_to_free_swap(page); |
309381fea mm: dump page whe... |
1410 |
VM_BUG_ON_PAGE(PageActive(page), page); |
ad6b67041 mm: remove SWAP_M... |
1411 |
if (!PageMlocked(page)) { |
9de4f22a6 mm: code cleanup ... |
1412 |
int type = page_is_file_lru(page); |
ad6b67041 mm: remove SWAP_M... |
1413 |
SetPageActive(page); |
98879b3b9 mm: vmscan: corre... |
1414 |
stat->nr_activate[type] += nr_pages; |
2262185c5 mm: per-cgroup me... |
1415 |
count_memcg_page_event(page, PGACTIVATE); |
ad6b67041 mm: remove SWAP_M... |
1416 |
} |
1da177e4c Linux-2.6.12-rc2 |
1417 1418 1419 1420 |
keep_locked: unlock_page(page); keep: list_add(&page->lru, &ret_pages); |
309381fea mm: dump page whe... |
1421 |
VM_BUG_ON_PAGE(PageLRU(page) || PageUnevictable(page), page); |
1da177e4c Linux-2.6.12-rc2 |
1422 |
} |
abe4c3b50 vmscan: set up pa... |
1423 |
|
98879b3b9 mm: vmscan: corre... |
1424 |
pgactivate = stat->nr_activate[0] + stat->nr_activate[1]; |
747db954c mm: memcontrol: u... |
1425 |
mem_cgroup_uncharge_list(&free_pages); |
72b252aed mm: send one IPI ... |
1426 |
try_to_unmap_flush(); |
2d4894b5d mm: remove cold p... |
1427 |
free_unref_page_list(&free_pages); |
abe4c3b50 vmscan: set up pa... |
1428 |
|
1da177e4c Linux-2.6.12-rc2 |
1429 |
list_splice(&ret_pages, page_list); |
886cf1901 mm: move recent_r... |
1430 |
count_vm_events(PGACTIVATE, pgactivate); |
060f005f0 mm/vmscan.c: do n... |
1431 |
|
05ff51376 [PATCH] vmscan re... |
1432 |
return nr_reclaimed; |
1da177e4c Linux-2.6.12-rc2 |
1433 |
} |
730ec8c01 mm/vmscan.c: chan... |
1434 |
unsigned int reclaim_clean_pages_from_list(struct zone *zone, |
02c6de8d7 mm: cma: discard ... |
1435 1436 1437 1438 1439 1440 1441 |
struct list_head *page_list) { struct scan_control sc = { .gfp_mask = GFP_KERNEL, .priority = DEF_PRIORITY, .may_unmap = 1, }; |
1f318a9b0 mm/vmscan: count ... |
1442 |
struct reclaim_stat stat; |
730ec8c01 mm/vmscan.c: chan... |
1443 |
unsigned int nr_reclaimed; |
02c6de8d7 mm: cma: discard ... |
1444 1445 1446 1447 |
struct page *page, *next; LIST_HEAD(clean_pages); list_for_each_entry_safe(page, next, page_list, lru) { |
9de4f22a6 mm: code cleanup ... |
1448 |
if (page_is_file_lru(page) && !PageDirty(page) && |
a58f2cef2 mm/vmscan.c: fix ... |
1449 |
!__PageMovable(page) && !PageUnevictable(page)) { |
02c6de8d7 mm: cma: discard ... |
1450 1451 1452 1453 |
ClearPageActive(page); list_move(&page->lru, &clean_pages); } } |
1f318a9b0 mm/vmscan: count ... |
1454 |
nr_reclaimed = shrink_page_list(&clean_pages, zone->zone_pgdat, &sc, |
dd156e3fc mm/rmap: always d... |
1455 |
&stat, true); |
02c6de8d7 mm: cma: discard ... |
1456 |
list_splice(&clean_pages, page_list); |
2da9f6305 mm/vmscan: fix NR... |
1457 1458 |
mod_node_page_state(zone->zone_pgdat, NR_ISOLATED_FILE, -(long)nr_reclaimed); |
1f318a9b0 mm/vmscan: count ... |
1459 1460 1461 1462 1463 1464 1465 1466 1467 |
/* * Since lazyfree pages are isolated from file LRU from the beginning, * they will rotate back to anonymous LRU in the end if it failed to * discard so isolated count will be mismatched. * Compensate the isolated count for both LRU lists. */ mod_node_page_state(zone->zone_pgdat, NR_ISOLATED_ANON, stat.nr_lazyfree_fail); mod_node_page_state(zone->zone_pgdat, NR_ISOLATED_FILE, |
2da9f6305 mm/vmscan: fix NR... |
1468 |
-(long)stat.nr_lazyfree_fail); |
1f318a9b0 mm/vmscan: count ... |
1469 |
return nr_reclaimed; |
02c6de8d7 mm: cma: discard ... |
1470 |
} |
5ad333eb6 Lumpy Reclaim V4 |
1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 |
/* * 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... |
1481 |
int __isolate_lru_page(struct page *page, isolate_mode_t mode) |
5ad333eb6 Lumpy Reclaim V4 |
1482 1483 1484 1485 1486 1487 |
{ int ret = -EINVAL; /* Only take pages on the LRU. */ if (!PageLRU(page)) return ret; |
e46a28790 CMA: migrate mloc... |
1488 1489 |
/* Compaction should not handle unevictable pages but CMA can do so */ if (PageUnevictable(page) && !(mode & ISOLATE_UNEVICTABLE)) |
894bc3104 Unevictable LRU I... |
1490 |
return ret; |
5ad333eb6 Lumpy Reclaim V4 |
1491 |
ret = -EBUSY; |
08e552c69 memcg: synchroniz... |
1492 |
|
c82449352 mm: compaction: m... |
1493 1494 1495 1496 1497 |
/* * 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... |
1498 1499 1500 |
* 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 ... |
1501 |
if (mode & ISOLATE_ASYNC_MIGRATE) { |
c82449352 mm: compaction: m... |
1502 1503 1504 1505 1506 1507 |
/* 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... |
1508 |
bool migrate_dirty; |
c82449352 mm: compaction: m... |
1509 |
|
c82449352 mm: compaction: m... |
1510 1511 1512 |
/* * Only pages without mappings or that have a * ->migratepage callback are possible to migrate |
69d763fc6 mm: pin address_s... |
1513 1514 1515 1516 1517 |
* 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... |
1518 |
*/ |
69d763fc6 mm: pin address_s... |
1519 1520 |
if (!trylock_page(page)) return ret; |
c82449352 mm: compaction: m... |
1521 |
mapping = page_mapping(page); |
145e1a71e mm: fix the NULL ... |
1522 |
migrate_dirty = !mapping || mapping->a_ops->migratepage; |
69d763fc6 mm: pin address_s... |
1523 1524 |
unlock_page(page); if (!migrate_dirty) |
c82449352 mm: compaction: m... |
1525 1526 1527 |
return ret; } } |
39deaf858 mm: compaction: m... |
1528 |
|
f80c06736 mm: zone_reclaim:... |
1529 1530 |
if ((mode & ISOLATE_UNMAPPED) && page_mapped(page)) return ret; |
5ad333eb6 Lumpy Reclaim V4 |
1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 |
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... |
1543 1544 1545 |
/* * Update LRU sizes after isolating pages. The LRU size updates must |
55b65a57c mm/vmsan: fix som... |
1546 |
* be complete before mem_cgroup_update_lru_size due to a sanity check. |
7ee36a14f mm, vmscan: Updat... |
1547 1548 |
*/ static __always_inline void update_lru_sizes(struct lruvec *lruvec, |
b4536f0c8 mm, memcg: fix th... |
1549 |
enum lru_list lru, unsigned long *nr_zone_taken) |
7ee36a14f mm, vmscan: Updat... |
1550 |
{ |
7ee36a14f mm, vmscan: Updat... |
1551 |
int zid; |
7ee36a14f mm, vmscan: Updat... |
1552 1553 1554 |
for (zid = 0; zid < MAX_NR_ZONES; zid++) { if (!nr_zone_taken[zid]) continue; |
a892cb6b9 mm/vmscan.c: use ... |
1555 |
update_lru_size(lruvec, lru, zid, -nr_zone_taken[zid]); |
b4536f0c8 mm, memcg: fix th... |
1556 |
} |
7ee36a14f mm, vmscan: Updat... |
1557 |
} |
f4b7e272b mm: remove zone_l... |
1558 1559 |
/** * pgdat->lru_lock is heavily contended. Some of the functions that |
1da177e4c Linux-2.6.12-rc2 |
1560 1561 1562 1563 1564 1565 1566 1567 |
* 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 ... |
1568 |
* @nr_to_scan: The number of eligible pages to look through on the list. |
5dc35979e mm/vmscan: push l... |
1569 |
* @lruvec: The LRU vector to pull pages from. |
1da177e4c Linux-2.6.12-rc2 |
1570 |
* @dst: The temp list to put pages on to. |
f626012db mm: remove isolat... |
1571 |
* @nr_scanned: The number of pages that were scanned. |
fe2c2a106 vmscan: reclaim a... |
1572 |
* @sc: The scan_control struct for this reclaim session |
3cb994517 mm: push lru inde... |
1573 |
* @lru: LRU list id for isolating |
1da177e4c Linux-2.6.12-rc2 |
1574 1575 1576 |
* * returns how many pages were moved onto *@dst. */ |
69e05944a [PATCH] vmscan: u... |
1577 |
static unsigned long isolate_lru_pages(unsigned long nr_to_scan, |
5dc35979e mm/vmscan: push l... |
1578 |
struct lruvec *lruvec, struct list_head *dst, |
fe2c2a106 vmscan: reclaim a... |
1579 |
unsigned long *nr_scanned, struct scan_control *sc, |
a9e7c39fa mm/vmscan.c: remo... |
1580 |
enum lru_list lru) |
1da177e4c Linux-2.6.12-rc2 |
1581 |
{ |
75b00af77 mm: trivial clean... |
1582 |
struct list_head *src = &lruvec->lists[lru]; |
69e05944a [PATCH] vmscan: u... |
1583 |
unsigned long nr_taken = 0; |
599d0c954 mm, vmscan: move ... |
1584 |
unsigned long nr_zone_taken[MAX_NR_ZONES] = { 0 }; |
7cc30fcfd mm: vmstat: accou... |
1585 |
unsigned long nr_skipped[MAX_NR_ZONES] = { 0, }; |
3db65812d Revert "mm, vmsca... |
1586 |
unsigned long skipped = 0; |
791b48b64 mm: vmscan: scan ... |
1587 |
unsigned long scan, total_scan, nr_pages; |
b2e18757f mm, vmscan: begin... |
1588 |
LIST_HEAD(pages_skipped); |
a9e7c39fa mm/vmscan.c: remo... |
1589 |
isolate_mode_t mode = (sc->may_unmap ? 0 : ISOLATE_UNMAPPED); |
1da177e4c Linux-2.6.12-rc2 |
1590 |
|
98879b3b9 mm: vmscan: corre... |
1591 |
total_scan = 0; |
791b48b64 mm: vmscan: scan ... |
1592 |
scan = 0; |
98879b3b9 mm: vmscan: corre... |
1593 |
while (scan < nr_to_scan && !list_empty(src)) { |
5ad333eb6 Lumpy Reclaim V4 |
1594 |
struct page *page; |
5ad333eb6 Lumpy Reclaim V4 |
1595 |
|
1da177e4c Linux-2.6.12-rc2 |
1596 1597 |
page = lru_to_page(src); prefetchw_prev_lru_page(page, src, flags); |
309381fea mm: dump page whe... |
1598 |
VM_BUG_ON_PAGE(!PageLRU(page), page); |
8d438f96d [PATCH] mm: PageL... |
1599 |
|
d8c6546b1 mm: introduce com... |
1600 |
nr_pages = compound_nr(page); |
98879b3b9 mm: vmscan: corre... |
1601 |
total_scan += nr_pages; |
b2e18757f mm, vmscan: begin... |
1602 1603 |
if (page_zonenum(page) > sc->reclaim_idx) { list_move(&page->lru, &pages_skipped); |
98879b3b9 mm: vmscan: corre... |
1604 |
nr_skipped[page_zonenum(page)] += nr_pages; |
b2e18757f mm, vmscan: begin... |
1605 1606 |
continue; } |
791b48b64 mm: vmscan: scan ... |
1607 1608 1609 1610 1611 |
/* * 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... |
1612 1613 1614 1615 |
* * 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 ... |
1616 |
*/ |
98879b3b9 mm: vmscan: corre... |
1617 |
scan += nr_pages; |
f3fd4a619 mm: remove lru ty... |
1618 |
switch (__isolate_lru_page(page, mode)) { |
5ad333eb6 Lumpy Reclaim V4 |
1619 |
case 0: |
599d0c954 mm, vmscan: move ... |
1620 1621 |
nr_taken += nr_pages; nr_zone_taken[page_zonenum(page)] += nr_pages; |
5ad333eb6 Lumpy Reclaim V4 |
1622 |
list_move(&page->lru, dst); |
5ad333eb6 Lumpy Reclaim V4 |
1623 1624 1625 1626 1627 1628 |
break; case -EBUSY: /* else it is being freed elsewhere */ list_move(&page->lru, src); continue; |
46453a6e1 [PATCH] mm: never... |
1629 |
|
5ad333eb6 Lumpy Reclaim V4 |
1630 1631 1632 |
default: BUG(); } |
1da177e4c Linux-2.6.12-rc2 |
1633 |
} |
b2e18757f mm, vmscan: begin... |
1634 1635 1636 1637 1638 1639 1640 |
/* * 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... |
1641 1642 |
if (!list_empty(&pages_skipped)) { int zid; |
3db65812d Revert "mm, vmsca... |
1643 |
list_splice(&pages_skipped, src); |
7cc30fcfd mm: vmstat: accou... |
1644 1645 1646 1647 1648 |
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 ... |
1649 |
skipped += nr_skipped[zid]; |
7cc30fcfd mm: vmstat: accou... |
1650 1651 |
} } |
791b48b64 mm: vmscan: scan ... |
1652 |
*nr_scanned = total_scan; |
1265e3a69 mm, vmscan: show ... |
1653 |
trace_mm_vmscan_lru_isolate(sc->reclaim_idx, sc->order, nr_to_scan, |
791b48b64 mm: vmscan: scan ... |
1654 |
total_scan, skipped, nr_taken, mode, lru); |
b4536f0c8 mm, memcg: fix th... |
1655 |
update_lru_sizes(lruvec, lru, nr_zone_taken); |
1da177e4c Linux-2.6.12-rc2 |
1656 1657 |
return nr_taken; } |
62695a84e vmscan: move isol... |
1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 |
/** * 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... |
1669 1670 1671 |
* 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... |
1672 1673 1674 1675 1676 |
* * The vmstat statistic corresponding to the list on which the page was * found will be decremented. * * Restrictions: |
a5d09bed7 mm: docs: add bla... |
1677 |
* |
62695a84e vmscan: move isol... |
1678 |
* (1) Must be called with an elevated refcount on the page. This is a |
01c4776ba mm/vmscan: fix co... |
1679 |
* fundamental difference from isolate_lru_pages (which is called |
62695a84e vmscan: move isol... |
1680 1681 1682 1683 1684 1685 1686 |
* 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... |
1687 |
VM_BUG_ON_PAGE(!page_count(page), page); |
cf2a82ee4 mm: downgrade VM_... |
1688 |
WARN_RATELIMIT(PageTail(page), "trying to isolate tail page"); |
0c917313a mm: strictly requ... |
1689 |
|
62695a84e vmscan: move isol... |
1690 |
if (PageLRU(page)) { |
f4b7e272b mm: remove zone_l... |
1691 |
pg_data_t *pgdat = page_pgdat(page); |
fa9add641 mm/memcg: apply a... |
1692 |
struct lruvec *lruvec; |
62695a84e vmscan: move isol... |
1693 |
|
f4b7e272b mm: remove zone_l... |
1694 1695 |
spin_lock_irq(&pgdat->lru_lock); lruvec = mem_cgroup_page_lruvec(page, pgdat); |
0c917313a mm: strictly requ... |
1696 |
if (PageLRU(page)) { |
894bc3104 Unevictable LRU I... |
1697 |
int lru = page_lru(page); |
0c917313a mm: strictly requ... |
1698 |
get_page(page); |
62695a84e vmscan: move isol... |
1699 |
ClearPageLRU(page); |
fa9add641 mm/memcg: apply a... |
1700 1701 |
del_page_from_lru_list(page, lruvec, lru); ret = 0; |
62695a84e vmscan: move isol... |
1702 |
} |
f4b7e272b mm: remove zone_l... |
1703 |
spin_unlock_irq(&pgdat->lru_lock); |
62695a84e vmscan: move isol... |
1704 1705 1706 |
} return ret; } |
5ad333eb6 Lumpy Reclaim V4 |
1707 |
/* |
d37dd5dcb vmscan: comment t... |
1708 |
* A direct reclaimer may isolate SWAP_CLUSTER_MAX pages from the LRU list and |
178821b89 mm/vmscan.c: fix ... |
1709 |
* then get rescheduled. When there are massive number of tasks doing page |
d37dd5dcb vmscan: comment t... |
1710 1711 1712 |
* 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 ... |
1713 |
*/ |
599d0c954 mm, vmscan: move ... |
1714 |
static int too_many_isolated(struct pglist_data *pgdat, int file, |
35cd78156 vmscan: throttle ... |
1715 1716 1717 1718 1719 1720 |
struct scan_control *sc) { unsigned long inactive, isolated; if (current_is_kswapd()) return 0; |
b5ead35e7 mm: vmscan: namin... |
1721 |
if (!writeback_throttling_sane(sc)) |
35cd78156 vmscan: throttle ... |
1722 1723 1724 |
return 0; if (file) { |
599d0c954 mm, vmscan: move ... |
1725 1726 |
inactive = node_page_state(pgdat, NR_INACTIVE_FILE); isolated = node_page_state(pgdat, NR_ISOLATED_FILE); |
35cd78156 vmscan: throttle ... |
1727 |
} else { |
599d0c954 mm, vmscan: move ... |
1728 1729 |
inactive = node_page_state(pgdat, NR_INACTIVE_ANON); isolated = node_page_state(pgdat, NR_ISOLATED_ANON); |
35cd78156 vmscan: throttle ... |
1730 |
} |
3cf23841b mm/vmscan.c: avoi... |
1731 1732 1733 1734 1735 |
/* * 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... |
1736 |
if ((sc->gfp_mask & (__GFP_IO | __GFP_FS)) == (__GFP_IO | __GFP_FS)) |
3cf23841b mm/vmscan.c: avoi... |
1737 |
inactive >>= 3; |
35cd78156 vmscan: throttle ... |
1738 1739 |
return isolated > inactive; } |
a222f3415 mm: generalize pu... |
1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 |
/* * 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... |
1762 |
{ |
599d0c954 mm, vmscan: move ... |
1763 |
struct pglist_data *pgdat = lruvec_pgdat(lruvec); |
a222f3415 mm: generalize pu... |
1764 |
int nr_pages, nr_moved = 0; |
3f79768f2 mm: rearrange put... |
1765 |
LIST_HEAD(pages_to_free); |
a222f3415 mm: generalize pu... |
1766 1767 |
struct page *page; enum lru_list lru; |
666356297 vmscan: set up pa... |
1768 |
|
a222f3415 mm: generalize pu... |
1769 1770 |
while (!list_empty(list)) { page = lru_to_page(list); |
309381fea mm: dump page whe... |
1771 |
VM_BUG_ON_PAGE(PageLRU(page), page); |
39b5f29ac mm: remove vma ar... |
1772 |
if (unlikely(!page_evictable(page))) { |
a222f3415 mm: generalize pu... |
1773 |
list_del(&page->lru); |
599d0c954 mm, vmscan: move ... |
1774 |
spin_unlock_irq(&pgdat->lru_lock); |
666356297 vmscan: set up pa... |
1775 |
putback_lru_page(page); |
599d0c954 mm, vmscan: move ... |
1776 |
spin_lock_irq(&pgdat->lru_lock); |
666356297 vmscan: set up pa... |
1777 1778 |
continue; } |
599d0c954 mm, vmscan: move ... |
1779 |
lruvec = mem_cgroup_page_lruvec(page, pgdat); |
fa9add641 mm/memcg: apply a... |
1780 |
|
7a608572a Revert "mm: batch... |
1781 |
SetPageLRU(page); |
666356297 vmscan: set up pa... |
1782 |
lru = page_lru(page); |
a222f3415 mm: generalize pu... |
1783 |
|
6c357848b mm: replace hpage... |
1784 |
nr_pages = thp_nr_pages(page); |
a222f3415 mm: generalize pu... |
1785 1786 |
update_lru_size(lruvec, lru, page_zonenum(page), nr_pages); list_move(&page->lru, &lruvec->lists[lru]); |
fa9add641 mm/memcg: apply a... |
1787 |
|
2bcf88796 mm: take pagevecs... |
1788 1789 1790 |
if (put_page_testzero(page)) { __ClearPageLRU(page); __ClearPageActive(page); |
fa9add641 mm/memcg: apply a... |
1791 |
del_page_from_lru_list(page, lruvec, lru); |
2bcf88796 mm: take pagevecs... |
1792 1793 |
if (unlikely(PageCompound(page))) { |
599d0c954 mm, vmscan: move ... |
1794 |
spin_unlock_irq(&pgdat->lru_lock); |
ff45fc3ca mm: simplify call... |
1795 |
destroy_compound_page(page); |
599d0c954 mm, vmscan: move ... |
1796 |
spin_lock_irq(&pgdat->lru_lock); |
2bcf88796 mm: take pagevecs... |
1797 1798 |
} else list_add(&page->lru, &pages_to_free); |
a222f3415 mm: generalize pu... |
1799 1800 |
} else { nr_moved += nr_pages; |
31d8fcac0 mm: workingset: a... |
1801 1802 |
if (PageActive(page)) workingset_age_nonresident(lruvec, nr_pages); |
666356297 vmscan: set up pa... |
1803 1804 |
} } |
666356297 vmscan: set up pa... |
1805 |
|
3f79768f2 mm: rearrange put... |
1806 1807 1808 |
/* * To save our caller's stack, now use input list for pages to free. */ |
a222f3415 mm: generalize pu... |
1809 1810 1811 |
list_splice(&pages_to_free, list); return nr_moved; |
666356297 vmscan: set up pa... |
1812 1813 1814 |
} /* |
399ba0b95 mm/vmscan.c: avoi... |
1815 |
* If a kernel thread (such as nfsd for loop-back mounts) services |
a37b0715d mm/writeback: rep... |
1816 |
* a backing device by writing to the page cache it sets PF_LOCAL_THROTTLE. |
399ba0b95 mm/vmscan.c: avoi... |
1817 1818 1819 1820 1821 |
* 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) { |
a37b0715d mm/writeback: rep... |
1822 |
return !(current->flags & PF_LOCAL_THROTTLE) || |
399ba0b95 mm/vmscan.c: avoi... |
1823 1824 1825 1826 1827 |
current->backing_dev_info == NULL || bdi_write_congested(current->backing_dev_info); } /* |
b2e18757f mm, vmscan: begin... |
1828 |
* shrink_inactive_list() is a helper for shrink_node(). It returns the number |
1742f19fa [PATCH] vmscan: r... |
1829 |
* of reclaimed pages |
1da177e4c Linux-2.6.12-rc2 |
1830 |
*/ |
666356297 vmscan: set up pa... |
1831 |
static noinline_for_stack unsigned long |
1a93be0e7 mm/vmscan: push l... |
1832 |
shrink_inactive_list(unsigned long nr_to_scan, struct lruvec *lruvec, |
9e3b2f8cd mm/vmscan: store ... |
1833 |
struct scan_control *sc, enum lru_list lru) |
1da177e4c Linux-2.6.12-rc2 |
1834 1835 |
{ LIST_HEAD(page_list); |
e247dbce5 vmscan: simplify ... |
1836 |
unsigned long nr_scanned; |
730ec8c01 mm/vmscan.c: chan... |
1837 |
unsigned int nr_reclaimed = 0; |
e247dbce5 vmscan: simplify ... |
1838 |
unsigned long nr_taken; |
060f005f0 mm/vmscan.c: do n... |
1839 |
struct reclaim_stat stat; |
497a6c1b0 mm: keep separate... |
1840 |
bool file = is_file_lru(lru); |
f46b79120 mm/vmscan.c: simp... |
1841 |
enum vm_event_item item; |
599d0c954 mm, vmscan: move ... |
1842 |
struct pglist_data *pgdat = lruvec_pgdat(lruvec); |
db73ee0d4 mm, vmscan: do no... |
1843 |
bool stalled = false; |
78dc583d3 vmscan: low order... |
1844 |
|
599d0c954 mm, vmscan: move ... |
1845 |
while (unlikely(too_many_isolated(pgdat, file, sc))) { |
db73ee0d4 mm, vmscan: do no... |
1846 1847 1848 1849 1850 1851 |
if (stalled) return 0; /* wait a bit for the reclaimer. */ msleep(100); stalled = true; |
35cd78156 vmscan: throttle ... |
1852 1853 1854 1855 1856 |
/* 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 |
1857 |
lru_add_drain(); |
f80c06736 mm: zone_reclaim:... |
1858 |
|
599d0c954 mm, vmscan: move ... |
1859 |
spin_lock_irq(&pgdat->lru_lock); |
b35ea17b7 mm: shrink_inacti... |
1860 |
|
5dc35979e mm/vmscan: push l... |
1861 |
nr_taken = isolate_lru_pages(nr_to_scan, lruvec, &page_list, |
a9e7c39fa mm/vmscan.c: remo... |
1862 |
&nr_scanned, sc, lru); |
95d918fc0 mm/vmscan: remove... |
1863 |
|
599d0c954 mm, vmscan: move ... |
1864 |
__mod_node_page_state(pgdat, NR_ISOLATED_ANON + file, nr_taken); |
f46b79120 mm/vmscan.c: simp... |
1865 |
item = current_is_kswapd() ? PGSCAN_KSWAPD : PGSCAN_DIRECT; |
b5ead35e7 mm: vmscan: namin... |
1866 |
if (!cgroup_reclaim(sc)) |
f46b79120 mm/vmscan.c: simp... |
1867 1868 |
__count_vm_events(item, nr_scanned); __count_memcg_events(lruvec_memcg(lruvec), item, nr_scanned); |
497a6c1b0 mm: keep separate... |
1869 |
__count_vm_events(PGSCAN_ANON + file, nr_scanned); |
599d0c954 mm, vmscan: move ... |
1870 |
spin_unlock_irq(&pgdat->lru_lock); |
b35ea17b7 mm: shrink_inacti... |
1871 |
|
d563c0501 vmscan: handle is... |
1872 |
if (nr_taken == 0) |
666356297 vmscan: set up pa... |
1873 |
return 0; |
5ad333eb6 Lumpy Reclaim V4 |
1874 |
|
dd156e3fc mm/rmap: always d... |
1875 |
nr_reclaimed = shrink_page_list(&page_list, pgdat, sc, &stat, false); |
c661b078f synchronous lumpy... |
1876 |
|
599d0c954 mm, vmscan: move ... |
1877 |
spin_lock_irq(&pgdat->lru_lock); |
3f79768f2 mm: rearrange put... |
1878 |
|
497a6c1b0 mm: keep separate... |
1879 1880 1881 |
move_pages_to_lru(lruvec, &page_list); __mod_node_page_state(pgdat, NR_ISOLATED_ANON + file, -nr_taken); |
96f8bf4fb mm: vmscan: recla... |
1882 |
lru_note_cost(lruvec, file, stat.nr_pageout); |
f46b79120 mm/vmscan.c: simp... |
1883 |
item = current_is_kswapd() ? PGSTEAL_KSWAPD : PGSTEAL_DIRECT; |
b5ead35e7 mm: vmscan: namin... |
1884 |
if (!cgroup_reclaim(sc)) |
f46b79120 mm/vmscan.c: simp... |
1885 1886 |
__count_vm_events(item, nr_reclaimed); __count_memcg_events(lruvec_memcg(lruvec), item, nr_reclaimed); |
497a6c1b0 mm: keep separate... |
1887 |
__count_vm_events(PGSTEAL_ANON + file, nr_reclaimed); |
3f79768f2 mm: rearrange put... |
1888 |
|
599d0c954 mm, vmscan: move ... |
1889 |
spin_unlock_irq(&pgdat->lru_lock); |
3f79768f2 mm: rearrange put... |
1890 |
|
747db954c mm: memcontrol: u... |
1891 |
mem_cgroup_uncharge_list(&page_list); |
2d4894b5d mm: remove cold p... |
1892 |
free_unref_page_list(&page_list); |
e11da5b4f tracing, vmscan: ... |
1893 |
|
92df3a723 mm: vmscan: throt... |
1894 |
/* |
1c610d5f9 mm/vmscan: wake u... |
1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 |
* 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 ... |
1907 1908 1909 1910 1911 1912 1913 1914 |
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 ... |
1915 |
|
599d0c954 mm, vmscan: move ... |
1916 |
trace_mm_vmscan_lru_shrink_inactive(pgdat->node_id, |
d51d1e645 mm, vmscan, traci... |
1917 |
nr_scanned, nr_reclaimed, &stat, sc->priority, file); |
05ff51376 [PATCH] vmscan re... |
1918 |
return nr_reclaimed; |
1da177e4c Linux-2.6.12-rc2 |
1919 |
} |
f626012db mm: remove isolat... |
1920 |
static void shrink_active_list(unsigned long nr_to_scan, |
1a93be0e7 mm/vmscan: push l... |
1921 |
struct lruvec *lruvec, |
f16015fbf mm: vmscan: disti... |
1922 |
struct scan_control *sc, |
9e3b2f8cd mm/vmscan: store ... |
1923 |
enum lru_list lru) |
1da177e4c Linux-2.6.12-rc2 |
1924 |
{ |
44c241f16 mm: rename pgmove... |
1925 |
unsigned long nr_taken; |
f626012db mm: remove isolat... |
1926 |
unsigned long nr_scanned; |
6fe6b7e35 vmscan: report vm... |
1927 |
unsigned long vm_flags; |
1da177e4c Linux-2.6.12-rc2 |
1928 |
LIST_HEAD(l_hold); /* The pages which were snipped off */ |
8cab4754d vmscan: make mapp... |
1929 |
LIST_HEAD(l_active); |
b69408e88 vmscan: Use an in... |
1930 |
LIST_HEAD(l_inactive); |
1da177e4c Linux-2.6.12-rc2 |
1931 |
struct page *page; |
9d998b4f1 mm, vmscan: add a... |
1932 1933 |
unsigned nr_deactivate, nr_activate; unsigned nr_rotated = 0; |
3cb994517 mm: push lru inde... |
1934 |
int file = is_file_lru(lru); |
599d0c954 mm, vmscan: move ... |
1935 |
struct pglist_data *pgdat = lruvec_pgdat(lruvec); |
1da177e4c Linux-2.6.12-rc2 |
1936 1937 |
lru_add_drain(); |
f80c06736 mm: zone_reclaim:... |
1938 |
|
599d0c954 mm, vmscan: move ... |
1939 |
spin_lock_irq(&pgdat->lru_lock); |
925b7673c mm: make per-memc... |
1940 |
|
5dc35979e mm/vmscan: push l... |
1941 |
nr_taken = isolate_lru_pages(nr_to_scan, lruvec, &l_hold, |
a9e7c39fa mm/vmscan.c: remo... |
1942 |
&nr_scanned, sc, lru); |
89b5fae53 mm: vmscan: disti... |
1943 |
|
599d0c954 mm, vmscan: move ... |
1944 |
__mod_node_page_state(pgdat, NR_ISOLATED_ANON + file, nr_taken); |
1cfb419b3 per-zone and recl... |
1945 |
|
912c05720 mm: vmscan: consi... |
1946 1947 |
if (!cgroup_reclaim(sc)) __count_vm_events(PGREFILL, nr_scanned); |
2fa2690ca mm/vmscan.c: don'... |
1948 |
__count_memcg_events(lruvec_memcg(lruvec), PGREFILL, nr_scanned); |
9d5e6a9f2 mm: update_lru_si... |
1949 |
|
599d0c954 mm, vmscan: move ... |
1950 |
spin_unlock_irq(&pgdat->lru_lock); |
1da177e4c Linux-2.6.12-rc2 |
1951 |
|
1da177e4c Linux-2.6.12-rc2 |
1952 1953 1954 1955 |
while (!list_empty(&l_hold)) { cond_resched(); page = lru_to_page(&l_hold); list_del(&page->lru); |
7e9cd4842 vmscan: fix pagec... |
1956 |
|
39b5f29ac mm: remove vma ar... |
1957 |
if (unlikely(!page_evictable(page))) { |
894bc3104 Unevictable LRU I... |
1958 1959 1960 |
putback_lru_page(page); continue; } |
cc715d99e mm: vmscan: forci... |
1961 1962 1963 1964 1965 1966 1967 |
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 ... |
1968 1969 |
if (page_referenced(page, 0, sc->target_mem_cgroup, &vm_flags)) { |
8cab4754d vmscan: make mapp... |
1970 1971 1972 1973 1974 1975 1976 1977 1978 |
/* * 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. */ |
9de4f22a6 mm: code cleanup ... |
1979 |
if ((vm_flags & VM_EXEC) && page_is_file_lru(page)) { |
6c357848b mm: replace hpage... |
1980 |
nr_rotated += thp_nr_pages(page); |
8cab4754d vmscan: make mapp... |
1981 1982 1983 1984 |
list_add(&page->lru, &l_active); continue; } } |
7e9cd4842 vmscan: fix pagec... |
1985 |
|
5205e56ee vmscan: move Clea... |
1986 |
ClearPageActive(page); /* we are de-activating */ |
1899ad18c mm: workingset: t... |
1987 |
SetPageWorkingset(page); |
1da177e4c Linux-2.6.12-rc2 |
1988 1989 |
list_add(&page->lru, &l_inactive); } |
b555749aa vmscan: shrink_ac... |
1990 |
/* |
8cab4754d vmscan: make mapp... |
1991 |
* Move pages back to the lru list. |
b555749aa vmscan: shrink_ac... |
1992 |
*/ |
599d0c954 mm, vmscan: move ... |
1993 |
spin_lock_irq(&pgdat->lru_lock); |
556adecba vmscan: second ch... |
1994 |
|
a222f3415 mm: generalize pu... |
1995 1996 |
nr_activate = move_pages_to_lru(lruvec, &l_active); nr_deactivate = move_pages_to_lru(lruvec, &l_inactive); |
f372d89e5 mm: remove pages_... |
1997 1998 |
/* Keep all free pages in l_active list */ list_splice(&l_inactive, &l_active); |
9851ac135 mm: move nr_deact... |
1999 2000 2001 |
__count_vm_events(PGDEACTIVATE, nr_deactivate); __count_memcg_events(lruvec_memcg(lruvec), PGDEACTIVATE, nr_deactivate); |
599d0c954 mm, vmscan: move ... |
2002 2003 |
__mod_node_page_state(pgdat, NR_ISOLATED_ANON + file, -nr_taken); spin_unlock_irq(&pgdat->lru_lock); |
2bcf88796 mm: take pagevecs... |
2004 |
|
f372d89e5 mm: remove pages_... |
2005 2006 |
mem_cgroup_uncharge_list(&l_active); free_unref_page_list(&l_active); |
9d998b4f1 mm, vmscan: add a... |
2007 2008 |
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 |
2009 |
} |
1a4e58cce mm: introduce MAD... |
2010 2011 |
unsigned long reclaim_pages(struct list_head *page_list) { |
f661d007f mm: vmscan: repla... |
2012 |
int nid = NUMA_NO_NODE; |
730ec8c01 mm/vmscan.c: chan... |
2013 |
unsigned int nr_reclaimed = 0; |
1a4e58cce mm: introduce MAD... |
2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 |
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); |
f661d007f mm: vmscan: repla... |
2027 |
if (nid == NUMA_NO_NODE) { |
1a4e58cce mm: introduce MAD... |
2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 |
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), |
dd156e3fc mm/rmap: always d... |
2040 |
&sc, &dummy_stat, false); |
1a4e58cce mm: introduce MAD... |
2041 2042 2043 2044 2045 |
while (!list_empty(&node_page_list)) { page = lru_to_page(&node_page_list); list_del(&page->lru); putback_lru_page(page); } |
f661d007f mm: vmscan: repla... |
2046 |
nid = NUMA_NO_NODE; |
1a4e58cce mm: introduce MAD... |
2047 2048 2049 2050 2051 |
} if (!list_empty(&node_page_list)) { nr_reclaimed += shrink_page_list(&node_page_list, NODE_DATA(nid), |
dd156e3fc mm/rmap: always d... |
2052 |
&sc, &dummy_stat, false); |
1a4e58cce mm: introduce MAD... |
2053 2054 2055 2056 2057 2058 2059 2060 2061 |
while (!list_empty(&node_page_list)) { page = lru_to_page(&node_page_list); list_del(&page->lru); putback_lru_page(page); } } return nr_reclaimed; } |
b91ac3743 mm: vmscan: enfor... |
2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 |
static unsigned long shrink_list(enum lru_list lru, unsigned long nr_to_scan, struct lruvec *lruvec, struct scan_control *sc) { if (is_active_lru(lru)) { if (sc->may_deactivate & (1 << is_file_lru(lru))) shrink_active_list(nr_to_scan, lruvec, sc, lru); else sc->skipped_deactivate = 1; return 0; } return shrink_inactive_list(nr_to_scan, lruvec, sc, lru); } |
59dc76b0d mm: vmscan: reduc... |
2075 2076 2077 |
/* * The inactive anon list should be small enough that the VM never has * to do too much work. |
14797e236 memcg: add inacti... |
2078 |
* |
59dc76b0d mm: vmscan: reduc... |
2079 2080 2081 |
* 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... |
2082 |
* |
59dc76b0d mm: vmscan: reduc... |
2083 2084 |
* 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... |
2085 |
* |
2a2e48854 mm: vmscan: fix I... |
2086 2087 |
* If that fails and refaulting is observed, the inactive list grows. * |
59dc76b0d mm: vmscan: reduc... |
2088 |
* The inactive_ratio is the target ratio of ACTIVE to INACTIVE pages |
3a50d14d0 mm: remove unused... |
2089 |
* on this LRU, maintained by the pageout code. An inactive_ratio |
59dc76b0d mm: vmscan: reduc... |
2090 |
* of 3 means 3:1 or 25% of the pages are kept on the inactive list. |
56e49d218 vmscan: evict use... |
2091 |
* |
59dc76b0d mm: vmscan: reduc... |
2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 |
* 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... |
2102 |
*/ |
b91ac3743 mm: vmscan: enfor... |
2103 |
static bool inactive_is_low(struct lruvec *lruvec, enum lru_list inactive_lru) |
56e49d218 vmscan: evict use... |
2104 |
{ |
b91ac3743 mm: vmscan: enfor... |
2105 |
enum lru_list active_lru = inactive_lru + LRU_ACTIVE; |
2a2e48854 mm: vmscan: fix I... |
2106 2107 |
unsigned long inactive, active; unsigned long inactive_ratio; |
59dc76b0d mm: vmscan: reduc... |
2108 |
unsigned long gb; |
e3790144c mm: refactor inac... |
2109 |
|
b91ac3743 mm: vmscan: enfor... |
2110 2111 |
inactive = lruvec_page_state(lruvec, NR_LRU_BASE + inactive_lru); active = lruvec_page_state(lruvec, NR_LRU_BASE + active_lru); |
f8d1a3116 mm: consider whet... |
2112 |
|
b91ac3743 mm: vmscan: enfor... |
2113 |
gb = (inactive + active) >> (30 - PAGE_SHIFT); |
4002570c5 mm/vmscan: restor... |
2114 |
if (gb) |
b91ac3743 mm: vmscan: enfor... |
2115 2116 2117 |
inactive_ratio = int_sqrt(10 * gb); else inactive_ratio = 1; |
fd5388037 mm, vmscan: clean... |
2118 |
|
59dc76b0d mm: vmscan: reduc... |
2119 |
return inactive * inactive_ratio < active; |
b39415b27 vmscan: do not ev... |
2120 |
} |
9a2651140 mm: vmscan: clean... |
2121 2122 2123 2124 2125 2126 |
enum scan_balance { SCAN_EQUAL, SCAN_FRACT, SCAN_ANON, SCAN_FILE, }; |
4f98a2fee vmscan: split LRU... |
2127 2128 2129 2130 2131 2132 |
/* * 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... |
2133 2134 |
* 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... |
2135 |
*/ |
afaf07a65 mm: vmscan: turn ... |
2136 2137 |
static void get_scan_count(struct lruvec *lruvec, struct scan_control *sc, unsigned long *nr) |
4f98a2fee vmscan: split LRU... |
2138 |
{ |
afaf07a65 mm: vmscan: turn ... |
2139 |
struct mem_cgroup *memcg = lruvec_memcg(lruvec); |
d483a5dd0 mm: vmscan: limit... |
2140 |
unsigned long anon_cost, file_cost, total_cost; |
333776785 mm: vmscan: pass ... |
2141 |
int swappiness = mem_cgroup_swappiness(memcg); |
ed0173733 mm: use self-expl... |
2142 |
u64 fraction[ANON_AND_FILE]; |
9a2651140 mm: vmscan: clean... |
2143 |
u64 denominator = 0; /* gcc */ |
9a2651140 mm: vmscan: clean... |
2144 |
enum scan_balance scan_balance; |
4f98a2fee vmscan: split LRU... |
2145 |
unsigned long ap, fp; |
4111304da mm: enum lru_list... |
2146 |
enum lru_list lru; |
76a33fc38 vmscan: prevent g... |
2147 2148 |
/* If we have no swap space, do not bother scanning anon pages. */ |
d8b38438a mm: vmscan: do no... |
2149 |
if (!sc->may_swap || mem_cgroup_get_nr_swap_pages(memcg) <= 0) { |
9a2651140 mm: vmscan: clean... |
2150 |
scan_balance = SCAN_FILE; |
76a33fc38 vmscan: prevent g... |
2151 2152 |
goto out; } |
4f98a2fee vmscan: split LRU... |
2153 |
|
10316b313 mm: vmscan: clari... |
2154 2155 2156 2157 2158 2159 2160 |
/* * 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. */ |
b5ead35e7 mm: vmscan: namin... |
2161 |
if (cgroup_reclaim(sc) && !swappiness) { |
9a2651140 mm: vmscan: clean... |
2162 |
scan_balance = SCAN_FILE; |
10316b313 mm: vmscan: clari... |
2163 2164 2165 2166 2167 2168 2169 2170 |
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 ... |
2171 |
if (!sc->priority && swappiness) { |
9a2651140 mm: vmscan: clean... |
2172 |
scan_balance = SCAN_EQUAL; |
10316b313 mm: vmscan: clari... |
2173 2174 |
goto out; } |
11d16c25b mm: vmscan: impro... |
2175 |
/* |
53138cea7 mm: vmscan: move ... |
2176 |
* If the system is almost out of file pages, force-scan anon. |
623762517 revert "mm: vmsca... |
2177 |
*/ |
b91ac3743 mm: vmscan: enfor... |
2178 |
if (sc->file_is_tiny) { |
53138cea7 mm: vmscan: move ... |
2179 2180 |
scan_balance = SCAN_ANON; goto out; |
623762517 revert "mm: vmsca... |
2181 2182 2183 |
} /* |
b91ac3743 mm: vmscan: enfor... |
2184 2185 |
* If there is enough inactive page cache, we do not reclaim * anything from the anonymous working right now. |
7c5bd705d mm: memcg: only e... |
2186 |
*/ |
b91ac3743 mm: vmscan: enfor... |
2187 |
if (sc->cache_trim_mode) { |
9a2651140 mm: vmscan: clean... |
2188 |
scan_balance = SCAN_FILE; |
7c5bd705d mm: memcg: only e... |
2189 2190 |
goto out; } |
9a2651140 mm: vmscan: clean... |
2191 |
scan_balance = SCAN_FRACT; |
7c5bd705d mm: memcg: only e... |
2192 |
/* |
314b57fb0 mm: balance LRU l... |
2193 2194 2195 2196 2197 2198 2199 2200 |
* Calculate the pressure balance between anon and file pages. * * The amount of pressure we put on each LRU is inversely * proportional to the cost of reclaiming each list, as * determined by the share of pages that are refaulting, times * the relative IO cost of bringing back a swapped out * anonymous page vs reloading a filesystem page (swappiness). * |
d483a5dd0 mm: vmscan: limit... |
2201 2202 2203 2204 |
* Although we limit that influence to ensure no list gets * left behind completely: at least a third of the pressure is * applied, before swappiness. * |
314b57fb0 mm: balance LRU l... |
2205 |
* With swappiness at 100, anon and file have equal IO cost. |
58c37f6e0 vmscan: protect r... |
2206 |
*/ |
d483a5dd0 mm: vmscan: limit... |
2207 2208 2209 2210 |
total_cost = sc->anon_cost + sc->file_cost; anon_cost = total_cost + sc->anon_cost; file_cost = total_cost + sc->file_cost; total_cost = anon_cost + file_cost; |
58c37f6e0 vmscan: protect r... |
2211 |
|
d483a5dd0 mm: vmscan: limit... |
2212 2213 |
ap = swappiness * (total_cost + 1); ap /= anon_cost + 1; |
4f98a2fee vmscan: split LRU... |
2214 |
|
d483a5dd0 mm: vmscan: limit... |
2215 2216 |
fp = (200 - swappiness) * (total_cost + 1); fp /= file_cost + 1; |
4f98a2fee vmscan: split LRU... |
2217 |
|
76a33fc38 vmscan: prevent g... |
2218 2219 |
fraction[0] = ap; fraction[1] = fp; |
a4fe1631f mm: vmscan: drop ... |
2220 |
denominator = ap + fp; |
76a33fc38 vmscan: prevent g... |
2221 |
out: |
688035f72 mm: don't avoid h... |
2222 2223 |
for_each_evictable_lru(lru) { int file = is_file_lru(lru); |
9783aa991 mm, memcg: propor... |
2224 |
unsigned long lruvec_size; |
688035f72 mm: don't avoid h... |
2225 |
unsigned long scan; |
1bc63fb12 mm, memcg: make s... |
2226 |
unsigned long protection; |
9783aa991 mm, memcg: propor... |
2227 2228 |
lruvec_size = lruvec_lru_size(lruvec, lru, sc->reclaim_idx); |
22f7496f0 mm, memcg: avoid ... |
2229 2230 |
protection = mem_cgroup_protection(sc->target_mem_cgroup, memcg, |
1bc63fb12 mm, memcg: make s... |
2231 |
sc->memcg_low_reclaim); |
9783aa991 mm, memcg: propor... |
2232 |
|
1bc63fb12 mm, memcg: make s... |
2233 |
if (protection) { |
9783aa991 mm, memcg: propor... |
2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 |
/* * 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... |
2246 2247 2248 2249 |
* * 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... |
2250 |
* |
9de7ca46a mm, memcg: make m... |
2251 2252 2253 2254 2255 2256 2257 2258 |
* 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... |
2259 2260 2261 |
* 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... |
2262 |
*/ |
1bc63fb12 mm, memcg: make s... |
2263 2264 2265 2266 2267 2268 2269 |
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... |
2270 2271 |
/* |
1bc63fb12 mm, memcg: make s... |
2272 |
* Minimally target SWAP_CLUSTER_MAX pages to keep |
55b65a57c mm/vmsan: fix som... |
2273 |
* reclaim moving forwards, avoiding decrementing |
9de7ca46a mm, memcg: make m... |
2274 |
* sc->priority further than desirable. |
9783aa991 mm, memcg: propor... |
2275 |
*/ |
1bc63fb12 mm, memcg: make s... |
2276 |
scan = max(scan, SWAP_CLUSTER_MAX); |
9783aa991 mm, memcg: propor... |
2277 2278 2279 2280 2281 |
} else { scan = lruvec_size; } scan >>= sc->priority; |
6b4f7799c mm: vmscan: invok... |
2282 |
|
688035f72 mm: don't avoid h... |
2283 2284 2285 2286 2287 |
/* * 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... |
2288 |
scan = min(lruvec_size, SWAP_CLUSTER_MAX); |
6b4f7799c mm: vmscan: invok... |
2289 |
|
688035f72 mm: don't avoid h... |
2290 2291 2292 2293 2294 |
switch (scan_balance) { case SCAN_EQUAL: /* Scan lists relative to size */ break; case SCAN_FRACT: |
9a2651140 mm: vmscan: clean... |
2295 |
/* |
688035f72 mm: don't avoid h... |
2296 2297 |
* Scan types proportional to swappiness and * their relative recent reclaim efficiency. |
76073c646 mm/vmscan.c: don'... |
2298 2299 2300 |
* Make sure we don't miss the last page on * the offlined memory cgroups because of a * round-off error. |
9a2651140 mm: vmscan: clean... |
2301 |
*/ |
76073c646 mm/vmscan.c: don'... |
2302 2303 2304 |
scan = mem_cgroup_online(memcg) ? div64_u64(scan * fraction[file], denominator) : DIV64_U64_ROUND_UP(scan * fraction[file], |
68600f623 mm: don't miss th... |
2305 |
denominator); |
688035f72 mm: don't avoid h... |
2306 2307 2308 2309 |
break; case SCAN_FILE: case SCAN_ANON: /* Scan one type exclusively */ |
e072bff60 mm/vmscan.c: clea... |
2310 |
if ((scan_balance == SCAN_FILE) != file) |
688035f72 mm: don't avoid h... |
2311 |
scan = 0; |
688035f72 mm: don't avoid h... |
2312 2313 2314 2315 |
break; default: /* Look ma, no brain */ BUG(); |
9a2651140 mm: vmscan: clean... |
2316 |
} |
688035f72 mm: don't avoid h... |
2317 |
|
688035f72 mm: don't avoid h... |
2318 |
nr[lru] = scan; |
76a33fc38 vmscan: prevent g... |
2319 |
} |
6e08a369e vmscan: cleanup t... |
2320 |
} |
4f98a2fee vmscan: split LRU... |
2321 |
|
afaf07a65 mm: vmscan: turn ... |
2322 |
static void shrink_lruvec(struct lruvec *lruvec, struct scan_control *sc) |
9b4f98cda mm: vmscan: compa... |
2323 2324 |
{ unsigned long nr[NR_LRU_LISTS]; |
e82e0561d mm: vmscan: obey ... |
2325 |
unsigned long targets[NR_LRU_LISTS]; |
9b4f98cda mm: vmscan: compa... |
2326 2327 2328 2329 2330 |
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... |
2331 |
bool scan_adjusted; |
9b4f98cda mm: vmscan: compa... |
2332 |
|
afaf07a65 mm: vmscan: turn ... |
2333 |
get_scan_count(lruvec, sc, nr); |
9b4f98cda mm: vmscan: compa... |
2334 |
|
e82e0561d mm: vmscan: obey ... |
2335 2336 |
/* Record the original scan target for proportional adjustments later */ memcpy(targets, nr, sizeof(nr)); |
1a501907b mm: vmscan: use p... |
2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 |
/* * 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. */ |
b5ead35e7 mm: vmscan: namin... |
2348 |
scan_adjusted = (!cgroup_reclaim(sc) && !current_is_kswapd() && |
1a501907b mm: vmscan: use p... |
2349 |
sc->priority == DEF_PRIORITY); |
9b4f98cda mm: vmscan: compa... |
2350 2351 2352 |
blk_start_plug(&plug); while (nr[LRU_INACTIVE_ANON] || nr[LRU_ACTIVE_FILE] || nr[LRU_INACTIVE_FILE]) { |
e82e0561d mm: vmscan: obey ... |
2353 2354 |
unsigned long nr_anon, nr_file, percentage; unsigned long nr_scanned; |
9b4f98cda mm: vmscan: compa... |
2355 2356 2357 2358 2359 2360 |
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 ... |
2361 |
lruvec, sc); |
9b4f98cda mm: vmscan: compa... |
2362 2363 |
} } |
e82e0561d mm: vmscan: obey ... |
2364 |
|
bd041733c mm, vmscan: add c... |
2365 |
cond_resched(); |
e82e0561d mm: vmscan: obey ... |
2366 2367 |
if (nr_reclaimed < nr_to_reclaim || scan_adjusted) continue; |
9b4f98cda mm: vmscan: compa... |
2368 |
/* |
e82e0561d mm: vmscan: obey ... |
2369 |
* For kswapd and memcg, reclaim at least the number of pages |
1a501907b mm: vmscan: use p... |
2370 |
* requested. Ensure that the anon and file LRUs are scanned |
e82e0561d mm: vmscan: obey ... |
2371 2372 2373 2374 2375 2376 |
* 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... |
2377 2378 2379 2380 2381 2382 2383 2384 |
/* * 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 ... |
2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 |
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... |
2416 2417 2418 2419 2420 2421 2422 2423 |
} 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. */ |
b91ac3743 mm: vmscan: enfor... |
2424 |
if (total_swap_pages && inactive_is_low(lruvec, LRU_INACTIVE_ANON)) |
9b4f98cda mm: vmscan: compa... |
2425 2426 |
shrink_active_list(SWAP_CLUSTER_MAX, lruvec, sc, LRU_ACTIVE_ANON); |
9b4f98cda mm: vmscan: compa... |
2427 |
} |
23b9da55c mm: vmscan: remov... |
2428 |
/* Use reclaim/compaction for costly allocs or under memory pressure */ |
9e3b2f8cd mm/vmscan: store ... |
2429 |
static bool in_reclaim_compaction(struct scan_control *sc) |
23b9da55c mm: vmscan: remov... |
2430 |
{ |
d84da3f9e mm: use IS_ENABLE... |
2431 |
if (IS_ENABLED(CONFIG_COMPACTION) && sc->order && |
23b9da55c mm: vmscan: remov... |
2432 |
(sc->order > PAGE_ALLOC_COSTLY_ORDER || |
9e3b2f8cd mm/vmscan: store ... |
2433 |
sc->priority < DEF_PRIORITY - 2)) |
23b9da55c mm: vmscan: remov... |
2434 2435 2436 2437 |
return true; return false; } |
1da177e4c Linux-2.6.12-rc2 |
2438 |
/* |
23b9da55c mm: vmscan: remov... |
2439 2440 2441 |
* 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 |
df3a45f9d mm/vmscan: update... |
2442 |
* calls try_to_compact_pages() that it will have enough free pages to succeed. |
23b9da55c mm: vmscan: remov... |
2443 |
* It will give up earlier than that if there is difficulty reclaiming pages. |
3e7d34497 mm: vmscan: recla... |
2444 |
*/ |
a9dd0a831 mm, vmscan: make ... |
2445 |
static inline bool should_continue_reclaim(struct pglist_data *pgdat, |
3e7d34497 mm: vmscan: recla... |
2446 |
unsigned long nr_reclaimed, |
3e7d34497 mm: vmscan: recla... |
2447 2448 2449 2450 |
struct scan_control *sc) { unsigned long pages_for_compaction; unsigned long inactive_lru_pages; |
a9dd0a831 mm, vmscan: make ... |
2451 |
int z; |
3e7d34497 mm: vmscan: recla... |
2452 2453 |
/* If not in reclaim/compaction mode, stop */ |
9e3b2f8cd mm/vmscan: store ... |
2454 |
if (!in_reclaim_compaction(sc)) |
3e7d34497 mm: vmscan: recla... |
2455 |
return false; |
5ee04716c mm, reclaim: clea... |
2456 2457 2458 2459 2460 2461 2462 2463 2464 2465 2466 2467 |
/* * 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... |
2468 |
|
3e7d34497 mm: vmscan: recla... |
2469 |
/* If compaction would go ahead or the allocation would succeed, stop */ |
a9dd0a831 mm, vmscan: make ... |
2470 2471 |
for (z = 0; z <= sc->reclaim_idx; z++) { struct zone *zone = &pgdat->node_zones[z]; |
6aa303def mm, vmscan: only ... |
2472 |
if (!managed_zone(zone)) |
a9dd0a831 mm, vmscan: make ... |
2473 2474 2475 |
continue; switch (compaction_suitable(zone, sc->order, 0, sc->reclaim_idx)) { |
cf378319d mm, compaction: r... |
2476 |
case COMPACT_SUCCESS: |
a9dd0a831 mm, vmscan: make ... |
2477 2478 2479 2480 2481 2482 |
case COMPACT_CONTINUE: return false; default: /* check next zone */ ; } |
3e7d34497 mm: vmscan: recla... |
2483 |
} |
1c6c15971 mm, reclaim: make... |
2484 2485 2486 2487 2488 2489 2490 2491 2492 |
/* * 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... |
2493 |
return inactive_lru_pages > pages_for_compaction; |
3e7d34497 mm: vmscan: recla... |
2494 |
} |
0f6a5cff4 mm: vmscan: split... |
2495 |
static void shrink_node_memcgs(pg_data_t *pgdat, struct scan_control *sc) |
1da177e4c Linux-2.6.12-rc2 |
2496 |
{ |
0f6a5cff4 mm: vmscan: split... |
2497 |
struct mem_cgroup *target_memcg = sc->target_mem_cgroup; |
d2af33970 mm: vmscan: repla... |
2498 |
struct mem_cgroup *memcg; |
1da177e4c Linux-2.6.12-rc2 |
2499 |
|
0f6a5cff4 mm: vmscan: split... |
2500 |
memcg = mem_cgroup_iter(target_memcg, NULL, NULL); |
d2af33970 mm: vmscan: repla... |
2501 |
do { |
afaf07a65 mm: vmscan: turn ... |
2502 |
struct lruvec *lruvec = mem_cgroup_lruvec(memcg, pgdat); |
d2af33970 mm: vmscan: repla... |
2503 2504 |
unsigned long reclaimed; unsigned long scanned; |
5660048cc mm: move memcg hi... |
2505 |
|
e3336cab2 mm: memcg: fix me... |
2506 2507 2508 2509 2510 2511 2512 |
/* * 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(); |
45c7f7e1e mm, memcg: decoup... |
2513 2514 2515 |
mem_cgroup_calculate_protection(target_memcg, memcg); if (mem_cgroup_below_min(memcg)) { |
d2af33970 mm: vmscan: repla... |
2516 2517 2518 2519 2520 |
/* * Hard protection. * If there is no reclaimable memory, OOM. */ continue; |
45c7f7e1e mm, memcg: decoup... |
2521 |
} else if (mem_cgroup_below_low(memcg)) { |
d2af33970 mm: vmscan: repla... |
2522 2523 2524 2525 2526 2527 2528 2529 |
/* * Soft protection. * Respect the protection only as long as * there is an unprotected supply * of reclaimable memory from other cgroups. */ if (!sc->memcg_low_reclaim) { sc->memcg_low_skipped = 1; |
bf8d5d52f memcg: introduce ... |
2530 |
continue; |
241994ed8 mm: memcontrol: d... |
2531 |
} |
d2af33970 mm: vmscan: repla... |
2532 |
memcg_memory_event(memcg, MEMCG_LOW); |
d2af33970 mm: vmscan: repla... |
2533 |
} |
241994ed8 mm: memcontrol: d... |
2534 |
|
d2af33970 mm: vmscan: repla... |
2535 2536 |
reclaimed = sc->nr_reclaimed; scanned = sc->nr_scanned; |
afaf07a65 mm: vmscan: turn ... |
2537 2538 |
shrink_lruvec(lruvec, sc); |
70ddf637e memcg: add memory... |
2539 |
|
d2af33970 mm: vmscan: repla... |
2540 2541 |
shrink_slab(sc->gfp_mask, pgdat->node_id, memcg, sc->priority); |
6b4f7799c mm: vmscan: invok... |
2542 |
|
d2af33970 mm: vmscan: repla... |
2543 2544 2545 2546 |
/* Record the group's reclaim efficiency */ vmpressure(sc->gfp_mask, memcg, false, sc->nr_scanned - scanned, sc->nr_reclaimed - reclaimed); |
70ddf637e memcg: add memory... |
2547 |
|
0f6a5cff4 mm: vmscan: split... |
2548 2549 |
} while ((memcg = mem_cgroup_iter(target_memcg, memcg, NULL))); } |
6c9e0907f mm/vmscan.c: remo... |
2550 |
static void shrink_node(pg_data_t *pgdat, struct scan_control *sc) |
0f6a5cff4 mm: vmscan: split... |
2551 2552 |
{ struct reclaim_state *reclaim_state = current->reclaim_state; |
0f6a5cff4 mm: vmscan: split... |
2553 |
unsigned long nr_reclaimed, nr_scanned; |
1b05117df mm: vmscan: harmo... |
2554 |
struct lruvec *target_lruvec; |
0f6a5cff4 mm: vmscan: split... |
2555 |
bool reclaimable = false; |
b91ac3743 mm: vmscan: enfor... |
2556 |
unsigned long file; |
0f6a5cff4 mm: vmscan: split... |
2557 |
|
1b05117df mm: vmscan: harmo... |
2558 |
target_lruvec = mem_cgroup_lruvec(sc->target_mem_cgroup, pgdat); |
0f6a5cff4 mm: vmscan: split... |
2559 2560 2561 2562 2563 |
again: memset(&sc->nr, 0, sizeof(sc->nr)); nr_reclaimed = sc->nr_reclaimed; nr_scanned = sc->nr_scanned; |
53138cea7 mm: vmscan: move ... |
2564 |
/* |
7cf111bc3 mm: vmscan: deter... |
2565 2566 2567 2568 2569 2570 2571 2572 |
* Determine the scan balance between anon and file LRUs. */ spin_lock_irq(&pgdat->lru_lock); sc->anon_cost = target_lruvec->anon_cost; sc->file_cost = target_lruvec->file_cost; spin_unlock_irq(&pgdat->lru_lock); /* |
b91ac3743 mm: vmscan: enfor... |
2573 2574 2575 2576 2577 |
* Target desirable inactive:active list ratios for the anon * and file LRU lists. */ if (!sc->force_deactivate) { unsigned long refaults; |
170b04b7a mm/workingset: pr... |
2578 2579 2580 2581 |
refaults = lruvec_page_state(target_lruvec, WORKINGSET_ACTIVATE_ANON); if (refaults != target_lruvec->refaults[0] || inactive_is_low(target_lruvec, LRU_INACTIVE_ANON)) |
b91ac3743 mm: vmscan: enfor... |
2582 2583 2584 2585 2586 2587 2588 2589 2590 2591 |
sc->may_deactivate |= DEACTIVATE_ANON; else sc->may_deactivate &= ~DEACTIVATE_ANON; /* * When refaults are being observed, it means a new * workingset is being established. Deactivate to get * rid of any stale active pages quickly. */ refaults = lruvec_page_state(target_lruvec, |
170b04b7a mm/workingset: pr... |
2592 2593 |
WORKINGSET_ACTIVATE_FILE); if (refaults != target_lruvec->refaults[1] || |
b91ac3743 mm: vmscan: enfor... |
2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 |
inactive_is_low(target_lruvec, LRU_INACTIVE_FILE)) sc->may_deactivate |= DEACTIVATE_FILE; else sc->may_deactivate &= ~DEACTIVATE_FILE; } else sc->may_deactivate = DEACTIVATE_ANON | DEACTIVATE_FILE; /* * If we have plenty of inactive file pages that aren't * thrashing, try to reclaim those first before touching * anonymous pages. */ file = lruvec_page_state(target_lruvec, NR_INACTIVE_FILE); if (file >> sc->priority && !(sc->may_deactivate & DEACTIVATE_FILE)) sc->cache_trim_mode = 1; else sc->cache_trim_mode = 0; /* |
53138cea7 mm: vmscan: move ... |
2613 2614 2615 2616 2617 2618 2619 2620 2621 |
* 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 (!cgroup_reclaim(sc)) { |
53138cea7 mm: vmscan: move ... |
2622 |
unsigned long total_high_wmark = 0; |
b91ac3743 mm: vmscan: enfor... |
2623 2624 |
unsigned long free, anon; int z; |
53138cea7 mm: vmscan: move ... |
2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 |
free = sum_zone_node_page_state(pgdat->node_id, NR_FREE_PAGES); file = 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]; if (!managed_zone(zone)) continue; total_high_wmark += high_wmark_pages(zone); } |
b91ac3743 mm: vmscan: enfor... |
2637 2638 2639 2640 2641 2642 2643 2644 2645 2646 2647 |
/* * Consider anon: if that's low too, this isn't a * runaway file reclaim problem, but rather just * extreme pressure. Reclaim as per usual then. */ anon = node_page_state(pgdat, NR_INACTIVE_ANON); sc->file_is_tiny = file + free <= total_high_wmark && !(sc->may_deactivate & DEACTIVATE_ANON) && anon >> sc->priority; |
53138cea7 mm: vmscan: move ... |
2648 |
} |
0f6a5cff4 mm: vmscan: split... |
2649 |
shrink_node_memcgs(pgdat, sc); |
2344d7e44 mm: vmscan: remov... |
2650 |
|
d2af33970 mm: vmscan: repla... |
2651 2652 2653 2654 |
if (reclaim_state) { sc->nr_reclaimed += reclaim_state->reclaimed_slab; reclaim_state->reclaimed_slab = 0; } |
d108c7721 mm/vmscan: don't ... |
2655 |
|
d2af33970 mm: vmscan: repla... |
2656 |
/* Record the subtree's reclaim efficiency */ |
1b05117df mm: vmscan: harmo... |
2657 |
vmpressure(sc->gfp_mask, sc->target_mem_cgroup, true, |
d2af33970 mm: vmscan: repla... |
2658 2659 |
sc->nr_scanned - nr_scanned, sc->nr_reclaimed - nr_reclaimed); |
d108c7721 mm/vmscan: don't ... |
2660 |
|
d2af33970 mm: vmscan: repla... |
2661 2662 |
if (sc->nr_reclaimed - nr_reclaimed) reclaimable = true; |
d108c7721 mm/vmscan: don't ... |
2663 |
|
d2af33970 mm: vmscan: repla... |
2664 2665 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 |
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 ... |
2684 |
|
d2af33970 mm: vmscan: repla... |
2685 2686 2687 |
/* Allow kswapd to start writing pages during reclaim.*/ if (sc->nr.unqueued_dirty == sc->nr.file_taken) set_bit(PGDAT_DIRTY, &pgdat->flags); |
e3c1ac586 mm/vmscan: don't ... |
2688 2689 |
/* |
1eba09c15 mm/vmscan.c: dele... |
2690 |
* If kswapd scans pages marked for immediate |
d2af33970 mm: vmscan: repla... |
2691 2692 2693 |
* reclaim and under writeback (nr_immediate), it * implies that pages are cycling through the LRU * faster than they are written so also forcibly stall. |
d108c7721 mm/vmscan: don't ... |
2694 |
*/ |
d2af33970 mm: vmscan: repla... |
2695 2696 2697 2698 2699 |
if (sc->nr.immediate) congestion_wait(BLK_RW_ASYNC, HZ/10); } /* |
1b05117df mm: vmscan: harmo... |
2700 2701 2702 2703 |
* Tag a node/memcg as congested if all the dirty pages * scanned were backed by a congested BDI and * wait_iff_congested will stall. * |
d2af33970 mm: vmscan: repla... |
2704 2705 2706 |
* Legacy memcg will stall in page writeback so avoid forcibly * stalling in wait_iff_congested(). */ |
1b05117df mm: vmscan: harmo... |
2707 2708 |
if ((current_is_kswapd() || (cgroup_reclaim(sc) && writeback_throttling_sane(sc))) && |
d2af33970 mm: vmscan: repla... |
2709 |
sc->nr.dirty && sc->nr.dirty == sc->nr.congested) |
1b05117df mm: vmscan: harmo... |
2710 |
set_bit(LRUVEC_CONGESTED, &target_lruvec->flags); |
d2af33970 mm: vmscan: repla... |
2711 2712 2713 2714 2715 2716 2717 |
/* * 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. */ |
1b05117df mm: vmscan: harmo... |
2718 2719 2720 |
if (!current_is_kswapd() && current_may_throttle() && !sc->hibernation_mode && test_bit(LRUVEC_CONGESTED, &target_lruvec->flags)) |
d2af33970 mm: vmscan: repla... |
2721 |
wait_iff_congested(BLK_RW_ASYNC, HZ/10); |
d108c7721 mm/vmscan: don't ... |
2722 |
|
d2af33970 mm: vmscan: repla... |
2723 2724 2725 |
if (should_continue_reclaim(pgdat, sc->nr_reclaimed - nr_reclaimed, sc)) goto again; |
2344d7e44 mm: vmscan: remov... |
2726 |
|
c73322d09 mm: fix 100% CPU ... |
2727 2728 2729 2730 2731 2732 2733 2734 |
/* * 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; |
f16015fbf mm: vmscan: disti... |
2735 |
} |
53853e2d2 mm, compaction: d... |
2736 |
/* |
fdd4c6149 mm, vmscan: make ... |
2737 2738 2739 |
* 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... |
2740 |
*/ |
4f588331b mm, vmscan: avoid... |
2741 |
static inline bool compaction_ready(struct zone *zone, struct scan_control *sc) |
fe4b1b244 mm: vmscan: when ... |
2742 |
{ |
31483b6ad mm, vmscan: remov... |
2743 |
unsigned long watermark; |
fdd4c6149 mm, vmscan: make ... |
2744 |
enum compact_result suitable; |
fe4b1b244 mm: vmscan: when ... |
2745 |
|
fdd4c6149 mm, vmscan: make ... |
2746 2747 2748 2749 2750 2751 2752 |
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 ... |
2753 |
|
53853e2d2 mm, compaction: d... |
2754 |
/* |
fdd4c6149 mm, vmscan: make ... |
2755 2756 2757 2758 2759 2760 2761 |
* 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... |
2762 |
*/ |
fdd4c6149 mm, vmscan: make ... |
2763 |
watermark = high_wmark_pages(zone) + compact_gap(sc->order); |
fe4b1b244 mm: vmscan: when ... |
2764 |
|
fdd4c6149 mm, vmscan: make ... |
2765 |
return zone_watermark_ok_safe(zone, 0, watermark, sc->reclaim_idx); |
fe4b1b244 mm: vmscan: when ... |
2766 |
} |
1da177e4c Linux-2.6.12-rc2 |
2767 2768 2769 2770 2771 |
/* * 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 |
2772 2773 2774 |
* 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... |
2775 |
static void shrink_zones(struct zonelist *zonelist, struct scan_control *sc) |
1da177e4c Linux-2.6.12-rc2 |
2776 |
{ |
dd1a239f6 mm: have zonelist... |
2777 |
struct zoneref *z; |
54a6eb5c4 mm: use two zonel... |
2778 |
struct zone *zone; |
0608f43da revert "memcg, vm... |
2779 2780 |
unsigned long nr_soft_reclaimed; unsigned long nr_soft_scanned; |
619d0d76c mm/vmscan: restor... |
2781 |
gfp_t orig_mask; |
79dafcdca mm, vmscan: by de... |
2782 |
pg_data_t *last_pgdat = NULL; |
1cfb419b3 per-zone and recl... |
2783 |
|
cc715d99e mm: vmscan: forci... |
2784 2785 2786 2787 2788 |
/* * 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... |
2789 |
orig_mask = sc->gfp_mask; |
b2e18757f mm, vmscan: begin... |
2790 |
if (buffer_heads_over_limit) { |
cc715d99e mm: vmscan: forci... |
2791 |
sc->gfp_mask |= __GFP_HIGHMEM; |
4f588331b mm, vmscan: avoid... |
2792 |
sc->reclaim_idx = gfp_zone(sc->gfp_mask); |
b2e18757f mm, vmscan: begin... |
2793 |
} |
cc715d99e mm: vmscan: forci... |
2794 |
|
d4debc66d vmscan: remove un... |
2795 |
for_each_zone_zonelist_nodemask(zone, z, zonelist, |
b2e18757f mm, vmscan: begin... |
2796 |
sc->reclaim_idx, sc->nodemask) { |
b2e18757f mm, vmscan: begin... |
2797 |
/* |
1cfb419b3 per-zone and recl... |
2798 2799 2800 |
* Take care memory controller reclaiming has small influence * to global LRU. */ |
b5ead35e7 mm: vmscan: namin... |
2801 |
if (!cgroup_reclaim(sc)) { |
344736f29 cpuset: simplify ... |
2802 2803 |
if (!cpuset_zone_allowed(zone, GFP_KERNEL | __GFP_HARDWALL)) |
1cfb419b3 per-zone and recl... |
2804 |
continue; |
65ec02cb9 mm: vmscan: move ... |
2805 |
|
0b06496a3 mm: vmscan: rewor... |
2806 2807 2808 2809 2810 2811 2812 2813 2814 2815 2816 |
/* * 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... |
2817 |
compaction_ready(zone, sc)) { |
0b06496a3 mm: vmscan: rewor... |
2818 2819 |
sc->compaction_ready = true; continue; |
e0887c19b vmscan: limit dir... |
2820 |
} |
0b06496a3 mm: vmscan: rewor... |
2821 |
|
0608f43da revert "memcg, vm... |
2822 |
/* |
79dafcdca mm, vmscan: by de... |
2823 2824 2825 2826 2827 2828 2829 2830 2831 |
* 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... |
2832 2833 2834 2835 2836 2837 |
* 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... |
2838 |
nr_soft_reclaimed = mem_cgroup_soft_limit_reclaim(zone->zone_pgdat, |
0608f43da revert "memcg, vm... |
2839 2840 2841 2842 |
sc->order, sc->gfp_mask, &nr_soft_scanned); sc->nr_reclaimed += nr_soft_reclaimed; sc->nr_scanned += nr_soft_scanned; |
ac34a1a3c memcg: fix direct... |
2843 |
/* need some check for avoid more shrink_zone() */ |
1cfb419b3 per-zone and recl... |
2844 |
} |
408d85441 [PATCH] oom: use ... |
2845 |
|
79dafcdca mm, vmscan: by de... |
2846 2847 2848 2849 |
/* 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... |
2850 |
shrink_node(zone->zone_pgdat, sc); |
1da177e4c Linux-2.6.12-rc2 |
2851 |
} |
e0c23279c vmscan: abort rec... |
2852 |
|
65ec02cb9 mm: vmscan: move ... |
2853 |
/* |
619d0d76c mm/vmscan: restor... |
2854 2855 2856 2857 |
* 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 |
2858 |
} |
4f98a2fee vmscan: split LRU... |
2859 |
|
b910718a9 mm: vmscan: detec... |
2860 |
static void snapshot_refaults(struct mem_cgroup *target_memcg, pg_data_t *pgdat) |
2a2e48854 mm: vmscan: fix I... |
2861 |
{ |
b910718a9 mm: vmscan: detec... |
2862 2863 |
struct lruvec *target_lruvec; unsigned long refaults; |
2a2e48854 mm: vmscan: fix I... |
2864 |
|
b910718a9 mm: vmscan: detec... |
2865 |
target_lruvec = mem_cgroup_lruvec(target_memcg, pgdat); |
170b04b7a mm/workingset: pr... |
2866 2867 2868 2869 |
refaults = lruvec_page_state(target_lruvec, WORKINGSET_ACTIVATE_ANON); target_lruvec->refaults[0] = refaults; refaults = lruvec_page_state(target_lruvec, WORKINGSET_ACTIVATE_FILE); target_lruvec->refaults[1] = refaults; |
2a2e48854 mm: vmscan: fix I... |
2870 |
} |
1da177e4c Linux-2.6.12-rc2 |
2871 2872 2873 2874 2875 2876 2877 2878 |
/* * 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... |
2879 2880 2881 2882 |
* 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... |
2883 2884 2885 |
* * returns: 0, if no pages reclaimed * else, the number of pages reclaimed |
1da177e4c Linux-2.6.12-rc2 |
2886 |
*/ |
dac1d27bc mm: use zonelists... |
2887 |
static unsigned long do_try_to_free_pages(struct zonelist *zonelist, |
3115cd914 mm: vmscan: remov... |
2888 |
struct scan_control *sc) |
1da177e4c Linux-2.6.12-rc2 |
2889 |
{ |
241994ed8 mm: memcontrol: d... |
2890 |
int initial_priority = sc->priority; |
2a2e48854 mm: vmscan: fix I... |
2891 2892 2893 |
pg_data_t *last_pgdat; struct zoneref *z; struct zone *zone; |
241994ed8 mm: memcontrol: d... |
2894 |
retry: |
873b47717 per-task-delay-ac... |
2895 |
delayacct_freepages_start(); |
b5ead35e7 mm: vmscan: namin... |
2896 |
if (!cgroup_reclaim(sc)) |
7cc30fcfd mm: vmstat: accou... |
2897 |
__count_zid_vm_events(ALLOCSTALL, sc->reclaim_idx, 1); |
1da177e4c Linux-2.6.12-rc2 |
2898 |
|
9e3b2f8cd mm/vmscan: store ... |
2899 |
do { |
70ddf637e memcg: add memory... |
2900 2901 |
vmpressure_prio(sc->gfp_mask, sc->target_mem_cgroup, sc->priority); |
66e1707bc Memory controller... |
2902 |
sc->nr_scanned = 0; |
0a0337e0d mm, oom: rework o... |
2903 |
shrink_zones(zonelist, sc); |
c6a8a8c58 vmscan: recalcula... |
2904 |
|
bb21c7ce1 vmscan: fix do_tr... |
2905 |
if (sc->nr_reclaimed >= sc->nr_to_reclaim) |
0b06496a3 mm: vmscan: rewor... |
2906 2907 2908 2909 |
break; if (sc->compaction_ready) break; |
1da177e4c Linux-2.6.12-rc2 |
2910 2911 |
/* |
0e50ce3b5 mm: use up free s... |
2912 2913 2914 2915 2916 |
* 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... |
2917 |
} while (--sc->priority >= 0); |
bb21c7ce1 vmscan: fix do_tr... |
2918 |
|
2a2e48854 mm: vmscan: fix I... |
2919 2920 2921 2922 2923 2924 |
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; |
1b05117df mm: vmscan: harmo... |
2925 |
|
2a2e48854 mm: vmscan: fix I... |
2926 |
snapshot_refaults(sc->target_mem_cgroup, zone->zone_pgdat); |
1b05117df mm: vmscan: harmo... |
2927 2928 2929 2930 2931 2932 2933 2934 |
if (cgroup_reclaim(sc)) { struct lruvec *lruvec; lruvec = mem_cgroup_lruvec(sc->target_mem_cgroup, zone->zone_pgdat); clear_bit(LRUVEC_CONGESTED, &lruvec->flags); } |
2a2e48854 mm: vmscan: fix I... |
2935 |
} |
873b47717 per-task-delay-ac... |
2936 |
delayacct_freepages_end(); |
bb21c7ce1 vmscan: fix do_tr... |
2937 2938 |
if (sc->nr_reclaimed) return sc->nr_reclaimed; |
0cee34fd7 mm: vmscan: check... |
2939 |
/* Aborted reclaim to try compaction? don't OOM, then */ |
0b06496a3 mm: vmscan: rewor... |
2940 |
if (sc->compaction_ready) |
7335084d4 mm: vmscan: do no... |
2941 |
return 1; |
b91ac3743 mm: vmscan: enfor... |
2942 2943 2944 2945 2946 2947 2948 2949 2950 2951 2952 2953 2954 2955 2956 |
/* * We make inactive:active ratio decisions based on the node's * composition of memory, but a restrictive reclaim_idx or a * memory.low cgroup setting can exempt large amounts of * memory from reclaim. Neither of which are very common, so * instead of doing costly eligibility calculations of the * entire cgroup subtree up front, we assume the estimates are * good, and retry with forcible deactivation if that fails. */ if (sc->skipped_deactivate) { sc->priority = initial_priority; sc->force_deactivate = 1; sc->skipped_deactivate = 0; goto retry; } |
241994ed8 mm: memcontrol: d... |
2957 |
/* Untapped cgroup reserves? Don't OOM, retry. */ |
d6622f636 mm/vmscan: more r... |
2958 |
if (sc->memcg_low_skipped) { |
241994ed8 mm: memcontrol: d... |
2959 |
sc->priority = initial_priority; |
b91ac3743 mm: vmscan: enfor... |
2960 |
sc->force_deactivate = 0; |
d6622f636 mm/vmscan: more r... |
2961 2962 |
sc->memcg_low_reclaim = 1; sc->memcg_low_skipped = 0; |
241994ed8 mm: memcontrol: d... |
2963 2964 |
goto retry; } |
bb21c7ce1 vmscan: fix do_tr... |
2965 |
return 0; |
1da177e4c Linux-2.6.12-rc2 |
2966 |
} |
c73322d09 mm: fix 100% CPU ... |
2967 |
static bool allow_direct_reclaim(pg_data_t *pgdat) |
5515061d2 mm: throttle dire... |
2968 2969 2970 2971 2972 2973 |
{ struct zone *zone; unsigned long pfmemalloc_reserve = 0; unsigned long free_pages = 0; int i; bool wmark_ok; |
c73322d09 mm: fix 100% CPU ... |
2974 2975 |
if (pgdat->kswapd_failures >= MAX_RECLAIM_RETRIES) return true; |
5515061d2 mm: throttle dire... |
2976 2977 |
for (i = 0; i <= ZONE_NORMAL; i++) { zone = &pgdat->node_zones[i]; |
d450abd81 mm: fix check for... |
2978 2979 2980 2981 |
if (!managed_zone(zone)) continue; if (!zone_reclaimable_pages(zone)) |
675becce1 mm: vmscan: do no... |
2982 |
continue; |
5515061d2 mm: throttle dire... |
2983 2984 2985 |
pfmemalloc_reserve += min_wmark_pages(zone); free_pages += zone_page_state(zone, NR_FREE_PAGES); } |
675becce1 mm: vmscan: do no... |
2986 2987 2988 |
/* If there are no reserves (unexpected config) then do not throttle */ if (!pfmemalloc_reserve) return true; |
5515061d2 mm: throttle dire... |
2989 2990 2991 2992 |
wmark_ok = free_pages > pfmemalloc_reserve / 2; /* kswapd must be awake if processes are being throttled */ if (!wmark_ok && waitqueue_active(&pgdat->kswapd_wait)) { |
97a225e69 mm/page_alloc: in... |
2993 2994 |
if (READ_ONCE(pgdat->kswapd_highest_zoneidx) > ZONE_NORMAL) WRITE_ONCE(pgdat->kswapd_highest_zoneidx, ZONE_NORMAL); |
5644e1fbb mm/vmscan.c: fix ... |
2995 |
|
5515061d2 mm: throttle dire... |
2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 |
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... |
3006 3007 3008 3009 |
* 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... |
3010 |
*/ |
50694c28f mm: vmscan: check... |
3011 |
static bool throttle_direct_reclaim(gfp_t gfp_mask, struct zonelist *zonelist, |
5515061d2 mm: throttle dire... |
3012 3013 |
nodemask_t *nodemask) { |
675becce1 mm: vmscan: do no... |
3014 |
struct zoneref *z; |
5515061d2 mm: throttle dire... |
3015 |
struct zone *zone; |
675becce1 mm: vmscan: do no... |
3016 |
pg_data_t *pgdat = NULL; |
5515061d2 mm: throttle dire... |
3017 3018 3019 3020 3021 3022 3023 3024 3025 |
/* * 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... |
3026 3027 3028 3029 3030 3031 3032 3033 |
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... |
3034 |
|
675becce1 mm: vmscan: do no... |
3035 3036 3037 3038 3039 3040 3041 3042 3043 3044 3045 3046 3047 3048 3049 |
/* * 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... |
3050 |
gfp_zone(gfp_mask), nodemask) { |
675becce1 mm: vmscan: do no... |
3051 3052 3053 3054 3055 |
if (zone_idx(zone) > ZONE_NORMAL) continue; /* Throttle based on the first usable node */ pgdat = zone->zone_pgdat; |
c73322d09 mm: fix 100% CPU ... |
3056 |
if (allow_direct_reclaim(pgdat)) |
675becce1 mm: vmscan: do no... |
3057 3058 3059 3060 3061 3062 |
goto out; break; } /* If no zone was usable by the allocation flags then do not throttle */ if (!pgdat) |
50694c28f mm: vmscan: check... |
3063 |
goto out; |
5515061d2 mm: throttle dire... |
3064 |
|
68243e76e mm: account for t... |
3065 3066 |
/* Account for the throttling */ count_vm_event(PGSCAN_DIRECT_THROTTLE); |
5515061d2 mm: throttle dire... |
3067 3068 3069 3070 3071 3072 3073 3074 3075 3076 |
/* * 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 ... |
3077 |
allow_direct_reclaim(pgdat), HZ); |
50694c28f mm: vmscan: check... |
3078 3079 |
goto check_pending; |
5515061d2 mm: throttle dire... |
3080 3081 3082 3083 |
} /* Throttle until kswapd wakes the process */ wait_event_killable(zone->zone_pgdat->pfmemalloc_wait, |
c73322d09 mm: fix 100% CPU ... |
3084 |
allow_direct_reclaim(pgdat)); |
50694c28f mm: vmscan: check... |
3085 3086 3087 3088 3089 3090 3091 |
check_pending: if (fatal_signal_pending(current)) return true; out: return false; |
5515061d2 mm: throttle dire... |
3092 |
} |
dac1d27bc mm: use zonelists... |
3093 |
unsigned long try_to_free_pages(struct zonelist *zonelist, int order, |
327c0e968 vmscan: fix it to... |
3094 |
gfp_t gfp_mask, nodemask_t *nodemask) |
66e1707bc Memory controller... |
3095 |
{ |
33906bc5c vmscan: tracing: ... |
3096 |
unsigned long nr_reclaimed; |
66e1707bc Memory controller... |
3097 |
struct scan_control sc = { |
ee814fe23 mm: vmscan: clean... |
3098 |
.nr_to_reclaim = SWAP_CLUSTER_MAX, |
f2f43e566 mm/vmscan.c: fix ... |
3099 |
.gfp_mask = current_gfp_context(gfp_mask), |
b2e18757f mm, vmscan: begin... |
3100 |
.reclaim_idx = gfp_zone(gfp_mask), |
ee814fe23 mm: vmscan: clean... |
3101 3102 3103 |
.order = order, .nodemask = nodemask, .priority = DEF_PRIORITY, |
66e1707bc Memory controller... |
3104 |
.may_writepage = !laptop_mode, |
a6dc60f89 vmscan: rename sc... |
3105 |
.may_unmap = 1, |
2e2e42598 vmscan,memcg: rei... |
3106 |
.may_swap = 1, |
66e1707bc Memory controller... |
3107 |
}; |
5515061d2 mm: throttle dire... |
3108 |
/* |
bb451fdf3 mm/vmscan.c: cond... |
3109 3110 3111 3112 3113 3114 3115 3116 |
* 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... |
3117 3118 3119 |
* 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... |
3120 |
*/ |
f2f43e566 mm/vmscan.c: fix ... |
3121 |
if (throttle_direct_reclaim(sc.gfp_mask, zonelist, nodemask)) |
5515061d2 mm: throttle dire... |
3122 |
return 1; |
1732d2b01 mm/vmscan.c: add ... |
3123 |
set_task_reclaim_state(current, &sc.reclaim_state); |
3481c37ff mm/vmscan: drop m... |
3124 |
trace_mm_vmscan_direct_reclaim_begin(order, sc.gfp_mask); |
33906bc5c vmscan: tracing: ... |
3125 |
|
3115cd914 mm: vmscan: remov... |
3126 |
nr_reclaimed = do_try_to_free_pages(zonelist, &sc); |
33906bc5c vmscan: tracing: ... |
3127 3128 |
trace_mm_vmscan_direct_reclaim_end(nr_reclaimed); |
1732d2b01 mm/vmscan.c: add ... |
3129 |
set_task_reclaim_state(current, NULL); |
33906bc5c vmscan: tracing: ... |
3130 3131 |
return nr_reclaimed; |
66e1707bc Memory controller... |
3132 |
} |
c255a4580 memcg: rename con... |
3133 |
#ifdef CONFIG_MEMCG |
66e1707bc Memory controller... |
3134 |
|
d2e5fb927 mm, memcg: do not... |
3135 |
/* Only used by soft limit reclaim. Do not reuse for anything else. */ |
a9dd0a831 mm, vmscan: make ... |
3136 |
unsigned long mem_cgroup_shrink_node(struct mem_cgroup *memcg, |
4e4169535 memory controller... |
3137 |
gfp_t gfp_mask, bool noswap, |
ef8f23279 mm, memcg: move m... |
3138 |
pg_data_t *pgdat, |
0ae5e89c6 memcg: count the ... |
3139 |
unsigned long *nr_scanned) |
4e4169535 memory controller... |
3140 |
{ |
afaf07a65 mm: vmscan: turn ... |
3141 |
struct lruvec *lruvec = mem_cgroup_lruvec(memcg, pgdat); |
4e4169535 memory controller... |
3142 |
struct scan_control sc = { |
b8f5c5664 memcg: sc.nr_to_r... |
3143 |
.nr_to_reclaim = SWAP_CLUSTER_MAX, |
ee814fe23 mm: vmscan: clean... |
3144 |
.target_mem_cgroup = memcg, |
4e4169535 memory controller... |
3145 3146 |
.may_writepage = !laptop_mode, .may_unmap = 1, |
b2e18757f mm, vmscan: begin... |
3147 |
.reclaim_idx = MAX_NR_ZONES - 1, |
4e4169535 memory controller... |
3148 |
.may_swap = !noswap, |
4e4169535 memory controller... |
3149 |
}; |
0ae5e89c6 memcg: count the ... |
3150 |
|
d2e5fb927 mm, memcg: do not... |
3151 |
WARN_ON_ONCE(!current->reclaim_state); |
4e4169535 memory controller... |
3152 3153 |
sc.gfp_mask = (gfp_mask & GFP_RECLAIM_MASK) | (GFP_HIGHUSER_MOVABLE & ~GFP_RECLAIM_MASK); |
bdce6d9eb memcg, vmscan: ad... |
3154 |
|
9e3b2f8cd mm/vmscan: store ... |
3155 |
trace_mm_vmscan_memcg_softlimit_reclaim_begin(sc.order, |
3481c37ff mm/vmscan: drop m... |
3156 |
sc.gfp_mask); |
bdce6d9eb memcg, vmscan: ad... |
3157 |
|
4e4169535 memory controller... |
3158 3159 3160 |
/* * 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 ... |
3161 |
* if we don't reclaim here, the shrink_node from balance_pgdat |
4e4169535 memory controller... |
3162 3163 3164 |
* will pick up pages from other mem cgroup's as well. We hack * the priority and make it zero. */ |
afaf07a65 mm: vmscan: turn ... |
3165 |
shrink_lruvec(lruvec, &sc); |
bdce6d9eb memcg, vmscan: ad... |
3166 3167 |
trace_mm_vmscan_memcg_softlimit_reclaim_end(sc.nr_reclaimed); |
0ae5e89c6 memcg: count the ... |
3168 |
*nr_scanned = sc.nr_scanned; |
0308f7cf1 mm/vmscan.c: calc... |
3169 |
|
4e4169535 memory controller... |
3170 3171 |
return sc.nr_reclaimed; } |
72835c86c mm: unify remaini... |
3172 |
unsigned long try_to_free_mem_cgroup_pages(struct mem_cgroup *memcg, |
b70a2a21d mm: memcontrol: f... |
3173 |
unsigned long nr_pages, |
a7885eb8a memcg: swappiness |
3174 |
gfp_t gfp_mask, |
b70a2a21d mm: memcontrol: f... |
3175 |
bool may_swap) |
66e1707bc Memory controller... |
3176 |
{ |
bdce6d9eb memcg, vmscan: ad... |
3177 |
unsigned long nr_reclaimed; |
499118e96 mm: introduce mem... |
3178 |
unsigned int noreclaim_flag; |
66e1707bc Memory controller... |
3179 |
struct scan_control sc = { |
b70a2a21d mm: memcontrol: f... |
3180 |
.nr_to_reclaim = max(nr_pages, SWAP_CLUSTER_MAX), |
7dea19f9e mm: introduce mem... |
3181 |
.gfp_mask = (current_gfp_context(gfp_mask) & GFP_RECLAIM_MASK) | |
a09ed5e00 vmscan: change sh... |
3182 |
(GFP_HIGHUSER_MOVABLE & ~GFP_RECLAIM_MASK), |
b2e18757f mm, vmscan: begin... |
3183 |
.reclaim_idx = MAX_NR_ZONES - 1, |
ee814fe23 mm: vmscan: clean... |
3184 3185 3186 3187 |
.target_mem_cgroup = memcg, .priority = DEF_PRIORITY, .may_writepage = !laptop_mode, .may_unmap = 1, |
b70a2a21d mm: memcontrol: f... |
3188 |
.may_swap = may_swap, |
a09ed5e00 vmscan: change sh... |
3189 |
}; |
889976dbc memcg: reclaim me... |
3190 |
/* |
fa40d1ee9 mm: vmscan: memco... |
3191 3192 3193 |
* Traverse the ZONELIST_FALLBACK zonelist of the current node to put * equal pressure on all the nodes. This is based on the assumption that * the reclaim does not bail out early. |
889976dbc memcg: reclaim me... |
3194 |
*/ |
fa40d1ee9 mm: vmscan: memco... |
3195 |
struct zonelist *zonelist = node_zonelist(numa_node_id(), sc.gfp_mask); |
889976dbc memcg: reclaim me... |
3196 |
|
fa40d1ee9 mm: vmscan: memco... |
3197 |
set_task_reclaim_state(current, &sc.reclaim_state); |
3481c37ff mm/vmscan: drop m... |
3198 |
trace_mm_vmscan_memcg_reclaim_begin(0, sc.gfp_mask); |
499118e96 mm: introduce mem... |
3199 |
noreclaim_flag = memalloc_noreclaim_save(); |
eb414681d psi: pressure sta... |
3200 |
|
3115cd914 mm: vmscan: remov... |
3201 |
nr_reclaimed = do_try_to_free_pages(zonelist, &sc); |
eb414681d psi: pressure sta... |
3202 |
|
499118e96 mm: introduce mem... |
3203 |
memalloc_noreclaim_restore(noreclaim_flag); |
bdce6d9eb memcg, vmscan: ad... |
3204 |
trace_mm_vmscan_memcg_reclaim_end(nr_reclaimed); |
1732d2b01 mm/vmscan.c: add ... |
3205 |
set_task_reclaim_state(current, NULL); |
bdce6d9eb memcg, vmscan: ad... |
3206 3207 |
return nr_reclaimed; |
66e1707bc Memory controller... |
3208 3209 |
} #endif |
1d82de618 mm, vmscan: make ... |
3210 |
static void age_active_anon(struct pglist_data *pgdat, |
ef8f23279 mm, memcg: move m... |
3211 |
struct scan_control *sc) |
f16015fbf mm: vmscan: disti... |
3212 |
{ |
b95a2f2d4 mm: vmscan: conve... |
3213 |
struct mem_cgroup *memcg; |
b91ac3743 mm: vmscan: enfor... |
3214 |
struct lruvec *lruvec; |
f16015fbf mm: vmscan: disti... |
3215 |
|
b95a2f2d4 mm: vmscan: conve... |
3216 3217 |
if (!total_swap_pages) return; |
b91ac3743 mm: vmscan: enfor... |
3218 3219 3220 |
lruvec = mem_cgroup_lruvec(NULL, pgdat); if (!inactive_is_low(lruvec, LRU_INACTIVE_ANON)) return; |
b95a2f2d4 mm: vmscan: conve... |
3221 3222 |
memcg = mem_cgroup_iter(NULL, NULL, NULL); do { |
b91ac3743 mm: vmscan: enfor... |
3223 3224 3225 |
lruvec = mem_cgroup_lruvec(memcg, pgdat); shrink_active_list(SWAP_CLUSTER_MAX, lruvec, sc, LRU_ACTIVE_ANON); |
b95a2f2d4 mm: vmscan: conve... |
3226 3227 |
memcg = mem_cgroup_iter(NULL, memcg, NULL); } while (memcg); |
f16015fbf mm: vmscan: disti... |
3228 |
} |
97a225e69 mm/page_alloc: in... |
3229 |
static bool pgdat_watermark_boosted(pg_data_t *pgdat, int highest_zoneidx) |
1c30844d2 mm: reclaim small... |
3230 3231 3232 3233 3234 3235 3236 |
{ 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 |
1eba09c15 mm/vmscan.c: dele... |
3237 |
* should not be checked at the same time as reclaim would |
1c30844d2 mm: reclaim small... |
3238 3239 3240 |
* start prematurely when there is no boosting and a lower * zone is balanced. */ |
97a225e69 mm/page_alloc: in... |
3241 |
for (i = highest_zoneidx; i >= 0; i--) { |
1c30844d2 mm: reclaim small... |
3242 3243 3244 3245 3246 3247 3248 3249 3250 3251 |
zone = pgdat->node_zones + i; if (!managed_zone(zone)) continue; if (zone->watermark_boost) return true; } return false; } |
e716f2eb2 mm, vmscan: preve... |
3252 3253 |
/* * Returns true if there is an eligible zone balanced for the request order |
97a225e69 mm/page_alloc: in... |
3254 |
* and highest_zoneidx |
e716f2eb2 mm, vmscan: preve... |
3255 |
*/ |
97a225e69 mm/page_alloc: in... |
3256 |
static bool pgdat_balanced(pg_data_t *pgdat, int order, int highest_zoneidx) |
60cefed48 mm: vmscan: fix e... |
3257 |
{ |
e716f2eb2 mm, vmscan: preve... |
3258 3259 3260 |
int i; unsigned long mark = -1; struct zone *zone; |
60cefed48 mm: vmscan: fix e... |
3261 |
|
1c30844d2 mm: reclaim small... |
3262 3263 3264 3265 |
/* * Check watermarks bottom-up as lower zones are more likely to * meet watermarks. */ |
97a225e69 mm/page_alloc: in... |
3266 |
for (i = 0; i <= highest_zoneidx; i++) { |
e716f2eb2 mm, vmscan: preve... |
3267 |
zone = pgdat->node_zones + i; |
6256c6b49 mm, vmscan: remov... |
3268 |
|
e716f2eb2 mm, vmscan: preve... |
3269 3270 3271 3272 |
if (!managed_zone(zone)) continue; mark = high_wmark_pages(zone); |
97a225e69 mm/page_alloc: in... |
3273 |
if (zone_watermark_ok_safe(zone, order, mark, highest_zoneidx)) |
e716f2eb2 mm, vmscan: preve... |
3274 3275 3276 3277 |
return true; } /* |
97a225e69 mm/page_alloc: in... |
3278 |
* If a node has no populated zone within highest_zoneidx, it does not |
e716f2eb2 mm, vmscan: preve... |
3279 3280 3281 3282 3283 3284 3285 |
* 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... |
3286 |
} |
631b6e083 mm, vmscan: only ... |
3287 3288 3289 |
/* Clear pgdat state for congested, dirty or under writeback. */ static void clear_pgdat_congested(pg_data_t *pgdat) { |
1b05117df mm: vmscan: harmo... |
3290 3291 3292 |
struct lruvec *lruvec = mem_cgroup_lruvec(NULL, pgdat); clear_bit(LRUVEC_CONGESTED, &lruvec->flags); |
631b6e083 mm, vmscan: only ... |
3293 3294 3295 |
clear_bit(PGDAT_DIRTY, &pgdat->flags); clear_bit(PGDAT_WRITEBACK, &pgdat->flags); } |
1741c8775 mm: kswapd: keep ... |
3296 |
/* |
5515061d2 mm: throttle dire... |
3297 3298 3299 3300 3301 |
* 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 */ |
97a225e69 mm/page_alloc: in... |
3302 3303 |
static bool prepare_kswapd_sleep(pg_data_t *pgdat, int order, int highest_zoneidx) |
f50de2d38 vmscan: have kswa... |
3304 |
{ |
5515061d2 mm: throttle dire... |
3305 |
/* |
9e5e36617 mm, vmscan: preve... |
3306 |
* The throttled processes are normally woken up in balance_pgdat() as |
c73322d09 mm: fix 100% CPU ... |
3307 |
* soon as allow_direct_reclaim() is true. But there is a potential |
9e5e36617 mm, vmscan: preve... |
3308 3309 3310 3311 3312 3313 3314 3315 3316 |
* 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... |
3317 |
*/ |
9e5e36617 mm, vmscan: preve... |
3318 3319 |
if (waitqueue_active(&pgdat->pfmemalloc_wait)) wake_up_all(&pgdat->pfmemalloc_wait); |
f50de2d38 vmscan: have kswa... |
3320 |
|
c73322d09 mm: fix 100% CPU ... |
3321 3322 3323 |
/* Hopeless node, leave it to direct reclaim */ if (pgdat->kswapd_failures >= MAX_RECLAIM_RETRIES) return true; |
97a225e69 mm/page_alloc: in... |
3324 |
if (pgdat_balanced(pgdat, order, highest_zoneidx)) { |
e716f2eb2 mm, vmscan: preve... |
3325 3326 |
clear_pgdat_congested(pgdat); return true; |
1d82de618 mm, vmscan: make ... |
3327 |
} |
333b0a459 mm, vmscan: fix z... |
3328 |
return false; |
f50de2d38 vmscan: have kswa... |
3329 |
} |
1da177e4c Linux-2.6.12-rc2 |
3330 |
/* |
1d82de618 mm, vmscan: make ... |
3331 3332 |
* kswapd shrinks a node of pages that are at or below the highest usable * zone that is currently unbalanced. |
b8e83b942 mm: vmscan: flatt... |
3333 3334 |
* * Returns true if kswapd scanned at least the requested number of pages to |
283aba9f9 mm: vmscan: block... |
3335 3336 |
* 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... |
3337 |
*/ |
1d82de618 mm, vmscan: make ... |
3338 |
static bool kswapd_shrink_node(pg_data_t *pgdat, |
accf62422 mm, kswapd: repla... |
3339 |
struct scan_control *sc) |
75485363c mm: vmscan: limit... |
3340 |
{ |
1d82de618 mm, vmscan: make ... |
3341 3342 |
struct zone *zone; int z; |
75485363c mm: vmscan: limit... |
3343 |
|
1d82de618 mm, vmscan: make ... |
3344 3345 |
/* Reclaim a number of pages proportional to the number of zones */ sc->nr_to_reclaim = 0; |
970a39a36 mm, vmscan: avoid... |
3346 |
for (z = 0; z <= sc->reclaim_idx; z++) { |
1d82de618 mm, vmscan: make ... |
3347 |
zone = pgdat->node_zones + z; |
6aa303def mm, vmscan: only ... |
3348 |
if (!managed_zone(zone)) |
1d82de618 mm, vmscan: make ... |
3349 |
continue; |
7c954f6de mm: vmscan: move ... |
3350 |
|
1d82de618 mm, vmscan: make ... |
3351 3352 |
sc->nr_to_reclaim += max(high_wmark_pages(zone), SWAP_CLUSTER_MAX); } |
7c954f6de mm: vmscan: move ... |
3353 3354 |
/* |
1d82de618 mm, vmscan: make ... |
3355 3356 |
* 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 ... |
3357 |
*/ |
970a39a36 mm, vmscan: avoid... |
3358 |
shrink_node(pgdat, sc); |
283aba9f9 mm: vmscan: block... |
3359 |
|
7c954f6de mm: vmscan: move ... |
3360 |
/* |
1d82de618 mm, vmscan: make ... |
3361 3362 3363 3364 3365 |
* 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 ... |
3366 |
*/ |
9861a62c3 mm, compaction: c... |
3367 |
if (sc->order && sc->nr_reclaimed >= compact_gap(sc->order)) |
1d82de618 mm, vmscan: make ... |
3368 |
sc->order = 0; |
7c954f6de mm: vmscan: move ... |
3369 |
|
b8e83b942 mm: vmscan: flatt... |
3370 |
return sc->nr_scanned >= sc->nr_to_reclaim; |
75485363c mm: vmscan: limit... |
3371 3372 3373 |
} /* |
1d82de618 mm, vmscan: make ... |
3374 3375 3376 |
* 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 |
3377 |
* |
1d82de618 mm, vmscan: make ... |
3378 |
* Returns the order kswapd finished reclaiming at. |
1da177e4c Linux-2.6.12-rc2 |
3379 3380 |
* * kswapd scans the zones in the highmem->normal->dma direction. It skips |
418589663 page allocator: u... |
3381 |
* zones which have free_pages > high_wmark_pages(zone), but once a zone is |
8bb4e7a2e mm: fix some typo... |
3382 |
* found to have free_pages <= high_wmark_pages(zone), any page in that zone |
1d82de618 mm, vmscan: make ... |
3383 3384 |
* or lower is eligible for reclaim until at least one usable zone is * balanced. |
1da177e4c Linux-2.6.12-rc2 |
3385 |
*/ |
97a225e69 mm/page_alloc: in... |
3386 |
static int balance_pgdat(pg_data_t *pgdat, int order, int highest_zoneidx) |
1da177e4c Linux-2.6.12-rc2 |
3387 |
{ |
1da177e4c Linux-2.6.12-rc2 |
3388 |
int i; |
0608f43da revert "memcg, vm... |
3389 3390 |
unsigned long nr_soft_reclaimed; unsigned long nr_soft_scanned; |
eb414681d psi: pressure sta... |
3391 |
unsigned long pflags; |
1c30844d2 mm: reclaim small... |
3392 3393 3394 |
unsigned long nr_boost_reclaim; unsigned long zone_boosts[MAX_NR_ZONES] = { 0, }; bool boosted; |
1d82de618 mm, vmscan: make ... |
3395 |
struct zone *zone; |
179e96395 [PATCH] vmscan: s... |
3396 3397 |
struct scan_control sc = { .gfp_mask = GFP_KERNEL, |
ee814fe23 mm: vmscan: clean... |
3398 |
.order = order, |
a6dc60f89 vmscan: rename sc... |
3399 |
.may_unmap = 1, |
179e96395 [PATCH] vmscan: s... |
3400 |
}; |
93781325d lockdep: fix fs_r... |
3401 |
|
1732d2b01 mm/vmscan.c: add ... |
3402 |
set_task_reclaim_state(current, &sc.reclaim_state); |
eb414681d psi: pressure sta... |
3403 |
psi_memstall_enter(&pflags); |
93781325d lockdep: fix fs_r... |
3404 |
__fs_reclaim_acquire(); |
f8891e5e1 [PATCH] Light wei... |
3405 |
count_vm_event(PAGEOUTRUN); |
1da177e4c Linux-2.6.12-rc2 |
3406 |
|
1c30844d2 mm: reclaim small... |
3407 3408 3409 3410 3411 3412 |
/* * 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; |
97a225e69 mm/page_alloc: in... |
3413 |
for (i = 0; i <= highest_zoneidx; i++) { |
1c30844d2 mm: reclaim small... |
3414 3415 3416 3417 3418 3419 3420 3421 3422 3423 3424 |
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 ... |
3425 |
do { |
c73322d09 mm: fix 100% CPU ... |
3426 |
unsigned long nr_reclaimed = sc.nr_reclaimed; |
b8e83b942 mm: vmscan: flatt... |
3427 |
bool raise_priority = true; |
1c30844d2 mm: reclaim small... |
3428 |
bool balanced; |
93781325d lockdep: fix fs_r... |
3429 |
bool ret; |
b8e83b942 mm: vmscan: flatt... |
3430 |
|
97a225e69 mm/page_alloc: in... |
3431 |
sc.reclaim_idx = highest_zoneidx; |
1da177e4c Linux-2.6.12-rc2 |
3432 |
|
86c79f6b5 mm: vmscan: do no... |
3433 |
/* |
84c7a7771 mm, vmscan: Have ... |
3434 3435 3436 3437 3438 3439 3440 3441 |
* 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... |
3442 3443 3444 3445 |
*/ if (buffer_heads_over_limit) { for (i = MAX_NR_ZONES - 1; i >= 0; i--) { zone = pgdat->node_zones + i; |
6aa303def mm, vmscan: only ... |
3446 |
if (!managed_zone(zone)) |
86c79f6b5 mm: vmscan: do no... |
3447 |
continue; |
cc715d99e mm: vmscan: forci... |
3448 |
|
970a39a36 mm, vmscan: avoid... |
3449 |
sc.reclaim_idx = i; |
e1dbeda60 [PATCH] balance_p... |
3450 |
break; |
1da177e4c Linux-2.6.12-rc2 |
3451 |
} |
1da177e4c Linux-2.6.12-rc2 |
3452 |
} |
dafcb73e3 mm: avoid calling... |
3453 |
|
86c79f6b5 mm: vmscan: do no... |
3454 |
/* |
1c30844d2 mm: reclaim small... |
3455 3456 3457 3458 3459 3460 |
* 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. */ |
97a225e69 mm/page_alloc: in... |
3461 |
balanced = pgdat_balanced(pgdat, sc.order, highest_zoneidx); |
1c30844d2 mm: reclaim small... |
3462 3463 3464 3465 3466 3467 3468 3469 3470 |
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... |
3471 |
*/ |
1c30844d2 mm: reclaim small... |
3472 |
if (!nr_boost_reclaim && balanced) |
e716f2eb2 mm, vmscan: preve... |
3473 |
goto out; |
e1dbeda60 [PATCH] balance_p... |
3474 |
|
1c30844d2 mm: reclaim small... |
3475 3476 3477 3478 3479 3480 3481 3482 3483 3484 3485 3486 |
/* 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... |
3487 |
|
1da177e4c Linux-2.6.12-rc2 |
3488 |
/* |
1d82de618 mm, vmscan: make ... |
3489 3490 3491 3492 3493 |
* 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... |
3494 |
age_active_anon(pgdat, &sc); |
1d82de618 mm, vmscan: make ... |
3495 3496 |
/* |
b7ea3c417 mm: vmscan: check... |
3497 3498 3499 |
* If we're getting trouble reclaiming, start doing writepage * even in laptop mode. */ |
047d72c30 mm: remove seemin... |
3500 |
if (sc.priority < DEF_PRIORITY - 2) |
b7ea3c417 mm: vmscan: check... |
3501 |
sc.may_writepage = 1; |
1d82de618 mm, vmscan: make ... |
3502 3503 3504 |
/* Call soft limit reclaim before calling shrink_node. */ sc.nr_scanned = 0; nr_soft_scanned = 0; |
ef8f23279 mm, memcg: move m... |
3505 |
nr_soft_reclaimed = mem_cgroup_soft_limit_reclaim(pgdat, sc.order, |
1d82de618 mm, vmscan: make ... |
3506 3507 |
sc.gfp_mask, &nr_soft_scanned); sc.nr_reclaimed += nr_soft_reclaimed; |
b7ea3c417 mm: vmscan: check... |
3508 |
/* |
1d82de618 mm, vmscan: make ... |
3509 3510 3511 |
* 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 |
3512 |
*/ |
970a39a36 mm, vmscan: avoid... |
3513 |
if (kswapd_shrink_node(pgdat, &sc)) |
1d82de618 mm, vmscan: make ... |
3514 |
raise_priority = false; |
5515061d2 mm: throttle dire... |
3515 3516 3517 3518 3519 3520 3521 |
/* * 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 ... |
3522 |
allow_direct_reclaim(pgdat)) |
cfc511557 mm, vmscan: wake ... |
3523 |
wake_up_all(&pgdat->pfmemalloc_wait); |
5515061d2 mm: throttle dire... |
3524 |
|
b8e83b942 mm: vmscan: flatt... |
3525 |
/* Check if kswapd should be suspending */ |
93781325d lockdep: fix fs_r... |
3526 3527 3528 3529 |
__fs_reclaim_release(); ret = try_to_freeze(); __fs_reclaim_acquire(); if (ret || kthread_should_stop()) |
b8e83b942 mm: vmscan: flatt... |
3530 |
break; |
8357376d3 [PATCH] swsusp: I... |
3531 |
|
73ce02e96 mm: stop kswapd's... |
3532 |
/* |
b8e83b942 mm: vmscan: flatt... |
3533 3534 |
* Raise priority if scanning rate is too low or there was no * progress in reclaiming pages |
73ce02e96 mm: stop kswapd's... |
3535 |
*/ |
c73322d09 mm: fix 100% CPU ... |
3536 |
nr_reclaimed = sc.nr_reclaimed - nr_reclaimed; |
1c30844d2 mm: reclaim small... |
3537 3538 3539 3540 3541 3542 3543 3544 3545 |
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 ... |
3546 |
if (raise_priority || !nr_reclaimed) |
b8e83b942 mm: vmscan: flatt... |
3547 |
sc.priority--; |
1d82de618 mm, vmscan: make ... |
3548 |
} while (sc.priority >= 1); |
1da177e4c Linux-2.6.12-rc2 |
3549 |
|
c73322d09 mm: fix 100% CPU ... |
3550 3551 |
if (!sc.nr_reclaimed) pgdat->kswapd_failures++; |
b8e83b942 mm: vmscan: flatt... |
3552 |
out: |
1c30844d2 mm: reclaim small... |
3553 3554 3555 |
/* If reclaim was boosted, account for the reclaim done in this pass */ if (boosted) { unsigned long flags; |
97a225e69 mm/page_alloc: in... |
3556 |
for (i = 0; i <= highest_zoneidx; i++) { |
1c30844d2 mm: reclaim small... |
3557 3558 3559 3560 3561 3562 3563 3564 3565 3566 3567 3568 3569 3570 |
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. */ |
97a225e69 mm/page_alloc: in... |
3571 |
wakeup_kcompactd(pgdat, pageblock_order, highest_zoneidx); |
1c30844d2 mm: reclaim small... |
3572 |
} |
2a2e48854 mm: vmscan: fix I... |
3573 |
snapshot_refaults(NULL, pgdat); |
93781325d lockdep: fix fs_r... |
3574 |
__fs_reclaim_release(); |
eb414681d psi: pressure sta... |
3575 |
psi_memstall_leave(&pflags); |
1732d2b01 mm/vmscan.c: add ... |
3576 |
set_task_reclaim_state(current, NULL); |
e5ca8071f mm/vmscan.c: add ... |
3577 |
|
0abdee2bd mm: kswapd: use t... |
3578 |
/* |
1d82de618 mm, vmscan: make ... |
3579 3580 3581 3582 |
* 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... |
3583 |
*/ |
1d82de618 mm, vmscan: make ... |
3584 |
return sc.order; |
1da177e4c Linux-2.6.12-rc2 |
3585 |
} |
e716f2eb2 mm, vmscan: preve... |
3586 |
/* |
97a225e69 mm/page_alloc: in... |
3587 3588 3589 3590 3591 |
* The pgdat->kswapd_highest_zoneidx 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... |
3592 |
*/ |
97a225e69 mm/page_alloc: in... |
3593 3594 |
static enum zone_type kswapd_highest_zoneidx(pg_data_t *pgdat, enum zone_type prev_highest_zoneidx) |
e716f2eb2 mm, vmscan: preve... |
3595 |
{ |
97a225e69 mm/page_alloc: in... |
3596 |
enum zone_type curr_idx = READ_ONCE(pgdat->kswapd_highest_zoneidx); |
5644e1fbb mm/vmscan.c: fix ... |
3597 |
|
97a225e69 mm/page_alloc: in... |
3598 |
return curr_idx == MAX_NR_ZONES ? prev_highest_zoneidx : curr_idx; |
e716f2eb2 mm, vmscan: preve... |
3599 |
} |
38087d9b0 mm, vmscan: simpl... |
3600 |
static void kswapd_try_to_sleep(pg_data_t *pgdat, int alloc_order, int reclaim_order, |
97a225e69 mm/page_alloc: in... |
3601 |
unsigned int highest_zoneidx) |
f0bc0a60b vmscan: factor ou... |
3602 3603 3604 3605 3606 3607 3608 3609 |
{ 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... |
3610 3611 3612 3613 3614 3615 3616 |
/* * 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. */ |
97a225e69 mm/page_alloc: in... |
3617 |
if (prepare_kswapd_sleep(pgdat, reclaim_order, highest_zoneidx)) { |
fd901c953 mm: wake kcompact... |
3618 3619 3620 3621 3622 3623 3624 3625 3626 3627 3628 3629 |
/* * 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. */ |
97a225e69 mm/page_alloc: in... |
3630 |
wakeup_kcompactd(pgdat, alloc_order, highest_zoneidx); |
fd901c953 mm: wake kcompact... |
3631 |
|
f0bc0a60b vmscan: factor ou... |
3632 |
remaining = schedule_timeout(HZ/10); |
38087d9b0 mm, vmscan: simpl... |
3633 3634 |
/* |
97a225e69 mm/page_alloc: in... |
3635 |
* If woken prematurely then reset kswapd_highest_zoneidx and |
38087d9b0 mm, vmscan: simpl... |
3636 3637 3638 3639 |
* order. The values will either be from a wakeup request or * the previous request that slept prematurely. */ if (remaining) { |
97a225e69 mm/page_alloc: in... |
3640 3641 3642 |
WRITE_ONCE(pgdat->kswapd_highest_zoneidx, kswapd_highest_zoneidx(pgdat, highest_zoneidx)); |
5644e1fbb mm/vmscan.c: fix ... |
3643 3644 3645 |
if (READ_ONCE(pgdat->kswapd_order) < reclaim_order) WRITE_ONCE(pgdat->kswapd_order, reclaim_order); |
38087d9b0 mm, vmscan: simpl... |
3646 |
} |
f0bc0a60b vmscan: factor ou... |
3647 3648 3649 3650 3651 3652 3653 3654 |
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... |
3655 |
if (!remaining && |
97a225e69 mm/page_alloc: in... |
3656 |
prepare_kswapd_sleep(pgdat, reclaim_order, highest_zoneidx)) { |
f0bc0a60b vmscan: factor ou... |
3657 3658 3659 3660 3661 3662 3663 3664 3665 3666 3667 |
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... |
3668 3669 3670 |
if (!kthread_should_stop()) schedule(); |
f0bc0a60b vmscan: factor ou... |
3671 3672 3673 3674 3675 3676 3677 3678 3679 |
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 |
3680 3681 |
/* * The background pageout daemon, started as a kernel thread |
4f98a2fee vmscan: split LRU... |
3682 |
* from the init process. |
1da177e4c Linux-2.6.12-rc2 |
3683 3684 3685 3686 3687 3688 3689 3690 3691 3692 3693 3694 |
* * 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... |
3695 |
unsigned int alloc_order, reclaim_order; |
97a225e69 mm/page_alloc: in... |
3696 |
unsigned int highest_zoneidx = MAX_NR_ZONES - 1; |
1da177e4c Linux-2.6.12-rc2 |
3697 3698 |
pg_data_t *pgdat = (pg_data_t*)p; struct task_struct *tsk = current; |
a70f73028 cpumask: replace ... |
3699 |
const struct cpumask *cpumask = cpumask_of_node(pgdat->node_id); |
1da177e4c Linux-2.6.12-rc2 |
3700 |
|
174596a0b cpumask: convert mm/ |
3701 |
if (!cpumask_empty(cpumask)) |
c5f59f083 nodemask: use new... |
3702 |
set_cpus_allowed_ptr(tsk, cpumask); |
1da177e4c Linux-2.6.12-rc2 |
3703 3704 3705 3706 3707 3708 3709 3710 3711 3712 3713 3714 3715 |
/* * 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... |
3716 |
tsk->flags |= PF_MEMALLOC | PF_SWAPWRITE | PF_KSWAPD; |
831441862 Freezer: make ker... |
3717 |
set_freezable(); |
1da177e4c Linux-2.6.12-rc2 |
3718 |
|
5644e1fbb mm/vmscan.c: fix ... |
3719 |
WRITE_ONCE(pgdat->kswapd_order, 0); |
97a225e69 mm/page_alloc: in... |
3720 |
WRITE_ONCE(pgdat->kswapd_highest_zoneidx, MAX_NR_ZONES); |
1da177e4c Linux-2.6.12-rc2 |
3721 |
for ( ; ; ) { |
6f6313d48 mm/vmscan.c: try_... |
3722 |
bool ret; |
3e1d1d28d [PATCH] Cleanup p... |
3723 |
|
5644e1fbb mm/vmscan.c: fix ... |
3724 |
alloc_order = reclaim_order = READ_ONCE(pgdat->kswapd_order); |
97a225e69 mm/page_alloc: in... |
3725 3726 |
highest_zoneidx = kswapd_highest_zoneidx(pgdat, highest_zoneidx); |
e716f2eb2 mm, vmscan: preve... |
3727 |
|
38087d9b0 mm, vmscan: simpl... |
3728 3729 |
kswapd_try_sleep: kswapd_try_to_sleep(pgdat, alloc_order, reclaim_order, |
97a225e69 mm/page_alloc: in... |
3730 |
highest_zoneidx); |
215ddd666 mm: vmscan: only ... |
3731 |
|
97a225e69 mm/page_alloc: in... |
3732 |
/* Read the new order and highest_zoneidx */ |
5644e1fbb mm/vmscan.c: fix ... |
3733 |
alloc_order = reclaim_order = READ_ONCE(pgdat->kswapd_order); |
97a225e69 mm/page_alloc: in... |
3734 3735 |
highest_zoneidx = kswapd_highest_zoneidx(pgdat, highest_zoneidx); |
5644e1fbb mm/vmscan.c: fix ... |
3736 |
WRITE_ONCE(pgdat->kswapd_order, 0); |
97a225e69 mm/page_alloc: in... |
3737 |
WRITE_ONCE(pgdat->kswapd_highest_zoneidx, MAX_NR_ZONES); |
1da177e4c Linux-2.6.12-rc2 |
3738 |
|
8fe23e057 mm: clear node in... |
3739 3740 3741 3742 3743 3744 3745 3746 |
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... |
3747 3748 3749 3750 3751 3752 3753 3754 3755 3756 3757 |
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). */ |
97a225e69 mm/page_alloc: in... |
3758 |
trace_mm_vmscan_kswapd_wake(pgdat->node_id, highest_zoneidx, |
e5146b12e mm, vmscan: add c... |
3759 |
alloc_order); |
97a225e69 mm/page_alloc: in... |
3760 3761 |
reclaim_order = balance_pgdat(pgdat, alloc_order, highest_zoneidx); |
38087d9b0 mm, vmscan: simpl... |
3762 3763 |
if (reclaim_order < alloc_order) goto kswapd_try_sleep; |
1da177e4c Linux-2.6.12-rc2 |
3764 |
} |
b0a8cc58e mm: bugfix: set c... |
3765 |
|
71abdc15a mm: vmscan: clear... |
3766 |
tsk->flags &= ~(PF_MEMALLOC | PF_SWAPWRITE | PF_KSWAPD); |
71abdc15a mm: vmscan: clear... |
3767 |
|
1da177e4c Linux-2.6.12-rc2 |
3768 3769 3770 3771 |
return 0; } /* |
5ecd9d403 mm, page_alloc: w... |
3772 3773 3774 3775 3776 |
* 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 |
3777 |
*/ |
5ecd9d403 mm, page_alloc: w... |
3778 |
void wakeup_kswapd(struct zone *zone, gfp_t gfp_flags, int order, |
97a225e69 mm/page_alloc: in... |
3779 |
enum zone_type highest_zoneidx) |
1da177e4c Linux-2.6.12-rc2 |
3780 3781 |
{ pg_data_t *pgdat; |
5644e1fbb mm/vmscan.c: fix ... |
3782 |
enum zone_type curr_idx; |
1da177e4c Linux-2.6.12-rc2 |
3783 |
|
6aa303def mm, vmscan: only ... |
3784 |
if (!managed_zone(zone)) |
1da177e4c Linux-2.6.12-rc2 |
3785 |
return; |
5ecd9d403 mm, page_alloc: w... |
3786 |
if (!cpuset_zone_allowed(zone, gfp_flags)) |
1da177e4c Linux-2.6.12-rc2 |
3787 |
return; |
5644e1fbb mm/vmscan.c: fix ... |
3788 |
|
88f5acf88 mm: page allocato... |
3789 |
pgdat = zone->zone_pgdat; |
97a225e69 mm/page_alloc: in... |
3790 |
curr_idx = READ_ONCE(pgdat->kswapd_highest_zoneidx); |
5644e1fbb mm/vmscan.c: fix ... |
3791 |
|
97a225e69 mm/page_alloc: in... |
3792 3793 |
if (curr_idx == MAX_NR_ZONES || curr_idx < highest_zoneidx) WRITE_ONCE(pgdat->kswapd_highest_zoneidx, highest_zoneidx); |
5644e1fbb mm/vmscan.c: fix ... |
3794 3795 3796 |
if (READ_ONCE(pgdat->kswapd_order) < order) WRITE_ONCE(pgdat->kswapd_order, order); |
dffcac2cb mm/vmscan.c: prev... |
3797 |
|
8d0986e28 [PATCH] vm: kswap... |
3798 |
if (!waitqueue_active(&pgdat->kswapd_wait)) |
1da177e4c Linux-2.6.12-rc2 |
3799 |
return; |
e1a556374 mm, vmscan: only ... |
3800 |
|
5ecd9d403 mm, page_alloc: w... |
3801 3802 |
/* Hopeless node, leave it to direct reclaim if possible */ if (pgdat->kswapd_failures >= MAX_RECLAIM_RETRIES || |
97a225e69 mm/page_alloc: in... |
3803 3804 |
(pgdat_balanced(pgdat, order, highest_zoneidx) && !pgdat_watermark_boosted(pgdat, highest_zoneidx))) { |
5ecd9d403 mm, page_alloc: w... |
3805 3806 3807 3808 3809 3810 3811 3812 |
/* * 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)) |
97a225e69 mm/page_alloc: in... |
3813 |
wakeup_kcompactd(pgdat, order, highest_zoneidx); |
e716f2eb2 mm, vmscan: preve... |
3814 |
return; |
5ecd9d403 mm, page_alloc: w... |
3815 |
} |
88f5acf88 mm: page allocato... |
3816 |
|
97a225e69 mm/page_alloc: in... |
3817 |
trace_mm_vmscan_wakeup_kswapd(pgdat->node_id, highest_zoneidx, order, |
5ecd9d403 mm, page_alloc: w... |
3818 |
gfp_flags); |
8d0986e28 [PATCH] vm: kswap... |
3819 |
wake_up_interruptible(&pgdat->kswapd_wait); |
1da177e4c Linux-2.6.12-rc2 |
3820 |
} |
c6f37f121 PM/Suspend: Do no... |
3821 |
#ifdef CONFIG_HIBERNATION |
1da177e4c Linux-2.6.12-rc2 |
3822 |
/* |
7b51755c3 vmscan: kill hibe... |
3823 |
* Try to free `nr_to_reclaim' of memory, system-wide, and return the number of |
d6277db4a [PATCH] swsusp: r... |
3824 3825 3826 3827 3828 |
* 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 |
3829 |
*/ |
7b51755c3 vmscan: kill hibe... |
3830 |
unsigned long shrink_all_memory(unsigned long nr_to_reclaim) |
1da177e4c Linux-2.6.12-rc2 |
3831 |
{ |
d6277db4a [PATCH] swsusp: r... |
3832 |
struct scan_control sc = { |
ee814fe23 mm: vmscan: clean... |
3833 |
.nr_to_reclaim = nr_to_reclaim, |
7b51755c3 vmscan: kill hibe... |
3834 |
.gfp_mask = GFP_HIGHUSER_MOVABLE, |
b2e18757f mm, vmscan: begin... |
3835 |
.reclaim_idx = MAX_NR_ZONES - 1, |
ee814fe23 mm: vmscan: clean... |
3836 |
.priority = DEF_PRIORITY, |
d6277db4a [PATCH] swsusp: r... |
3837 |
.may_writepage = 1, |
ee814fe23 mm: vmscan: clean... |
3838 3839 |
.may_unmap = 1, .may_swap = 1, |
7b51755c3 vmscan: kill hibe... |
3840 |
.hibernation_mode = 1, |
1da177e4c Linux-2.6.12-rc2 |
3841 |
}; |
a09ed5e00 vmscan: change sh... |
3842 |
struct zonelist *zonelist = node_zonelist(numa_node_id(), sc.gfp_mask); |
7b51755c3 vmscan: kill hibe... |
3843 |
unsigned long nr_reclaimed; |
499118e96 mm: introduce mem... |
3844 |
unsigned int noreclaim_flag; |
1da177e4c Linux-2.6.12-rc2 |
3845 |
|
d92a8cfcb locking/lockdep: ... |
3846 |
fs_reclaim_acquire(sc.gfp_mask); |
93781325d lockdep: fix fs_r... |
3847 |
noreclaim_flag = memalloc_noreclaim_save(); |
1732d2b01 mm/vmscan.c: add ... |
3848 |
set_task_reclaim_state(current, &sc.reclaim_state); |
d6277db4a [PATCH] swsusp: r... |
3849 |
|
3115cd914 mm: vmscan: remov... |
3850 |
nr_reclaimed = do_try_to_free_pages(zonelist, &sc); |
d979677c4 mm: shrink_all_me... |
3851 |
|
1732d2b01 mm/vmscan.c: add ... |
3852 |
set_task_reclaim_state(current, NULL); |
499118e96 mm: introduce mem... |
3853 |
memalloc_noreclaim_restore(noreclaim_flag); |
93781325d lockdep: fix fs_r... |
3854 |
fs_reclaim_release(sc.gfp_mask); |
d6277db4a [PATCH] swsusp: r... |
3855 |
|
7b51755c3 vmscan: kill hibe... |
3856 |
return nr_reclaimed; |
1da177e4c Linux-2.6.12-rc2 |
3857 |
} |
c6f37f121 PM/Suspend: Do no... |
3858 |
#endif /* CONFIG_HIBERNATION */ |
1da177e4c Linux-2.6.12-rc2 |
3859 |
|
3218ae14b [PATCH] pgdat all... |
3860 3861 3862 3863 3864 3865 3866 3867 3868 3869 3870 3871 3872 3873 3874 |
/* * 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... |
3875 |
BUG_ON(system_state < SYSTEM_RUNNING); |
d5dc0ad92 mm/vmscan: fix er... |
3876 3877 3878 |
pr_err("Failed to start kswapd on node %d ", nid); ret = PTR_ERR(pgdat->kswapd); |
d72515b85 mm/vmscan: fix er... |
3879 |
pgdat->kswapd = NULL; |
3218ae14b [PATCH] pgdat all... |
3880 3881 3882 |
} return ret; } |
8fe23e057 mm: clear node in... |
3883 |
/* |
d8adde17e memory hotplug: f... |
3884 |
* Called by memory hotplug when all memory in a node is offlined. Caller must |
bfc8c9013 mem-hotplug: impl... |
3885 |
* hold mem_hotplug_begin/end(). |
8fe23e057 mm: clear node in... |
3886 3887 3888 3889 |
*/ void kswapd_stop(int nid) { struct task_struct *kswapd = NODE_DATA(nid)->kswapd; |
d8adde17e memory hotplug: f... |
3890 |
if (kswapd) { |
8fe23e057 mm: clear node in... |
3891 |
kthread_stop(kswapd); |
d8adde17e memory hotplug: f... |
3892 3893 |
NODE_DATA(nid)->kswapd = NULL; } |
8fe23e057 mm: clear node in... |
3894 |
} |
1da177e4c Linux-2.6.12-rc2 |
3895 3896 |
static int __init kswapd_init(void) { |
6b700b5b3 mm/vmscan.c: remo... |
3897 |
int nid; |
69e05944a [PATCH] vmscan: u... |
3898 |
|
1da177e4c Linux-2.6.12-rc2 |
3899 |
swap_setup(); |
48fb2e240 vmscan: use N_MEM... |
3900 |
for_each_node_state(nid, N_MEMORY) |
3218ae14b [PATCH] pgdat all... |
3901 |
kswapd_run(nid); |
1da177e4c Linux-2.6.12-rc2 |
3902 3903 3904 3905 |
return 0; } module_init(kswapd_init) |
9eeff2395 [PATCH] Zone recl... |
3906 3907 3908 |
#ifdef CONFIG_NUMA /* |
a5f5f91da mm: convert zone_... |
3909 |
* Node reclaim mode |
9eeff2395 [PATCH] Zone recl... |
3910 |
* |
a5f5f91da mm: convert zone_... |
3911 |
* If non-zero call node_reclaim when the number of free pages falls below |
9eeff2395 [PATCH] Zone recl... |
3912 |
* the watermarks. |
9eeff2395 [PATCH] Zone recl... |
3913 |
*/ |
a5f5f91da mm: convert zone_... |
3914 |
int node_reclaim_mode __read_mostly; |
9eeff2395 [PATCH] Zone recl... |
3915 |
|
648b5cf36 mm/vmscan: remove... |
3916 3917 |
#define RECLAIM_WRITE (1<<0) /* Writeout pages during reclaim */ #define RECLAIM_UNMAP (1<<1) /* Unmap pages during reclaim */ |
1b2ffb789 [PATCH] Zone recl... |
3918 |
|
9eeff2395 [PATCH] Zone recl... |
3919 |
/* |
a5f5f91da mm: convert zone_... |
3920 |
* Priority for NODE_RECLAIM. This determines the fraction of pages |
a92f71263 [PATCH] zone_recl... |
3921 3922 3923 |
* of a node considered for each zone_reclaim. 4 scans 1/16th of * a zone. */ |
a5f5f91da mm: convert zone_... |
3924 |
#define NODE_RECLAIM_PRIORITY 4 |
a92f71263 [PATCH] zone_recl... |
3925 |
|
9eeff2395 [PATCH] Zone recl... |
3926 |
/* |
a5f5f91da mm: convert zone_... |
3927 |
* Percentage of pages in a zone that must be unmapped for node_reclaim to |
9614634fe [PATCH] ZVC/zone_... |
3928 3929 3930 3931 3932 |
* occur. */ int sysctl_min_unmapped_ratio = 1; /* |
0ff38490c [PATCH] zone_recl... |
3933 3934 3935 3936 |
* 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... |
3937 |
static inline unsigned long node_unmapped_file_pages(struct pglist_data *pgdat) |
90afa5de6 vmscan: properly ... |
3938 |
{ |
11fb99898 mm: move most fil... |
3939 3940 3941 |
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 ... |
3942 3943 3944 3945 3946 3947 3948 3949 3950 3951 |
/* * 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_... |
3952 |
static unsigned long node_pagecache_reclaimable(struct pglist_data *pgdat) |
90afa5de6 vmscan: properly ... |
3953 |
{ |
d031a1579 mm/vmscan.c: fix ... |
3954 3955 |
unsigned long nr_pagecache_reclaimable; unsigned long delta = 0; |
90afa5de6 vmscan: properly ... |
3956 3957 |
/* |
95bbc0c72 mm: rename RECLAI... |
3958 |
* If RECLAIM_UNMAP is set, then all file pages are considered |
90afa5de6 vmscan: properly ... |
3959 |
* potentially reclaimable. Otherwise, we have to worry about |
11fb99898 mm: move most fil... |
3960 |
* pages like swapcache and node_unmapped_file_pages() provides |
90afa5de6 vmscan: properly ... |
3961 3962 |
* a better estimate */ |
a5f5f91da mm: convert zone_... |
3963 3964 |
if (node_reclaim_mode & RECLAIM_UNMAP) nr_pagecache_reclaimable = node_page_state(pgdat, NR_FILE_PAGES); |
90afa5de6 vmscan: properly ... |
3965 |
else |
a5f5f91da mm: convert zone_... |
3966 |
nr_pagecache_reclaimable = node_unmapped_file_pages(pgdat); |
90afa5de6 vmscan: properly ... |
3967 3968 |
/* If we can't clean pages, remove dirty pages from consideration */ |
a5f5f91da mm: convert zone_... |
3969 3970 |
if (!(node_reclaim_mode & RECLAIM_WRITE)) delta += node_page_state(pgdat, NR_FILE_DIRTY); |
90afa5de6 vmscan: properly ... |
3971 3972 3973 3974 3975 3976 3977 |
/* 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... |
3978 |
/* |
a5f5f91da mm: convert zone_... |
3979 |
* Try to free up some pages from this node through reclaim. |
9eeff2395 [PATCH] Zone recl... |
3980 |
*/ |
a5f5f91da mm: convert zone_... |
3981 |
static int __node_reclaim(struct pglist_data *pgdat, gfp_t gfp_mask, unsigned int order) |
9eeff2395 [PATCH] Zone recl... |
3982 |
{ |
7fb2d46d3 [PATCH] zone_recl... |
3983 |
/* Minimum pages needed in order to stay on node */ |
69e05944a [PATCH] vmscan: u... |
3984 |
const unsigned long nr_pages = 1 << order; |
9eeff2395 [PATCH] Zone recl... |
3985 |
struct task_struct *p = current; |
499118e96 mm: introduce mem... |
3986 |
unsigned int noreclaim_flag; |
179e96395 [PATCH] vmscan: s... |
3987 |
struct scan_control sc = { |
62b726c1b mm/vmscan.c:__zon... |
3988 |
.nr_to_reclaim = max(nr_pages, SWAP_CLUSTER_MAX), |
f2f43e566 mm/vmscan.c: fix ... |
3989 |
.gfp_mask = current_gfp_context(gfp_mask), |
bd2f6199c vmscan: respect h... |
3990 |
.order = order, |
a5f5f91da mm: convert zone_... |
3991 3992 3993 |
.priority = NODE_RECLAIM_PRIORITY, .may_writepage = !!(node_reclaim_mode & RECLAIM_WRITE), .may_unmap = !!(node_reclaim_mode & RECLAIM_UNMAP), |
ee814fe23 mm: vmscan: clean... |
3994 |
.may_swap = 1, |
f2f43e566 mm/vmscan.c: fix ... |
3995 |
.reclaim_idx = gfp_zone(gfp_mask), |
179e96395 [PATCH] vmscan: s... |
3996 |
}; |
9eeff2395 [PATCH] Zone recl... |
3997 |
|
132bb8cfc mm/vmscan: add tr... |
3998 3999 |
trace_mm_vmscan_node_reclaim_begin(pgdat->node_id, order, sc.gfp_mask); |
9eeff2395 [PATCH] Zone recl... |
4000 |
cond_resched(); |
93781325d lockdep: fix fs_r... |
4001 |
fs_reclaim_acquire(sc.gfp_mask); |
d4f7796e9 [PATCH] vmscan: f... |
4002 |
/* |
95bbc0c72 mm: rename RECLAI... |
4003 |
* We need to be able to allocate from the reserves for RECLAIM_UNMAP |
d4f7796e9 [PATCH] vmscan: f... |
4004 |
* and we also need to be able to write out pages for RECLAIM_WRITE |
95bbc0c72 mm: rename RECLAI... |
4005 |
* and RECLAIM_UNMAP. |
d4f7796e9 [PATCH] vmscan: f... |
4006 |
*/ |
499118e96 mm: introduce mem... |
4007 4008 |
noreclaim_flag = memalloc_noreclaim_save(); p->flags |= PF_SWAPWRITE; |
1732d2b01 mm/vmscan.c: add ... |
4009 |
set_task_reclaim_state(p, &sc.reclaim_state); |
c84db23c6 [PATCH] zone_recl... |
4010 |
|
a5f5f91da mm: convert zone_... |
4011 |
if (node_pagecache_reclaimable(pgdat) > pgdat->min_unmapped_pages) { |
0ff38490c [PATCH] zone_recl... |
4012 |
/* |
894befec4 mm/vmscan: update... |
4013 |
* Free memory by calling shrink node with increasing |
0ff38490c [PATCH] zone_recl... |
4014 4015 |
* priorities until we have enough memory freed. */ |
0ff38490c [PATCH] zone_recl... |
4016 |
do { |
970a39a36 mm, vmscan: avoid... |
4017 |
shrink_node(pgdat, &sc); |
9e3b2f8cd mm/vmscan: store ... |
4018 |
} while (sc.nr_reclaimed < nr_pages && --sc.priority >= 0); |
0ff38490c [PATCH] zone_recl... |
4019 |
} |
c84db23c6 [PATCH] zone_recl... |
4020 |
|
1732d2b01 mm/vmscan.c: add ... |
4021 |
set_task_reclaim_state(p, NULL); |
499118e96 mm: introduce mem... |
4022 4023 |
current->flags &= ~PF_SWAPWRITE; memalloc_noreclaim_restore(noreclaim_flag); |
93781325d lockdep: fix fs_r... |
4024 |
fs_reclaim_release(sc.gfp_mask); |
132bb8cfc mm/vmscan: add tr... |
4025 4026 |
trace_mm_vmscan_node_reclaim_end(sc.nr_reclaimed); |
a79311c14 vmscan: bail out ... |
4027 |
return sc.nr_reclaimed >= nr_pages; |
9eeff2395 [PATCH] Zone recl... |
4028 |
} |
179e96395 [PATCH] vmscan: s... |
4029 |
|
a5f5f91da mm: convert zone_... |
4030 |
int node_reclaim(struct pglist_data *pgdat, gfp_t gfp_mask, unsigned int order) |
179e96395 [PATCH] vmscan: s... |
4031 |
{ |
d773ed6b8 mm: test and set ... |
4032 |
int ret; |
179e96395 [PATCH] vmscan: s... |
4033 4034 |
/* |
a5f5f91da mm: convert zone_... |
4035 |
* Node reclaim reclaims unmapped file backed pages and |
0ff38490c [PATCH] zone_recl... |
4036 |
* slab pages if we are over the defined limits. |
34aa1330f [PATCH] zoned vm ... |
4037 |
* |
9614634fe [PATCH] ZVC/zone_... |
4038 4039 |
* 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_... |
4040 4041 |
* 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_... |
4042 |
* unmapped file backed pages. |
179e96395 [PATCH] vmscan: s... |
4043 |
*/ |
a5f5f91da mm: convert zone_... |
4044 |
if (node_pagecache_reclaimable(pgdat) <= pgdat->min_unmapped_pages && |
d42f3245c mm: memcg: conver... |
4045 4046 |
node_page_state_pages(pgdat, NR_SLAB_RECLAIMABLE_B) <= pgdat->min_slab_pages) |
a5f5f91da mm: convert zone_... |
4047 |
return NODE_RECLAIM_FULL; |
179e96395 [PATCH] vmscan: s... |
4048 4049 |
/* |
d773ed6b8 mm: test and set ... |
4050 |
* Do not scan if the allocation should not be delayed. |
179e96395 [PATCH] vmscan: s... |
4051 |
*/ |
d0164adc8 mm, page_alloc: d... |
4052 |
if (!gfpflags_allow_blocking(gfp_mask) || (current->flags & PF_MEMALLOC)) |
a5f5f91da mm: convert zone_... |
4053 |
return NODE_RECLAIM_NOSCAN; |
179e96395 [PATCH] vmscan: s... |
4054 4055 |
/* |
a5f5f91da mm: convert zone_... |
4056 |
* Only run node reclaim on the local node or on nodes that do not |
179e96395 [PATCH] vmscan: s... |
4057 4058 4059 4060 |
* 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_... |
4061 4062 |
if (node_state(pgdat->node_id, N_CPU) && pgdat->node_id != numa_node_id()) return NODE_RECLAIM_NOSCAN; |
d773ed6b8 mm: test and set ... |
4063 |
|
a5f5f91da mm: convert zone_... |
4064 4065 |
if (test_and_set_bit(PGDAT_RECLAIM_LOCKED, &pgdat->flags)) return NODE_RECLAIM_NOSCAN; |
fa5e084e4 vmscan: do not un... |
4066 |
|
a5f5f91da mm: convert zone_... |
4067 4068 |
ret = __node_reclaim(pgdat, gfp_mask, order); clear_bit(PGDAT_RECLAIM_LOCKED, &pgdat->flags); |
d773ed6b8 mm: test and set ... |
4069 |
|
24cf72518 vmscan: count the... |
4070 4071 |
if (!ret) count_vm_event(PGSCAN_ZONE_RECLAIM_FAILED); |
d773ed6b8 mm: test and set ... |
4072 |
return ret; |
179e96395 [PATCH] vmscan: s... |
4073 |
} |
9eeff2395 [PATCH] Zone recl... |
4074 |
#endif |
894bc3104 Unevictable LRU I... |
4075 |
|
89e004ea5 SHM_LOCKED pages ... |
4076 |
/** |
64e3d12f7 mm, drm/i915: mar... |
4077 4078 4079 |
* 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 ... |
4080 |
* |
64e3d12f7 mm, drm/i915: mar... |
4081 4082 4083 |
* 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 ... |
4084 |
*/ |
64e3d12f7 mm, drm/i915: mar... |
4085 |
void check_move_unevictable_pages(struct pagevec *pvec) |
89e004ea5 SHM_LOCKED pages ... |
4086 |
{ |
925b7673c mm: make per-memc... |
4087 |
struct lruvec *lruvec; |
785b99feb mm, vmscan: relea... |
4088 |
struct pglist_data *pgdat = NULL; |
245132643 SHM_UNLOCK: fix U... |
4089 4090 4091 |
int pgscanned = 0; int pgrescued = 0; int i; |
89e004ea5 SHM_LOCKED pages ... |
4092 |
|
64e3d12f7 mm, drm/i915: mar... |
4093 4094 |
for (i = 0; i < pvec->nr; i++) { struct page *page = pvec->pages[i]; |
785b99feb mm, vmscan: relea... |
4095 |
struct pglist_data *pagepgdat = page_pgdat(page); |
8d8869ca5 mm: fix check_mov... |
4096 4097 4098 4099 4100 4101 4102 |
int nr_pages; if (PageTransTail(page)) continue; nr_pages = thp_nr_pages(page); pgscanned += nr_pages; |
89e004ea5 SHM_LOCKED pages ... |
4103 |
|
785b99feb mm, vmscan: relea... |
4104 4105 4106 4107 4108 |
if (pagepgdat != pgdat) { if (pgdat) spin_unlock_irq(&pgdat->lru_lock); pgdat = pagepgdat; spin_lock_irq(&pgdat->lru_lock); |
245132643 SHM_UNLOCK: fix U... |
4109 |
} |
785b99feb mm, vmscan: relea... |
4110 |
lruvec = mem_cgroup_page_lruvec(page, pgdat); |
89e004ea5 SHM_LOCKED pages ... |
4111 |
|
245132643 SHM_UNLOCK: fix U... |
4112 4113 |
if (!PageLRU(page) || !PageUnevictable(page)) continue; |
89e004ea5 SHM_LOCKED pages ... |
4114 |
|
39b5f29ac mm: remove vma ar... |
4115 |
if (page_evictable(page)) { |
245132643 SHM_UNLOCK: fix U... |
4116 |
enum lru_list lru = page_lru_base_type(page); |
309381fea mm: dump page whe... |
4117 |
VM_BUG_ON_PAGE(PageActive(page), page); |
245132643 SHM_UNLOCK: fix U... |
4118 |
ClearPageUnevictable(page); |
fa9add641 mm/memcg: apply a... |
4119 4120 |
del_page_from_lru_list(page, lruvec, LRU_UNEVICTABLE); add_page_to_lru_list(page, lruvec, lru); |
8d8869ca5 mm: fix check_mov... |
4121 |
pgrescued += nr_pages; |
89e004ea5 SHM_LOCKED pages ... |
4122 |
} |
245132643 SHM_UNLOCK: fix U... |
4123 |
} |
89e004ea5 SHM_LOCKED pages ... |
4124 |
|
785b99feb mm, vmscan: relea... |
4125 |
if (pgdat) { |
245132643 SHM_UNLOCK: fix U... |
4126 4127 |
__count_vm_events(UNEVICTABLE_PGRESCUED, pgrescued); __count_vm_events(UNEVICTABLE_PGSCANNED, pgscanned); |
785b99feb mm, vmscan: relea... |
4128 |
spin_unlock_irq(&pgdat->lru_lock); |
89e004ea5 SHM_LOCKED pages ... |
4129 |
} |
89e004ea5 SHM_LOCKED pages ... |
4130 |
} |
64e3d12f7 mm, drm/i915: mar... |
4131 |
EXPORT_SYMBOL_GPL(check_move_unevictable_pages); |