Blame view
mm/vmscan.c
96 KB
1da177e4c
|
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 |
/* * 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. */ #include <linux/mm.h> #include <linux/module.h> |
5a0e3ad6a
|
16 |
#include <linux/gfp.h> |
1da177e4c
|
17 18 19 20 21 |
#include <linux/kernel_stat.h> #include <linux/swap.h> #include <linux/pagemap.h> #include <linux/init.h> #include <linux/highmem.h> |
e129b5c23
|
22 |
#include <linux/vmstat.h> |
1da177e4c
|
23 24 25 26 27 28 |
#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
|
29 30 31 32 33 |
#include <linux/backing-dev.h> #include <linux/rmap.h> #include <linux/topology.h> #include <linux/cpu.h> #include <linux/cpuset.h> |
3e7d34497
|
34 |
#include <linux/compaction.h> |
1da177e4c
|
35 36 |
#include <linux/notifier.h> #include <linux/rwsem.h> |
248a0301e
|
37 |
#include <linux/delay.h> |
3218ae14b
|
38 |
#include <linux/kthread.h> |
7dfb71030
|
39 |
#include <linux/freezer.h> |
66e1707bc
|
40 |
#include <linux/memcontrol.h> |
873b47717
|
41 |
#include <linux/delayacct.h> |
af936a160
|
42 |
#include <linux/sysctl.h> |
929bea7c7
|
43 |
#include <linux/oom.h> |
268bb0ce3
|
44 |
#include <linux/prefetch.h> |
1da177e4c
|
45 46 47 48 49 |
#include <asm/tlbflush.h> #include <asm/div64.h> #include <linux/swapops.h> |
0f8053a50
|
50 |
#include "internal.h" |
33906bc5c
|
51 52 |
#define CREATE_TRACE_POINTS #include <trace/events/vmscan.h> |
1da177e4c
|
53 |
struct scan_control { |
1da177e4c
|
54 55 |
/* Incremented by the number of inactive pages that were scanned */ unsigned long nr_scanned; |
a79311c14
|
56 57 |
/* Number of pages freed so far during a call to shrink_zones() */ unsigned long nr_reclaimed; |
22fba3354
|
58 59 |
/* How many pages shrink_list() should reclaim */ unsigned long nr_to_reclaim; |
7b51755c3
|
60 |
unsigned long hibernation_mode; |
1da177e4c
|
61 |
/* This context's GFP mask */ |
6daa0e286
|
62 |
gfp_t gfp_mask; |
1da177e4c
|
63 64 |
int may_writepage; |
a6dc60f89
|
65 66 |
/* Can mapped pages be reclaimed? */ int may_unmap; |
f1fd1067e
|
67 |
|
2e2e42598
|
68 69 |
/* Can pages be swapped as part of reclaim? */ int may_swap; |
5ad333eb6
|
70 |
int order; |
66e1707bc
|
71 |
|
5f53e7629
|
72 |
/* |
f16015fbf
|
73 74 75 76 |
* 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
|
77 |
|
327c0e968
|
78 79 80 81 82 |
/* * Nodemask of nodes allowed by the caller. If NULL, all nodes * are scanned. */ nodemask_t *nodemask; |
1da177e4c
|
83 |
}; |
f16015fbf
|
84 85 86 87 |
struct mem_cgroup_zone { struct mem_cgroup *mem_cgroup; struct zone *zone; }; |
1da177e4c
|
88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 |
#define lru_to_page(_head) (list_entry((_head)->prev, struct page, lru)) #ifdef ARCH_HAS_PREFETCH #define prefetch_prev_lru_page(_page, _base, _field) \ do { \ if ((_page)->lru.prev != _base) { \ struct page *prev; \ \ prev = lru_to_page(&(_page->lru)); \ prefetch(&prev->_field); \ } \ } while (0) #else #define prefetch_prev_lru_page(_page, _base, _field) do { } while (0) #endif #ifdef ARCH_HAS_PREFETCHW #define prefetchw_prev_lru_page(_page, _base, _field) \ do { \ if ((_page)->lru.prev != _base) { \ struct page *prev; \ \ prev = lru_to_page(&(_page->lru)); \ prefetchw(&prev->_field); \ } \ } while (0) #else #define prefetchw_prev_lru_page(_page, _base, _field) do { } while (0) #endif /* * From 0 .. 100. Higher means more swappy. */ int vm_swappiness = 60; |
bd1e22b8e
|
122 |
long vm_total_pages; /* The total number of pages which the VM controls */ |
1da177e4c
|
123 124 125 |
static LIST_HEAD(shrinker_list); static DECLARE_RWSEM(shrinker_rwsem); |
00f0b8259
|
126 |
#ifdef CONFIG_CGROUP_MEM_RES_CTLR |
89b5fae53
|
127 128 |
static bool global_reclaim(struct scan_control *sc) { |
f16015fbf
|
129 |
return !sc->target_mem_cgroup; |
89b5fae53
|
130 |
} |
f16015fbf
|
131 |
static bool scanning_global_lru(struct mem_cgroup_zone *mz) |
89b5fae53
|
132 |
{ |
f16015fbf
|
133 |
return !mz->mem_cgroup; |
89b5fae53
|
134 |
} |
91a45470f
|
135 |
#else |
89b5fae53
|
136 137 138 139 |
static bool global_reclaim(struct scan_control *sc) { return true; } |
f16015fbf
|
140 |
static bool scanning_global_lru(struct mem_cgroup_zone *mz) |
89b5fae53
|
141 142 143 |
{ return true; } |
91a45470f
|
144 |
#endif |
f16015fbf
|
145 |
static struct zone_reclaim_stat *get_reclaim_stat(struct mem_cgroup_zone *mz) |
6e9015716
|
146 |
{ |
f16015fbf
|
147 148 |
if (!scanning_global_lru(mz)) return mem_cgroup_get_reclaim_stat(mz->mem_cgroup, mz->zone); |
3e2f41f1f
|
149 |
|
f16015fbf
|
150 |
return &mz->zone->reclaim_stat; |
6e9015716
|
151 |
} |
f16015fbf
|
152 153 |
static unsigned long zone_nr_lru_pages(struct mem_cgroup_zone *mz, enum lru_list lru) |
c9f299d98
|
154 |
{ |
f16015fbf
|
155 156 157 158 159 |
if (!scanning_global_lru(mz)) return mem_cgroup_zone_nr_lru_pages(mz->mem_cgroup, zone_to_nid(mz->zone), zone_idx(mz->zone), BIT(lru)); |
a3d8e0549
|
160 |
|
f16015fbf
|
161 |
return zone_page_state(mz->zone, NR_LRU_BASE + lru); |
c9f299d98
|
162 |
} |
1da177e4c
|
163 164 165 |
/* * Add a shrinker callback to be called from the vm */ |
8e1f936b7
|
166 |
void register_shrinker(struct shrinker *shrinker) |
1da177e4c
|
167 |
{ |
83aeeada7
|
168 |
atomic_long_set(&shrinker->nr_in_batch, 0); |
8e1f936b7
|
169 170 171 |
down_write(&shrinker_rwsem); list_add_tail(&shrinker->list, &shrinker_list); up_write(&shrinker_rwsem); |
1da177e4c
|
172 |
} |
8e1f936b7
|
173 |
EXPORT_SYMBOL(register_shrinker); |
1da177e4c
|
174 175 176 177 |
/* * Remove one */ |
8e1f936b7
|
178 |
void unregister_shrinker(struct shrinker *shrinker) |
1da177e4c
|
179 180 181 182 |
{ down_write(&shrinker_rwsem); list_del(&shrinker->list); up_write(&shrinker_rwsem); |
1da177e4c
|
183 |
} |
8e1f936b7
|
184 |
EXPORT_SYMBOL(unregister_shrinker); |
1da177e4c
|
185 |
|
1495f230f
|
186 187 188 189 190 191 192 |
static inline int do_shrinker_shrink(struct shrinker *shrinker, struct shrink_control *sc, unsigned long nr_to_scan) { sc->nr_to_scan = nr_to_scan; return (*shrinker->shrink)(shrinker, sc); } |
1da177e4c
|
193 194 195 196 197 198 199 200 201 |
#define SHRINK_BATCH 128 /* * Call the shrink functions to age shrinkable caches * * Here we assume it costs one seek to replace a lru page and that it also * takes a seek to recreate a cache object. With this in mind we age equal * percentages of the lru and ageable caches. This should balance the seeks * generated by these structures. * |
183ff22bb
|
202 |
* If the vm encountered mapped pages on the LRU it increase the pressure on |
1da177e4c
|
203 204 205 206 207 208 209 |
* slab to avoid swapping. * * We do weird things to avoid (scanned*seeks*entries) overflowing 32 bits. * * `lru_pages' represents the number of on-LRU pages in all the zones which * are eligible for the caller's allocation attempt. It is used for balancing * slab reclaim versus page reclaim. |
b15e0905f
|
210 211 |
* * Returns the number of slab objects which we shrunk. |
1da177e4c
|
212 |
*/ |
a09ed5e00
|
213 |
unsigned long shrink_slab(struct shrink_control *shrink, |
1495f230f
|
214 |
unsigned long nr_pages_scanned, |
a09ed5e00
|
215 |
unsigned long lru_pages) |
1da177e4c
|
216 217 |
{ struct shrinker *shrinker; |
69e05944a
|
218 |
unsigned long ret = 0; |
1da177e4c
|
219 |
|
1495f230f
|
220 221 |
if (nr_pages_scanned == 0) nr_pages_scanned = SWAP_CLUSTER_MAX; |
1da177e4c
|
222 |
|
f06590bd7
|
223 224 225 226 227 |
if (!down_read_trylock(&shrinker_rwsem)) { /* Assume we'll be able to shrink next time */ ret = 1; goto out; } |
1da177e4c
|
228 229 230 |
list_for_each_entry(shrinker, &shrinker_list, list) { unsigned long long delta; |
635697c66
|
231 232 |
long total_scan; long max_pass; |
095760730
|
233 |
int shrink_ret = 0; |
acf92b485
|
234 235 |
long nr; long new_nr; |
e9299f505
|
236 237 |
long batch_size = shrinker->batch ? shrinker->batch : SHRINK_BATCH; |
1da177e4c
|
238 |
|
635697c66
|
239 240 241 |
max_pass = do_shrinker_shrink(shrinker, shrink, 0); if (max_pass <= 0) continue; |
acf92b485
|
242 243 244 245 246 |
/* * 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. */ |
83aeeada7
|
247 |
nr = atomic_long_xchg(&shrinker->nr_in_batch, 0); |
acf92b485
|
248 249 |
total_scan = nr; |
1495f230f
|
250 |
delta = (4 * nr_pages_scanned) / shrinker->seeks; |
ea164d73a
|
251 |
delta *= max_pass; |
1da177e4c
|
252 |
do_div(delta, lru_pages + 1); |
acf92b485
|
253 254 |
total_scan += delta; if (total_scan < 0) { |
88c3bd707
|
255 256 257 |
printk(KERN_ERR "shrink_slab: %pF negative objects to " "delete nr=%ld ", |
acf92b485
|
258 259 |
shrinker->shrink, total_scan); total_scan = max_pass; |
ea164d73a
|
260 261 262 |
} /* |
3567b59aa
|
263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 |
* 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 >>> * max_pass. This is bad for sustaining a working set in * memory. * * Hence only allow the shrinker to scan the entire cache when * a large delta change is calculated directly. */ if (delta < max_pass / 4) total_scan = min(total_scan, max_pass / 2); /* |
ea164d73a
|
278 279 280 281 |
* Avoid risking looping forever due to too large nr value: * never try to free more than twice the estimate number of * freeable entries. */ |
acf92b485
|
282 283 |
if (total_scan > max_pass * 2) total_scan = max_pass * 2; |
1da177e4c
|
284 |
|
acf92b485
|
285 |
trace_mm_shrink_slab_start(shrinker, shrink, nr, |
095760730
|
286 287 |
nr_pages_scanned, lru_pages, max_pass, delta, total_scan); |
e9299f505
|
288 |
while (total_scan >= batch_size) { |
b15e0905f
|
289 |
int nr_before; |
1da177e4c
|
290 |
|
1495f230f
|
291 292 |
nr_before = do_shrinker_shrink(shrinker, shrink, 0); shrink_ret = do_shrinker_shrink(shrinker, shrink, |
e9299f505
|
293 |
batch_size); |
1da177e4c
|
294 295 |
if (shrink_ret == -1) break; |
b15e0905f
|
296 297 |
if (shrink_ret < nr_before) ret += nr_before - shrink_ret; |
e9299f505
|
298 299 |
count_vm_events(SLABS_SCANNED, batch_size); total_scan -= batch_size; |
1da177e4c
|
300 301 302 |
cond_resched(); } |
acf92b485
|
303 304 305 306 307 |
/* * 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. */ |
83aeeada7
|
308 309 310 311 312 |
if (total_scan > 0) new_nr = atomic_long_add_return(total_scan, &shrinker->nr_in_batch); else new_nr = atomic_long_read(&shrinker->nr_in_batch); |
acf92b485
|
313 314 |
trace_mm_shrink_slab_end(shrinker, shrink_ret, nr, new_nr); |
1da177e4c
|
315 316 |
} up_read(&shrinker_rwsem); |
f06590bd7
|
317 318 |
out: cond_resched(); |
b15e0905f
|
319 |
return ret; |
1da177e4c
|
320 |
} |
1da177e4c
|
321 322 |
static inline int is_page_cache_freeable(struct page *page) { |
ceddc3a52
|
323 324 325 326 327 |
/* * A freeable page cache page is referenced only by the caller * that isolated the page, the page cache radix tree and * optional buffer heads at page->private. */ |
edcf4748c
|
328 |
return page_count(page) - page_has_private(page) == 2; |
1da177e4c
|
329 |
} |
7d3579e8e
|
330 331 |
static int may_write_to_queue(struct backing_dev_info *bdi, struct scan_control *sc) |
1da177e4c
|
332 |
{ |
930d91525
|
333 |
if (current->flags & PF_SWAPWRITE) |
1da177e4c
|
334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 |
return 1; if (!bdi_write_congested(bdi)) return 1; if (bdi == current->backing_dev_info) 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
|
357 |
lock_page(page); |
3e9f45bd1
|
358 359 |
if (page_mapping(page) == mapping) mapping_set_error(mapping, error); |
1da177e4c
|
360 361 |
unlock_page(page); } |
04e62a29b
|
362 363 364 365 366 367 368 369 370 371 372 |
/* 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
|
373 |
/* |
1742f19fa
|
374 375 |
* pageout is called by shrink_page_list() for each dirty page. * Calls ->writepage(). |
1da177e4c
|
376 |
*/ |
c661b078f
|
377 |
static pageout_t pageout(struct page *page, struct address_space *mapping, |
7d3579e8e
|
378 |
struct scan_control *sc) |
1da177e4c
|
379 380 381 382 383 384 385 386 |
{ /* * 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. * |
6aceb53be
|
387 |
* If this process is currently in __generic_file_aio_write() against |
1da177e4c
|
388 389 390 391 392 393 394 |
* 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
|
395 396 397 398 399 400 401 402 |
*/ 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
|
403 |
if (page_has_private(page)) { |
1da177e4c
|
404 405 |
if (try_to_free_buffers(page)) { ClearPageDirty(page); |
d40cee245
|
406 407 |
printk("%s: orphaned page ", __func__); |
1da177e4c
|
408 409 410 411 412 413 414 |
return PAGE_CLEAN; } } return PAGE_KEEP; } if (mapping->a_ops->writepage == NULL) return PAGE_ACTIVATE; |
0e093d997
|
415 |
if (!may_write_to_queue(mapping->backing_dev_info, sc)) |
1da177e4c
|
416 417 418 419 420 421 422 |
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
|
423 424 |
.range_start = 0, .range_end = LLONG_MAX, |
1da177e4c
|
425 426 427 428 429 430 431 |
.for_reclaim = 1, }; SetPageReclaim(page); res = mapping->a_ops->writepage(page, &wbc); if (res < 0) handle_write_error(mapping, page, res); |
994fc28c7
|
432 |
if (res == AOP_WRITEPAGE_ACTIVATE) { |
1da177e4c
|
433 434 435 |
ClearPageReclaim(page); return PAGE_ACTIVATE; } |
c661b078f
|
436 |
|
1da177e4c
|
437 438 439 440 |
if (!PageWriteback(page)) { /* synchronous write or broken a_ops? */ ClearPageReclaim(page); } |
23b9da55c
|
441 |
trace_mm_vmscan_writepage(page, trace_reclaim_flags(page)); |
e129b5c23
|
442 |
inc_zone_page_state(page, NR_VMSCAN_WRITE); |
1da177e4c
|
443 444 445 446 447 |
return PAGE_SUCCESS; } return PAGE_CLEAN; } |
a649fd927
|
448 |
/* |
e286781d5
|
449 450 |
* Same as remove_mapping, but if the page is removed from the mapping, it * gets returned with a refcount of 0. |
a649fd927
|
451 |
*/ |
e286781d5
|
452 |
static int __remove_mapping(struct address_space *mapping, struct page *page) |
49d2e9cc4
|
453 |
{ |
28e4d965e
|
454 455 |
BUG_ON(!PageLocked(page)); BUG_ON(mapping != page_mapping(page)); |
49d2e9cc4
|
456 |
|
19fd62312
|
457 |
spin_lock_irq(&mapping->tree_lock); |
49d2e9cc4
|
458 |
/* |
0fd0e6b05
|
459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 |
* 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 * load is not satisfied before that of page->_count. * * Note that if SetPageDirty is always performed via set_page_dirty, * and thus under tree_lock, then this ordering is not required. |
49d2e9cc4
|
482 |
*/ |
e286781d5
|
483 |
if (!page_freeze_refs(page, 2)) |
49d2e9cc4
|
484 |
goto cannot_free; |
e286781d5
|
485 486 487 |
/* note: atomic_cmpxchg in page_freeze_refs provides the smp_rmb */ if (unlikely(PageDirty(page))) { page_unfreeze_refs(page, 2); |
49d2e9cc4
|
488 |
goto cannot_free; |
e286781d5
|
489 |
} |
49d2e9cc4
|
490 491 492 493 |
if (PageSwapCache(page)) { swp_entry_t swap = { .val = page_private(page) }; __delete_from_swap_cache(page); |
19fd62312
|
494 |
spin_unlock_irq(&mapping->tree_lock); |
cb4b86ba4
|
495 |
swapcache_free(swap, page); |
e286781d5
|
496 |
} else { |
6072d13c4
|
497 498 499 |
void (*freepage)(struct page *); freepage = mapping->a_ops->freepage; |
e64a782fe
|
500 |
__delete_from_page_cache(page); |
19fd62312
|
501 |
spin_unlock_irq(&mapping->tree_lock); |
e767e0561
|
502 |
mem_cgroup_uncharge_cache_page(page); |
6072d13c4
|
503 504 505 |
if (freepage != NULL) freepage(page); |
49d2e9cc4
|
506 |
} |
49d2e9cc4
|
507 508 509 |
return 1; cannot_free: |
19fd62312
|
510 |
spin_unlock_irq(&mapping->tree_lock); |
49d2e9cc4
|
511 512 |
return 0; } |
1da177e4c
|
513 |
/* |
e286781d5
|
514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 |
* 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) { if (__remove_mapping(mapping, page)) { /* * Unfreezing the refcount with 1 rather than 2 effectively * drops the pagecache ref for us without requiring another * atomic operation. */ page_unfreeze_refs(page, 1); return 1; } return 0; } |
894bc3104
|
532 533 534 535 536 537 538 539 540 |
/** * 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
|
541 542 543 544 |
void putback_lru_page(struct page *page) { int lru; int active = !!TestClearPageActive(page); |
bbfd28eee
|
545 |
int was_unevictable = PageUnevictable(page); |
894bc3104
|
546 547 548 549 550 551 552 553 554 555 556 557 558 |
VM_BUG_ON(PageLRU(page)); redo: ClearPageUnevictable(page); if (page_evictable(page, NULL)) { /* * For evictable pages, we can use the cache. * In event of a race, worst case is we end up with an * unevictable page on [in]active list. * We know how to handle that. */ |
401a8e1c1
|
559 |
lru = active + page_lru_base_type(page); |
894bc3104
|
560 561 562 563 564 565 566 567 |
lru_cache_add_lru(page, lru); } else { /* * Put unevictable pages directly on zone's unevictable * list. */ lru = LRU_UNEVICTABLE; add_page_to_unevictable_list(page); |
6a7b95481
|
568 |
/* |
21ee9f398
|
569 570 571 |
* When racing with an mlock or AS_UNEVICTABLE clearing * (page is unlocked) make sure that if the other thread * does not observe our setting of PG_lru and fails |
245132643
|
572 |
* isolation/check_move_unevictable_pages, |
21ee9f398
|
573 |
* we see PG_mlocked/AS_UNEVICTABLE cleared below and move |
6a7b95481
|
574 575 |
* the page back to the evictable list. * |
21ee9f398
|
576 |
* The other side is TestClearPageMlocked() or shmem_lock(). |
6a7b95481
|
577 578 |
*/ smp_mb(); |
894bc3104
|
579 |
} |
894bc3104
|
580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 |
/* * page's status can change while we move it among lru. If an evictable * page is on unevictable list, it never be freed. To avoid that, * check after we added it to the list, again. */ if (lru == LRU_UNEVICTABLE && page_evictable(page, NULL)) { if (!isolate_lru_page(page)) { put_page(page); goto redo; } /* This means someone else dropped this page from LRU * So, it will be freed or putback to LRU again. There is * nothing to do here. */ } |
bbfd28eee
|
596 597 598 599 |
if (was_unevictable && lru != LRU_UNEVICTABLE) count_vm_event(UNEVICTABLE_PGRESCUED); else if (!was_unevictable && lru == LRU_UNEVICTABLE) count_vm_event(UNEVICTABLE_PGCULLED); |
894bc3104
|
600 601 |
put_page(page); /* drop ref from isolate */ } |
dfc8d636c
|
602 603 604 |
enum page_references { PAGEREF_RECLAIM, PAGEREF_RECLAIM_CLEAN, |
645747462
|
605 |
PAGEREF_KEEP, |
dfc8d636c
|
606 607 608 609 |
PAGEREF_ACTIVATE, }; static enum page_references page_check_references(struct page *page, |
f16015fbf
|
610 |
struct mem_cgroup_zone *mz, |
dfc8d636c
|
611 612 |
struct scan_control *sc) { |
645747462
|
613 |
int referenced_ptes, referenced_page; |
dfc8d636c
|
614 |
unsigned long vm_flags; |
dfc8d636c
|
615 |
|
c3ac9a8ad
|
616 617 |
referenced_ptes = page_referenced(page, 1, sc->target_mem_cgroup, &vm_flags); |
645747462
|
618 |
referenced_page = TestClearPageReferenced(page); |
dfc8d636c
|
619 |
|
dfc8d636c
|
620 621 622 623 624 625 |
/* * 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
|
626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 |
if (referenced_ptes) { if (PageAnon(page)) return PAGEREF_ACTIVATE; /* * All mapped pages start out with page table * references from the instantiating fault, so we need * to look twice if a mapped file page is used more * than once. * * Mark it and spare it for another trip around the * inactive list. Another page table reference will * lead to its activation. * * Note: the mark is set for activated pages as well * so that recently deactivated but used pages are * quickly recovered. */ SetPageReferenced(page); |
34dbc67a6
|
644 |
if (referenced_page || referenced_ptes > 1) |
645747462
|
645 |
return PAGEREF_ACTIVATE; |
c909e9936
|
646 647 648 649 650 |
/* * Activate file-backed executable pages after first usage. */ if (vm_flags & VM_EXEC) return PAGEREF_ACTIVATE; |
645747462
|
651 652 |
return PAGEREF_KEEP; } |
dfc8d636c
|
653 654 |
/* Reclaim if clean, defer dirty pages to writeback */ |
2e30244a7
|
655 |
if (referenced_page && !PageSwapBacked(page)) |
645747462
|
656 657 658 |
return PAGEREF_RECLAIM_CLEAN; return PAGEREF_RECLAIM; |
dfc8d636c
|
659 |
} |
e286781d5
|
660 |
/* |
1742f19fa
|
661 |
* shrink_page_list() returns the number of reclaimed pages |
1da177e4c
|
662 |
*/ |
1742f19fa
|
663 |
static unsigned long shrink_page_list(struct list_head *page_list, |
f16015fbf
|
664 |
struct mem_cgroup_zone *mz, |
f84f6e2b0
|
665 |
struct scan_control *sc, |
92df3a723
|
666 667 668 |
int priority, unsigned long *ret_nr_dirty, unsigned long *ret_nr_writeback) |
1da177e4c
|
669 670 |
{ LIST_HEAD(ret_pages); |
abe4c3b50
|
671 |
LIST_HEAD(free_pages); |
1da177e4c
|
672 |
int pgactivate = 0; |
0e093d997
|
673 674 |
unsigned long nr_dirty = 0; unsigned long nr_congested = 0; |
05ff51376
|
675 |
unsigned long nr_reclaimed = 0; |
92df3a723
|
676 |
unsigned long nr_writeback = 0; |
1da177e4c
|
677 678 |
cond_resched(); |
1da177e4c
|
679 |
while (!list_empty(page_list)) { |
dfc8d636c
|
680 |
enum page_references references; |
1da177e4c
|
681 682 683 |
struct address_space *mapping; struct page *page; int may_enter_fs; |
1da177e4c
|
684 685 686 687 688 |
cond_resched(); page = lru_to_page(page_list); list_del(&page->lru); |
529ae9aaa
|
689 |
if (!trylock_page(page)) |
1da177e4c
|
690 |
goto keep; |
725d704ec
|
691 |
VM_BUG_ON(PageActive(page)); |
f16015fbf
|
692 |
VM_BUG_ON(page_zone(page) != mz->zone); |
1da177e4c
|
693 694 |
sc->nr_scanned++; |
80e434260
|
695 |
|
b291f0003
|
696 697 |
if (unlikely(!page_evictable(page, NULL))) goto cull_mlocked; |
894bc3104
|
698 |
|
a6dc60f89
|
699 |
if (!sc->may_unmap && page_mapped(page)) |
80e434260
|
700 |
goto keep_locked; |
1da177e4c
|
701 702 703 |
/* Double the slab pressure for mapped and swapcache pages */ if (page_mapped(page) || PageSwapCache(page)) sc->nr_scanned++; |
c661b078f
|
704 705 706 707 |
may_enter_fs = (sc->gfp_mask & __GFP_FS) || (PageSwapCache(page) && (sc->gfp_mask & __GFP_IO)); if (PageWriteback(page)) { |
92df3a723
|
708 |
nr_writeback++; |
41ac1999c
|
709 710 |
unlock_page(page); goto keep; |
c661b078f
|
711 |
} |
1da177e4c
|
712 |
|
f16015fbf
|
713 |
references = page_check_references(page, mz, sc); |
dfc8d636c
|
714 715 |
switch (references) { case PAGEREF_ACTIVATE: |
1da177e4c
|
716 |
goto activate_locked; |
645747462
|
717 718 |
case PAGEREF_KEEP: goto keep_locked; |
dfc8d636c
|
719 720 721 722 |
case PAGEREF_RECLAIM: case PAGEREF_RECLAIM_CLEAN: ; /* try to reclaim the page below */ } |
1da177e4c
|
723 |
|
1da177e4c
|
724 725 726 727 |
/* * Anonymous process memory has backing store? * Try to allocate it some swap space here. */ |
b291f0003
|
728 |
if (PageAnon(page) && !PageSwapCache(page)) { |
63eb6b93c
|
729 730 |
if (!(sc->gfp_mask & __GFP_IO)) goto keep_locked; |
ac47b003d
|
731 |
if (!add_to_swap(page)) |
1da177e4c
|
732 |
goto activate_locked; |
63eb6b93c
|
733 |
may_enter_fs = 1; |
b291f0003
|
734 |
} |
1da177e4c
|
735 736 |
mapping = page_mapping(page); |
1da177e4c
|
737 738 739 740 741 742 |
/* * The page is mapped into the page tables of one or more * processes. Try to unmap it here. */ if (page_mapped(page) && mapping) { |
14fa31b89
|
743 |
switch (try_to_unmap(page, TTU_UNMAP)) { |
1da177e4c
|
744 745 746 747 |
case SWAP_FAIL: goto activate_locked; case SWAP_AGAIN: goto keep_locked; |
b291f0003
|
748 749 |
case SWAP_MLOCK: goto cull_mlocked; |
1da177e4c
|
750 751 752 753 754 755 |
case SWAP_SUCCESS: ; /* try to free the page below */ } } if (PageDirty(page)) { |
0e093d997
|
756 |
nr_dirty++; |
ee72886d8
|
757 758 |
/* * Only kswapd can writeback filesystem pages to |
f84f6e2b0
|
759 760 |
* avoid risk of stack overflow but do not writeback * unless under significant pressure. |
ee72886d8
|
761 |
*/ |
f84f6e2b0
|
762 763 |
if (page_is_file_cache(page) && (!current_is_kswapd() || priority >= DEF_PRIORITY - 2)) { |
49ea7eb65
|
764 765 766 767 768 769 770 771 |
/* * Immediately reclaim when written back. * Similar in principal to deactivate_page() * except we already have the page isolated * and know it's dirty */ inc_zone_page_state(page, NR_VMSCAN_IMMEDIATE); SetPageReclaim(page); |
ee72886d8
|
772 773 |
goto keep_locked; } |
dfc8d636c
|
774 |
if (references == PAGEREF_RECLAIM_CLEAN) |
1da177e4c
|
775 |
goto keep_locked; |
4dd4b9202
|
776 |
if (!may_enter_fs) |
1da177e4c
|
777 |
goto keep_locked; |
52a8363ea
|
778 |
if (!sc->may_writepage) |
1da177e4c
|
779 780 781 |
goto keep_locked; /* Page is dirty, try to write it out here */ |
7d3579e8e
|
782 |
switch (pageout(page, mapping, sc)) { |
1da177e4c
|
783 |
case PAGE_KEEP: |
0e093d997
|
784 |
nr_congested++; |
1da177e4c
|
785 786 787 788 |
goto keep_locked; case PAGE_ACTIVATE: goto activate_locked; case PAGE_SUCCESS: |
7d3579e8e
|
789 |
if (PageWriteback(page)) |
41ac1999c
|
790 |
goto keep; |
7d3579e8e
|
791 |
if (PageDirty(page)) |
1da177e4c
|
792 |
goto keep; |
7d3579e8e
|
793 |
|
1da177e4c
|
794 795 796 797 |
/* * A synchronous write - probably a ramdisk. Go * ahead and try to reclaim the page. */ |
529ae9aaa
|
798 |
if (!trylock_page(page)) |
1da177e4c
|
799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 |
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
|
818 |
* will do this, as well as the blockdev mapping. |
1da177e4c
|
819 820 821 822 823 824 825 826 827 828 |
* 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
|
829 |
if (page_has_private(page)) { |
1da177e4c
|
830 831 |
if (!try_to_release_page(page, sc->gfp_mask)) goto activate_locked; |
e286781d5
|
832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 |
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
|
848 |
} |
e286781d5
|
849 |
if (!mapping || !__remove_mapping(mapping, page)) |
49d2e9cc4
|
850 |
goto keep_locked; |
1da177e4c
|
851 |
|
a978d6f52
|
852 853 854 855 856 857 858 859 |
/* * At this point, we have no other references and there is * no way to pick any more up (removed from LRU, removed * from pagecache). Can use non-atomic bitops now (and * we obviously don't have to worry about waking up a process * waiting on the page lock, because there are no references. */ __clear_page_locked(page); |
e286781d5
|
860 |
free_it: |
05ff51376
|
861 |
nr_reclaimed++; |
abe4c3b50
|
862 863 864 865 866 867 |
/* * Is there need to periodically free_page_list? It would * appear not as the counts should be low */ list_add(&page->lru, &free_pages); |
1da177e4c
|
868 |
continue; |
b291f0003
|
869 |
cull_mlocked: |
63d6c5ad7
|
870 871 |
if (PageSwapCache(page)) try_to_free_swap(page); |
b291f0003
|
872 873 874 |
unlock_page(page); putback_lru_page(page); continue; |
1da177e4c
|
875 |
activate_locked: |
68a22394c
|
876 877 |
/* Not a candidate for swapping, so reclaim swap space. */ if (PageSwapCache(page) && vm_swap_full()) |
a2c43eed8
|
878 |
try_to_free_swap(page); |
894bc3104
|
879 |
VM_BUG_ON(PageActive(page)); |
1da177e4c
|
880 881 882 883 884 885 |
SetPageActive(page); pgactivate++; keep_locked: unlock_page(page); keep: list_add(&page->lru, &ret_pages); |
b291f0003
|
886 |
VM_BUG_ON(PageLRU(page) || PageUnevictable(page)); |
1da177e4c
|
887 |
} |
abe4c3b50
|
888 |
|
0e093d997
|
889 890 891 892 893 894 |
/* * Tag a zone as congested if all the dirty pages encountered were * backed by a congested BDI. In this case, reclaimers should just * back off and wait for congestion to clear because further reclaim * will encounter the same problem */ |
89b5fae53
|
895 |
if (nr_dirty && nr_dirty == nr_congested && global_reclaim(sc)) |
f16015fbf
|
896 |
zone_set_flag(mz->zone, ZONE_CONGESTED); |
0e093d997
|
897 |
|
cc59850ef
|
898 |
free_hot_cold_page_list(&free_pages, 1); |
abe4c3b50
|
899 |
|
1da177e4c
|
900 |
list_splice(&ret_pages, page_list); |
f8891e5e1
|
901 |
count_vm_events(PGACTIVATE, pgactivate); |
92df3a723
|
902 903 |
*ret_nr_dirty += nr_dirty; *ret_nr_writeback += nr_writeback; |
05ff51376
|
904 |
return nr_reclaimed; |
1da177e4c
|
905 |
} |
5ad333eb6
|
906 907 908 909 910 911 912 913 914 915 |
/* * 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. */ |
4356f21d0
|
916 |
int __isolate_lru_page(struct page *page, isolate_mode_t mode, int file) |
5ad333eb6
|
917 |
{ |
4356f21d0
|
918 |
bool all_lru_mode; |
5ad333eb6
|
919 920 921 922 923 |
int ret = -EINVAL; /* Only take pages on the LRU. */ if (!PageLRU(page)) return ret; |
4356f21d0
|
924 925 |
all_lru_mode = (mode & (ISOLATE_ACTIVE|ISOLATE_INACTIVE)) == (ISOLATE_ACTIVE|ISOLATE_INACTIVE); |
5ad333eb6
|
926 927 928 929 930 |
/* * When checking the active state, we need to be sure we are * dealing with comparible boolean values. Take the logical not * of each. */ |
4356f21d0
|
931 |
if (!all_lru_mode && !PageActive(page) != !(mode & ISOLATE_ACTIVE)) |
5ad333eb6
|
932 |
return ret; |
4356f21d0
|
933 |
if (!all_lru_mode && !!page_is_file_cache(page) != file) |
4f98a2fee
|
934 |
return ret; |
c53919adc
|
935 |
/* Do not give back unevictable pages for compaction */ |
894bc3104
|
936 937 |
if (PageUnevictable(page)) return ret; |
5ad333eb6
|
938 |
ret = -EBUSY; |
08e552c69
|
939 |
|
c82449352
|
940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 |
/* * 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. * * ISOLATE_CLEAN means that only clean pages should be isolated. This * is used by reclaim when it is cannot write to backing storage * * ISOLATE_ASYNC_MIGRATE is used to indicate that it only wants to pages * that it is possible to migrate without blocking */ if (mode & (ISOLATE_CLEAN|ISOLATE_ASYNC_MIGRATE)) { /* All the caller can do on PageWriteback is block */ if (PageWriteback(page)) return ret; if (PageDirty(page)) { struct address_space *mapping; /* ISOLATE_CLEAN means only clean pages */ if (mode & ISOLATE_CLEAN) return ret; /* * Only pages without mappings or that have a * ->migratepage callback are possible to migrate * without blocking */ mapping = page_mapping(page); if (mapping && !mapping->a_ops->migratepage) return ret; } } |
39deaf858
|
973 |
|
f80c06736
|
974 975 |
if ((mode & ISOLATE_UNMAPPED) && page_mapped(page)) return ret; |
5ad333eb6
|
976 977 978 979 980 981 982 983 984 985 986 987 |
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; } |
49d2e9cc4
|
988 |
/* |
1da177e4c
|
989 990 991 992 993 994 995 996 997 998 |
* zone->lru_lock is heavily contended. Some of the functions that * 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. * * @nr_to_scan: The number of pages to look through on the list. |
f626012db
|
999 |
* @mz: The mem_cgroup_zone to pull pages from. |
1da177e4c
|
1000 |
* @dst: The temp list to put pages on to. |
f626012db
|
1001 |
* @nr_scanned: The number of pages that were scanned. |
fe2c2a106
|
1002 |
* @sc: The scan_control struct for this reclaim session |
5ad333eb6
|
1003 |
* @mode: One of the LRU isolation modes |
f626012db
|
1004 |
* @active: True [1] if isolating active pages |
4f98a2fee
|
1005 |
* @file: True [1] if isolating file [!anon] pages |
1da177e4c
|
1006 1007 1008 |
* * returns how many pages were moved onto *@dst. */ |
69e05944a
|
1009 |
static unsigned long isolate_lru_pages(unsigned long nr_to_scan, |
f626012db
|
1010 |
struct mem_cgroup_zone *mz, struct list_head *dst, |
fe2c2a106
|
1011 1012 |
unsigned long *nr_scanned, struct scan_control *sc, isolate_mode_t mode, int active, int file) |
1da177e4c
|
1013 |
{ |
f626012db
|
1014 1015 |
struct lruvec *lruvec; struct list_head *src; |
69e05944a
|
1016 |
unsigned long nr_taken = 0; |
c9b02d970
|
1017 |
unsigned long scan; |
f626012db
|
1018 1019 1020 1021 1022 1023 1024 1025 |
int lru = LRU_BASE; lruvec = mem_cgroup_zone_lruvec(mz->zone, mz->mem_cgroup); if (active) lru += LRU_ACTIVE; if (file) lru += LRU_FILE; src = &lruvec->lists[lru]; |
1da177e4c
|
1026 |
|
c9b02d970
|
1027 |
for (scan = 0; scan < nr_to_scan && !list_empty(src); scan++) { |
5ad333eb6
|
1028 |
struct page *page; |
5ad333eb6
|
1029 |
|
1da177e4c
|
1030 1031 |
page = lru_to_page(src); prefetchw_prev_lru_page(page, src, flags); |
725d704ec
|
1032 |
VM_BUG_ON(!PageLRU(page)); |
8d438f96d
|
1033 |
|
4f98a2fee
|
1034 |
switch (__isolate_lru_page(page, mode, file)) { |
5ad333eb6
|
1035 |
case 0: |
925b7673c
|
1036 |
mem_cgroup_lru_del(page); |
5ad333eb6
|
1037 |
list_move(&page->lru, dst); |
2c888cfbc
|
1038 |
nr_taken += hpage_nr_pages(page); |
5ad333eb6
|
1039 1040 1041 1042 1043 1044 |
break; case -EBUSY: /* else it is being freed elsewhere */ list_move(&page->lru, src); continue; |
46453a6e1
|
1045 |
|
5ad333eb6
|
1046 1047 1048 |
default: BUG(); } |
1da177e4c
|
1049 |
} |
f626012db
|
1050 |
*nr_scanned = scan; |
a8a94d151
|
1051 |
|
fe2c2a106
|
1052 |
trace_mm_vmscan_lru_isolate(sc->order, |
a8a94d151
|
1053 1054 |
nr_to_scan, scan, nr_taken, |
ea4d349ff
|
1055 |
mode, file); |
1da177e4c
|
1056 1057 |
return nr_taken; } |
62695a84e
|
1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 |
/** * 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
|
1069 1070 1071 |
* 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
|
1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 |
* * The vmstat statistic corresponding to the list on which the page was * found will be decremented. * * Restrictions: * (1) Must be called with an elevated refcount on the page. This is a * fundamentnal difference from isolate_lru_pages (which is called * without a stable reference). * (2) the lru_lock must not be held. * (3) interrupts must be enabled. */ int isolate_lru_page(struct page *page) { int ret = -EBUSY; |
0c917313a
|
1086 |
VM_BUG_ON(!page_count(page)); |
62695a84e
|
1087 1088 1089 1090 |
if (PageLRU(page)) { struct zone *zone = page_zone(page); spin_lock_irq(&zone->lru_lock); |
0c917313a
|
1091 |
if (PageLRU(page)) { |
894bc3104
|
1092 |
int lru = page_lru(page); |
62695a84e
|
1093 |
ret = 0; |
0c917313a
|
1094 |
get_page(page); |
62695a84e
|
1095 |
ClearPageLRU(page); |
4f98a2fee
|
1096 |
|
4f98a2fee
|
1097 |
del_page_from_lru_list(zone, page, lru); |
62695a84e
|
1098 1099 1100 1101 1102 |
} spin_unlock_irq(&zone->lru_lock); } return ret; } |
5ad333eb6
|
1103 |
/* |
35cd78156
|
1104 1105 1106 1107 1108 1109 1110 1111 1112 |
* Are there way too many processes in the direct reclaim path already? */ static int too_many_isolated(struct zone *zone, int file, struct scan_control *sc) { unsigned long inactive, isolated; if (current_is_kswapd()) return 0; |
89b5fae53
|
1113 |
if (!global_reclaim(sc)) |
35cd78156
|
1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 |
return 0; if (file) { inactive = zone_page_state(zone, NR_INACTIVE_FILE); isolated = zone_page_state(zone, NR_ISOLATED_FILE); } else { inactive = zone_page_state(zone, NR_INACTIVE_ANON); isolated = zone_page_state(zone, NR_ISOLATED_ANON); } return isolated > inactive; } |
666356297
|
1126 |
static noinline_for_stack void |
3f79768f2
|
1127 1128 |
putback_inactive_pages(struct mem_cgroup_zone *mz, struct list_head *page_list) |
666356297
|
1129 |
{ |
f16015fbf
|
1130 |
struct zone_reclaim_stat *reclaim_stat = get_reclaim_stat(mz); |
3f79768f2
|
1131 1132 |
struct zone *zone = mz->zone; LIST_HEAD(pages_to_free); |
666356297
|
1133 |
|
666356297
|
1134 1135 1136 |
/* * Put back any unfreeable pages. */ |
666356297
|
1137 |
while (!list_empty(page_list)) { |
3f79768f2
|
1138 |
struct page *page = lru_to_page(page_list); |
666356297
|
1139 |
int lru; |
3f79768f2
|
1140 |
|
666356297
|
1141 1142 1143 1144 1145 1146 1147 1148 |
VM_BUG_ON(PageLRU(page)); list_del(&page->lru); if (unlikely(!page_evictable(page, NULL))) { spin_unlock_irq(&zone->lru_lock); putback_lru_page(page); spin_lock_irq(&zone->lru_lock); continue; } |
7a608572a
|
1149 |
SetPageLRU(page); |
666356297
|
1150 |
lru = page_lru(page); |
7a608572a
|
1151 |
add_page_to_lru_list(zone, page, lru); |
666356297
|
1152 1153 |
if (is_active_lru(lru)) { int file = is_file_lru(lru); |
9992af102
|
1154 1155 |
int numpages = hpage_nr_pages(page); reclaim_stat->recent_rotated[file] += numpages; |
666356297
|
1156 |
} |
2bcf88796
|
1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 |
if (put_page_testzero(page)) { __ClearPageLRU(page); __ClearPageActive(page); del_page_from_lru_list(zone, page, lru); if (unlikely(PageCompound(page))) { spin_unlock_irq(&zone->lru_lock); (*get_compound_page_dtor(page))(page); spin_lock_irq(&zone->lru_lock); } else list_add(&page->lru, &pages_to_free); |
666356297
|
1168 1169 |
} } |
666356297
|
1170 |
|
3f79768f2
|
1171 1172 1173 1174 |
/* * To save our caller's stack, now use input list for pages to free. */ list_splice(&pages_to_free, page_list); |
666356297
|
1175 |
} |
f16015fbf
|
1176 1177 |
static noinline_for_stack void update_isolated_counts(struct mem_cgroup_zone *mz, |
3f79768f2
|
1178 |
struct list_head *page_list, |
f16015fbf
|
1179 |
unsigned long *nr_anon, |
3f79768f2
|
1180 |
unsigned long *nr_file) |
1489fa14c
|
1181 |
{ |
f16015fbf
|
1182 |
struct zone *zone = mz->zone; |
1489fa14c
|
1183 |
unsigned int count[NR_LRU_LISTS] = { 0, }; |
3f79768f2
|
1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 |
unsigned long nr_active = 0; struct page *page; int lru; /* * Count pages and clear active flags */ list_for_each_entry(page, page_list, lru) { int numpages = hpage_nr_pages(page); lru = page_lru_base_type(page); if (PageActive(page)) { lru += LRU_ACTIVE; ClearPageActive(page); nr_active += numpages; } count[lru] += numpages; } |
1489fa14c
|
1201 |
|
d563c0501
|
1202 |
preempt_disable(); |
1489fa14c
|
1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 |
__count_vm_events(PGDEACTIVATE, nr_active); __mod_zone_page_state(zone, NR_ACTIVE_FILE, -count[LRU_ACTIVE_FILE]); __mod_zone_page_state(zone, NR_INACTIVE_FILE, -count[LRU_INACTIVE_FILE]); __mod_zone_page_state(zone, NR_ACTIVE_ANON, -count[LRU_ACTIVE_ANON]); __mod_zone_page_state(zone, NR_INACTIVE_ANON, -count[LRU_INACTIVE_ANON]); *nr_anon = count[LRU_ACTIVE_ANON] + count[LRU_INACTIVE_ANON]; *nr_file = count[LRU_ACTIVE_FILE] + count[LRU_INACTIVE_FILE]; |
1489fa14c
|
1216 |
|
d563c0501
|
1217 1218 1219 |
__mod_zone_page_state(zone, NR_ISOLATED_ANON, *nr_anon); __mod_zone_page_state(zone, NR_ISOLATED_FILE, *nr_file); preempt_enable(); |
1489fa14c
|
1220 |
} |
666356297
|
1221 |
/* |
1742f19fa
|
1222 1223 |
* shrink_inactive_list() is a helper for shrink_zone(). It returns the number * of reclaimed pages |
1da177e4c
|
1224 |
*/ |
666356297
|
1225 |
static noinline_for_stack unsigned long |
f16015fbf
|
1226 1227 |
shrink_inactive_list(unsigned long nr_to_scan, struct mem_cgroup_zone *mz, struct scan_control *sc, int priority, int file) |
1da177e4c
|
1228 1229 |
{ LIST_HEAD(page_list); |
e247dbce5
|
1230 |
unsigned long nr_scanned; |
05ff51376
|
1231 |
unsigned long nr_reclaimed = 0; |
e247dbce5
|
1232 |
unsigned long nr_taken; |
e247dbce5
|
1233 1234 |
unsigned long nr_anon; unsigned long nr_file; |
92df3a723
|
1235 1236 |
unsigned long nr_dirty = 0; unsigned long nr_writeback = 0; |
613172891
|
1237 |
isolate_mode_t isolate_mode = ISOLATE_INACTIVE; |
f16015fbf
|
1238 |
struct zone *zone = mz->zone; |
d563c0501
|
1239 |
struct zone_reclaim_stat *reclaim_stat = get_reclaim_stat(mz); |
78dc583d3
|
1240 |
|
35cd78156
|
1241 |
while (unlikely(too_many_isolated(zone, file, sc))) { |
58355c787
|
1242 |
congestion_wait(BLK_RW_ASYNC, HZ/10); |
35cd78156
|
1243 1244 1245 1246 1247 |
/* We are about to die and free our memory. Return now. */ if (fatal_signal_pending(current)) return SWAP_CLUSTER_MAX; } |
1da177e4c
|
1248 |
lru_add_drain(); |
f80c06736
|
1249 1250 |
if (!sc->may_unmap) |
613172891
|
1251 |
isolate_mode |= ISOLATE_UNMAPPED; |
f80c06736
|
1252 |
if (!sc->may_writepage) |
613172891
|
1253 |
isolate_mode |= ISOLATE_CLEAN; |
f80c06736
|
1254 |
|
1da177e4c
|
1255 |
spin_lock_irq(&zone->lru_lock); |
b35ea17b7
|
1256 |
|
fe2c2a106
|
1257 1258 |
nr_taken = isolate_lru_pages(nr_to_scan, mz, &page_list, &nr_scanned, sc, isolate_mode, 0, file); |
89b5fae53
|
1259 |
if (global_reclaim(sc)) { |
e247dbce5
|
1260 1261 1262 1263 1264 1265 1266 |
zone->pages_scanned += nr_scanned; if (current_is_kswapd()) __count_zone_vm_events(PGSCAN_KSWAPD, zone, nr_scanned); else __count_zone_vm_events(PGSCAN_DIRECT, zone, nr_scanned); |
e247dbce5
|
1267 |
} |
d563c0501
|
1268 |
spin_unlock_irq(&zone->lru_lock); |
b35ea17b7
|
1269 |
|
d563c0501
|
1270 |
if (nr_taken == 0) |
666356297
|
1271 |
return 0; |
5ad333eb6
|
1272 |
|
3f79768f2
|
1273 |
update_isolated_counts(mz, &page_list, &nr_anon, &nr_file); |
f16015fbf
|
1274 |
nr_reclaimed = shrink_page_list(&page_list, mz, sc, priority, |
92df3a723
|
1275 |
&nr_dirty, &nr_writeback); |
c661b078f
|
1276 |
|
3f79768f2
|
1277 |
spin_lock_irq(&zone->lru_lock); |
d563c0501
|
1278 1279 |
reclaim_stat->recent_scanned[0] += nr_anon; reclaim_stat->recent_scanned[1] += nr_file; |
904249aa6
|
1280 1281 1282 1283 1284 1285 1286 1287 |
if (global_reclaim(sc)) { if (current_is_kswapd()) __count_zone_vm_events(PGSTEAL_KSWAPD, zone, nr_reclaimed); else __count_zone_vm_events(PGSTEAL_DIRECT, zone, nr_reclaimed); } |
a74609faf
|
1288 |
|
3f79768f2
|
1289 1290 1291 1292 1293 1294 1295 1296 |
putback_inactive_pages(mz, &page_list); __mod_zone_page_state(zone, NR_ISOLATED_ANON, -nr_anon); __mod_zone_page_state(zone, NR_ISOLATED_FILE, -nr_file); spin_unlock_irq(&zone->lru_lock); free_hot_cold_page_list(&page_list, 1); |
e11da5b4f
|
1297 |
|
92df3a723
|
1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 |
/* * 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. * * This scales the number of dirty pages that must be under writeback * before throttling depending on priority. It is a simple backoff * function that has the most effect in the range DEF_PRIORITY to * DEF_PRIORITY-2 which is the priority reclaim is considered to be * in trouble and reclaim is considered to be in trouble. * * DEF_PRIORITY 100% isolated pages must be PageWriteback to throttle * DEF_PRIORITY-1 50% must be PageWriteback * DEF_PRIORITY-2 25% must be PageWriteback, kswapd in trouble * ... * DEF_PRIORITY-6 For SWAP_CLUSTER_MAX isolated pages, throttle if any * isolated page is PageWriteback */ if (nr_writeback && nr_writeback >= (nr_taken >> (DEF_PRIORITY-priority))) wait_iff_congested(zone, BLK_RW_ASYNC, HZ/10); |
e11da5b4f
|
1323 1324 1325 1326 |
trace_mm_vmscan_lru_shrink_inactive(zone->zone_pgdat->node_id, zone_idx(zone), nr_scanned, nr_reclaimed, priority, |
23b9da55c
|
1327 |
trace_shrink_flags(file)); |
05ff51376
|
1328 |
return nr_reclaimed; |
1da177e4c
|
1329 |
} |
3bb1a852a
|
1330 |
/* |
1da177e4c
|
1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 |
* This moves pages from the active list to the inactive 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->_count against each page. * But we had to alter page->flags anyway. */ |
1cfb419b3
|
1347 |
|
3eb4140f0
|
1348 1349 |
static void move_active_pages_to_lru(struct zone *zone, struct list_head *list, |
2bcf88796
|
1350 |
struct list_head *pages_to_free, |
3eb4140f0
|
1351 1352 1353 |
enum lru_list lru) { unsigned long pgmoved = 0; |
3eb4140f0
|
1354 |
struct page *page; |
3eb4140f0
|
1355 |
while (!list_empty(list)) { |
925b7673c
|
1356 |
struct lruvec *lruvec; |
3eb4140f0
|
1357 |
page = lru_to_page(list); |
3eb4140f0
|
1358 1359 1360 |
VM_BUG_ON(PageLRU(page)); SetPageLRU(page); |
925b7673c
|
1361 1362 |
lruvec = mem_cgroup_lru_add_list(zone, page, lru); list_move(&page->lru, &lruvec->lists[lru]); |
2c888cfbc
|
1363 |
pgmoved += hpage_nr_pages(page); |
3eb4140f0
|
1364 |
|
2bcf88796
|
1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 |
if (put_page_testzero(page)) { __ClearPageLRU(page); __ClearPageActive(page); del_page_from_lru_list(zone, page, lru); if (unlikely(PageCompound(page))) { spin_unlock_irq(&zone->lru_lock); (*get_compound_page_dtor(page))(page); spin_lock_irq(&zone->lru_lock); } else list_add(&page->lru, pages_to_free); |
3eb4140f0
|
1376 1377 1378 1379 1380 1381 |
} } __mod_zone_page_state(zone, NR_LRU_BASE + lru, pgmoved); if (!is_active_lru(lru)) __count_vm_events(PGDEACTIVATE, pgmoved); } |
1cfb419b3
|
1382 |
|
f626012db
|
1383 |
static void shrink_active_list(unsigned long nr_to_scan, |
f16015fbf
|
1384 1385 1386 |
struct mem_cgroup_zone *mz, struct scan_control *sc, int priority, int file) |
1da177e4c
|
1387 |
{ |
44c241f16
|
1388 |
unsigned long nr_taken; |
f626012db
|
1389 |
unsigned long nr_scanned; |
6fe6b7e35
|
1390 |
unsigned long vm_flags; |
1da177e4c
|
1391 |
LIST_HEAD(l_hold); /* The pages which were snipped off */ |
8cab4754d
|
1392 |
LIST_HEAD(l_active); |
b69408e88
|
1393 |
LIST_HEAD(l_inactive); |
1da177e4c
|
1394 |
struct page *page; |
f16015fbf
|
1395 |
struct zone_reclaim_stat *reclaim_stat = get_reclaim_stat(mz); |
44c241f16
|
1396 |
unsigned long nr_rotated = 0; |
613172891
|
1397 |
isolate_mode_t isolate_mode = ISOLATE_ACTIVE; |
f16015fbf
|
1398 |
struct zone *zone = mz->zone; |
1da177e4c
|
1399 1400 |
lru_add_drain(); |
f80c06736
|
1401 1402 |
if (!sc->may_unmap) |
613172891
|
1403 |
isolate_mode |= ISOLATE_UNMAPPED; |
f80c06736
|
1404 |
if (!sc->may_writepage) |
613172891
|
1405 |
isolate_mode |= ISOLATE_CLEAN; |
f80c06736
|
1406 |
|
1da177e4c
|
1407 |
spin_lock_irq(&zone->lru_lock); |
925b7673c
|
1408 |
|
fe2c2a106
|
1409 |
nr_taken = isolate_lru_pages(nr_to_scan, mz, &l_hold, &nr_scanned, sc, |
613172891
|
1410 |
isolate_mode, 1, file); |
89b5fae53
|
1411 |
if (global_reclaim(sc)) |
f626012db
|
1412 |
zone->pages_scanned += nr_scanned; |
89b5fae53
|
1413 |
|
b7c46d151
|
1414 |
reclaim_stat->recent_scanned[file] += nr_taken; |
1cfb419b3
|
1415 |
|
f626012db
|
1416 |
__count_zone_vm_events(PGREFILL, zone, nr_scanned); |
4f98a2fee
|
1417 |
if (file) |
44c241f16
|
1418 |
__mod_zone_page_state(zone, NR_ACTIVE_FILE, -nr_taken); |
4f98a2fee
|
1419 |
else |
44c241f16
|
1420 |
__mod_zone_page_state(zone, NR_ACTIVE_ANON, -nr_taken); |
a731286de
|
1421 |
__mod_zone_page_state(zone, NR_ISOLATED_ANON + file, nr_taken); |
1da177e4c
|
1422 |
spin_unlock_irq(&zone->lru_lock); |
1da177e4c
|
1423 1424 1425 1426 |
while (!list_empty(&l_hold)) { cond_resched(); page = lru_to_page(&l_hold); list_del(&page->lru); |
7e9cd4842
|
1427 |
|
894bc3104
|
1428 1429 1430 1431 |
if (unlikely(!page_evictable(page, NULL))) { putback_lru_page(page); continue; } |
cc715d99e
|
1432 1433 1434 1435 1436 1437 1438 |
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
|
1439 1440 |
if (page_referenced(page, 0, sc->target_mem_cgroup, &vm_flags)) { |
9992af102
|
1441 |
nr_rotated += hpage_nr_pages(page); |
8cab4754d
|
1442 1443 1444 1445 1446 1447 1448 1449 1450 |
/* * Identify referenced, file-backed active pages and * give them one more trip around the active list. So * that executable code get better chances to stay in * memory under moderate memory pressure. Anon pages * are not likely to be evicted by use-once streaming * IO, plus JVM can create lots of anon VM_EXEC pages, * so we ignore them here. */ |
41e20983f
|
1451 |
if ((vm_flags & VM_EXEC) && page_is_file_cache(page)) { |
8cab4754d
|
1452 1453 1454 1455 |
list_add(&page->lru, &l_active); continue; } } |
7e9cd4842
|
1456 |
|
5205e56ee
|
1457 |
ClearPageActive(page); /* we are de-activating */ |
1da177e4c
|
1458 1459 |
list_add(&page->lru, &l_inactive); } |
b555749aa
|
1460 |
/* |
8cab4754d
|
1461 |
* Move pages back to the lru list. |
b555749aa
|
1462 |
*/ |
2a1dc5097
|
1463 |
spin_lock_irq(&zone->lru_lock); |
4f98a2fee
|
1464 |
/* |
8cab4754d
|
1465 1466 1467 1468 |
* Count referenced pages from currently used mappings as rotated, * even though only some of them are actually re-activated. This * helps balance scan pressure between file and anonymous pages in * get_scan_ratio. |
7e9cd4842
|
1469 |
*/ |
b7c46d151
|
1470 |
reclaim_stat->recent_rotated[file] += nr_rotated; |
556adecba
|
1471 |
|
2bcf88796
|
1472 |
move_active_pages_to_lru(zone, &l_active, &l_hold, |
3eb4140f0
|
1473 |
LRU_ACTIVE + file * LRU_FILE); |
2bcf88796
|
1474 |
move_active_pages_to_lru(zone, &l_inactive, &l_hold, |
3eb4140f0
|
1475 |
LRU_BASE + file * LRU_FILE); |
a731286de
|
1476 |
__mod_zone_page_state(zone, NR_ISOLATED_ANON + file, -nr_taken); |
f8891e5e1
|
1477 |
spin_unlock_irq(&zone->lru_lock); |
2bcf88796
|
1478 1479 |
free_hot_cold_page_list(&l_hold, 1); |
1da177e4c
|
1480 |
} |
74e3f3c33
|
1481 |
#ifdef CONFIG_SWAP |
14797e236
|
1482 |
static int inactive_anon_is_low_global(struct zone *zone) |
f89eb90e3
|
1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 |
{ unsigned long active, inactive; active = zone_page_state(zone, NR_ACTIVE_ANON); inactive = zone_page_state(zone, NR_INACTIVE_ANON); if (inactive * zone->inactive_ratio < active) return 1; return 0; } |
14797e236
|
1494 1495 1496 1497 1498 1499 1500 1501 |
/** * inactive_anon_is_low - check if anonymous pages need to be deactivated * @zone: zone to check * @sc: scan control of this context * * Returns true if the zone does not have enough inactive anon pages, * meaning some active anon pages need to be deactivated. */ |
f16015fbf
|
1502 |
static int inactive_anon_is_low(struct mem_cgroup_zone *mz) |
14797e236
|
1503 |
{ |
74e3f3c33
|
1504 1505 1506 1507 1508 1509 |
/* * If we don't have swap space, anonymous page deactivation * is pointless. */ if (!total_swap_pages) return 0; |
f16015fbf
|
1510 1511 1512 1513 1514 |
if (!scanning_global_lru(mz)) return mem_cgroup_inactive_anon_is_low(mz->mem_cgroup, mz->zone); return inactive_anon_is_low_global(mz->zone); |
14797e236
|
1515 |
} |
74e3f3c33
|
1516 |
#else |
f16015fbf
|
1517 |
static inline int inactive_anon_is_low(struct mem_cgroup_zone *mz) |
74e3f3c33
|
1518 1519 1520 1521 |
{ return 0; } #endif |
14797e236
|
1522 |
|
56e49d218
|
1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 |
static int inactive_file_is_low_global(struct zone *zone) { unsigned long active, inactive; active = zone_page_state(zone, NR_ACTIVE_FILE); inactive = zone_page_state(zone, NR_INACTIVE_FILE); return (active > inactive); } /** * inactive_file_is_low - check if file pages need to be deactivated |
f16015fbf
|
1535 |
* @mz: memory cgroup and zone to check |
56e49d218
|
1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 |
* * When the system is doing streaming IO, memory pressure here * ensures that active file pages get deactivated, until more * than half of the file pages are on the inactive list. * * Once we get to that situation, protect the system's working * set from being evicted by disabling active file page aging. * * This uses a different ratio than the anonymous pages, because * the page cache uses a use-once replacement algorithm. */ |
f16015fbf
|
1547 |
static int inactive_file_is_low(struct mem_cgroup_zone *mz) |
56e49d218
|
1548 |
{ |
f16015fbf
|
1549 1550 1551 |
if (!scanning_global_lru(mz)) return mem_cgroup_inactive_file_is_low(mz->mem_cgroup, mz->zone); |
56e49d218
|
1552 |
|
f16015fbf
|
1553 |
return inactive_file_is_low_global(mz->zone); |
56e49d218
|
1554 |
} |
f16015fbf
|
1555 |
static int inactive_list_is_low(struct mem_cgroup_zone *mz, int file) |
b39415b27
|
1556 1557 |
{ if (file) |
f16015fbf
|
1558 |
return inactive_file_is_low(mz); |
b39415b27
|
1559 |
else |
f16015fbf
|
1560 |
return inactive_anon_is_low(mz); |
b39415b27
|
1561 |
} |
4f98a2fee
|
1562 |
static unsigned long shrink_list(enum lru_list lru, unsigned long nr_to_scan, |
f16015fbf
|
1563 1564 |
struct mem_cgroup_zone *mz, struct scan_control *sc, int priority) |
b69408e88
|
1565 |
{ |
4f98a2fee
|
1566 |
int file = is_file_lru(lru); |
b39415b27
|
1567 |
if (is_active_lru(lru)) { |
f16015fbf
|
1568 1569 |
if (inactive_list_is_low(mz, file)) shrink_active_list(nr_to_scan, mz, sc, priority, file); |
556adecba
|
1570 1571 |
return 0; } |
f16015fbf
|
1572 |
return shrink_inactive_list(nr_to_scan, mz, sc, priority, file); |
4f98a2fee
|
1573 |
} |
f16015fbf
|
1574 1575 |
static int vmscan_swappiness(struct mem_cgroup_zone *mz, struct scan_control *sc) |
1f4c025b5
|
1576 |
{ |
89b5fae53
|
1577 |
if (global_reclaim(sc)) |
1f4c025b5
|
1578 |
return vm_swappiness; |
f16015fbf
|
1579 |
return mem_cgroup_swappiness(mz->mem_cgroup); |
1f4c025b5
|
1580 |
} |
4f98a2fee
|
1581 1582 1583 1584 1585 1586 |
/* * 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. * |
76a33fc38
|
1587 |
* nr[0] = anon pages to scan; nr[1] = file pages to scan |
4f98a2fee
|
1588 |
*/ |
f16015fbf
|
1589 1590 |
static void get_scan_count(struct mem_cgroup_zone *mz, struct scan_control *sc, unsigned long *nr, int priority) |
4f98a2fee
|
1591 1592 1593 1594 |
{ unsigned long anon, file, free; unsigned long anon_prio, file_prio; unsigned long ap, fp; |
f16015fbf
|
1595 |
struct zone_reclaim_stat *reclaim_stat = get_reclaim_stat(mz); |
76a33fc38
|
1596 |
u64 fraction[2], denominator; |
4111304da
|
1597 |
enum lru_list lru; |
76a33fc38
|
1598 |
int noswap = 0; |
a4d3e9e76
|
1599 |
bool force_scan = false; |
246e87a93
|
1600 |
|
f11c0ca50
|
1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 |
/* * If the zone or memcg is small, nr[l] can be 0. This * results in no scanning on this priority and a potential * priority drop. Global direct reclaim can go to the next * zone and tends to have no problems. Global kswapd is for * zone balancing and it needs to scan a minimum amount. When * reclaiming for a memcg, a priority drop can cause high * latencies, so it's better to scan a minimum amount there as * well. */ |
b95a2f2d4
|
1611 |
if (current_is_kswapd() && mz->zone->all_unreclaimable) |
a4d3e9e76
|
1612 |
force_scan = true; |
89b5fae53
|
1613 |
if (!global_reclaim(sc)) |
a4d3e9e76
|
1614 |
force_scan = true; |
76a33fc38
|
1615 1616 1617 1618 1619 1620 1621 1622 1623 |
/* If we have no swap space, do not bother scanning anon pages. */ if (!sc->may_swap || (nr_swap_pages <= 0)) { noswap = 1; fraction[0] = 0; fraction[1] = 1; denominator = 1; goto out; } |
4f98a2fee
|
1624 |
|
f16015fbf
|
1625 1626 1627 1628 |
anon = zone_nr_lru_pages(mz, LRU_ACTIVE_ANON) + zone_nr_lru_pages(mz, LRU_INACTIVE_ANON); file = zone_nr_lru_pages(mz, LRU_ACTIVE_FILE) + zone_nr_lru_pages(mz, LRU_INACTIVE_FILE); |
a4d3e9e76
|
1629 |
|
89b5fae53
|
1630 |
if (global_reclaim(sc)) { |
f16015fbf
|
1631 |
free = zone_page_state(mz->zone, NR_FREE_PAGES); |
eeee9a8cd
|
1632 1633 |
/* If we have very few page cache pages, force-scan anon pages. */ |
f16015fbf
|
1634 |
if (unlikely(file + free <= high_wmark_pages(mz->zone))) { |
76a33fc38
|
1635 1636 1637 1638 |
fraction[0] = 1; fraction[1] = 0; denominator = 1; goto out; |
eeee9a8cd
|
1639 |
} |
4f98a2fee
|
1640 1641 1642 |
} /* |
58c37f6e0
|
1643 1644 1645 |
* With swappiness at 100, anonymous and file have the same priority. * This scanning priority is essentially the inverse of IO cost. */ |
f16015fbf
|
1646 1647 |
anon_prio = vmscan_swappiness(mz, sc); file_prio = 200 - vmscan_swappiness(mz, sc); |
58c37f6e0
|
1648 1649 |
/* |
4f98a2fee
|
1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 |
* OK, so we have swap space and a fair amount of page cache * pages. We use the recently rotated / recently scanned * ratios to determine how valuable each cache is. * * Because workloads change over time (and to avoid overflow) * we keep these statistics as a floating average, which ends * up weighing recent references more than old ones. * * anon in [0], file in [1] */ |
f16015fbf
|
1660 |
spin_lock_irq(&mz->zone->lru_lock); |
6e9015716
|
1661 |
if (unlikely(reclaim_stat->recent_scanned[0] > anon / 4)) { |
6e9015716
|
1662 1663 |
reclaim_stat->recent_scanned[0] /= 2; reclaim_stat->recent_rotated[0] /= 2; |
4f98a2fee
|
1664 |
} |
6e9015716
|
1665 |
if (unlikely(reclaim_stat->recent_scanned[1] > file / 4)) { |
6e9015716
|
1666 1667 |
reclaim_stat->recent_scanned[1] /= 2; reclaim_stat->recent_rotated[1] /= 2; |
4f98a2fee
|
1668 1669 1670 |
} /* |
00d8089c5
|
1671 1672 1673 |
* The amount of pressure on anon vs file pages is inversely * proportional to the fraction of recently scanned pages on * each list that were recently referenced and in active use. |
4f98a2fee
|
1674 |
*/ |
6e9015716
|
1675 1676 |
ap = (anon_prio + 1) * (reclaim_stat->recent_scanned[0] + 1); ap /= reclaim_stat->recent_rotated[0] + 1; |
4f98a2fee
|
1677 |
|
6e9015716
|
1678 1679 |
fp = (file_prio + 1) * (reclaim_stat->recent_scanned[1] + 1); fp /= reclaim_stat->recent_rotated[1] + 1; |
f16015fbf
|
1680 |
spin_unlock_irq(&mz->zone->lru_lock); |
4f98a2fee
|
1681 |
|
76a33fc38
|
1682 1683 1684 1685 |
fraction[0] = ap; fraction[1] = fp; denominator = ap + fp + 1; out: |
4111304da
|
1686 1687 |
for_each_evictable_lru(lru) { int file = is_file_lru(lru); |
76a33fc38
|
1688 |
unsigned long scan; |
6e08a369e
|
1689 |
|
4111304da
|
1690 |
scan = zone_nr_lru_pages(mz, lru); |
76a33fc38
|
1691 1692 |
if (priority || noswap) { scan >>= priority; |
f11c0ca50
|
1693 1694 |
if (!scan && force_scan) scan = SWAP_CLUSTER_MAX; |
76a33fc38
|
1695 1696 |
scan = div64_u64(scan * fraction[file], denominator); } |
4111304da
|
1697 |
nr[lru] = scan; |
76a33fc38
|
1698 |
} |
6e08a369e
|
1699 |
} |
4f98a2fee
|
1700 |
|
23b9da55c
|
1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 |
/* Use reclaim/compaction for costly allocs or under memory pressure */ static bool in_reclaim_compaction(int priority, struct scan_control *sc) { if (COMPACTION_BUILD && sc->order && (sc->order > PAGE_ALLOC_COSTLY_ORDER || priority < DEF_PRIORITY - 2)) return true; return false; } |
1da177e4c
|
1711 |
/* |
23b9da55c
|
1712 1713 1714 1715 1716 |
* Reclaim/compaction is used for high-order allocation requests. It reclaims * order-0 pages before compacting the zone. should_continue_reclaim() returns * true if more pages should be reclaimed such that when the page allocator * calls try_to_compact_zone() that it will have enough free pages to succeed. * It will give up earlier than that if there is difficulty reclaiming pages. |
3e7d34497
|
1717 |
*/ |
f16015fbf
|
1718 |
static inline bool should_continue_reclaim(struct mem_cgroup_zone *mz, |
3e7d34497
|
1719 1720 |
unsigned long nr_reclaimed, unsigned long nr_scanned, |
23b9da55c
|
1721 |
int priority, |
3e7d34497
|
1722 1723 1724 1725 1726 1727 |
struct scan_control *sc) { unsigned long pages_for_compaction; unsigned long inactive_lru_pages; /* If not in reclaim/compaction mode, stop */ |
23b9da55c
|
1728 |
if (!in_reclaim_compaction(priority, sc)) |
3e7d34497
|
1729 |
return false; |
2876592f2
|
1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 |
/* Consider stopping depending on scan and reclaim activity */ if (sc->gfp_mask & __GFP_REPEAT) { /* * For __GFP_REPEAT allocations, stop reclaiming if the * full LRU list has been scanned and we are still failing * to reclaim pages. This full LRU scan is potentially * expensive but a __GFP_REPEAT caller really wants to succeed */ if (!nr_reclaimed && !nr_scanned) return false; } else { /* * For non-__GFP_REPEAT allocations which can presumably * fail without consequence, 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 faster at the risk reclaim/compaction and * the resulting allocation attempt fails */ if (!nr_reclaimed) return false; } |
3e7d34497
|
1752 1753 1754 1755 1756 1757 |
/* * If we have not reclaimed enough pages for compaction and the * inactive lists are large enough, continue reclaiming */ pages_for_compaction = (2UL << sc->order); |
f16015fbf
|
1758 |
inactive_lru_pages = zone_nr_lru_pages(mz, LRU_INACTIVE_FILE); |
86cfd3a45
|
1759 |
if (nr_swap_pages > 0) |
f16015fbf
|
1760 |
inactive_lru_pages += zone_nr_lru_pages(mz, LRU_INACTIVE_ANON); |
3e7d34497
|
1761 1762 1763 1764 1765 |
if (sc->nr_reclaimed < pages_for_compaction && inactive_lru_pages > pages_for_compaction) return true; /* If compaction would go ahead or the allocation would succeed, stop */ |
f16015fbf
|
1766 |
switch (compaction_suitable(mz->zone, sc->order)) { |
3e7d34497
|
1767 1768 1769 1770 1771 1772 1773 1774 1775 |
case COMPACT_PARTIAL: case COMPACT_CONTINUE: return false; default: return true; } } /* |
1da177e4c
|
1776 1777 |
* This is a basic per-zone page freer. Used by both kswapd and direct reclaim. */ |
f16015fbf
|
1778 1779 |
static void shrink_mem_cgroup_zone(int priority, struct mem_cgroup_zone *mz, struct scan_control *sc) |
1da177e4c
|
1780 |
{ |
b69408e88
|
1781 |
unsigned long nr[NR_LRU_LISTS]; |
8695949a1
|
1782 |
unsigned long nr_to_scan; |
4111304da
|
1783 |
enum lru_list lru; |
f0fdc5e8e
|
1784 |
unsigned long nr_reclaimed, nr_scanned; |
22fba3354
|
1785 |
unsigned long nr_to_reclaim = sc->nr_to_reclaim; |
3da367c3e
|
1786 |
struct blk_plug plug; |
e0f79b8f1
|
1787 |
|
3e7d34497
|
1788 1789 |
restart: nr_reclaimed = 0; |
f0fdc5e8e
|
1790 |
nr_scanned = sc->nr_scanned; |
f16015fbf
|
1791 |
get_scan_count(mz, sc, nr, priority); |
1da177e4c
|
1792 |
|
3da367c3e
|
1793 |
blk_start_plug(&plug); |
556adecba
|
1794 1795 |
while (nr[LRU_INACTIVE_ANON] || nr[LRU_ACTIVE_FILE] || nr[LRU_INACTIVE_FILE]) { |
4111304da
|
1796 1797 |
for_each_evictable_lru(lru) { if (nr[lru]) { |
ece74b2e7
|
1798 |
nr_to_scan = min_t(unsigned long, |
4111304da
|
1799 1800 |
nr[lru], SWAP_CLUSTER_MAX); nr[lru] -= nr_to_scan; |
1da177e4c
|
1801 |
|
4111304da
|
1802 |
nr_reclaimed += shrink_list(lru, nr_to_scan, |
f16015fbf
|
1803 |
mz, sc, priority); |
b69408e88
|
1804 |
} |
1da177e4c
|
1805 |
} |
a79311c14
|
1806 1807 1808 1809 1810 1811 1812 1813 |
/* * On large memory systems, scan >> priority can become * really large. This is fine for the starting priority; * we want to put equal scanning pressure on each zone. * However, if the VM has a harder time of freeing pages, * with multiple processes reclaiming pages, the total * freeing target can get unreasonably large. */ |
41c930881
|
1814 |
if (nr_reclaimed >= nr_to_reclaim && priority < DEF_PRIORITY) |
a79311c14
|
1815 |
break; |
1da177e4c
|
1816 |
} |
3da367c3e
|
1817 |
blk_finish_plug(&plug); |
3e7d34497
|
1818 |
sc->nr_reclaimed += nr_reclaimed; |
01dbe5c9b
|
1819 |
|
556adecba
|
1820 1821 1822 1823 |
/* * Even if we did not try to evict anon pages at all, we want to * rebalance the anon lru active/inactive ratio. */ |
f16015fbf
|
1824 1825 |
if (inactive_anon_is_low(mz)) shrink_active_list(SWAP_CLUSTER_MAX, mz, sc, priority, 0); |
556adecba
|
1826 |
|
3e7d34497
|
1827 |
/* reclaim/compaction might need reclaim to continue */ |
f16015fbf
|
1828 |
if (should_continue_reclaim(mz, nr_reclaimed, |
23b9da55c
|
1829 1830 |
sc->nr_scanned - nr_scanned, priority, sc)) |
3e7d34497
|
1831 |
goto restart; |
232ea4d69
|
1832 |
throttle_vm_writeout(sc->gfp_mask); |
1da177e4c
|
1833 |
} |
f16015fbf
|
1834 1835 1836 |
static void shrink_zone(int priority, struct zone *zone, struct scan_control *sc) { |
5660048cc
|
1837 1838 |
struct mem_cgroup *root = sc->target_mem_cgroup; struct mem_cgroup_reclaim_cookie reclaim = { |
f16015fbf
|
1839 |
.zone = zone, |
5660048cc
|
1840 |
.priority = priority, |
f16015fbf
|
1841 |
}; |
5660048cc
|
1842 |
struct mem_cgroup *memcg; |
5660048cc
|
1843 1844 1845 1846 1847 1848 |
memcg = mem_cgroup_iter(root, NULL, &reclaim); do { struct mem_cgroup_zone mz = { .mem_cgroup = memcg, .zone = zone, }; |
f16015fbf
|
1849 |
|
5660048cc
|
1850 1851 1852 1853 1854 1855 |
shrink_mem_cgroup_zone(priority, &mz, sc); /* * Limit reclaim has historically picked one memcg and * scanned it with decreasing priority levels until * nr_to_reclaim had been reclaimed. This priority * cycle is thus over after a single memcg. |
b95a2f2d4
|
1856 1857 1858 1859 |
* * Direct reclaim and kswapd, on the other hand, have * to scan all memory cgroups to fulfill the overall * scan target for the zone. |
5660048cc
|
1860 1861 1862 1863 1864 1865 1866 |
*/ if (!global_reclaim(sc)) { mem_cgroup_iter_break(root, memcg); break; } memcg = mem_cgroup_iter(root, memcg, &reclaim); } while (memcg); |
f16015fbf
|
1867 |
} |
fe4b1b244
|
1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 |
/* Returns true if compaction should go ahead for a high-order request */ static inline bool compaction_ready(struct zone *zone, struct scan_control *sc) { unsigned long balance_gap, watermark; bool watermark_ok; /* Do not consider compaction for orders reclaim is meant to satisfy */ if (sc->order <= PAGE_ALLOC_COSTLY_ORDER) return false; /* * Compaction takes time to run and there are potentially other * callers using the pages just freed. Continue reclaiming until * there is a buffer of free pages available to give compaction * a reasonable chance of completing and allocating the page */ balance_gap = min(low_wmark_pages(zone), (zone->present_pages + KSWAPD_ZONE_BALANCE_GAP_RATIO-1) / KSWAPD_ZONE_BALANCE_GAP_RATIO); watermark = high_wmark_pages(zone) + balance_gap + (2UL << sc->order); watermark_ok = zone_watermark_ok_safe(zone, 0, watermark, 0, 0); /* * If compaction is deferred, reclaim up to a point where * compaction will have a chance of success when re-enabled */ |
aff622495
|
1894 |
if (compaction_deferred(zone, sc->order)) |
fe4b1b244
|
1895 1896 1897 1898 1899 1900 1901 1902 |
return watermark_ok; /* If compaction is not ready to start, keep reclaiming */ if (!compaction_suitable(zone, sc->order)) return false; return watermark_ok; } |
1da177e4c
|
1903 1904 1905 1906 1907 |
/* * 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. * |
418589663
|
1908 1909 |
* We reclaim from a zone even if that zone is over high_wmark_pages(zone). * Because: |
1da177e4c
|
1910 1911 |
* a) The caller may be trying to free *extra* pages to satisfy a higher-order * allocation or |
418589663
|
1912 1913 1914 |
* b) The target zone may be at high_wmark_pages(zone) but the lower zones * must go *over* high_wmark_pages(zone) to satisfy the `incremental min' * zone defense algorithm. |
1da177e4c
|
1915 |
* |
1da177e4c
|
1916 1917 |
* If a zone is deemed to be full of pinned pages then just give it a light * scan then give up on it. |
e0c23279c
|
1918 1919 |
* * This function returns true if a zone is being reclaimed for a costly |
fe4b1b244
|
1920 |
* high-order allocation and compaction is ready to begin. This indicates to |
0cee34fd7
|
1921 1922 |
* the caller that it should consider retrying the allocation instead of * further reclaim. |
1da177e4c
|
1923 |
*/ |
e0c23279c
|
1924 |
static bool shrink_zones(int priority, struct zonelist *zonelist, |
05ff51376
|
1925 |
struct scan_control *sc) |
1da177e4c
|
1926 |
{ |
dd1a239f6
|
1927 |
struct zoneref *z; |
54a6eb5c4
|
1928 |
struct zone *zone; |
d149e3b25
|
1929 1930 |
unsigned long nr_soft_reclaimed; unsigned long nr_soft_scanned; |
0cee34fd7
|
1931 |
bool aborted_reclaim = false; |
1cfb419b3
|
1932 |
|
cc715d99e
|
1933 1934 1935 1936 1937 1938 1939 |
/* * 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 */ if (buffer_heads_over_limit) sc->gfp_mask |= __GFP_HIGHMEM; |
d4debc66d
|
1940 1941 |
for_each_zone_zonelist_nodemask(zone, z, zonelist, gfp_zone(sc->gfp_mask), sc->nodemask) { |
f3fe65122
|
1942 |
if (!populated_zone(zone)) |
1da177e4c
|
1943 |
continue; |
1cfb419b3
|
1944 1945 1946 1947 |
/* * Take care memory controller reclaiming has small influence * to global LRU. */ |
89b5fae53
|
1948 |
if (global_reclaim(sc)) { |
1cfb419b3
|
1949 1950 |
if (!cpuset_zone_allowed_hardwall(zone, GFP_KERNEL)) continue; |
93e4a89a8
|
1951 |
if (zone->all_unreclaimable && priority != DEF_PRIORITY) |
1cfb419b3
|
1952 |
continue; /* Let kswapd poll it */ |
e0887c19b
|
1953 1954 |
if (COMPACTION_BUILD) { /* |
e0c23279c
|
1955 1956 1957 1958 1959 |
* 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 |
c7cfa37b7
|
1960 1961 |
* noticeable problem, like transparent huge * page allocations. |
e0887c19b
|
1962 |
*/ |
fe4b1b244
|
1963 |
if (compaction_ready(zone, sc)) { |
0cee34fd7
|
1964 |
aborted_reclaim = true; |
e0887c19b
|
1965 |
continue; |
e0c23279c
|
1966 |
} |
e0887c19b
|
1967 |
} |
ac34a1a3c
|
1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 |
/* * 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; nr_soft_reclaimed = mem_cgroup_soft_limit_reclaim(zone, sc->order, sc->gfp_mask, &nr_soft_scanned); sc->nr_reclaimed += nr_soft_reclaimed; sc->nr_scanned += nr_soft_scanned; /* need some check for avoid more shrink_zone() */ |
1cfb419b3
|
1981 |
} |
408d85441
|
1982 |
|
a79311c14
|
1983 |
shrink_zone(priority, zone, sc); |
1da177e4c
|
1984 |
} |
e0c23279c
|
1985 |
|
0cee34fd7
|
1986 |
return aborted_reclaim; |
d1908362a
|
1987 1988 1989 1990 1991 1992 |
} static bool zone_reclaimable(struct zone *zone) { return zone->pages_scanned < zone_reclaimable_pages(zone) * 6; } |
929bea7c7
|
1993 |
/* All zones in zonelist are unreclaimable? */ |
d1908362a
|
1994 1995 1996 1997 1998 |
static bool all_unreclaimable(struct zonelist *zonelist, struct scan_control *sc) { struct zoneref *z; struct zone *zone; |
d1908362a
|
1999 2000 2001 2002 2003 2004 2005 |
for_each_zone_zonelist_nodemask(zone, z, zonelist, gfp_zone(sc->gfp_mask), sc->nodemask) { if (!populated_zone(zone)) continue; if (!cpuset_zone_allowed_hardwall(zone, GFP_KERNEL)) continue; |
929bea7c7
|
2006 2007 |
if (!zone->all_unreclaimable) return false; |
d1908362a
|
2008 |
} |
929bea7c7
|
2009 |
return true; |
1da177e4c
|
2010 |
} |
4f98a2fee
|
2011 |
|
1da177e4c
|
2012 2013 2014 2015 2016 2017 2018 2019 |
/* * 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
|
2020 2021 2022 2023 |
* 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
|
2024 2025 2026 |
* * returns: 0, if no pages reclaimed * else, the number of pages reclaimed |
1da177e4c
|
2027 |
*/ |
dac1d27bc
|
2028 |
static unsigned long do_try_to_free_pages(struct zonelist *zonelist, |
a09ed5e00
|
2029 2030 |
struct scan_control *sc, struct shrink_control *shrink) |
1da177e4c
|
2031 2032 |
{ int priority; |
69e05944a
|
2033 |
unsigned long total_scanned = 0; |
1da177e4c
|
2034 |
struct reclaim_state *reclaim_state = current->reclaim_state; |
dd1a239f6
|
2035 |
struct zoneref *z; |
54a6eb5c4
|
2036 |
struct zone *zone; |
22fba3354
|
2037 |
unsigned long writeback_threshold; |
0cee34fd7
|
2038 |
bool aborted_reclaim; |
1da177e4c
|
2039 |
|
873b47717
|
2040 |
delayacct_freepages_start(); |
89b5fae53
|
2041 |
if (global_reclaim(sc)) |
1cfb419b3
|
2042 |
count_vm_event(ALLOCSTALL); |
1da177e4c
|
2043 2044 |
for (priority = DEF_PRIORITY; priority >= 0; priority--) { |
66e1707bc
|
2045 |
sc->nr_scanned = 0; |
0cee34fd7
|
2046 |
aborted_reclaim = shrink_zones(priority, zonelist, sc); |
e0c23279c
|
2047 |
|
66e1707bc
|
2048 2049 2050 2051 |
/* * Don't shrink slabs when reclaiming memory from * over limit cgroups */ |
89b5fae53
|
2052 |
if (global_reclaim(sc)) { |
c6a8a8c58
|
2053 |
unsigned long lru_pages = 0; |
d4debc66d
|
2054 2055 |
for_each_zone_zonelist(zone, z, zonelist, gfp_zone(sc->gfp_mask)) { |
c6a8a8c58
|
2056 2057 2058 2059 2060 |
if (!cpuset_zone_allowed_hardwall(zone, GFP_KERNEL)) continue; lru_pages += zone_reclaimable_pages(zone); } |
1495f230f
|
2061 |
shrink_slab(shrink, sc->nr_scanned, lru_pages); |
91a45470f
|
2062 |
if (reclaim_state) { |
a79311c14
|
2063 |
sc->nr_reclaimed += reclaim_state->reclaimed_slab; |
91a45470f
|
2064 2065 |
reclaim_state->reclaimed_slab = 0; } |
1da177e4c
|
2066 |
} |
66e1707bc
|
2067 |
total_scanned += sc->nr_scanned; |
bb21c7ce1
|
2068 |
if (sc->nr_reclaimed >= sc->nr_to_reclaim) |
1da177e4c
|
2069 |
goto out; |
1da177e4c
|
2070 2071 2072 2073 2074 2075 2076 2077 |
/* * Try to write back as many pages as we just scanned. This * tends to cause slow streaming writers to write data to the * disk smoothly, at the dirtying rate, which is nice. But * that's undesirable in laptop mode, where we *want* lumpy * writeout. So in laptop mode, write out the whole world. */ |
22fba3354
|
2078 2079 |
writeback_threshold = sc->nr_to_reclaim + sc->nr_to_reclaim / 2; if (total_scanned > writeback_threshold) { |
0e175a183
|
2080 2081 |
wakeup_flusher_threads(laptop_mode ? 0 : total_scanned, WB_REASON_TRY_TO_FREE_PAGES); |
66e1707bc
|
2082 |
sc->may_writepage = 1; |
1da177e4c
|
2083 2084 2085 |
} /* Take a nap, wait for some writeback to complete */ |
7b51755c3
|
2086 |
if (!sc->hibernation_mode && sc->nr_scanned && |
0e093d997
|
2087 2088 2089 2090 |
priority < DEF_PRIORITY - 2) { struct zone *preferred_zone; first_zones_zonelist(zonelist, gfp_zone(sc->gfp_mask), |
f33261d75
|
2091 2092 |
&cpuset_current_mems_allowed, &preferred_zone); |
0e093d997
|
2093 2094 |
wait_iff_congested(preferred_zone, BLK_RW_ASYNC, HZ/10); } |
1da177e4c
|
2095 |
} |
bb21c7ce1
|
2096 |
|
1da177e4c
|
2097 |
out: |
873b47717
|
2098 |
delayacct_freepages_end(); |
bb21c7ce1
|
2099 2100 |
if (sc->nr_reclaimed) return sc->nr_reclaimed; |
929bea7c7
|
2101 2102 2103 2104 2105 2106 2107 |
/* * As hibernation is going on, kswapd is freezed so that it can't mark * the zone into all_unreclaimable. Thus bypassing all_unreclaimable * check. */ if (oom_killer_disabled) return 0; |
0cee34fd7
|
2108 2109 |
/* Aborted reclaim to try compaction? don't OOM, then */ if (aborted_reclaim) |
7335084d4
|
2110 |
return 1; |
bb21c7ce1
|
2111 |
/* top priority shrink_zones still had more to do? don't OOM, then */ |
89b5fae53
|
2112 |
if (global_reclaim(sc) && !all_unreclaimable(zonelist, sc)) |
bb21c7ce1
|
2113 2114 2115 |
return 1; return 0; |
1da177e4c
|
2116 |
} |
dac1d27bc
|
2117 |
unsigned long try_to_free_pages(struct zonelist *zonelist, int order, |
327c0e968
|
2118 |
gfp_t gfp_mask, nodemask_t *nodemask) |
66e1707bc
|
2119 |
{ |
33906bc5c
|
2120 |
unsigned long nr_reclaimed; |
66e1707bc
|
2121 2122 2123 |
struct scan_control sc = { .gfp_mask = gfp_mask, .may_writepage = !laptop_mode, |
22fba3354
|
2124 |
.nr_to_reclaim = SWAP_CLUSTER_MAX, |
a6dc60f89
|
2125 |
.may_unmap = 1, |
2e2e42598
|
2126 |
.may_swap = 1, |
66e1707bc
|
2127 |
.order = order, |
f16015fbf
|
2128 |
.target_mem_cgroup = NULL, |
327c0e968
|
2129 |
.nodemask = nodemask, |
66e1707bc
|
2130 |
}; |
a09ed5e00
|
2131 2132 2133 |
struct shrink_control shrink = { .gfp_mask = sc.gfp_mask, }; |
66e1707bc
|
2134 |
|
33906bc5c
|
2135 2136 2137 |
trace_mm_vmscan_direct_reclaim_begin(order, sc.may_writepage, gfp_mask); |
a09ed5e00
|
2138 |
nr_reclaimed = do_try_to_free_pages(zonelist, &sc, &shrink); |
33906bc5c
|
2139 2140 2141 2142 |
trace_mm_vmscan_direct_reclaim_end(nr_reclaimed); return nr_reclaimed; |
66e1707bc
|
2143 |
} |
00f0b8259
|
2144 |
#ifdef CONFIG_CGROUP_MEM_RES_CTLR |
66e1707bc
|
2145 |
|
72835c86c
|
2146 |
unsigned long mem_cgroup_shrink_node_zone(struct mem_cgroup *memcg, |
4e4169535
|
2147 |
gfp_t gfp_mask, bool noswap, |
0ae5e89c6
|
2148 2149 |
struct zone *zone, unsigned long *nr_scanned) |
4e4169535
|
2150 2151 |
{ struct scan_control sc = { |
0ae5e89c6
|
2152 |
.nr_scanned = 0, |
b8f5c5664
|
2153 |
.nr_to_reclaim = SWAP_CLUSTER_MAX, |
4e4169535
|
2154 2155 2156 |
.may_writepage = !laptop_mode, .may_unmap = 1, .may_swap = !noswap, |
4e4169535
|
2157 |
.order = 0, |
72835c86c
|
2158 |
.target_mem_cgroup = memcg, |
4e4169535
|
2159 |
}; |
5660048cc
|
2160 |
struct mem_cgroup_zone mz = { |
72835c86c
|
2161 |
.mem_cgroup = memcg, |
5660048cc
|
2162 2163 |
.zone = zone, }; |
0ae5e89c6
|
2164 |
|
4e4169535
|
2165 2166 |
sc.gfp_mask = (gfp_mask & GFP_RECLAIM_MASK) | (GFP_HIGHUSER_MOVABLE & ~GFP_RECLAIM_MASK); |
bdce6d9eb
|
2167 2168 2169 2170 |
trace_mm_vmscan_memcg_softlimit_reclaim_begin(0, sc.may_writepage, sc.gfp_mask); |
4e4169535
|
2171 2172 2173 2174 2175 2176 2177 |
/* * NOTE: Although we can get the priority field, using it * here is not a good idea, since it limits the pages we can scan. * if we don't reclaim here, the shrink_zone from balance_pgdat * will pick up pages from other mem cgroup's as well. We hack * the priority and make it zero. */ |
5660048cc
|
2178 |
shrink_mem_cgroup_zone(0, &mz, &sc); |
bdce6d9eb
|
2179 2180 |
trace_mm_vmscan_memcg_softlimit_reclaim_end(sc.nr_reclaimed); |
0ae5e89c6
|
2181 |
*nr_scanned = sc.nr_scanned; |
4e4169535
|
2182 2183 |
return sc.nr_reclaimed; } |
72835c86c
|
2184 |
unsigned long try_to_free_mem_cgroup_pages(struct mem_cgroup *memcg, |
a7885eb8a
|
2185 |
gfp_t gfp_mask, |
185efc0f9
|
2186 |
bool noswap) |
66e1707bc
|
2187 |
{ |
4e4169535
|
2188 |
struct zonelist *zonelist; |
bdce6d9eb
|
2189 |
unsigned long nr_reclaimed; |
889976dbc
|
2190 |
int nid; |
66e1707bc
|
2191 |
struct scan_control sc = { |
66e1707bc
|
2192 |
.may_writepage = !laptop_mode, |
a6dc60f89
|
2193 |
.may_unmap = 1, |
2e2e42598
|
2194 |
.may_swap = !noswap, |
22fba3354
|
2195 |
.nr_to_reclaim = SWAP_CLUSTER_MAX, |
66e1707bc
|
2196 |
.order = 0, |
72835c86c
|
2197 |
.target_mem_cgroup = memcg, |
327c0e968
|
2198 |
.nodemask = NULL, /* we don't care the placement */ |
a09ed5e00
|
2199 2200 2201 2202 2203 |
.gfp_mask = (gfp_mask & GFP_RECLAIM_MASK) | (GFP_HIGHUSER_MOVABLE & ~GFP_RECLAIM_MASK), }; struct shrink_control shrink = { .gfp_mask = sc.gfp_mask, |
66e1707bc
|
2204 |
}; |
66e1707bc
|
2205 |
|
889976dbc
|
2206 2207 2208 2209 2210 |
/* * Unlike direct reclaim via alloc_pages(), memcg's reclaim doesn't * take care of from where we get pages. So the node where we start the * scan does not need to be the current node. */ |
72835c86c
|
2211 |
nid = mem_cgroup_select_victim_node(memcg); |
889976dbc
|
2212 2213 |
zonelist = NODE_DATA(nid)->node_zonelists; |
bdce6d9eb
|
2214 2215 2216 2217 |
trace_mm_vmscan_memcg_reclaim_begin(0, sc.may_writepage, sc.gfp_mask); |
a09ed5e00
|
2218 |
nr_reclaimed = do_try_to_free_pages(zonelist, &sc, &shrink); |
bdce6d9eb
|
2219 2220 2221 2222 |
trace_mm_vmscan_memcg_reclaim_end(nr_reclaimed); return nr_reclaimed; |
66e1707bc
|
2223 2224 |
} #endif |
f16015fbf
|
2225 2226 2227 |
static void age_active_anon(struct zone *zone, struct scan_control *sc, int priority) { |
b95a2f2d4
|
2228 |
struct mem_cgroup *memcg; |
f16015fbf
|
2229 |
|
b95a2f2d4
|
2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 |
if (!total_swap_pages) return; memcg = mem_cgroup_iter(NULL, NULL, NULL); do { struct mem_cgroup_zone mz = { .mem_cgroup = memcg, .zone = zone, }; if (inactive_anon_is_low(&mz)) shrink_active_list(SWAP_CLUSTER_MAX, &mz, sc, priority, 0); memcg = mem_cgroup_iter(NULL, memcg, NULL); } while (memcg); |
f16015fbf
|
2246 |
} |
1741c8775
|
2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 |
/* * pgdat_balanced is used when checking if a node is balanced for high-order * allocations. Only zones that meet watermarks and are in a zone allowed * by the callers classzone_idx are added to balanced_pages. The total of * balanced pages must be at least 25% of the zones allowed by classzone_idx * for the node to be considered balanced. Forcing all zones to be balanced * for high orders can cause excessive reclaim when there are imbalanced zones. * The choice of 25% is due to * o a 16M DMA zone that is balanced will not balance a zone on any * reasonable sized machine * o On all other machines, the top zone must be at least a reasonable |
25985edce
|
2258 |
* percentage of the middle zones. For example, on 32-bit x86, highmem |
1741c8775
|
2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 |
* would need to be at least 256M for it to be balance a whole node. * Similarly, on x86-64 the Normal zone would need to be at least 1G * to balance a node on its own. These seemed like reasonable ratios. */ static bool pgdat_balanced(pg_data_t *pgdat, unsigned long balanced_pages, int classzone_idx) { unsigned long present_pages = 0; int i; for (i = 0; i <= classzone_idx; i++) present_pages += pgdat->node_zones[i].present_pages; |
4746efded
|
2271 2272 |
/* A special case here: if zone has no page, we think it's balanced */ return balanced_pages >= (present_pages >> 2); |
1741c8775
|
2273 |
} |
f50de2d38
|
2274 |
/* is kswapd sleeping prematurely? */ |
dc83edd94
|
2275 2276 |
static bool sleeping_prematurely(pg_data_t *pgdat, int order, long remaining, int classzone_idx) |
f50de2d38
|
2277 |
{ |
bb3ab5968
|
2278 |
int i; |
1741c8775
|
2279 2280 |
unsigned long balanced = 0; bool all_zones_ok = true; |
f50de2d38
|
2281 2282 2283 |
/* If a direct reclaimer woke kswapd within HZ/10, it's premature */ if (remaining) |
dc83edd94
|
2284 |
return true; |
f50de2d38
|
2285 |
|
0abdee2bd
|
2286 |
/* Check the watermark levels */ |
08951e545
|
2287 |
for (i = 0; i <= classzone_idx; i++) { |
bb3ab5968
|
2288 2289 2290 2291 |
struct zone *zone = pgdat->node_zones + i; if (!populated_zone(zone)) continue; |
355b09c47
|
2292 2293 2294 2295 2296 2297 2298 2299 |
/* * balance_pgdat() skips over all_unreclaimable after * DEF_PRIORITY. Effectively, it considers them balanced so * they must be considered balanced here as well if kswapd * is to sleep */ if (zone->all_unreclaimable) { balanced += zone->present_pages; |
de3fab393
|
2300 |
continue; |
355b09c47
|
2301 |
} |
de3fab393
|
2302 |
|
88f5acf88
|
2303 |
if (!zone_watermark_ok_safe(zone, order, high_wmark_pages(zone), |
da175d06b
|
2304 |
i, 0)) |
1741c8775
|
2305 2306 2307 |
all_zones_ok = false; else balanced += zone->present_pages; |
bb3ab5968
|
2308 |
} |
f50de2d38
|
2309 |
|
1741c8775
|
2310 2311 2312 2313 2314 2315 |
/* * For high-order requests, the balanced zones must contain at least * 25% of the nodes pages for kswapd to sleep. For order-0, all zones * must be balanced */ if (order) |
afc7e326a
|
2316 |
return !pgdat_balanced(pgdat, balanced, classzone_idx); |
1741c8775
|
2317 2318 |
else return !all_zones_ok; |
f50de2d38
|
2319 |
} |
1da177e4c
|
2320 2321 |
/* * For kswapd, balance_pgdat() will work across all this node's zones until |
418589663
|
2322 |
* they are all at high_wmark_pages(zone). |
1da177e4c
|
2323 |
* |
0abdee2bd
|
2324 |
* Returns the final order kswapd was reclaiming at |
1da177e4c
|
2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 |
* * There is special handling here for zones which are full of pinned pages. * This can happen if the pages are all mlocked, or if they are all used by * device drivers (say, ZONE_DMA). Or if they are all in use by hugetlb. * What we do is to detect the case where all pages in the zone have been * scanned twice and there has been zero successful reclaim. Mark the zone as * dead and from now on, only perform a short scan. Basically we're polling * the zone for when the problem goes away. * * kswapd scans the zones in the highmem->normal->dma direction. It skips |
418589663
|
2335 2336 2337 2338 2339 |
* zones which have free_pages > high_wmark_pages(zone), but once a zone is * found to have free_pages <= high_wmark_pages(zone), we scan that zone and the * lower zones regardless of the number of free pages in the lower zones. This * interoperates with the page allocator fallback scheme to ensure that aging * of pages is balanced across the zones. |
1da177e4c
|
2340 |
*/ |
995047488
|
2341 |
static unsigned long balance_pgdat(pg_data_t *pgdat, int order, |
dc83edd94
|
2342 |
int *classzone_idx) |
1da177e4c
|
2343 |
{ |
1da177e4c
|
2344 |
int all_zones_ok; |
1741c8775
|
2345 |
unsigned long balanced; |
1da177e4c
|
2346 2347 |
int priority; int i; |
995047488
|
2348 |
int end_zone = 0; /* Inclusive. 0 = ZONE_DMA */ |
69e05944a
|
2349 |
unsigned long total_scanned; |
1da177e4c
|
2350 |
struct reclaim_state *reclaim_state = current->reclaim_state; |
0ae5e89c6
|
2351 2352 |
unsigned long nr_soft_reclaimed; unsigned long nr_soft_scanned; |
179e96395
|
2353 2354 |
struct scan_control sc = { .gfp_mask = GFP_KERNEL, |
a6dc60f89
|
2355 |
.may_unmap = 1, |
2e2e42598
|
2356 |
.may_swap = 1, |
22fba3354
|
2357 2358 2359 2360 2361 |
/* * kswapd doesn't want to be bailed out while reclaim. because * we want to put equal scanning pressure on each zone. */ .nr_to_reclaim = ULONG_MAX, |
5ad333eb6
|
2362 |
.order = order, |
f16015fbf
|
2363 |
.target_mem_cgroup = NULL, |
179e96395
|
2364 |
}; |
a09ed5e00
|
2365 2366 2367 |
struct shrink_control shrink = { .gfp_mask = sc.gfp_mask, }; |
1da177e4c
|
2368 2369 |
loop_again: total_scanned = 0; |
a79311c14
|
2370 |
sc.nr_reclaimed = 0; |
c0bbbc73d
|
2371 |
sc.may_writepage = !laptop_mode; |
f8891e5e1
|
2372 |
count_vm_event(PAGEOUTRUN); |
1da177e4c
|
2373 |
|
1da177e4c
|
2374 |
for (priority = DEF_PRIORITY; priority >= 0; priority--) { |
1da177e4c
|
2375 |
unsigned long lru_pages = 0; |
bb3ab5968
|
2376 |
int has_under_min_watermark_zone = 0; |
1da177e4c
|
2377 2378 |
all_zones_ok = 1; |
1741c8775
|
2379 |
balanced = 0; |
1da177e4c
|
2380 |
|
d6277db4a
|
2381 2382 2383 2384 2385 2386 |
/* * Scan in the highmem->dma direction for the highest * zone which needs scanning */ for (i = pgdat->nr_zones - 1; i >= 0; i--) { struct zone *zone = pgdat->node_zones + i; |
1da177e4c
|
2387 |
|
d6277db4a
|
2388 2389 |
if (!populated_zone(zone)) continue; |
1da177e4c
|
2390 |
|
93e4a89a8
|
2391 |
if (zone->all_unreclaimable && priority != DEF_PRIORITY) |
d6277db4a
|
2392 |
continue; |
1da177e4c
|
2393 |
|
556adecba
|
2394 2395 2396 2397 |
/* * Do some background aging of the anon list, to give * pages a chance to be referenced before reclaiming. */ |
f16015fbf
|
2398 |
age_active_anon(zone, &sc, priority); |
556adecba
|
2399 |
|
cc715d99e
|
2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 |
/* * If the number of buffer_heads in the machine * exceeds the maximum allowed level and this node * has a highmem zone, force kswapd to reclaim from * it to relieve lowmem pressure. */ if (buffer_heads_over_limit && is_highmem_idx(i)) { end_zone = i; break; } |
88f5acf88
|
2410 |
if (!zone_watermark_ok_safe(zone, order, |
418589663
|
2411 |
high_wmark_pages(zone), 0, 0)) { |
d6277db4a
|
2412 |
end_zone = i; |
e1dbeda60
|
2413 |
break; |
439423f68
|
2414 2415 2416 |
} else { /* If balanced, clear the congested flag */ zone_clear_flag(zone, ZONE_CONGESTED); |
1da177e4c
|
2417 |
} |
1da177e4c
|
2418 |
} |
e1dbeda60
|
2419 2420 |
if (i < 0) goto out; |
1da177e4c
|
2421 2422 |
for (i = 0; i <= end_zone; i++) { struct zone *zone = pgdat->node_zones + i; |
adea02a1b
|
2423 |
lru_pages += zone_reclaimable_pages(zone); |
1da177e4c
|
2424 2425 2426 2427 2428 2429 2430 2431 2432 2433 2434 2435 2436 |
} /* * Now scan the zone in the dma->highmem direction, stopping * at the last zone which needs scanning. * * We do this because the page allocator works in the opposite * direction. This prevents the page allocator from allocating * pages behind kswapd's direction of progress, which would * cause too much scanning of the lower zones. */ for (i = 0; i <= end_zone; i++) { struct zone *zone = pgdat->node_zones + i; |
fe2c2a106
|
2437 |
int nr_slab, testorder; |
8afdcece4
|
2438 |
unsigned long balance_gap; |
1da177e4c
|
2439 |
|
f3fe65122
|
2440 |
if (!populated_zone(zone)) |
1da177e4c
|
2441 |
continue; |
93e4a89a8
|
2442 |
if (zone->all_unreclaimable && priority != DEF_PRIORITY) |
1da177e4c
|
2443 |
continue; |
1da177e4c
|
2444 |
sc.nr_scanned = 0; |
4e4169535
|
2445 |
|
0ae5e89c6
|
2446 |
nr_soft_scanned = 0; |
4e4169535
|
2447 2448 |
/* * Call soft limit reclaim before calling shrink_zone. |
4e4169535
|
2449 |
*/ |
0ae5e89c6
|
2450 2451 2452 2453 2454 |
nr_soft_reclaimed = mem_cgroup_soft_limit_reclaim(zone, order, sc.gfp_mask, &nr_soft_scanned); sc.nr_reclaimed += nr_soft_reclaimed; total_scanned += nr_soft_scanned; |
00918b6ab
|
2455 |
|
32a4330d4
|
2456 |
/* |
8afdcece4
|
2457 2458 2459 2460 2461 2462 |
* We put equal pressure on every zone, unless * one zone has way too many pages free * already. The "too many pages" is defined * as the high wmark plus a "gap" where the * gap is either the low watermark or 1% * of the zone, whichever is smaller. |
32a4330d4
|
2463 |
*/ |
8afdcece4
|
2464 2465 2466 2467 |
balance_gap = min(low_wmark_pages(zone), (zone->present_pages + KSWAPD_ZONE_BALANCE_GAP_RATIO-1) / KSWAPD_ZONE_BALANCE_GAP_RATIO); |
fe2c2a106
|
2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 |
/* * Kswapd reclaims only single pages with compaction * enabled. Trying too hard to reclaim until contiguous * free pages have become available can hurt performance * by evicting too much useful data from memory. * Do not reclaim more than needed for compaction. */ testorder = order; if (COMPACTION_BUILD && order && compaction_suitable(zone, order) != COMPACT_SKIPPED) testorder = 0; |
cc715d99e
|
2480 |
if ((buffer_heads_over_limit && is_highmem_idx(i)) || |
643ac9fc5
|
2481 |
!zone_watermark_ok_safe(zone, testorder, |
8afdcece4
|
2482 |
high_wmark_pages(zone) + balance_gap, |
d7868dae8
|
2483 |
end_zone, 0)) { |
a79311c14
|
2484 |
shrink_zone(priority, zone, &sc); |
5a03b051e
|
2485 |
|
d7868dae8
|
2486 2487 2488 2489 2490 2491 2492 2493 |
reclaim_state->reclaimed_slab = 0; nr_slab = shrink_slab(&shrink, sc.nr_scanned, lru_pages); sc.nr_reclaimed += reclaim_state->reclaimed_slab; total_scanned += sc.nr_scanned; if (nr_slab == 0 && !zone_reclaimable(zone)) zone->all_unreclaimable = 1; } |
1da177e4c
|
2494 2495 2496 2497 2498 2499 |
/* * If we've done a decent amount of scanning and * the reclaim ratio is low, start doing writepage * even in laptop mode */ if (total_scanned > SWAP_CLUSTER_MAX * 2 && |
a79311c14
|
2500 |
total_scanned > sc.nr_reclaimed + sc.nr_reclaimed / 2) |
1da177e4c
|
2501 |
sc.may_writepage = 1; |
bb3ab5968
|
2502 |
|
215ddd666
|
2503 2504 2505 |
if (zone->all_unreclaimable) { if (end_zone && end_zone == i) end_zone--; |
d7868dae8
|
2506 |
continue; |
215ddd666
|
2507 |
} |
d7868dae8
|
2508 |
|
fe2c2a106
|
2509 |
if (!zone_watermark_ok_safe(zone, testorder, |
45973d74f
|
2510 2511 2512 2513 2514 2515 2516 |
high_wmark_pages(zone), end_zone, 0)) { all_zones_ok = 0; /* * We are still under min water mark. This * means that we have a GFP_ATOMIC allocation * failure risk. Hurry up! */ |
88f5acf88
|
2517 |
if (!zone_watermark_ok_safe(zone, order, |
45973d74f
|
2518 2519 |
min_wmark_pages(zone), end_zone, 0)) has_under_min_watermark_zone = 1; |
0e093d997
|
2520 2521 2522 2523 2524 2525 2526 2527 2528 |
} else { /* * If a zone reaches its high watermark, * consider it to be no longer congested. It's * possible there are dirty pages backed by * congested BDIs but as pressure is relieved, * spectulatively avoid congestion waits */ zone_clear_flag(zone, ZONE_CONGESTED); |
dc83edd94
|
2529 |
if (i <= *classzone_idx) |
1741c8775
|
2530 |
balanced += zone->present_pages; |
45973d74f
|
2531 |
} |
bb3ab5968
|
2532 |
|
1da177e4c
|
2533 |
} |
dc83edd94
|
2534 |
if (all_zones_ok || (order && pgdat_balanced(pgdat, balanced, *classzone_idx))) |
1da177e4c
|
2535 2536 2537 2538 2539 |
break; /* kswapd: all done */ /* * OK, kswapd is getting into trouble. Take a nap, then take * another pass across the zones. */ |
bb3ab5968
|
2540 2541 2542 2543 2544 2545 |
if (total_scanned && (priority < DEF_PRIORITY - 2)) { if (has_under_min_watermark_zone) count_vm_event(KSWAPD_SKIP_CONGESTION_WAIT); else congestion_wait(BLK_RW_ASYNC, HZ/10); } |
1da177e4c
|
2546 2547 2548 2549 2550 2551 2552 |
/* * We do this so kswapd doesn't build up large priorities for * example when it is freeing in parallel with allocators. It * matches the direct reclaim path behaviour in terms of impact * on zone->*_priority. */ |
a79311c14
|
2553 |
if (sc.nr_reclaimed >= SWAP_CLUSTER_MAX) |
1da177e4c
|
2554 2555 2556 |
break; } out: |
995047488
|
2557 2558 2559 |
/* * order-0: All zones must meet high watermark for a balanced node |
1741c8775
|
2560 2561 |
* high-order: Balanced zones must make up at least 25% of the node * for the node to be balanced |
995047488
|
2562 |
*/ |
dc83edd94
|
2563 |
if (!(all_zones_ok || (order && pgdat_balanced(pgdat, balanced, *classzone_idx)))) { |
1da177e4c
|
2564 |
cond_resched(); |
8357376d3
|
2565 2566 |
try_to_freeze(); |
73ce02e96
|
2567 2568 2569 2570 2571 2572 2573 2574 2575 2576 2577 2578 2579 2580 2581 2582 |
/* * Fragmentation may mean that the system cannot be * rebalanced for high-order allocations in all zones. * At this point, if nr_reclaimed < SWAP_CLUSTER_MAX, * it means the zones have been fully scanned and are still * not balanced. For high-order allocations, there is * little point trying all over again as kswapd may * infinite loop. * * Instead, recheck all watermarks at order-0 as they * are the most important. If watermarks are ok, kswapd will go * back to sleep. High-order users can still perform direct * reclaim if they wish. */ if (sc.nr_reclaimed < SWAP_CLUSTER_MAX) order = sc.order = 0; |
1da177e4c
|
2583 2584 |
goto loop_again; } |
995047488
|
2585 2586 2587 2588 2589 2590 2591 2592 2593 |
/* * If kswapd was reclaiming at a higher order, it has the option of * sleeping without all zones being balanced. Before it does, it must * ensure that the watermarks for order-0 on *all* zones are met and * that the congestion flags are cleared. The congestion flag must * be cleared as kswapd is the only mechanism that clears the flag * and it is potentially going to sleep here. */ if (order) { |
7be62de99
|
2594 |
int zones_need_compaction = 1; |
995047488
|
2595 2596 2597 2598 2599 2600 2601 2602 |
for (i = 0; i <= end_zone; i++) { struct zone *zone = pgdat->node_zones + i; if (!populated_zone(zone)) continue; if (zone->all_unreclaimable && priority != DEF_PRIORITY) continue; |
fe2c2a106
|
2603 |
/* Would compaction fail due to lack of free memory? */ |
496b919b3
|
2604 2605 |
if (COMPACTION_BUILD && compaction_suitable(zone, order) == COMPACT_SKIPPED) |
fe2c2a106
|
2606 |
goto loop_again; |
995047488
|
2607 2608 2609 2610 2611 2612 |
/* Confirm the zone is balanced for order-0 */ if (!zone_watermark_ok(zone, 0, high_wmark_pages(zone), 0, 0)) { order = sc.order = 0; goto loop_again; } |
7be62de99
|
2613 2614 2615 2616 |
/* Check if the memory needs to be defragmented. */ if (zone_watermark_ok(zone, order, low_wmark_pages(zone), *classzone_idx, 0)) zones_need_compaction = 0; |
995047488
|
2617 2618 2619 |
/* If balanced, clear the congested flag */ zone_clear_flag(zone, ZONE_CONGESTED); } |
7be62de99
|
2620 2621 2622 |
if (zones_need_compaction) compact_pgdat(pgdat, order); |
995047488
|
2623 |
} |
0abdee2bd
|
2624 2625 2626 2627 2628 2629 |
/* * Return the order we were reclaiming at so sleeping_prematurely() * makes a decision on the order we were last reclaiming at. However, * if another caller entered the allocator slow path while kswapd * was awake, order will remain at the higher level */ |
dc83edd94
|
2630 |
*classzone_idx = end_zone; |
0abdee2bd
|
2631 |
return order; |
1da177e4c
|
2632 |
} |
dc83edd94
|
2633 |
static void kswapd_try_to_sleep(pg_data_t *pgdat, int order, int classzone_idx) |
f0bc0a60b
|
2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 |
{ long remaining = 0; DEFINE_WAIT(wait); if (freezing(current) || kthread_should_stop()) return; prepare_to_wait(&pgdat->kswapd_wait, &wait, TASK_INTERRUPTIBLE); /* Try to sleep for a short interval */ |
dc83edd94
|
2644 |
if (!sleeping_prematurely(pgdat, order, remaining, classzone_idx)) { |
f0bc0a60b
|
2645 2646 2647 2648 2649 2650 2651 2652 2653 |
remaining = schedule_timeout(HZ/10); 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. */ |
dc83edd94
|
2654 |
if (!sleeping_prematurely(pgdat, order, remaining, classzone_idx)) { |
f0bc0a60b
|
2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 |
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); schedule(); 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
|
2676 2677 |
/* * The background pageout daemon, started as a kernel thread |
4f98a2fee
|
2678 |
* from the init process. |
1da177e4c
|
2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 |
* * 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) { |
215ddd666
|
2691 |
unsigned long order, new_order; |
d2ebd0f6b
|
2692 |
unsigned balanced_order; |
215ddd666
|
2693 |
int classzone_idx, new_classzone_idx; |
d2ebd0f6b
|
2694 |
int balanced_classzone_idx; |
1da177e4c
|
2695 2696 |
pg_data_t *pgdat = (pg_data_t*)p; struct task_struct *tsk = current; |
f0bc0a60b
|
2697 |
|
1da177e4c
|
2698 2699 2700 |
struct reclaim_state reclaim_state = { .reclaimed_slab = 0, }; |
a70f73028
|
2701 |
const struct cpumask *cpumask = cpumask_of_node(pgdat->node_id); |
1da177e4c
|
2702 |
|
cf40bd16f
|
2703 |
lockdep_set_current_reclaim_state(GFP_KERNEL); |
174596a0b
|
2704 |
if (!cpumask_empty(cpumask)) |
c5f59f083
|
2705 |
set_cpus_allowed_ptr(tsk, cpumask); |
1da177e4c
|
2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 |
current->reclaim_state = &reclaim_state; /* * 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
|
2720 |
tsk->flags |= PF_MEMALLOC | PF_SWAPWRITE | PF_KSWAPD; |
831441862
|
2721 |
set_freezable(); |
1da177e4c
|
2722 |
|
215ddd666
|
2723 |
order = new_order = 0; |
d2ebd0f6b
|
2724 |
balanced_order = 0; |
215ddd666
|
2725 |
classzone_idx = new_classzone_idx = pgdat->nr_zones - 1; |
d2ebd0f6b
|
2726 |
balanced_classzone_idx = classzone_idx; |
1da177e4c
|
2727 |
for ( ; ; ) { |
8fe23e057
|
2728 |
int ret; |
3e1d1d28d
|
2729 |
|
215ddd666
|
2730 2731 2732 2733 2734 |
/* * If the last balance_pgdat was unsuccessful it's unlikely a * new request of a similar or harder type will succeed soon * so consider going to sleep on the basis we reclaimed at */ |
d2ebd0f6b
|
2735 2736 |
if (balanced_classzone_idx >= new_classzone_idx && balanced_order == new_order) { |
215ddd666
|
2737 2738 2739 2740 2741 |
new_order = pgdat->kswapd_max_order; new_classzone_idx = pgdat->classzone_idx; pgdat->kswapd_max_order = 0; pgdat->classzone_idx = pgdat->nr_zones - 1; } |
995047488
|
2742 |
if (order < new_order || classzone_idx > new_classzone_idx) { |
1da177e4c
|
2743 2744 |
/* * Don't sleep if someone wants a larger 'order' |
995047488
|
2745 |
* allocation or has tigher zone constraints |
1da177e4c
|
2746 2747 |
*/ order = new_order; |
995047488
|
2748 |
classzone_idx = new_classzone_idx; |
1da177e4c
|
2749 |
} else { |
d2ebd0f6b
|
2750 2751 |
kswapd_try_to_sleep(pgdat, balanced_order, balanced_classzone_idx); |
1da177e4c
|
2752 |
order = pgdat->kswapd_max_order; |
995047488
|
2753 |
classzone_idx = pgdat->classzone_idx; |
f0dfcde09
|
2754 2755 |
new_order = order; new_classzone_idx = classzone_idx; |
4d40502ea
|
2756 |
pgdat->kswapd_max_order = 0; |
215ddd666
|
2757 |
pgdat->classzone_idx = pgdat->nr_zones - 1; |
1da177e4c
|
2758 |
} |
1da177e4c
|
2759 |
|
8fe23e057
|
2760 2761 2762 2763 2764 2765 2766 2767 |
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 */ |
33906bc5c
|
2768 2769 |
if (!ret) { trace_mm_vmscan_kswapd_wake(pgdat->node_id, order); |
d2ebd0f6b
|
2770 2771 2772 |
balanced_classzone_idx = classzone_idx; balanced_order = balance_pgdat(pgdat, order, &balanced_classzone_idx); |
33906bc5c
|
2773 |
} |
1da177e4c
|
2774 2775 2776 2777 2778 2779 2780 |
} return 0; } /* * A zone is low on free memory, so wake its kswapd task to service it. */ |
995047488
|
2781 |
void wakeup_kswapd(struct zone *zone, int order, enum zone_type classzone_idx) |
1da177e4c
|
2782 2783 |
{ pg_data_t *pgdat; |
f3fe65122
|
2784 |
if (!populated_zone(zone)) |
1da177e4c
|
2785 |
return; |
88f5acf88
|
2786 |
if (!cpuset_zone_allowed_hardwall(zone, GFP_KERNEL)) |
1da177e4c
|
2787 |
return; |
88f5acf88
|
2788 |
pgdat = zone->zone_pgdat; |
995047488
|
2789 |
if (pgdat->kswapd_max_order < order) { |
1da177e4c
|
2790 |
pgdat->kswapd_max_order = order; |
995047488
|
2791 2792 |
pgdat->classzone_idx = min(pgdat->classzone_idx, classzone_idx); } |
8d0986e28
|
2793 |
if (!waitqueue_active(&pgdat->kswapd_wait)) |
1da177e4c
|
2794 |
return; |
88f5acf88
|
2795 2796 2797 2798 |
if (zone_watermark_ok_safe(zone, order, low_wmark_pages(zone), 0, 0)) return; trace_mm_vmscan_wakeup_kswapd(pgdat->node_id, zone_idx(zone), order); |
8d0986e28
|
2799 |
wake_up_interruptible(&pgdat->kswapd_wait); |
1da177e4c
|
2800 |
} |
adea02a1b
|
2801 2802 2803 2804 2805 2806 2807 2808 |
/* * The reclaimable count would be mostly accurate. * The less reclaimable pages may be * - mlocked pages, which will be moved to unevictable list when encountered * - mapped pages, which may require several travels to be reclaimed * - dirty pages, which is not "instantly" reclaimable */ unsigned long global_reclaimable_pages(void) |
4f98a2fee
|
2809 |
{ |
adea02a1b
|
2810 2811 2812 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 |
int nr; nr = global_page_state(NR_ACTIVE_FILE) + global_page_state(NR_INACTIVE_FILE); if (nr_swap_pages > 0) nr += global_page_state(NR_ACTIVE_ANON) + global_page_state(NR_INACTIVE_ANON); return nr; } unsigned long zone_reclaimable_pages(struct zone *zone) { int nr; nr = zone_page_state(zone, NR_ACTIVE_FILE) + zone_page_state(zone, NR_INACTIVE_FILE); if (nr_swap_pages > 0) nr += zone_page_state(zone, NR_ACTIVE_ANON) + zone_page_state(zone, NR_INACTIVE_ANON); return nr; |
4f98a2fee
|
2834 |
} |
c6f37f121
|
2835 |
#ifdef CONFIG_HIBERNATION |
1da177e4c
|
2836 |
/* |
7b51755c3
|
2837 |
* Try to free `nr_to_reclaim' of memory, system-wide, and return the number of |
d6277db4a
|
2838 2839 2840 2841 2842 |
* 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
|
2843 |
*/ |
7b51755c3
|
2844 |
unsigned long shrink_all_memory(unsigned long nr_to_reclaim) |
1da177e4c
|
2845 |
{ |
d6277db4a
|
2846 |
struct reclaim_state reclaim_state; |
d6277db4a
|
2847 |
struct scan_control sc = { |
7b51755c3
|
2848 2849 2850 |
.gfp_mask = GFP_HIGHUSER_MOVABLE, .may_swap = 1, .may_unmap = 1, |
d6277db4a
|
2851 |
.may_writepage = 1, |
7b51755c3
|
2852 2853 |
.nr_to_reclaim = nr_to_reclaim, .hibernation_mode = 1, |
7b51755c3
|
2854 |
.order = 0, |
1da177e4c
|
2855 |
}; |
a09ed5e00
|
2856 2857 2858 2859 |
struct shrink_control shrink = { .gfp_mask = sc.gfp_mask, }; struct zonelist *zonelist = node_zonelist(numa_node_id(), sc.gfp_mask); |
7b51755c3
|
2860 2861 |
struct task_struct *p = current; unsigned long nr_reclaimed; |
1da177e4c
|
2862 |
|
7b51755c3
|
2863 2864 2865 2866 |
p->flags |= PF_MEMALLOC; lockdep_set_current_reclaim_state(sc.gfp_mask); reclaim_state.reclaimed_slab = 0; p->reclaim_state = &reclaim_state; |
d6277db4a
|
2867 |
|
a09ed5e00
|
2868 |
nr_reclaimed = do_try_to_free_pages(zonelist, &sc, &shrink); |
d979677c4
|
2869 |
|
7b51755c3
|
2870 2871 2872 |
p->reclaim_state = NULL; lockdep_clear_current_reclaim_state(); p->flags &= ~PF_MEMALLOC; |
d6277db4a
|
2873 |
|
7b51755c3
|
2874 |
return nr_reclaimed; |
1da177e4c
|
2875 |
} |
c6f37f121
|
2876 |
#endif /* CONFIG_HIBERNATION */ |
1da177e4c
|
2877 |
|
1da177e4c
|
2878 2879 2880 2881 |
/* It's optimal to keep kswapds on the same CPUs as their memory, but not required for correctness. So if the last cpu in a node goes away, we get changed to run anywhere: as the first one comes back, restore their cpu bindings. */ |
9c7b216d2
|
2882 |
static int __devinit cpu_callback(struct notifier_block *nfb, |
69e05944a
|
2883 |
unsigned long action, void *hcpu) |
1da177e4c
|
2884 |
{ |
58c0a4a78
|
2885 |
int nid; |
1da177e4c
|
2886 |
|
8bb784428
|
2887 |
if (action == CPU_ONLINE || action == CPU_ONLINE_FROZEN) { |
58c0a4a78
|
2888 |
for_each_node_state(nid, N_HIGH_MEMORY) { |
c5f59f083
|
2889 |
pg_data_t *pgdat = NODE_DATA(nid); |
a70f73028
|
2890 2891 2892 |
const struct cpumask *mask; mask = cpumask_of_node(pgdat->node_id); |
c5f59f083
|
2893 |
|
3e5979453
|
2894 |
if (cpumask_any_and(cpu_online_mask, mask) < nr_cpu_ids) |
1da177e4c
|
2895 |
/* One of our CPUs online: restore mask */ |
c5f59f083
|
2896 |
set_cpus_allowed_ptr(pgdat->kswapd, mask); |
1da177e4c
|
2897 2898 2899 2900 |
} } return NOTIFY_OK; } |
1da177e4c
|
2901 |
|
3218ae14b
|
2902 2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 2917 2918 2919 2920 2921 2922 2923 |
/* * 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 */ BUG_ON(system_state == SYSTEM_BOOTING); printk("Failed to start kswapd on node %d ",nid); ret = -1; } return ret; } |
8fe23e057
|
2924 2925 2926 2927 2928 2929 2930 2931 2932 2933 |
/* * Called by memory hotplug when all memory in a node is offlined. */ void kswapd_stop(int nid) { struct task_struct *kswapd = NODE_DATA(nid)->kswapd; if (kswapd) kthread_stop(kswapd); } |
1da177e4c
|
2934 2935 |
static int __init kswapd_init(void) { |
3218ae14b
|
2936 |
int nid; |
69e05944a
|
2937 |
|
1da177e4c
|
2938 |
swap_setup(); |
9422ffba4
|
2939 |
for_each_node_state(nid, N_HIGH_MEMORY) |
3218ae14b
|
2940 |
kswapd_run(nid); |
1da177e4c
|
2941 2942 2943 2944 2945 |
hotcpu_notifier(cpu_callback, 0); return 0; } module_init(kswapd_init) |
9eeff2395
|
2946 2947 2948 2949 2950 2951 2952 |
#ifdef CONFIG_NUMA /* * Zone reclaim mode * * If non-zero call zone_reclaim when the number of free pages falls below * the watermarks. |
9eeff2395
|
2953 2954 |
*/ int zone_reclaim_mode __read_mostly; |
1b2ffb789
|
2955 |
#define RECLAIM_OFF 0 |
7d03431cf
|
2956 |
#define RECLAIM_ZONE (1<<0) /* Run shrink_inactive_list on the zone */ |
1b2ffb789
|
2957 2958 |
#define RECLAIM_WRITE (1<<1) /* Writeout pages during reclaim */ #define RECLAIM_SWAP (1<<2) /* Swap pages out during reclaim */ |
9eeff2395
|
2959 |
/* |
a92f71263
|
2960 2961 2962 2963 2964 |
* Priority for ZONE_RECLAIM. This determines the fraction of pages * of a node considered for each zone_reclaim. 4 scans 1/16th of * a zone. */ #define ZONE_RECLAIM_PRIORITY 4 |
9eeff2395
|
2965 |
/* |
9614634fe
|
2966 2967 2968 2969 2970 2971 |
* Percentage of pages in a zone that must be unmapped for zone_reclaim to * occur. */ int sysctl_min_unmapped_ratio = 1; /* |
0ff38490c
|
2972 2973 2974 2975 |
* 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; |
90afa5de6
|
2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012 3013 3014 3015 3016 |
static inline unsigned long zone_unmapped_file_pages(struct zone *zone) { unsigned long file_mapped = zone_page_state(zone, NR_FILE_MAPPED); unsigned long file_lru = zone_page_state(zone, NR_INACTIVE_FILE) + zone_page_state(zone, NR_ACTIVE_FILE); /* * 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 */ static long zone_pagecache_reclaimable(struct zone *zone) { long nr_pagecache_reclaimable; long delta = 0; /* * If RECLAIM_SWAP is set, then all file pages are considered * potentially reclaimable. Otherwise, we have to worry about * pages like swapcache and zone_unmapped_file_pages() provides * a better estimate */ if (zone_reclaim_mode & RECLAIM_SWAP) nr_pagecache_reclaimable = zone_page_state(zone, NR_FILE_PAGES); else nr_pagecache_reclaimable = zone_unmapped_file_pages(zone); /* If we can't clean pages, remove dirty pages from consideration */ if (!(zone_reclaim_mode & RECLAIM_WRITE)) delta += zone_page_state(zone, NR_FILE_DIRTY); /* Watch for any possible underflows due to delta */ if (unlikely(delta > nr_pagecache_reclaimable)) delta = nr_pagecache_reclaimable; return nr_pagecache_reclaimable - delta; } |
0ff38490c
|
3017 |
/* |
9eeff2395
|
3018 3019 |
* Try to free up some pages from this zone through reclaim. */ |
179e96395
|
3020 |
static int __zone_reclaim(struct zone *zone, gfp_t gfp_mask, unsigned int order) |
9eeff2395
|
3021 |
{ |
7fb2d46d3
|
3022 |
/* Minimum pages needed in order to stay on node */ |
69e05944a
|
3023 |
const unsigned long nr_pages = 1 << order; |
9eeff2395
|
3024 3025 |
struct task_struct *p = current; struct reclaim_state reclaim_state; |
8695949a1
|
3026 |
int priority; |
179e96395
|
3027 3028 |
struct scan_control sc = { .may_writepage = !!(zone_reclaim_mode & RECLAIM_WRITE), |
a6dc60f89
|
3029 |
.may_unmap = !!(zone_reclaim_mode & RECLAIM_SWAP), |
2e2e42598
|
3030 |
.may_swap = 1, |
22fba3354
|
3031 3032 |
.nr_to_reclaim = max_t(unsigned long, nr_pages, SWAP_CLUSTER_MAX), |
179e96395
|
3033 |
.gfp_mask = gfp_mask, |
bd2f6199c
|
3034 |
.order = order, |
179e96395
|
3035 |
}; |
a09ed5e00
|
3036 3037 3038 |
struct shrink_control shrink = { .gfp_mask = sc.gfp_mask, }; |
157480489
|
3039 |
unsigned long nr_slab_pages0, nr_slab_pages1; |
9eeff2395
|
3040 |
|
9eeff2395
|
3041 |
cond_resched(); |
d4f7796e9
|
3042 3043 3044 3045 3046 3047 |
/* * We need to be able to allocate from the reserves for RECLAIM_SWAP * and we also need to be able to write out pages for RECLAIM_WRITE * and RECLAIM_SWAP. */ p->flags |= PF_MEMALLOC | PF_SWAPWRITE; |
76ca542d8
|
3048 |
lockdep_set_current_reclaim_state(gfp_mask); |
9eeff2395
|
3049 3050 |
reclaim_state.reclaimed_slab = 0; p->reclaim_state = &reclaim_state; |
c84db23c6
|
3051 |
|
90afa5de6
|
3052 |
if (zone_pagecache_reclaimable(zone) > zone->min_unmapped_pages) { |
0ff38490c
|
3053 3054 3055 3056 3057 3058 |
/* * Free memory by calling shrink zone with increasing * priorities until we have enough memory freed. */ priority = ZONE_RECLAIM_PRIORITY; do { |
a79311c14
|
3059 |
shrink_zone(priority, zone, &sc); |
0ff38490c
|
3060 |
priority--; |
a79311c14
|
3061 |
} while (priority >= 0 && sc.nr_reclaimed < nr_pages); |
0ff38490c
|
3062 |
} |
c84db23c6
|
3063 |
|
157480489
|
3064 3065 |
nr_slab_pages0 = zone_page_state(zone, NR_SLAB_RECLAIMABLE); if (nr_slab_pages0 > zone->min_slab_pages) { |
2a16e3f4b
|
3066 |
/* |
7fb2d46d3
|
3067 |
* shrink_slab() does not currently allow us to determine how |
0ff38490c
|
3068 3069 3070 3071 |
* many pages were freed in this zone. So we take the current * number of slab pages and shake the slab until it is reduced * by the same nr_pages that we used for reclaiming unmapped * pages. |
2a16e3f4b
|
3072 |
* |
0ff38490c
|
3073 3074 |
* Note that shrink_slab will free memory on all zones and may * take a long time. |
2a16e3f4b
|
3075 |
*/ |
4dc4b3d97
|
3076 3077 3078 3079 |
for (;;) { unsigned long lru_pages = zone_reclaimable_pages(zone); /* No reclaimable slab or very low memory pressure */ |
1495f230f
|
3080 |
if (!shrink_slab(&shrink, sc.nr_scanned, lru_pages)) |
4dc4b3d97
|
3081 3082 3083 3084 3085 3086 3087 3088 |
break; /* Freed enough memory */ nr_slab_pages1 = zone_page_state(zone, NR_SLAB_RECLAIMABLE); if (nr_slab_pages1 + nr_pages <= nr_slab_pages0) break; } |
83e33a471
|
3089 3090 3091 3092 3093 |
/* * Update nr_reclaimed by the number of slab pages we * reclaimed from this zone. */ |
157480489
|
3094 3095 3096 |
nr_slab_pages1 = zone_page_state(zone, NR_SLAB_RECLAIMABLE); if (nr_slab_pages1 < nr_slab_pages0) sc.nr_reclaimed += nr_slab_pages0 - nr_slab_pages1; |
2a16e3f4b
|
3097 |
} |
9eeff2395
|
3098 |
p->reclaim_state = NULL; |
d4f7796e9
|
3099 |
current->flags &= ~(PF_MEMALLOC | PF_SWAPWRITE); |
76ca542d8
|
3100 |
lockdep_clear_current_reclaim_state(); |
a79311c14
|
3101 |
return sc.nr_reclaimed >= nr_pages; |
9eeff2395
|
3102 |
} |
179e96395
|
3103 3104 3105 |
int zone_reclaim(struct zone *zone, gfp_t gfp_mask, unsigned int order) { |
179e96395
|
3106 |
int node_id; |
d773ed6b8
|
3107 |
int ret; |
179e96395
|
3108 3109 |
/* |
0ff38490c
|
3110 3111 |
* Zone reclaim reclaims unmapped file backed pages and * slab pages if we are over the defined limits. |
34aa1330f
|
3112 |
* |
9614634fe
|
3113 3114 3115 3116 3117 |
* A small portion of unmapped file backed pages is needed for * file I/O otherwise pages read by file I/O will be immediately * thrown out if the zone is overallocated. So we do not reclaim * if less than a specified percentage of the zone is used by * unmapped file backed pages. |
179e96395
|
3118 |
*/ |
90afa5de6
|
3119 3120 |
if (zone_pagecache_reclaimable(zone) <= zone->min_unmapped_pages && zone_page_state(zone, NR_SLAB_RECLAIMABLE) <= zone->min_slab_pages) |
fa5e084e4
|
3121 |
return ZONE_RECLAIM_FULL; |
179e96395
|
3122 |
|
93e4a89a8
|
3123 |
if (zone->all_unreclaimable) |
fa5e084e4
|
3124 |
return ZONE_RECLAIM_FULL; |
d773ed6b8
|
3125 |
|
179e96395
|
3126 |
/* |
d773ed6b8
|
3127 |
* Do not scan if the allocation should not be delayed. |
179e96395
|
3128 |
*/ |
d773ed6b8
|
3129 |
if (!(gfp_mask & __GFP_WAIT) || (current->flags & PF_MEMALLOC)) |
fa5e084e4
|
3130 |
return ZONE_RECLAIM_NOSCAN; |
179e96395
|
3131 3132 3133 3134 3135 3136 3137 |
/* * Only run zone reclaim on the local zone or on zones that do not * have associated processors. This will favor the local processor * over remote processors and spread off node memory allocations * as wide as possible. */ |
89fa30242
|
3138 |
node_id = zone_to_nid(zone); |
37c0708db
|
3139 |
if (node_state(node_id, N_CPU) && node_id != numa_node_id()) |
fa5e084e4
|
3140 |
return ZONE_RECLAIM_NOSCAN; |
d773ed6b8
|
3141 3142 |
if (zone_test_and_set_flag(zone, ZONE_RECLAIM_LOCKED)) |
fa5e084e4
|
3143 |
return ZONE_RECLAIM_NOSCAN; |
d773ed6b8
|
3144 3145 |
ret = __zone_reclaim(zone, gfp_mask, order); zone_clear_flag(zone, ZONE_RECLAIM_LOCKED); |
24cf72518
|
3146 3147 |
if (!ret) count_vm_event(PGSCAN_ZONE_RECLAIM_FAILED); |
d773ed6b8
|
3148 |
return ret; |
179e96395
|
3149 |
} |
9eeff2395
|
3150 |
#endif |
894bc3104
|
3151 |
|
894bc3104
|
3152 3153 3154 3155 3156 3157 |
/* * page_evictable - test whether a page is evictable * @page: the page to test * @vma: the VMA in which the page is or will be mapped, may be NULL * * Test whether page is evictable--i.e., should be placed on active/inactive |
b291f0003
|
3158 3159 |
* lists vs unevictable list. The vma argument is !NULL when called from the * fault path to determine how to instantate a new page. |
894bc3104
|
3160 3161 |
* * Reasons page might not be evictable: |
ba9ddf493
|
3162 |
* (1) page's mapping marked unevictable |
b291f0003
|
3163 |
* (2) page is part of an mlocked VMA |
ba9ddf493
|
3164 |
* |
894bc3104
|
3165 3166 3167 |
*/ int page_evictable(struct page *page, struct vm_area_struct *vma) { |
ba9ddf493
|
3168 3169 |
if (mapping_unevictable(page_mapping(page))) return 0; |
096a7cf44
|
3170 |
if (PageMlocked(page) || (vma && mlocked_vma_newpage(vma, page))) |
b291f0003
|
3171 |
return 0; |
894bc3104
|
3172 3173 3174 |
return 1; } |
89e004ea5
|
3175 |
|
85046579b
|
3176 |
#ifdef CONFIG_SHMEM |
89e004ea5
|
3177 |
/** |
245132643
|
3178 3179 3180 |
* check_move_unevictable_pages - check pages for evictability and move to appropriate zone lru list * @pages: array of pages to check * @nr_pages: number of pages to check |
89e004ea5
|
3181 |
* |
245132643
|
3182 |
* Checks pages for evictability and moves them to the appropriate lru list. |
85046579b
|
3183 3184 |
* * This function is only used for SysV IPC SHM_UNLOCK. |
89e004ea5
|
3185 |
*/ |
245132643
|
3186 |
void check_move_unevictable_pages(struct page **pages, int nr_pages) |
89e004ea5
|
3187 |
{ |
925b7673c
|
3188 |
struct lruvec *lruvec; |
245132643
|
3189 3190 3191 3192 |
struct zone *zone = NULL; int pgscanned = 0; int pgrescued = 0; int i; |
89e004ea5
|
3193 |
|
245132643
|
3194 3195 3196 |
for (i = 0; i < nr_pages; i++) { struct page *page = pages[i]; struct zone *pagezone; |
89e004ea5
|
3197 |
|
245132643
|
3198 3199 3200 3201 3202 3203 3204 3205 |
pgscanned++; pagezone = page_zone(page); if (pagezone != zone) { if (zone) spin_unlock_irq(&zone->lru_lock); zone = pagezone; spin_lock_irq(&zone->lru_lock); } |
89e004ea5
|
3206 |
|
245132643
|
3207 3208 |
if (!PageLRU(page) || !PageUnevictable(page)) continue; |
89e004ea5
|
3209 |
|
245132643
|
3210 3211 3212 3213 3214 3215 3216 3217 3218 3219 3220 |
if (page_evictable(page, NULL)) { enum lru_list lru = page_lru_base_type(page); VM_BUG_ON(PageActive(page)); ClearPageUnevictable(page); __dec_zone_state(zone, NR_UNEVICTABLE); lruvec = mem_cgroup_lru_move_lists(zone, page, LRU_UNEVICTABLE, lru); list_move(&page->lru, &lruvec->lists[lru]); __inc_zone_state(zone, NR_INACTIVE_ANON + lru); pgrescued++; |
89e004ea5
|
3221 |
} |
245132643
|
3222 |
} |
89e004ea5
|
3223 |
|
245132643
|
3224 3225 3226 3227 |
if (zone) { __count_vm_events(UNEVICTABLE_PGRESCUED, pgrescued); __count_vm_events(UNEVICTABLE_PGSCANNED, pgscanned); spin_unlock_irq(&zone->lru_lock); |
89e004ea5
|
3228 |
} |
89e004ea5
|
3229 |
} |
85046579b
|
3230 |
#endif /* CONFIG_SHMEM */ |
af936a160
|
3231 |
|
264e56d82
|
3232 |
static void warn_scan_unevictable_pages(void) |
af936a160
|
3233 |
{ |
264e56d82
|
3234 |
printk_once(KERN_WARNING |
25bd91bd2
|
3235 |
"%s: The scan_unevictable_pages sysctl/node-interface has been " |
264e56d82
|
3236 |
"disabled for lack of a legitimate use case. If you have " |
25bd91bd2
|
3237 3238 3239 |
"one, please send an email to linux-mm@kvack.org. ", current->comm); |
af936a160
|
3240 3241 3242 3243 3244 3245 3246 3247 3248 |
} /* * scan_unevictable_pages [vm] sysctl handler. On demand re-scan of * all nodes' unevictable lists for evictable pages */ unsigned long scan_unevictable_pages; int scan_unevictable_handler(struct ctl_table *table, int write, |
8d65af789
|
3249 |
void __user *buffer, |
af936a160
|
3250 3251 |
size_t *length, loff_t *ppos) { |
264e56d82
|
3252 |
warn_scan_unevictable_pages(); |
8d65af789
|
3253 |
proc_doulongvec_minmax(table, write, buffer, length, ppos); |
af936a160
|
3254 3255 3256 |
scan_unevictable_pages = 0; return 0; } |
e4455abb5
|
3257 |
#ifdef CONFIG_NUMA |
af936a160
|
3258 3259 3260 3261 |
/* * per node 'scan_unevictable_pages' attribute. On demand re-scan of * a specified node's per zone unevictable lists for evictable pages. */ |
10fbcf4c6
|
3262 3263 |
static ssize_t read_scan_unevictable_node(struct device *dev, struct device_attribute *attr, |
af936a160
|
3264 3265 |
char *buf) { |
264e56d82
|
3266 |
warn_scan_unevictable_pages(); |
af936a160
|
3267 3268 3269 |
return sprintf(buf, "0 "); /* always zero; should fit... */ } |
10fbcf4c6
|
3270 3271 |
static ssize_t write_scan_unevictable_node(struct device *dev, struct device_attribute *attr, |
af936a160
|
3272 3273 |
const char *buf, size_t count) { |
264e56d82
|
3274 |
warn_scan_unevictable_pages(); |
af936a160
|
3275 3276 |
return 1; } |
10fbcf4c6
|
3277 |
static DEVICE_ATTR(scan_unevictable_pages, S_IRUGO | S_IWUSR, |
af936a160
|
3278 3279 3280 3281 3282 |
read_scan_unevictable_node, write_scan_unevictable_node); int scan_unevictable_register_node(struct node *node) { |
10fbcf4c6
|
3283 |
return device_create_file(&node->dev, &dev_attr_scan_unevictable_pages); |
af936a160
|
3284 3285 3286 3287 |
} void scan_unevictable_unregister_node(struct node *node) { |
10fbcf4c6
|
3288 |
device_remove_file(&node->dev, &dev_attr_scan_unevictable_pages); |
af936a160
|
3289 |
} |
e4455abb5
|
3290 |
#endif |