Blame view
mm/page-writeback.c
83.7 KB
1da177e4c
|
1 |
/* |
f30c22695
|
2 |
* mm/page-writeback.c |
1da177e4c
|
3 4 |
* * Copyright (C) 2002, Linus Torvalds. |
90eec103b
|
5 |
* Copyright (C) 2007 Red Hat, Inc., Peter Zijlstra |
1da177e4c
|
6 7 8 9 |
* * Contains functions related to writing back dirty pages at the * address_space level. * |
e1f8e8744
|
10 |
* 10Apr2002 Andrew Morton |
1da177e4c
|
11 12 13 14 |
* Initial version */ #include <linux/kernel.h> |
b95f1b31b
|
15 |
#include <linux/export.h> |
1da177e4c
|
16 17 18 19 20 21 22 23 24 |
#include <linux/spinlock.h> #include <linux/fs.h> #include <linux/mm.h> #include <linux/swap.h> #include <linux/slab.h> #include <linux/pagemap.h> #include <linux/writeback.h> #include <linux/init.h> #include <linux/backing-dev.h> |
55e829af0
|
25 |
#include <linux/task_io_accounting_ops.h> |
1da177e4c
|
26 27 |
#include <linux/blkdev.h> #include <linux/mpage.h> |
d08b3851d
|
28 |
#include <linux/rmap.h> |
1da177e4c
|
29 30 31 32 33 34 |
#include <linux/percpu.h> #include <linux/notifier.h> #include <linux/smp.h> #include <linux/sysctl.h> #include <linux/cpu.h> #include <linux/syscalls.h> |
ff01bb483
|
35 |
#include <linux/buffer_head.h> /* __set_page_dirty_buffers */ |
811d736f9
|
36 |
#include <linux/pagevec.h> |
eb608e3a3
|
37 |
#include <linux/timer.h> |
8bd75c77b
|
38 |
#include <linux/sched/rt.h> |
6e543d578
|
39 |
#include <linux/mm_inline.h> |
028c2dd18
|
40 |
#include <trace/events/writeback.h> |
1da177e4c
|
41 |
|
6e543d578
|
42 |
#include "internal.h" |
1da177e4c
|
43 |
/* |
ffd1f609a
|
44 45 46 47 48 |
* Sleep at most 200ms at a time in balance_dirty_pages(). */ #define MAX_PAUSE max(HZ/5, 1) /* |
5b9b35743
|
49 50 51 52 53 54 |
* Try to keep balance_dirty_pages() call intervals higher than this many pages * by raising pause time to max_pause when falls below it. */ #define DIRTY_POLL_THRESH (128 >> (PAGE_SHIFT - 10)) /* |
e98be2d59
|
55 56 57 |
* Estimate write bandwidth at 200ms intervals. */ #define BANDWIDTH_INTERVAL max(HZ/5, 1) |
6c14ae1e9
|
58 |
#define RATELIMIT_CALC_SHIFT 10 |
e98be2d59
|
59 |
/* |
1da177e4c
|
60 61 62 63 |
* After a CPU has dirtied this many pages, balance_dirty_pages_ratelimited * will look to see if it needs to force writeback or throttling. */ static long ratelimit_pages = 32; |
1da177e4c
|
64 65 66 |
/* The following parameters are exported via /proc/sys/vm */ /* |
5b0830cb9
|
67 |
* Start background writeback (via writeback threads) at this percentage |
1da177e4c
|
68 |
*/ |
1b5e62b42
|
69 |
int dirty_background_ratio = 10; |
1da177e4c
|
70 71 |
/* |
2da02997e
|
72 73 74 75 76 77 |
* dirty_background_bytes starts at 0 (disabled) so that it is a function of * dirty_background_ratio * the amount of dirtyable memory */ unsigned long dirty_background_bytes; /* |
195cf453d
|
78 79 80 81 82 83 |
* free highmem will not be subtracted from the total free memory * for calculating free ratios if vm_highmem_is_dirtyable is true */ int vm_highmem_is_dirtyable; /* |
1da177e4c
|
84 85 |
* The generator of dirty data starts writeback at this percentage */ |
1b5e62b42
|
86 |
int vm_dirty_ratio = 20; |
1da177e4c
|
87 88 |
/* |
2da02997e
|
89 90 91 92 93 94 |
* vm_dirty_bytes starts at 0 (disabled) so that it is a function of * vm_dirty_ratio * the amount of dirtyable memory */ unsigned long vm_dirty_bytes; /* |
704503d83
|
95 |
* The interval between `kupdate'-style writebacks |
1da177e4c
|
96 |
*/ |
22ef37eed
|
97 |
unsigned int dirty_writeback_interval = 5 * 100; /* centiseconds */ |
1da177e4c
|
98 |
|
91913a294
|
99 |
EXPORT_SYMBOL_GPL(dirty_writeback_interval); |
1da177e4c
|
100 |
/* |
704503d83
|
101 |
* The longest time for which data is allowed to remain dirty |
1da177e4c
|
102 |
*/ |
22ef37eed
|
103 |
unsigned int dirty_expire_interval = 30 * 100; /* centiseconds */ |
1da177e4c
|
104 105 106 107 108 109 110 |
/* * Flag that makes the machine dump writes/reads and block dirtyings. */ int block_dump; /* |
ed5b43f15
|
111 112 |
* Flag that puts the machine in "laptop mode". Doubles as a timeout in jiffies: * a full sync is triggered after this time elapses without any disk activity. |
1da177e4c
|
113 114 115 116 117 118 |
*/ int laptop_mode; EXPORT_SYMBOL(laptop_mode); /* End of sysctl-exported parameters */ |
dcc25ae76
|
119 |
struct wb_domain global_wb_domain; |
1da177e4c
|
120 |
|
2bc00aef0
|
121 122 |
/* consolidated parameters for balance_dirty_pages() and its subroutines */ struct dirty_throttle_control { |
e9f07dfd7
|
123 124 |
#ifdef CONFIG_CGROUP_WRITEBACK struct wb_domain *dom; |
9fc3a43e1
|
125 |
struct dirty_throttle_control *gdtc; /* only set in memcg dtc's */ |
e9f07dfd7
|
126 |
#endif |
2bc00aef0
|
127 |
struct bdi_writeback *wb; |
e9770b348
|
128 |
struct fprop_local_percpu *wb_completions; |
eb608e3a3
|
129 |
|
9fc3a43e1
|
130 |
unsigned long avail; /* dirtyable */ |
2bc00aef0
|
131 132 133 134 135 136 |
unsigned long dirty; /* file_dirty + write + nfs */ unsigned long thresh; /* dirty threshold */ unsigned long bg_thresh; /* dirty background threshold */ unsigned long wb_dirty; /* per-wb counterparts */ unsigned long wb_thresh; |
970fb01ad
|
137 |
unsigned long wb_bg_thresh; |
daddfa3cb
|
138 139 |
unsigned long pos_ratio; |
2bc00aef0
|
140 |
}; |
eb608e3a3
|
141 142 143 144 145 146 |
/* * Length of period for aging writeout fractions of bdis. This is an * arbitrarily chosen number. The longer the period, the slower fractions will * reflect changes in current writeout rate. */ #define VM_COMPLETIONS_PERIOD_LEN (3*HZ) |
04fbfdc14
|
147 |
|
693108a8a
|
148 |
#ifdef CONFIG_CGROUP_WRITEBACK |
d60d1bddd
|
149 150 151 |
#define GDTC_INIT(__wb) .wb = (__wb), \ .dom = &global_wb_domain, \ .wb_completions = &(__wb)->completions |
9fc3a43e1
|
152 |
#define GDTC_INIT_NO_WB .dom = &global_wb_domain |
d60d1bddd
|
153 154 155 156 157 |
#define MDTC_INIT(__wb, __gdtc) .wb = (__wb), \ .dom = mem_cgroup_wb_domain(__wb), \ .wb_completions = &(__wb)->memcg_completions, \ .gdtc = __gdtc |
c2aa723a6
|
158 159 160 161 162 |
static bool mdtc_valid(struct dirty_throttle_control *dtc) { return dtc->dom; } |
e9f07dfd7
|
163 164 165 166 167 |
static struct wb_domain *dtc_dom(struct dirty_throttle_control *dtc) { return dtc->dom; } |
9fc3a43e1
|
168 169 170 171 |
static struct dirty_throttle_control *mdtc_gdtc(struct dirty_throttle_control *mdtc) { return mdtc->gdtc; } |
841710aa6
|
172 173 174 175 |
static struct fprop_local_percpu *wb_memcg_completions(struct bdi_writeback *wb) { return &wb->memcg_completions; } |
693108a8a
|
176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 |
static void wb_min_max_ratio(struct bdi_writeback *wb, unsigned long *minp, unsigned long *maxp) { unsigned long this_bw = wb->avg_write_bandwidth; unsigned long tot_bw = atomic_long_read(&wb->bdi->tot_write_bandwidth); unsigned long long min = wb->bdi->min_ratio; unsigned long long max = wb->bdi->max_ratio; /* * @wb may already be clean by the time control reaches here and * the total may not include its bw. */ if (this_bw < tot_bw) { if (min) { min *= this_bw; do_div(min, tot_bw); } if (max < 100) { max *= this_bw; do_div(max, tot_bw); } } *minp = min; *maxp = max; } #else /* CONFIG_CGROUP_WRITEBACK */ |
d60d1bddd
|
204 205 |
#define GDTC_INIT(__wb) .wb = (__wb), \ .wb_completions = &(__wb)->completions |
9fc3a43e1
|
206 |
#define GDTC_INIT_NO_WB |
c2aa723a6
|
207 208 209 210 211 212 |
#define MDTC_INIT(__wb, __gdtc) static bool mdtc_valid(struct dirty_throttle_control *dtc) { return false; } |
e9f07dfd7
|
213 214 215 216 217 |
static struct wb_domain *dtc_dom(struct dirty_throttle_control *dtc) { return &global_wb_domain; } |
9fc3a43e1
|
218 219 220 221 |
static struct dirty_throttle_control *mdtc_gdtc(struct dirty_throttle_control *mdtc) { return NULL; } |
841710aa6
|
222 223 224 225 |
static struct fprop_local_percpu *wb_memcg_completions(struct bdi_writeback *wb) { return NULL; } |
693108a8a
|
226 227 228 229 230 231 232 233 |
static void wb_min_max_ratio(struct bdi_writeback *wb, unsigned long *minp, unsigned long *maxp) { *minp = wb->bdi->min_ratio; *maxp = wb->bdi->max_ratio; } #endif /* CONFIG_CGROUP_WRITEBACK */ |
04fbfdc14
|
234 |
/* |
a756cf590
|
235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 |
* In a memory zone, there is a certain amount of pages we consider * available for the page cache, which is essentially the number of * free and reclaimable pages, minus some zone reserves to protect * lowmem and the ability to uphold the zone's watermarks without * requiring writeback. * * This number of dirtyable pages is the base value of which the * user-configurable dirty ratio is the effictive number of pages that * are allowed to be actually dirtied. Per individual zone, or * globally by using the sum of dirtyable pages over all zones. * * Because the user is allowed to specify the dirty limit globally as * absolute number of bytes, calculating the per-zone dirty limit can * require translating the configured limit into a percentage of * global dirtyable memory first. */ |
a804552b9
|
251 252 253 254 255 256 257 258 259 260 261 262 |
/** * zone_dirtyable_memory - number of dirtyable pages in a zone * @zone: the zone * * Returns the zone's number of pages potentially available for dirty * page cache. This is the base value for the per-zone dirty limits. */ static unsigned long zone_dirtyable_memory(struct zone *zone) { unsigned long nr_pages; nr_pages = zone_page_state(zone, NR_FREE_PAGES); |
a8d014373
|
263 264 265 266 267 268 |
/* * Pages reserved for the kernel should not be considered * dirtyable, to prevent a situation where reclaim has to * clean pages in order to balance the zones. */ nr_pages -= min(nr_pages, zone->totalreserve_pages); |
a804552b9
|
269 |
|
a1c3bfb2f
|
270 271 |
nr_pages += zone_page_state(zone, NR_INACTIVE_FILE); nr_pages += zone_page_state(zone, NR_ACTIVE_FILE); |
a804552b9
|
272 273 274 |
return nr_pages; } |
1edf22348
|
275 276 277 278 279 |
static unsigned long highmem_dirtyable_memory(unsigned long total) { #ifdef CONFIG_HIGHMEM int node; unsigned long x = 0; |
09b4ab3c4
|
280 |
int i; |
1edf22348
|
281 282 |
for_each_node_state(node, N_HIGH_MEMORY) { |
09b4ab3c4
|
283 284 |
for (i = 0; i < MAX_NR_ZONES; i++) { struct zone *z = &NODE_DATA(node)->node_zones[i]; |
1edf22348
|
285 |
|
09b4ab3c4
|
286 287 288 |
if (is_highmem(z)) x += zone_dirtyable_memory(z); } |
1edf22348
|
289 290 |
} /* |
c8b74c2f6
|
291 292 293 294 295 296 297 298 299 300 301 302 |
* Unreclaimable memory (kernel memory or anonymous memory * without swap) can bring down the dirtyable pages below * the zone's dirty balance reserve and the above calculation * will underflow. However we still want to add in nodes * which are below threshold (negative values) to get a more * accurate calculation but make sure that the total never * underflows. */ if ((long)x < 0) x = 0; /* |
1edf22348
|
303 304 305 306 307 308 309 310 311 312 313 314 |
* Make sure that the number of highmem pages is never larger * than the number of the total dirtyable memory. This can only * occur in very strange VM situations but we want to make sure * that this does not occur. */ return min(x, total); #else return 0; #endif } /** |
ccafa2879
|
315 |
* global_dirtyable_memory - number of globally dirtyable pages |
1edf22348
|
316 |
* |
ccafa2879
|
317 318 |
* Returns the global number of pages potentially available for dirty * page cache. This is the base value for the global dirty limits. |
1edf22348
|
319 |
*/ |
18cf8cf8b
|
320 |
static unsigned long global_dirtyable_memory(void) |
1edf22348
|
321 322 |
{ unsigned long x; |
a804552b9
|
323 |
x = global_page_state(NR_FREE_PAGES); |
a8d014373
|
324 325 326 327 328 329 |
/* * Pages reserved for the kernel should not be considered * dirtyable, to prevent a situation where reclaim has to * clean pages in order to balance the zones. */ x -= min(x, totalreserve_pages); |
1edf22348
|
330 |
|
a1c3bfb2f
|
331 332 |
x += global_page_state(NR_INACTIVE_FILE); x += global_page_state(NR_ACTIVE_FILE); |
a804552b9
|
333 |
|
1edf22348
|
334 335 336 337 338 |
if (!vm_highmem_is_dirtyable) x -= highmem_dirtyable_memory(x); return x + 1; /* Ensure that we never return 0 */ } |
9fc3a43e1
|
339 340 341 |
/** * domain_dirty_limits - calculate thresh and bg_thresh for a wb_domain * @dtc: dirty_throttle_control of interest |
ccafa2879
|
342 |
* |
9fc3a43e1
|
343 344 345 346 |
* Calculate @dtc->thresh and ->bg_thresh considering * vm_dirty_{bytes|ratio} and dirty_background_{bytes|ratio}. The caller * must ensure that @dtc->avail is set before calling this function. The * dirty limits will be lifted by 1/4 for PF_LESS_THROTTLE (ie. nfsd) and |
ccafa2879
|
347 348 |
* real-time tasks. */ |
9fc3a43e1
|
349 |
static void domain_dirty_limits(struct dirty_throttle_control *dtc) |
ccafa2879
|
350 |
{ |
9fc3a43e1
|
351 352 353 354 355 356 357 358 |
const unsigned long available_memory = dtc->avail; struct dirty_throttle_control *gdtc = mdtc_gdtc(dtc); unsigned long bytes = vm_dirty_bytes; unsigned long bg_bytes = dirty_background_bytes; unsigned long ratio = vm_dirty_ratio; unsigned long bg_ratio = dirty_background_ratio; unsigned long thresh; unsigned long bg_thresh; |
ccafa2879
|
359 |
struct task_struct *tsk; |
9fc3a43e1
|
360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 |
/* gdtc is !NULL iff @dtc is for memcg domain */ if (gdtc) { unsigned long global_avail = gdtc->avail; /* * The byte settings can't be applied directly to memcg * domains. Convert them to ratios by scaling against * globally available memory. */ if (bytes) ratio = min(DIV_ROUND_UP(bytes, PAGE_SIZE) * 100 / global_avail, 100UL); if (bg_bytes) bg_ratio = min(DIV_ROUND_UP(bg_bytes, PAGE_SIZE) * 100 / global_avail, 100UL); bytes = bg_bytes = 0; } if (bytes) thresh = DIV_ROUND_UP(bytes, PAGE_SIZE); |
ccafa2879
|
380 |
else |
9fc3a43e1
|
381 |
thresh = (ratio * available_memory) / 100; |
ccafa2879
|
382 |
|
9fc3a43e1
|
383 384 |
if (bg_bytes) bg_thresh = DIV_ROUND_UP(bg_bytes, PAGE_SIZE); |
ccafa2879
|
385 |
else |
9fc3a43e1
|
386 |
bg_thresh = (bg_ratio * available_memory) / 100; |
ccafa2879
|
387 |
|
9fc3a43e1
|
388 389 |
if (bg_thresh >= thresh) bg_thresh = thresh / 2; |
ccafa2879
|
390 391 |
tsk = current; if (tsk->flags & PF_LESS_THROTTLE || rt_task(tsk)) { |
a53eaff8c
|
392 393 |
bg_thresh += bg_thresh / 4 + global_wb_domain.dirty_limit / 32; thresh += thresh / 4 + global_wb_domain.dirty_limit / 32; |
ccafa2879
|
394 |
} |
9fc3a43e1
|
395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 |
dtc->thresh = thresh; dtc->bg_thresh = bg_thresh; /* we should eventually report the domain in the TP */ if (!gdtc) trace_global_dirty_state(bg_thresh, thresh); } /** * global_dirty_limits - background-writeback and dirty-throttling thresholds * @pbackground: out parameter for bg_thresh * @pdirty: out parameter for thresh * * Calculate bg_thresh and thresh for global_wb_domain. See * domain_dirty_limits() for details. */ void global_dirty_limits(unsigned long *pbackground, unsigned long *pdirty) { struct dirty_throttle_control gdtc = { GDTC_INIT_NO_WB }; gdtc.avail = global_dirtyable_memory(); domain_dirty_limits(&gdtc); *pbackground = gdtc.bg_thresh; *pdirty = gdtc.thresh; |
ccafa2879
|
420 |
} |
a756cf590
|
421 |
/** |
a756cf590
|
422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 |
* zone_dirty_limit - maximum number of dirty pages allowed in a zone * @zone: the zone * * Returns the maximum number of dirty pages allowed in a zone, based * on the zone's dirtyable memory. */ static unsigned long zone_dirty_limit(struct zone *zone) { unsigned long zone_memory = zone_dirtyable_memory(zone); struct task_struct *tsk = current; unsigned long dirty; if (vm_dirty_bytes) dirty = DIV_ROUND_UP(vm_dirty_bytes, PAGE_SIZE) * zone_memory / global_dirtyable_memory(); else dirty = vm_dirty_ratio * zone_memory / 100; if (tsk->flags & PF_LESS_THROTTLE || rt_task(tsk)) dirty += dirty / 4; return dirty; } /** * zone_dirty_ok - tells whether a zone is within its dirty limits * @zone: the zone to check * * Returns %true when the dirty pages in @zone are within the zone's * dirty limit, %false if the limit is exceeded. */ bool zone_dirty_ok(struct zone *zone) { unsigned long limit = zone_dirty_limit(zone); return zone_page_state(zone, NR_FILE_DIRTY) + zone_page_state(zone, NR_UNSTABLE_NFS) + zone_page_state(zone, NR_WRITEBACK) <= limit; } |
2da02997e
|
461 |
int dirty_background_ratio_handler(struct ctl_table *table, int write, |
8d65af789
|
462 |
void __user *buffer, size_t *lenp, |
2da02997e
|
463 464 465 |
loff_t *ppos) { int ret; |
8d65af789
|
466 |
ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos); |
2da02997e
|
467 468 469 470 471 472 |
if (ret == 0 && write) dirty_background_bytes = 0; return ret; } int dirty_background_bytes_handler(struct ctl_table *table, int write, |
8d65af789
|
473 |
void __user *buffer, size_t *lenp, |
2da02997e
|
474 475 476 |
loff_t *ppos) { int ret; |
8d65af789
|
477 |
ret = proc_doulongvec_minmax(table, write, buffer, lenp, ppos); |
2da02997e
|
478 479 480 481 |
if (ret == 0 && write) dirty_background_ratio = 0; return ret; } |
04fbfdc14
|
482 |
int dirty_ratio_handler(struct ctl_table *table, int write, |
8d65af789
|
483 |
void __user *buffer, size_t *lenp, |
04fbfdc14
|
484 485 486 |
loff_t *ppos) { int old_ratio = vm_dirty_ratio; |
2da02997e
|
487 |
int ret; |
8d65af789
|
488 |
ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos); |
04fbfdc14
|
489 |
if (ret == 0 && write && vm_dirty_ratio != old_ratio) { |
eb608e3a3
|
490 |
writeback_set_ratelimit(); |
2da02997e
|
491 492 493 494 |
vm_dirty_bytes = 0; } return ret; } |
2da02997e
|
495 |
int dirty_bytes_handler(struct ctl_table *table, int write, |
8d65af789
|
496 |
void __user *buffer, size_t *lenp, |
2da02997e
|
497 498 |
loff_t *ppos) { |
fc3501d41
|
499 |
unsigned long old_bytes = vm_dirty_bytes; |
2da02997e
|
500 |
int ret; |
8d65af789
|
501 |
ret = proc_doulongvec_minmax(table, write, buffer, lenp, ppos); |
2da02997e
|
502 |
if (ret == 0 && write && vm_dirty_bytes != old_bytes) { |
eb608e3a3
|
503 |
writeback_set_ratelimit(); |
2da02997e
|
504 |
vm_dirty_ratio = 0; |
04fbfdc14
|
505 506 507 |
} return ret; } |
eb608e3a3
|
508 509 510 511 512 513 514 515 |
static unsigned long wp_next_time(unsigned long cur_time) { cur_time += VM_COMPLETIONS_PERIOD_LEN; /* 0 has a special meaning... */ if (!cur_time) return 1; return cur_time; } |
c7981433e
|
516 517 518 |
static void wb_domain_writeout_inc(struct wb_domain *dom, struct fprop_local_percpu *completions, unsigned int max_prop_frac) |
04fbfdc14
|
519 |
{ |
c7981433e
|
520 521 |
__fprop_inc_percpu_max(&dom->completions, completions, max_prop_frac); |
eb608e3a3
|
522 |
/* First event after period switching was turned off? */ |
380c27ca3
|
523 |
if (!unlikely(dom->period_time)) { |
eb608e3a3
|
524 525 526 527 528 529 |
/* * We can race with other __bdi_writeout_inc calls here but * it does not cause any harm since the resulting time when * timer will fire and what is in writeout_period_time will be * roughly the same. */ |
380c27ca3
|
530 531 |
dom->period_time = wp_next_time(jiffies); mod_timer(&dom->period_timer, dom->period_time); |
eb608e3a3
|
532 |
} |
04fbfdc14
|
533 |
} |
c7981433e
|
534 535 536 537 538 |
/* * Increment @wb's writeout completion count and the global writeout * completion count. Called from test_clear_page_writeback(). */ static inline void __wb_writeout_inc(struct bdi_writeback *wb) |
dd5656e59
|
539 |
{ |
841710aa6
|
540 |
struct wb_domain *cgdom; |
dd5656e59
|
541 |
|
c7981433e
|
542 543 544 |
__inc_wb_stat(wb, WB_WRITTEN); wb_domain_writeout_inc(&global_wb_domain, &wb->completions, wb->bdi->max_prop_frac); |
841710aa6
|
545 546 547 548 549 |
cgdom = mem_cgroup_wb_domain(wb); if (cgdom) wb_domain_writeout_inc(cgdom, wb_memcg_completions(wb), wb->bdi->max_prop_frac); |
dd5656e59
|
550 |
} |
dd5656e59
|
551 |
|
93f78d882
|
552 |
void wb_writeout_inc(struct bdi_writeback *wb) |
04fbfdc14
|
553 |
{ |
dd5656e59
|
554 555 556 |
unsigned long flags; local_irq_save(flags); |
93f78d882
|
557 |
__wb_writeout_inc(wb); |
dd5656e59
|
558 |
local_irq_restore(flags); |
04fbfdc14
|
559 |
} |
93f78d882
|
560 |
EXPORT_SYMBOL_GPL(wb_writeout_inc); |
04fbfdc14
|
561 |
|
04fbfdc14
|
562 |
/* |
eb608e3a3
|
563 564 565 566 567 |
* On idle system, we can be called long after we scheduled because we use * deferred timers so count with missed periods. */ static void writeout_period(unsigned long t) { |
380c27ca3
|
568 569 |
struct wb_domain *dom = (void *)t; int miss_periods = (jiffies - dom->period_time) / |
eb608e3a3
|
570 |
VM_COMPLETIONS_PERIOD_LEN; |
380c27ca3
|
571 572 |
if (fprop_new_period(&dom->completions, miss_periods + 1)) { dom->period_time = wp_next_time(dom->period_time + |
eb608e3a3
|
573 |
miss_periods * VM_COMPLETIONS_PERIOD_LEN); |
380c27ca3
|
574 |
mod_timer(&dom->period_timer, dom->period_time); |
eb608e3a3
|
575 576 577 578 579 |
} else { /* * Aging has zeroed all fractions. Stop wasting CPU on period * updates. */ |
380c27ca3
|
580 |
dom->period_time = 0; |
eb608e3a3
|
581 582 |
} } |
380c27ca3
|
583 584 585 |
int wb_domain_init(struct wb_domain *dom, gfp_t gfp) { memset(dom, 0, sizeof(*dom)); |
dcc25ae76
|
586 587 |
spin_lock_init(&dom->lock); |
380c27ca3
|
588 589 590 |
init_timer_deferrable(&dom->period_timer); dom->period_timer.function = writeout_period; dom->period_timer.data = (unsigned long)dom; |
dcc25ae76
|
591 592 |
dom->dirty_limit_tstamp = jiffies; |
380c27ca3
|
593 594 |
return fprop_global_init(&dom->completions, gfp); } |
841710aa6
|
595 596 597 598 599 600 601 |
#ifdef CONFIG_CGROUP_WRITEBACK void wb_domain_exit(struct wb_domain *dom) { del_timer_sync(&dom->period_timer); fprop_global_destroy(&dom->completions); } #endif |
eb608e3a3
|
602 |
/* |
d08c429b0
|
603 604 605 |
* bdi_min_ratio keeps the sum of the minimum dirty shares of all * registered backing devices, which, for obvious reasons, can not * exceed 100%. |
189d3c4a9
|
606 |
*/ |
189d3c4a9
|
607 608 609 610 611 |
static unsigned int bdi_min_ratio; int bdi_set_min_ratio(struct backing_dev_info *bdi, unsigned int min_ratio) { int ret = 0; |
189d3c4a9
|
612 |
|
cfc4ba536
|
613 |
spin_lock_bh(&bdi_lock); |
a42dde041
|
614 |
if (min_ratio > bdi->max_ratio) { |
189d3c4a9
|
615 |
ret = -EINVAL; |
a42dde041
|
616 617 618 619 620 621 622 623 624 |
} else { min_ratio -= bdi->min_ratio; if (bdi_min_ratio + min_ratio < 100) { bdi_min_ratio += min_ratio; bdi->min_ratio += min_ratio; } else { ret = -EINVAL; } } |
cfc4ba536
|
625 |
spin_unlock_bh(&bdi_lock); |
a42dde041
|
626 627 628 629 630 631 |
return ret; } int bdi_set_max_ratio(struct backing_dev_info *bdi, unsigned max_ratio) { |
a42dde041
|
632 633 634 635 |
int ret = 0; if (max_ratio > 100) return -EINVAL; |
cfc4ba536
|
636 |
spin_lock_bh(&bdi_lock); |
a42dde041
|
637 638 639 640 |
if (bdi->min_ratio > max_ratio) { ret = -EINVAL; } else { bdi->max_ratio = max_ratio; |
eb608e3a3
|
641 |
bdi->max_prop_frac = (FPROP_FRAC_BASE * max_ratio) / 100; |
a42dde041
|
642 |
} |
cfc4ba536
|
643 |
spin_unlock_bh(&bdi_lock); |
189d3c4a9
|
644 645 646 |
return ret; } |
a42dde041
|
647 |
EXPORT_SYMBOL(bdi_set_max_ratio); |
189d3c4a9
|
648 |
|
6c14ae1e9
|
649 650 651 652 653 |
static unsigned long dirty_freerun_ceiling(unsigned long thresh, unsigned long bg_thresh) { return (thresh + bg_thresh) / 2; } |
c7981433e
|
654 655 |
static unsigned long hard_dirty_limit(struct wb_domain *dom, unsigned long thresh) |
ffd1f609a
|
656 |
{ |
dcc25ae76
|
657 |
return max(thresh, dom->dirty_limit); |
ffd1f609a
|
658 |
} |
c5edf9cdc
|
659 660 661 662 663 664 |
/* * Memory which can be further allocated to a memcg domain is capped by * system-wide clean memory excluding the amount being used in the domain. */ static void mdtc_calc_avail(struct dirty_throttle_control *mdtc, unsigned long filepages, unsigned long headroom) |
c2aa723a6
|
665 666 |
{ struct dirty_throttle_control *gdtc = mdtc_gdtc(mdtc); |
c5edf9cdc
|
667 668 669 |
unsigned long clean = filepages - min(filepages, mdtc->dirty); unsigned long global_clean = gdtc->avail - min(gdtc->avail, gdtc->dirty); unsigned long other_clean = global_clean - min(global_clean, clean); |
c2aa723a6
|
670 |
|
c5edf9cdc
|
671 |
mdtc->avail = filepages + min(headroom, other_clean); |
ffd1f609a
|
672 |
} |
6f7186562
|
673 |
/** |
b1cbc6d40
|
674 675 |
* __wb_calc_thresh - @wb's share of dirty throttling threshold * @dtc: dirty_throttle_context of interest |
1babe1838
|
676 |
* |
a88a341a7
|
677 |
* Returns @wb's dirty limit in pages. The term "dirty" in the context of |
6f7186562
|
678 |
* dirty balancing includes all PG_dirty, PG_writeback and NFS unstable pages. |
aed21ad28
|
679 680 681 682 683 684 |
* * Note that balance_dirty_pages() will only seriously take it as a hard limit * when sleeping max_pause per page is not enough to keep the dirty pages under * control. For example, when the device is completely stalled due to some error * conditions, or when there are 1000 dd tasks writing to a slow 10MB/s USB key. * In the other normal situations, it acts more gently by throttling the tasks |
a88a341a7
|
685 |
* more (rather than completely block them) when the wb dirty pages go high. |
1babe1838
|
686 |
* |
6f7186562
|
687 |
* It allocates high/low dirty limits to fast/slow devices, in order to prevent |
1babe1838
|
688 689 690 |
* - starving fast devices * - piling up dirty pages (that will take long time to sync) on slow devices * |
a88a341a7
|
691 |
* The wb's share of dirty limit will be adapting to its throughput and |
1babe1838
|
692 693 |
* bounded by the bdi->min_ratio and/or bdi->max_ratio parameters, if set. */ |
b1cbc6d40
|
694 |
static unsigned long __wb_calc_thresh(struct dirty_throttle_control *dtc) |
16c4042f0
|
695 |
{ |
e9f07dfd7
|
696 |
struct wb_domain *dom = dtc_dom(dtc); |
b1cbc6d40
|
697 |
unsigned long thresh = dtc->thresh; |
0d960a383
|
698 |
u64 wb_thresh; |
16c4042f0
|
699 |
long numerator, denominator; |
693108a8a
|
700 |
unsigned long wb_min_ratio, wb_max_ratio; |
04fbfdc14
|
701 |
|
16c4042f0
|
702 |
/* |
0d960a383
|
703 |
* Calculate this BDI's share of the thresh ratio. |
16c4042f0
|
704 |
*/ |
e9770b348
|
705 |
fprop_fraction_percpu(&dom->completions, dtc->wb_completions, |
380c27ca3
|
706 |
&numerator, &denominator); |
04fbfdc14
|
707 |
|
0d960a383
|
708 709 710 |
wb_thresh = (thresh * (100 - bdi_min_ratio)) / 100; wb_thresh *= numerator; do_div(wb_thresh, denominator); |
04fbfdc14
|
711 |
|
b1cbc6d40
|
712 |
wb_min_max_ratio(dtc->wb, &wb_min_ratio, &wb_max_ratio); |
04fbfdc14
|
713 |
|
0d960a383
|
714 715 716 |
wb_thresh += (thresh * wb_min_ratio) / 100; if (wb_thresh > (thresh * wb_max_ratio) / 100) wb_thresh = thresh * wb_max_ratio / 100; |
16c4042f0
|
717 |
|
0d960a383
|
718 |
return wb_thresh; |
1da177e4c
|
719 |
} |
b1cbc6d40
|
720 721 722 723 724 |
unsigned long wb_calc_thresh(struct bdi_writeback *wb, unsigned long thresh) { struct dirty_throttle_control gdtc = { GDTC_INIT(wb), .thresh = thresh }; return __wb_calc_thresh(&gdtc); |
1da177e4c
|
725 |
} |
6c14ae1e9
|
726 |
/* |
5a5374856
|
727 728 729 730 731 732 733 734 735 736 737 738 739 |
* setpoint - dirty 3 * f(dirty) := 1.0 + (----------------) * limit - setpoint * * it's a 3rd order polynomial that subjects to * * (1) f(freerun) = 2.0 => rampup dirty_ratelimit reasonably fast * (2) f(setpoint) = 1.0 => the balance point * (3) f(limit) = 0 => the hard limit * (4) df/dx <= 0 => negative feedback control * (5) the closer to setpoint, the smaller |df/dx| (and the reverse) * => fast response on large errors; small oscillation near setpoint */ |
d5c9fde3d
|
740 |
static long long pos_ratio_polynom(unsigned long setpoint, |
5a5374856
|
741 742 743 744 745 |
unsigned long dirty, unsigned long limit) { long long pos_ratio; long x; |
d5c9fde3d
|
746 |
x = div64_s64(((s64)setpoint - (s64)dirty) << RATELIMIT_CALC_SHIFT, |
464d1387a
|
747 |
(limit - setpoint) | 1); |
5a5374856
|
748 749 750 751 752 753 754 755 756 |
pos_ratio = x; pos_ratio = pos_ratio * x >> RATELIMIT_CALC_SHIFT; pos_ratio = pos_ratio * x >> RATELIMIT_CALC_SHIFT; pos_ratio += 1 << RATELIMIT_CALC_SHIFT; return clamp(pos_ratio, 0LL, 2LL << RATELIMIT_CALC_SHIFT); } /* |
6c14ae1e9
|
757 758 759 760 |
* Dirty position control. * * (o) global/bdi setpoints * |
de1fff37b
|
761 |
* We want the dirty pages be balanced around the global/wb setpoints. |
6c14ae1e9
|
762 763 764 765 766 767 768 769 770 |
* When the number of dirty pages is higher/lower than the setpoint, the * dirty position control ratio (and hence task dirty ratelimit) will be * decreased/increased to bring the dirty pages back to the setpoint. * * pos_ratio = 1 << RATELIMIT_CALC_SHIFT * * if (dirty < setpoint) scale up pos_ratio * if (dirty > setpoint) scale down pos_ratio * |
de1fff37b
|
771 772 |
* if (wb_dirty < wb_setpoint) scale up pos_ratio * if (wb_dirty > wb_setpoint) scale down pos_ratio |
6c14ae1e9
|
773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 |
* * task_ratelimit = dirty_ratelimit * pos_ratio >> RATELIMIT_CALC_SHIFT * * (o) global control line * * ^ pos_ratio * | * | |<===== global dirty control scope ======>| * 2.0 .............* * | .* * | . * * | . * * | . * * | . * * | . * * 1.0 ................................* * | . . * * | . . * * | . . * * | . . * * | . . * * 0 +------------.------------------.----------------------*-------------> * freerun^ setpoint^ limit^ dirty pages * |
de1fff37b
|
797 |
* (o) wb control line |
6c14ae1e9
|
798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 |
* * ^ pos_ratio * | * | * * | * * | * * | * * | * |<=========== span ============>| * 1.0 .......................* * | . * * | . * * | . * * | . * * | . * * | . * * | . * * | . * * | . * * | . * * | . * * 1/4 ...............................................* * * * * * * * * * * * * | . . * | . . * | . . * 0 +----------------------.-------------------------------.-------------> |
de1fff37b
|
823 |
* wb_setpoint^ x_intercept^ |
6c14ae1e9
|
824 |
* |
de1fff37b
|
825 |
* The wb control line won't drop below pos_ratio=1/4, so that wb_dirty can |
6c14ae1e9
|
826 827 |
* be smoothly throttled down to normal if it starts high in situations like * - start writing to a slow SD card and a fast disk at the same time. The SD |
de1fff37b
|
828 829 |
* card's wb_dirty may rush to many times higher than wb_setpoint. * - the wb dirty thresh drops quickly due to change of JBOD workload |
6c14ae1e9
|
830 |
*/ |
daddfa3cb
|
831 |
static void wb_position_ratio(struct dirty_throttle_control *dtc) |
6c14ae1e9
|
832 |
{ |
2bc00aef0
|
833 |
struct bdi_writeback *wb = dtc->wb; |
a88a341a7
|
834 |
unsigned long write_bw = wb->avg_write_bandwidth; |
2bc00aef0
|
835 |
unsigned long freerun = dirty_freerun_ceiling(dtc->thresh, dtc->bg_thresh); |
c7981433e
|
836 |
unsigned long limit = hard_dirty_limit(dtc_dom(dtc), dtc->thresh); |
2bc00aef0
|
837 |
unsigned long wb_thresh = dtc->wb_thresh; |
6c14ae1e9
|
838 839 |
unsigned long x_intercept; unsigned long setpoint; /* dirty pages' target balance point */ |
de1fff37b
|
840 |
unsigned long wb_setpoint; |
6c14ae1e9
|
841 842 843 |
unsigned long span; long long pos_ratio; /* for scaling up/down the rate limit */ long x; |
daddfa3cb
|
844 |
dtc->pos_ratio = 0; |
2bc00aef0
|
845 |
if (unlikely(dtc->dirty >= limit)) |
daddfa3cb
|
846 |
return; |
6c14ae1e9
|
847 848 849 850 |
/* * global setpoint * |
5a5374856
|
851 852 853 |
* See comment for pos_ratio_polynom(). */ setpoint = (freerun + limit) / 2; |
2bc00aef0
|
854 |
pos_ratio = pos_ratio_polynom(setpoint, dtc->dirty, limit); |
5a5374856
|
855 856 857 858 |
/* * The strictlimit feature is a tool preventing mistrusted filesystems * from growing a large number of dirty pages before throttling. For |
de1fff37b
|
859 860 |
* such filesystems balance_dirty_pages always checks wb counters * against wb limits. Even if global "nr_dirty" is under "freerun". |
5a5374856
|
861 862 863 864 |
* This is especially important for fuse which sets bdi->max_ratio to * 1% by default. Without strictlimit feature, fuse writeback may * consume arbitrary amount of RAM because it is accounted in * NR_WRITEBACK_TEMP which is not involved in calculating "nr_dirty". |
6c14ae1e9
|
865 |
* |
a88a341a7
|
866 |
* Here, in wb_position_ratio(), we calculate pos_ratio based on |
de1fff37b
|
867 |
* two values: wb_dirty and wb_thresh. Let's consider an example: |
5a5374856
|
868 869 |
* total amount of RAM is 16GB, bdi->max_ratio is equal to 1%, global * limits are set by default to 10% and 20% (background and throttle). |
de1fff37b
|
870 |
* Then wb_thresh is 1% of 20% of 16GB. This amounts to ~8K pages. |
0d960a383
|
871 |
* wb_calc_thresh(wb, bg_thresh) is about ~4K pages. wb_setpoint is |
de1fff37b
|
872 |
* about ~6K pages (as the average of background and throttle wb |
5a5374856
|
873 |
* limits). The 3rd order polynomial will provide positive feedback if |
de1fff37b
|
874 |
* wb_dirty is under wb_setpoint and vice versa. |
6c14ae1e9
|
875 |
* |
5a5374856
|
876 |
* Note, that we cannot use global counters in these calculations |
de1fff37b
|
877 |
* because we want to throttle process writing to a strictlimit wb |
5a5374856
|
878 879 |
* much earlier than global "freerun" is reached (~23MB vs. ~2.3GB * in the example above). |
6c14ae1e9
|
880 |
*/ |
a88a341a7
|
881 |
if (unlikely(wb->bdi->capabilities & BDI_CAP_STRICTLIMIT)) { |
de1fff37b
|
882 |
long long wb_pos_ratio; |
5a5374856
|
883 |
|
daddfa3cb
|
884 885 886 887 888 |
if (dtc->wb_dirty < 8) { dtc->pos_ratio = min_t(long long, pos_ratio * 2, 2 << RATELIMIT_CALC_SHIFT); return; } |
5a5374856
|
889 |
|
2bc00aef0
|
890 |
if (dtc->wb_dirty >= wb_thresh) |
daddfa3cb
|
891 |
return; |
5a5374856
|
892 |
|
970fb01ad
|
893 894 |
wb_setpoint = dirty_freerun_ceiling(wb_thresh, dtc->wb_bg_thresh); |
5a5374856
|
895 |
|
de1fff37b
|
896 |
if (wb_setpoint == 0 || wb_setpoint == wb_thresh) |
daddfa3cb
|
897 |
return; |
5a5374856
|
898 |
|
2bc00aef0
|
899 |
wb_pos_ratio = pos_ratio_polynom(wb_setpoint, dtc->wb_dirty, |
de1fff37b
|
900 |
wb_thresh); |
5a5374856
|
901 902 |
/* |
de1fff37b
|
903 904 |
* Typically, for strictlimit case, wb_setpoint << setpoint * and pos_ratio >> wb_pos_ratio. In the other words global |
5a5374856
|
905 |
* state ("dirty") is not limiting factor and we have to |
de1fff37b
|
906 |
* make decision based on wb counters. But there is an |
5a5374856
|
907 908 |
* important case when global pos_ratio should get precedence: * global limits are exceeded (e.g. due to activities on other |
de1fff37b
|
909 |
* wb's) while given strictlimit wb is below limit. |
5a5374856
|
910 |
* |
de1fff37b
|
911 |
* "pos_ratio * wb_pos_ratio" would work for the case above, |
5a5374856
|
912 |
* but it would look too non-natural for the case of all |
de1fff37b
|
913 |
* activity in the system coming from a single strictlimit wb |
5a5374856
|
914 915 916 917 |
* with bdi->max_ratio == 100%. * * Note that min() below somewhat changes the dynamics of the * control system. Normally, pos_ratio value can be well over 3 |
de1fff37b
|
918 |
* (when globally we are at freerun and wb is well below wb |
5a5374856
|
919 920 921 922 |
* setpoint). Now the maximum pos_ratio in the same situation * is 2. We might want to tweak this if we observe the control * system is too slow to adapt. */ |
daddfa3cb
|
923 924 |
dtc->pos_ratio = min(pos_ratio, wb_pos_ratio); return; |
5a5374856
|
925 |
} |
6c14ae1e9
|
926 927 928 |
/* * We have computed basic pos_ratio above based on global situation. If |
de1fff37b
|
929 |
* the wb is over/under its share of dirty pages, we want to scale |
6c14ae1e9
|
930 931 932 933 |
* pos_ratio further down/up. That is done by the following mechanism. */ /* |
de1fff37b
|
934 |
* wb setpoint |
6c14ae1e9
|
935 |
* |
de1fff37b
|
936 |
* f(wb_dirty) := 1.0 + k * (wb_dirty - wb_setpoint) |
6c14ae1e9
|
937 |
* |
de1fff37b
|
938 |
* x_intercept - wb_dirty |
6c14ae1e9
|
939 |
* := -------------------------- |
de1fff37b
|
940 |
* x_intercept - wb_setpoint |
6c14ae1e9
|
941 |
* |
de1fff37b
|
942 |
* The main wb control line is a linear function that subjects to |
6c14ae1e9
|
943 |
* |
de1fff37b
|
944 945 946 |
* (1) f(wb_setpoint) = 1.0 * (2) k = - 1 / (8 * write_bw) (in single wb case) * or equally: x_intercept = wb_setpoint + 8 * write_bw |
6c14ae1e9
|
947 |
* |
de1fff37b
|
948 |
* For single wb case, the dirty pages are observed to fluctuate |
6c14ae1e9
|
949 |
* regularly within range |
de1fff37b
|
950 |
* [wb_setpoint - write_bw/2, wb_setpoint + write_bw/2] |
6c14ae1e9
|
951 952 953 |
* for various filesystems, where (2) can yield in a reasonable 12.5% * fluctuation range for pos_ratio. * |
de1fff37b
|
954 |
* For JBOD case, wb_thresh (not wb_dirty!) could fluctuate up to its |
6c14ae1e9
|
955 |
* own size, so move the slope over accordingly and choose a slope that |
de1fff37b
|
956 |
* yields 100% pos_ratio fluctuation on suddenly doubled wb_thresh. |
6c14ae1e9
|
957 |
*/ |
2bc00aef0
|
958 959 |
if (unlikely(wb_thresh > dtc->thresh)) wb_thresh = dtc->thresh; |
aed21ad28
|
960 |
/* |
de1fff37b
|
961 |
* It's very possible that wb_thresh is close to 0 not because the |
aed21ad28
|
962 963 964 965 966 |
* device is slow, but that it has remained inactive for long time. * Honour such devices a reasonable good (hopefully IO efficient) * threshold, so that the occasional writes won't be blocked and active * writes can rampup the threshold quickly. */ |
2bc00aef0
|
967 |
wb_thresh = max(wb_thresh, (limit - dtc->dirty) / 8); |
6c14ae1e9
|
968 |
/* |
de1fff37b
|
969 970 |
* scale global setpoint to wb's: * wb_setpoint = setpoint * wb_thresh / thresh |
6c14ae1e9
|
971 |
*/ |
e4bc13adf
|
972 |
x = div_u64((u64)wb_thresh << 16, dtc->thresh | 1); |
de1fff37b
|
973 |
wb_setpoint = setpoint * (u64)x >> 16; |
6c14ae1e9
|
974 |
/* |
de1fff37b
|
975 976 |
* Use span=(8*write_bw) in single wb case as indicated by * (thresh - wb_thresh ~= 0) and transit to wb_thresh in JBOD case. |
6c14ae1e9
|
977 |
* |
de1fff37b
|
978 979 980 |
* wb_thresh thresh - wb_thresh * span = --------- * (8 * write_bw) + ------------------ * wb_thresh * thresh thresh |
6c14ae1e9
|
981 |
*/ |
2bc00aef0
|
982 |
span = (dtc->thresh - wb_thresh + 8 * write_bw) * (u64)x >> 16; |
de1fff37b
|
983 |
x_intercept = wb_setpoint + span; |
6c14ae1e9
|
984 |
|
2bc00aef0
|
985 986 |
if (dtc->wb_dirty < x_intercept - span / 4) { pos_ratio = div64_u64(pos_ratio * (x_intercept - dtc->wb_dirty), |
e4bc13adf
|
987 |
(x_intercept - wb_setpoint) | 1); |
6c14ae1e9
|
988 989 |
} else pos_ratio /= 4; |
8927f66c4
|
990 |
/* |
de1fff37b
|
991 |
* wb reserve area, safeguard against dirty pool underrun and disk idle |
8927f66c4
|
992 993 994 |
* It may push the desired control point of global dirty pages higher * than setpoint. */ |
de1fff37b
|
995 |
x_intercept = wb_thresh / 2; |
2bc00aef0
|
996 997 998 999 |
if (dtc->wb_dirty < x_intercept) { if (dtc->wb_dirty > x_intercept / 8) pos_ratio = div_u64(pos_ratio * x_intercept, dtc->wb_dirty); |
50657fc4d
|
1000 |
else |
8927f66c4
|
1001 1002 |
pos_ratio *= 8; } |
daddfa3cb
|
1003 |
dtc->pos_ratio = pos_ratio; |
6c14ae1e9
|
1004 |
} |
a88a341a7
|
1005 1006 1007 |
static void wb_update_write_bandwidth(struct bdi_writeback *wb, unsigned long elapsed, unsigned long written) |
e98be2d59
|
1008 1009 |
{ const unsigned long period = roundup_pow_of_two(3 * HZ); |
a88a341a7
|
1010 1011 |
unsigned long avg = wb->avg_write_bandwidth; unsigned long old = wb->write_bandwidth; |
e98be2d59
|
1012 1013 1014 1015 1016 1017 1018 1019 |
u64 bw; /* * bw = written * HZ / elapsed * * bw * elapsed + write_bandwidth * (period - elapsed) * write_bandwidth = --------------------------------------------------- * period |
c72efb658
|
1020 1021 1022 |
* * @written may have decreased due to account_page_redirty(). * Avoid underflowing @bw calculation. |
e98be2d59
|
1023 |
*/ |
a88a341a7
|
1024 |
bw = written - min(written, wb->written_stamp); |
e98be2d59
|
1025 1026 1027 1028 1029 1030 |
bw *= HZ; if (unlikely(elapsed > period)) { do_div(bw, elapsed); avg = bw; goto out; } |
a88a341a7
|
1031 |
bw += (u64)wb->write_bandwidth * (period - elapsed); |
e98be2d59
|
1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 |
bw >>= ilog2(period); /* * one more level of smoothing, for filtering out sudden spikes */ if (avg > old && old >= (unsigned long)bw) avg -= (avg - old) >> 3; if (avg < old && old <= (unsigned long)bw) avg += (old - avg) >> 3; out: |
95a46c65e
|
1044 1045 1046 1047 1048 1049 1050 |
/* keep avg > 0 to guarantee that tot > 0 if there are dirty wbs */ avg = max(avg, 1LU); if (wb_has_dirty_io(wb)) { long delta = avg - wb->avg_write_bandwidth; WARN_ON_ONCE(atomic_long_add_return(delta, &wb->bdi->tot_write_bandwidth) <= 0); } |
a88a341a7
|
1051 1052 |
wb->write_bandwidth = bw; wb->avg_write_bandwidth = avg; |
e98be2d59
|
1053 |
} |
2bc00aef0
|
1054 |
static void update_dirty_limit(struct dirty_throttle_control *dtc) |
c42843f2f
|
1055 |
{ |
e9f07dfd7
|
1056 |
struct wb_domain *dom = dtc_dom(dtc); |
2bc00aef0
|
1057 |
unsigned long thresh = dtc->thresh; |
dcc25ae76
|
1058 |
unsigned long limit = dom->dirty_limit; |
c42843f2f
|
1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 |
/* * Follow up in one step. */ if (limit < thresh) { limit = thresh; goto update; } /* * Follow down slowly. Use the higher one as the target, because thresh * may drop below dirty. This is exactly the reason to introduce |
dcc25ae76
|
1071 |
* dom->dirty_limit which is guaranteed to lie above the dirty pages. |
c42843f2f
|
1072 |
*/ |
2bc00aef0
|
1073 |
thresh = max(thresh, dtc->dirty); |
c42843f2f
|
1074 1075 1076 1077 1078 1079 |
if (limit > thresh) { limit -= (limit - thresh) >> 5; goto update; } return; update: |
dcc25ae76
|
1080 |
dom->dirty_limit = limit; |
c42843f2f
|
1081 |
} |
e9f07dfd7
|
1082 |
static void domain_update_bandwidth(struct dirty_throttle_control *dtc, |
c42843f2f
|
1083 1084 |
unsigned long now) { |
e9f07dfd7
|
1085 |
struct wb_domain *dom = dtc_dom(dtc); |
c42843f2f
|
1086 1087 1088 1089 |
/* * check locklessly first to optimize away locking for the most time */ |
dcc25ae76
|
1090 |
if (time_before(now, dom->dirty_limit_tstamp + BANDWIDTH_INTERVAL)) |
c42843f2f
|
1091 |
return; |
dcc25ae76
|
1092 1093 |
spin_lock(&dom->lock); if (time_after_eq(now, dom->dirty_limit_tstamp + BANDWIDTH_INTERVAL)) { |
2bc00aef0
|
1094 |
update_dirty_limit(dtc); |
dcc25ae76
|
1095 |
dom->dirty_limit_tstamp = now; |
c42843f2f
|
1096 |
} |
dcc25ae76
|
1097 |
spin_unlock(&dom->lock); |
c42843f2f
|
1098 |
} |
be3ffa276
|
1099 |
/* |
de1fff37b
|
1100 |
* Maintain wb->dirty_ratelimit, the base dirty throttle rate. |
be3ffa276
|
1101 |
* |
de1fff37b
|
1102 |
* Normal wb tasks will be curbed at or below it in long term. |
be3ffa276
|
1103 1104 |
* Obviously it should be around (write_bw / N) when there are N dd tasks. */ |
2bc00aef0
|
1105 |
static void wb_update_dirty_ratelimit(struct dirty_throttle_control *dtc, |
a88a341a7
|
1106 1107 |
unsigned long dirtied, unsigned long elapsed) |
be3ffa276
|
1108 |
{ |
2bc00aef0
|
1109 1110 1111 |
struct bdi_writeback *wb = dtc->wb; unsigned long dirty = dtc->dirty; unsigned long freerun = dirty_freerun_ceiling(dtc->thresh, dtc->bg_thresh); |
c7981433e
|
1112 |
unsigned long limit = hard_dirty_limit(dtc_dom(dtc), dtc->thresh); |
7381131cb
|
1113 |
unsigned long setpoint = (freerun + limit) / 2; |
a88a341a7
|
1114 1115 |
unsigned long write_bw = wb->avg_write_bandwidth; unsigned long dirty_ratelimit = wb->dirty_ratelimit; |
be3ffa276
|
1116 1117 1118 |
unsigned long dirty_rate; unsigned long task_ratelimit; unsigned long balanced_dirty_ratelimit; |
7381131cb
|
1119 1120 |
unsigned long step; unsigned long x; |
d59b1087a
|
1121 |
unsigned long shift; |
be3ffa276
|
1122 1123 1124 1125 1126 |
/* * The dirty rate will match the writeout rate in long term, except * when dirty pages are truncated by userspace or re-dirtied by FS. */ |
a88a341a7
|
1127 |
dirty_rate = (dirtied - wb->dirtied_stamp) * HZ / elapsed; |
be3ffa276
|
1128 |
|
be3ffa276
|
1129 1130 1131 1132 |
/* * task_ratelimit reflects each dd's dirty rate for the past 200ms. */ task_ratelimit = (u64)dirty_ratelimit * |
daddfa3cb
|
1133 |
dtc->pos_ratio >> RATELIMIT_CALC_SHIFT; |
be3ffa276
|
1134 1135 1136 1137 |
task_ratelimit++; /* it helps rampup dirty_ratelimit from tiny values */ /* * A linear estimation of the "balanced" throttle rate. The theory is, |
de1fff37b
|
1138 |
* if there are N dd tasks, each throttled at task_ratelimit, the wb's |
be3ffa276
|
1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 |
* dirty_rate will be measured to be (N * task_ratelimit). So the below * formula will yield the balanced rate limit (write_bw / N). * * Note that the expanded form is not a pure rate feedback: * rate_(i+1) = rate_(i) * (write_bw / dirty_rate) (1) * but also takes pos_ratio into account: * rate_(i+1) = rate_(i) * (write_bw / dirty_rate) * pos_ratio (2) * * (1) is not realistic because pos_ratio also takes part in balancing * the dirty rate. Consider the state * pos_ratio = 0.5 (3) * rate = 2 * (write_bw / N) (4) * If (1) is used, it will stuck in that state! Because each dd will * be throttled at * task_ratelimit = pos_ratio * rate = (write_bw / N) (5) * yielding * dirty_rate = N * task_ratelimit = write_bw (6) * put (6) into (1) we get * rate_(i+1) = rate_(i) (7) * * So we end up using (2) to always keep * rate_(i+1) ~= (write_bw / N) (8) * regardless of the value of pos_ratio. As long as (8) is satisfied, * pos_ratio is able to drive itself to 1.0, which is not only where * the dirty count meet the setpoint, but also where the slope of * pos_ratio is most flat and hence task_ratelimit is least fluctuated. */ balanced_dirty_ratelimit = div_u64((u64)task_ratelimit * write_bw, dirty_rate | 1); |
bdaac4902
|
1168 1169 1170 1171 1172 |
/* * balanced_dirty_ratelimit ~= (write_bw / N) <= write_bw */ if (unlikely(balanced_dirty_ratelimit > write_bw)) balanced_dirty_ratelimit = write_bw; |
be3ffa276
|
1173 |
|
7381131cb
|
1174 1175 1176 |
/* * We could safely do this and return immediately: * |
de1fff37b
|
1177 |
* wb->dirty_ratelimit = balanced_dirty_ratelimit; |
7381131cb
|
1178 1179 |
* * However to get a more stable dirty_ratelimit, the below elaborated |
331cbdeed
|
1180 |
* code makes use of task_ratelimit to filter out singular points and |
7381131cb
|
1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 |
* limit the step size. * * The below code essentially only uses the relative value of * * task_ratelimit - dirty_ratelimit * = (pos_ratio - 1) * dirty_ratelimit * * which reflects the direction and size of dirty position error. */ /* * dirty_ratelimit will follow balanced_dirty_ratelimit iff * task_ratelimit is on the same side of dirty_ratelimit, too. * For example, when * - dirty_ratelimit > balanced_dirty_ratelimit * - dirty_ratelimit > task_ratelimit (dirty pages are above setpoint) * lowering dirty_ratelimit will help meet both the position and rate * control targets. Otherwise, don't update dirty_ratelimit if it will * only help meet the rate target. After all, what the users ultimately * feel and care are stable dirty rate and small position error. * * |task_ratelimit - dirty_ratelimit| is used to limit the step size |
331cbdeed
|
1203 |
* and filter out the singular points of balanced_dirty_ratelimit. Which |
7381131cb
|
1204 1205 1206 1207 1208 |
* keeps jumping around randomly and can even leap far away at times * due to the small 200ms estimation period of dirty_rate (we want to * keep that period small to reduce time lags). */ step = 0; |
5a5374856
|
1209 1210 |
/* |
de1fff37b
|
1211 |
* For strictlimit case, calculations above were based on wb counters |
a88a341a7
|
1212 |
* and limits (starting from pos_ratio = wb_position_ratio() and up to |
5a5374856
|
1213 |
* balanced_dirty_ratelimit = task_ratelimit * write_bw / dirty_rate). |
de1fff37b
|
1214 1215 |
* Hence, to calculate "step" properly, we have to use wb_dirty as * "dirty" and wb_setpoint as "setpoint". |
5a5374856
|
1216 |
* |
de1fff37b
|
1217 1218 |
* We rampup dirty_ratelimit forcibly if wb_dirty is low because * it's possible that wb_thresh is close to zero due to inactivity |
970fb01ad
|
1219 |
* of backing device. |
5a5374856
|
1220 |
*/ |
a88a341a7
|
1221 |
if (unlikely(wb->bdi->capabilities & BDI_CAP_STRICTLIMIT)) { |
2bc00aef0
|
1222 1223 1224 |
dirty = dtc->wb_dirty; if (dtc->wb_dirty < 8) setpoint = dtc->wb_dirty + 1; |
5a5374856
|
1225 |
else |
970fb01ad
|
1226 |
setpoint = (dtc->wb_thresh + dtc->wb_bg_thresh) / 2; |
5a5374856
|
1227 |
} |
7381131cb
|
1228 |
if (dirty < setpoint) { |
a88a341a7
|
1229 |
x = min3(wb->balanced_dirty_ratelimit, |
7c809968f
|
1230 |
balanced_dirty_ratelimit, task_ratelimit); |
7381131cb
|
1231 1232 1233 |
if (dirty_ratelimit < x) step = x - dirty_ratelimit; } else { |
a88a341a7
|
1234 |
x = max3(wb->balanced_dirty_ratelimit, |
7c809968f
|
1235 |
balanced_dirty_ratelimit, task_ratelimit); |
7381131cb
|
1236 1237 1238 1239 1240 1241 1242 1243 1244 |
if (dirty_ratelimit > x) step = dirty_ratelimit - x; } /* * Don't pursue 100% rate matching. It's impossible since the balanced * rate itself is constantly fluctuating. So decrease the track speed * when it gets close to the target. Helps eliminate pointless tremors. */ |
d59b1087a
|
1245 1246 1247 1248 1249 |
shift = dirty_ratelimit / (2 * step + 1); if (shift < BITS_PER_LONG) step = DIV_ROUND_UP(step >> shift, 8); else step = 0; |
7381131cb
|
1250 1251 1252 1253 1254 |
if (dirty_ratelimit < balanced_dirty_ratelimit) dirty_ratelimit += step; else dirty_ratelimit -= step; |
a88a341a7
|
1255 1256 |
wb->dirty_ratelimit = max(dirty_ratelimit, 1UL); wb->balanced_dirty_ratelimit = balanced_dirty_ratelimit; |
b48c104d2
|
1257 |
|
5634cc2aa
|
1258 |
trace_bdi_dirty_ratelimit(wb, dirty_rate, task_ratelimit); |
be3ffa276
|
1259 |
} |
c2aa723a6
|
1260 1261 |
static void __wb_update_bandwidth(struct dirty_throttle_control *gdtc, struct dirty_throttle_control *mdtc, |
8a7317995
|
1262 1263 |
unsigned long start_time, bool update_ratelimit) |
e98be2d59
|
1264 |
{ |
c2aa723a6
|
1265 |
struct bdi_writeback *wb = gdtc->wb; |
e98be2d59
|
1266 |
unsigned long now = jiffies; |
a88a341a7
|
1267 |
unsigned long elapsed = now - wb->bw_time_stamp; |
be3ffa276
|
1268 |
unsigned long dirtied; |
e98be2d59
|
1269 |
unsigned long written; |
8a7317995
|
1270 |
lockdep_assert_held(&wb->list_lock); |
e98be2d59
|
1271 1272 1273 1274 1275 |
/* * rate-limit, only update once every 200ms. */ if (elapsed < BANDWIDTH_INTERVAL) return; |
a88a341a7
|
1276 1277 |
dirtied = percpu_counter_read(&wb->stat[WB_DIRTIED]); written = percpu_counter_read(&wb->stat[WB_WRITTEN]); |
e98be2d59
|
1278 1279 1280 1281 1282 |
/* * Skip quiet periods when disk bandwidth is under-utilized. * (at least 1s idle time between two flusher runs) */ |
a88a341a7
|
1283 |
if (elapsed > HZ && time_before(wb->bw_time_stamp, start_time)) |
e98be2d59
|
1284 |
goto snapshot; |
8a7317995
|
1285 |
if (update_ratelimit) { |
c2aa723a6
|
1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 |
domain_update_bandwidth(gdtc, now); wb_update_dirty_ratelimit(gdtc, dirtied, elapsed); /* * @mdtc is always NULL if !CGROUP_WRITEBACK but the * compiler has no way to figure that out. Help it. */ if (IS_ENABLED(CONFIG_CGROUP_WRITEBACK) && mdtc) { domain_update_bandwidth(mdtc, now); wb_update_dirty_ratelimit(mdtc, dirtied, elapsed); } |
be3ffa276
|
1297 |
} |
a88a341a7
|
1298 |
wb_update_write_bandwidth(wb, elapsed, written); |
e98be2d59
|
1299 1300 |
snapshot: |
a88a341a7
|
1301 1302 1303 |
wb->dirtied_stamp = dirtied; wb->written_stamp = written; wb->bw_time_stamp = now; |
e98be2d59
|
1304 |
} |
8a7317995
|
1305 |
void wb_update_bandwidth(struct bdi_writeback *wb, unsigned long start_time) |
e98be2d59
|
1306 |
{ |
2bc00aef0
|
1307 |
struct dirty_throttle_control gdtc = { GDTC_INIT(wb) }; |
c2aa723a6
|
1308 |
__wb_update_bandwidth(&gdtc, NULL, start_time, false); |
e98be2d59
|
1309 |
} |
1da177e4c
|
1310 |
/* |
d0e1d66b5
|
1311 |
* After a task dirtied this many pages, balance_dirty_pages_ratelimited() |
9d823e8f6
|
1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 |
* will look to see if it needs to start dirty throttling. * * If dirty_poll_interval is too low, big NUMA machines will call the expensive * global_page_state() too often. So scale it near-sqrt to the safety margin * (the number of pages we may dirty without exceeding the dirty limits). */ static unsigned long dirty_poll_interval(unsigned long dirty, unsigned long thresh) { if (thresh > dirty) return 1UL << (ilog2(thresh - dirty) >> 1); return 1; } |
a88a341a7
|
1326 |
static unsigned long wb_max_pause(struct bdi_writeback *wb, |
de1fff37b
|
1327 |
unsigned long wb_dirty) |
c8462cc9d
|
1328 |
{ |
a88a341a7
|
1329 |
unsigned long bw = wb->avg_write_bandwidth; |
e3b6c655b
|
1330 |
unsigned long t; |
c8462cc9d
|
1331 |
|
7ccb9ad53
|
1332 1333 1334 1335 1336 1337 1338 |
/* * Limit pause time for small memory systems. If sleeping for too long * time, a small pool of dirty/writeback pages may go empty and disk go * idle. * * 8 serves as the safety ratio. */ |
de1fff37b
|
1339 |
t = wb_dirty / (1 + bw / roundup_pow_of_two(1 + HZ / 8)); |
7ccb9ad53
|
1340 |
t++; |
e3b6c655b
|
1341 |
return min_t(unsigned long, t, MAX_PAUSE); |
7ccb9ad53
|
1342 |
} |
a88a341a7
|
1343 1344 1345 1346 1347 |
static long wb_min_pause(struct bdi_writeback *wb, long max_pause, unsigned long task_ratelimit, unsigned long dirty_ratelimit, int *nr_dirtied_pause) |
c8462cc9d
|
1348 |
{ |
a88a341a7
|
1349 1350 |
long hi = ilog2(wb->avg_write_bandwidth); long lo = ilog2(wb->dirty_ratelimit); |
7ccb9ad53
|
1351 1352 1353 |
long t; /* target pause */ long pause; /* estimated next pause */ int pages; /* target nr_dirtied_pause */ |
c8462cc9d
|
1354 |
|
7ccb9ad53
|
1355 1356 |
/* target for 10ms pause on 1-dd case */ t = max(1, HZ / 100); |
c8462cc9d
|
1357 1358 1359 1360 1361 |
/* * Scale up pause time for concurrent dirtiers in order to reduce CPU * overheads. * |
7ccb9ad53
|
1362 |
* (N * 10ms) on 2^N concurrent tasks. |
c8462cc9d
|
1363 1364 |
*/ if (hi > lo) |
7ccb9ad53
|
1365 |
t += (hi - lo) * (10 * HZ) / 1024; |
c8462cc9d
|
1366 1367 |
/* |
7ccb9ad53
|
1368 1369 1370 1371 1372 1373 1374 1375 |
* This is a bit convoluted. We try to base the next nr_dirtied_pause * on the much more stable dirty_ratelimit. However the next pause time * will be computed based on task_ratelimit and the two rate limits may * depart considerably at some time. Especially if task_ratelimit goes * below dirty_ratelimit/2 and the target pause is max_pause, the next * pause time will be max_pause*2 _trimmed down_ to max_pause. As a * result task_ratelimit won't be executed faithfully, which could * eventually bring down dirty_ratelimit. |
c8462cc9d
|
1376 |
* |
7ccb9ad53
|
1377 1378 1379 1380 1381 1382 1383 |
* We apply two rules to fix it up: * 1) try to estimate the next pause time and if necessary, use a lower * nr_dirtied_pause so as not to exceed max_pause. When this happens, * nr_dirtied_pause will be "dancing" with task_ratelimit. * 2) limit the target pause time to max_pause/2, so that the normal * small fluctuations of task_ratelimit won't trigger rule (1) and * nr_dirtied_pause will remain as stable as dirty_ratelimit. |
c8462cc9d
|
1384 |
*/ |
7ccb9ad53
|
1385 1386 |
t = min(t, 1 + max_pause / 2); pages = dirty_ratelimit * t / roundup_pow_of_two(HZ); |
c8462cc9d
|
1387 1388 |
/* |
5b9b35743
|
1389 1390 1391 1392 1393 1394 |
* Tiny nr_dirtied_pause is found to hurt I/O performance in the test * case fio-mmap-randwrite-64k, which does 16*{sync read, async write}. * When the 16 consecutive reads are often interrupted by some dirty * throttling pause during the async writes, cfq will go into idles * (deadline is fine). So push nr_dirtied_pause as high as possible * until reaches DIRTY_POLL_THRESH=32 pages. |
c8462cc9d
|
1395 |
*/ |
5b9b35743
|
1396 1397 1398 1399 1400 1401 1402 1403 |
if (pages < DIRTY_POLL_THRESH) { t = max_pause; pages = dirty_ratelimit * t / roundup_pow_of_two(HZ); if (pages > DIRTY_POLL_THRESH) { pages = DIRTY_POLL_THRESH; t = HZ * DIRTY_POLL_THRESH / dirty_ratelimit; } } |
7ccb9ad53
|
1404 1405 1406 1407 1408 |
pause = HZ * pages / (task_ratelimit + 1); if (pause > max_pause) { t = max_pause; pages = task_ratelimit * t / roundup_pow_of_two(HZ); } |
c8462cc9d
|
1409 |
|
7ccb9ad53
|
1410 |
*nr_dirtied_pause = pages; |
c8462cc9d
|
1411 |
/* |
7ccb9ad53
|
1412 |
* The minimal pause time will normally be half the target pause time. |
c8462cc9d
|
1413 |
*/ |
5b9b35743
|
1414 |
return pages >= DIRTY_POLL_THRESH ? 1 + t / 2 : t; |
c8462cc9d
|
1415 |
} |
970fb01ad
|
1416 |
static inline void wb_dirty_limits(struct dirty_throttle_control *dtc) |
5a5374856
|
1417 |
{ |
2bc00aef0
|
1418 |
struct bdi_writeback *wb = dtc->wb; |
93f78d882
|
1419 |
unsigned long wb_reclaimable; |
5a5374856
|
1420 1421 |
/* |
de1fff37b
|
1422 |
* wb_thresh is not treated as some limiting factor as |
5a5374856
|
1423 |
* dirty_thresh, due to reasons |
de1fff37b
|
1424 |
* - in JBOD setup, wb_thresh can fluctuate a lot |
5a5374856
|
1425 |
* - in a system with HDD and USB key, the USB key may somehow |
de1fff37b
|
1426 1427 |
* go into state (wb_dirty >> wb_thresh) either because * wb_dirty starts high, or because wb_thresh drops low. |
5a5374856
|
1428 |
* In this case we don't want to hard throttle the USB key |
de1fff37b
|
1429 1430 |
* dirtiers for 100 seconds until wb_dirty drops under * wb_thresh. Instead the auxiliary wb control line in |
a88a341a7
|
1431 |
* wb_position_ratio() will let the dirtier task progress |
de1fff37b
|
1432 |
* at some rate <= (write_bw / 2) for bringing down wb_dirty. |
5a5374856
|
1433 |
*/ |
b1cbc6d40
|
1434 |
dtc->wb_thresh = __wb_calc_thresh(dtc); |
970fb01ad
|
1435 1436 |
dtc->wb_bg_thresh = dtc->thresh ? div_u64((u64)dtc->wb_thresh * dtc->bg_thresh, dtc->thresh) : 0; |
5a5374856
|
1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 |
/* * In order to avoid the stacked BDI deadlock we need * to ensure we accurately count the 'dirty' pages when * the threshold is low. * * Otherwise it would be possible to get thresh+n pages * reported dirty, even though there are thresh-m pages * actually dirty; with m+n sitting in the percpu * deltas. */ |
2bc00aef0
|
1448 |
if (dtc->wb_thresh < 2 * wb_stat_error(wb)) { |
93f78d882
|
1449 |
wb_reclaimable = wb_stat_sum(wb, WB_RECLAIMABLE); |
2bc00aef0
|
1450 |
dtc->wb_dirty = wb_reclaimable + wb_stat_sum(wb, WB_WRITEBACK); |
5a5374856
|
1451 |
} else { |
93f78d882
|
1452 |
wb_reclaimable = wb_stat(wb, WB_RECLAIMABLE); |
2bc00aef0
|
1453 |
dtc->wb_dirty = wb_reclaimable + wb_stat(wb, WB_WRITEBACK); |
5a5374856
|
1454 1455 |
} } |
9d823e8f6
|
1456 |
/* |
1da177e4c
|
1457 1458 |
* balance_dirty_pages() must be called by processes which are generating dirty * data. It looks at the number of dirty pages in the machine and will force |
143dfe861
|
1459 |
* the caller to wait once crossing the (background_thresh + dirty_thresh) / 2. |
5b0830cb9
|
1460 1461 |
* If we're over `background_thresh' then the writeback threads are woken to * perform some writeout. |
1da177e4c
|
1462 |
*/ |
3a2e9a5a2
|
1463 |
static void balance_dirty_pages(struct address_space *mapping, |
dfb8ae567
|
1464 |
struct bdi_writeback *wb, |
143dfe861
|
1465 |
unsigned long pages_dirtied) |
1da177e4c
|
1466 |
{ |
2bc00aef0
|
1467 |
struct dirty_throttle_control gdtc_stor = { GDTC_INIT(wb) }; |
c2aa723a6
|
1468 |
struct dirty_throttle_control mdtc_stor = { MDTC_INIT(wb, &gdtc_stor) }; |
2bc00aef0
|
1469 |
struct dirty_throttle_control * const gdtc = &gdtc_stor; |
c2aa723a6
|
1470 1471 1472 |
struct dirty_throttle_control * const mdtc = mdtc_valid(&mdtc_stor) ? &mdtc_stor : NULL; struct dirty_throttle_control *sdtc; |
143dfe861
|
1473 |
unsigned long nr_reclaimable; /* = file_dirty + unstable_nfs */ |
83712358b
|
1474 |
long period; |
7ccb9ad53
|
1475 1476 1477 1478 |
long pause; long max_pause; long min_pause; int nr_dirtied_pause; |
e50e37201
|
1479 |
bool dirty_exceeded = false; |
143dfe861
|
1480 |
unsigned long task_ratelimit; |
7ccb9ad53
|
1481 |
unsigned long dirty_ratelimit; |
dfb8ae567
|
1482 |
struct backing_dev_info *bdi = wb->bdi; |
5a5374856
|
1483 |
bool strictlimit = bdi->capabilities & BDI_CAP_STRICTLIMIT; |
e98be2d59
|
1484 |
unsigned long start_time = jiffies; |
1da177e4c
|
1485 1486 |
for (;;) { |
83712358b
|
1487 |
unsigned long now = jiffies; |
2bc00aef0
|
1488 |
unsigned long dirty, thresh, bg_thresh; |
50e55bf62
|
1489 1490 1491 |
unsigned long m_dirty = 0; /* stop bogus uninit warnings */ unsigned long m_thresh = 0; unsigned long m_bg_thresh = 0; |
83712358b
|
1492 |
|
143dfe861
|
1493 1494 1495 1496 1497 1498 |
/* * Unstable writes are a feature of certain networked * filesystems (i.e. NFS) in which data may have been * written to the server's write cache, but has not yet * been flushed to permanent storage. */ |
5fce25a9d
|
1499 1500 |
nr_reclaimable = global_page_state(NR_FILE_DIRTY) + global_page_state(NR_UNSTABLE_NFS); |
9fc3a43e1
|
1501 |
gdtc->avail = global_dirtyable_memory(); |
2bc00aef0
|
1502 |
gdtc->dirty = nr_reclaimable + global_page_state(NR_WRITEBACK); |
5fce25a9d
|
1503 |
|
9fc3a43e1
|
1504 |
domain_dirty_limits(gdtc); |
16c4042f0
|
1505 |
|
5a5374856
|
1506 |
if (unlikely(strictlimit)) { |
970fb01ad
|
1507 |
wb_dirty_limits(gdtc); |
5a5374856
|
1508 |
|
2bc00aef0
|
1509 1510 |
dirty = gdtc->wb_dirty; thresh = gdtc->wb_thresh; |
970fb01ad
|
1511 |
bg_thresh = gdtc->wb_bg_thresh; |
5a5374856
|
1512 |
} else { |
2bc00aef0
|
1513 1514 1515 |
dirty = gdtc->dirty; thresh = gdtc->thresh; bg_thresh = gdtc->bg_thresh; |
5a5374856
|
1516 |
} |
c2aa723a6
|
1517 |
if (mdtc) { |
c5edf9cdc
|
1518 |
unsigned long filepages, headroom, writeback; |
c2aa723a6
|
1519 1520 1521 1522 1523 |
/* * If @wb belongs to !root memcg, repeat the same * basic calculations for the memcg domain. */ |
c5edf9cdc
|
1524 1525 |
mem_cgroup_wb_stats(wb, &filepages, &headroom, &mdtc->dirty, &writeback); |
c2aa723a6
|
1526 |
mdtc->dirty += writeback; |
c5edf9cdc
|
1527 |
mdtc_calc_avail(mdtc, filepages, headroom); |
c2aa723a6
|
1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 |
domain_dirty_limits(mdtc); if (unlikely(strictlimit)) { wb_dirty_limits(mdtc); m_dirty = mdtc->wb_dirty; m_thresh = mdtc->wb_thresh; m_bg_thresh = mdtc->wb_bg_thresh; } else { m_dirty = mdtc->dirty; m_thresh = mdtc->thresh; m_bg_thresh = mdtc->bg_thresh; } |
5a5374856
|
1541 |
} |
16c4042f0
|
1542 1543 1544 |
/* * Throttle it only when the background writeback cannot * catch-up. This avoids (excessively) small writeouts |
de1fff37b
|
1545 |
* when the wb limits are ramping up in case of !strictlimit. |
5a5374856
|
1546 |
* |
de1fff37b
|
1547 1548 |
* In strictlimit case make decision based on the wb counters * and limits. Small writeouts when the wb limits are ramping |
5a5374856
|
1549 |
* up are the price we consciously pay for strictlimit-ing. |
c2aa723a6
|
1550 1551 1552 |
* * If memcg domain is in effect, @dirty should be under * both global and memcg freerun ceilings. |
16c4042f0
|
1553 |
*/ |
c2aa723a6
|
1554 1555 1556 1557 1558 |
if (dirty <= dirty_freerun_ceiling(thresh, bg_thresh) && (!mdtc || m_dirty <= dirty_freerun_ceiling(m_thresh, m_bg_thresh))) { unsigned long intv = dirty_poll_interval(dirty, thresh); unsigned long m_intv = ULONG_MAX; |
83712358b
|
1559 1560 |
current->dirty_paused_when = now; current->nr_dirtied = 0; |
c2aa723a6
|
1561 1562 1563 |
if (mdtc) m_intv = dirty_poll_interval(m_dirty, m_thresh); current->nr_dirtied_pause = min(intv, m_intv); |
16c4042f0
|
1564 |
break; |
83712358b
|
1565 |
} |
16c4042f0
|
1566 |
|
bc05873dc
|
1567 |
if (unlikely(!writeback_in_progress(wb))) |
9ecf4866c
|
1568 |
wb_start_background_writeback(wb); |
143dfe861
|
1569 |
|
c2aa723a6
|
1570 1571 1572 1573 |
/* * Calculate global domain's pos_ratio and select the * global dtc by default. */ |
5a5374856
|
1574 |
if (!strictlimit) |
970fb01ad
|
1575 |
wb_dirty_limits(gdtc); |
5fce25a9d
|
1576 |
|
2bc00aef0
|
1577 1578 |
dirty_exceeded = (gdtc->wb_dirty > gdtc->wb_thresh) && ((gdtc->dirty > gdtc->thresh) || strictlimit); |
daddfa3cb
|
1579 1580 |
wb_position_ratio(gdtc); |
c2aa723a6
|
1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 |
sdtc = gdtc; if (mdtc) { /* * If memcg domain is in effect, calculate its * pos_ratio. @wb should satisfy constraints from * both global and memcg domains. Choose the one * w/ lower pos_ratio. */ if (!strictlimit) wb_dirty_limits(mdtc); dirty_exceeded |= (mdtc->wb_dirty > mdtc->wb_thresh) && ((mdtc->dirty > mdtc->thresh) || strictlimit); wb_position_ratio(mdtc); if (mdtc->pos_ratio < gdtc->pos_ratio) sdtc = mdtc; } |
daddfa3cb
|
1600 |
|
a88a341a7
|
1601 1602 |
if (dirty_exceeded && !wb->dirty_exceeded) wb->dirty_exceeded = 1; |
1da177e4c
|
1603 |
|
8a7317995
|
1604 1605 1606 |
if (time_is_before_jiffies(wb->bw_time_stamp + BANDWIDTH_INTERVAL)) { spin_lock(&wb->list_lock); |
c2aa723a6
|
1607 |
__wb_update_bandwidth(gdtc, mdtc, start_time, true); |
8a7317995
|
1608 1609 |
spin_unlock(&wb->list_lock); } |
e98be2d59
|
1610 |
|
c2aa723a6
|
1611 |
/* throttle according to the chosen dtc */ |
a88a341a7
|
1612 |
dirty_ratelimit = wb->dirty_ratelimit; |
c2aa723a6
|
1613 |
task_ratelimit = ((u64)dirty_ratelimit * sdtc->pos_ratio) >> |
3a73dbbc9
|
1614 |
RATELIMIT_CALC_SHIFT; |
c2aa723a6
|
1615 |
max_pause = wb_max_pause(wb, sdtc->wb_dirty); |
a88a341a7
|
1616 1617 1618 |
min_pause = wb_min_pause(wb, max_pause, task_ratelimit, dirty_ratelimit, &nr_dirtied_pause); |
7ccb9ad53
|
1619 |
|
3a73dbbc9
|
1620 |
if (unlikely(task_ratelimit == 0)) { |
83712358b
|
1621 |
period = max_pause; |
c8462cc9d
|
1622 |
pause = max_pause; |
143dfe861
|
1623 |
goto pause; |
04fbfdc14
|
1624 |
} |
83712358b
|
1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 |
period = HZ * pages_dirtied / task_ratelimit; pause = period; if (current->dirty_paused_when) pause -= now - current->dirty_paused_when; /* * For less than 1s think time (ext3/4 may block the dirtier * for up to 800ms from time to time on 1-HDD; so does xfs, * however at much less frequency), try to compensate it in * future periods by updating the virtual time; otherwise just * do a reset, as it may be a light dirtier. */ |
7ccb9ad53
|
1636 |
if (pause < min_pause) { |
5634cc2aa
|
1637 |
trace_balance_dirty_pages(wb, |
c2aa723a6
|
1638 1639 1640 1641 1642 |
sdtc->thresh, sdtc->bg_thresh, sdtc->dirty, sdtc->wb_thresh, sdtc->wb_dirty, |
ece13ac31
|
1643 1644 1645 |
dirty_ratelimit, task_ratelimit, pages_dirtied, |
83712358b
|
1646 |
period, |
7ccb9ad53
|
1647 |
min(pause, 0L), |
ece13ac31
|
1648 |
start_time); |
83712358b
|
1649 1650 1651 1652 1653 1654 |
if (pause < -HZ) { current->dirty_paused_when = now; current->nr_dirtied = 0; } else if (period) { current->dirty_paused_when += period; current->nr_dirtied = 0; |
7ccb9ad53
|
1655 1656 |
} else if (current->nr_dirtied_pause <= pages_dirtied) current->nr_dirtied_pause += pages_dirtied; |
57fc978cf
|
1657 |
break; |
04fbfdc14
|
1658 |
} |
7ccb9ad53
|
1659 1660 1661 1662 1663 |
if (unlikely(pause > max_pause)) { /* for occasional dropped task_ratelimit */ now += min(pause - max_pause, max_pause); pause = max_pause; } |
143dfe861
|
1664 1665 |
pause: |
5634cc2aa
|
1666 |
trace_balance_dirty_pages(wb, |
c2aa723a6
|
1667 1668 1669 1670 1671 |
sdtc->thresh, sdtc->bg_thresh, sdtc->dirty, sdtc->wb_thresh, sdtc->wb_dirty, |
ece13ac31
|
1672 1673 1674 |
dirty_ratelimit, task_ratelimit, pages_dirtied, |
83712358b
|
1675 |
period, |
ece13ac31
|
1676 1677 |
pause, start_time); |
499d05ecf
|
1678 |
__set_current_state(TASK_KILLABLE); |
d25105e89
|
1679 |
io_schedule_timeout(pause); |
87c6a9b25
|
1680 |
|
83712358b
|
1681 1682 |
current->dirty_paused_when = now + pause; current->nr_dirtied = 0; |
7ccb9ad53
|
1683 |
current->nr_dirtied_pause = nr_dirtied_pause; |
83712358b
|
1684 |
|
ffd1f609a
|
1685 |
/* |
2bc00aef0
|
1686 1687 |
* This is typically equal to (dirty < thresh) and can also * keep "1000+ dd on a slow USB stick" under control. |
ffd1f609a
|
1688 |
*/ |
1df647197
|
1689 |
if (task_ratelimit) |
ffd1f609a
|
1690 |
break; |
499d05ecf
|
1691 |
|
c5c6343c4
|
1692 1693 |
/* * In the case of an unresponding NFS server and the NFS dirty |
de1fff37b
|
1694 |
* pages exceeds dirty_thresh, give the other good wb's a pipe |
c5c6343c4
|
1695 1696 1697 1698 |
* to go through, so that tasks on them still remain responsive. * * In theory 1 page is enough to keep the comsumer-producer * pipe going: the flusher cleans 1 page => the task dirties 1 |
de1fff37b
|
1699 |
* more page. However wb_dirty has accounting errors. So use |
93f78d882
|
1700 |
* the larger and more IO friendly wb_stat_error. |
c5c6343c4
|
1701 |
*/ |
c2aa723a6
|
1702 |
if (sdtc->wb_dirty <= wb_stat_error(wb)) |
c5c6343c4
|
1703 |
break; |
499d05ecf
|
1704 1705 |
if (fatal_signal_pending(current)) break; |
1da177e4c
|
1706 |
} |
a88a341a7
|
1707 1708 |
if (!dirty_exceeded && wb->dirty_exceeded) wb->dirty_exceeded = 0; |
1da177e4c
|
1709 |
|
bc05873dc
|
1710 |
if (writeback_in_progress(wb)) |
5b0830cb9
|
1711 |
return; |
1da177e4c
|
1712 1713 1714 1715 1716 1717 1718 1719 1720 |
/* * In laptop mode, we wait until hitting the higher threshold before * starting background writeout, and then write out all the way down * to the lower threshold. So slow writers cause minimal disk activity. * * In normal mode, we start background writeout at the lower * background_thresh, to keep the amount of dirty memory low. */ |
143dfe861
|
1721 1722 |
if (laptop_mode) return; |
2bc00aef0
|
1723 |
if (nr_reclaimable > gdtc->bg_thresh) |
9ecf4866c
|
1724 |
wb_start_background_writeback(wb); |
1da177e4c
|
1725 |
} |
9d823e8f6
|
1726 |
static DEFINE_PER_CPU(int, bdp_ratelimits); |
245b2e70e
|
1727 |
|
54848d73f
|
1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 |
/* * Normal tasks are throttled by * loop { * dirty tsk->nr_dirtied_pause pages; * take a snap in balance_dirty_pages(); * } * However there is a worst case. If every task exit immediately when dirtied * (tsk->nr_dirtied_pause - 1) pages, balance_dirty_pages() will never be * called to throttle the page dirties. The solution is to save the not yet * throttled page dirties in dirty_throttle_leaks on task exit and charge them * randomly into the running tasks. This works well for the above worst case, * as the new task will pick up and accumulate the old task's leaked dirty * count and eventually get throttled. */ DEFINE_PER_CPU(int, dirty_throttle_leaks) = 0; |
1da177e4c
|
1743 |
/** |
d0e1d66b5
|
1744 |
* balance_dirty_pages_ratelimited - balance dirty memory state |
67be2dd1b
|
1745 |
* @mapping: address_space which was dirtied |
1da177e4c
|
1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 |
* * Processes which are dirtying memory should call in here once for each page * which was newly dirtied. The function will periodically check the system's * dirty state and will initiate writeback if needed. * * On really big machines, get_writeback_state is expensive, so try to avoid * calling it too often (ratelimiting). But once we're over the dirty memory * limit we decrease the ratelimiting by a lot, to prevent individual processes * from overshooting the limit by (ratelimit_pages) each. */ |
d0e1d66b5
|
1756 |
void balance_dirty_pages_ratelimited(struct address_space *mapping) |
1da177e4c
|
1757 |
{ |
dfb8ae567
|
1758 1759 1760 |
struct inode *inode = mapping->host; struct backing_dev_info *bdi = inode_to_bdi(inode); struct bdi_writeback *wb = NULL; |
9d823e8f6
|
1761 1762 |
int ratelimit; int *p; |
1da177e4c
|
1763 |
|
36715cef0
|
1764 1765 |
if (!bdi_cap_account_dirty(bdi)) return; |
dfb8ae567
|
1766 1767 1768 1769 |
if (inode_cgwb_enabled(inode)) wb = wb_get_create_current(bdi, GFP_KERNEL); if (!wb) wb = &bdi->wb; |
9d823e8f6
|
1770 |
ratelimit = current->nr_dirtied_pause; |
a88a341a7
|
1771 |
if (wb->dirty_exceeded) |
9d823e8f6
|
1772 |
ratelimit = min(ratelimit, 32 >> (PAGE_SHIFT - 10)); |
9d823e8f6
|
1773 |
preempt_disable(); |
1da177e4c
|
1774 |
/* |
9d823e8f6
|
1775 1776 1777 1778 |
* This prevents one CPU to accumulate too many dirtied pages without * calling into balance_dirty_pages(), which can happen when there are * 1000+ tasks, all of them start dirtying pages at exactly the same * time, hence all honoured too large initial task->nr_dirtied_pause. |
1da177e4c
|
1779 |
*/ |
7c8e0181e
|
1780 |
p = this_cpu_ptr(&bdp_ratelimits); |
9d823e8f6
|
1781 |
if (unlikely(current->nr_dirtied >= ratelimit)) |
fa5a734e4
|
1782 |
*p = 0; |
d3bc1fef9
|
1783 1784 1785 |
else if (unlikely(*p >= ratelimit_pages)) { *p = 0; ratelimit = 0; |
1da177e4c
|
1786 |
} |
54848d73f
|
1787 1788 1789 1790 1791 |
/* * Pick up the dirtied pages by the exited tasks. This avoids lots of * short-lived tasks (eg. gcc invocations in a kernel build) escaping * the dirty throttling and livelock other long-run dirtiers. */ |
7c8e0181e
|
1792 |
p = this_cpu_ptr(&dirty_throttle_leaks); |
54848d73f
|
1793 |
if (*p > 0 && current->nr_dirtied < ratelimit) { |
d0e1d66b5
|
1794 |
unsigned long nr_pages_dirtied; |
54848d73f
|
1795 1796 1797 |
nr_pages_dirtied = min(*p, ratelimit - current->nr_dirtied); *p -= nr_pages_dirtied; current->nr_dirtied += nr_pages_dirtied; |
1da177e4c
|
1798 |
} |
fa5a734e4
|
1799 |
preempt_enable(); |
9d823e8f6
|
1800 1801 |
if (unlikely(current->nr_dirtied >= ratelimit)) |
dfb8ae567
|
1802 1803 1804 |
balance_dirty_pages(mapping, wb, current->nr_dirtied); wb_put(wb); |
1da177e4c
|
1805 |
} |
d0e1d66b5
|
1806 |
EXPORT_SYMBOL(balance_dirty_pages_ratelimited); |
1da177e4c
|
1807 |
|
aa661bbe1
|
1808 1809 1810 1811 1812 1813 1814 1815 1816 |
/** * wb_over_bg_thresh - does @wb need to be written back? * @wb: bdi_writeback of interest * * Determines whether background writeback should keep writing @wb or it's * clean enough. Returns %true if writeback should continue. */ bool wb_over_bg_thresh(struct bdi_writeback *wb) { |
947e9762a
|
1817 |
struct dirty_throttle_control gdtc_stor = { GDTC_INIT(wb) }; |
c2aa723a6
|
1818 |
struct dirty_throttle_control mdtc_stor = { MDTC_INIT(wb, &gdtc_stor) }; |
947e9762a
|
1819 |
struct dirty_throttle_control * const gdtc = &gdtc_stor; |
c2aa723a6
|
1820 1821 |
struct dirty_throttle_control * const mdtc = mdtc_valid(&mdtc_stor) ? &mdtc_stor : NULL; |
aa661bbe1
|
1822 |
|
947e9762a
|
1823 1824 1825 1826 1827 1828 1829 1830 |
/* * Similar to balance_dirty_pages() but ignores pages being written * as we're trying to decide whether to put more under writeback. */ gdtc->avail = global_dirtyable_memory(); gdtc->dirty = global_page_state(NR_FILE_DIRTY) + global_page_state(NR_UNSTABLE_NFS); domain_dirty_limits(gdtc); |
aa661bbe1
|
1831 |
|
947e9762a
|
1832 |
if (gdtc->dirty > gdtc->bg_thresh) |
aa661bbe1
|
1833 |
return true; |
74d369443
|
1834 1835 |
if (wb_stat(wb, WB_RECLAIMABLE) > wb_calc_thresh(gdtc->wb, gdtc->bg_thresh)) |
aa661bbe1
|
1836 |
return true; |
c2aa723a6
|
1837 |
if (mdtc) { |
c5edf9cdc
|
1838 |
unsigned long filepages, headroom, writeback; |
c2aa723a6
|
1839 |
|
c5edf9cdc
|
1840 1841 1842 |
mem_cgroup_wb_stats(wb, &filepages, &headroom, &mdtc->dirty, &writeback); mdtc_calc_avail(mdtc, filepages, headroom); |
c2aa723a6
|
1843 1844 1845 1846 |
domain_dirty_limits(mdtc); /* ditto, ignore writeback */ if (mdtc->dirty > mdtc->bg_thresh) return true; |
74d369443
|
1847 1848 |
if (wb_stat(wb, WB_RECLAIMABLE) > wb_calc_thresh(mdtc->wb, mdtc->bg_thresh)) |
c2aa723a6
|
1849 1850 |
return true; } |
aa661bbe1
|
1851 1852 |
return false; } |
232ea4d69
|
1853 |
void throttle_vm_writeout(gfp_t gfp_mask) |
1da177e4c
|
1854 |
{ |
364aeb284
|
1855 1856 |
unsigned long background_thresh; unsigned long dirty_thresh; |
1da177e4c
|
1857 1858 |
for ( ; ; ) { |
16c4042f0
|
1859 |
global_dirty_limits(&background_thresh, &dirty_thresh); |
c7981433e
|
1860 |
dirty_thresh = hard_dirty_limit(&global_wb_domain, dirty_thresh); |
1da177e4c
|
1861 1862 1863 1864 1865 1866 |
/* * Boost the allowable dirty threshold a bit for page * allocators so they don't get DoS'ed by heavy writers */ dirty_thresh += dirty_thresh / 10; /* wheeee... */ |
c24f21bda
|
1867 1868 1869 |
if (global_page_state(NR_UNSTABLE_NFS) + global_page_state(NR_WRITEBACK) <= dirty_thresh) break; |
8aa7e847d
|
1870 |
congestion_wait(BLK_RW_ASYNC, HZ/10); |
369f2389e
|
1871 1872 1873 1874 1875 1876 1877 1878 |
/* * The caller might hold locks which can prevent IO completion * or progress in the filesystem. So we cannot just sit here * waiting for IO to complete. */ if ((gfp_mask & (__GFP_FS|__GFP_IO)) != (__GFP_FS|__GFP_IO)) break; |
1da177e4c
|
1879 1880 |
} } |
1da177e4c
|
1881 |
/* |
1da177e4c
|
1882 1883 |
* sysctl handler for /proc/sys/vm/dirty_writeback_centisecs */ |
cccad5b98
|
1884 |
int dirty_writeback_centisecs_handler(struct ctl_table *table, int write, |
8d65af789
|
1885 |
void __user *buffer, size_t *length, loff_t *ppos) |
1da177e4c
|
1886 |
{ |
8d65af789
|
1887 |
proc_dointvec(table, write, buffer, length, ppos); |
1da177e4c
|
1888 1889 |
return 0; } |
c2c4986ed
|
1890 |
#ifdef CONFIG_BLOCK |
31373d09d
|
1891 |
void laptop_mode_timer_fn(unsigned long data) |
1da177e4c
|
1892 |
{ |
31373d09d
|
1893 1894 1895 |
struct request_queue *q = (struct request_queue *)data; int nr_pages = global_page_state(NR_FILE_DIRTY) + global_page_state(NR_UNSTABLE_NFS); |
a06fd6b10
|
1896 |
struct bdi_writeback *wb; |
1da177e4c
|
1897 |
|
31373d09d
|
1898 1899 1900 1901 |
/* * We want to write everything out, not just down to the dirty * threshold */ |
a06fd6b10
|
1902 1903 |
if (!bdi_has_dirty_io(&q->backing_dev_info)) return; |
9ad18ab93
|
1904 |
rcu_read_lock(); |
b817525a4
|
1905 |
list_for_each_entry_rcu(wb, &q->backing_dev_info.wb_list, bdi_node) |
a06fd6b10
|
1906 1907 1908 |
if (wb_has_dirty_io(wb)) wb_start_writeback(wb, nr_pages, true, WB_REASON_LAPTOP_TIMER); |
9ad18ab93
|
1909 |
rcu_read_unlock(); |
1da177e4c
|
1910 1911 1912 1913 1914 1915 1916 |
} /* * We've spun up the disk and we're in laptop mode: schedule writeback * of all dirty data a few seconds from now. If the flush is already scheduled * then push it back - the user is still using the disk. */ |
31373d09d
|
1917 |
void laptop_io_completion(struct backing_dev_info *info) |
1da177e4c
|
1918 |
{ |
31373d09d
|
1919 |
mod_timer(&info->laptop_mode_wb_timer, jiffies + laptop_mode); |
1da177e4c
|
1920 1921 1922 1923 1924 1925 1926 1927 1928 |
} /* * We're in laptop mode and we've just synced. The sync's writes will have * caused another writeback to be scheduled by laptop_io_completion. * Nothing needs to be written back anymore, so we unschedule the writeback. */ void laptop_sync_completion(void) { |
31373d09d
|
1929 1930 1931 1932 1933 1934 1935 1936 |
struct backing_dev_info *bdi; rcu_read_lock(); list_for_each_entry_rcu(bdi, &bdi_list, bdi_list) del_timer(&bdi->laptop_mode_wb_timer); rcu_read_unlock(); |
1da177e4c
|
1937 |
} |
c2c4986ed
|
1938 |
#endif |
1da177e4c
|
1939 1940 1941 1942 1943 1944 1945 1946 1947 |
/* * If ratelimit_pages is too high then we can get into dirty-data overload * if a large number of processes all perform writes at the same time. * If it is too low then SMP machines will call the (expensive) * get_writeback_state too often. * * Here we set ratelimit_pages to a level which ensures that when all CPUs are * dirtying in parallel, we cannot go more than 3% (1/32) over the dirty memory |
9d823e8f6
|
1948 |
* thresholds. |
1da177e4c
|
1949 |
*/ |
2d1d43f6a
|
1950 |
void writeback_set_ratelimit(void) |
1da177e4c
|
1951 |
{ |
dcc25ae76
|
1952 |
struct wb_domain *dom = &global_wb_domain; |
9d823e8f6
|
1953 1954 |
unsigned long background_thresh; unsigned long dirty_thresh; |
dcc25ae76
|
1955 |
|
9d823e8f6
|
1956 |
global_dirty_limits(&background_thresh, &dirty_thresh); |
dcc25ae76
|
1957 |
dom->dirty_limit = dirty_thresh; |
9d823e8f6
|
1958 |
ratelimit_pages = dirty_thresh / (num_online_cpus() * 32); |
1da177e4c
|
1959 1960 |
if (ratelimit_pages < 16) ratelimit_pages = 16; |
1da177e4c
|
1961 |
} |
0db0628d9
|
1962 |
static int |
2f60d628f
|
1963 1964 |
ratelimit_handler(struct notifier_block *self, unsigned long action, void *hcpu) |
1da177e4c
|
1965 |
{ |
2f60d628f
|
1966 1967 1968 1969 1970 1971 1972 1973 1974 |
switch (action & ~CPU_TASKS_FROZEN) { case CPU_ONLINE: case CPU_DEAD: writeback_set_ratelimit(); return NOTIFY_OK; default: return NOTIFY_DONE; } |
1da177e4c
|
1975 |
} |
0db0628d9
|
1976 |
static struct notifier_block ratelimit_nb = { |
1da177e4c
|
1977 1978 1979 1980 1981 |
.notifier_call = ratelimit_handler, .next = NULL, }; /* |
dc6e29da9
|
1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 |
* Called early on to tune the page writeback dirty limits. * * We used to scale dirty pages according to how total memory * related to pages that could be allocated for buffers (by * comparing nr_free_buffer_pages() to vm_total_pages. * * However, that was when we used "dirty_ratio" to scale with * all memory, and we don't do that any more. "dirty_ratio" * is now applied to total non-HIGHPAGE memory (by subtracting * totalhigh_pages from vm_total_pages), and as such we can't * get into the old insane situation any more where we had * large amounts of dirty pages compared to a small amount of * non-HIGHMEM memory. * * But we might still want to scale the dirty_ratio by how * much memory the box has.. |
1da177e4c
|
1998 1999 2000 |
*/ void __init page_writeback_init(void) { |
a50fcb512
|
2001 |
BUG_ON(wb_domain_init(&global_wb_domain, GFP_KERNEL)); |
2d1d43f6a
|
2002 |
writeback_set_ratelimit(); |
1da177e4c
|
2003 2004 |
register_cpu_notifier(&ratelimit_nb); } |
811d736f9
|
2005 |
/** |
f446daaea
|
2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 |
* tag_pages_for_writeback - tag pages to be written by write_cache_pages * @mapping: address space structure to write * @start: starting page index * @end: ending page index (inclusive) * * This function scans the page range from @start to @end (inclusive) and tags * all pages that have DIRTY tag set with a special TOWRITE tag. The idea is * that write_cache_pages (or whoever calls this function) will then use * TOWRITE tag to identify pages eligible for writeback. This mechanism is * used to avoid livelocking of writeback by a process steadily creating new * dirty pages in the file (thus it is important for this function to be quick * so that it can tag pages faster than a dirtying process can create them). */ /* * We tag pages in batches of WRITEBACK_TAG_BATCH to reduce tree_lock latency. */ |
f446daaea
|
2022 2023 2024 |
void tag_pages_for_writeback(struct address_space *mapping, pgoff_t start, pgoff_t end) { |
3c111a071
|
2025 |
#define WRITEBACK_TAG_BATCH 4096 |
f446daaea
|
2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 |
unsigned long tagged; do { spin_lock_irq(&mapping->tree_lock); tagged = radix_tree_range_tag_if_tagged(&mapping->page_tree, &start, end, WRITEBACK_TAG_BATCH, PAGECACHE_TAG_DIRTY, PAGECACHE_TAG_TOWRITE); spin_unlock_irq(&mapping->tree_lock); WARN_ON_ONCE(tagged > WRITEBACK_TAG_BATCH); cond_resched(); |
d5ed3a4af
|
2036 2037 |
/* We check 'start' to handle wrapping when end == ~0UL */ } while (tagged >= WRITEBACK_TAG_BATCH && start); |
f446daaea
|
2038 2039 2040 2041 |
} EXPORT_SYMBOL(tag_pages_for_writeback); /** |
0ea971801
|
2042 |
* write_cache_pages - walk the list of dirty pages of the given address space and write all of them. |
811d736f9
|
2043 2044 |
* @mapping: address space structure to write * @wbc: subtract the number of written pages from *@wbc->nr_to_write |
0ea971801
|
2045 2046 |
* @writepage: function called for each page * @data: data passed to writepage function |
811d736f9
|
2047 |
* |
0ea971801
|
2048 |
* If a page is already under I/O, write_cache_pages() skips it, even |
811d736f9
|
2049 2050 2051 2052 2053 2054 |
* if it's dirty. This is desirable behaviour for memory-cleaning writeback, * but it is INCORRECT for data-integrity system calls such as fsync(). fsync() * and msync() need to guarantee that all the data which was dirty at the time * the call was made get new I/O started against them. If wbc->sync_mode is * WB_SYNC_ALL then we were called for data integrity and we must wait for * existing IO to complete. |
f446daaea
|
2055 2056 2057 2058 2059 2060 2061 |
* * To avoid livelocks (when other process dirties new pages), we first tag * pages which should be written back with TOWRITE tag and only then start * writing them. For data-integrity sync we have to be careful so that we do * not miss some pages (e.g., because some other process has cleared TOWRITE * tag we set). The rule we follow is that TOWRITE tag can be cleared only * by the process clearing the DIRTY tag (and submitting the page for IO). |
811d736f9
|
2062 |
*/ |
0ea971801
|
2063 2064 2065 |
int write_cache_pages(struct address_space *mapping, struct writeback_control *wbc, writepage_t writepage, void *data) |
811d736f9
|
2066 |
{ |
811d736f9
|
2067 2068 |
int ret = 0; int done = 0; |
811d736f9
|
2069 2070 |
struct pagevec pvec; int nr_pages; |
31a12666d
|
2071 |
pgoff_t uninitialized_var(writeback_index); |
811d736f9
|
2072 2073 |
pgoff_t index; pgoff_t end; /* Inclusive */ |
bd19e012f
|
2074 |
pgoff_t done_index; |
31a12666d
|
2075 |
int cycled; |
811d736f9
|
2076 |
int range_whole = 0; |
f446daaea
|
2077 |
int tag; |
811d736f9
|
2078 |
|
811d736f9
|
2079 2080 |
pagevec_init(&pvec, 0); if (wbc->range_cyclic) { |
31a12666d
|
2081 2082 2083 2084 2085 2086 |
writeback_index = mapping->writeback_index; /* prev offset */ index = writeback_index; if (index == 0) cycled = 1; else cycled = 0; |
811d736f9
|
2087 2088 |
end = -1; } else { |
09cbfeaf1
|
2089 2090 |
index = wbc->range_start >> PAGE_SHIFT; end = wbc->range_end >> PAGE_SHIFT; |
811d736f9
|
2091 2092 |
if (wbc->range_start == 0 && wbc->range_end == LLONG_MAX) range_whole = 1; |
31a12666d
|
2093 |
cycled = 1; /* ignore range_cyclic tests */ |
811d736f9
|
2094 |
} |
6e6938b6d
|
2095 |
if (wbc->sync_mode == WB_SYNC_ALL || wbc->tagged_writepages) |
f446daaea
|
2096 2097 2098 |
tag = PAGECACHE_TAG_TOWRITE; else tag = PAGECACHE_TAG_DIRTY; |
811d736f9
|
2099 |
retry: |
6e6938b6d
|
2100 |
if (wbc->sync_mode == WB_SYNC_ALL || wbc->tagged_writepages) |
f446daaea
|
2101 |
tag_pages_for_writeback(mapping, index, end); |
bd19e012f
|
2102 |
done_index = index; |
5a3d5c981
|
2103 2104 |
while (!done && (index <= end)) { int i; |
f446daaea
|
2105 |
nr_pages = pagevec_lookup_tag(&pvec, mapping, &index, tag, |
5a3d5c981
|
2106 2107 2108 |
min(end - index, (pgoff_t)PAGEVEC_SIZE-1) + 1); if (nr_pages == 0) break; |
811d736f9
|
2109 |
|
811d736f9
|
2110 2111 2112 2113 |
for (i = 0; i < nr_pages; i++) { struct page *page = pvec.pages[i]; /* |
d5482cdf8
|
2114 2115 2116 2117 2118 |
* At this point, the page may be truncated or * invalidated (changing page->mapping to NULL), or * even swizzled back from swapper_space to tmpfs file * mapping. However, page->index will not change * because we have a reference on the page. |
811d736f9
|
2119 |
*/ |
d5482cdf8
|
2120 2121 2122 2123 2124 2125 2126 2127 |
if (page->index > end) { /* * can't be range_cyclic (1st pass) because * end == -1 in that case. */ done = 1; break; } |
cf15b07cf
|
2128 |
done_index = page->index; |
d5482cdf8
|
2129 |
|
811d736f9
|
2130 |
lock_page(page); |
5a3d5c981
|
2131 2132 2133 2134 2135 2136 2137 2138 |
/* * Page truncated or invalidated. We can freely skip it * then, even for data integrity operations: the page * has disappeared concurrently, so there could be no * real expectation of this data interity operation * even if there is now a new, dirty page at the same * pagecache address. */ |
811d736f9
|
2139 |
if (unlikely(page->mapping != mapping)) { |
5a3d5c981
|
2140 |
continue_unlock: |
811d736f9
|
2141 2142 2143 |
unlock_page(page); continue; } |
515f4a037
|
2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 |
if (!PageDirty(page)) { /* someone wrote it for us */ goto continue_unlock; } if (PageWriteback(page)) { if (wbc->sync_mode != WB_SYNC_NONE) wait_on_page_writeback(page); else goto continue_unlock; } |
811d736f9
|
2155 |
|
515f4a037
|
2156 2157 |
BUG_ON(PageWriteback(page)); if (!clear_page_dirty_for_io(page)) |
5a3d5c981
|
2158 |
goto continue_unlock; |
811d736f9
|
2159 |
|
de1414a65
|
2160 |
trace_wbc_writepage(wbc, inode_to_bdi(mapping->host)); |
0ea971801
|
2161 |
ret = (*writepage)(page, wbc, data); |
00266770b
|
2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 |
if (unlikely(ret)) { if (ret == AOP_WRITEPAGE_ACTIVATE) { unlock_page(page); ret = 0; } else { /* * done_index is set past this page, * so media errors will not choke * background writeout for the entire * file. This has consequences for * range_cyclic semantics (ie. it may * not be suitable for data integrity * writeout). */ |
cf15b07cf
|
2176 |
done_index = page->index + 1; |
00266770b
|
2177 2178 2179 |
done = 1; break; } |
0b5649278
|
2180 |
} |
00266770b
|
2181 |
|
546a19242
|
2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 |
/* * We stop writing back only if we are not doing * integrity sync. In case of integrity sync we have to * keep going until we have written all the pages * we tagged for writeback prior to entering this loop. */ if (--wbc->nr_to_write <= 0 && wbc->sync_mode == WB_SYNC_NONE) { done = 1; break; |
05fe478dd
|
2192 |
} |
811d736f9
|
2193 2194 2195 2196 |
} pagevec_release(&pvec); cond_resched(); } |
3a4c6800f
|
2197 |
if (!cycled && !done) { |
811d736f9
|
2198 |
/* |
31a12666d
|
2199 |
* range_cyclic: |
811d736f9
|
2200 2201 2202 |
* We hit the last page and there is more work to be done: wrap * back to the start of the file */ |
31a12666d
|
2203 |
cycled = 1; |
811d736f9
|
2204 |
index = 0; |
31a12666d
|
2205 |
end = writeback_index - 1; |
811d736f9
|
2206 2207 |
goto retry; } |
0b5649278
|
2208 2209 |
if (wbc->range_cyclic || (range_whole && wbc->nr_to_write > 0)) mapping->writeback_index = done_index; |
06d6cf695
|
2210 |
|
811d736f9
|
2211 2212 |
return ret; } |
0ea971801
|
2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 |
EXPORT_SYMBOL(write_cache_pages); /* * Function used by generic_writepages to call the real writepage * function and set the mapping flags on error */ static int __writepage(struct page *page, struct writeback_control *wbc, void *data) { struct address_space *mapping = data; int ret = mapping->a_ops->writepage(page, wbc); mapping_set_error(mapping, ret); return ret; } /** * generic_writepages - walk the list of dirty pages of the given address space and writepage() all of them. * @mapping: address space structure to write * @wbc: subtract the number of written pages from *@wbc->nr_to_write * * This is a library function, which implements the writepages() * address_space_operation. */ int generic_writepages(struct address_space *mapping, struct writeback_control *wbc) { |
9b6096a65
|
2239 2240 |
struct blk_plug plug; int ret; |
0ea971801
|
2241 2242 2243 |
/* deal with chardevs and other special file */ if (!mapping->a_ops->writepage) return 0; |
9b6096a65
|
2244 2245 2246 2247 |
blk_start_plug(&plug); ret = write_cache_pages(mapping, wbc, __writepage, mapping); blk_finish_plug(&plug); return ret; |
0ea971801
|
2248 |
} |
811d736f9
|
2249 2250 |
EXPORT_SYMBOL(generic_writepages); |
1da177e4c
|
2251 2252 |
int do_writepages(struct address_space *mapping, struct writeback_control *wbc) { |
22905f775
|
2253 |
int ret; |
1da177e4c
|
2254 2255 2256 |
if (wbc->nr_to_write <= 0) return 0; if (mapping->a_ops->writepages) |
d08b3851d
|
2257 |
ret = mapping->a_ops->writepages(mapping, wbc); |
22905f775
|
2258 2259 |
else ret = generic_writepages(mapping, wbc); |
22905f775
|
2260 |
return ret; |
1da177e4c
|
2261 2262 2263 2264 |
} /** * write_one_page - write out a single page and optionally wait on I/O |
67be2dd1b
|
2265 2266 |
* @page: the page to write * @wait: if true, wait on writeout |
1da177e4c
|
2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 |
* * The page must be locked by the caller and will be unlocked upon return. * * write_one_page() returns a negative error code if I/O failed. */ int write_one_page(struct page *page, int wait) { struct address_space *mapping = page->mapping; int ret = 0; struct writeback_control wbc = { .sync_mode = WB_SYNC_ALL, .nr_to_write = 1, }; BUG_ON(!PageLocked(page)); if (wait) wait_on_page_writeback(page); if (clear_page_dirty_for_io(page)) { |
09cbfeaf1
|
2287 |
get_page(page); |
1da177e4c
|
2288 2289 2290 2291 2292 2293 |
ret = mapping->a_ops->writepage(page, &wbc); if (ret == 0 && wait) { wait_on_page_writeback(page); if (PageError(page)) ret = -EIO; } |
09cbfeaf1
|
2294 |
put_page(page); |
1da177e4c
|
2295 2296 2297 2298 2299 2300 2301 2302 |
} else { unlock_page(page); } return ret; } EXPORT_SYMBOL(write_one_page); /* |
767193253
|
2303 2304 2305 2306 2307 |
* For address_spaces which do not use buffers nor write back. */ int __set_page_dirty_no_writeback(struct page *page) { if (!PageDirty(page)) |
c3f0da631
|
2308 |
return !TestSetPageDirty(page); |
767193253
|
2309 2310 2311 2312 |
return 0; } /* |
e3a7cca1e
|
2313 |
* Helper function for set_page_dirty family. |
c4843a759
|
2314 |
* |
81f8c3a46
|
2315 |
* Caller must hold lock_page_memcg(). |
c4843a759
|
2316 |
* |
e3a7cca1e
|
2317 2318 |
* NOTE: This relies on being atomic wrt interrupts. */ |
62cccb8c8
|
2319 |
void account_page_dirtied(struct page *page, struct address_space *mapping) |
e3a7cca1e
|
2320 |
{ |
52ebea749
|
2321 |
struct inode *inode = mapping->host; |
9fb0a7da0
|
2322 |
trace_writeback_dirty_page(page, mapping); |
e3a7cca1e
|
2323 |
if (mapping_cap_account_dirty(mapping)) { |
52ebea749
|
2324 |
struct bdi_writeback *wb; |
de1414a65
|
2325 |
|
52ebea749
|
2326 2327 |
inode_attach_wb(inode, page); wb = inode_to_wb(inode); |
de1414a65
|
2328 |
|
62cccb8c8
|
2329 |
mem_cgroup_inc_page_stat(page, MEM_CGROUP_STAT_DIRTY); |
e3a7cca1e
|
2330 |
__inc_zone_page_state(page, NR_FILE_DIRTY); |
ea941f0e2
|
2331 |
__inc_zone_page_state(page, NR_DIRTIED); |
52ebea749
|
2332 2333 |
__inc_wb_stat(wb, WB_RECLAIMABLE); __inc_wb_stat(wb, WB_DIRTIED); |
09cbfeaf1
|
2334 |
task_io_account_write(PAGE_SIZE); |
d3bc1fef9
|
2335 2336 |
current->nr_dirtied++; this_cpu_inc(bdp_ratelimits); |
e3a7cca1e
|
2337 2338 |
} } |
679ceace8
|
2339 |
EXPORT_SYMBOL(account_page_dirtied); |
e3a7cca1e
|
2340 2341 |
/* |
b9ea25152
|
2342 2343 |
* Helper function for deaccounting dirty page without writeback. * |
81f8c3a46
|
2344 |
* Caller must hold lock_page_memcg(). |
b9ea25152
|
2345 |
*/ |
c4843a759
|
2346 |
void account_page_cleaned(struct page *page, struct address_space *mapping, |
62cccb8c8
|
2347 |
struct bdi_writeback *wb) |
b9ea25152
|
2348 2349 |
{ if (mapping_cap_account_dirty(mapping)) { |
62cccb8c8
|
2350 |
mem_cgroup_dec_page_stat(page, MEM_CGROUP_STAT_DIRTY); |
b9ea25152
|
2351 |
dec_zone_page_state(page, NR_FILE_DIRTY); |
682aa8e1a
|
2352 |
dec_wb_stat(wb, WB_RECLAIMABLE); |
09cbfeaf1
|
2353 |
task_io_account_cancelled_write(PAGE_SIZE); |
b9ea25152
|
2354 2355 |
} } |
b9ea25152
|
2356 2357 |
/* |
1da177e4c
|
2358 2359 2360 2361 2362 2363 2364 |
* For address_spaces which do not use buffers. Just tag the page as dirty in * its radix tree. * * This is also used when a single buffer is being dirtied: we want to set the * page dirty in that case, but not all the buffers. This is a "bottom-up" * dirtying, whereas __set_page_dirty_buffers() is a "top-down" dirtying. * |
2d6d7f982
|
2365 2366 2367 |
* The caller must ensure this doesn't race with truncation. Most will simply * hold the page lock, but e.g. zap_pte_range() calls with the page mapped and * the pte lock held, which also locks out truncation. |
1da177e4c
|
2368 2369 2370 |
*/ int __set_page_dirty_nobuffers(struct page *page) { |
62cccb8c8
|
2371 |
lock_page_memcg(page); |
1da177e4c
|
2372 2373 |
if (!TestSetPageDirty(page)) { struct address_space *mapping = page_mapping(page); |
a85d9df1e
|
2374 |
unsigned long flags; |
1da177e4c
|
2375 |
|
c4843a759
|
2376 |
if (!mapping) { |
62cccb8c8
|
2377 |
unlock_page_memcg(page); |
8c08540f8
|
2378 |
return 1; |
c4843a759
|
2379 |
} |
8c08540f8
|
2380 |
|
a85d9df1e
|
2381 |
spin_lock_irqsave(&mapping->tree_lock, flags); |
2d6d7f982
|
2382 2383 |
BUG_ON(page_mapping(page) != mapping); WARN_ON_ONCE(!PagePrivate(page) && !PageUptodate(page)); |
62cccb8c8
|
2384 |
account_page_dirtied(page, mapping); |
2d6d7f982
|
2385 2386 |
radix_tree_tag_set(&mapping->page_tree, page_index(page), PAGECACHE_TAG_DIRTY); |
a85d9df1e
|
2387 |
spin_unlock_irqrestore(&mapping->tree_lock, flags); |
62cccb8c8
|
2388 |
unlock_page_memcg(page); |
c4843a759
|
2389 |
|
8c08540f8
|
2390 2391 2392 |
if (mapping->host) { /* !PageAnon && !swapper_space */ __mark_inode_dirty(mapping->host, I_DIRTY_PAGES); |
1da177e4c
|
2393 |
} |
4741c9fd3
|
2394 |
return 1; |
1da177e4c
|
2395 |
} |
62cccb8c8
|
2396 |
unlock_page_memcg(page); |
4741c9fd3
|
2397 |
return 0; |
1da177e4c
|
2398 2399 2400 2401 |
} EXPORT_SYMBOL(__set_page_dirty_nobuffers); /* |
2f800fbd7
|
2402 2403 2404 2405 2406 2407 2408 2409 2410 |
* Call this whenever redirtying a page, to de-account the dirty counters * (NR_DIRTIED, BDI_DIRTIED, tsk->nr_dirtied), so that they match the written * counters (NR_WRITTEN, BDI_WRITTEN) in long term. The mismatches will lead to * systematic errors in balanced_dirty_ratelimit and the dirty pages position * control. */ void account_page_redirty(struct page *page) { struct address_space *mapping = page->mapping; |
910181343
|
2411 |
|
2f800fbd7
|
2412 |
if (mapping && mapping_cap_account_dirty(mapping)) { |
682aa8e1a
|
2413 2414 2415 |
struct inode *inode = mapping->host; struct bdi_writeback *wb; bool locked; |
910181343
|
2416 |
|
682aa8e1a
|
2417 |
wb = unlocked_inode_to_wb_begin(inode, &locked); |
2f800fbd7
|
2418 2419 |
current->nr_dirtied--; dec_zone_page_state(page, NR_DIRTIED); |
910181343
|
2420 |
dec_wb_stat(wb, WB_DIRTIED); |
682aa8e1a
|
2421 |
unlocked_inode_to_wb_end(inode, locked); |
2f800fbd7
|
2422 2423 2424 2425 2426 |
} } EXPORT_SYMBOL(account_page_redirty); /* |
1da177e4c
|
2427 2428 2429 2430 2431 2432 |
* When a writepage implementation decides that it doesn't want to write this * page for some reason, it should redirty the locked page via * redirty_page_for_writepage() and it should then unlock the page and return 0 */ int redirty_page_for_writepage(struct writeback_control *wbc, struct page *page) { |
8d38633c3
|
2433 |
int ret; |
1da177e4c
|
2434 |
wbc->pages_skipped++; |
8d38633c3
|
2435 |
ret = __set_page_dirty_nobuffers(page); |
2f800fbd7
|
2436 |
account_page_redirty(page); |
8d38633c3
|
2437 |
return ret; |
1da177e4c
|
2438 2439 2440 2441 |
} EXPORT_SYMBOL(redirty_page_for_writepage); /* |
6746aff74
|
2442 2443 2444 2445 2446 2447 2448 |
* Dirty a page. * * For pages with a mapping this should be done under the page lock * for the benefit of asynchronous memory errors who prefer a consistent * dirty state. This rule can be broken in some special cases, * but should be better not to. * |
1da177e4c
|
2449 2450 2451 |
* If the mapping doesn't provide a set_page_dirty a_op, then * just fall through and assume that it wants buffer_heads. */ |
1cf6e7d83
|
2452 |
int set_page_dirty(struct page *page) |
1da177e4c
|
2453 2454 2455 2456 2457 |
{ struct address_space *mapping = page_mapping(page); if (likely(mapping)) { int (*spd)(struct page *) = mapping->a_ops->set_page_dirty; |
278df9f45
|
2458 2459 2460 2461 2462 2463 2464 2465 2466 2467 |
/* * readahead/lru_deactivate_page could remain * PG_readahead/PG_reclaim due to race with end_page_writeback * About readahead, if the page is written, the flags would be * reset. So no problem. * About lru_deactivate_page, if the page is redirty, the flag * will be reset. So no problem. but if the page is used by readahead * it will confuse readahead and make it restart the size rampup * process. But it's a trivial problem. */ |
a4bb3ecdc
|
2468 2469 |
if (PageReclaim(page)) ClearPageReclaim(page); |
9361401eb
|
2470 2471 2472 2473 2474 |
#ifdef CONFIG_BLOCK if (!spd) spd = __set_page_dirty_buffers; #endif return (*spd)(page); |
1da177e4c
|
2475 |
} |
4741c9fd3
|
2476 2477 2478 2479 |
if (!PageDirty(page)) { if (!TestSetPageDirty(page)) return 1; } |
1da177e4c
|
2480 2481 2482 2483 2484 2485 2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 |
return 0; } EXPORT_SYMBOL(set_page_dirty); /* * set_page_dirty() is racy if the caller has no reference against * page->mapping->host, and if the page is unlocked. This is because another * CPU could truncate the page off the mapping and then free the mapping. * * Usually, the page _is_ locked, or the caller is a user-space process which * holds a reference on the inode by having an open file. * * In other cases, the page should be locked before running set_page_dirty(). */ int set_page_dirty_lock(struct page *page) { int ret; |
7eaceacca
|
2497 |
lock_page(page); |
1da177e4c
|
2498 2499 2500 2501 2502 2503 2504 |
ret = set_page_dirty(page); unlock_page(page); return ret; } EXPORT_SYMBOL(set_page_dirty_lock); /* |
11f81becc
|
2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515 2516 2517 2518 |
* This cancels just the dirty bit on the kernel page itself, it does NOT * actually remove dirty bits on any mmap's that may be around. It also * leaves the page tagged dirty, so any sync activity will still find it on * the dirty lists, and in particular, clear_page_dirty_for_io() will still * look at the dirty bits in the VM. * * Doing this should *normally* only ever be done when a page is truncated, * and is not actually mapped anywhere at all. However, fs/buffer.c does * this when it notices that somebody has cleaned out all the buffers on a * page without actually doing it through the VM. Can you say "ext3 is * horribly ugly"? Thought you could. */ void cancel_dirty_page(struct page *page) { |
c4843a759
|
2519 2520 2521 |
struct address_space *mapping = page_mapping(page); if (mapping_cap_account_dirty(mapping)) { |
682aa8e1a
|
2522 2523 |
struct inode *inode = mapping->host; struct bdi_writeback *wb; |
682aa8e1a
|
2524 |
bool locked; |
c4843a759
|
2525 |
|
62cccb8c8
|
2526 |
lock_page_memcg(page); |
682aa8e1a
|
2527 |
wb = unlocked_inode_to_wb_begin(inode, &locked); |
c4843a759
|
2528 2529 |
if (TestClearPageDirty(page)) |
62cccb8c8
|
2530 |
account_page_cleaned(page, mapping, wb); |
c4843a759
|
2531 |
|
682aa8e1a
|
2532 |
unlocked_inode_to_wb_end(inode, locked); |
62cccb8c8
|
2533 |
unlock_page_memcg(page); |
c4843a759
|
2534 2535 2536 |
} else { ClearPageDirty(page); } |
11f81becc
|
2537 2538 2539 2540 |
} EXPORT_SYMBOL(cancel_dirty_page); /* |
1da177e4c
|
2541 2542 2543 2544 2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 |
* Clear a page's dirty flag, while caring for dirty memory accounting. * Returns true if the page was previously dirty. * * This is for preparing to put the page under writeout. We leave the page * tagged as dirty in the radix tree so that a concurrent write-for-sync * can discover it via a PAGECACHE_TAG_DIRTY walk. The ->writepage * implementation will run either set_page_writeback() or set_page_dirty(), * at which stage we bring the page's dirty flag and radix-tree dirty tag * back into sync. * * This incoherency between the page's dirty flag and radix-tree tag is * unfortunate, but it only exists while the page is locked. */ int clear_page_dirty_for_io(struct page *page) { struct address_space *mapping = page_mapping(page); |
c4843a759
|
2557 |
int ret = 0; |
1da177e4c
|
2558 |
|
79352894b
|
2559 |
BUG_ON(!PageLocked(page)); |
7658cc289
|
2560 |
if (mapping && mapping_cap_account_dirty(mapping)) { |
682aa8e1a
|
2561 2562 |
struct inode *inode = mapping->host; struct bdi_writeback *wb; |
682aa8e1a
|
2563 |
bool locked; |
7658cc289
|
2564 2565 2566 2567 2568 2569 2570 2571 2572 2573 2574 2575 2576 2577 2578 2579 2580 2581 2582 2583 2584 2585 2586 2587 |
/* * Yes, Virginia, this is indeed insane. * * We use this sequence to make sure that * (a) we account for dirty stats properly * (b) we tell the low-level filesystem to * mark the whole page dirty if it was * dirty in a pagetable. Only to then * (c) clean the page again and return 1 to * cause the writeback. * * This way we avoid all nasty races with the * dirty bit in multiple places and clearing * them concurrently from different threads. * * Note! Normally the "set_page_dirty(page)" * has no effect on the actual dirty bit - since * that will already usually be set. But we * need the side effects, and it can help us * avoid races. * * We basically use the page "master dirty bit" * as a serialization point for all the different * threads doing their things. |
7658cc289
|
2588 2589 2590 |
*/ if (page_mkclean(page)) set_page_dirty(page); |
79352894b
|
2591 2592 2593 |
/* * We carefully synchronise fault handlers against * installing a dirty pte and marking the page dirty |
2d6d7f982
|
2594 2595 2596 2597 |
* at this point. We do this by having them hold the * page lock while dirtying the page, and pages are * always locked coming in here, so we get the desired * exclusion. |
79352894b
|
2598 |
*/ |
682aa8e1a
|
2599 |
wb = unlocked_inode_to_wb_begin(inode, &locked); |
7658cc289
|
2600 |
if (TestClearPageDirty(page)) { |
62cccb8c8
|
2601 |
mem_cgroup_dec_page_stat(page, MEM_CGROUP_STAT_DIRTY); |
8c08540f8
|
2602 |
dec_zone_page_state(page, NR_FILE_DIRTY); |
682aa8e1a
|
2603 |
dec_wb_stat(wb, WB_RECLAIMABLE); |
c4843a759
|
2604 |
ret = 1; |
1da177e4c
|
2605 |
} |
682aa8e1a
|
2606 |
unlocked_inode_to_wb_end(inode, locked); |
c4843a759
|
2607 |
return ret; |
1da177e4c
|
2608 |
} |
7658cc289
|
2609 |
return TestClearPageDirty(page); |
1da177e4c
|
2610 |
} |
58bb01a9c
|
2611 |
EXPORT_SYMBOL(clear_page_dirty_for_io); |
1da177e4c
|
2612 2613 2614 2615 |
int test_clear_page_writeback(struct page *page) { struct address_space *mapping = page_mapping(page); |
d7365e783
|
2616 |
int ret; |
1da177e4c
|
2617 |
|
62cccb8c8
|
2618 |
lock_page_memcg(page); |
1da177e4c
|
2619 |
if (mapping) { |
910181343
|
2620 2621 |
struct inode *inode = mapping->host; struct backing_dev_info *bdi = inode_to_bdi(inode); |
1da177e4c
|
2622 |
unsigned long flags; |
19fd62312
|
2623 |
spin_lock_irqsave(&mapping->tree_lock, flags); |
1da177e4c
|
2624 |
ret = TestClearPageWriteback(page); |
69cb51d18
|
2625 |
if (ret) { |
1da177e4c
|
2626 2627 2628 |
radix_tree_tag_clear(&mapping->page_tree, page_index(page), PAGECACHE_TAG_WRITEBACK); |
e4ad08fe6
|
2629 |
if (bdi_cap_account_writeback(bdi)) { |
910181343
|
2630 2631 2632 2633 |
struct bdi_writeback *wb = inode_to_wb(inode); __dec_wb_stat(wb, WB_WRITEBACK); __wb_writeout_inc(wb); |
04fbfdc14
|
2634 |
} |
69cb51d18
|
2635 |
} |
19fd62312
|
2636 |
spin_unlock_irqrestore(&mapping->tree_lock, flags); |
1da177e4c
|
2637 2638 2639 |
} else { ret = TestClearPageWriteback(page); } |
99b12e3d8
|
2640 |
if (ret) { |
62cccb8c8
|
2641 |
mem_cgroup_dec_page_stat(page, MEM_CGROUP_STAT_WRITEBACK); |
d688abf50
|
2642 |
dec_zone_page_state(page, NR_WRITEBACK); |
99b12e3d8
|
2643 2644 |
inc_zone_page_state(page, NR_WRITTEN); } |
62cccb8c8
|
2645 |
unlock_page_memcg(page); |
1da177e4c
|
2646 2647 |
return ret; } |
1c8349a17
|
2648 |
int __test_set_page_writeback(struct page *page, bool keep_write) |
1da177e4c
|
2649 2650 |
{ struct address_space *mapping = page_mapping(page); |
d7365e783
|
2651 |
int ret; |
1da177e4c
|
2652 |
|
62cccb8c8
|
2653 |
lock_page_memcg(page); |
1da177e4c
|
2654 |
if (mapping) { |
910181343
|
2655 2656 |
struct inode *inode = mapping->host; struct backing_dev_info *bdi = inode_to_bdi(inode); |
1da177e4c
|
2657 |
unsigned long flags; |
19fd62312
|
2658 |
spin_lock_irqsave(&mapping->tree_lock, flags); |
1da177e4c
|
2659 |
ret = TestSetPageWriteback(page); |
69cb51d18
|
2660 |
if (!ret) { |
1da177e4c
|
2661 2662 2663 |
radix_tree_tag_set(&mapping->page_tree, page_index(page), PAGECACHE_TAG_WRITEBACK); |
e4ad08fe6
|
2664 |
if (bdi_cap_account_writeback(bdi)) |
910181343
|
2665 |
__inc_wb_stat(inode_to_wb(inode), WB_WRITEBACK); |
69cb51d18
|
2666 |
} |
1da177e4c
|
2667 2668 2669 2670 |
if (!PageDirty(page)) radix_tree_tag_clear(&mapping->page_tree, page_index(page), PAGECACHE_TAG_DIRTY); |
1c8349a17
|
2671 2672 2673 2674 |
if (!keep_write) radix_tree_tag_clear(&mapping->page_tree, page_index(page), PAGECACHE_TAG_TOWRITE); |
19fd62312
|
2675 |
spin_unlock_irqrestore(&mapping->tree_lock, flags); |
1da177e4c
|
2676 2677 2678 |
} else { ret = TestSetPageWriteback(page); } |
3a3c02ecf
|
2679 |
if (!ret) { |
62cccb8c8
|
2680 |
mem_cgroup_inc_page_stat(page, MEM_CGROUP_STAT_WRITEBACK); |
3a3c02ecf
|
2681 2682 |
inc_zone_page_state(page, NR_WRITEBACK); } |
62cccb8c8
|
2683 |
unlock_page_memcg(page); |
1da177e4c
|
2684 2685 2686 |
return ret; } |
1c8349a17
|
2687 |
EXPORT_SYMBOL(__test_set_page_writeback); |
1da177e4c
|
2688 2689 |
/* |
001281881
|
2690 |
* Return true if any of the pages in the mapping are marked with the |
1da177e4c
|
2691 2692 2693 2694 |
* passed tag. */ int mapping_tagged(struct address_space *mapping, int tag) { |
72c478321
|
2695 |
return radix_tree_tagged(&mapping->page_tree, tag); |
1da177e4c
|
2696 2697 |
} EXPORT_SYMBOL(mapping_tagged); |
1d1d1a767
|
2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 |
/** * wait_for_stable_page() - wait for writeback to finish, if necessary. * @page: The page to wait on. * * This function determines if the given page is related to a backing device * that requires page contents to be held stable during writeback. If so, then * it will wait for any pending writeback to complete. */ void wait_for_stable_page(struct page *page) { |
de1414a65
|
2709 2710 |
if (bdi_cap_stable_pages_required(inode_to_bdi(page->mapping->host))) wait_on_page_writeback(page); |
1d1d1a767
|
2711 2712 |
} EXPORT_SYMBOL_GPL(wait_for_stable_page); |