page-writeback.c 38.8 KB
Newer Older
L
Linus Torvalds 已提交
1
/*
2
 * mm/page-writeback.c
L
Linus Torvalds 已提交
3 4
 *
 * Copyright (C) 2002, Linus Torvalds.
P
Peter Zijlstra 已提交
5
 * Copyright (C) 2007 Red Hat, Inc., Peter Zijlstra <pzijlstr@redhat.com>
L
Linus Torvalds 已提交
6 7 8 9
 *
 * Contains functions related to writing back dirty pages at the
 * address_space level.
 *
10
 * 10Apr2002	Andrew Morton
L
Linus Torvalds 已提交
11 12 13 14 15 16 17 18 19 20 21 22 23 24
 *		Initial version
 */

#include <linux/kernel.h>
#include <linux/module.h>
#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>
25
#include <linux/task_io_accounting_ops.h>
L
Linus Torvalds 已提交
26 27
#include <linux/blkdev.h>
#include <linux/mpage.h>
28
#include <linux/rmap.h>
L
Linus Torvalds 已提交
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>
35
#include <linux/buffer_head.h>
36
#include <linux/pagevec.h>
37
#include <trace/events/writeback.h>
L
Linus Torvalds 已提交
38 39 40 41 42 43 44 45 46 47

/*
 * 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;

/*
 * When balance_dirty_pages decides that the caller needs to perform some
 * non-background writeback, this is how many pages it will attempt to write.
48
 * It should be somewhat larger than dirtied pages to ensure that reasonably
L
Linus Torvalds 已提交
49 50
 * large amounts of I/O are submitted.
 */
51
static inline long sync_writeback_pages(unsigned long dirtied)
L
Linus Torvalds 已提交
52
{
53 54 55 56
	if (dirtied < ratelimit_pages)
		dirtied = ratelimit_pages;

	return dirtied + dirtied / 2;
L
Linus Torvalds 已提交
57 58 59 60 61
}

/* The following parameters are exported via /proc/sys/vm */

/*
62
 * Start background writeback (via writeback threads) at this percentage
L
Linus Torvalds 已提交
63
 */
64
int dirty_background_ratio = 10;
L
Linus Torvalds 已提交
65

66 67 68 69 70 71
/*
 * 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;

72 73 74 75 76 77
/*
 * 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;

L
Linus Torvalds 已提交
78 79 80
/*
 * The generator of dirty data starts writeback at this percentage
 */
81
int vm_dirty_ratio = 20;
L
Linus Torvalds 已提交
82

83 84 85 86 87 88
/*
 * 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;

L
Linus Torvalds 已提交
89
/*
90
 * The interval between `kupdate'-style writebacks
L
Linus Torvalds 已提交
91
 */
92
unsigned int dirty_writeback_interval = 5 * 100; /* centiseconds */
L
Linus Torvalds 已提交
93 94

/*
95
 * The longest time for which data is allowed to remain dirty
L
Linus Torvalds 已提交
96
 */
97
unsigned int dirty_expire_interval = 30 * 100; /* centiseconds */
L
Linus Torvalds 已提交
98 99 100 101 102 103 104

/*
 * Flag that makes the machine dump writes/reads and block dirtyings.
 */
int block_dump;

/*
105 106
 * 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.
L
Linus Torvalds 已提交
107 108 109 110 111 112 113 114
 */
int laptop_mode;

EXPORT_SYMBOL(laptop_mode);

/* End of sysctl-exported parameters */


P
Peter Zijlstra 已提交
115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131
/*
 * Scale the writeback cache size proportional to the relative writeout speeds.
 *
 * We do this by keeping a floating proportion between BDIs, based on page
 * writeback completions [end_page_writeback()]. Those devices that write out
 * pages fastest will get the larger share, while the slower will get a smaller
 * share.
 *
 * We use page writeout completions because we are interested in getting rid of
 * dirty pages. Having them written out is the primary goal.
 *
 * We introduce a concept of time, a period over which we measure these events,
 * because demand can/will vary over time. The length of this period itself is
 * measured in page writeback completions.
 *
 */
static struct prop_descriptor vm_completions;
P
Peter Zijlstra 已提交
132
static struct prop_descriptor vm_dirties;
P
Peter Zijlstra 已提交
133 134 135 136 137 138 139 140 141 142

/*
 * couple the period to the dirty_ratio:
 *
 *   period/2 ~ roundup_pow_of_two(dirty limit)
 */
static int calc_period_shift(void)
{
	unsigned long dirty_total;

143 144 145 146 147
	if (vm_dirty_bytes)
		dirty_total = vm_dirty_bytes / PAGE_SIZE;
	else
		dirty_total = (vm_dirty_ratio * determine_dirtyable_memory()) /
				100;
P
Peter Zijlstra 已提交
148 149 150 151
	return 2 + ilog2(dirty_total - 1);
}

/*
152
 * update the period when the dirty threshold changes.
P
Peter Zijlstra 已提交
153
 */
154 155 156 157 158 159 160 161
static void update_completion_period(void)
{
	int shift = calc_period_shift();
	prop_change_shift(&vm_completions, shift);
	prop_change_shift(&vm_dirties, shift);
}

int dirty_background_ratio_handler(struct ctl_table *table, int write,
162
		void __user *buffer, size_t *lenp,
163 164 165 166
		loff_t *ppos)
{
	int ret;

167
	ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
168 169 170 171 172 173
	if (ret == 0 && write)
		dirty_background_bytes = 0;
	return ret;
}

int dirty_background_bytes_handler(struct ctl_table *table, int write,
174
		void __user *buffer, size_t *lenp,
175 176 177 178
		loff_t *ppos)
{
	int ret;

179
	ret = proc_doulongvec_minmax(table, write, buffer, lenp, ppos);
180 181 182 183 184
	if (ret == 0 && write)
		dirty_background_ratio = 0;
	return ret;
}

P
Peter Zijlstra 已提交
185
int dirty_ratio_handler(struct ctl_table *table, int write,
186
		void __user *buffer, size_t *lenp,
P
Peter Zijlstra 已提交
187 188 189
		loff_t *ppos)
{
	int old_ratio = vm_dirty_ratio;
190 191
	int ret;

192
	ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
P
Peter Zijlstra 已提交
193
	if (ret == 0 && write && vm_dirty_ratio != old_ratio) {
194 195 196 197 198 199 200 201
		update_completion_period();
		vm_dirty_bytes = 0;
	}
	return ret;
}


int dirty_bytes_handler(struct ctl_table *table, int write,
202
		void __user *buffer, size_t *lenp,
203 204
		loff_t *ppos)
{
205
	unsigned long old_bytes = vm_dirty_bytes;
206 207
	int ret;

208
	ret = proc_doulongvec_minmax(table, write, buffer, lenp, ppos);
209 210 211
	if (ret == 0 && write && vm_dirty_bytes != old_bytes) {
		update_completion_period();
		vm_dirty_ratio = 0;
P
Peter Zijlstra 已提交
212 213 214 215 216 217 218 219 220 221
	}
	return ret;
}

/*
 * Increment the BDI's writeout completion count and the global writeout
 * completion count. Called from test_clear_page_writeback().
 */
static inline void __bdi_writeout_inc(struct backing_dev_info *bdi)
{
222 223
	__prop_inc_percpu_max(&vm_completions, &bdi->completions,
			      bdi->max_prop_frac);
P
Peter Zijlstra 已提交
224 225
}

226 227 228 229 230 231 232 233 234 235
void bdi_writeout_inc(struct backing_dev_info *bdi)
{
	unsigned long flags;

	local_irq_save(flags);
	__bdi_writeout_inc(bdi);
	local_irq_restore(flags);
}
EXPORT_SYMBOL_GPL(bdi_writeout_inc);

N
Nick Piggin 已提交
236
void task_dirty_inc(struct task_struct *tsk)
P
Peter Zijlstra 已提交
237 238 239 240
{
	prop_inc_single(&vm_dirties, &tsk->dirties);
}

P
Peter Zijlstra 已提交
241 242 243 244 245 246 247 248 249 250 251 252 253 254 255
/*
 * Obtain an accurate fraction of the BDI's portion.
 */
static void bdi_writeout_fraction(struct backing_dev_info *bdi,
		long *numerator, long *denominator)
{
	if (bdi_cap_writeback_dirty(bdi)) {
		prop_fraction_percpu(&vm_completions, &bdi->completions,
				numerator, denominator);
	} else {
		*numerator = 0;
		*denominator = 1;
	}
}

P
Peter Zijlstra 已提交
256 257 258 259 260 261 262 263
static inline void task_dirties_fraction(struct task_struct *tsk,
		long *numerator, long *denominator)
{
	prop_fraction_single(&vm_dirties, &tsk->dirties,
				numerator, denominator);
}

/*
264
 * task_dirty_limit - scale down dirty throttling threshold for one task
P
Peter Zijlstra 已提交
265 266 267 268
 *
 * task specific dirty limit:
 *
 *   dirty -= (dirty/8) * p_{t}
269 270 271 272 273 274 275
 *
 * To protect light/slow dirtying tasks from heavier/fast ones, we start
 * throttling individual tasks before reaching the bdi dirty limit.
 * Relatively low thresholds will be allocated to heavy dirtiers. So when
 * dirty pages grow large, heavy dirtiers will be throttled first, which will
 * effectively curb the growth of dirty pages. Light dirtiers with high enough
 * dirty threshold may never get throttled.
P
Peter Zijlstra 已提交
276
 */
277 278
static unsigned long task_dirty_limit(struct task_struct *tsk,
				       unsigned long bdi_dirty)
P
Peter Zijlstra 已提交
279 280
{
	long numerator, denominator;
281
	unsigned long dirty = bdi_dirty;
P
Peter Zijlstra 已提交
282 283 284 285 286 287 288 289
	u64 inv = dirty >> 3;

	task_dirties_fraction(tsk, &numerator, &denominator);
	inv *= numerator;
	do_div(inv, denominator);

	dirty -= inv;

290
	return max(dirty, bdi_dirty/2);
P
Peter Zijlstra 已提交
291 292
}

293 294 295 296 297 298 299 300 301
/*
 *
 */
static unsigned int bdi_min_ratio;

int bdi_set_min_ratio(struct backing_dev_info *bdi, unsigned int min_ratio)
{
	int ret = 0;

302
	spin_lock_bh(&bdi_lock);
303
	if (min_ratio > bdi->max_ratio) {
304
		ret = -EINVAL;
305 306 307 308 309 310 311 312 313
	} 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;
		}
	}
314
	spin_unlock_bh(&bdi_lock);
315 316 317 318 319 320 321 322 323 324 325

	return ret;
}

int bdi_set_max_ratio(struct backing_dev_info *bdi, unsigned max_ratio)
{
	int ret = 0;

	if (max_ratio > 100)
		return -EINVAL;

326
	spin_lock_bh(&bdi_lock);
327 328 329 330 331 332
	if (bdi->min_ratio > max_ratio) {
		ret = -EINVAL;
	} else {
		bdi->max_ratio = max_ratio;
		bdi->max_prop_frac = (PROP_FRAC_BASE * max_ratio) / 100;
	}
333
	spin_unlock_bh(&bdi_lock);
334 335 336

	return ret;
}
337
EXPORT_SYMBOL(bdi_set_max_ratio);
338

L
Linus Torvalds 已提交
339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355
/*
 * Work out the current dirty-memory clamping and background writeout
 * thresholds.
 *
 * The main aim here is to lower them aggressively if there is a lot of mapped
 * memory around.  To avoid stressing page reclaim with lots of unreclaimable
 * pages.  It is better to clamp down on writers than to start swapping, and
 * performing lots of scanning.
 *
 * We only allow 1/2 of the currently-unmapped memory to be dirtied.
 *
 * We don't permit the clamping level to fall below 5% - that is getting rather
 * excessive.
 *
 * We make sure that the background writeout level is below the adjusted
 * clamping level.
 */
356 357 358 359 360 361 362

static unsigned long highmem_dirtyable_memory(unsigned long total)
{
#ifdef CONFIG_HIGHMEM
	int node;
	unsigned long x = 0;

363
	for_each_node_state(node, N_HIGH_MEMORY) {
364 365 366
		struct zone *z =
			&NODE_DATA(node)->node_zones[ZONE_HIGHMEM];

367 368
		x += zone_page_state(z, NR_FREE_PAGES) +
		     zone_reclaimable_pages(z);
369 370 371 372 373 374 375 376 377 378 379 380 381
	}
	/*
	 * 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
}

S
Steven Rostedt 已提交
382 383 384 385 386 387 388
/**
 * determine_dirtyable_memory - amount of memory that may be used
 *
 * Returns the numebr of pages that can currently be freed and used
 * by the kernel for direct mappings.
 */
unsigned long determine_dirtyable_memory(void)
389 390 391
{
	unsigned long x;

392
	x = global_page_state(NR_FREE_PAGES) + global_reclaimable_pages();
393 394 395 396

	if (!vm_highmem_is_dirtyable)
		x -= highmem_dirtyable_memory(x);

397 398 399
	return x + 1;	/* Ensure that we never return 0 */
}

400 401 402 403 404 405 406 407 408
/**
 * global_dirty_limits - background-writeback and dirty-throttling thresholds
 *
 * Calculate the dirty thresholds based on sysctl parameters
 * - vm.dirty_background_ratio  or  vm.dirty_background_bytes
 * - vm.dirty_ratio             or  vm.dirty_bytes
 * The dirty limits will be lifted by 1/4 for PF_LESS_THROTTLE (ie. nfsd) and
 * runtime tasks.
 */
409
void global_dirty_limits(unsigned long *pbackground, unsigned long *pdirty)
L
Linus Torvalds 已提交
410
{
411 412
	unsigned long background;
	unsigned long dirty;
413
	unsigned long available_memory = determine_dirtyable_memory();
L
Linus Torvalds 已提交
414 415
	struct task_struct *tsk;

416 417 418 419 420 421 422 423 424 425
	if (vm_dirty_bytes)
		dirty = DIV_ROUND_UP(vm_dirty_bytes, PAGE_SIZE);
	else {
		int dirty_ratio;

		dirty_ratio = vm_dirty_ratio;
		if (dirty_ratio < 5)
			dirty_ratio = 5;
		dirty = (dirty_ratio * available_memory) / 100;
	}
L
Linus Torvalds 已提交
426

427 428 429 430
	if (dirty_background_bytes)
		background = DIV_ROUND_UP(dirty_background_bytes, PAGE_SIZE);
	else
		background = (dirty_background_ratio * available_memory) / 100;
L
Linus Torvalds 已提交
431

432 433
	if (background >= dirty)
		background = dirty / 2;
L
Linus Torvalds 已提交
434 435 436 437 438 439 440
	tsk = current;
	if (tsk->flags & PF_LESS_THROTTLE || rt_task(tsk)) {
		background += background / 4;
		dirty += dirty / 4;
	}
	*pbackground = background;
	*pdirty = dirty;
441
}
P
Peter Zijlstra 已提交
442

443 444 445 446 447 448 449 450 451 452 453
/**
 * bdi_dirty_limit - @bdi's share of dirty throttling threshold
 *
 * Allocate high/low dirty limits to fast/slow devices, in order to prevent
 * - starving fast devices
 * - piling up dirty pages (that will take long time to sync) on slow devices
 *
 * The bdi's share of dirty limit will be adapting to its throughput and
 * bounded by the bdi->min_ratio and/or bdi->max_ratio parameters, if set.
 */
unsigned long bdi_dirty_limit(struct backing_dev_info *bdi, unsigned long dirty)
454 455 456
{
	u64 bdi_dirty;
	long numerator, denominator;
P
Peter Zijlstra 已提交
457

458 459 460 461
	/*
	 * Calculate this BDI's share of the dirty ratio.
	 */
	bdi_writeout_fraction(bdi, &numerator, &denominator);
P
Peter Zijlstra 已提交
462

463 464 465
	bdi_dirty = (dirty * (100 - bdi_min_ratio)) / 100;
	bdi_dirty *= numerator;
	do_div(bdi_dirty, denominator);
P
Peter Zijlstra 已提交
466

467 468 469 470 471
	bdi_dirty += (dirty * bdi->min_ratio) / 100;
	if (bdi_dirty > (dirty * bdi->max_ratio) / 100)
		bdi_dirty = dirty * bdi->max_ratio / 100;

	return bdi_dirty;
L
Linus Torvalds 已提交
472 473 474 475 476 477
}

/*
 * 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
 * the caller to perform writeback if the system is over `vm_dirty_ratio'.
478 479
 * If we're over `background_thresh' then the writeback threads are woken to
 * perform some writeout.
L
Linus Torvalds 已提交
480
 */
481 482
static void balance_dirty_pages(struct address_space *mapping,
				unsigned long write_chunk)
L
Linus Torvalds 已提交
483
{
484 485
	long nr_reclaimable, bdi_nr_reclaimable;
	long nr_writeback, bdi_nr_writeback;
486 487 488
	unsigned long background_thresh;
	unsigned long dirty_thresh;
	unsigned long bdi_thresh;
L
Linus Torvalds 已提交
489
	unsigned long pages_written = 0;
490
	unsigned long pause = 1;
491
	bool dirty_exceeded = false;
L
Linus Torvalds 已提交
492 493 494 495 496 497 498
	struct backing_dev_info *bdi = mapping->backing_dev_info;

	for (;;) {
		struct writeback_control wbc = {
			.sync_mode	= WB_SYNC_NONE,
			.older_than_this = NULL,
			.nr_to_write	= write_chunk,
499
			.range_cyclic	= 1,
L
Linus Torvalds 已提交
500 501
		};

502 503 504 505
		nr_reclaimable = global_page_state(NR_FILE_DIRTY) +
					global_page_state(NR_UNSTABLE_NFS);
		nr_writeback = global_page_state(NR_WRITEBACK);

506 507 508 509 510 511 512 513 514 515 516 517 518 519
		global_dirty_limits(&background_thresh, &dirty_thresh);

		/*
		 * Throttle it only when the background writeback cannot
		 * catch-up. This avoids (excessively) small writeouts
		 * when the bdi limits are ramping up.
		 */
		if (nr_reclaimable + nr_writeback <
				(background_thresh + dirty_thresh) / 2)
			break;

		bdi_thresh = bdi_dirty_limit(bdi, dirty_thresh);
		bdi_thresh = task_dirty_limit(current, bdi_thresh);

520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536
		/*
		 * 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.
		 */
		if (bdi_thresh < 2*bdi_stat_error(bdi)) {
			bdi_nr_reclaimable = bdi_stat_sum(bdi, BDI_RECLAIMABLE);
			bdi_nr_writeback = bdi_stat_sum(bdi, BDI_WRITEBACK);
		} else {
			bdi_nr_reclaimable = bdi_stat(bdi, BDI_RECLAIMABLE);
			bdi_nr_writeback = bdi_stat(bdi, BDI_WRITEBACK);
		}
537

538 539 540 541 542 543 544 545 546 547 548
		/*
		 * The bdi thresh is somehow "soft" limit derived from the
		 * global "hard" limit. The former helps to prevent heavy IO
		 * bdi or process from holding back light ones; The latter is
		 * the last resort safeguard.
		 */
		dirty_exceeded =
			(bdi_nr_reclaimable + bdi_nr_writeback >= bdi_thresh)
			|| (nr_reclaimable + nr_writeback >= dirty_thresh);

		if (!dirty_exceeded)
P
Peter Zijlstra 已提交
549
			break;
L
Linus Torvalds 已提交
550

P
Peter Zijlstra 已提交
551 552
		if (!bdi->dirty_exceeded)
			bdi->dirty_exceeded = 1;
L
Linus Torvalds 已提交
553 554 555 556 557 558

		/* Note: nr_reclaimable denotes nr_dirty + nr_unstable.
		 * 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.
559 560 561
		 * Only move pages to writeback if this bdi is over its
		 * threshold otherwise wait until the disk writes catch
		 * up.
L
Linus Torvalds 已提交
562
		 */
563
		trace_wbc_balance_dirty_start(&wbc, bdi);
564
		if (bdi_nr_reclaimable > bdi_thresh) {
565
			writeback_inodes_wb(&bdi->wb, &wbc);
L
Linus Torvalds 已提交
566
			pages_written += write_chunk - wbc.nr_to_write;
567
			trace_wbc_balance_dirty_written(&wbc, bdi);
568 569
			if (pages_written >= write_chunk)
				break;		/* We've done our duty */
P
Peter Zijlstra 已提交
570
		}
571
		trace_wbc_balance_dirty_wait(&wbc, bdi);
572 573
		__set_current_state(TASK_INTERRUPTIBLE);
		io_schedule_timeout(pause);
574 575 576 577 578 579 580 581

		/*
		 * Increase the delay for each loop, up to our previous
		 * default of taking a 100ms nap.
		 */
		pause <<= 1;
		if (pause > HZ / 10)
			pause = HZ / 10;
L
Linus Torvalds 已提交
582 583
	}

584
	if (!dirty_exceeded && bdi->dirty_exceeded)
P
Peter Zijlstra 已提交
585
		bdi->dirty_exceeded = 0;
L
Linus Torvalds 已提交
586 587

	if (writeback_in_progress(bdi))
588
		return;
L
Linus Torvalds 已提交
589 590 591 592 593 594 595 596 597 598

	/*
	 * 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.
	 */
	if ((laptop_mode && pages_written) ||
599
	    (!laptop_mode && (nr_reclaimable > background_thresh)))
600
		bdi_start_background_writeback(bdi);
L
Linus Torvalds 已提交
601 602
}

603
void set_page_dirty_balance(struct page *page, int page_mkwrite)
P
Peter Zijlstra 已提交
604
{
605
	if (set_page_dirty(page) || page_mkwrite) {
P
Peter Zijlstra 已提交
606 607 608 609 610 611 612
		struct address_space *mapping = page_mapping(page);

		if (mapping)
			balance_dirty_pages_ratelimited(mapping);
	}
}

613 614
static DEFINE_PER_CPU(unsigned long, bdp_ratelimits) = 0;

L
Linus Torvalds 已提交
615
/**
616
 * balance_dirty_pages_ratelimited_nr - balance dirty memory state
617
 * @mapping: address_space which was dirtied
618
 * @nr_pages_dirtied: number of pages which the caller has just dirtied
L
Linus Torvalds 已提交
619 620 621 622 623 624 625 626 627 628
 *
 * 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.
 */
629 630
void balance_dirty_pages_ratelimited_nr(struct address_space *mapping,
					unsigned long nr_pages_dirtied)
L
Linus Torvalds 已提交
631
{
632 633
	unsigned long ratelimit;
	unsigned long *p;
L
Linus Torvalds 已提交
634 635

	ratelimit = ratelimit_pages;
P
Peter Zijlstra 已提交
636
	if (mapping->backing_dev_info->dirty_exceeded)
L
Linus Torvalds 已提交
637 638 639 640 641 642
		ratelimit = 8;

	/*
	 * Check the rate limiting. Also, we do not want to throttle real-time
	 * tasks in balance_dirty_pages(). Period.
	 */
643
	preempt_disable();
644
	p =  &__get_cpu_var(bdp_ratelimits);
645 646
	*p += nr_pages_dirtied;
	if (unlikely(*p >= ratelimit)) {
647
		ratelimit = sync_writeback_pages(*p);
648 649
		*p = 0;
		preempt_enable();
650
		balance_dirty_pages(mapping, ratelimit);
L
Linus Torvalds 已提交
651 652
		return;
	}
653
	preempt_enable();
L
Linus Torvalds 已提交
654
}
655
EXPORT_SYMBOL(balance_dirty_pages_ratelimited_nr);
L
Linus Torvalds 已提交
656

657
void throttle_vm_writeout(gfp_t gfp_mask)
L
Linus Torvalds 已提交
658
{
659 660
	unsigned long background_thresh;
	unsigned long dirty_thresh;
L
Linus Torvalds 已提交
661 662

        for ( ; ; ) {
663
		global_dirty_limits(&background_thresh, &dirty_thresh);
L
Linus Torvalds 已提交
664 665 666 667 668 669 670

                /*
                 * 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... */

671 672 673
                if (global_page_state(NR_UNSTABLE_NFS) +
			global_page_state(NR_WRITEBACK) <= dirty_thresh)
                        	break;
674
                congestion_wait(BLK_RW_ASYNC, HZ/10);
675 676 677 678 679 680 681 682

		/*
		 * 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;
L
Linus Torvalds 已提交
683 684 685 686 687 688 689
        }
}

/*
 * sysctl handler for /proc/sys/vm/dirty_writeback_centisecs
 */
int dirty_writeback_centisecs_handler(ctl_table *table, int write,
690
	void __user *buffer, size_t *length, loff_t *ppos)
L
Linus Torvalds 已提交
691
{
692
	proc_dointvec(table, write, buffer, length, ppos);
693
	bdi_arm_supers_timer();
L
Linus Torvalds 已提交
694 695 696
	return 0;
}

697
#ifdef CONFIG_BLOCK
698
void laptop_mode_timer_fn(unsigned long data)
L
Linus Torvalds 已提交
699
{
700 701 702
	struct request_queue *q = (struct request_queue *)data;
	int nr_pages = global_page_state(NR_FILE_DIRTY) +
		global_page_state(NR_UNSTABLE_NFS);
L
Linus Torvalds 已提交
703

704 705 706 707 708
	/*
	 * We want to write everything out, not just down to the dirty
	 * threshold
	 */
	if (bdi_has_dirty_io(&q->backing_dev_info))
709
		bdi_start_writeback(&q->backing_dev_info, nr_pages);
L
Linus Torvalds 已提交
710 711 712 713 714 715 716
}

/*
 * 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.
 */
717
void laptop_io_completion(struct backing_dev_info *info)
L
Linus Torvalds 已提交
718
{
719
	mod_timer(&info->laptop_mode_wb_timer, jiffies + laptop_mode);
L
Linus Torvalds 已提交
720 721 722 723 724 725 726 727 728
}

/*
 * 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)
{
729 730 731 732 733 734 735 736
	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();
L
Linus Torvalds 已提交
737
}
738
#endif
L
Linus Torvalds 已提交
739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756

/*
 * 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
 * thresholds before writeback cuts in.
 *
 * But the limit should not be set too high.  Because it also controls the
 * amount of memory which the balance_dirty_pages() caller has to write back.
 * If this is too large then the caller will block on the IO queue all the
 * time.  So limit it to four megabytes - the balance_dirty_pages() caller
 * will write six megabyte chunks, max.
 */

757
void writeback_set_ratelimit(void)
L
Linus Torvalds 已提交
758
{
759
	ratelimit_pages = vm_total_pages / (num_online_cpus() * 32);
L
Linus Torvalds 已提交
760 761 762 763 764 765
	if (ratelimit_pages < 16)
		ratelimit_pages = 16;
	if (ratelimit_pages * PAGE_CACHE_SIZE > 4096 * 1024)
		ratelimit_pages = (4096 * 1024) / PAGE_CACHE_SIZE;
}

766
static int __cpuinit
L
Linus Torvalds 已提交
767 768
ratelimit_handler(struct notifier_block *self, unsigned long u, void *v)
{
769
	writeback_set_ratelimit();
770
	return NOTIFY_DONE;
L
Linus Torvalds 已提交
771 772
}

773
static struct notifier_block __cpuinitdata ratelimit_nb = {
L
Linus Torvalds 已提交
774 775 776 777 778
	.notifier_call	= ratelimit_handler,
	.next		= NULL,
};

/*
779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794
 * 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..
L
Linus Torvalds 已提交
795 796 797
 */
void __init page_writeback_init(void)
{
P
Peter Zijlstra 已提交
798 799
	int shift;

800
	writeback_set_ratelimit();
L
Linus Torvalds 已提交
801
	register_cpu_notifier(&ratelimit_nb);
P
Peter Zijlstra 已提交
802 803 804

	shift = calc_period_shift();
	prop_descriptor_init(&vm_completions, shift);
P
Peter Zijlstra 已提交
805
	prop_descriptor_init(&vm_dirties, shift);
L
Linus Torvalds 已提交
806 807
}

808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827
/**
 * 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.
 */
void tag_pages_for_writeback(struct address_space *mapping,
			     pgoff_t start, pgoff_t end)
{
R
Randy Dunlap 已提交
828
#define WRITEBACK_TAG_BATCH 4096
829 830 831 832 833 834 835 836 837 838 839 840 841 842
	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();
	} while (tagged >= WRITEBACK_TAG_BATCH);
}
EXPORT_SYMBOL(tag_pages_for_writeback);

843
/**
844
 * write_cache_pages - walk the list of dirty pages of the given address space and write all of them.
845 846
 * @mapping: address space structure to write
 * @wbc: subtract the number of written pages from *@wbc->nr_to_write
847 848
 * @writepage: function called for each page
 * @data: data passed to writepage function
849
 *
850
 * If a page is already under I/O, write_cache_pages() skips it, even
851 852 853 854 855 856
 * 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.
857 858 859 860 861 862 863
 *
 * 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).
864
 */
865 866 867
int write_cache_pages(struct address_space *mapping,
		      struct writeback_control *wbc, writepage_t writepage,
		      void *data)
868 869 870 871 872
{
	int ret = 0;
	int done = 0;
	struct pagevec pvec;
	int nr_pages;
N
Nick Piggin 已提交
873
	pgoff_t uninitialized_var(writeback_index);
874 875
	pgoff_t index;
	pgoff_t end;		/* Inclusive */
876
	pgoff_t done_index;
N
Nick Piggin 已提交
877
	int cycled;
878
	int range_whole = 0;
879
	int tag;
880 881 882

	pagevec_init(&pvec, 0);
	if (wbc->range_cyclic) {
N
Nick Piggin 已提交
883 884 885 886 887 888
		writeback_index = mapping->writeback_index; /* prev offset */
		index = writeback_index;
		if (index == 0)
			cycled = 1;
		else
			cycled = 0;
889 890 891 892 893 894
		end = -1;
	} else {
		index = wbc->range_start >> PAGE_CACHE_SHIFT;
		end = wbc->range_end >> PAGE_CACHE_SHIFT;
		if (wbc->range_start == 0 && wbc->range_end == LLONG_MAX)
			range_whole = 1;
N
Nick Piggin 已提交
895
		cycled = 1; /* ignore range_cyclic tests */
896
	}
897 898 899 900
	if (wbc->sync_mode == WB_SYNC_ALL)
		tag = PAGECACHE_TAG_TOWRITE;
	else
		tag = PAGECACHE_TAG_DIRTY;
901
retry:
902 903
	if (wbc->sync_mode == WB_SYNC_ALL)
		tag_pages_for_writeback(mapping, index, end);
904
	done_index = index;
N
Nick Piggin 已提交
905 906 907
	while (!done && (index <= end)) {
		int i;

908
		nr_pages = pagevec_lookup_tag(&pvec, mapping, &index, tag,
N
Nick Piggin 已提交
909 910 911
			      min(end - index, (pgoff_t)PAGEVEC_SIZE-1) + 1);
		if (nr_pages == 0)
			break;
912 913 914 915 916

		for (i = 0; i < nr_pages; i++) {
			struct page *page = pvec.pages[i];

			/*
917 918 919 920 921
			 * 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.
922
			 */
923 924 925 926 927 928 929 930 931 932 933
			if (page->index > end) {
				/*
				 * can't be range_cyclic (1st pass) because
				 * end == -1 in that case.
				 */
				done = 1;
				break;
			}

			done_index = page->index + 1;

934 935
			lock_page(page);

N
Nick Piggin 已提交
936 937 938 939 940 941 942 943
			/*
			 * 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.
			 */
944
			if (unlikely(page->mapping != mapping)) {
N
Nick Piggin 已提交
945
continue_unlock:
946 947 948 949
				unlock_page(page);
				continue;
			}

950 951 952 953 954 955 956 957 958 959 960
			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;
			}
961

962 963
			BUG_ON(PageWriteback(page));
			if (!clear_page_dirty_for_io(page))
N
Nick Piggin 已提交
964
				goto continue_unlock;
965

966
			trace_wbc_writepage(wbc, mapping->backing_dev_info);
967
			ret = (*writepage)(page, wbc, data);
968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984
			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).
					 */
					done = 1;
					break;
				}
985
			}
986

987 988
			if (wbc->nr_to_write > 0) {
				if (--wbc->nr_to_write == 0 &&
F
Federico Cuello 已提交
989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002
				    wbc->sync_mode == WB_SYNC_NONE) {
					/*
					 * We stop writing back only if we are
					 * not doing integrity sync. In case of
					 * integrity sync we have to keep going
					 * because someone may be concurrently
					 * dirtying pages, and we might have
					 * synced a lot of newly appeared dirty
					 * pages, but have not synced all of the
					 * old dirty pages.
					 */
					done = 1;
					break;
				}
1003
			}
1004 1005 1006 1007
		}
		pagevec_release(&pvec);
		cond_resched();
	}
1008
	if (!cycled && !done) {
1009
		/*
N
Nick Piggin 已提交
1010
		 * range_cyclic:
1011 1012 1013
		 * We hit the last page and there is more work to be done: wrap
		 * back to the start of the file
		 */
N
Nick Piggin 已提交
1014
		cycled = 1;
1015
		index = 0;
N
Nick Piggin 已提交
1016
		end = writeback_index - 1;
1017 1018
		goto retry;
	}
1019 1020
	if (wbc->range_cyclic || (range_whole && wbc->nr_to_write > 0))
		mapping->writeback_index = done_index;
1021

1022 1023
	return ret;
}
1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055
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)
{
	/* deal with chardevs and other special file */
	if (!mapping->a_ops->writepage)
		return 0;

	return write_cache_pages(mapping, wbc, __writepage, mapping);
}
1056 1057 1058

EXPORT_SYMBOL(generic_writepages);

L
Linus Torvalds 已提交
1059 1060
int do_writepages(struct address_space *mapping, struct writeback_control *wbc)
{
1061 1062
	int ret;

L
Linus Torvalds 已提交
1063 1064 1065
	if (wbc->nr_to_write <= 0)
		return 0;
	if (mapping->a_ops->writepages)
1066
		ret = mapping->a_ops->writepages(mapping, wbc);
1067 1068 1069
	else
		ret = generic_writepages(mapping, wbc);
	return ret;
L
Linus Torvalds 已提交
1070 1071 1072 1073
}

/**
 * write_one_page - write out a single page and optionally wait on I/O
1074 1075
 * @page: the page to write
 * @wait: if true, wait on writeout
L
Linus Torvalds 已提交
1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110
 *
 * 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)) {
		page_cache_get(page);
		ret = mapping->a_ops->writepage(page, &wbc);
		if (ret == 0 && wait) {
			wait_on_page_writeback(page);
			if (PageError(page))
				ret = -EIO;
		}
		page_cache_release(page);
	} else {
		unlock_page(page);
	}
	return ret;
}
EXPORT_SYMBOL(write_one_page);

1111 1112 1113 1114 1115 1116 1117 1118 1119 1120
/*
 * For address_spaces which do not use buffers nor write back.
 */
int __set_page_dirty_no_writeback(struct page *page)
{
	if (!PageDirty(page))
		SetPageDirty(page);
	return 0;
}

1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134
/*
 * Helper function for set_page_dirty family.
 * NOTE: This relies on being atomic wrt interrupts.
 */
void account_page_dirtied(struct page *page, struct address_space *mapping)
{
	if (mapping_cap_account_dirty(mapping)) {
		__inc_zone_page_state(page, NR_FILE_DIRTY);
		__inc_bdi_stat(mapping->backing_dev_info, BDI_RECLAIMABLE);
		task_dirty_inc(current);
		task_io_account_write(PAGE_CACHE_SIZE);
	}
}

L
Linus Torvalds 已提交
1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147
/*
 * 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.
 *
 * Most callers have locked the page, which pins the address_space in memory.
 * But zap_pte_range() does not lock the page, however in that case the
 * mapping is pinned by the vma's ->vm_file reference.
 *
 * We take care to handle the case where the page was truncated from the
S
Simon Arlott 已提交
1148
 * mapping by re-checking page_mapping() inside tree_lock.
L
Linus Torvalds 已提交
1149 1150 1151 1152 1153 1154 1155
 */
int __set_page_dirty_nobuffers(struct page *page)
{
	if (!TestSetPageDirty(page)) {
		struct address_space *mapping = page_mapping(page);
		struct address_space *mapping2;

1156 1157 1158
		if (!mapping)
			return 1;

N
Nick Piggin 已提交
1159
		spin_lock_irq(&mapping->tree_lock);
1160 1161 1162
		mapping2 = page_mapping(page);
		if (mapping2) { /* Race with truncate? */
			BUG_ON(mapping2 != mapping);
1163
			WARN_ON_ONCE(!PagePrivate(page) && !PageUptodate(page));
1164
			account_page_dirtied(page, mapping);
1165 1166 1167
			radix_tree_tag_set(&mapping->page_tree,
				page_index(page), PAGECACHE_TAG_DIRTY);
		}
N
Nick Piggin 已提交
1168
		spin_unlock_irq(&mapping->tree_lock);
1169 1170 1171
		if (mapping->host) {
			/* !PageAnon && !swapper_space */
			__mark_inode_dirty(mapping->host, I_DIRTY_PAGES);
L
Linus Torvalds 已提交
1172
		}
1173
		return 1;
L
Linus Torvalds 已提交
1174
	}
1175
	return 0;
L
Linus Torvalds 已提交
1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191
}
EXPORT_SYMBOL(__set_page_dirty_nobuffers);

/*
 * 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)
{
	wbc->pages_skipped++;
	return __set_page_dirty_nobuffers(page);
}
EXPORT_SYMBOL(redirty_page_for_writepage);

/*
1192 1193 1194 1195 1196 1197 1198
 * 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.
 *
L
Linus Torvalds 已提交
1199 1200 1201
 * If the mapping doesn't provide a set_page_dirty a_op, then
 * just fall through and assume that it wants buffer_heads.
 */
N
Nick Piggin 已提交
1202
int set_page_dirty(struct page *page)
L
Linus Torvalds 已提交
1203 1204 1205 1206 1207
{
	struct address_space *mapping = page_mapping(page);

	if (likely(mapping)) {
		int (*spd)(struct page *) = mapping->a_ops->set_page_dirty;
1208 1209 1210 1211 1212
#ifdef CONFIG_BLOCK
		if (!spd)
			spd = __set_page_dirty_buffers;
#endif
		return (*spd)(page);
L
Linus Torvalds 已提交
1213
	}
1214 1215 1216 1217
	if (!PageDirty(page)) {
		if (!TestSetPageDirty(page))
			return 1;
	}
L
Linus Torvalds 已提交
1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235
	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;

1236
	lock_page_nosync(page);
L
Linus Torvalds 已提交
1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260
	ret = set_page_dirty(page);
	unlock_page(page);
	return ret;
}
EXPORT_SYMBOL(set_page_dirty_lock);

/*
 * 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);

1261 1262
	BUG_ON(!PageLocked(page));

1263
	ClearPageReclaim(page);
1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291
	if (mapping && mapping_cap_account_dirty(mapping)) {
		/*
		 * 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.
		 */
		if (page_mkclean(page))
			set_page_dirty(page);
1292 1293 1294 1295 1296 1297 1298 1299 1300 1301
		/*
		 * We carefully synchronise fault handlers against
		 * installing a dirty pte and marking the page dirty
		 * at this point. We do this by having them hold the
		 * page lock at some point after installing their
		 * pte, but before marking the page dirty.
		 * Pages are always locked coming in here, so we get
		 * the desired exclusion. See mm/memory.c:do_wp_page()
		 * for more comments.
		 */
1302
		if (TestClearPageDirty(page)) {
1303
			dec_zone_page_state(page, NR_FILE_DIRTY);
1304 1305
			dec_bdi_stat(mapping->backing_dev_info,
					BDI_RECLAIMABLE);
1306
			return 1;
L
Linus Torvalds 已提交
1307
		}
1308
		return 0;
L
Linus Torvalds 已提交
1309
	}
1310
	return TestClearPageDirty(page);
L
Linus Torvalds 已提交
1311
}
1312
EXPORT_SYMBOL(clear_page_dirty_for_io);
L
Linus Torvalds 已提交
1313 1314 1315 1316 1317 1318 1319

int test_clear_page_writeback(struct page *page)
{
	struct address_space *mapping = page_mapping(page);
	int ret;

	if (mapping) {
P
Peter Zijlstra 已提交
1320
		struct backing_dev_info *bdi = mapping->backing_dev_info;
L
Linus Torvalds 已提交
1321 1322
		unsigned long flags;

N
Nick Piggin 已提交
1323
		spin_lock_irqsave(&mapping->tree_lock, flags);
L
Linus Torvalds 已提交
1324
		ret = TestClearPageWriteback(page);
P
Peter Zijlstra 已提交
1325
		if (ret) {
L
Linus Torvalds 已提交
1326 1327 1328
			radix_tree_tag_clear(&mapping->page_tree,
						page_index(page),
						PAGECACHE_TAG_WRITEBACK);
1329
			if (bdi_cap_account_writeback(bdi)) {
P
Peter Zijlstra 已提交
1330
				__dec_bdi_stat(bdi, BDI_WRITEBACK);
P
Peter Zijlstra 已提交
1331 1332
				__bdi_writeout_inc(bdi);
			}
P
Peter Zijlstra 已提交
1333
		}
N
Nick Piggin 已提交
1334
		spin_unlock_irqrestore(&mapping->tree_lock, flags);
L
Linus Torvalds 已提交
1335 1336 1337
	} else {
		ret = TestClearPageWriteback(page);
	}
1338 1339
	if (ret)
		dec_zone_page_state(page, NR_WRITEBACK);
L
Linus Torvalds 已提交
1340 1341 1342 1343 1344 1345 1346 1347 1348
	return ret;
}

int test_set_page_writeback(struct page *page)
{
	struct address_space *mapping = page_mapping(page);
	int ret;

	if (mapping) {
P
Peter Zijlstra 已提交
1349
		struct backing_dev_info *bdi = mapping->backing_dev_info;
L
Linus Torvalds 已提交
1350 1351
		unsigned long flags;

N
Nick Piggin 已提交
1352
		spin_lock_irqsave(&mapping->tree_lock, flags);
L
Linus Torvalds 已提交
1353
		ret = TestSetPageWriteback(page);
P
Peter Zijlstra 已提交
1354
		if (!ret) {
L
Linus Torvalds 已提交
1355 1356 1357
			radix_tree_tag_set(&mapping->page_tree,
						page_index(page),
						PAGECACHE_TAG_WRITEBACK);
1358
			if (bdi_cap_account_writeback(bdi))
P
Peter Zijlstra 已提交
1359 1360
				__inc_bdi_stat(bdi, BDI_WRITEBACK);
		}
L
Linus Torvalds 已提交
1361 1362 1363 1364
		if (!PageDirty(page))
			radix_tree_tag_clear(&mapping->page_tree,
						page_index(page),
						PAGECACHE_TAG_DIRTY);
1365 1366 1367
		radix_tree_tag_clear(&mapping->page_tree,
				     page_index(page),
				     PAGECACHE_TAG_TOWRITE);
N
Nick Piggin 已提交
1368
		spin_unlock_irqrestore(&mapping->tree_lock, flags);
L
Linus Torvalds 已提交
1369 1370 1371
	} else {
		ret = TestSetPageWriteback(page);
	}
1372 1373
	if (!ret)
		inc_zone_page_state(page, NR_WRITEBACK);
L
Linus Torvalds 已提交
1374 1375 1376 1377 1378 1379
	return ret;

}
EXPORT_SYMBOL(test_set_page_writeback);

/*
N
Nick Piggin 已提交
1380
 * Return true if any of the pages in the mapping are marked with the
L
Linus Torvalds 已提交
1381 1382 1383 1384 1385
 * passed tag.
 */
int mapping_tagged(struct address_space *mapping, int tag)
{
	int ret;
N
Nick Piggin 已提交
1386
	rcu_read_lock();
L
Linus Torvalds 已提交
1387
	ret = radix_tree_tagged(&mapping->page_tree, tag);
N
Nick Piggin 已提交
1388
	rcu_read_unlock();
L
Linus Torvalds 已提交
1389 1390 1391
	return ret;
}
EXPORT_SYMBOL(mapping_tagged);