page-writeback.c 36.1 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>
L
Linus Torvalds 已提交
37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59

/*
 * 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.
 * It should be somewhat larger than RATELIMIT_PAGES to ensure that reasonably
 * large amounts of I/O are submitted.
 */
static inline long sync_writeback_pages(void)
{
	return ratelimit_pages + ratelimit_pages / 2;
}

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

/*
 * Start background writeback (via pdflush) at this percentage
 */
60
int dirty_background_ratio = 10;
L
Linus Torvalds 已提交
61

62 63 64 65 66 67
/*
 * 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;

68 69 70 71 72 73
/*
 * 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
74 75 76
/*
 * The generator of dirty data starts writeback at this percentage
 */
77
int vm_dirty_ratio = 20;
L
Linus Torvalds 已提交
78

79 80 81 82 83 84
/*
 * 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
85
/*
86
 * The interval between `kupdate'-style writebacks
L
Linus Torvalds 已提交
87
 */
88
unsigned int dirty_writeback_interval = 5 * 100; /* centiseconds */
L
Linus Torvalds 已提交
89 90

/*
91
 * The longest time for which data is allowed to remain dirty
L
Linus Torvalds 已提交
92
 */
93
unsigned int dirty_expire_interval = 30 * 100; /* centiseconds */
L
Linus Torvalds 已提交
94 95 96 97 98 99 100

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

/*
101 102
 * 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
103 104 105 106 107 108 109 110
 */
int laptop_mode;

EXPORT_SYMBOL(laptop_mode);

/* End of sysctl-exported parameters */


P
Peter Zijlstra 已提交
111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127
/*
 * 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
128
static struct prop_descriptor vm_dirties;
P
Peter Zijlstra 已提交
129 130 131 132 133 134 135 136 137 138

/*
 * 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;

139 140 141 142 143
	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
144 145 146 147
	return 2 + ilog2(dirty_total - 1);
}

/*
148
 * update the period when the dirty threshold changes.
P
Peter Zijlstra 已提交
149
 */
150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180
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,
		struct file *filp, void __user *buffer, size_t *lenp,
		loff_t *ppos)
{
	int ret;

	ret = proc_dointvec_minmax(table, write, filp, buffer, lenp, ppos);
	if (ret == 0 && write)
		dirty_background_bytes = 0;
	return ret;
}

int dirty_background_bytes_handler(struct ctl_table *table, int write,
		struct file *filp, void __user *buffer, size_t *lenp,
		loff_t *ppos)
{
	int ret;

	ret = proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos);
	if (ret == 0 && write)
		dirty_background_ratio = 0;
	return ret;
}

P
Peter Zijlstra 已提交
181 182 183 184 185
int dirty_ratio_handler(struct ctl_table *table, int write,
		struct file *filp, void __user *buffer, size_t *lenp,
		loff_t *ppos)
{
	int old_ratio = vm_dirty_ratio;
186 187 188
	int ret;

	ret = proc_dointvec_minmax(table, write, filp, buffer, lenp, ppos);
P
Peter Zijlstra 已提交
189
	if (ret == 0 && write && vm_dirty_ratio != old_ratio) {
190 191 192 193 194 195 196 197 198 199 200
		update_completion_period();
		vm_dirty_bytes = 0;
	}
	return ret;
}


int dirty_bytes_handler(struct ctl_table *table, int write,
		struct file *filp, void __user *buffer, size_t *lenp,
		loff_t *ppos)
{
201
	unsigned long old_bytes = vm_dirty_bytes;
202 203 204 205 206 207
	int ret;

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

222 223 224 225 226 227 228 229 230 231
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
232
void task_dirty_inc(struct task_struct *tsk)
P
Peter Zijlstra 已提交
233 234 235 236
{
	prop_inc_single(&vm_dirties, &tsk->dirties);
}

P
Peter Zijlstra 已提交
237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256
/*
 * 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;
	}
}

/*
 * Clip the earned share of dirty pages to that which is actually available.
 * This avoids exceeding the total dirty_limit when the floating averages
 * fluctuate too quickly.
 */
257 258
static void clip_bdi_dirty_limit(struct backing_dev_info *bdi,
		unsigned long dirty, unsigned long *pbdi_dirty)
P
Peter Zijlstra 已提交
259
{
260
	unsigned long avail_dirty;
P
Peter Zijlstra 已提交
261

262
	avail_dirty = global_page_state(NR_FILE_DIRTY) +
P
Peter Zijlstra 已提交
263
		 global_page_state(NR_WRITEBACK) +
264
		 global_page_state(NR_UNSTABLE_NFS) +
265
		 global_page_state(NR_WRITEBACK_TEMP);
P
Peter Zijlstra 已提交
266

267 268 269
	if (avail_dirty < dirty)
		avail_dirty = dirty - avail_dirty;
	else
P
Peter Zijlstra 已提交
270 271 272 273 274 275 276 277
		avail_dirty = 0;

	avail_dirty += bdi_stat(bdi, BDI_RECLAIMABLE) +
		bdi_stat(bdi, BDI_WRITEBACK);

	*pbdi_dirty = min(*pbdi_dirty, avail_dirty);
}

P
Peter Zijlstra 已提交
278 279 280 281 282 283 284 285 286 287 288 289 290 291
static inline void task_dirties_fraction(struct task_struct *tsk,
		long *numerator, long *denominator)
{
	prop_fraction_single(&vm_dirties, &tsk->dirties,
				numerator, denominator);
}

/*
 * scale the dirty limit
 *
 * task specific dirty limit:
 *
 *   dirty -= (dirty/8) * p_{t}
 */
292
static void task_dirty_limit(struct task_struct *tsk, unsigned long *pdirty)
P
Peter Zijlstra 已提交
293 294
{
	long numerator, denominator;
295
	unsigned long dirty = *pdirty;
P
Peter Zijlstra 已提交
296 297 298 299 300 301 302 303 304 305 306 307 308
	u64 inv = dirty >> 3;

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

	dirty -= inv;
	if (dirty < *pdirty/2)
		dirty = *pdirty/2;

	*pdirty = dirty;
}

309 310 311 312 313 314 315 316 317
/*
 *
 */
static unsigned int bdi_min_ratio;

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

318
	spin_lock(&bdi_lock);
319
	if (min_ratio > bdi->max_ratio) {
320
		ret = -EINVAL;
321 322 323 324 325 326 327 328 329
	} 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;
		}
	}
330
	spin_unlock(&bdi_lock);
331 332 333 334 335 336 337 338 339 340 341

	return ret;
}

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

	if (max_ratio > 100)
		return -EINVAL;

342
	spin_lock(&bdi_lock);
343 344 345 346 347 348
	if (bdi->min_ratio > max_ratio) {
		ret = -EINVAL;
	} else {
		bdi->max_ratio = max_ratio;
		bdi->max_prop_frac = (PROP_FRAC_BASE * max_ratio) / 100;
	}
349
	spin_unlock(&bdi_lock);
350 351 352

	return ret;
}
353
EXPORT_SYMBOL(bdi_set_max_ratio);
354

L
Linus Torvalds 已提交
355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371
/*
 * 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.
 */
372 373 374 375 376 377 378

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

379
	for_each_node_state(node, N_HIGH_MEMORY) {
380 381 382
		struct zone *z =
			&NODE_DATA(node)->node_zones[ZONE_HIGHMEM];

383
		x += zone_page_state(z, NR_FREE_PAGES) + zone_lru_pages(z);
384 385 386 387 388 389 390 391 392 393 394 395 396
	}
	/*
	 * 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
397 398 399 400 401 402 403
/**
 * 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)
404 405 406
{
	unsigned long x;

407
	x = global_page_state(NR_FREE_PAGES) + global_lru_pages();
408 409 410 411

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

412 413 414
	return x + 1;	/* Ensure that we never return 0 */
}

415
void
416 417
get_dirty_limits(unsigned long *pbackground, unsigned long *pdirty,
		 unsigned long *pbdi_dirty, struct backing_dev_info *bdi)
L
Linus Torvalds 已提交
418
{
419 420
	unsigned long background;
	unsigned long dirty;
421
	unsigned long available_memory = determine_dirtyable_memory();
L
Linus Torvalds 已提交
422 423
	struct task_struct *tsk;

424 425 426 427 428 429 430 431 432 433
	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
434

435 436 437 438
	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
439

440 441
	if (background >= dirty)
		background = dirty / 2;
L
Linus Torvalds 已提交
442 443 444 445 446 447 448
	tsk = current;
	if (tsk->flags & PF_LESS_THROTTLE || rt_task(tsk)) {
		background += background / 4;
		dirty += dirty / 4;
	}
	*pbackground = background;
	*pdirty = dirty;
P
Peter Zijlstra 已提交
449 450

	if (bdi) {
451
		u64 bdi_dirty;
P
Peter Zijlstra 已提交
452 453 454 455 456 457 458
		long numerator, denominator;

		/*
		 * Calculate this BDI's share of the dirty ratio.
		 */
		bdi_writeout_fraction(bdi, &numerator, &denominator);

459
		bdi_dirty = (dirty * (100 - bdi_min_ratio)) / 100;
P
Peter Zijlstra 已提交
460 461
		bdi_dirty *= numerator;
		do_div(bdi_dirty, denominator);
462
		bdi_dirty += (dirty * bdi->min_ratio) / 100;
463 464
		if (bdi_dirty > (dirty * bdi->max_ratio) / 100)
			bdi_dirty = dirty * bdi->max_ratio / 100;
P
Peter Zijlstra 已提交
465 466 467

		*pbdi_dirty = bdi_dirty;
		clip_bdi_dirty_limit(bdi, dirty, pbdi_dirty);
P
Peter Zijlstra 已提交
468
		task_dirty_limit(current, pbdi_dirty);
P
Peter Zijlstra 已提交
469
	}
L
Linus Torvalds 已提交
470 471 472 473 474 475 476 477 478 479 480
}

/*
 * 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'.
 * If we're over `background_thresh' then pdflush is woken to perform some
 * writeout.
 */
static void balance_dirty_pages(struct address_space *mapping)
{
481 482
	long nr_reclaimable, bdi_nr_reclaimable;
	long nr_writeback, bdi_nr_writeback;
483 484 485
	unsigned long background_thresh;
	unsigned long dirty_thresh;
	unsigned long bdi_thresh;
L
Linus Torvalds 已提交
486 487 488 489 490 491 492 493 494 495 496
	unsigned long pages_written = 0;
	unsigned long write_chunk = sync_writeback_pages();

	struct backing_dev_info *bdi = mapping->backing_dev_info;

	for (;;) {
		struct writeback_control wbc = {
			.bdi		= bdi,
			.sync_mode	= WB_SYNC_NONE,
			.older_than_this = NULL,
			.nr_to_write	= write_chunk,
497
			.range_cyclic	= 1,
L
Linus Torvalds 已提交
498 499
		};

P
Peter Zijlstra 已提交
500 501
		get_dirty_limits(&background_thresh, &dirty_thresh,
				&bdi_thresh, bdi);
502 503 504 505 506

		nr_reclaimable = global_page_state(NR_FILE_DIRTY) +
					global_page_state(NR_UNSTABLE_NFS);
		nr_writeback = global_page_state(NR_WRITEBACK);

P
Peter Zijlstra 已提交
507 508
		bdi_nr_reclaimable = bdi_stat(bdi, BDI_RECLAIMABLE);
		bdi_nr_writeback = bdi_stat(bdi, BDI_WRITEBACK);
509

P
Peter Zijlstra 已提交
510 511
		if (bdi_nr_reclaimable + bdi_nr_writeback <= bdi_thresh)
			break;
L
Linus Torvalds 已提交
512

513 514 515 516 517 518 519 520 521
		/*
		 * 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;

P
Peter Zijlstra 已提交
522 523
		if (!bdi->dirty_exceeded)
			bdi->dirty_exceeded = 1;
L
Linus Torvalds 已提交
524 525 526 527 528 529

		/* 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.
530 531 532
		 * 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
533
		 */
534
		if (bdi_nr_reclaimable > bdi_thresh) {
535
			writeback_inodes_wbc(&wbc);
L
Linus Torvalds 已提交
536
			pages_written += write_chunk - wbc.nr_to_write;
P
Peter Zijlstra 已提交
537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556
			get_dirty_limits(&background_thresh, &dirty_thresh,
				       &bdi_thresh, bdi);
		}

		/*
		 * 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 if (bdi_nr_reclaimable) {
			bdi_nr_reclaimable = bdi_stat(bdi, BDI_RECLAIMABLE);
			bdi_nr_writeback = bdi_stat(bdi, BDI_WRITEBACK);
L
Linus Torvalds 已提交
557
		}
P
Peter Zijlstra 已提交
558 559 560 561 562 563

		if (bdi_nr_reclaimable + bdi_nr_writeback <= bdi_thresh)
			break;
		if (pages_written >= write_chunk)
			break;		/* We've done our duty */

564
		schedule_timeout(1);
L
Linus Torvalds 已提交
565 566
	}

P
Peter Zijlstra 已提交
567 568 569
	if (bdi_nr_reclaimable + bdi_nr_writeback < bdi_thresh &&
			bdi->dirty_exceeded)
		bdi->dirty_exceeded = 0;
L
Linus Torvalds 已提交
570 571 572 573 574 575 576 577 578 579 580 581 582

	if (writeback_in_progress(bdi))
		return;		/* pdflush is already working this queue */

	/*
	 * 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) ||
583 584 585 586 587 588 589 590 591 592 593 594
	    (!laptop_mode && ((nr_writeback = global_page_state(NR_FILE_DIRTY)
					  + global_page_state(NR_UNSTABLE_NFS))
					  > background_thresh))) {
		struct writeback_control wbc = {
			.bdi		= bdi,
			.sync_mode	= WB_SYNC_NONE,
			.nr_to_write	= nr_writeback,
		};


		bdi_start_writeback(&wbc);
	}
L
Linus Torvalds 已提交
595 596
}

597
void set_page_dirty_balance(struct page *page, int page_mkwrite)
P
Peter Zijlstra 已提交
598
{
599
	if (set_page_dirty(page) || page_mkwrite) {
P
Peter Zijlstra 已提交
600 601 602 603 604 605 606
		struct address_space *mapping = page_mapping(page);

		if (mapping)
			balance_dirty_pages_ratelimited(mapping);
	}
}

607 608
static DEFINE_PER_CPU(unsigned long, bdp_ratelimits) = 0;

L
Linus Torvalds 已提交
609
/**
610
 * balance_dirty_pages_ratelimited_nr - balance dirty memory state
611
 * @mapping: address_space which was dirtied
612
 * @nr_pages_dirtied: number of pages which the caller has just dirtied
L
Linus Torvalds 已提交
613 614 615 616 617 618 619 620 621 622
 *
 * 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.
 */
623 624
void balance_dirty_pages_ratelimited_nr(struct address_space *mapping,
					unsigned long nr_pages_dirtied)
L
Linus Torvalds 已提交
625
{
626 627
	unsigned long ratelimit;
	unsigned long *p;
L
Linus Torvalds 已提交
628 629

	ratelimit = ratelimit_pages;
P
Peter Zijlstra 已提交
630
	if (mapping->backing_dev_info->dirty_exceeded)
L
Linus Torvalds 已提交
631 632 633 634 635 636
		ratelimit = 8;

	/*
	 * Check the rate limiting. Also, we do not want to throttle real-time
	 * tasks in balance_dirty_pages(). Period.
	 */
637
	preempt_disable();
638
	p =  &__get_cpu_var(bdp_ratelimits);
639 640 641 642
	*p += nr_pages_dirtied;
	if (unlikely(*p >= ratelimit)) {
		*p = 0;
		preempt_enable();
L
Linus Torvalds 已提交
643 644 645
		balance_dirty_pages(mapping);
		return;
	}
646
	preempt_enable();
L
Linus Torvalds 已提交
647
}
648
EXPORT_SYMBOL(balance_dirty_pages_ratelimited_nr);
L
Linus Torvalds 已提交
649

650
void throttle_vm_writeout(gfp_t gfp_mask)
L
Linus Torvalds 已提交
651
{
652 653
	unsigned long background_thresh;
	unsigned long dirty_thresh;
L
Linus Torvalds 已提交
654 655

        for ( ; ; ) {
P
Peter Zijlstra 已提交
656
		get_dirty_limits(&background_thresh, &dirty_thresh, NULL, NULL);
L
Linus Torvalds 已提交
657 658 659 660 661 662 663

                /*
                 * 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... */

664 665 666
                if (global_page_state(NR_UNSTABLE_NFS) +
			global_page_state(NR_WRITEBACK) <= dirty_thresh)
                        	break;
667
                congestion_wait(BLK_RW_ASYNC, HZ/10);
668 669 670 671 672 673 674 675

		/*
		 * 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
676 677 678 679 680
        }
}

static void laptop_timer_fn(unsigned long unused);

681
static DEFINE_TIMER(laptop_mode_wb_timer, laptop_timer_fn, 0, 0);
L
Linus Torvalds 已提交
682 683 684 685 686

/*
 * sysctl handler for /proc/sys/vm/dirty_writeback_centisecs
 */
int dirty_writeback_centisecs_handler(ctl_table *table, int write,
687
	struct file *file, void __user *buffer, size_t *length, loff_t *ppos)
L
Linus Torvalds 已提交
688
{
689
	proc_dointvec(table, write, file, buffer, length, ppos);
L
Linus Torvalds 已提交
690 691 692
	return 0;
}

693
static void do_laptop_sync(struct work_struct *work)
L
Linus Torvalds 已提交
694
{
695 696
	wakeup_flusher_threads(0);
	kfree(work);
L
Linus Torvalds 已提交
697 698 699 700
}

static void laptop_timer_fn(unsigned long unused)
{
701 702 703 704 705 706 707
	struct work_struct *work;

	work = kmalloc(sizeof(*work), GFP_ATOMIC);
	if (work) {
		INIT_WORK(work, do_laptop_sync);
		schedule_work(work);
	}
L
Linus Torvalds 已提交
708 709 710 711 712 713 714 715 716
}

/*
 * 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.
 */
void laptop_io_completion(void)
{
717
	mod_timer(&laptop_mode_wb_timer, jiffies + laptop_mode);
L
Linus Torvalds 已提交
718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746
}

/*
 * 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)
{
	del_timer(&laptop_mode_wb_timer);
}

/*
 * 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.
 */

747
void writeback_set_ratelimit(void)
L
Linus Torvalds 已提交
748
{
749
	ratelimit_pages = vm_total_pages / (num_online_cpus() * 32);
L
Linus Torvalds 已提交
750 751 752 753 754 755
	if (ratelimit_pages < 16)
		ratelimit_pages = 16;
	if (ratelimit_pages * PAGE_CACHE_SIZE > 4096 * 1024)
		ratelimit_pages = (4096 * 1024) / PAGE_CACHE_SIZE;
}

756
static int __cpuinit
L
Linus Torvalds 已提交
757 758
ratelimit_handler(struct notifier_block *self, unsigned long u, void *v)
{
759
	writeback_set_ratelimit();
760
	return NOTIFY_DONE;
L
Linus Torvalds 已提交
761 762
}

763
static struct notifier_block __cpuinitdata ratelimit_nb = {
L
Linus Torvalds 已提交
764 765 766 767 768
	.notifier_call	= ratelimit_handler,
	.next		= NULL,
};

/*
769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784
 * 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
785 786 787
 */
void __init page_writeback_init(void)
{
P
Peter Zijlstra 已提交
788 789
	int shift;

790
	writeback_set_ratelimit();
L
Linus Torvalds 已提交
791
	register_cpu_notifier(&ratelimit_nb);
P
Peter Zijlstra 已提交
792 793 794

	shift = calc_period_shift();
	prop_descriptor_init(&vm_completions, shift);
P
Peter Zijlstra 已提交
795
	prop_descriptor_init(&vm_dirties, shift);
L
Linus Torvalds 已提交
796 797
}

798
/**
799
 * write_cache_pages - walk the list of dirty pages of the given address space and write all of them.
800 801
 * @mapping: address space structure to write
 * @wbc: subtract the number of written pages from *@wbc->nr_to_write
802 803
 * @writepage: function called for each page
 * @data: data passed to writepage function
804
 *
805
 * If a page is already under I/O, write_cache_pages() skips it, even
806 807 808 809 810 811 812
 * 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.
 */
813 814 815
int write_cache_pages(struct address_space *mapping,
		      struct writeback_control *wbc, writepage_t writepage,
		      void *data)
816 817 818 819 820 821
{
	struct backing_dev_info *bdi = mapping->backing_dev_info;
	int ret = 0;
	int done = 0;
	struct pagevec pvec;
	int nr_pages;
N
Nick Piggin 已提交
822
	pgoff_t uninitialized_var(writeback_index);
823 824
	pgoff_t index;
	pgoff_t end;		/* Inclusive */
825
	pgoff_t done_index;
N
Nick Piggin 已提交
826
	int cycled;
827
	int range_whole = 0;
828
	long nr_to_write = wbc->nr_to_write;
829 830 831 832 833 834 835 836

	if (wbc->nonblocking && bdi_write_congested(bdi)) {
		wbc->encountered_congestion = 1;
		return 0;
	}

	pagevec_init(&pvec, 0);
	if (wbc->range_cyclic) {
N
Nick Piggin 已提交
837 838 839 840 841 842
		writeback_index = mapping->writeback_index; /* prev offset */
		index = writeback_index;
		if (index == 0)
			cycled = 1;
		else
			cycled = 0;
843 844 845 846 847 848
		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
849
		cycled = 1; /* ignore range_cyclic tests */
850 851
	}
retry:
852
	done_index = index;
N
Nick Piggin 已提交
853 854 855 856 857 858 859 860
	while (!done && (index <= end)) {
		int i;

		nr_pages = pagevec_lookup_tag(&pvec, mapping, &index,
			      PAGECACHE_TAG_DIRTY,
			      min(end - index, (pgoff_t)PAGEVEC_SIZE-1) + 1);
		if (nr_pages == 0)
			break;
861 862 863 864 865

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

			/*
866 867 868 869 870
			 * 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.
871
			 */
872 873 874 875 876 877 878 879 880 881 882
			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;

883 884
			lock_page(page);

N
Nick Piggin 已提交
885 886 887 888 889 890 891 892
			/*
			 * 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.
			 */
893
			if (unlikely(page->mapping != mapping)) {
N
Nick Piggin 已提交
894
continue_unlock:
895 896 897 898
				unlock_page(page);
				continue;
			}

899 900 901 902 903 904 905 906 907 908 909
			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;
			}
910

911 912
			BUG_ON(PageWriteback(page));
			if (!clear_page_dirty_for_io(page))
N
Nick Piggin 已提交
913
				goto continue_unlock;
914

915
			ret = (*writepage)(page, wbc, data);
916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934
			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;
				}
 			}

F
Federico Cuello 已提交
935
			if (nr_to_write > 0) {
936
				nr_to_write--;
F
Federico Cuello 已提交
937 938 939 940 941 942 943 944 945 946 947 948 949 950 951
				if (nr_to_write == 0 &&
				    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;
				}
952
			}
953

954 955 956
			if (wbc->nonblocking && bdi_write_congested(bdi)) {
				wbc->encountered_congestion = 1;
				done = 1;
957
				break;
958 959 960 961 962
			}
		}
		pagevec_release(&pvec);
		cond_resched();
	}
963
	if (!cycled && !done) {
964
		/*
N
Nick Piggin 已提交
965
		 * range_cyclic:
966 967 968
		 * 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
969
		cycled = 1;
970
		index = 0;
N
Nick Piggin 已提交
971
		end = writeback_index - 1;
972 973
		goto retry;
	}
974 975
	if (!wbc->no_nrwrite_index_update) {
		if (wbc->range_cyclic || (range_whole && nr_to_write > 0))
976
			mapping->writeback_index = done_index;
977 978
		wbc->nr_to_write = nr_to_write;
	}
979

980 981
	return ret;
}
982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013
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);
}
1014 1015 1016

EXPORT_SYMBOL(generic_writepages);

L
Linus Torvalds 已提交
1017 1018
int do_writepages(struct address_space *mapping, struct writeback_control *wbc)
{
1019 1020
	int ret;

L
Linus Torvalds 已提交
1021 1022
	if (wbc->nr_to_write <= 0)
		return 0;
1023
	wbc->for_writepages = 1;
L
Linus Torvalds 已提交
1024
	if (mapping->a_ops->writepages)
1025
		ret = mapping->a_ops->writepages(mapping, wbc);
1026 1027 1028 1029
	else
		ret = generic_writepages(mapping, wbc);
	wbc->for_writepages = 0;
	return ret;
L
Linus Torvalds 已提交
1030 1031 1032 1033
}

/**
 * write_one_page - write out a single page and optionally wait on I/O
1034 1035
 * @page: the page to write
 * @wait: if true, wait on writeout
L
Linus Torvalds 已提交
1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070
 *
 * 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);

1071 1072 1073 1074 1075 1076 1077 1078 1079 1080
/*
 * 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;
}

1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094
/*
 * 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
1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107
/*
 * 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
1108
 * mapping by re-checking page_mapping() inside tree_lock.
L
Linus Torvalds 已提交
1109 1110 1111 1112 1113 1114 1115
 */
int __set_page_dirty_nobuffers(struct page *page)
{
	if (!TestSetPageDirty(page)) {
		struct address_space *mapping = page_mapping(page);
		struct address_space *mapping2;

1116 1117 1118
		if (!mapping)
			return 1;

N
Nick Piggin 已提交
1119
		spin_lock_irq(&mapping->tree_lock);
1120 1121 1122
		mapping2 = page_mapping(page);
		if (mapping2) { /* Race with truncate? */
			BUG_ON(mapping2 != mapping);
1123
			WARN_ON_ONCE(!PagePrivate(page) && !PageUptodate(page));
1124
			account_page_dirtied(page, mapping);
1125 1126 1127
			radix_tree_tag_set(&mapping->page_tree,
				page_index(page), PAGECACHE_TAG_DIRTY);
		}
N
Nick Piggin 已提交
1128
		spin_unlock_irq(&mapping->tree_lock);
1129 1130 1131
		if (mapping->host) {
			/* !PageAnon && !swapper_space */
			__mark_inode_dirty(mapping->host, I_DIRTY_PAGES);
L
Linus Torvalds 已提交
1132
		}
1133
		return 1;
L
Linus Torvalds 已提交
1134
	}
1135
	return 0;
L
Linus Torvalds 已提交
1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154
}
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);

/*
 * 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
1155
int set_page_dirty(struct page *page)
L
Linus Torvalds 已提交
1156 1157 1158 1159 1160
{
	struct address_space *mapping = page_mapping(page);

	if (likely(mapping)) {
		int (*spd)(struct page *) = mapping->a_ops->set_page_dirty;
1161 1162 1163 1164 1165
#ifdef CONFIG_BLOCK
		if (!spd)
			spd = __set_page_dirty_buffers;
#endif
		return (*spd)(page);
L
Linus Torvalds 已提交
1166
	}
1167 1168 1169 1170
	if (!PageDirty(page)) {
		if (!TestSetPageDirty(page))
			return 1;
	}
L
Linus Torvalds 已提交
1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188
	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;

1189
	lock_page_nosync(page);
L
Linus Torvalds 已提交
1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213
	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);

1214 1215
	BUG_ON(!PageLocked(page));

1216
	ClearPageReclaim(page);
1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244
	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);
1245 1246 1247 1248 1249 1250 1251 1252 1253 1254
		/*
		 * 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.
		 */
1255
		if (TestClearPageDirty(page)) {
1256
			dec_zone_page_state(page, NR_FILE_DIRTY);
1257 1258
			dec_bdi_stat(mapping->backing_dev_info,
					BDI_RECLAIMABLE);
1259
			return 1;
L
Linus Torvalds 已提交
1260
		}
1261
		return 0;
L
Linus Torvalds 已提交
1262
	}
1263
	return TestClearPageDirty(page);
L
Linus Torvalds 已提交
1264
}
1265
EXPORT_SYMBOL(clear_page_dirty_for_io);
L
Linus Torvalds 已提交
1266 1267 1268 1269 1270 1271 1272

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
1273
		struct backing_dev_info *bdi = mapping->backing_dev_info;
L
Linus Torvalds 已提交
1274 1275
		unsigned long flags;

N
Nick Piggin 已提交
1276
		spin_lock_irqsave(&mapping->tree_lock, flags);
L
Linus Torvalds 已提交
1277
		ret = TestClearPageWriteback(page);
P
Peter Zijlstra 已提交
1278
		if (ret) {
L
Linus Torvalds 已提交
1279 1280 1281
			radix_tree_tag_clear(&mapping->page_tree,
						page_index(page),
						PAGECACHE_TAG_WRITEBACK);
1282
			if (bdi_cap_account_writeback(bdi)) {
P
Peter Zijlstra 已提交
1283
				__dec_bdi_stat(bdi, BDI_WRITEBACK);
P
Peter Zijlstra 已提交
1284 1285
				__bdi_writeout_inc(bdi);
			}
P
Peter Zijlstra 已提交
1286
		}
N
Nick Piggin 已提交
1287
		spin_unlock_irqrestore(&mapping->tree_lock, flags);
L
Linus Torvalds 已提交
1288 1289 1290
	} else {
		ret = TestClearPageWriteback(page);
	}
1291 1292
	if (ret)
		dec_zone_page_state(page, NR_WRITEBACK);
L
Linus Torvalds 已提交
1293 1294 1295 1296 1297 1298 1299 1300 1301
	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
1302
		struct backing_dev_info *bdi = mapping->backing_dev_info;
L
Linus Torvalds 已提交
1303 1304
		unsigned long flags;

N
Nick Piggin 已提交
1305
		spin_lock_irqsave(&mapping->tree_lock, flags);
L
Linus Torvalds 已提交
1306
		ret = TestSetPageWriteback(page);
P
Peter Zijlstra 已提交
1307
		if (!ret) {
L
Linus Torvalds 已提交
1308 1309 1310
			radix_tree_tag_set(&mapping->page_tree,
						page_index(page),
						PAGECACHE_TAG_WRITEBACK);
1311
			if (bdi_cap_account_writeback(bdi))
P
Peter Zijlstra 已提交
1312 1313
				__inc_bdi_stat(bdi, BDI_WRITEBACK);
		}
L
Linus Torvalds 已提交
1314 1315 1316 1317
		if (!PageDirty(page))
			radix_tree_tag_clear(&mapping->page_tree,
						page_index(page),
						PAGECACHE_TAG_DIRTY);
N
Nick Piggin 已提交
1318
		spin_unlock_irqrestore(&mapping->tree_lock, flags);
L
Linus Torvalds 已提交
1319 1320 1321
	} else {
		ret = TestSetPageWriteback(page);
	}
1322 1323
	if (!ret)
		inc_zone_page_state(page, NR_WRITEBACK);
L
Linus Torvalds 已提交
1324 1325 1326 1327 1328 1329
	return ret;

}
EXPORT_SYMBOL(test_set_page_writeback);

/*
N
Nick Piggin 已提交
1330
 * Return true if any of the pages in the mapping are marked with the
L
Linus Torvalds 已提交
1331 1332 1333 1334 1335
 * passed tag.
 */
int mapping_tagged(struct address_space *mapping, int tag)
{
	int ret;
N
Nick Piggin 已提交
1336
	rcu_read_lock();
L
Linus Torvalds 已提交
1337
	ret = radix_tree_tagged(&mapping->page_tree, tag);
N
Nick Piggin 已提交
1338
	rcu_read_unlock();
L
Linus Torvalds 已提交
1339 1340 1341
	return ret;
}
EXPORT_SYMBOL(mapping_tagged);