fs-writeback.c 67.2 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10
/*
 * fs/fs-writeback.c
 *
 * Copyright (C) 2002, Linus Torvalds.
 *
 * Contains all the functions related to writing back and waiting
 * upon dirty inodes against superblocks, and writing back dirty
 * pages against inodes.  ie: data writeback.  Writeout of the
 * inode itself is not handled here.
 *
11
 * 10Apr2002	Andrew Morton
L
Linus Torvalds 已提交
12 13 14 15 16
 *		Split out of fs/inode.c
 *		Additions for address_space-based writeback
 */

#include <linux/kernel.h>
17
#include <linux/export.h>
L
Linus Torvalds 已提交
18
#include <linux/spinlock.h>
19
#include <linux/slab.h>
L
Linus Torvalds 已提交
20 21 22
#include <linux/sched.h>
#include <linux/fs.h>
#include <linux/mm.h>
23
#include <linux/pagemap.h>
24
#include <linux/kthread.h>
L
Linus Torvalds 已提交
25 26 27
#include <linux/writeback.h>
#include <linux/blkdev.h>
#include <linux/backing-dev.h>
28
#include <linux/tracepoint.h>
29
#include <linux/device.h>
30
#include <linux/memcontrol.h>
31
#include "internal.h"
L
Linus Torvalds 已提交
32

33 34 35 36 37
/*
 * 4MB minimal write chunk size
 */
#define MIN_WRITEBACK_PAGES	(4096UL >> (PAGE_CACHE_SHIFT - 10))

38 39 40 41
struct wb_completion {
	atomic_t		cnt;
};

42 43 44
/*
 * Passed into wb_writeback(), essentially a subset of writeback_control
 */
45
struct wb_writeback_work {
46 47
	long nr_pages;
	struct super_block *sb;
48
	unsigned long *older_than_this;
49
	enum writeback_sync_modes sync_mode;
50
	unsigned int tagged_writepages:1;
51 52 53
	unsigned int for_kupdate:1;
	unsigned int range_cyclic:1;
	unsigned int for_background:1;
54
	unsigned int for_sync:1;	/* sync(2) WB_SYNC_ALL writeback */
55
	unsigned int auto_free:1;	/* free on completion */
56
	enum wb_reason reason;		/* why was writeback initiated? */
57

58
	struct list_head list;		/* pending work list */
59
	struct wb_completion *done;	/* set if the caller waits */
60 61
};

62 63 64 65 66 67 68 69 70 71 72 73 74
/*
 * If one wants to wait for one or more wb_writeback_works, each work's
 * ->done should be set to a wb_completion defined using the following
 * macro.  Once all work items are issued with wb_queue_work(), the caller
 * can wait for the completion of all using wb_wait_for_completion().  Work
 * items which are waited upon aren't freed automatically on completion.
 */
#define DEFINE_WB_COMPLETION_ONSTACK(cmpl)				\
	struct wb_completion cmpl = {					\
		.cnt		= ATOMIC_INIT(1),			\
	}


75 76 77 78 79 80 81 82 83 84 85 86
/*
 * If an inode is constantly having its pages dirtied, but then the
 * updates stop dirtytime_expire_interval seconds in the past, it's
 * possible for the worst case time between when an inode has its
 * timestamps updated and when they finally get written out to be two
 * dirtytime_expire_intervals.  We set the default to 12 hours (in
 * seconds), which means most of the time inodes will have their
 * timestamps written to disk after 12 hours, but in the worst case a
 * few inodes might not their timestamps updated for 24 hours.
 */
unsigned int dirtytime_expire_interval = 12 * 60 * 60;

N
Nick Piggin 已提交
87 88
static inline struct inode *wb_inode(struct list_head *head)
{
89
	return list_entry(head, struct inode, i_io_list);
N
Nick Piggin 已提交
90 91
}

92 93 94 95 96 97 98 99
/*
 * Include the creation of the trace points after defining the
 * wb_writeback_work structure and inline functions so that the definition
 * remains local to this file.
 */
#define CREATE_TRACE_POINTS
#include <trace/events/writeback.h>

100 101
EXPORT_TRACEPOINT_SYMBOL_GPL(wbc_writepage);

102 103 104 105 106 107
static bool wb_io_lists_populated(struct bdi_writeback *wb)
{
	if (wb_has_dirty_io(wb)) {
		return false;
	} else {
		set_bit(WB_has_dirty_io, &wb->state);
108
		WARN_ON_ONCE(!wb->avg_write_bandwidth);
109 110
		atomic_long_add(wb->avg_write_bandwidth,
				&wb->bdi->tot_write_bandwidth);
111 112 113 114 115 116 117
		return true;
	}
}

static void wb_io_lists_depopulated(struct bdi_writeback *wb)
{
	if (wb_has_dirty_io(wb) && list_empty(&wb->b_dirty) &&
118
	    list_empty(&wb->b_io) && list_empty(&wb->b_more_io)) {
119
		clear_bit(WB_has_dirty_io, &wb->state);
120 121
		WARN_ON_ONCE(atomic_long_sub_return(wb->avg_write_bandwidth,
					&wb->bdi->tot_write_bandwidth) < 0);
122
	}
123 124 125
}

/**
126
 * inode_io_list_move_locked - move an inode onto a bdi_writeback IO list
127 128 129 130
 * @inode: inode to be moved
 * @wb: target bdi_writeback
 * @head: one of @wb->b_{dirty|io|more_io}
 *
131
 * Move @inode->i_io_list to @list of @wb and set %WB_has_dirty_io.
132 133 134
 * Returns %true if @inode is the first occupant of the !dirty_time IO
 * lists; otherwise, %false.
 */
135
static bool inode_io_list_move_locked(struct inode *inode,
136 137 138 139 140
				      struct bdi_writeback *wb,
				      struct list_head *head)
{
	assert_spin_locked(&wb->list_lock);

141
	list_move(&inode->i_io_list, head);
142 143 144 145 146 147 148 149 150 151

	/* dirty_time doesn't count as dirty_io until expiration */
	if (head != &wb->b_dirty_time)
		return wb_io_lists_populated(wb);

	wb_io_lists_depopulated(wb);
	return false;
}

/**
152
 * inode_io_list_del_locked - remove an inode from its bdi_writeback IO list
153 154 155 156 157 158
 * @inode: inode to be removed
 * @wb: bdi_writeback @inode is being removed from
 *
 * Remove @inode which may be on one of @wb->b_{dirty|io|more_io} lists and
 * clear %WB_has_dirty_io if all are empty afterwards.
 */
159
static void inode_io_list_del_locked(struct inode *inode,
160 161 162 163
				     struct bdi_writeback *wb)
{
	assert_spin_locked(&wb->list_lock);

164
	list_del_init(&inode->i_io_list);
165 166 167
	wb_io_lists_depopulated(wb);
}

168
static void wb_wakeup(struct bdi_writeback *wb)
J
Jan Kara 已提交
169
{
170 171 172 173
	spin_lock_bh(&wb->work_lock);
	if (test_bit(WB_registered, &wb->state))
		mod_delayed_work(bdi_wq, &wb->dwork, 0);
	spin_unlock_bh(&wb->work_lock);
J
Jan Kara 已提交
174 175
}

176 177
static void wb_queue_work(struct bdi_writeback *wb,
			  struct wb_writeback_work *work)
178
{
179
	trace_writeback_queue(wb, work);
180

181
	spin_lock_bh(&wb->work_lock);
182
	if (!test_bit(WB_registered, &wb->state))
J
Jan Kara 已提交
183
		goto out_unlock;
184 185
	if (work->done)
		atomic_inc(&work->done->cnt);
186 187
	list_add_tail(&work->list, &wb->work_list);
	mod_delayed_work(bdi_wq, &wb->dwork, 0);
J
Jan Kara 已提交
188
out_unlock:
189
	spin_unlock_bh(&wb->work_lock);
L
Linus Torvalds 已提交
190 191
}

192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209
/**
 * wb_wait_for_completion - wait for completion of bdi_writeback_works
 * @bdi: bdi work items were issued to
 * @done: target wb_completion
 *
 * Wait for one or more work items issued to @bdi with their ->done field
 * set to @done, which should have been defined with
 * DEFINE_WB_COMPLETION_ONSTACK().  This function returns after all such
 * work items are completed.  Work items which are waited upon aren't freed
 * automatically on completion.
 */
static void wb_wait_for_completion(struct backing_dev_info *bdi,
				   struct wb_completion *done)
{
	atomic_dec(&done->cnt);		/* put down the initial count */
	wait_event(bdi->wb_waitq, !atomic_read(&done->cnt));
}

210 211
#ifdef CONFIG_CGROUP_WRITEBACK

212 213 214 215 216 217 218 219 220 221 222 223 224 225
/* parameters for foreign inode detection, see wb_detach_inode() */
#define WB_FRN_TIME_SHIFT	13	/* 1s = 2^13, upto 8 secs w/ 16bit */
#define WB_FRN_TIME_AVG_SHIFT	3	/* avg = avg * 7/8 + new * 1/8 */
#define WB_FRN_TIME_CUT_DIV	2	/* ignore rounds < avg / 2 */
#define WB_FRN_TIME_PERIOD	(2 * (1 << WB_FRN_TIME_SHIFT))	/* 2s */

#define WB_FRN_HIST_SLOTS	16	/* inode->i_wb_frn_history is 16bit */
#define WB_FRN_HIST_UNIT	(WB_FRN_TIME_PERIOD / WB_FRN_HIST_SLOTS)
					/* each slot's duration is 2s / 16 */
#define WB_FRN_HIST_THR_SLOTS	(WB_FRN_HIST_SLOTS / 2)
					/* if foreign slots >= 8, switch */
#define WB_FRN_HIST_MAX_SLOTS	(WB_FRN_HIST_THR_SLOTS / 2 + 1)
					/* one round can affect upto 5 slots */

226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255
void __inode_attach_wb(struct inode *inode, struct page *page)
{
	struct backing_dev_info *bdi = inode_to_bdi(inode);
	struct bdi_writeback *wb = NULL;

	if (inode_cgwb_enabled(inode)) {
		struct cgroup_subsys_state *memcg_css;

		if (page) {
			memcg_css = mem_cgroup_css_from_page(page);
			wb = wb_get_create(bdi, memcg_css, GFP_ATOMIC);
		} else {
			/* must pin memcg_css, see wb_get_create() */
			memcg_css = task_get_css(current, memory_cgrp_id);
			wb = wb_get_create(bdi, memcg_css, GFP_ATOMIC);
			css_put(memcg_css);
		}
	}

	if (!wb)
		wb = &bdi->wb;

	/*
	 * There may be multiple instances of this function racing to
	 * update the same inode.  Use cmpxchg() to tell the winner.
	 */
	if (unlikely(cmpxchg(&inode->i_wb, NULL, wb)))
		wb_put(wb);
}

256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282
/**
 * locked_inode_to_wb_and_lock_list - determine a locked inode's wb and lock it
 * @inode: inode of interest with i_lock held
 *
 * Returns @inode's wb with its list_lock held.  @inode->i_lock must be
 * held on entry and is released on return.  The returned wb is guaranteed
 * to stay @inode's associated wb until its list_lock is released.
 */
static struct bdi_writeback *
locked_inode_to_wb_and_lock_list(struct inode *inode)
	__releases(&inode->i_lock)
	__acquires(&wb->list_lock)
{
	while (true) {
		struct bdi_writeback *wb = inode_to_wb(inode);

		/*
		 * inode_to_wb() association is protected by both
		 * @inode->i_lock and @wb->list_lock but list_lock nests
		 * outside i_lock.  Drop i_lock and verify that the
		 * association hasn't changed after acquiring list_lock.
		 */
		wb_get(wb);
		spin_unlock(&inode->i_lock);
		spin_lock(&wb->list_lock);
		wb_put(wb);		/* not gonna deref it anymore */

283 284
		/* i_wb may have changed inbetween, can't use inode_to_wb() */
		if (likely(wb == inode->i_wb))
285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306
			return wb;	/* @inode already has ref */

		spin_unlock(&wb->list_lock);
		cpu_relax();
		spin_lock(&inode->i_lock);
	}
}

/**
 * inode_to_wb_and_lock_list - determine an inode's wb and lock it
 * @inode: inode of interest
 *
 * Same as locked_inode_to_wb_and_lock_list() but @inode->i_lock isn't held
 * on entry.
 */
static struct bdi_writeback *inode_to_wb_and_lock_list(struct inode *inode)
	__acquires(&wb->list_lock)
{
	spin_lock(&inode->i_lock);
	return locked_inode_to_wb_and_lock_list(inode);
}

307 308 309 310 311 312 313 314 315 316 317 318 319
struct inode_switch_wbs_context {
	struct inode		*inode;
	struct bdi_writeback	*new_wb;

	struct rcu_head		rcu_head;
	struct work_struct	work;
};

static void inode_switch_wbs_work_fn(struct work_struct *work)
{
	struct inode_switch_wbs_context *isw =
		container_of(work, struct inode_switch_wbs_context, work);
	struct inode *inode = isw->inode;
320 321
	struct address_space *mapping = inode->i_mapping;
	struct bdi_writeback *old_wb = inode->i_wb;
322
	struct bdi_writeback *new_wb = isw->new_wb;
323 324 325
	struct radix_tree_iter iter;
	bool switched = false;
	void **slot;
326 327 328 329 330 331

	/*
	 * By the time control reaches here, RCU grace period has passed
	 * since I_WB_SWITCH assertion and all wb stat update transactions
	 * between unlocked_inode_to_wb_begin/end() are guaranteed to be
	 * synchronizing against mapping->tree_lock.
332 333 334 335
	 *
	 * Grabbing old_wb->list_lock, inode->i_lock and mapping->tree_lock
	 * gives us exclusion against all wb related operations on @inode
	 * including IO list manipulations and stat updates.
336
	 */
337 338 339 340 341 342 343
	if (old_wb < new_wb) {
		spin_lock(&old_wb->list_lock);
		spin_lock_nested(&new_wb->list_lock, SINGLE_DEPTH_NESTING);
	} else {
		spin_lock(&new_wb->list_lock);
		spin_lock_nested(&old_wb->list_lock, SINGLE_DEPTH_NESTING);
	}
344
	spin_lock(&inode->i_lock);
345 346 347 348
	spin_lock_irq(&mapping->tree_lock);

	/*
	 * Once I_FREEING is visible under i_lock, the eviction path owns
349
	 * the inode and we shouldn't modify ->i_io_list.
350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387
	 */
	if (unlikely(inode->i_state & I_FREEING))
		goto skip_switch;

	/*
	 * Count and transfer stats.  Note that PAGECACHE_TAG_DIRTY points
	 * to possibly dirty pages while PAGECACHE_TAG_WRITEBACK points to
	 * pages actually under underwriteback.
	 */
	radix_tree_for_each_tagged(slot, &mapping->page_tree, &iter, 0,
				   PAGECACHE_TAG_DIRTY) {
		struct page *page = radix_tree_deref_slot_protected(slot,
							&mapping->tree_lock);
		if (likely(page) && PageDirty(page)) {
			__dec_wb_stat(old_wb, WB_RECLAIMABLE);
			__inc_wb_stat(new_wb, WB_RECLAIMABLE);
		}
	}

	radix_tree_for_each_tagged(slot, &mapping->page_tree, &iter, 0,
				   PAGECACHE_TAG_WRITEBACK) {
		struct page *page = radix_tree_deref_slot_protected(slot,
							&mapping->tree_lock);
		if (likely(page)) {
			WARN_ON_ONCE(!PageWriteback(page));
			__dec_wb_stat(old_wb, WB_WRITEBACK);
			__inc_wb_stat(new_wb, WB_WRITEBACK);
		}
	}

	wb_get(new_wb);

	/*
	 * Transfer to @new_wb's IO list if necessary.  The specific list
	 * @inode was on is ignored and the inode is put on ->b_dirty which
	 * is always correct including from ->b_dirty_time.  The transfer
	 * preserves @inode->dirtied_when ordering.
	 */
388
	if (!list_empty(&inode->i_io_list)) {
389 390
		struct inode *pos;

391
		inode_io_list_del_locked(inode, old_wb);
392
		inode->i_wb = new_wb;
393
		list_for_each_entry(pos, &new_wb->b_dirty, i_io_list)
394 395 396
			if (time_after_eq(inode->dirtied_when,
					  pos->dirtied_when))
				break;
397
		inode_io_list_move_locked(inode, new_wb, pos->i_io_list.prev);
398 399 400
	} else {
		inode->i_wb = new_wb;
	}
401

402
	/* ->i_wb_frn updates may race wbc_detach_inode() but doesn't matter */
403 404 405
	inode->i_wb_frn_winner = 0;
	inode->i_wb_frn_avg_time = 0;
	inode->i_wb_frn_history = 0;
406 407
	switched = true;
skip_switch:
408 409 410 411 412 413
	/*
	 * Paired with load_acquire in unlocked_inode_to_wb_begin() and
	 * ensures that the new wb is visible if they see !I_WB_SWITCH.
	 */
	smp_store_release(&inode->i_state, inode->i_state & ~I_WB_SWITCH);

414
	spin_unlock_irq(&mapping->tree_lock);
415
	spin_unlock(&inode->i_lock);
416 417
	spin_unlock(&new_wb->list_lock);
	spin_unlock(&old_wb->list_lock);
418

419 420 421 422
	if (switched) {
		wb_wakeup(new_wb);
		wb_put(old_wb);
	}
423
	wb_put(new_wb);
424 425

	iput(inode);
426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497
	kfree(isw);
}

static void inode_switch_wbs_rcu_fn(struct rcu_head *rcu_head)
{
	struct inode_switch_wbs_context *isw = container_of(rcu_head,
				struct inode_switch_wbs_context, rcu_head);

	/* needs to grab bh-unsafe locks, bounce to work item */
	INIT_WORK(&isw->work, inode_switch_wbs_work_fn);
	schedule_work(&isw->work);
}

/**
 * inode_switch_wbs - change the wb association of an inode
 * @inode: target inode
 * @new_wb_id: ID of the new wb
 *
 * Switch @inode's wb association to the wb identified by @new_wb_id.  The
 * switching is performed asynchronously and may fail silently.
 */
static void inode_switch_wbs(struct inode *inode, int new_wb_id)
{
	struct backing_dev_info *bdi = inode_to_bdi(inode);
	struct cgroup_subsys_state *memcg_css;
	struct inode_switch_wbs_context *isw;

	/* noop if seems to be already in progress */
	if (inode->i_state & I_WB_SWITCH)
		return;

	isw = kzalloc(sizeof(*isw), GFP_ATOMIC);
	if (!isw)
		return;

	/* find and pin the new wb */
	rcu_read_lock();
	memcg_css = css_from_id(new_wb_id, &memory_cgrp_subsys);
	if (memcg_css)
		isw->new_wb = wb_get_create(bdi, memcg_css, GFP_ATOMIC);
	rcu_read_unlock();
	if (!isw->new_wb)
		goto out_free;

	/* while holding I_WB_SWITCH, no one else can update the association */
	spin_lock(&inode->i_lock);
	if (inode->i_state & (I_WB_SWITCH | I_FREEING) ||
	    inode_to_wb(inode) == isw->new_wb) {
		spin_unlock(&inode->i_lock);
		goto out_free;
	}
	inode->i_state |= I_WB_SWITCH;
	spin_unlock(&inode->i_lock);

	ihold(inode);
	isw->inode = inode;

	/*
	 * In addition to synchronizing among switchers, I_WB_SWITCH tells
	 * the RCU protected stat update paths to grab the mapping's
	 * tree_lock so that stat transfer can synchronize against them.
	 * Let's continue after I_WB_SWITCH is guaranteed to be visible.
	 */
	call_rcu(&isw->rcu_head, inode_switch_wbs_rcu_fn);
	return;

out_free:
	if (isw->new_wb)
		wb_put(isw->new_wb);
	kfree(isw);
}

498 499 500 501 502 503 504 505 506 507 508 509 510
/**
 * wbc_attach_and_unlock_inode - associate wbc with target inode and unlock it
 * @wbc: writeback_control of interest
 * @inode: target inode
 *
 * @inode is locked and about to be written back under the control of @wbc.
 * Record @inode's writeback context into @wbc and unlock the i_lock.  On
 * writeback completion, wbc_detach_inode() should be called.  This is used
 * to track the cgroup writeback context.
 */
void wbc_attach_and_unlock_inode(struct writeback_control *wbc,
				 struct inode *inode)
{
511 512 513 514 515
	if (!inode_cgwb_enabled(inode)) {
		spin_unlock(&inode->i_lock);
		return;
	}

516
	wbc->wb = inode_to_wb(inode);
517 518 519 520 521 522 523 524 525
	wbc->inode = inode;

	wbc->wb_id = wbc->wb->memcg_css->id;
	wbc->wb_lcand_id = inode->i_wb_frn_winner;
	wbc->wb_tcand_id = 0;
	wbc->wb_bytes = 0;
	wbc->wb_lcand_bytes = 0;
	wbc->wb_tcand_bytes = 0;

526 527
	wb_get(wbc->wb);
	spin_unlock(&inode->i_lock);
528 529 530 531 532 533 534

	/*
	 * A dying wb indicates that the memcg-blkcg mapping has changed
	 * and a new wb is already serving the memcg.  Switch immediately.
	 */
	if (unlikely(wb_dying(wbc->wb)))
		inode_switch_wbs(inode, wbc->wb_id);
535 536 537
}

/**
538 539
 * wbc_detach_inode - disassociate wbc from inode and perform foreign detection
 * @wbc: writeback_control of the just finished writeback
540 541 542
 *
 * To be called after a writeback attempt of an inode finishes and undoes
 * wbc_attach_and_unlock_inode().  Can be called under any context.
543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572
 *
 * As concurrent write sharing of an inode is expected to be very rare and
 * memcg only tracks page ownership on first-use basis severely confining
 * the usefulness of such sharing, cgroup writeback tracks ownership
 * per-inode.  While the support for concurrent write sharing of an inode
 * is deemed unnecessary, an inode being written to by different cgroups at
 * different points in time is a lot more common, and, more importantly,
 * charging only by first-use can too readily lead to grossly incorrect
 * behaviors (single foreign page can lead to gigabytes of writeback to be
 * incorrectly attributed).
 *
 * To resolve this issue, cgroup writeback detects the majority dirtier of
 * an inode and transfers the ownership to it.  To avoid unnnecessary
 * oscillation, the detection mechanism keeps track of history and gives
 * out the switch verdict only if the foreign usage pattern is stable over
 * a certain amount of time and/or writeback attempts.
 *
 * On each writeback attempt, @wbc tries to detect the majority writer
 * using Boyer-Moore majority vote algorithm.  In addition to the byte
 * count from the majority voting, it also counts the bytes written for the
 * current wb and the last round's winner wb (max of last round's current
 * wb, the winner from two rounds ago, and the last round's majority
 * candidate).  Keeping track of the historical winner helps the algorithm
 * to semi-reliably detect the most active writer even when it's not the
 * absolute majority.
 *
 * Once the winner of the round is determined, whether the winner is
 * foreign or not and how much IO time the round consumed is recorded in
 * inode->i_wb_frn_history.  If the amount of recorded foreign IO time is
 * over a certain threshold, the switch verdict is given.
573 574 575
 */
void wbc_detach_inode(struct writeback_control *wbc)
{
576 577
	struct bdi_writeback *wb = wbc->wb;
	struct inode *inode = wbc->inode;
578 579
	unsigned long avg_time, max_bytes, max_time;
	u16 history;
580 581
	int max_id;

582 583 584 585 586 587
	if (!wb)
		return;

	history = inode->i_wb_frn_history;
	avg_time = inode->i_wb_frn_avg_time;

588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639
	/* pick the winner of this round */
	if (wbc->wb_bytes >= wbc->wb_lcand_bytes &&
	    wbc->wb_bytes >= wbc->wb_tcand_bytes) {
		max_id = wbc->wb_id;
		max_bytes = wbc->wb_bytes;
	} else if (wbc->wb_lcand_bytes >= wbc->wb_tcand_bytes) {
		max_id = wbc->wb_lcand_id;
		max_bytes = wbc->wb_lcand_bytes;
	} else {
		max_id = wbc->wb_tcand_id;
		max_bytes = wbc->wb_tcand_bytes;
	}

	/*
	 * Calculate the amount of IO time the winner consumed and fold it
	 * into the running average kept per inode.  If the consumed IO
	 * time is lower than avag / WB_FRN_TIME_CUT_DIV, ignore it for
	 * deciding whether to switch or not.  This is to prevent one-off
	 * small dirtiers from skewing the verdict.
	 */
	max_time = DIV_ROUND_UP((max_bytes >> PAGE_SHIFT) << WB_FRN_TIME_SHIFT,
				wb->avg_write_bandwidth);
	if (avg_time)
		avg_time += (max_time >> WB_FRN_TIME_AVG_SHIFT) -
			    (avg_time >> WB_FRN_TIME_AVG_SHIFT);
	else
		avg_time = max_time;	/* immediate catch up on first run */

	if (max_time >= avg_time / WB_FRN_TIME_CUT_DIV) {
		int slots;

		/*
		 * The switch verdict is reached if foreign wb's consume
		 * more than a certain proportion of IO time in a
		 * WB_FRN_TIME_PERIOD.  This is loosely tracked by 16 slot
		 * history mask where each bit represents one sixteenth of
		 * the period.  Determine the number of slots to shift into
		 * history from @max_time.
		 */
		slots = min(DIV_ROUND_UP(max_time, WB_FRN_HIST_UNIT),
			    (unsigned long)WB_FRN_HIST_MAX_SLOTS);
		history <<= slots;
		if (wbc->wb_id != max_id)
			history |= (1U << slots) - 1;

		/*
		 * Switch if the current wb isn't the consistent winner.
		 * If there are multiple closely competing dirtiers, the
		 * inode may switch across them repeatedly over time, which
		 * is okay.  The main goal is avoiding keeping an inode on
		 * the wrong wb for an extended period of time.
		 */
640 641
		if (hweight32(history) > WB_FRN_HIST_THR_SLOTS)
			inode_switch_wbs(inode, max_id);
642 643 644 645 646 647 648 649 650 651
	}

	/*
	 * Multiple instances of this function may race to update the
	 * following fields but we don't mind occassional inaccuracies.
	 */
	inode->i_wb_frn_winner = max_id;
	inode->i_wb_frn_avg_time = min(avg_time, (unsigned long)U16_MAX);
	inode->i_wb_frn_history = history;

652 653 654 655
	wb_put(wbc->wb);
	wbc->wb = NULL;
}

656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699
/**
 * wbc_account_io - account IO issued during writeback
 * @wbc: writeback_control of the writeback in progress
 * @page: page being written out
 * @bytes: number of bytes being written out
 *
 * @bytes from @page are about to written out during the writeback
 * controlled by @wbc.  Keep the book for foreign inode detection.  See
 * wbc_detach_inode().
 */
void wbc_account_io(struct writeback_control *wbc, struct page *page,
		    size_t bytes)
{
	int id;

	/*
	 * pageout() path doesn't attach @wbc to the inode being written
	 * out.  This is intentional as we don't want the function to block
	 * behind a slow cgroup.  Ultimately, we want pageout() to kick off
	 * regular writeback instead of writing things out itself.
	 */
	if (!wbc->wb)
		return;

	rcu_read_lock();
	id = mem_cgroup_css_from_page(page)->id;
	rcu_read_unlock();

	if (id == wbc->wb_id) {
		wbc->wb_bytes += bytes;
		return;
	}

	if (id == wbc->wb_lcand_id)
		wbc->wb_lcand_bytes += bytes;

	/* Boyer-Moore majority vote algorithm */
	if (!wbc->wb_tcand_bytes)
		wbc->wb_tcand_id = id;
	if (id == wbc->wb_tcand_id)
		wbc->wb_tcand_bytes += bytes;
	else
		wbc->wb_tcand_bytes -= min(bytes, wbc->wb_tcand_bytes);
}
700
EXPORT_SYMBOL_GPL(wbc_account_io);
701

702 703
/**
 * inode_congested - test whether an inode is congested
704
 * @inode: inode to test for congestion (may be NULL)
705 706 707 708 709 710 711 712 713
 * @cong_bits: mask of WB_[a]sync_congested bits to test
 *
 * Tests whether @inode is congested.  @cong_bits is the mask of congestion
 * bits to test and the return value is the mask of set bits.
 *
 * If cgroup writeback is enabled for @inode, the congestion state is
 * determined by whether the cgwb (cgroup bdi_writeback) for the blkcg
 * associated with @inode is congested; otherwise, the root wb's congestion
 * state is used.
714 715 716
 *
 * @inode is allowed to be NULL as this function is often called on
 * mapping->host which is NULL for the swapper space.
717 718 719
 */
int inode_congested(struct inode *inode, int cong_bits)
{
720 721 722 723
	/*
	 * Once set, ->i_wb never becomes NULL while the inode is alive.
	 * Start transaction iff ->i_wb is visible.
	 */
724
	if (inode && inode_to_wb_is_valid(inode)) {
725 726 727 728 729 730 731
		struct bdi_writeback *wb;
		bool locked, congested;

		wb = unlocked_inode_to_wb_begin(inode, &locked);
		congested = wb_congested(wb, cong_bits);
		unlocked_inode_to_wb_end(inode, locked);
		return congested;
732 733 734 735 736 737
	}

	return wb_congested(&inode_to_bdi(inode)->wb, cong_bits);
}
EXPORT_SYMBOL_GPL(inode_congested);

738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765
/**
 * wb_split_bdi_pages - split nr_pages to write according to bandwidth
 * @wb: target bdi_writeback to split @nr_pages to
 * @nr_pages: number of pages to write for the whole bdi
 *
 * Split @wb's portion of @nr_pages according to @wb's write bandwidth in
 * relation to the total write bandwidth of all wb's w/ dirty inodes on
 * @wb->bdi.
 */
static long wb_split_bdi_pages(struct bdi_writeback *wb, long nr_pages)
{
	unsigned long this_bw = wb->avg_write_bandwidth;
	unsigned long tot_bw = atomic_long_read(&wb->bdi->tot_write_bandwidth);

	if (nr_pages == LONG_MAX)
		return LONG_MAX;

	/*
	 * This may be called on clean wb's and proportional distribution
	 * may not make sense, just use the original @nr_pages in those
	 * cases.  In general, we wanna err on the side of writing more.
	 */
	if (!tot_bw || this_bw >= tot_bw)
		return nr_pages;
	else
		return DIV_ROUND_UP_ULL((u64)nr_pages * this_bw, tot_bw);
}

766 767 768 769 770 771 772 773 774 775 776 777 778 779 780
/**
 * bdi_split_work_to_wbs - split a wb_writeback_work to all wb's of a bdi
 * @bdi: target backing_dev_info
 * @base_work: wb_writeback_work to issue
 * @skip_if_busy: skip wb's which already have writeback in progress
 *
 * Split and issue @base_work to all wb's (bdi_writeback's) of @bdi which
 * have dirty inodes.  If @base_work->nr_page isn't %LONG_MAX, it's
 * distributed to the busy wbs according to each wb's proportion in the
 * total active write bandwidth of @bdi.
 */
static void bdi_split_work_to_wbs(struct backing_dev_info *bdi,
				  struct wb_writeback_work *base_work,
				  bool skip_if_busy)
{
781 782 783
	struct bdi_writeback *last_wb = NULL;
	struct bdi_writeback *wb = list_entry_rcu(&bdi->wb_list,
						struct bdi_writeback, bdi_node);
784 785 786 787

	might_sleep();
restart:
	rcu_read_lock();
788
	list_for_each_entry_continue_rcu(wb, &bdi->wb_list, bdi_node) {
789 790 791 792 793
		DEFINE_WB_COMPLETION_ONSTACK(fallback_work_done);
		struct wb_writeback_work fallback_work;
		struct wb_writeback_work *work;
		long nr_pages;

794 795 796 797 798
		if (last_wb) {
			wb_put(last_wb);
			last_wb = NULL;
		}

799 800 801 802 803 804
		/* SYNC_ALL writes out I_DIRTY_TIME too */
		if (!wb_has_dirty_io(wb) &&
		    (base_work->sync_mode == WB_SYNC_NONE ||
		     list_empty(&wb->b_dirty_time)))
			continue;
		if (skip_if_busy && writeback_in_progress(wb))
805 806
			continue;

807 808 809 810 811 812 813 814 815
		nr_pages = wb_split_bdi_pages(wb, base_work->nr_pages);

		work = kmalloc(sizeof(*work), GFP_ATOMIC);
		if (work) {
			*work = *base_work;
			work->nr_pages = nr_pages;
			work->auto_free = 1;
			wb_queue_work(wb, work);
			continue;
816
		}
817 818 819 820 821 822 823 824 825 826

		/* alloc failed, execute synchronously using on-stack fallback */
		work = &fallback_work;
		*work = *base_work;
		work->nr_pages = nr_pages;
		work->auto_free = 0;
		work->done = &fallback_work_done;

		wb_queue_work(wb, work);

827 828 829 830 831 832 833 834
		/*
		 * Pin @wb so that it stays on @bdi->wb_list.  This allows
		 * continuing iteration from @wb after dropping and
		 * regrabbing rcu read lock.
		 */
		wb_get(wb);
		last_wb = wb;

835 836 837
		rcu_read_unlock();
		wb_wait_for_completion(bdi, &fallback_work_done);
		goto restart;
838 839
	}
	rcu_read_unlock();
840 841 842

	if (last_wb)
		wb_put(last_wb);
843 844
}

845 846
#else	/* CONFIG_CGROUP_WRITEBACK */

847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867
static struct bdi_writeback *
locked_inode_to_wb_and_lock_list(struct inode *inode)
	__releases(&inode->i_lock)
	__acquires(&wb->list_lock)
{
	struct bdi_writeback *wb = inode_to_wb(inode);

	spin_unlock(&inode->i_lock);
	spin_lock(&wb->list_lock);
	return wb;
}

static struct bdi_writeback *inode_to_wb_and_lock_list(struct inode *inode)
	__acquires(&wb->list_lock)
{
	struct bdi_writeback *wb = inode_to_wb(inode);

	spin_lock(&wb->list_lock);
	return wb;
}

868 869 870 871 872
static long wb_split_bdi_pages(struct bdi_writeback *wb, long nr_pages)
{
	return nr_pages;
}

873 874 875 876 877 878
static void bdi_split_work_to_wbs(struct backing_dev_info *bdi,
				  struct wb_writeback_work *base_work,
				  bool skip_if_busy)
{
	might_sleep();

879
	if (!skip_if_busy || !writeback_in_progress(&bdi->wb)) {
880 881 882 883 884
		base_work->auto_free = 0;
		wb_queue_work(&bdi->wb, base_work);
	}
}

885 886
#endif	/* CONFIG_CGROUP_WRITEBACK */

887 888
void wb_start_writeback(struct bdi_writeback *wb, long nr_pages,
			bool range_cyclic, enum wb_reason reason)
889
{
890 891 892 893 894 895 896 897 898 899 900
	struct wb_writeback_work *work;

	if (!wb_has_dirty_io(wb))
		return;

	/*
	 * This is WB_SYNC_NONE writeback, so if allocation fails just
	 * wakeup the thread for old dirty data writeback
	 */
	work = kzalloc(sizeof(*work), GFP_ATOMIC);
	if (!work) {
901
		trace_writeback_nowork(wb);
902 903 904 905 906 907 908 909
		wb_wakeup(wb);
		return;
	}

	work->sync_mode	= WB_SYNC_NONE;
	work->nr_pages	= nr_pages;
	work->range_cyclic = range_cyclic;
	work->reason	= reason;
910
	work->auto_free	= 1;
911 912

	wb_queue_work(wb, work);
913
}
914

915
/**
916 917
 * wb_start_background_writeback - start background writeback
 * @wb: bdi_writback to write from
918 919
 *
 * Description:
920
 *   This makes sure WB_SYNC_NONE background writeback happens. When
921
 *   this function returns, it is only guaranteed that for given wb
922 923
 *   some IO is happening if we are over background dirty threshold.
 *   Caller need not hold sb s_umount semaphore.
924
 */
925
void wb_start_background_writeback(struct bdi_writeback *wb)
926
{
927 928 929 930
	/*
	 * We just wake up the flusher thread. It will perform background
	 * writeback as soon as there is no other work to do.
	 */
931
	trace_writeback_wake_background(wb);
932
	wb_wakeup(wb);
L
Linus Torvalds 已提交
933 934
}

935 936 937
/*
 * Remove the inode from the writeback list it is on.
 */
938
void inode_io_list_del(struct inode *inode)
939
{
940
	struct bdi_writeback *wb;
941

942
	wb = inode_to_wb_and_lock_list(inode);
943
	inode_io_list_del_locked(inode, wb);
944
	spin_unlock(&wb->list_lock);
945 946
}

947 948 949 950 951
/*
 * Redirty an inode: set its when-it-was dirtied timestamp and move it to the
 * furthest end of its superblock's dirty-inode list.
 *
 * Before stamping the inode's ->dirtied_when, we check to see whether it is
952
 * already the most-recently-dirtied inode on the b_dirty list.  If that is
953 954 955
 * the case then the inode must have been redirtied while it was being written
 * out and we don't reset its dirtied_when.
 */
956
static void redirty_tail(struct inode *inode, struct bdi_writeback *wb)
957
{
958
	if (!list_empty(&wb->b_dirty)) {
959
		struct inode *tail;
960

N
Nick Piggin 已提交
961
		tail = wb_inode(wb->b_dirty.next);
962
		if (time_before(inode->dirtied_when, tail->dirtied_when))
963 964
			inode->dirtied_when = jiffies;
	}
965
	inode_io_list_move_locked(inode, wb, &wb->b_dirty);
966 967
}

968
/*
969
 * requeue inode for re-scanning after bdi->b_io list is exhausted.
970
 */
971
static void requeue_io(struct inode *inode, struct bdi_writeback *wb)
972
{
973
	inode_io_list_move_locked(inode, wb, &wb->b_more_io);
974 975
}

J
Joern Engel 已提交
976 977
static void inode_sync_complete(struct inode *inode)
{
978
	inode->i_state &= ~I_SYNC;
979 980
	/* If inode is clean an unused, put it into LRU now... */
	inode_add_lru(inode);
981
	/* Waiters must see I_SYNC cleared before being woken up */
J
Joern Engel 已提交
982 983 984 985
	smp_mb();
	wake_up_bit(&inode->i_state, __I_SYNC);
}

986 987 988 989 990 991 992 993
static bool inode_dirtied_after(struct inode *inode, unsigned long t)
{
	bool ret = time_after(inode->dirtied_when, t);
#ifndef CONFIG_64BIT
	/*
	 * For inodes being constantly redirtied, dirtied_when can get stuck.
	 * It _appears_ to be in the future, but is actually in distant past.
	 * This test is necessary to prevent such wrapped-around relative times
994
	 * from permanently stopping the whole bdi writeback.
995 996 997 998 999 1000
	 */
	ret = ret && time_before_eq(inode->dirtied_when, jiffies);
#endif
	return ret;
}

1001 1002
#define EXPIRE_DIRTY_ATIME 0x0001

1003
/*
1004
 * Move expired (dirtied before work->older_than_this) dirty inodes from
J
Jan Kara 已提交
1005
 * @delaying_queue to @dispatch_queue.
1006
 */
1007
static int move_expired_inodes(struct list_head *delaying_queue,
1008
			       struct list_head *dispatch_queue,
1009
			       int flags,
1010
			       struct wb_writeback_work *work)
1011
{
1012 1013
	unsigned long *older_than_this = NULL;
	unsigned long expire_time;
1014 1015
	LIST_HEAD(tmp);
	struct list_head *pos, *node;
1016
	struct super_block *sb = NULL;
1017
	struct inode *inode;
1018
	int do_sb_sort = 0;
1019
	int moved = 0;
1020

1021 1022
	if ((flags & EXPIRE_DIRTY_ATIME) == 0)
		older_than_this = work->older_than_this;
1023 1024
	else if (!work->for_sync) {
		expire_time = jiffies - (dirtytime_expire_interval * HZ);
1025 1026
		older_than_this = &expire_time;
	}
1027
	while (!list_empty(delaying_queue)) {
N
Nick Piggin 已提交
1028
		inode = wb_inode(delaying_queue->prev);
1029 1030
		if (older_than_this &&
		    inode_dirtied_after(inode, *older_than_this))
1031
			break;
1032
		list_move(&inode->i_io_list, &tmp);
1033
		moved++;
1034 1035
		if (flags & EXPIRE_DIRTY_ATIME)
			set_bit(__I_DIRTY_TIME_EXPIRED, &inode->i_state);
1036 1037
		if (sb_is_blkdev_sb(inode->i_sb))
			continue;
1038 1039 1040
		if (sb && sb != inode->i_sb)
			do_sb_sort = 1;
		sb = inode->i_sb;
1041 1042
	}

1043 1044 1045
	/* just one sb in list, splice to dispatch_queue and we're done */
	if (!do_sb_sort) {
		list_splice(&tmp, dispatch_queue);
1046
		goto out;
1047 1048
	}

1049 1050
	/* Move inodes from one superblock together */
	while (!list_empty(&tmp)) {
N
Nick Piggin 已提交
1051
		sb = wb_inode(tmp.prev)->i_sb;
1052
		list_for_each_prev_safe(pos, node, &tmp) {
N
Nick Piggin 已提交
1053
			inode = wb_inode(pos);
1054
			if (inode->i_sb == sb)
1055
				list_move(&inode->i_io_list, dispatch_queue);
1056
		}
1057
	}
1058 1059
out:
	return moved;
1060 1061 1062 1063
}

/*
 * Queue all expired dirty inodes for io, eldest first.
1064 1065 1066 1067 1068 1069 1070 1071
 * Before
 *         newly dirtied     b_dirty    b_io    b_more_io
 *         =============>    gf         edc     BA
 * After
 *         newly dirtied     b_dirty    b_io    b_more_io
 *         =============>    g          fBAedc
 *                                           |
 *                                           +--> dequeue for IO
1072
 */
1073
static void queue_io(struct bdi_writeback *wb, struct wb_writeback_work *work)
1074
{
1075
	int moved;
1076

1077
	assert_spin_locked(&wb->list_lock);
1078
	list_splice_init(&wb->b_more_io, &wb->b_io);
1079 1080 1081
	moved = move_expired_inodes(&wb->b_dirty, &wb->b_io, 0, work);
	moved += move_expired_inodes(&wb->b_dirty_time, &wb->b_io,
				     EXPIRE_DIRTY_ATIME, work);
1082 1083
	if (moved)
		wb_io_lists_populated(wb);
1084
	trace_writeback_queue_io(wb, work, moved);
1085 1086
}

1087
static int write_inode(struct inode *inode, struct writeback_control *wbc)
1088
{
T
Tejun Heo 已提交
1089 1090 1091 1092 1093 1094 1095 1096
	int ret;

	if (inode->i_sb->s_op->write_inode && !is_bad_inode(inode)) {
		trace_writeback_write_inode_start(inode, wbc);
		ret = inode->i_sb->s_op->write_inode(inode, wbc);
		trace_writeback_write_inode(inode, wbc);
		return ret;
	}
1097
	return 0;
1098 1099
}

L
Linus Torvalds 已提交
1100
/*
1101 1102
 * Wait for writeback on an inode to complete. Called with i_lock held.
 * Caller must make sure inode cannot go away when we drop i_lock.
1103
 */
1104 1105 1106
static void __inode_wait_for_writeback(struct inode *inode)
	__releases(inode->i_lock)
	__acquires(inode->i_lock)
1107 1108 1109 1110 1111
{
	DEFINE_WAIT_BIT(wq, &inode->i_state, __I_SYNC);
	wait_queue_head_t *wqh;

	wqh = bit_waitqueue(&inode->i_state, __I_SYNC);
1112 1113
	while (inode->i_state & I_SYNC) {
		spin_unlock(&inode->i_lock);
1114 1115
		__wait_on_bit(wqh, &wq, bit_wait,
			      TASK_UNINTERRUPTIBLE);
1116
		spin_lock(&inode->i_lock);
1117
	}
1118 1119
}

1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149
/*
 * Wait for writeback on an inode to complete. Caller must have inode pinned.
 */
void inode_wait_for_writeback(struct inode *inode)
{
	spin_lock(&inode->i_lock);
	__inode_wait_for_writeback(inode);
	spin_unlock(&inode->i_lock);
}

/*
 * Sleep until I_SYNC is cleared. This function must be called with i_lock
 * held and drops it. It is aimed for callers not holding any inode reference
 * so once i_lock is dropped, inode can go away.
 */
static void inode_sleep_on_writeback(struct inode *inode)
	__releases(inode->i_lock)
{
	DEFINE_WAIT(wait);
	wait_queue_head_t *wqh = bit_waitqueue(&inode->i_state, __I_SYNC);
	int sleep;

	prepare_to_wait(wqh, &wait, TASK_UNINTERRUPTIBLE);
	sleep = inode->i_state & I_SYNC;
	spin_unlock(&inode->i_lock);
	if (sleep)
		schedule();
	finish_wait(wqh, &wait);
}

1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172
/*
 * Find proper writeback list for the inode depending on its current state and
 * possibly also change of its state while we were doing writeback.  Here we
 * handle things such as livelock prevention or fairness of writeback among
 * inodes. This function can be called only by flusher thread - noone else
 * processes all inodes in writeback lists and requeueing inodes behind flusher
 * thread's back can have unexpected consequences.
 */
static void requeue_inode(struct inode *inode, struct bdi_writeback *wb,
			  struct writeback_control *wbc)
{
	if (inode->i_state & I_FREEING)
		return;

	/*
	 * Sync livelock prevention. Each inode is tagged and synced in one
	 * shot. If still dirty, it will be redirty_tail()'ed below.  Update
	 * the dirty time to prevent enqueue and sync it again.
	 */
	if ((inode->i_state & I_DIRTY) &&
	    (wbc->sync_mode == WB_SYNC_ALL || wbc->tagged_writepages))
		inode->dirtied_when = jiffies;

1173 1174 1175 1176 1177 1178 1179 1180 1181
	if (wbc->pages_skipped) {
		/*
		 * writeback is not making progress due to locked
		 * buffers. Skip this inode for now.
		 */
		redirty_tail(inode, wb);
		return;
	}

1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206
	if (mapping_tagged(inode->i_mapping, PAGECACHE_TAG_DIRTY)) {
		/*
		 * We didn't write back all the pages.  nfs_writepages()
		 * sometimes bales out without doing anything.
		 */
		if (wbc->nr_to_write <= 0) {
			/* Slice used up. Queue for next turn. */
			requeue_io(inode, wb);
		} else {
			/*
			 * Writeback blocked by something other than
			 * congestion. Delay the inode for some time to
			 * avoid spinning on the CPU (100% iowait)
			 * retrying writeback of the dirty page/inode
			 * that cannot be performed immediately.
			 */
			redirty_tail(inode, wb);
		}
	} else if (inode->i_state & I_DIRTY) {
		/*
		 * Filesystems can dirty the inode during writeback operations,
		 * such as delayed allocation during submission or metadata
		 * updates after data IO completion.
		 */
		redirty_tail(inode, wb);
1207
	} else if (inode->i_state & I_DIRTY_TIME) {
1208
		inode->dirtied_when = jiffies;
1209
		inode_io_list_move_locked(inode, wb, &wb->b_dirty_time);
1210 1211
	} else {
		/* The inode is clean. Remove from writeback lists. */
1212
		inode_io_list_del_locked(inode, wb);
1213 1214 1215
	}
}

1216
/*
1217 1218 1219
 * Write out an inode and its dirty pages. Do not update the writeback list
 * linkage. That is left to the caller. The caller is also responsible for
 * setting I_SYNC flag and calling inode_sync_complete() to clear it.
L
Linus Torvalds 已提交
1220 1221
 */
static int
1222
__writeback_single_inode(struct inode *inode, struct writeback_control *wbc)
L
Linus Torvalds 已提交
1223 1224
{
	struct address_space *mapping = inode->i_mapping;
1225
	long nr_to_write = wbc->nr_to_write;
1226
	unsigned dirty;
L
Linus Torvalds 已提交
1227 1228
	int ret;

1229
	WARN_ON(!(inode->i_state & I_SYNC));
L
Linus Torvalds 已提交
1230

T
Tejun Heo 已提交
1231 1232
	trace_writeback_single_inode_start(inode, wbc, nr_to_write);

L
Linus Torvalds 已提交
1233 1234
	ret = do_writepages(mapping, wbc);

1235 1236 1237
	/*
	 * Make sure to wait on the data before writing out the metadata.
	 * This is important for filesystems that modify metadata on data
1238 1239 1240
	 * I/O completion. We don't do it for sync(2) writeback because it has a
	 * separate, external IO completion path and ->sync_fs for guaranteeing
	 * inode metadata is written back correctly.
1241
	 */
1242
	if (wbc->sync_mode == WB_SYNC_ALL && !wbc->for_sync) {
1243
		int err = filemap_fdatawait(mapping);
L
Linus Torvalds 已提交
1244 1245 1246 1247
		if (ret == 0)
			ret = err;
	}

1248 1249 1250 1251 1252
	/*
	 * Some filesystems may redirty the inode during the writeback
	 * due to delalloc, clear dirty metadata flags right before
	 * write_inode()
	 */
1253
	spin_lock(&inode->i_lock);
1254

1255
	dirty = inode->i_state & I_DIRTY;
1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266
	if (inode->i_state & I_DIRTY_TIME) {
		if ((dirty & (I_DIRTY_SYNC | I_DIRTY_DATASYNC)) ||
		    unlikely(inode->i_state & I_DIRTY_TIME_EXPIRED) ||
		    unlikely(time_after(jiffies,
					(inode->dirtied_time_when +
					 dirtytime_expire_interval * HZ)))) {
			dirty |= I_DIRTY_TIME | I_DIRTY_TIME_EXPIRED;
			trace_writeback_lazytime(inode);
		}
	} else
		inode->i_state &= ~I_DIRTY_TIME_EXPIRED;
1267
	inode->i_state &= ~dirty;
1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284

	/*
	 * Paired with smp_mb() in __mark_inode_dirty().  This allows
	 * __mark_inode_dirty() to test i_state without grabbing i_lock -
	 * either they see the I_DIRTY bits cleared or we see the dirtied
	 * inode.
	 *
	 * I_DIRTY_PAGES is always cleared together above even if @mapping
	 * still has dirty pages.  The flag is reinstated after smp_mb() if
	 * necessary.  This guarantees that either __mark_inode_dirty()
	 * sees clear I_DIRTY_PAGES or we see PAGECACHE_TAG_DIRTY.
	 */
	smp_mb();

	if (mapping_tagged(mapping, PAGECACHE_TAG_DIRTY))
		inode->i_state |= I_DIRTY_PAGES;

1285
	spin_unlock(&inode->i_lock);
1286

1287 1288
	if (dirty & I_DIRTY_TIME)
		mark_inode_dirty_sync(inode);
1289
	/* Don't write the inode if only I_DIRTY_PAGES was set */
1290
	if (dirty & ~I_DIRTY_PAGES) {
1291
		int err = write_inode(inode, wbc);
L
Linus Torvalds 已提交
1292 1293 1294
		if (ret == 0)
			ret = err;
	}
1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322
	trace_writeback_single_inode(inode, wbc, nr_to_write);
	return ret;
}

/*
 * Write out an inode's dirty pages. Either the caller has an active reference
 * on the inode or the inode has I_WILL_FREE set.
 *
 * This function is designed to be called for writing back one inode which
 * we go e.g. from filesystem. Flusher thread uses __writeback_single_inode()
 * and does more profound writeback list handling in writeback_sb_inodes().
 */
static int
writeback_single_inode(struct inode *inode, struct bdi_writeback *wb,
		       struct writeback_control *wbc)
{
	int ret = 0;

	spin_lock(&inode->i_lock);
	if (!atomic_read(&inode->i_count))
		WARN_ON(!(inode->i_state & (I_WILL_FREE|I_FREEING)));
	else
		WARN_ON(inode->i_state & I_WILL_FREE);

	if (inode->i_state & I_SYNC) {
		if (wbc->sync_mode != WB_SYNC_ALL)
			goto out;
		/*
1323 1324 1325
		 * It's a data-integrity sync. We must wait. Since callers hold
		 * inode reference or inode has I_WILL_FREE set, it cannot go
		 * away under us.
1326
		 */
1327
		__inode_wait_for_writeback(inode);
1328 1329 1330
	}
	WARN_ON(inode->i_state & I_SYNC);
	/*
J
Jan Kara 已提交
1331 1332 1333 1334 1335 1336
	 * Skip inode if it is clean and we have no outstanding writeback in
	 * WB_SYNC_ALL mode. We don't want to mess with writeback lists in this
	 * function since flusher thread may be doing for example sync in
	 * parallel and if we move the inode, it could get skipped. So here we
	 * make sure inode is on some writeback list and leave it there unless
	 * we have completely cleaned the inode.
1337
	 */
1338
	if (!(inode->i_state & I_DIRTY_ALL) &&
J
Jan Kara 已提交
1339 1340
	    (wbc->sync_mode != WB_SYNC_ALL ||
	     !mapping_tagged(inode->i_mapping, PAGECACHE_TAG_WRITEBACK)))
1341 1342
		goto out;
	inode->i_state |= I_SYNC;
1343
	wbc_attach_and_unlock_inode(wbc, inode);
1344

1345
	ret = __writeback_single_inode(inode, wbc);
L
Linus Torvalds 已提交
1346

1347
	wbc_detach_inode(wbc);
1348
	spin_lock(&wb->list_lock);
1349
	spin_lock(&inode->i_lock);
1350 1351 1352 1353
	/*
	 * If inode is clean, remove it from writeback lists. Otherwise don't
	 * touch it. See comment above for explanation.
	 */
1354
	if (!(inode->i_state & I_DIRTY_ALL))
1355
		inode_io_list_del_locked(inode, wb);
1356
	spin_unlock(&wb->list_lock);
J
Joern Engel 已提交
1357
	inode_sync_complete(inode);
1358 1359
out:
	spin_unlock(&inode->i_lock);
L
Linus Torvalds 已提交
1360 1361 1362
	return ret;
}

1363
static long writeback_chunk_size(struct bdi_writeback *wb,
1364
				 struct wb_writeback_work *work)
1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382
{
	long pages;

	/*
	 * WB_SYNC_ALL mode does livelock avoidance by syncing dirty
	 * inodes/pages in one big loop. Setting wbc.nr_to_write=LONG_MAX
	 * here avoids calling into writeback_inodes_wb() more than once.
	 *
	 * The intended call sequence for WB_SYNC_ALL writeback is:
	 *
	 *      wb_writeback()
	 *          writeback_sb_inodes()       <== called only once
	 *              write_cache_pages()     <== called once for each inode
	 *                   (quickly) tag currently dirty pages
	 *                   (maybe slowly) sync all tagged pages
	 */
	if (work->sync_mode == WB_SYNC_ALL || work->tagged_writepages)
		pages = LONG_MAX;
1383
	else {
1384
		pages = min(wb->avg_write_bandwidth / 2,
1385
			    global_wb_domain.dirty_limit / DIRTY_SCOPE);
1386 1387 1388 1389
		pages = min(pages, work->nr_pages);
		pages = round_down(pages + MIN_WRITEBACK_PAGES,
				   MIN_WRITEBACK_PAGES);
	}
1390 1391 1392 1393

	return pages;
}

1394 1395
/*
 * Write a portion of b_io inodes which belong to @sb.
1396
 *
1397
 * Return the number of pages and/or inodes written.
1398 1399 1400 1401
 *
 * NOTE! This is called with wb->list_lock held, and will
 * unlock and relock that for each inode it ends up doing
 * IO for.
1402
 */
1403 1404 1405
static long writeback_sb_inodes(struct super_block *sb,
				struct bdi_writeback *wb,
				struct wb_writeback_work *work)
L
Linus Torvalds 已提交
1406
{
1407 1408 1409 1410 1411
	struct writeback_control wbc = {
		.sync_mode		= work->sync_mode,
		.tagged_writepages	= work->tagged_writepages,
		.for_kupdate		= work->for_kupdate,
		.for_background		= work->for_background,
1412
		.for_sync		= work->for_sync,
1413 1414 1415 1416 1417 1418 1419 1420
		.range_cyclic		= work->range_cyclic,
		.range_start		= 0,
		.range_end		= LLONG_MAX,
	};
	unsigned long start_time = jiffies;
	long write_chunk;
	long wrote = 0;  /* count both pages and inodes */

1421
	while (!list_empty(&wb->b_io)) {
N
Nick Piggin 已提交
1422
		struct inode *inode = wb_inode(wb->b_io.prev);
1423 1424

		if (inode->i_sb != sb) {
1425
			if (work->sb) {
1426 1427 1428 1429 1430
				/*
				 * We only want to write back data for this
				 * superblock, move all inodes not belonging
				 * to it back onto the dirty list.
				 */
1431
				redirty_tail(inode, wb);
1432 1433 1434 1435 1436 1437 1438 1439
				continue;
			}

			/*
			 * The inode belongs to a different superblock.
			 * Bounce back to the caller to unpin this and
			 * pin the next superblock.
			 */
1440
			break;
1441 1442
		}

1443
		/*
W
Wanpeng Li 已提交
1444 1445
		 * Don't bother with new inodes or inodes being freed, first
		 * kind does not need periodic writeout yet, and for the latter
1446 1447
		 * kind writeout is handled by the freer.
		 */
1448
		spin_lock(&inode->i_lock);
1449
		if (inode->i_state & (I_NEW | I_FREEING | I_WILL_FREE)) {
1450
			spin_unlock(&inode->i_lock);
1451
			redirty_tail(inode, wb);
1452 1453
			continue;
		}
1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468
		if ((inode->i_state & I_SYNC) && wbc.sync_mode != WB_SYNC_ALL) {
			/*
			 * If this inode is locked for writeback and we are not
			 * doing writeback-for-data-integrity, move it to
			 * b_more_io so that writeback can proceed with the
			 * other inodes on s_io.
			 *
			 * We'll have another go at writing back this inode
			 * when we completed a full scan of b_io.
			 */
			spin_unlock(&inode->i_lock);
			requeue_io(inode, wb);
			trace_writeback_sb_inodes_requeue(inode);
			continue;
		}
1469 1470
		spin_unlock(&wb->list_lock);

1471 1472 1473 1474 1475
		/*
		 * We already requeued the inode if it had I_SYNC set and we
		 * are doing WB_SYNC_NONE writeback. So this catches only the
		 * WB_SYNC_ALL case.
		 */
1476 1477 1478 1479
		if (inode->i_state & I_SYNC) {
			/* Wait for I_SYNC. This function drops i_lock... */
			inode_sleep_on_writeback(inode);
			/* Inode may be gone, start again */
1480
			spin_lock(&wb->list_lock);
1481 1482
			continue;
		}
1483
		inode->i_state |= I_SYNC;
1484
		wbc_attach_and_unlock_inode(&wbc, inode);
1485

1486
		write_chunk = writeback_chunk_size(wb, work);
1487 1488
		wbc.nr_to_write = write_chunk;
		wbc.pages_skipped = 0;
1489

1490 1491 1492 1493
		/*
		 * We use I_SYNC to pin the inode in memory. While it is set
		 * evict_inode() will wait so the inode cannot be freed.
		 */
1494
		__writeback_single_inode(inode, &wbc);
1495

1496
		wbc_detach_inode(&wbc);
1497 1498
		work->nr_pages -= write_chunk - wbc.nr_to_write;
		wrote += write_chunk - wbc.nr_to_write;
1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513

		if (need_resched()) {
			/*
			 * We're trying to balance between building up a nice
			 * long list of IOs to improve our merge rate, and
			 * getting those IOs out quickly for anyone throttling
			 * in balance_dirty_pages().  cond_resched() doesn't
			 * unplug, so get our IOs out the door before we
			 * give up the CPU.
			 */
			blk_flush_plug(current);
			cond_resched();
		}


1514 1515
		spin_lock(&wb->list_lock);
		spin_lock(&inode->i_lock);
1516
		if (!(inode->i_state & I_DIRTY_ALL))
1517
			wrote++;
1518 1519
		requeue_inode(inode, wb, &wbc);
		inode_sync_complete(inode);
1520
		spin_unlock(&inode->i_lock);
1521

1522 1523 1524 1525 1526 1527 1528 1529 1530
		/*
		 * bail out to wb_writeback() often enough to check
		 * background threshold and other termination conditions.
		 */
		if (wrote) {
			if (time_is_before_jiffies(start_time + HZ / 10UL))
				break;
			if (work->nr_pages <= 0)
				break;
1531
		}
L
Linus Torvalds 已提交
1532
	}
1533
	return wrote;
1534 1535
}

1536 1537
static long __writeback_inodes_wb(struct bdi_writeback *wb,
				  struct wb_writeback_work *work)
1538
{
1539 1540
	unsigned long start_time = jiffies;
	long wrote = 0;
N
Nick Piggin 已提交
1541

1542
	while (!list_empty(&wb->b_io)) {
N
Nick Piggin 已提交
1543
		struct inode *inode = wb_inode(wb->b_io.prev);
1544
		struct super_block *sb = inode->i_sb;
1545

1546
		if (!trylock_super(sb)) {
1547
			/*
1548
			 * trylock_super() may fail consistently due to
1549 1550 1551 1552
			 * s_umount being grabbed by someone else. Don't use
			 * requeue_io() to avoid busy retrying the inode/sb.
			 */
			redirty_tail(inode, wb);
1553
			continue;
1554
		}
1555
		wrote += writeback_sb_inodes(sb, wb, work);
1556
		up_read(&sb->s_umount);
1557

1558 1559 1560 1561 1562 1563 1564
		/* refer to the same tests at the end of writeback_sb_inodes */
		if (wrote) {
			if (time_is_before_jiffies(start_time + HZ / 10UL))
				break;
			if (work->nr_pages <= 0)
				break;
		}
1565
	}
1566
	/* Leave any unwritten inodes on b_io */
1567
	return wrote;
1568 1569
}

1570
static long writeback_inodes_wb(struct bdi_writeback *wb, long nr_pages,
1571
				enum wb_reason reason)
1572
{
1573 1574 1575 1576
	struct wb_writeback_work work = {
		.nr_pages	= nr_pages,
		.sync_mode	= WB_SYNC_NONE,
		.range_cyclic	= 1,
1577
		.reason		= reason,
1578
	};
1579
	struct blk_plug plug;
1580

1581
	blk_start_plug(&plug);
1582
	spin_lock(&wb->list_lock);
W
Wu Fengguang 已提交
1583
	if (list_empty(&wb->b_io))
1584
		queue_io(wb, &work);
1585
	__writeback_inodes_wb(wb, &work);
1586
	spin_unlock(&wb->list_lock);
1587
	blk_finish_plug(&plug);
1588

1589 1590
	return nr_pages - work.nr_pages;
}
1591 1592 1593

/*
 * Explicit flushing or periodic writeback of "old" data.
1594
 *
1595 1596 1597 1598
 * Define "old": the first time one of an inode's pages is dirtied, we mark the
 * dirtying-time in the inode's address_space.  So this periodic writeback code
 * just walks the superblock inode list, writing back any inodes which are
 * older than a specific point in time.
1599
 *
1600 1601 1602
 * Try to run once per dirty_writeback_interval.  But if a writeback event
 * takes longer than a dirty_writeback_interval interval, then leave a
 * one-second gap.
1603
 *
1604 1605
 * older_than_this takes precedence over nr_to_write.  So we'll only write back
 * all dirty pages if they are all attached to "old" mappings.
1606
 */
1607
static long wb_writeback(struct bdi_writeback *wb,
1608
			 struct wb_writeback_work *work)
1609
{
1610
	unsigned long wb_start = jiffies;
1611
	long nr_pages = work->nr_pages;
1612
	unsigned long oldest_jif;
J
Jan Kara 已提交
1613
	struct inode *inode;
1614
	long progress;
1615
	struct blk_plug plug;
1616

1617 1618
	oldest_jif = jiffies;
	work->older_than_this = &oldest_jif;
N
Nick Piggin 已提交
1619

1620
	blk_start_plug(&plug);
1621
	spin_lock(&wb->list_lock);
1622 1623
	for (;;) {
		/*
1624
		 * Stop writeback when nr_pages has been consumed
1625
		 */
1626
		if (work->nr_pages <= 0)
1627
			break;
1628

1629 1630 1631 1632 1633 1634 1635
		/*
		 * Background writeout and kupdate-style writeback may
		 * run forever. Stop them if there is other work to do
		 * so that e.g. sync can proceed. They'll be restarted
		 * after the other works are all done.
		 */
		if ((work->for_background || work->for_kupdate) &&
1636
		    !list_empty(&wb->work_list))
1637 1638
			break;

N
Nick Piggin 已提交
1639
		/*
1640 1641
		 * For background writeout, stop when we are below the
		 * background dirty threshold
N
Nick Piggin 已提交
1642
		 */
1643
		if (work->for_background && !wb_over_bg_thresh(wb))
1644
			break;
N
Nick Piggin 已提交
1645

1646 1647 1648 1649 1650 1651
		/*
		 * Kupdate and background works are special and we want to
		 * include all inodes that need writing. Livelock avoidance is
		 * handled by these works yielding to any other work so we are
		 * safe.
		 */
1652
		if (work->for_kupdate) {
1653
			oldest_jif = jiffies -
1654
				msecs_to_jiffies(dirty_expire_interval * 10);
1655
		} else if (work->for_background)
1656
			oldest_jif = jiffies;
1657

1658
		trace_writeback_start(wb, work);
1659
		if (list_empty(&wb->b_io))
1660
			queue_io(wb, work);
1661
		if (work->sb)
1662
			progress = writeback_sb_inodes(work->sb, wb, work);
1663
		else
1664
			progress = __writeback_inodes_wb(wb, work);
1665
		trace_writeback_written(wb, work);
1666

1667
		wb_update_bandwidth(wb, wb_start);
1668 1669

		/*
1670 1671 1672 1673 1674 1675
		 * Did we write something? Try for more
		 *
		 * Dirty inodes are moved to b_io for writeback in batches.
		 * The completion of the current batch does not necessarily
		 * mean the overall work is done. So we keep looping as long
		 * as made some progress on cleaning pages or inodes.
1676
		 */
1677
		if (progress)
1678 1679
			continue;
		/*
1680
		 * No more inodes for IO, bail
1681
		 */
1682
		if (list_empty(&wb->b_more_io))
1683
			break;
1684 1685 1686 1687 1688 1689
		/*
		 * Nothing written. Wait for some inode to
		 * become available for writeback. Otherwise
		 * we'll just busyloop.
		 */
		if (!list_empty(&wb->b_more_io))  {
1690
			trace_writeback_wait(wb, work);
N
Nick Piggin 已提交
1691
			inode = wb_inode(wb->b_more_io.prev);
1692
			spin_lock(&inode->i_lock);
1693
			spin_unlock(&wb->list_lock);
1694 1695
			/* This function drops i_lock... */
			inode_sleep_on_writeback(inode);
1696
			spin_lock(&wb->list_lock);
1697 1698
		}
	}
1699
	spin_unlock(&wb->list_lock);
1700
	blk_finish_plug(&plug);
1701

1702
	return nr_pages - work->nr_pages;
1703 1704 1705
}

/*
1706
 * Return the next wb_writeback_work struct that hasn't been processed yet.
1707
 */
1708
static struct wb_writeback_work *get_next_work_item(struct bdi_writeback *wb)
1709
{
1710
	struct wb_writeback_work *work = NULL;
1711

1712 1713 1714
	spin_lock_bh(&wb->work_lock);
	if (!list_empty(&wb->work_list)) {
		work = list_entry(wb->work_list.next,
1715 1716
				  struct wb_writeback_work, list);
		list_del_init(&work->list);
1717
	}
1718
	spin_unlock_bh(&wb->work_lock);
1719
	return work;
1720 1721
}

1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732
/*
 * Add in the number of potentially dirty inodes, because each inode
 * write can dirty pagecache in the underlying blockdev.
 */
static unsigned long get_nr_dirty_pages(void)
{
	return global_page_state(NR_FILE_DIRTY) +
		global_page_state(NR_UNSTABLE_NFS) +
		get_nr_dirty_inodes();
}

1733 1734
static long wb_check_background_flush(struct bdi_writeback *wb)
{
1735
	if (wb_over_bg_thresh(wb)) {
1736 1737 1738 1739 1740 1741

		struct wb_writeback_work work = {
			.nr_pages	= LONG_MAX,
			.sync_mode	= WB_SYNC_NONE,
			.for_background	= 1,
			.range_cyclic	= 1,
1742
			.reason		= WB_REASON_BACKGROUND,
1743 1744 1745 1746 1747 1748 1749 1750
		};

		return wb_writeback(wb, &work);
	}

	return 0;
}

1751 1752 1753 1754 1755
static long wb_check_old_data_flush(struct bdi_writeback *wb)
{
	unsigned long expired;
	long nr_pages;

1756 1757 1758 1759 1760 1761
	/*
	 * When set to zero, disable periodic writeback
	 */
	if (!dirty_writeback_interval)
		return 0;

1762 1763 1764 1765 1766 1767
	expired = wb->last_old_flush +
			msecs_to_jiffies(dirty_writeback_interval * 10);
	if (time_before(jiffies, expired))
		return 0;

	wb->last_old_flush = jiffies;
1768
	nr_pages = get_nr_dirty_pages();
1769

1770
	if (nr_pages) {
1771
		struct wb_writeback_work work = {
1772 1773 1774 1775
			.nr_pages	= nr_pages,
			.sync_mode	= WB_SYNC_NONE,
			.for_kupdate	= 1,
			.range_cyclic	= 1,
1776
			.reason		= WB_REASON_PERIODIC,
1777 1778
		};

1779
		return wb_writeback(wb, &work);
1780
	}
1781 1782 1783 1784 1785 1786 1787

	return 0;
}

/*
 * Retrieve work items and do the writeback they describe
 */
1788
static long wb_do_writeback(struct bdi_writeback *wb)
1789
{
1790
	struct wb_writeback_work *work;
1791
	long wrote = 0;
1792

1793
	set_bit(WB_writeback_running, &wb->state);
1794
	while ((work = get_next_work_item(wb)) != NULL) {
1795
		struct wb_completion *done = work->done;
1796

1797
		trace_writeback_exec(wb, work);
1798

1799
		wrote += wb_writeback(wb, work);
1800

1801
		if (work->auto_free)
1802
			kfree(work);
1803 1804
		if (done && atomic_dec_and_test(&done->cnt))
			wake_up_all(&wb->bdi->wb_waitq);
1805 1806 1807 1808 1809 1810
	}

	/*
	 * Check for periodic writeback, kupdated() style
	 */
	wrote += wb_check_old_data_flush(wb);
1811
	wrote += wb_check_background_flush(wb);
1812
	clear_bit(WB_writeback_running, &wb->state);
1813 1814 1815 1816 1817 1818

	return wrote;
}

/*
 * Handle writeback of dirty data for the device backed by this bdi. Also
1819
 * reschedules periodically and does kupdated style flushing.
1820
 */
1821
void wb_workfn(struct work_struct *work)
1822
{
1823 1824
	struct bdi_writeback *wb = container_of(to_delayed_work(work),
						struct bdi_writeback, dwork);
1825 1826
	long pages_written;

1827
	set_worker_desc("flush-%s", dev_name(wb->bdi->dev));
P
Peter Zijlstra 已提交
1828
	current->flags |= PF_SWAPWRITE;
1829

1830
	if (likely(!current_is_workqueue_rescuer() ||
1831
		   !test_bit(WB_registered, &wb->state))) {
1832
		/*
1833
		 * The normal path.  Keep writing back @wb until its
1834
		 * work_list is empty.  Note that this path is also taken
1835
		 * if @wb is shutting down even when we're running off the
1836
		 * rescuer as work_list needs to be drained.
1837
		 */
1838
		do {
1839
			pages_written = wb_do_writeback(wb);
1840
			trace_writeback_pages_written(pages_written);
1841
		} while (!list_empty(&wb->work_list));
1842 1843 1844 1845 1846 1847
	} else {
		/*
		 * bdi_wq can't get enough workers and we're running off
		 * the emergency worker.  Don't hog it.  Hopefully, 1024 is
		 * enough for efficient IO.
		 */
1848
		pages_written = writeback_inodes_wb(wb, 1024,
1849
						    WB_REASON_FORKER_THREAD);
1850
		trace_writeback_pages_written(pages_written);
1851 1852
	}

1853
	if (!list_empty(&wb->work_list))
1854 1855
		mod_delayed_work(bdi_wq, &wb->dwork, 0);
	else if (wb_has_dirty_io(wb) && dirty_writeback_interval)
1856
		wb_wakeup_delayed(wb);
1857

1858
	current->flags &= ~PF_SWAPWRITE;
1859 1860 1861
}

/*
1862 1863
 * Start writeback of `nr_pages' pages.  If `nr_pages' is zero, write back
 * the whole world.
1864
 */
1865
void wakeup_flusher_threads(long nr_pages, enum wb_reason reason)
1866
{
1867
	struct backing_dev_info *bdi;
1868

1869 1870
	if (!nr_pages)
		nr_pages = get_nr_dirty_pages();
1871

1872
	rcu_read_lock();
1873 1874 1875 1876 1877 1878
	list_for_each_entry_rcu(bdi, &bdi_list, bdi_list) {
		struct bdi_writeback *wb;

		if (!bdi_has_dirty_io(bdi))
			continue;

1879
		list_for_each_entry_rcu(wb, &bdi->wb_list, bdi_node)
1880 1881 1882
			wb_start_writeback(wb, wb_split_bdi_pages(wb, nr_pages),
					   false, reason);
	}
1883
	rcu_read_unlock();
L
Linus Torvalds 已提交
1884 1885
}

1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909
/*
 * Wake up bdi's periodically to make sure dirtytime inodes gets
 * written back periodically.  We deliberately do *not* check the
 * b_dirtytime list in wb_has_dirty_io(), since this would cause the
 * kernel to be constantly waking up once there are any dirtytime
 * inodes on the system.  So instead we define a separate delayed work
 * function which gets called much more rarely.  (By default, only
 * once every 12 hours.)
 *
 * If there is any other write activity going on in the file system,
 * this function won't be necessary.  But if the only thing that has
 * happened on the file system is a dirtytime inode caused by an atime
 * update, we need this infrastructure below to make sure that inode
 * eventually gets pushed out to disk.
 */
static void wakeup_dirtytime_writeback(struct work_struct *w);
static DECLARE_DELAYED_WORK(dirtytime_work, wakeup_dirtytime_writeback);

static void wakeup_dirtytime_writeback(struct work_struct *w)
{
	struct backing_dev_info *bdi;

	rcu_read_lock();
	list_for_each_entry_rcu(bdi, &bdi_list, bdi_list) {
1910 1911
		struct bdi_writeback *wb;

1912
		list_for_each_entry_rcu(wb, &bdi->wb_list, bdi_node)
1913 1914
			if (!list_empty(&wb->b_dirty_time))
				wb_wakeup(wb);
1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926
	}
	rcu_read_unlock();
	schedule_delayed_work(&dirtytime_work, dirtytime_expire_interval * HZ);
}

static int __init start_dirtytime_writeback(void)
{
	schedule_delayed_work(&dirtytime_work, dirtytime_expire_interval * HZ);
	return 0;
}
__initcall(start_dirtytime_writeback);

1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937
int dirtytime_interval_handler(struct ctl_table *table, int write,
			       void __user *buffer, size_t *lenp, loff_t *ppos)
{
	int ret;

	ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
	if (ret == 0 && write)
		mod_delayed_work(system_wq, &dirtytime_work, 0);
	return ret;
}

1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965
static noinline void block_dump___mark_inode_dirty(struct inode *inode)
{
	if (inode->i_ino || strcmp(inode->i_sb->s_id, "bdev")) {
		struct dentry *dentry;
		const char *name = "?";

		dentry = d_find_alias(inode);
		if (dentry) {
			spin_lock(&dentry->d_lock);
			name = (const char *) dentry->d_name.name;
		}
		printk(KERN_DEBUG
		       "%s(%d): dirtied inode %lu (%s) on %s\n",
		       current->comm, task_pid_nr(current), inode->i_ino,
		       name, inode->i_sb->s_id);
		if (dentry) {
			spin_unlock(&dentry->d_lock);
			dput(dentry);
		}
	}
}

/**
 *	__mark_inode_dirty -	internal function
 *	@inode: inode to mark
 *	@flags: what kind of dirty (i.e. I_DIRTY_SYNC)
 *	Mark an inode as dirty. Callers should use mark_inode_dirty or
 *  	mark_inode_dirty_sync.
L
Linus Torvalds 已提交
1966
 *
1967 1968 1969 1970 1971 1972 1973 1974 1975
 * Put the inode on the super block's dirty list.
 *
 * CAREFUL! We mark it dirty unconditionally, but move it onto the
 * dirty list only if it is hashed or if it refers to a blockdev.
 * If it was not hashed, it will never be added to the dirty list
 * even if it is later hashed, as it will have been marked dirty already.
 *
 * In short, make sure you hash any inodes _before_ you start marking
 * them dirty.
L
Linus Torvalds 已提交
1976
 *
1977 1978 1979 1980 1981 1982
 * Note that for blockdevs, inode->dirtied_when represents the dirtying time of
 * the block-special inode (/dev/hda1) itself.  And the ->dirtied_when field of
 * the kernel-internal blockdev inode represents the dirtying time of the
 * blockdev's pages.  This is why for I_DIRTY_PAGES we always use
 * page->mapping->host, so the page-dirtying time is recorded in the internal
 * blockdev inode.
L
Linus Torvalds 已提交
1983
 */
1984
#define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC)
1985
void __mark_inode_dirty(struct inode *inode, int flags)
L
Linus Torvalds 已提交
1986
{
1987
	struct super_block *sb = inode->i_sb;
1988 1989 1990
	int dirtytime;

	trace_writeback_mark_inode_dirty(inode, flags);
L
Linus Torvalds 已提交
1991

1992 1993 1994 1995
	/*
	 * Don't do this for I_DIRTY_PAGES - that doesn't actually
	 * dirty the inode itself
	 */
1996
	if (flags & (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_TIME)) {
T
Tejun Heo 已提交
1997 1998
		trace_writeback_dirty_inode_start(inode, flags);

1999
		if (sb->s_op->dirty_inode)
2000
			sb->s_op->dirty_inode(inode, flags);
T
Tejun Heo 已提交
2001 2002

		trace_writeback_dirty_inode(inode, flags);
2003
	}
2004 2005 2006
	if (flags & I_DIRTY_INODE)
		flags &= ~I_DIRTY_TIME;
	dirtytime = flags & I_DIRTY_TIME;
2007 2008

	/*
2009 2010
	 * Paired with smp_mb() in __writeback_single_inode() for the
	 * following lockless i_state test.  See there for details.
2011 2012 2013
	 */
	smp_mb();

2014 2015
	if (((inode->i_state & flags) == flags) ||
	    (dirtytime && (inode->i_state & I_DIRTY_INODE)))
2016 2017 2018 2019 2020
		return;

	if (unlikely(block_dump))
		block_dump___mark_inode_dirty(inode);

2021
	spin_lock(&inode->i_lock);
2022 2023
	if (dirtytime && (inode->i_state & I_DIRTY_INODE))
		goto out_unlock_inode;
2024 2025 2026
	if ((inode->i_state & flags) != flags) {
		const int was_dirty = inode->i_state & I_DIRTY;

2027 2028
		inode_attach_wb(inode, NULL);

2029 2030
		if (flags & I_DIRTY_INODE)
			inode->i_state &= ~I_DIRTY_TIME;
2031 2032 2033 2034 2035 2036 2037 2038
		inode->i_state |= flags;

		/*
		 * If the inode is being synced, just update its dirty state.
		 * The unlocker will place the inode on the appropriate
		 * superblock list, based upon its state.
		 */
		if (inode->i_state & I_SYNC)
2039
			goto out_unlock_inode;
2040 2041 2042 2043 2044 2045

		/*
		 * Only add valid (hashed) inodes to the superblock's
		 * dirty list.  Add blockdev inodes as well.
		 */
		if (!S_ISBLK(inode->i_mode)) {
A
Al Viro 已提交
2046
			if (inode_unhashed(inode))
2047
				goto out_unlock_inode;
2048
		}
A
Al Viro 已提交
2049
		if (inode->i_state & I_FREEING)
2050
			goto out_unlock_inode;
2051 2052 2053 2054 2055 2056

		/*
		 * If the inode was already on b_dirty/b_io/b_more_io, don't
		 * reposition it (that would break b_dirty time-ordering).
		 */
		if (!was_dirty) {
2057
			struct bdi_writeback *wb;
2058
			struct list_head *dirty_list;
2059
			bool wakeup_bdi = false;
2060

2061
			wb = locked_inode_to_wb_and_lock_list(inode);
2062

2063 2064 2065
			WARN(bdi_cap_writeback_dirty(wb->bdi) &&
			     !test_bit(WB_registered, &wb->state),
			     "bdi-%s not registered\n", wb->bdi->name);
2066 2067

			inode->dirtied_when = jiffies;
2068 2069
			if (dirtytime)
				inode->dirtied_time_when = jiffies;
2070

2071
			if (inode->i_state & (I_DIRTY_INODE | I_DIRTY_PAGES))
2072
				dirty_list = &wb->b_dirty;
2073
			else
2074
				dirty_list = &wb->b_dirty_time;
2075

2076
			wakeup_bdi = inode_io_list_move_locked(inode, wb,
2077 2078
							       dirty_list);

2079
			spin_unlock(&wb->list_lock);
2080
			trace_writeback_dirty_inode_enqueue(inode);
2081

2082 2083 2084 2085 2086 2087
			/*
			 * If this is the first dirty inode for this bdi,
			 * we have to wake-up the corresponding bdi thread
			 * to make sure background write-back happens
			 * later.
			 */
2088 2089
			if (bdi_cap_writeback_dirty(wb->bdi) && wakeup_bdi)
				wb_wakeup_delayed(wb);
2090
			return;
L
Linus Torvalds 已提交
2091 2092
		}
	}
2093 2094
out_unlock_inode:
	spin_unlock(&inode->i_lock);
2095

2096 2097 2098
}
EXPORT_SYMBOL(__mark_inode_dirty);

2099 2100 2101 2102 2103 2104 2105 2106 2107
/*
 * The @s_sync_lock is used to serialise concurrent sync operations
 * to avoid lock contention problems with concurrent wait_sb_inodes() calls.
 * Concurrent callers will block on the s_sync_lock rather than doing contending
 * walks. The queueing maintains sync(2) required behaviour as all the IO that
 * has been issued up to the time this function is enter is guaranteed to be
 * completed by the time we have gained the lock and waited for all IO that is
 * in progress regardless of the order callers are granted the lock.
 */
2108
static void wait_sb_inodes(struct super_block *sb)
2109 2110 2111 2112 2113 2114 2115
{
	struct inode *inode, *old_inode = NULL;

	/*
	 * We need to be protected against the filesystem going from
	 * r/o to r/w or vice versa.
	 */
2116
	WARN_ON(!rwsem_is_locked(&sb->s_umount));
2117

2118
	mutex_lock(&sb->s_sync_lock);
2119
	spin_lock(&sb->s_inode_list_lock);
2120 2121 2122 2123 2124 2125 2126 2127

	/*
	 * Data integrity sync. Must wait for all pages under writeback,
	 * because there may have been pages dirtied before our sync
	 * call, but which had writeout started before we write it out.
	 * In which case, the inode may not be on the dirty list, but
	 * we still have to wait for that writeout.
	 */
2128
	list_for_each_entry(inode, &sb->s_inodes, i_sb_list) {
2129
		struct address_space *mapping = inode->i_mapping;
2130

2131 2132 2133 2134
		spin_lock(&inode->i_lock);
		if ((inode->i_state & (I_FREEING|I_WILL_FREE|I_NEW)) ||
		    (mapping->nrpages == 0)) {
			spin_unlock(&inode->i_lock);
2135
			continue;
2136
		}
2137
		__iget(inode);
2138
		spin_unlock(&inode->i_lock);
2139
		spin_unlock(&sb->s_inode_list_lock);
2140

2141
		/*
2142 2143
		 * We hold a reference to 'inode' so it couldn't have been
		 * removed from s_inodes list while we dropped the
2144
		 * s_inode_list_lock.  We cannot iput the inode now as we can
2145
		 * be holding the last reference and we cannot iput it under
2146
		 * s_inode_list_lock. So we keep the reference and iput it
2147
		 * later.
2148 2149 2150 2151 2152 2153 2154 2155
		 */
		iput(old_inode);
		old_inode = inode;

		filemap_fdatawait(mapping);

		cond_resched();

2156
		spin_lock(&sb->s_inode_list_lock);
2157
	}
2158
	spin_unlock(&sb->s_inode_list_lock);
2159
	iput(old_inode);
2160
	mutex_unlock(&sb->s_sync_lock);
L
Linus Torvalds 已提交
2161 2162
}

2163 2164
static void __writeback_inodes_sb_nr(struct super_block *sb, unsigned long nr,
				     enum wb_reason reason, bool skip_if_busy)
L
Linus Torvalds 已提交
2165
{
2166
	DEFINE_WB_COMPLETION_ONSTACK(done);
2167
	struct wb_writeback_work work = {
2168 2169 2170 2171 2172
		.sb			= sb,
		.sync_mode		= WB_SYNC_NONE,
		.tagged_writepages	= 1,
		.done			= &done,
		.nr_pages		= nr,
2173
		.reason			= reason,
2174
	};
2175
	struct backing_dev_info *bdi = sb->s_bdi;
2176

2177
	if (!bdi_has_dirty_io(bdi) || bdi == &noop_backing_dev_info)
2178
		return;
2179
	WARN_ON(!rwsem_is_locked(&sb->s_umount));
2180

2181
	bdi_split_work_to_wbs(sb->s_bdi, &work, skip_if_busy);
2182
	wb_wait_for_completion(bdi, &done);
2183
}
2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199 2200

/**
 * writeback_inodes_sb_nr -	writeback dirty inodes from given super_block
 * @sb: the superblock
 * @nr: the number of pages to write
 * @reason: reason why some writeback work initiated
 *
 * Start writeback on some inodes on this super_block. No guarantees are made
 * on how many (if any) will be written, and this function does not wait
 * for IO completion of submitted IO.
 */
void writeback_inodes_sb_nr(struct super_block *sb,
			    unsigned long nr,
			    enum wb_reason reason)
{
	__writeback_inodes_sb_nr(sb, nr, reason, false);
}
2201 2202 2203 2204 2205
EXPORT_SYMBOL(writeback_inodes_sb_nr);

/**
 * writeback_inodes_sb	-	writeback dirty inodes from given super_block
 * @sb: the superblock
2206
 * @reason: reason why some writeback work was initiated
2207 2208 2209 2210 2211
 *
 * Start writeback on some inodes on this super_block. No guarantees are made
 * on how many (if any) will be written, and this function does not wait
 * for IO completion of submitted IO.
 */
2212
void writeback_inodes_sb(struct super_block *sb, enum wb_reason reason)
2213
{
2214
	return writeback_inodes_sb_nr(sb, get_nr_dirty_pages(), reason);
2215
}
2216
EXPORT_SYMBOL(writeback_inodes_sb);
2217

2218
/**
2219
 * try_to_writeback_inodes_sb_nr - try to start writeback if none underway
2220
 * @sb: the superblock
2221 2222
 * @nr: the number of pages to write
 * @reason: the reason of writeback
2223
 *
2224
 * Invoke writeback_inodes_sb_nr if no writeback is currently underway.
2225 2226
 * Returns 1 if writeback was started, 0 if not.
 */
2227 2228
bool try_to_writeback_inodes_sb_nr(struct super_block *sb, unsigned long nr,
				   enum wb_reason reason)
2229
{
2230
	if (!down_read_trylock(&sb->s_umount))
2231
		return false;
2232

2233
	__writeback_inodes_sb_nr(sb, nr, reason, true);
2234
	up_read(&sb->s_umount);
2235
	return true;
2236
}
2237
EXPORT_SYMBOL(try_to_writeback_inodes_sb_nr);
2238

2239
/**
2240
 * try_to_writeback_inodes_sb - try to start writeback if none underway
2241
 * @sb: the superblock
2242
 * @reason: reason why some writeback work was initiated
2243
 *
2244
 * Implement by try_to_writeback_inodes_sb_nr()
2245 2246
 * Returns 1 if writeback was started, 0 if not.
 */
2247
bool try_to_writeback_inodes_sb(struct super_block *sb, enum wb_reason reason)
2248
{
2249
	return try_to_writeback_inodes_sb_nr(sb, get_nr_dirty_pages(), reason);
2250
}
2251
EXPORT_SYMBOL(try_to_writeback_inodes_sb);
2252

2253 2254
/**
 * sync_inodes_sb	-	sync sb inode pages
2255
 * @sb: the superblock
2256 2257
 *
 * This function writes and waits on any dirty inode belonging to this
2258
 * super_block.
2259
 */
2260
void sync_inodes_sb(struct super_block *sb)
2261
{
2262
	DEFINE_WB_COMPLETION_ONSTACK(done);
2263
	struct wb_writeback_work work = {
2264 2265 2266 2267
		.sb		= sb,
		.sync_mode	= WB_SYNC_ALL,
		.nr_pages	= LONG_MAX,
		.range_cyclic	= 0,
2268
		.done		= &done,
2269
		.reason		= WB_REASON_SYNC,
2270
		.for_sync	= 1,
2271
	};
2272
	struct backing_dev_info *bdi = sb->s_bdi;
2273

2274 2275 2276 2277 2278 2279
	/*
	 * Can't skip on !bdi_has_dirty() because we should wait for !dirty
	 * inodes under writeback and I_DIRTY_TIME inodes ignored by
	 * bdi_has_dirty() need to be written out too.
	 */
	if (bdi == &noop_backing_dev_info)
2280
		return;
2281 2282
	WARN_ON(!rwsem_is_locked(&sb->s_umount));

2283
	bdi_split_work_to_wbs(bdi, &work, false);
2284
	wb_wait_for_completion(bdi, &done);
2285

2286
	wait_sb_inodes(sb);
L
Linus Torvalds 已提交
2287
}
2288
EXPORT_SYMBOL(sync_inodes_sb);
L
Linus Torvalds 已提交
2289 2290

/**
2291 2292 2293 2294 2295 2296
 * write_inode_now	-	write an inode to disk
 * @inode: inode to write to disk
 * @sync: whether the write should be synchronous or not
 *
 * This function commits an inode to disk immediately if it is dirty. This is
 * primarily needed by knfsd.
L
Linus Torvalds 已提交
2297
 *
2298
 * The caller must either have a ref on the inode or must have set I_WILL_FREE.
L
Linus Torvalds 已提交
2299 2300 2301
 */
int write_inode_now(struct inode *inode, int sync)
{
2302
	struct bdi_writeback *wb = &inode_to_bdi(inode)->wb;
L
Linus Torvalds 已提交
2303 2304
	struct writeback_control wbc = {
		.nr_to_write = LONG_MAX,
2305
		.sync_mode = sync ? WB_SYNC_ALL : WB_SYNC_NONE,
2306 2307
		.range_start = 0,
		.range_end = LLONG_MAX,
L
Linus Torvalds 已提交
2308 2309 2310
	};

	if (!mapping_cap_writeback_dirty(inode->i_mapping))
2311
		wbc.nr_to_write = 0;
L
Linus Torvalds 已提交
2312 2313

	might_sleep();
2314
	return writeback_single_inode(inode, wb, &wbc);
L
Linus Torvalds 已提交
2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330
}
EXPORT_SYMBOL(write_inode_now);

/**
 * sync_inode - write an inode and its pages to disk.
 * @inode: the inode to sync
 * @wbc: controls the writeback mode
 *
 * sync_inode() will write an inode and its pages to disk.  It will also
 * correctly update the inode on its superblock's dirty inode lists and will
 * update inode->i_state.
 *
 * The caller must have a ref on the inode.
 */
int sync_inode(struct inode *inode, struct writeback_control *wbc)
{
2331
	return writeback_single_inode(inode, &inode_to_bdi(inode)->wb, wbc);
L
Linus Torvalds 已提交
2332 2333
}
EXPORT_SYMBOL(sync_inode);
C
Christoph Hellwig 已提交
2334 2335

/**
A
Andrew Morton 已提交
2336
 * sync_inode_metadata - write an inode to disk
C
Christoph Hellwig 已提交
2337 2338 2339
 * @inode: the inode to sync
 * @wait: wait for I/O to complete.
 *
A
Andrew Morton 已提交
2340
 * Write an inode to disk and adjust its dirty state after completion.
C
Christoph Hellwig 已提交
2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353
 *
 * Note: only writes the actual inode, no associated data or other metadata.
 */
int sync_inode_metadata(struct inode *inode, int wait)
{
	struct writeback_control wbc = {
		.sync_mode = wait ? WB_SYNC_ALL : WB_SYNC_NONE,
		.nr_to_write = 0, /* metadata-only */
	};

	return sync_inode(inode, &wbc);
}
EXPORT_SYMBOL(sync_inode_metadata);