extent_io.c 147.5 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11
#include <linux/bitops.h>
#include <linux/slab.h>
#include <linux/bio.h>
#include <linux/mm.h>
#include <linux/pagemap.h>
#include <linux/page-flags.h>
#include <linux/spinlock.h>
#include <linux/blkdev.h>
#include <linux/swap.h>
#include <linux/writeback.h>
#include <linux/pagevec.h>
12
#include <linux/prefetch.h>
D
Dan Magenheimer 已提交
13
#include <linux/cleancache.h>
14 15
#include "extent_io.h"
#include "extent_map.h"
16 17
#include "ctree.h"
#include "btrfs_inode.h"
18
#include "volumes.h"
19
#include "check-integrity.h"
20
#include "locking.h"
21
#include "rcu-string.h"
22
#include "backref.h"
23 24 25

static struct kmem_cache *extent_state_cache;
static struct kmem_cache *extent_buffer_cache;
26
static struct bio_set *btrfs_bioset;
27

28 29 30 31 32
static inline bool extent_state_in_tree(const struct extent_state *state)
{
	return !RB_EMPTY_NODE(&state->rb_node);
}

33
#ifdef CONFIG_BTRFS_DEBUG
34 35
static LIST_HEAD(buffers);
static LIST_HEAD(states);
C
Chris Mason 已提交
36

C
Chris Mason 已提交
37
static DEFINE_SPINLOCK(leak_lock);
38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66

static inline
void btrfs_leak_debug_add(struct list_head *new, struct list_head *head)
{
	unsigned long flags;

	spin_lock_irqsave(&leak_lock, flags);
	list_add(new, head);
	spin_unlock_irqrestore(&leak_lock, flags);
}

static inline
void btrfs_leak_debug_del(struct list_head *entry)
{
	unsigned long flags;

	spin_lock_irqsave(&leak_lock, flags);
	list_del(entry);
	spin_unlock_irqrestore(&leak_lock, flags);
}

static inline
void btrfs_leak_debug_check(void)
{
	struct extent_state *state;
	struct extent_buffer *eb;

	while (!list_empty(&states)) {
		state = list_entry(states.next, struct extent_state, leak_list);
67
		pr_err("BTRFS: state leak: start %llu end %llu state %u in tree %d refs %d\n",
68 69
		       state->start, state->end, state->state,
		       extent_state_in_tree(state),
70
		       atomic_read(&state->refs));
71 72 73 74 75 76
		list_del(&state->leak_list);
		kmem_cache_free(extent_state_cache, state);
	}

	while (!list_empty(&buffers)) {
		eb = list_entry(buffers.next, struct extent_buffer, leak_list);
77
		printk(KERN_ERR "BTRFS: buffer leak start %llu len %lu "
78 79
		       "refs %d\n",
		       eb->start, eb->len, atomic_read(&eb->refs));
80 81 82 83
		list_del(&eb->leak_list);
		kmem_cache_free(extent_buffer_cache, eb);
	}
}
84

85 86
#define btrfs_debug_check_extent_io_range(tree, start, end)		\
	__btrfs_debug_check_extent_io_range(__func__, (tree), (start), (end))
87
static inline void __btrfs_debug_check_extent_io_range(const char *caller,
88
		struct extent_io_tree *tree, u64 start, u64 end)
89
{
90 91
	struct inode *inode;
	u64 isize;
92

93 94
	if (!tree->mapping)
		return;
95

96 97
	inode = tree->mapping->host;
	isize = i_size_read(inode);
98
	if (end >= PAGE_SIZE && (end % 2) == 0 && end != isize - 1) {
99 100
		btrfs_debug_rl(BTRFS_I(inode)->root->fs_info,
		    "%s: ino %llu isize %llu odd range [%llu,%llu]",
101
				caller, btrfs_ino(inode), isize, start, end);
102 103
	}
}
104 105 106 107
#else
#define btrfs_leak_debug_add(new, head)	do {} while (0)
#define btrfs_leak_debug_del(entry)	do {} while (0)
#define btrfs_leak_debug_check()	do {} while (0)
108
#define btrfs_debug_check_extent_io_range(c, s, e)	do {} while (0)
C
Chris Mason 已提交
109
#endif
110 111 112 113 114 115 116 117 118 119 120 121 122

#define BUFFER_LRU_MAX 64

struct tree_entry {
	u64 start;
	u64 end;
	struct rb_node rb_node;
};

struct extent_page_data {
	struct bio *bio;
	struct extent_io_tree *tree;
	get_extent_t *get_extent;
123
	unsigned long bio_flags;
124 125 126 127

	/* tells writepage not to lock the state bits for this range
	 * it still does the unlocking
	 */
128 129 130 131
	unsigned int extent_locked:1;

	/* tells the submit_bio code to use a WRITE_SYNC */
	unsigned int sync_io:1;
132 133
};

134 135 136 137 138 139 140 141 142 143
static void add_extent_changeset(struct extent_state *state, unsigned bits,
				 struct extent_changeset *changeset,
				 int set)
{
	int ret;

	if (!changeset)
		return;
	if (set && (state->state & bits) == bits)
		return;
144 145
	if (!set && (state->state & bits) == 0)
		return;
146 147 148 149 150 151 152
	changeset->bytes_changed += state->end - state->start + 1;
	ret = ulist_add(changeset->range_changed, state->start, state->end,
			GFP_ATOMIC);
	/* ENOMEM */
	BUG_ON(ret < 0);
}

153
static noinline void flush_write_bio(void *data);
154 155 156
static inline struct btrfs_fs_info *
tree_fs_info(struct extent_io_tree *tree)
{
157 158
	if (!tree->mapping)
		return NULL;
159 160
	return btrfs_sb(tree->mapping->host->i_sb);
}
161

162 163
int __init extent_io_init(void)
{
D
David Sterba 已提交
164
	extent_state_cache = kmem_cache_create("btrfs_extent_state",
165 166
			sizeof(struct extent_state), 0,
			SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD, NULL);
167 168 169
	if (!extent_state_cache)
		return -ENOMEM;

D
David Sterba 已提交
170
	extent_buffer_cache = kmem_cache_create("btrfs_extent_buffer",
171 172
			sizeof(struct extent_buffer), 0,
			SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD, NULL);
173 174
	if (!extent_buffer_cache)
		goto free_state_cache;
175 176 177 178 179

	btrfs_bioset = bioset_create(BIO_POOL_SIZE,
				     offsetof(struct btrfs_io_bio, bio));
	if (!btrfs_bioset)
		goto free_buffer_cache;
180 181 182 183

	if (bioset_integrity_create(btrfs_bioset, BIO_POOL_SIZE))
		goto free_bioset;

184 185
	return 0;

186 187 188 189
free_bioset:
	bioset_free(btrfs_bioset);
	btrfs_bioset = NULL;

190 191 192 193
free_buffer_cache:
	kmem_cache_destroy(extent_buffer_cache);
	extent_buffer_cache = NULL;

194 195
free_state_cache:
	kmem_cache_destroy(extent_state_cache);
196
	extent_state_cache = NULL;
197 198 199 200 201
	return -ENOMEM;
}

void extent_io_exit(void)
{
202
	btrfs_leak_debug_check();
203 204 205 206 207 208

	/*
	 * Make sure all delayed rcu free are flushed before we
	 * destroy caches.
	 */
	rcu_barrier();
209 210 211 212
	if (extent_state_cache)
		kmem_cache_destroy(extent_state_cache);
	if (extent_buffer_cache)
		kmem_cache_destroy(extent_buffer_cache);
213 214
	if (btrfs_bioset)
		bioset_free(btrfs_bioset);
215 216 217
}

void extent_io_tree_init(struct extent_io_tree *tree,
218
			 struct address_space *mapping)
219
{
220
	tree->state = RB_ROOT;
221 222
	tree->ops = NULL;
	tree->dirty_bytes = 0;
223
	spin_lock_init(&tree->lock);
224 225 226
	tree->mapping = mapping;
}

227
static struct extent_state *alloc_extent_state(gfp_t mask)
228 229 230 231
{
	struct extent_state *state;

	state = kmem_cache_alloc(extent_state_cache, mask);
232
	if (!state)
233 234
		return state;
	state->state = 0;
235
	state->failrec = NULL;
236
	RB_CLEAR_NODE(&state->rb_node);
237
	btrfs_leak_debug_add(&state->leak_list, &states);
238 239
	atomic_set(&state->refs, 1);
	init_waitqueue_head(&state->wq);
240
	trace_alloc_extent_state(state, mask, _RET_IP_);
241 242 243
	return state;
}

244
void free_extent_state(struct extent_state *state)
245 246 247 248
{
	if (!state)
		return;
	if (atomic_dec_and_test(&state->refs)) {
249
		WARN_ON(extent_state_in_tree(state));
250
		btrfs_leak_debug_del(&state->leak_list);
251
		trace_free_extent_state(state, _RET_IP_);
252 253 254 255
		kmem_cache_free(extent_state_cache, state);
	}
}

256 257 258
static struct rb_node *tree_insert(struct rb_root *root,
				   struct rb_node *search_start,
				   u64 offset,
259 260 261
				   struct rb_node *node,
				   struct rb_node ***p_in,
				   struct rb_node **parent_in)
262
{
263
	struct rb_node **p;
C
Chris Mason 已提交
264
	struct rb_node *parent = NULL;
265 266
	struct tree_entry *entry;

267 268 269 270 271 272
	if (p_in && parent_in) {
		p = *p_in;
		parent = *parent_in;
		goto do_insert;
	}

273
	p = search_start ? &search_start : &root->rb_node;
C
Chris Mason 已提交
274
	while (*p) {
275 276 277 278 279 280 281 282 283 284 285
		parent = *p;
		entry = rb_entry(parent, struct tree_entry, rb_node);

		if (offset < entry->start)
			p = &(*p)->rb_left;
		else if (offset > entry->end)
			p = &(*p)->rb_right;
		else
			return parent;
	}

286
do_insert:
287 288 289 290 291
	rb_link_node(node, parent, p);
	rb_insert_color(node, root);
	return NULL;
}

292
static struct rb_node *__etree_search(struct extent_io_tree *tree, u64 offset,
293 294 295 296
				      struct rb_node **prev_ret,
				      struct rb_node **next_ret,
				      struct rb_node ***p_ret,
				      struct rb_node **parent_ret)
297
{
298
	struct rb_root *root = &tree->state;
299
	struct rb_node **n = &root->rb_node;
300 301 302 303 304
	struct rb_node *prev = NULL;
	struct rb_node *orig_prev = NULL;
	struct tree_entry *entry;
	struct tree_entry *prev_entry = NULL;

305 306 307
	while (*n) {
		prev = *n;
		entry = rb_entry(prev, struct tree_entry, rb_node);
308 309 310
		prev_entry = entry;

		if (offset < entry->start)
311
			n = &(*n)->rb_left;
312
		else if (offset > entry->end)
313
			n = &(*n)->rb_right;
C
Chris Mason 已提交
314
		else
315
			return *n;
316 317
	}

318 319 320 321 322
	if (p_ret)
		*p_ret = n;
	if (parent_ret)
		*parent_ret = prev;

323 324
	if (prev_ret) {
		orig_prev = prev;
C
Chris Mason 已提交
325
		while (prev && offset > prev_entry->end) {
326 327 328 329 330 331 332 333 334
			prev = rb_next(prev);
			prev_entry = rb_entry(prev, struct tree_entry, rb_node);
		}
		*prev_ret = prev;
		prev = orig_prev;
	}

	if (next_ret) {
		prev_entry = rb_entry(prev, struct tree_entry, rb_node);
C
Chris Mason 已提交
335
		while (prev && offset < prev_entry->start) {
336 337 338 339 340 341 342 343
			prev = rb_prev(prev);
			prev_entry = rb_entry(prev, struct tree_entry, rb_node);
		}
		*next_ret = prev;
	}
	return NULL;
}

344 345 346 347 348
static inline struct rb_node *
tree_search_for_insert(struct extent_io_tree *tree,
		       u64 offset,
		       struct rb_node ***p_ret,
		       struct rb_node **parent_ret)
349
{
350
	struct rb_node *prev = NULL;
351
	struct rb_node *ret;
352

353
	ret = __etree_search(tree, offset, &prev, NULL, p_ret, parent_ret);
C
Chris Mason 已提交
354
	if (!ret)
355 356 357 358
		return prev;
	return ret;
}

359 360 361 362 363 364
static inline struct rb_node *tree_search(struct extent_io_tree *tree,
					  u64 offset)
{
	return tree_search_for_insert(tree, offset, NULL, NULL);
}

J
Josef Bacik 已提交
365 366 367 368 369 370 371 372
static void merge_cb(struct extent_io_tree *tree, struct extent_state *new,
		     struct extent_state *other)
{
	if (tree->ops && tree->ops->merge_extent_hook)
		tree->ops->merge_extent_hook(tree->mapping->host, new,
					     other);
}

373 374 375 376 377 378 379 380 381
/*
 * utility function to look for merge candidates inside a given range.
 * Any extents with matching state are merged together into a single
 * extent in the tree.  Extents with EXTENT_IO in their state field
 * are not merged because the end_io handlers need to be able to do
 * operations on them without sleeping (or doing allocations/splits).
 *
 * This should be called with the tree lock held.
 */
382 383
static void merge_state(struct extent_io_tree *tree,
		        struct extent_state *state)
384 385 386 387
{
	struct extent_state *other;
	struct rb_node *other_node;

388
	if (state->state & (EXTENT_IOBITS | EXTENT_BOUNDARY))
389
		return;
390 391 392 393 394 395

	other_node = rb_prev(&state->rb_node);
	if (other_node) {
		other = rb_entry(other_node, struct extent_state, rb_node);
		if (other->end == state->start - 1 &&
		    other->state == state->state) {
J
Josef Bacik 已提交
396
			merge_cb(tree, state, other);
397 398
			state->start = other->start;
			rb_erase(&other->rb_node, &tree->state);
399
			RB_CLEAR_NODE(&other->rb_node);
400 401 402 403 404 405 406 407
			free_extent_state(other);
		}
	}
	other_node = rb_next(&state->rb_node);
	if (other_node) {
		other = rb_entry(other_node, struct extent_state, rb_node);
		if (other->start == state->end + 1 &&
		    other->state == state->state) {
J
Josef Bacik 已提交
408
			merge_cb(tree, state, other);
409 410
			state->end = other->end;
			rb_erase(&other->rb_node, &tree->state);
411
			RB_CLEAR_NODE(&other->rb_node);
412
			free_extent_state(other);
413 414 415 416
		}
	}
}

417
static void set_state_cb(struct extent_io_tree *tree,
418
			 struct extent_state *state, unsigned *bits)
419
{
420 421
	if (tree->ops && tree->ops->set_bit_hook)
		tree->ops->set_bit_hook(tree->mapping->host, state, bits);
422 423 424
}

static void clear_state_cb(struct extent_io_tree *tree,
425
			   struct extent_state *state, unsigned *bits)
426
{
J
Josef Bacik 已提交
427 428
	if (tree->ops && tree->ops->clear_bit_hook)
		tree->ops->clear_bit_hook(tree->mapping->host, state, bits);
429 430
}

431
static void set_state_bits(struct extent_io_tree *tree,
432 433
			   struct extent_state *state, unsigned *bits,
			   struct extent_changeset *changeset);
434

435 436 437 438 439 440 441 442 443 444 445 446
/*
 * insert an extent_state struct into the tree.  'bits' are set on the
 * struct before it is inserted.
 *
 * This may return -EEXIST if the extent is already there, in which case the
 * state struct is freed.
 *
 * The tree lock is not taken internally.  This is a utility function and
 * probably isn't what you want to call (see set/clear_extent_bit).
 */
static int insert_state(struct extent_io_tree *tree,
			struct extent_state *state, u64 start, u64 end,
447 448
			struct rb_node ***p,
			struct rb_node **parent,
449
			unsigned *bits, struct extent_changeset *changeset)
450 451 452
{
	struct rb_node *node;

J
Julia Lawall 已提交
453
	if (end < start)
454
		WARN(1, KERN_ERR "BTRFS: end < start %llu %llu\n",
455
		       end, start);
456 457
	state->start = start;
	state->end = end;
J
Josef Bacik 已提交
458

459
	set_state_bits(tree, state, bits, changeset);
460

461
	node = tree_insert(&tree->state, NULL, end, &state->rb_node, p, parent);
462 463 464
	if (node) {
		struct extent_state *found;
		found = rb_entry(node, struct extent_state, rb_node);
465
		printk(KERN_ERR "BTRFS: found node %llu %llu on insert of "
466 467
		       "%llu %llu\n",
		       found->start, found->end, start, end);
468 469 470 471 472 473
		return -EEXIST;
	}
	merge_state(tree, state);
	return 0;
}

474
static void split_cb(struct extent_io_tree *tree, struct extent_state *orig,
J
Josef Bacik 已提交
475 476 477
		     u64 split)
{
	if (tree->ops && tree->ops->split_extent_hook)
478
		tree->ops->split_extent_hook(tree->mapping->host, orig, split);
J
Josef Bacik 已提交
479 480
}

481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498
/*
 * split a given extent state struct in two, inserting the preallocated
 * struct 'prealloc' as the newly created second half.  'split' indicates an
 * offset inside 'orig' where it should be split.
 *
 * Before calling,
 * the tree has 'orig' at [orig->start, orig->end].  After calling, there
 * are two extent state structs in the tree:
 * prealloc: [orig->start, split - 1]
 * orig: [ split, orig->end ]
 *
 * The tree locks are not taken by this function. They need to be held
 * by the caller.
 */
static int split_state(struct extent_io_tree *tree, struct extent_state *orig,
		       struct extent_state *prealloc, u64 split)
{
	struct rb_node *node;
J
Josef Bacik 已提交
499 500 501

	split_cb(tree, orig, split);

502 503 504 505 506
	prealloc->start = orig->start;
	prealloc->end = split - 1;
	prealloc->state = orig->state;
	orig->start = split;

507 508
	node = tree_insert(&tree->state, &orig->rb_node, prealloc->end,
			   &prealloc->rb_node, NULL, NULL);
509 510 511 512 513 514 515
	if (node) {
		free_extent_state(prealloc);
		return -EEXIST;
	}
	return 0;
}

516 517 518 519 520 521 522 523 524
static struct extent_state *next_state(struct extent_state *state)
{
	struct rb_node *next = rb_next(&state->rb_node);
	if (next)
		return rb_entry(next, struct extent_state, rb_node);
	else
		return NULL;
}

525 526
/*
 * utility function to clear some bits in an extent state struct.
527
 * it will optionally wake up any one waiting on this state (wake == 1).
528 529 530 531
 *
 * If no bits are set on the state struct after clearing things, the
 * struct is freed and removed from the tree
 */
532 533
static struct extent_state *clear_state_bit(struct extent_io_tree *tree,
					    struct extent_state *state,
534 535
					    unsigned *bits, int wake,
					    struct extent_changeset *changeset)
536
{
537
	struct extent_state *next;
538
	unsigned bits_to_clear = *bits & ~EXTENT_CTLBITS;
539

540
	if ((bits_to_clear & EXTENT_DIRTY) && (state->state & EXTENT_DIRTY)) {
541 542 543 544
		u64 range = state->end - state->start + 1;
		WARN_ON(range > tree->dirty_bytes);
		tree->dirty_bytes -= range;
	}
545
	clear_state_cb(tree, state, bits);
546
	add_extent_changeset(state, bits_to_clear, changeset, 0);
547
	state->state &= ~bits_to_clear;
548 549
	if (wake)
		wake_up(&state->wq);
550
	if (state->state == 0) {
551
		next = next_state(state);
552
		if (extent_state_in_tree(state)) {
553
			rb_erase(&state->rb_node, &tree->state);
554
			RB_CLEAR_NODE(&state->rb_node);
555 556 557 558 559 560
			free_extent_state(state);
		} else {
			WARN_ON(1);
		}
	} else {
		merge_state(tree, state);
561
		next = next_state(state);
562
	}
563
	return next;
564 565
}

566 567 568 569 570 571 572 573 574
static struct extent_state *
alloc_extent_state_atomic(struct extent_state *prealloc)
{
	if (!prealloc)
		prealloc = alloc_extent_state(GFP_ATOMIC);

	return prealloc;
}

575
static void extent_io_tree_panic(struct extent_io_tree *tree, int err)
576 577 578 579 580 581
{
	btrfs_panic(tree_fs_info(tree), err, "Locking error: "
		    "Extent tree was modified by another "
		    "thread while locked.");
}

582 583 584 585 586 587 588 589 590 591
/*
 * clear some bits on a range in the tree.  This may require splitting
 * or inserting elements in the tree, so the gfp mask is used to
 * indicate which allocations or sleeping are allowed.
 *
 * pass 'wake' == 1 to kick any sleepers, and 'delete' == 1 to remove
 * the given range from the tree regardless of state (ie for truncate).
 *
 * the range [start, end] is inclusive.
 *
592
 * This takes the tree lock, and returns 0 on success and < 0 on error.
593
 */
594 595 596 597
static int __clear_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
			      unsigned bits, int wake, int delete,
			      struct extent_state **cached_state,
			      gfp_t mask, struct extent_changeset *changeset)
598 599
{
	struct extent_state *state;
600
	struct extent_state *cached;
601 602
	struct extent_state *prealloc = NULL;
	struct rb_node *node;
603
	u64 last_end;
604
	int err;
605
	int clear = 0;
606

607
	btrfs_debug_check_extent_io_range(tree, start, end);
608

609 610 611
	if (bits & EXTENT_DELALLOC)
		bits |= EXTENT_NORESERVE;

612 613 614 615
	if (delete)
		bits |= ~EXTENT_CTLBITS;
	bits |= EXTENT_FIRST_DELALLOC;

616 617
	if (bits & (EXTENT_IOBITS | EXTENT_BOUNDARY))
		clear = 1;
618
again:
619
	if (!prealloc && gfpflags_allow_blocking(mask)) {
620 621 622 623 624 625 626
		/*
		 * Don't care for allocation failure here because we might end
		 * up not needing the pre-allocated extent state at all, which
		 * is the case if we only have in the tree extent states that
		 * cover our input range and don't cover too any other range.
		 * If we end up needing a new extent state we allocate it later.
		 */
627 628 629
		prealloc = alloc_extent_state(mask);
	}

630
	spin_lock(&tree->lock);
631 632
	if (cached_state) {
		cached = *cached_state;
633 634 635 636 637 638

		if (clear) {
			*cached_state = NULL;
			cached_state = NULL;
		}

639 640
		if (cached && extent_state_in_tree(cached) &&
		    cached->start <= start && cached->end > start) {
641 642
			if (clear)
				atomic_dec(&cached->refs);
643
			state = cached;
644
			goto hit_next;
645
		}
646 647
		if (clear)
			free_extent_state(cached);
648
	}
649 650 651 652
	/*
	 * this search will find the extents that end after
	 * our range starts
	 */
653
	node = tree_search(tree, start);
654 655 656
	if (!node)
		goto out;
	state = rb_entry(node, struct extent_state, rb_node);
657
hit_next:
658 659 660
	if (state->start > end)
		goto out;
	WARN_ON(state->end < start);
661
	last_end = state->end;
662

663
	/* the state doesn't have the wanted bits, go ahead */
664 665
	if (!(state->state & bits)) {
		state = next_state(state);
666
		goto next;
667
	}
668

669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685
	/*
	 *     | ---- desired range ---- |
	 *  | state | or
	 *  | ------------- state -------------- |
	 *
	 * We need to split the extent we found, and may flip
	 * bits on second half.
	 *
	 * If the extent we found extends past our range, we
	 * just split and search again.  It'll get split again
	 * the next time though.
	 *
	 * If the extent we found is inside our range, we clear
	 * the desired bit on it.
	 */

	if (state->start < start) {
686 687
		prealloc = alloc_extent_state_atomic(prealloc);
		BUG_ON(!prealloc);
688
		err = split_state(tree, state, prealloc, start);
689 690 691
		if (err)
			extent_io_tree_panic(tree, err);

692 693 694 695
		prealloc = NULL;
		if (err)
			goto out;
		if (state->end <= end) {
696 697
			state = clear_state_bit(tree, state, &bits, wake,
						changeset);
698
			goto next;
699 700 701 702 703 704 705 706 707 708
		}
		goto search_again;
	}
	/*
	 * | ---- desired range ---- |
	 *                        | state |
	 * We need to split the extent, and clear the bit
	 * on the first half
	 */
	if (state->start <= end && state->end > end) {
709 710
		prealloc = alloc_extent_state_atomic(prealloc);
		BUG_ON(!prealloc);
711
		err = split_state(tree, state, prealloc, end + 1);
712 713 714
		if (err)
			extent_io_tree_panic(tree, err);

715 716
		if (wake)
			wake_up(&state->wq);
717

718
		clear_state_bit(tree, prealloc, &bits, wake, changeset);
J
Josef Bacik 已提交
719

720 721 722
		prealloc = NULL;
		goto out;
	}
723

724
	state = clear_state_bit(tree, state, &bits, wake, changeset);
725
next:
726 727 728
	if (last_end == (u64)-1)
		goto out;
	start = last_end + 1;
729
	if (start <= end && state && !need_resched())
730
		goto hit_next;
731 732 733
	goto search_again;

out:
734
	spin_unlock(&tree->lock);
735 736 737
	if (prealloc)
		free_extent_state(prealloc);

738
	return 0;
739 740 741 742

search_again:
	if (start > end)
		goto out;
743
	spin_unlock(&tree->lock);
744
	if (gfpflags_allow_blocking(mask))
745 746 747 748
		cond_resched();
	goto again;
}

749 750
static void wait_on_state(struct extent_io_tree *tree,
			  struct extent_state *state)
751 752
		__releases(tree->lock)
		__acquires(tree->lock)
753 754 755
{
	DEFINE_WAIT(wait);
	prepare_to_wait(&state->wq, &wait, TASK_UNINTERRUPTIBLE);
756
	spin_unlock(&tree->lock);
757
	schedule();
758
	spin_lock(&tree->lock);
759 760 761 762 763 764 765 766
	finish_wait(&state->wq, &wait);
}

/*
 * waits for one or more bits to clear on a range in the state tree.
 * The range [start, end] is inclusive.
 * The tree lock is taken by this function
 */
767 768
static void wait_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
			    unsigned long bits)
769 770 771 772
{
	struct extent_state *state;
	struct rb_node *node;

773
	btrfs_debug_check_extent_io_range(tree, start, end);
774

775
	spin_lock(&tree->lock);
776 777 778 779 780 781
again:
	while (1) {
		/*
		 * this search will find all the extents that end after
		 * our range starts
		 */
782
		node = tree_search(tree, start);
783
process_node:
784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803
		if (!node)
			break;

		state = rb_entry(node, struct extent_state, rb_node);

		if (state->start > end)
			goto out;

		if (state->state & bits) {
			start = state->start;
			atomic_inc(&state->refs);
			wait_on_state(tree, state);
			free_extent_state(state);
			goto again;
		}
		start = state->end + 1;

		if (start > end)
			break;

804 805 806 807
		if (!cond_resched_lock(&tree->lock)) {
			node = rb_next(node);
			goto process_node;
		}
808 809
	}
out:
810
	spin_unlock(&tree->lock);
811 812
}

813
static void set_state_bits(struct extent_io_tree *tree,
814
			   struct extent_state *state,
815
			   unsigned *bits, struct extent_changeset *changeset)
816
{
817
	unsigned bits_to_set = *bits & ~EXTENT_CTLBITS;
J
Josef Bacik 已提交
818

819
	set_state_cb(tree, state, bits);
820
	if ((bits_to_set & EXTENT_DIRTY) && !(state->state & EXTENT_DIRTY)) {
821 822 823
		u64 range = state->end - state->start + 1;
		tree->dirty_bytes += range;
	}
824
	add_extent_changeset(state, bits_to_set, changeset, 1);
825
	state->state |= bits_to_set;
826 827
}

828 829
static void cache_state_if_flags(struct extent_state *state,
				 struct extent_state **cached_ptr,
830
				 unsigned flags)
831 832
{
	if (cached_ptr && !(*cached_ptr)) {
833
		if (!flags || (state->state & flags)) {
834 835 836 837 838 839
			*cached_ptr = state;
			atomic_inc(&state->refs);
		}
	}
}

840 841 842 843 844 845 846
static void cache_state(struct extent_state *state,
			struct extent_state **cached_ptr)
{
	return cache_state_if_flags(state, cached_ptr,
				    EXTENT_IOBITS | EXTENT_BOUNDARY);
}

847
/*
848 849
 * set some bits on a range in the tree.  This may require allocations or
 * sleeping, so the gfp mask is used to indicate what is allowed.
850
 *
851 852 853
 * If any of the exclusive bits are set, this will fail with -EEXIST if some
 * part of the range already has the desired bits set.  The start of the
 * existing range is returned in failed_start in this case.
854
 *
855
 * [start, end] is inclusive This takes the tree lock.
856
 */
857

J
Jeff Mahoney 已提交
858 859
static int __must_check
__set_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
860
		 unsigned bits, unsigned exclusive_bits,
861
		 u64 *failed_start, struct extent_state **cached_state,
862
		 gfp_t mask, struct extent_changeset *changeset)
863 864 865 866
{
	struct extent_state *state;
	struct extent_state *prealloc = NULL;
	struct rb_node *node;
867 868
	struct rb_node **p;
	struct rb_node *parent;
869 870 871
	int err = 0;
	u64 last_start;
	u64 last_end;
872

873
	btrfs_debug_check_extent_io_range(tree, start, end);
874

875
	bits |= EXTENT_FIRST_DELALLOC;
876
again:
877
	if (!prealloc && gfpflags_allow_blocking(mask)) {
878
		prealloc = alloc_extent_state(mask);
879
		BUG_ON(!prealloc);
880 881
	}

882
	spin_lock(&tree->lock);
883 884
	if (cached_state && *cached_state) {
		state = *cached_state;
885
		if (state->start <= start && state->end > start &&
886
		    extent_state_in_tree(state)) {
887 888 889 890
			node = &state->rb_node;
			goto hit_next;
		}
	}
891 892 893 894
	/*
	 * this search will find all the extents that end after
	 * our range starts.
	 */
895
	node = tree_search_for_insert(tree, start, &p, &parent);
896
	if (!node) {
897 898
		prealloc = alloc_extent_state_atomic(prealloc);
		BUG_ON(!prealloc);
899
		err = insert_state(tree, prealloc, start, end,
900
				   &p, &parent, &bits, changeset);
901 902 903
		if (err)
			extent_io_tree_panic(tree, err);

904
		cache_state(prealloc, cached_state);
905 906 907 908
		prealloc = NULL;
		goto out;
	}
	state = rb_entry(node, struct extent_state, rb_node);
C
Chris Mason 已提交
909
hit_next:
910 911 912 913 914 915 916 917 918 919
	last_start = state->start;
	last_end = state->end;

	/*
	 * | ---- desired range ---- |
	 * | state |
	 *
	 * Just lock what we found and keep going
	 */
	if (state->start == start && state->end <= end) {
920
		if (state->state & exclusive_bits) {
921 922 923 924
			*failed_start = state->start;
			err = -EEXIST;
			goto out;
		}
925

926
		set_state_bits(tree, state, &bits, changeset);
927
		cache_state(state, cached_state);
928
		merge_state(tree, state);
929 930 931
		if (last_end == (u64)-1)
			goto out;
		start = last_end + 1;
932 933 934 935
		state = next_state(state);
		if (start < end && state && state->start == start &&
		    !need_resched())
			goto hit_next;
936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955
		goto search_again;
	}

	/*
	 *     | ---- desired range ---- |
	 * | state |
	 *   or
	 * | ------------- state -------------- |
	 *
	 * We need to split the extent we found, and may flip bits on
	 * second half.
	 *
	 * If the extent we found extends past our
	 * range, we just split and search again.  It'll get split
	 * again the next time though.
	 *
	 * If the extent we found is inside our range, we set the
	 * desired bit on it.
	 */
	if (state->start < start) {
956
		if (state->state & exclusive_bits) {
957 958 959 960
			*failed_start = start;
			err = -EEXIST;
			goto out;
		}
961 962 963

		prealloc = alloc_extent_state_atomic(prealloc);
		BUG_ON(!prealloc);
964
		err = split_state(tree, state, prealloc, start);
965 966 967
		if (err)
			extent_io_tree_panic(tree, err);

968 969 970 971
		prealloc = NULL;
		if (err)
			goto out;
		if (state->end <= end) {
972
			set_state_bits(tree, state, &bits, changeset);
973
			cache_state(state, cached_state);
974
			merge_state(tree, state);
975 976 977
			if (last_end == (u64)-1)
				goto out;
			start = last_end + 1;
978 979 980 981
			state = next_state(state);
			if (start < end && state && state->start == start &&
			    !need_resched())
				goto hit_next;
982 983 984 985 986 987 988 989 990 991 992 993 994 995 996
		}
		goto search_again;
	}
	/*
	 * | ---- desired range ---- |
	 *     | state | or               | state |
	 *
	 * There's a hole, we need to insert something in it and
	 * ignore the extent we found.
	 */
	if (state->start > start) {
		u64 this_end;
		if (end < last_start)
			this_end = end;
		else
C
Chris Mason 已提交
997
			this_end = last_start - 1;
998 999 1000

		prealloc = alloc_extent_state_atomic(prealloc);
		BUG_ON(!prealloc);
1001 1002 1003 1004 1005

		/*
		 * Avoid to free 'prealloc' if it can be merged with
		 * the later extent.
		 */
1006
		err = insert_state(tree, prealloc, start, this_end,
1007
				   NULL, NULL, &bits, changeset);
1008 1009 1010
		if (err)
			extent_io_tree_panic(tree, err);

J
Josef Bacik 已提交
1011 1012
		cache_state(prealloc, cached_state);
		prealloc = NULL;
1013 1014 1015 1016 1017 1018 1019 1020 1021 1022
		start = this_end + 1;
		goto search_again;
	}
	/*
	 * | ---- desired range ---- |
	 *                        | state |
	 * We need to split the extent, and set the bit
	 * on the first half
	 */
	if (state->start <= end && state->end > end) {
1023
		if (state->state & exclusive_bits) {
1024 1025 1026 1027
			*failed_start = start;
			err = -EEXIST;
			goto out;
		}
1028 1029 1030

		prealloc = alloc_extent_state_atomic(prealloc);
		BUG_ON(!prealloc);
1031
		err = split_state(tree, state, prealloc, end + 1);
1032 1033
		if (err)
			extent_io_tree_panic(tree, err);
1034

1035
		set_state_bits(tree, prealloc, &bits, changeset);
1036
		cache_state(prealloc, cached_state);
1037 1038 1039 1040 1041 1042 1043 1044
		merge_state(tree, prealloc);
		prealloc = NULL;
		goto out;
	}

	goto search_again;

out:
1045
	spin_unlock(&tree->lock);
1046 1047 1048 1049 1050 1051 1052 1053
	if (prealloc)
		free_extent_state(prealloc);

	return err;

search_again:
	if (start > end)
		goto out;
1054
	spin_unlock(&tree->lock);
1055
	if (gfpflags_allow_blocking(mask))
1056 1057 1058 1059
		cond_resched();
	goto again;
}

1060
int set_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
1061
		   unsigned bits, u64 * failed_start,
1062
		   struct extent_state **cached_state, gfp_t mask)
J
Jeff Mahoney 已提交
1063 1064
{
	return __set_extent_bit(tree, start, end, bits, 0, failed_start,
1065
				cached_state, mask, NULL);
J
Jeff Mahoney 已提交
1066 1067 1068
}


J
Josef Bacik 已提交
1069
/**
L
Liu Bo 已提交
1070 1071
 * convert_extent_bit - convert all bits in a given range from one bit to
 * 			another
J
Josef Bacik 已提交
1072 1073 1074 1075 1076
 * @tree:	the io tree to search
 * @start:	the start offset in bytes
 * @end:	the end offset in bytes (inclusive)
 * @bits:	the bits to set in this range
 * @clear_bits:	the bits to clear in this range
1077
 * @cached_state:	state that we're going to cache
J
Josef Bacik 已提交
1078 1079 1080 1081 1082 1083 1084 1085 1086
 * @mask:	the allocation mask
 *
 * This will go through and set bits for the given range.  If any states exist
 * already in this range they are set with the given bit and cleared of the
 * clear_bits.  This is only meant to be used by things that are mergeable, ie
 * converting from say DELALLOC to DIRTY.  This is not meant to be used with
 * boundary bits like LOCK.
 */
int convert_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
1087
		       unsigned bits, unsigned clear_bits,
1088
		       struct extent_state **cached_state, gfp_t mask)
J
Josef Bacik 已提交
1089 1090 1091 1092
{
	struct extent_state *state;
	struct extent_state *prealloc = NULL;
	struct rb_node *node;
1093 1094
	struct rb_node **p;
	struct rb_node *parent;
J
Josef Bacik 已提交
1095 1096 1097
	int err = 0;
	u64 last_start;
	u64 last_end;
1098
	bool first_iteration = true;
J
Josef Bacik 已提交
1099

1100
	btrfs_debug_check_extent_io_range(tree, start, end);
1101

J
Josef Bacik 已提交
1102
again:
1103
	if (!prealloc && gfpflags_allow_blocking(mask)) {
1104 1105 1106 1107 1108 1109 1110
		/*
		 * Best effort, don't worry if extent state allocation fails
		 * here for the first iteration. We might have a cached state
		 * that matches exactly the target range, in which case no
		 * extent state allocations are needed. We'll only know this
		 * after locking the tree.
		 */
J
Josef Bacik 已提交
1111
		prealloc = alloc_extent_state(mask);
1112
		if (!prealloc && !first_iteration)
J
Josef Bacik 已提交
1113 1114 1115 1116
			return -ENOMEM;
	}

	spin_lock(&tree->lock);
1117 1118 1119
	if (cached_state && *cached_state) {
		state = *cached_state;
		if (state->start <= start && state->end > start &&
1120
		    extent_state_in_tree(state)) {
1121 1122 1123 1124 1125
			node = &state->rb_node;
			goto hit_next;
		}
	}

J
Josef Bacik 已提交
1126 1127 1128 1129
	/*
	 * this search will find all the extents that end after
	 * our range starts.
	 */
1130
	node = tree_search_for_insert(tree, start, &p, &parent);
J
Josef Bacik 已提交
1131 1132
	if (!node) {
		prealloc = alloc_extent_state_atomic(prealloc);
1133 1134 1135 1136
		if (!prealloc) {
			err = -ENOMEM;
			goto out;
		}
1137
		err = insert_state(tree, prealloc, start, end,
1138
				   &p, &parent, &bits, NULL);
1139 1140
		if (err)
			extent_io_tree_panic(tree, err);
1141 1142
		cache_state(prealloc, cached_state);
		prealloc = NULL;
J
Josef Bacik 已提交
1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156
		goto out;
	}
	state = rb_entry(node, struct extent_state, rb_node);
hit_next:
	last_start = state->start;
	last_end = state->end;

	/*
	 * | ---- desired range ---- |
	 * | state |
	 *
	 * Just lock what we found and keep going
	 */
	if (state->start == start && state->end <= end) {
1157
		set_state_bits(tree, state, &bits, NULL);
1158
		cache_state(state, cached_state);
1159
		state = clear_state_bit(tree, state, &clear_bits, 0, NULL);
J
Josef Bacik 已提交
1160 1161 1162
		if (last_end == (u64)-1)
			goto out;
		start = last_end + 1;
1163 1164 1165
		if (start < end && state && state->start == start &&
		    !need_resched())
			goto hit_next;
J
Josef Bacik 已提交
1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186
		goto search_again;
	}

	/*
	 *     | ---- desired range ---- |
	 * | state |
	 *   or
	 * | ------------- state -------------- |
	 *
	 * We need to split the extent we found, and may flip bits on
	 * second half.
	 *
	 * If the extent we found extends past our
	 * range, we just split and search again.  It'll get split
	 * again the next time though.
	 *
	 * If the extent we found is inside our range, we set the
	 * desired bit on it.
	 */
	if (state->start < start) {
		prealloc = alloc_extent_state_atomic(prealloc);
1187 1188 1189 1190
		if (!prealloc) {
			err = -ENOMEM;
			goto out;
		}
J
Josef Bacik 已提交
1191
		err = split_state(tree, state, prealloc, start);
1192 1193
		if (err)
			extent_io_tree_panic(tree, err);
J
Josef Bacik 已提交
1194 1195 1196 1197
		prealloc = NULL;
		if (err)
			goto out;
		if (state->end <= end) {
1198
			set_state_bits(tree, state, &bits, NULL);
1199
			cache_state(state, cached_state);
1200 1201
			state = clear_state_bit(tree, state, &clear_bits, 0,
						NULL);
J
Josef Bacik 已提交
1202 1203 1204
			if (last_end == (u64)-1)
				goto out;
			start = last_end + 1;
1205 1206 1207
			if (start < end && state && state->start == start &&
			    !need_resched())
				goto hit_next;
J
Josef Bacik 已提交
1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225
		}
		goto search_again;
	}
	/*
	 * | ---- desired range ---- |
	 *     | state | or               | state |
	 *
	 * There's a hole, we need to insert something in it and
	 * ignore the extent we found.
	 */
	if (state->start > start) {
		u64 this_end;
		if (end < last_start)
			this_end = end;
		else
			this_end = last_start - 1;

		prealloc = alloc_extent_state_atomic(prealloc);
1226 1227 1228 1229
		if (!prealloc) {
			err = -ENOMEM;
			goto out;
		}
J
Josef Bacik 已提交
1230 1231 1232 1233 1234 1235

		/*
		 * Avoid to free 'prealloc' if it can be merged with
		 * the later extent.
		 */
		err = insert_state(tree, prealloc, start, this_end,
1236
				   NULL, NULL, &bits, NULL);
1237 1238
		if (err)
			extent_io_tree_panic(tree, err);
1239
		cache_state(prealloc, cached_state);
J
Josef Bacik 已提交
1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251
		prealloc = NULL;
		start = this_end + 1;
		goto search_again;
	}
	/*
	 * | ---- desired range ---- |
	 *                        | state |
	 * We need to split the extent, and set the bit
	 * on the first half
	 */
	if (state->start <= end && state->end > end) {
		prealloc = alloc_extent_state_atomic(prealloc);
1252 1253 1254 1255
		if (!prealloc) {
			err = -ENOMEM;
			goto out;
		}
J
Josef Bacik 已提交
1256 1257

		err = split_state(tree, state, prealloc, end + 1);
1258 1259
		if (err)
			extent_io_tree_panic(tree, err);
J
Josef Bacik 已提交
1260

1261
		set_state_bits(tree, prealloc, &bits, NULL);
1262
		cache_state(prealloc, cached_state);
1263
		clear_state_bit(tree, prealloc, &clear_bits, 0, NULL);
J
Josef Bacik 已提交
1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280
		prealloc = NULL;
		goto out;
	}

	goto search_again;

out:
	spin_unlock(&tree->lock);
	if (prealloc)
		free_extent_state(prealloc);

	return err;

search_again:
	if (start > end)
		goto out;
	spin_unlock(&tree->lock);
1281
	if (gfpflags_allow_blocking(mask))
J
Josef Bacik 已提交
1282
		cond_resched();
1283
	first_iteration = false;
J
Josef Bacik 已提交
1284 1285 1286
	goto again;
}

1287
/* wrappers around set/clear extent bit */
1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303
int set_record_extent_bits(struct extent_io_tree *tree, u64 start, u64 end,
			   unsigned bits, gfp_t mask,
			   struct extent_changeset *changeset)
{
	/*
	 * We don't support EXTENT_LOCKED yet, as current changeset will
	 * record any bits changed, so for EXTENT_LOCKED case, it will
	 * either fail with -EEXIST or changeset will record the whole
	 * range.
	 */
	BUG_ON(bits & EXTENT_LOCKED);

	return __set_extent_bit(tree, start, end, bits, 0, NULL, NULL, mask,
				changeset);
}

1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325
int clear_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
		     unsigned bits, int wake, int delete,
		     struct extent_state **cached, gfp_t mask)
{
	return __clear_extent_bit(tree, start, end, bits, wake, delete,
				  cached, mask, NULL);
}

int clear_record_extent_bits(struct extent_io_tree *tree, u64 start, u64 end,
			     unsigned bits, gfp_t mask,
			     struct extent_changeset *changeset)
{
	/*
	 * Don't support EXTENT_LOCKED case, same reason as
	 * set_record_extent_bits().
	 */
	BUG_ON(bits & EXTENT_LOCKED);

	return __clear_extent_bit(tree, start, end, bits, 0, 0, NULL, mask,
				  changeset);
}

C
Chris Mason 已提交
1326 1327 1328 1329
/*
 * either insert or lock state struct between start and end use mask to tell
 * us if waiting is desired.
 */
1330
int lock_extent_bits(struct extent_io_tree *tree, u64 start, u64 end,
1331
		     struct extent_state **cached_state)
1332 1333 1334
{
	int err;
	u64 failed_start;
1335

1336
	while (1) {
1337
		err = __set_extent_bit(tree, start, end, EXTENT_LOCKED,
J
Jeff Mahoney 已提交
1338
				       EXTENT_LOCKED, &failed_start,
1339
				       cached_state, GFP_NOFS, NULL);
1340
		if (err == -EEXIST) {
1341 1342
			wait_extent_bit(tree, failed_start, end, EXTENT_LOCKED);
			start = failed_start;
1343
		} else
1344 1345 1346 1347 1348 1349
			break;
		WARN_ON(start > end);
	}
	return err;
}

1350
int try_lock_extent(struct extent_io_tree *tree, u64 start, u64 end)
1351 1352 1353 1354
{
	int err;
	u64 failed_start;

J
Jeff Mahoney 已提交
1355
	err = __set_extent_bit(tree, start, end, EXTENT_LOCKED, EXTENT_LOCKED,
1356
			       &failed_start, NULL, GFP_NOFS, NULL);
Y
Yan Zheng 已提交
1357 1358 1359
	if (err == -EEXIST) {
		if (failed_start > start)
			clear_extent_bit(tree, start, failed_start - 1,
1360
					 EXTENT_LOCKED, 1, 0, NULL, GFP_NOFS);
1361
		return 0;
Y
Yan Zheng 已提交
1362
	}
1363 1364 1365
	return 1;
}

1366
void extent_range_clear_dirty_for_io(struct inode *inode, u64 start, u64 end)
1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380
{
	unsigned long index = start >> PAGE_CACHE_SHIFT;
	unsigned long end_index = end >> PAGE_CACHE_SHIFT;
	struct page *page;

	while (index <= end_index) {
		page = find_get_page(inode->i_mapping, index);
		BUG_ON(!page); /* Pages should be in the extent_io_tree */
		clear_page_dirty_for_io(page);
		page_cache_release(page);
		index++;
	}
}

1381
void extent_range_redirty_for_io(struct inode *inode, u64 start, u64 end)
1382 1383 1384 1385 1386 1387 1388 1389 1390
{
	unsigned long index = start >> PAGE_CACHE_SHIFT;
	unsigned long end_index = end >> PAGE_CACHE_SHIFT;
	struct page *page;

	while (index <= end_index) {
		page = find_get_page(inode->i_mapping, index);
		BUG_ON(!page); /* Pages should be in the extent_io_tree */
		__set_page_dirty_nobuffers(page);
1391
		account_page_redirty(page);
1392 1393 1394 1395 1396
		page_cache_release(page);
		index++;
	}
}

1397 1398 1399
/*
 * helper function to set both pages and extents in the tree writeback
 */
1400
static void set_range_writeback(struct extent_io_tree *tree, u64 start, u64 end)
1401 1402 1403 1404 1405 1406 1407
{
	unsigned long index = start >> PAGE_CACHE_SHIFT;
	unsigned long end_index = end >> PAGE_CACHE_SHIFT;
	struct page *page;

	while (index <= end_index) {
		page = find_get_page(tree->mapping, index);
1408
		BUG_ON(!page); /* Pages should be in the extent_io_tree */
1409 1410 1411 1412 1413 1414
		set_page_writeback(page);
		page_cache_release(page);
		index++;
	}
}

C
Chris Mason 已提交
1415 1416 1417 1418
/* find the first state struct with 'bits' set after 'start', and
 * return it.  tree->lock must be held.  NULL will returned if
 * nothing was found after 'start'
 */
1419 1420
static struct extent_state *
find_first_extent_bit_state(struct extent_io_tree *tree,
1421
			    u64 start, unsigned bits)
C
Chris Mason 已提交
1422 1423 1424 1425 1426 1427 1428 1429 1430
{
	struct rb_node *node;
	struct extent_state *state;

	/*
	 * this search will find all the extents that end after
	 * our range starts.
	 */
	node = tree_search(tree, start);
C
Chris Mason 已提交
1431
	if (!node)
C
Chris Mason 已提交
1432 1433
		goto out;

C
Chris Mason 已提交
1434
	while (1) {
C
Chris Mason 已提交
1435
		state = rb_entry(node, struct extent_state, rb_node);
C
Chris Mason 已提交
1436
		if (state->end >= start && (state->state & bits))
C
Chris Mason 已提交
1437
			return state;
C
Chris Mason 已提交
1438

C
Chris Mason 已提交
1439 1440 1441 1442 1443 1444 1445 1446
		node = rb_next(node);
		if (!node)
			break;
	}
out:
	return NULL;
}

1447 1448 1449 1450 1451
/*
 * find the first offset in the io tree with 'bits' set. zero is
 * returned if we find something, and *start_ret and *end_ret are
 * set to reflect the state struct that was found.
 *
1452
 * If nothing was found, 1 is returned. If found something, return 0.
1453 1454
 */
int find_first_extent_bit(struct extent_io_tree *tree, u64 start,
1455
			  u64 *start_ret, u64 *end_ret, unsigned bits,
1456
			  struct extent_state **cached_state)
1457 1458
{
	struct extent_state *state;
1459
	struct rb_node *n;
1460 1461 1462
	int ret = 1;

	spin_lock(&tree->lock);
1463 1464
	if (cached_state && *cached_state) {
		state = *cached_state;
1465
		if (state->end == start - 1 && extent_state_in_tree(state)) {
1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481
			n = rb_next(&state->rb_node);
			while (n) {
				state = rb_entry(n, struct extent_state,
						 rb_node);
				if (state->state & bits)
					goto got_it;
				n = rb_next(n);
			}
			free_extent_state(*cached_state);
			*cached_state = NULL;
			goto out;
		}
		free_extent_state(*cached_state);
		*cached_state = NULL;
	}

1482
	state = find_first_extent_bit_state(tree, start, bits);
1483
got_it:
1484
	if (state) {
1485
		cache_state_if_flags(state, cached_state, 0);
1486 1487 1488 1489
		*start_ret = state->start;
		*end_ret = state->end;
		ret = 0;
	}
1490
out:
1491 1492 1493 1494
	spin_unlock(&tree->lock);
	return ret;
}

C
Chris Mason 已提交
1495 1496 1497 1498 1499 1500
/*
 * find a contiguous range of bytes in the file marked as delalloc, not
 * more than 'max_bytes'.  start and end are used to return the range,
 *
 * 1 is returned if we find something, 0 if nothing was in the tree
 */
C
Chris Mason 已提交
1501
static noinline u64 find_delalloc_range(struct extent_io_tree *tree,
1502 1503
					u64 *start, u64 *end, u64 max_bytes,
					struct extent_state **cached_state)
1504 1505 1506 1507 1508 1509 1510
{
	struct rb_node *node;
	struct extent_state *state;
	u64 cur_start = *start;
	u64 found = 0;
	u64 total_bytes = 0;

1511
	spin_lock(&tree->lock);
C
Chris Mason 已提交
1512

1513 1514 1515 1516
	/*
	 * this search will find all the extents that end after
	 * our range starts.
	 */
1517
	node = tree_search(tree, cur_start);
1518
	if (!node) {
1519 1520
		if (!found)
			*end = (u64)-1;
1521 1522 1523
		goto out;
	}

C
Chris Mason 已提交
1524
	while (1) {
1525
		state = rb_entry(node, struct extent_state, rb_node);
1526 1527
		if (found && (state->start != cur_start ||
			      (state->state & EXTENT_BOUNDARY))) {
1528 1529 1530 1531 1532 1533 1534
			goto out;
		}
		if (!(state->state & EXTENT_DELALLOC)) {
			if (!found)
				*end = state->end;
			goto out;
		}
1535
		if (!found) {
1536
			*start = state->start;
1537 1538 1539
			*cached_state = state;
			atomic_inc(&state->refs);
		}
1540 1541 1542 1543 1544
		found++;
		*end = state->end;
		cur_start = state->end + 1;
		node = rb_next(node);
		total_bytes += state->end - state->start + 1;
1545
		if (total_bytes >= max_bytes)
1546 1547
			break;
		if (!node)
1548 1549 1550
			break;
	}
out:
1551
	spin_unlock(&tree->lock);
1552 1553 1554
	return found;
}

1555 1556 1557
static noinline void __unlock_for_delalloc(struct inode *inode,
					   struct page *locked_page,
					   u64 start, u64 end)
C
Chris Mason 已提交
1558 1559 1560 1561 1562 1563 1564 1565 1566
{
	int ret;
	struct page *pages[16];
	unsigned long index = start >> PAGE_CACHE_SHIFT;
	unsigned long end_index = end >> PAGE_CACHE_SHIFT;
	unsigned long nr_pages = end_index - index + 1;
	int i;

	if (index == locked_page->index && end_index == index)
1567
		return;
C
Chris Mason 已提交
1568

C
Chris Mason 已提交
1569
	while (nr_pages > 0) {
C
Chris Mason 已提交
1570
		ret = find_get_pages_contig(inode->i_mapping, index,
1571 1572
				     min_t(unsigned long, nr_pages,
				     ARRAY_SIZE(pages)), pages);
C
Chris Mason 已提交
1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603
		for (i = 0; i < ret; i++) {
			if (pages[i] != locked_page)
				unlock_page(pages[i]);
			page_cache_release(pages[i]);
		}
		nr_pages -= ret;
		index += ret;
		cond_resched();
	}
}

static noinline int lock_delalloc_pages(struct inode *inode,
					struct page *locked_page,
					u64 delalloc_start,
					u64 delalloc_end)
{
	unsigned long index = delalloc_start >> PAGE_CACHE_SHIFT;
	unsigned long start_index = index;
	unsigned long end_index = delalloc_end >> PAGE_CACHE_SHIFT;
	unsigned long pages_locked = 0;
	struct page *pages[16];
	unsigned long nrpages;
	int ret;
	int i;

	/* the caller is responsible for locking the start index */
	if (index == locked_page->index && index == end_index)
		return 0;

	/* skip the page at the start index */
	nrpages = end_index - index + 1;
C
Chris Mason 已提交
1604
	while (nrpages > 0) {
C
Chris Mason 已提交
1605
		ret = find_get_pages_contig(inode->i_mapping, index,
1606 1607
				     min_t(unsigned long,
				     nrpages, ARRAY_SIZE(pages)), pages);
C
Chris Mason 已提交
1608 1609 1610 1611 1612 1613 1614 1615 1616 1617
		if (ret == 0) {
			ret = -EAGAIN;
			goto done;
		}
		/* now we have an array of pages, lock them all */
		for (i = 0; i < ret; i++) {
			/*
			 * the caller is taking responsibility for
			 * locked_page
			 */
1618
			if (pages[i] != locked_page) {
C
Chris Mason 已提交
1619
				lock_page(pages[i]);
1620 1621
				if (!PageDirty(pages[i]) ||
				    pages[i]->mapping != inode->i_mapping) {
1622 1623 1624 1625 1626 1627
					ret = -EAGAIN;
					unlock_page(pages[i]);
					page_cache_release(pages[i]);
					goto done;
				}
			}
C
Chris Mason 已提交
1628
			page_cache_release(pages[i]);
1629
			pages_locked++;
C
Chris Mason 已提交
1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651
		}
		nrpages -= ret;
		index += ret;
		cond_resched();
	}
	ret = 0;
done:
	if (ret && pages_locked) {
		__unlock_for_delalloc(inode, locked_page,
			      delalloc_start,
			      ((u64)(start_index + pages_locked - 1)) <<
			      PAGE_CACHE_SHIFT);
	}
	return ret;
}

/*
 * find a contiguous range of bytes in the file marked as delalloc, not
 * more than 'max_bytes'.  start and end are used to return the range,
 *
 * 1 is returned if we find something, 0 if nothing was in the tree
 */
1652 1653 1654 1655
STATIC u64 find_lock_delalloc_range(struct inode *inode,
				    struct extent_io_tree *tree,
				    struct page *locked_page, u64 *start,
				    u64 *end, u64 max_bytes)
C
Chris Mason 已提交
1656 1657 1658 1659
{
	u64 delalloc_start;
	u64 delalloc_end;
	u64 found;
1660
	struct extent_state *cached_state = NULL;
C
Chris Mason 已提交
1661 1662 1663 1664 1665 1666 1667 1668
	int ret;
	int loops = 0;

again:
	/* step one, find a bunch of delalloc bytes starting at start */
	delalloc_start = *start;
	delalloc_end = 0;
	found = find_delalloc_range(tree, &delalloc_start, &delalloc_end,
1669
				    max_bytes, &cached_state);
C
Chris Mason 已提交
1670
	if (!found || delalloc_end <= *start) {
C
Chris Mason 已提交
1671 1672
		*start = delalloc_start;
		*end = delalloc_end;
1673
		free_extent_state(cached_state);
L
Liu Bo 已提交
1674
		return 0;
C
Chris Mason 已提交
1675 1676
	}

C
Chris Mason 已提交
1677 1678 1679 1680 1681
	/*
	 * start comes from the offset of locked_page.  We have to lock
	 * pages in order, so we can't process delalloc bytes before
	 * locked_page
	 */
C
Chris Mason 已提交
1682
	if (delalloc_start < *start)
C
Chris Mason 已提交
1683 1684
		delalloc_start = *start;

C
Chris Mason 已提交
1685 1686 1687
	/*
	 * make sure to limit the number of pages we try to lock down
	 */
1688 1689
	if (delalloc_end + 1 - delalloc_start > max_bytes)
		delalloc_end = delalloc_start + max_bytes - 1;
C
Chris Mason 已提交
1690

C
Chris Mason 已提交
1691 1692 1693 1694 1695 1696 1697
	/* step two, lock all the pages after the page that has start */
	ret = lock_delalloc_pages(inode, locked_page,
				  delalloc_start, delalloc_end);
	if (ret == -EAGAIN) {
		/* some of the pages are gone, lets avoid looping by
		 * shortening the size of the delalloc range we're searching
		 */
1698
		free_extent_state(cached_state);
1699
		cached_state = NULL;
C
Chris Mason 已提交
1700
		if (!loops) {
1701
			max_bytes = PAGE_CACHE_SIZE;
C
Chris Mason 已提交
1702 1703 1704 1705 1706 1707 1708
			loops = 1;
			goto again;
		} else {
			found = 0;
			goto out_failed;
		}
	}
1709
	BUG_ON(ret); /* Only valid values are 0 and -EAGAIN */
C
Chris Mason 已提交
1710 1711

	/* step three, lock the state bits for the whole range */
1712
	lock_extent_bits(tree, delalloc_start, delalloc_end, &cached_state);
C
Chris Mason 已提交
1713 1714 1715

	/* then test to make sure it is all still delalloc */
	ret = test_range_bit(tree, delalloc_start, delalloc_end,
1716
			     EXTENT_DELALLOC, 1, cached_state);
C
Chris Mason 已提交
1717
	if (!ret) {
1718 1719
		unlock_extent_cached(tree, delalloc_start, delalloc_end,
				     &cached_state, GFP_NOFS);
C
Chris Mason 已提交
1720 1721 1722 1723 1724
		__unlock_for_delalloc(inode, locked_page,
			      delalloc_start, delalloc_end);
		cond_resched();
		goto again;
	}
1725
	free_extent_state(cached_state);
C
Chris Mason 已提交
1726 1727 1728 1729 1730 1731
	*start = delalloc_start;
	*end = delalloc_end;
out_failed:
	return found;
}

1732
void extent_clear_unlock_delalloc(struct inode *inode, u64 start, u64 end,
1733
				 struct page *locked_page,
1734
				 unsigned clear_bits,
1735
				 unsigned long page_ops)
C
Chris Mason 已提交
1736
{
1737
	struct extent_io_tree *tree = &BTRFS_I(inode)->io_tree;
C
Chris Mason 已提交
1738 1739 1740 1741 1742 1743
	int ret;
	struct page *pages[16];
	unsigned long index = start >> PAGE_CACHE_SHIFT;
	unsigned long end_index = end >> PAGE_CACHE_SHIFT;
	unsigned long nr_pages = end_index - index + 1;
	int i;
1744

1745
	clear_extent_bit(tree, start, end, clear_bits, 1, 0, NULL, GFP_NOFS);
1746
	if (page_ops == 0)
1747
		return;
C
Chris Mason 已提交
1748

1749 1750 1751
	if ((page_ops & PAGE_SET_ERROR) && nr_pages > 0)
		mapping_set_error(inode->i_mapping, -EIO);

C
Chris Mason 已提交
1752
	while (nr_pages > 0) {
C
Chris Mason 已提交
1753
		ret = find_get_pages_contig(inode->i_mapping, index,
1754 1755
				     min_t(unsigned long,
				     nr_pages, ARRAY_SIZE(pages)), pages);
C
Chris Mason 已提交
1756
		for (i = 0; i < ret; i++) {
1757

1758
			if (page_ops & PAGE_SET_PRIVATE2)
1759 1760
				SetPagePrivate2(pages[i]);

C
Chris Mason 已提交
1761 1762 1763 1764
			if (pages[i] == locked_page) {
				page_cache_release(pages[i]);
				continue;
			}
1765
			if (page_ops & PAGE_CLEAR_DIRTY)
C
Chris Mason 已提交
1766
				clear_page_dirty_for_io(pages[i]);
1767
			if (page_ops & PAGE_SET_WRITEBACK)
C
Chris Mason 已提交
1768
				set_page_writeback(pages[i]);
1769 1770
			if (page_ops & PAGE_SET_ERROR)
				SetPageError(pages[i]);
1771
			if (page_ops & PAGE_END_WRITEBACK)
C
Chris Mason 已提交
1772
				end_page_writeback(pages[i]);
1773
			if (page_ops & PAGE_UNLOCK)
1774
				unlock_page(pages[i]);
C
Chris Mason 已提交
1775 1776 1777 1778 1779 1780 1781 1782
			page_cache_release(pages[i]);
		}
		nr_pages -= ret;
		index += ret;
		cond_resched();
	}
}

C
Chris Mason 已提交
1783 1784 1785 1786 1787
/*
 * count the number of bytes in the tree that have a given bit(s)
 * set.  This can be fairly slow, except for EXTENT_DIRTY which is
 * cached.  The total number found is returned.
 */
1788 1789
u64 count_range_bits(struct extent_io_tree *tree,
		     u64 *start, u64 search_end, u64 max_bytes,
1790
		     unsigned bits, int contig)
1791 1792 1793 1794 1795
{
	struct rb_node *node;
	struct extent_state *state;
	u64 cur_start = *start;
	u64 total_bytes = 0;
1796
	u64 last = 0;
1797 1798
	int found = 0;

1799
	if (WARN_ON(search_end <= cur_start))
1800 1801
		return 0;

1802
	spin_lock(&tree->lock);
1803 1804 1805 1806 1807 1808 1809 1810
	if (cur_start == 0 && bits == EXTENT_DIRTY) {
		total_bytes = tree->dirty_bytes;
		goto out;
	}
	/*
	 * this search will find all the extents that end after
	 * our range starts.
	 */
1811
	node = tree_search(tree, cur_start);
C
Chris Mason 已提交
1812
	if (!node)
1813 1814
		goto out;

C
Chris Mason 已提交
1815
	while (1) {
1816 1817 1818
		state = rb_entry(node, struct extent_state, rb_node);
		if (state->start > search_end)
			break;
1819 1820 1821
		if (contig && found && state->start > last + 1)
			break;
		if (state->end >= cur_start && (state->state & bits) == bits) {
1822 1823 1824 1825 1826
			total_bytes += min(search_end, state->end) + 1 -
				       max(cur_start, state->start);
			if (total_bytes >= max_bytes)
				break;
			if (!found) {
1827
				*start = max(cur_start, state->start);
1828 1829
				found = 1;
			}
1830 1831 1832
			last = state->end;
		} else if (contig && found) {
			break;
1833 1834 1835 1836 1837 1838
		}
		node = rb_next(node);
		if (!node)
			break;
	}
out:
1839
	spin_unlock(&tree->lock);
1840 1841
	return total_bytes;
}
1842

C
Chris Mason 已提交
1843 1844 1845 1846
/*
 * set the private field for a given byte offset in the tree.  If there isn't
 * an extent_state there already, this does nothing.
 */
1847 1848
static int set_state_failrec(struct extent_io_tree *tree, u64 start,
		struct io_failure_record *failrec)
1849 1850 1851 1852 1853
{
	struct rb_node *node;
	struct extent_state *state;
	int ret = 0;

1854
	spin_lock(&tree->lock);
1855 1856 1857 1858
	/*
	 * this search will find all the extents that end after
	 * our range starts.
	 */
1859
	node = tree_search(tree, start);
1860
	if (!node) {
1861 1862 1863 1864 1865 1866 1867 1868
		ret = -ENOENT;
		goto out;
	}
	state = rb_entry(node, struct extent_state, rb_node);
	if (state->start != start) {
		ret = -ENOENT;
		goto out;
	}
1869
	state->failrec = failrec;
1870
out:
1871
	spin_unlock(&tree->lock);
1872 1873 1874
	return ret;
}

1875 1876
static int get_state_failrec(struct extent_io_tree *tree, u64 start,
		struct io_failure_record **failrec)
1877 1878 1879 1880 1881
{
	struct rb_node *node;
	struct extent_state *state;
	int ret = 0;

1882
	spin_lock(&tree->lock);
1883 1884 1885 1886
	/*
	 * this search will find all the extents that end after
	 * our range starts.
	 */
1887
	node = tree_search(tree, start);
1888
	if (!node) {
1889 1890 1891 1892 1893 1894 1895 1896
		ret = -ENOENT;
		goto out;
	}
	state = rb_entry(node, struct extent_state, rb_node);
	if (state->start != start) {
		ret = -ENOENT;
		goto out;
	}
1897
	*failrec = state->failrec;
1898
out:
1899
	spin_unlock(&tree->lock);
1900 1901 1902 1903 1904
	return ret;
}

/*
 * searches a range in the state tree for a given mask.
1905
 * If 'filled' == 1, this returns 1 only if every extent in the tree
1906 1907 1908 1909
 * has the bits set.  Otherwise, 1 is returned if any bit in the
 * range is found set.
 */
int test_range_bit(struct extent_io_tree *tree, u64 start, u64 end,
1910
		   unsigned bits, int filled, struct extent_state *cached)
1911 1912 1913 1914 1915
{
	struct extent_state *state = NULL;
	struct rb_node *node;
	int bitset = 0;

1916
	spin_lock(&tree->lock);
1917
	if (cached && extent_state_in_tree(cached) && cached->start <= start &&
1918
	    cached->end > start)
1919 1920 1921
		node = &cached->rb_node;
	else
		node = tree_search(tree, start);
1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940
	while (node && start <= end) {
		state = rb_entry(node, struct extent_state, rb_node);

		if (filled && state->start > start) {
			bitset = 0;
			break;
		}

		if (state->start > end)
			break;

		if (state->state & bits) {
			bitset = 1;
			if (!filled)
				break;
		} else if (filled) {
			bitset = 0;
			break;
		}
1941 1942 1943 1944

		if (state->end == (u64)-1)
			break;

1945 1946 1947 1948 1949 1950 1951 1952 1953 1954
		start = state->end + 1;
		if (start > end)
			break;
		node = rb_next(node);
		if (!node) {
			if (filled)
				bitset = 0;
			break;
		}
	}
1955
	spin_unlock(&tree->lock);
1956 1957 1958 1959 1960 1961 1962
	return bitset;
}

/*
 * helper function to set a given page up to date if all the
 * extents in the tree for that page are up to date
 */
1963
static void check_page_uptodate(struct extent_io_tree *tree, struct page *page)
1964
{
M
Miao Xie 已提交
1965
	u64 start = page_offset(page);
1966
	u64 end = start + PAGE_CACHE_SIZE - 1;
1967
	if (test_range_bit(tree, start, end, EXTENT_UPTODATE, 1, NULL))
1968 1969 1970
		SetPageUptodate(page);
}

1971
int free_io_failure(struct inode *inode, struct io_failure_record *rec)
1972 1973 1974 1975 1976
{
	int ret;
	int err = 0;
	struct extent_io_tree *failure_tree = &BTRFS_I(inode)->io_failure_tree;

1977
	set_state_failrec(failure_tree, rec->start, NULL);
1978 1979 1980 1981 1982 1983
	ret = clear_extent_bits(failure_tree, rec->start,
				rec->start + rec->len - 1,
				EXTENT_LOCKED | EXTENT_DIRTY, GFP_NOFS);
	if (ret)
		err = ret;

D
David Woodhouse 已提交
1984 1985 1986 1987 1988
	ret = clear_extent_bits(&BTRFS_I(inode)->io_tree, rec->start,
				rec->start + rec->len - 1,
				EXTENT_DAMAGED, GFP_NOFS);
	if (ret && !err)
		err = ret;
1989 1990 1991 1992 1993 1994 1995 1996 1997 1998

	kfree(rec);
	return err;
}

/*
 * this bypasses the standard btrfs submit functions deliberately, as
 * the standard behavior is to write all copies in a raid setup. here we only
 * want to write the one bad copy. so we do the mapping for ourselves and issue
 * submit_bio directly.
1999
 * to avoid any synchronization issues, wait for the data after writing, which
2000 2001 2002 2003
 * actually prevents the read that triggered the error from finishing.
 * currently, there can be no more than two copies of every data bit. thus,
 * exactly one rewrite is required.
 */
2004 2005
int repair_io_failure(struct inode *inode, u64 start, u64 length, u64 logical,
		      struct page *page, unsigned int pg_offset, int mirror_num)
2006
{
2007
	struct btrfs_fs_info *fs_info = BTRFS_I(inode)->root->fs_info;
2008 2009 2010 2011 2012
	struct bio *bio;
	struct btrfs_device *dev;
	u64 map_length = 0;
	u64 sector;
	struct btrfs_bio *bbio = NULL;
D
David Woodhouse 已提交
2013
	struct btrfs_mapping_tree *map_tree = &fs_info->mapping_tree;
2014 2015
	int ret;

2016
	ASSERT(!(fs_info->sb->s_flags & MS_RDONLY));
2017 2018
	BUG_ON(!mirror_num);

D
David Woodhouse 已提交
2019 2020 2021 2022
	/* we can't repair anything in raid56 yet */
	if (btrfs_is_parity_mirror(map_tree, logical, length, mirror_num))
		return 0;

2023
	bio = btrfs_io_bio_alloc(GFP_NOFS, 1);
2024 2025
	if (!bio)
		return -EIO;
2026
	bio->bi_iter.bi_size = 0;
2027 2028
	map_length = length;

2029
	ret = btrfs_map_block(fs_info, WRITE, logical,
2030 2031 2032 2033 2034 2035 2036
			      &map_length, &bbio, mirror_num);
	if (ret) {
		bio_put(bio);
		return -EIO;
	}
	BUG_ON(mirror_num != bbio->mirror_num);
	sector = bbio->stripes[mirror_num-1].physical >> 9;
2037
	bio->bi_iter.bi_sector = sector;
2038
	dev = bbio->stripes[mirror_num-1].dev;
2039
	btrfs_put_bbio(bbio);
2040 2041 2042 2043 2044
	if (!dev || !dev->bdev || !dev->writeable) {
		bio_put(bio);
		return -EIO;
	}
	bio->bi_bdev = dev->bdev;
2045
	bio_add_page(bio, page, length, pg_offset);
2046

2047
	if (btrfsic_submit_bio_wait(WRITE_SYNC, bio)) {
2048 2049
		/* try to remap that extent elsewhere? */
		bio_put(bio);
2050
		btrfs_dev_stat_inc_and_print(dev, BTRFS_DEV_STAT_WRITE_ERRS);
2051 2052 2053
		return -EIO;
	}

2054 2055
	btrfs_info_rl_in_rcu(fs_info,
		"read error corrected: ino %llu off %llu (dev %s sector %llu)",
2056 2057
				  btrfs_ino(inode), start,
				  rcu_str_deref(dev->name), sector);
2058 2059 2060 2061
	bio_put(bio);
	return 0;
}

2062 2063 2064 2065 2066
int repair_eb_io_failure(struct btrfs_root *root, struct extent_buffer *eb,
			 int mirror_num)
{
	u64 start = eb->start;
	unsigned long i, num_pages = num_extent_pages(eb->start, eb->len);
2067
	int ret = 0;
2068

2069 2070 2071
	if (root->fs_info->sb->s_flags & MS_RDONLY)
		return -EROFS;

2072
	for (i = 0; i < num_pages; i++) {
2073
		struct page *p = eb->pages[i];
2074 2075 2076 2077

		ret = repair_io_failure(root->fs_info->btree_inode, start,
					PAGE_CACHE_SIZE, start, p,
					start - page_offset(p), mirror_num);
2078 2079 2080 2081 2082 2083 2084 2085
		if (ret)
			break;
		start += PAGE_CACHE_SIZE;
	}

	return ret;
}

2086 2087 2088 2089
/*
 * each time an IO finishes, we do a fast check in the IO failure tree
 * to see if we need to process or clean up an io_failure_record
 */
2090 2091
int clean_io_failure(struct inode *inode, u64 start, struct page *page,
		     unsigned int pg_offset)
2092 2093 2094
{
	u64 private;
	struct io_failure_record *failrec;
2095
	struct btrfs_fs_info *fs_info = BTRFS_I(inode)->root->fs_info;
2096 2097 2098 2099 2100 2101 2102 2103 2104 2105
	struct extent_state *state;
	int num_copies;
	int ret;

	private = 0;
	ret = count_range_bits(&BTRFS_I(inode)->io_failure_tree, &private,
				(u64)-1, 1, EXTENT_DIRTY, 0);
	if (!ret)
		return 0;

2106 2107
	ret = get_state_failrec(&BTRFS_I(inode)->io_failure_tree, start,
			&failrec);
2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118
	if (ret)
		return 0;

	BUG_ON(!failrec->this_mirror);

	if (failrec->in_validation) {
		/* there was no real error, just free the record */
		pr_debug("clean_io_failure: freeing dummy error at %llu\n",
			 failrec->start);
		goto out;
	}
2119 2120
	if (fs_info->sb->s_flags & MS_RDONLY)
		goto out;
2121 2122 2123 2124 2125 2126 2127

	spin_lock(&BTRFS_I(inode)->io_tree.lock);
	state = find_first_extent_bit_state(&BTRFS_I(inode)->io_tree,
					    failrec->start,
					    EXTENT_LOCKED);
	spin_unlock(&BTRFS_I(inode)->io_tree.lock);

2128 2129
	if (state && state->start <= failrec->start &&
	    state->end >= failrec->start + failrec->len - 1) {
2130 2131
		num_copies = btrfs_num_copies(fs_info, failrec->logical,
					      failrec->len);
2132
		if (num_copies > 1)  {
2133
			repair_io_failure(inode, start, failrec->len,
2134
					  failrec->logical, page,
2135
					  pg_offset, failrec->failed_mirror);
2136 2137 2138 2139
		}
	}

out:
2140
	free_io_failure(inode, failrec);
2141

2142
	return 0;
2143 2144
}

2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169
/*
 * Can be called when
 * - hold extent lock
 * - under ordered extent
 * - the inode is freeing
 */
void btrfs_free_io_failure_record(struct inode *inode, u64 start, u64 end)
{
	struct extent_io_tree *failure_tree = &BTRFS_I(inode)->io_failure_tree;
	struct io_failure_record *failrec;
	struct extent_state *state, *next;

	if (RB_EMPTY_ROOT(&failure_tree->state))
		return;

	spin_lock(&failure_tree->lock);
	state = find_first_extent_bit_state(failure_tree, start, EXTENT_DIRTY);
	while (state) {
		if (state->start > end)
			break;

		ASSERT(state->end <= end);

		next = next_state(state);

2170
		failrec = state->failrec;
2171 2172 2173 2174 2175 2176 2177 2178
		free_extent_state(state);
		kfree(failrec);

		state = next;
	}
	spin_unlock(&failure_tree->lock);
}

2179
int btrfs_get_io_failure_record(struct inode *inode, u64 start, u64 end,
2180
		struct io_failure_record **failrec_ret)
2181
{
2182
	struct io_failure_record *failrec;
2183 2184 2185 2186 2187 2188 2189
	struct extent_map *em;
	struct extent_io_tree *failure_tree = &BTRFS_I(inode)->io_failure_tree;
	struct extent_io_tree *tree = &BTRFS_I(inode)->io_tree;
	struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
	int ret;
	u64 logical;

2190
	ret = get_state_failrec(failure_tree, start, &failrec);
2191 2192 2193 2194
	if (ret) {
		failrec = kzalloc(sizeof(*failrec), GFP_NOFS);
		if (!failrec)
			return -ENOMEM;
2195

2196 2197 2198 2199 2200 2201 2202 2203 2204 2205 2206 2207 2208 2209
		failrec->start = start;
		failrec->len = end - start + 1;
		failrec->this_mirror = 0;
		failrec->bio_flags = 0;
		failrec->in_validation = 0;

		read_lock(&em_tree->lock);
		em = lookup_extent_mapping(em_tree, start, failrec->len);
		if (!em) {
			read_unlock(&em_tree->lock);
			kfree(failrec);
			return -EIO;
		}

2210
		if (em->start > start || em->start + em->len <= start) {
2211 2212 2213 2214
			free_extent_map(em);
			em = NULL;
		}
		read_unlock(&em_tree->lock);
2215
		if (!em) {
2216 2217 2218
			kfree(failrec);
			return -EIO;
		}
2219

2220 2221 2222 2223 2224 2225 2226 2227
		logical = start - em->start;
		logical = em->block_start + logical;
		if (test_bit(EXTENT_FLAG_COMPRESSED, &em->flags)) {
			logical = em->block_start;
			failrec->bio_flags = EXTENT_BIO_COMPRESSED;
			extent_set_compress_type(&failrec->bio_flags,
						 em->compress_type);
		}
2228 2229 2230 2231

		pr_debug("Get IO Failure Record: (new) logical=%llu, start=%llu, len=%llu\n",
			 logical, start, failrec->len);

2232 2233 2234 2235 2236 2237 2238
		failrec->logical = logical;
		free_extent_map(em);

		/* set the bits in the private failure tree */
		ret = set_extent_bits(failure_tree, start, end,
					EXTENT_LOCKED | EXTENT_DIRTY, GFP_NOFS);
		if (ret >= 0)
2239
			ret = set_state_failrec(failure_tree, start, failrec);
2240 2241 2242 2243 2244 2245 2246 2247 2248
		/* set the bits in the inode's tree */
		if (ret >= 0)
			ret = set_extent_bits(tree, start, end, EXTENT_DAMAGED,
						GFP_NOFS);
		if (ret < 0) {
			kfree(failrec);
			return ret;
		}
	} else {
2249
		pr_debug("Get IO Failure Record: (found) logical=%llu, start=%llu, len=%llu, validation=%d\n",
2250 2251 2252 2253 2254 2255 2256 2257
			 failrec->logical, failrec->start, failrec->len,
			 failrec->in_validation);
		/*
		 * when data can be on disk more than twice, add to failrec here
		 * (e.g. with a list for failed_mirror) to make
		 * clean_io_failure() clean all those errors at once.
		 */
	}
2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268

	*failrec_ret = failrec;

	return 0;
}

int btrfs_check_repairable(struct inode *inode, struct bio *failed_bio,
			   struct io_failure_record *failrec, int failed_mirror)
{
	int num_copies;

2269 2270
	num_copies = btrfs_num_copies(BTRFS_I(inode)->root->fs_info,
				      failrec->logical, failrec->len);
2271 2272 2273 2274 2275 2276
	if (num_copies == 1) {
		/*
		 * we only have a single copy of the data, so don't bother with
		 * all the retry and error correction code that follows. no
		 * matter what the error is, it is very likely to persist.
		 */
2277
		pr_debug("Check Repairable: cannot repair, num_copies=%d, next_mirror %d, failed_mirror %d\n",
2278
			 num_copies, failrec->this_mirror, failed_mirror);
2279
		return 0;
2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315
	}

	/*
	 * there are two premises:
	 *	a) deliver good data to the caller
	 *	b) correct the bad sectors on disk
	 */
	if (failed_bio->bi_vcnt > 1) {
		/*
		 * to fulfill b), we need to know the exact failing sectors, as
		 * we don't want to rewrite any more than the failed ones. thus,
		 * we need separate read requests for the failed bio
		 *
		 * if the following BUG_ON triggers, our validation request got
		 * merged. we need separate requests for our algorithm to work.
		 */
		BUG_ON(failrec->in_validation);
		failrec->in_validation = 1;
		failrec->this_mirror = failed_mirror;
	} else {
		/*
		 * we're ready to fulfill a) and b) alongside. get a good copy
		 * of the failed sector and if we succeed, we have setup
		 * everything for repair_io_failure to do the rest for us.
		 */
		if (failrec->in_validation) {
			BUG_ON(failrec->this_mirror != failed_mirror);
			failrec->in_validation = 0;
			failrec->this_mirror = 0;
		}
		failrec->failed_mirror = failed_mirror;
		failrec->this_mirror++;
		if (failrec->this_mirror == failed_mirror)
			failrec->this_mirror++;
	}

2316
	if (failrec->this_mirror > num_copies) {
2317
		pr_debug("Check Repairable: (fail) num_copies=%d, next_mirror %d, failed_mirror %d\n",
2318
			 num_copies, failrec->this_mirror, failed_mirror);
2319
		return 0;
2320 2321
	}

2322 2323 2324 2325 2326 2327 2328
	return 1;
}


struct bio *btrfs_create_repair_bio(struct inode *inode, struct bio *failed_bio,
				    struct io_failure_record *failrec,
				    struct page *page, int pg_offset, int icsum,
2329
				    bio_end_io_t *endio_func, void *data)
2330 2331 2332 2333 2334
{
	struct bio *bio;
	struct btrfs_io_bio *btrfs_failed_bio;
	struct btrfs_io_bio *btrfs_bio;

2335
	bio = btrfs_io_bio_alloc(GFP_NOFS, 1);
2336 2337 2338 2339
	if (!bio)
		return NULL;

	bio->bi_end_io = endio_func;
2340
	bio->bi_iter.bi_sector = failrec->logical >> 9;
2341
	bio->bi_bdev = BTRFS_I(inode)->root->fs_info->fs_devices->latest_bdev;
2342
	bio->bi_iter.bi_size = 0;
2343
	bio->bi_private = data;
2344

2345 2346 2347 2348 2349 2350 2351
	btrfs_failed_bio = btrfs_io_bio(failed_bio);
	if (btrfs_failed_bio->csum) {
		struct btrfs_fs_info *fs_info = BTRFS_I(inode)->root->fs_info;
		u16 csum_size = btrfs_super_csum_size(fs_info->super_copy);

		btrfs_bio = btrfs_io_bio(bio);
		btrfs_bio->csum = btrfs_bio->csum_inline;
2352 2353
		icsum *= csum_size;
		memcpy(btrfs_bio->csum, btrfs_failed_bio->csum + icsum,
2354 2355 2356
		       csum_size);
	}

2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400
	bio_add_page(bio, page, failrec->len, pg_offset);

	return bio;
}

/*
 * this is a generic handler for readpage errors (default
 * readpage_io_failed_hook). if other copies exist, read those and write back
 * good data to the failed position. does not investigate in remapping the
 * failed extent elsewhere, hoping the device will be smart enough to do this as
 * needed
 */

static int bio_readpage_error(struct bio *failed_bio, u64 phy_offset,
			      struct page *page, u64 start, u64 end,
			      int failed_mirror)
{
	struct io_failure_record *failrec;
	struct inode *inode = page->mapping->host;
	struct extent_io_tree *tree = &BTRFS_I(inode)->io_tree;
	struct bio *bio;
	int read_mode;
	int ret;

	BUG_ON(failed_bio->bi_rw & REQ_WRITE);

	ret = btrfs_get_io_failure_record(inode, start, end, &failrec);
	if (ret)
		return ret;

	ret = btrfs_check_repairable(inode, failed_bio, failrec, failed_mirror);
	if (!ret) {
		free_io_failure(inode, failrec);
		return -EIO;
	}

	if (failed_bio->bi_vcnt > 1)
		read_mode = READ_SYNC | REQ_FAILFAST_DEV;
	else
		read_mode = READ_SYNC;

	phy_offset >>= inode->i_sb->s_blocksize_bits;
	bio = btrfs_create_repair_bio(inode, failed_bio, failrec, page,
				      start - page_offset(page),
2401 2402
				      (int)phy_offset, failed_bio->bi_end_io,
				      NULL);
2403 2404 2405 2406
	if (!bio) {
		free_io_failure(inode, failrec);
		return -EIO;
	}
2407

2408 2409
	pr_debug("Repair Read Error: submitting new read[%#x] to this_mirror=%d, in_validation=%d\n",
		 read_mode, failrec->this_mirror, failrec->in_validation);
2410

2411 2412 2413
	ret = tree->ops->submit_bio_hook(inode, read_mode, bio,
					 failrec->this_mirror,
					 failrec->bio_flags, 0);
2414
	if (ret) {
2415
		free_io_failure(inode, failrec);
2416 2417 2418
		bio_put(bio);
	}

2419
	return ret;
2420 2421
}

2422 2423
/* lots and lots of room for performance fixes in the end_bio funcs */

2424
void end_extent_writepage(struct page *page, int err, u64 start, u64 end)
2425 2426 2427
{
	int uptodate = (err == 0);
	struct extent_io_tree *tree;
2428
	int ret = 0;
2429 2430 2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 2441

	tree = &BTRFS_I(page->mapping->host)->io_tree;

	if (tree->ops && tree->ops->writepage_end_io_hook) {
		ret = tree->ops->writepage_end_io_hook(page, start,
					       end, NULL, uptodate);
		if (ret)
			uptodate = 0;
	}

	if (!uptodate) {
		ClearPageUptodate(page);
		SetPageError(page);
2442 2443
		ret = ret < 0 ? ret : -EIO;
		mapping_set_error(page->mapping, ret);
2444 2445 2446
	}
}

2447 2448 2449 2450 2451 2452 2453 2454 2455
/*
 * after a writepage IO is done, we need to:
 * clear the uptodate bits on error
 * clear the writeback bits in the extent tree for this IO
 * end_page_writeback if the page has no more pending IO
 *
 * Scheduling is not allowed, so the extent state tree is expected
 * to have one and only one object corresponding to this IO.
 */
2456
static void end_bio_extent_writepage(struct bio *bio)
2457
{
2458
	struct bio_vec *bvec;
2459 2460
	u64 start;
	u64 end;
2461
	int i;
2462

2463
	bio_for_each_segment_all(bvec, bio, i) {
2464
		struct page *page = bvec->bv_page;
2465

2466 2467 2468 2469 2470
		/* We always issue full-page reads, but if some block
		 * in a page fails to read, blk_update_request() will
		 * advance bv_offset and adjust bv_len to compensate.
		 * Print a warning for nonzero offsets, and an error
		 * if they don't add up to a full page.  */
2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481
		if (bvec->bv_offset || bvec->bv_len != PAGE_CACHE_SIZE) {
			if (bvec->bv_offset + bvec->bv_len != PAGE_CACHE_SIZE)
				btrfs_err(BTRFS_I(page->mapping->host)->root->fs_info,
				   "partial page write in btrfs with offset %u and length %u",
					bvec->bv_offset, bvec->bv_len);
			else
				btrfs_info(BTRFS_I(page->mapping->host)->root->fs_info,
				   "incomplete page write in btrfs with offset %u and "
				   "length %u",
					bvec->bv_offset, bvec->bv_len);
		}
2482

2483 2484
		start = page_offset(page);
		end = start + bvec->bv_offset + bvec->bv_len - 1;
2485

2486
		end_extent_writepage(page, bio->bi_error, start, end);
2487
		end_page_writeback(page);
2488
	}
2489

2490 2491 2492
	bio_put(bio);
}

2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 2504
static void
endio_readpage_release_extent(struct extent_io_tree *tree, u64 start, u64 len,
			      int uptodate)
{
	struct extent_state *cached = NULL;
	u64 end = start + len - 1;

	if (uptodate && tree->track_uptodate)
		set_extent_uptodate(tree, start, end, &cached, GFP_ATOMIC);
	unlock_extent_cached(tree, start, end, &cached, GFP_ATOMIC);
}

2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515
/*
 * after a readpage IO is done, we need to:
 * clear the uptodate bits on error
 * set the uptodate bits if things worked
 * set the page up to date if all extents in the tree are uptodate
 * clear the lock bit in the extent tree
 * unlock the page if there are no other extents locked for it
 *
 * Scheduling is not allowed, so the extent state tree is expected
 * to have one and only one object corresponding to this IO.
 */
2516
static void end_bio_extent_readpage(struct bio *bio)
2517
{
2518
	struct bio_vec *bvec;
2519
	int uptodate = !bio->bi_error;
2520
	struct btrfs_io_bio *io_bio = btrfs_io_bio(bio);
2521
	struct extent_io_tree *tree;
2522
	u64 offset = 0;
2523 2524
	u64 start;
	u64 end;
2525
	u64 len;
2526 2527
	u64 extent_start = 0;
	u64 extent_len = 0;
2528
	int mirror;
2529
	int ret;
2530
	int i;
2531

2532
	bio_for_each_segment_all(bvec, bio, i) {
2533
		struct page *page = bvec->bv_page;
2534
		struct inode *inode = page->mapping->host;
2535

2536
		pr_debug("end_bio_extent_readpage: bi_sector=%llu, err=%d, "
2537 2538
			 "mirror=%u\n", (u64)bio->bi_iter.bi_sector,
			 bio->bi_error, io_bio->mirror_num);
2539
		tree = &BTRFS_I(inode)->io_tree;
2540

2541 2542 2543 2544 2545
		/* We always issue full-page reads, but if some block
		 * in a page fails to read, blk_update_request() will
		 * advance bv_offset and adjust bv_len to compensate.
		 * Print a warning for nonzero offsets, and an error
		 * if they don't add up to a full page.  */
2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556
		if (bvec->bv_offset || bvec->bv_len != PAGE_CACHE_SIZE) {
			if (bvec->bv_offset + bvec->bv_len != PAGE_CACHE_SIZE)
				btrfs_err(BTRFS_I(page->mapping->host)->root->fs_info,
				   "partial page read in btrfs with offset %u and length %u",
					bvec->bv_offset, bvec->bv_len);
			else
				btrfs_info(BTRFS_I(page->mapping->host)->root->fs_info,
				   "incomplete page read in btrfs with offset %u and "
				   "length %u",
					bvec->bv_offset, bvec->bv_len);
		}
2557

2558 2559
		start = page_offset(page);
		end = start + bvec->bv_offset + bvec->bv_len - 1;
2560
		len = bvec->bv_len;
2561

2562
		mirror = io_bio->mirror_num;
2563 2564
		if (likely(uptodate && tree->ops &&
			   tree->ops->readpage_end_io_hook)) {
2565 2566 2567
			ret = tree->ops->readpage_end_io_hook(io_bio, offset,
							      page, start, end,
							      mirror);
2568
			if (ret)
2569
				uptodate = 0;
2570
			else
2571
				clean_io_failure(inode, start, page, 0);
2572
		}
2573

2574 2575 2576 2577
		if (likely(uptodate))
			goto readpage_ok;

		if (tree->ops && tree->ops->readpage_io_failed_hook) {
2578
			ret = tree->ops->readpage_io_failed_hook(page, mirror);
2579
			if (!ret && !bio->bi_error)
2580
				uptodate = 1;
2581
		} else {
2582 2583 2584 2585 2586 2587 2588 2589 2590 2591
			/*
			 * The generic bio_readpage_error handles errors the
			 * following way: If possible, new read requests are
			 * created and submitted and will end up in
			 * end_bio_extent_readpage as well (if we're lucky, not
			 * in the !uptodate case). In that case it returns 0 and
			 * we just go on with the next page in our bio. If it
			 * can't handle the error it will return -EIO and we
			 * remain responsible for that page.
			 */
2592 2593
			ret = bio_readpage_error(bio, offset, page, start, end,
						 mirror);
2594
			if (ret == 0) {
2595
				uptodate = !bio->bi_error;
2596
				offset += len;
2597 2598 2599
				continue;
			}
		}
2600
readpage_ok:
2601
		if (likely(uptodate)) {
2602 2603
			loff_t i_size = i_size_read(inode);
			pgoff_t end_index = i_size >> PAGE_CACHE_SHIFT;
2604
			unsigned off;
2605 2606

			/* Zero out the end if this page straddles i_size */
2607 2608 2609
			off = i_size & (PAGE_CACHE_SIZE-1);
			if (page->index == end_index && off)
				zero_user_segment(page, off, PAGE_CACHE_SIZE);
2610
			SetPageUptodate(page);
2611
		} else {
2612 2613
			ClearPageUptodate(page);
			SetPageError(page);
2614
		}
2615
		unlock_page(page);
2616
		offset += len;
2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638

		if (unlikely(!uptodate)) {
			if (extent_len) {
				endio_readpage_release_extent(tree,
							      extent_start,
							      extent_len, 1);
				extent_start = 0;
				extent_len = 0;
			}
			endio_readpage_release_extent(tree, start,
						      end - start + 1, 0);
		} else if (!extent_len) {
			extent_start = start;
			extent_len = end + 1 - start;
		} else if (extent_start + extent_len == start) {
			extent_len += end + 1 - start;
		} else {
			endio_readpage_release_extent(tree, extent_start,
						      extent_len, uptodate);
			extent_start = start;
			extent_len = end + 1 - start;
		}
2639
	}
2640

2641 2642 2643
	if (extent_len)
		endio_readpage_release_extent(tree, extent_start, extent_len,
					      uptodate);
2644
	if (io_bio->end_io)
2645
		io_bio->end_io(io_bio, bio->bi_error);
2646 2647 2648
	bio_put(bio);
}

2649 2650 2651 2652
/*
 * this allocates from the btrfs_bioset.  We're returning a bio right now
 * but you can call btrfs_io_bio for the appropriate container_of magic
 */
2653 2654 2655
struct bio *
btrfs_bio_alloc(struct block_device *bdev, u64 first_sector, int nr_vecs,
		gfp_t gfp_flags)
2656
{
2657
	struct btrfs_io_bio *btrfs_bio;
2658 2659
	struct bio *bio;

2660
	bio = bio_alloc_bioset(gfp_flags, nr_vecs, btrfs_bioset);
2661 2662

	if (bio == NULL && (current->flags & PF_MEMALLOC)) {
2663 2664 2665 2666
		while (!bio && (nr_vecs /= 2)) {
			bio = bio_alloc_bioset(gfp_flags,
					       nr_vecs, btrfs_bioset);
		}
2667 2668 2669 2670
	}

	if (bio) {
		bio->bi_bdev = bdev;
2671
		bio->bi_iter.bi_sector = first_sector;
2672 2673 2674 2675
		btrfs_bio = btrfs_io_bio(bio);
		btrfs_bio->csum = NULL;
		btrfs_bio->csum_allocated = NULL;
		btrfs_bio->end_io = NULL;
2676 2677 2678 2679
	}
	return bio;
}

2680 2681
struct bio *btrfs_bio_clone(struct bio *bio, gfp_t gfp_mask)
{
2682 2683
	struct btrfs_io_bio *btrfs_bio;
	struct bio *new;
2684

2685 2686 2687 2688 2689 2690
	new = bio_clone_bioset(bio, gfp_mask, btrfs_bioset);
	if (new) {
		btrfs_bio = btrfs_io_bio(new);
		btrfs_bio->csum = NULL;
		btrfs_bio->csum_allocated = NULL;
		btrfs_bio->end_io = NULL;
2691 2692

#ifdef CONFIG_BLK_CGROUP
2693 2694 2695
		/* FIXME, put this into bio_clone_bioset */
		if (bio->bi_css)
			bio_associate_blkcg(new, bio->bi_css);
2696
#endif
2697 2698 2699
	}
	return new;
}
2700 2701 2702 2703

/* this also allocates from the btrfs_bioset */
struct bio *btrfs_io_bio_alloc(gfp_t gfp_mask, unsigned int nr_iovecs)
{
2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714
	struct btrfs_io_bio *btrfs_bio;
	struct bio *bio;

	bio = bio_alloc_bioset(gfp_mask, nr_iovecs, btrfs_bioset);
	if (bio) {
		btrfs_bio = btrfs_io_bio(bio);
		btrfs_bio->csum = NULL;
		btrfs_bio->csum_allocated = NULL;
		btrfs_bio->end_io = NULL;
	}
	return bio;
2715 2716 2717
}


2718 2719
static int __must_check submit_one_bio(int rw, struct bio *bio,
				       int mirror_num, unsigned long bio_flags)
2720 2721
{
	int ret = 0;
2722 2723 2724 2725 2726
	struct bio_vec *bvec = bio->bi_io_vec + bio->bi_vcnt - 1;
	struct page *page = bvec->bv_page;
	struct extent_io_tree *tree = bio->bi_private;
	u64 start;

M
Miao Xie 已提交
2727
	start = page_offset(page) + bvec->bv_offset;
2728

2729
	bio->bi_private = NULL;
2730 2731 2732

	bio_get(bio);

2733
	if (tree->ops && tree->ops->submit_bio_hook)
2734
		ret = tree->ops->submit_bio_hook(page->mapping->host, rw, bio,
2735
					   mirror_num, bio_flags, start);
2736
	else
2737
		btrfsic_submit_bio(rw, bio);
2738

2739 2740 2741 2742
	bio_put(bio);
	return ret;
}

2743
static int merge_bio(int rw, struct extent_io_tree *tree, struct page *page,
2744 2745 2746 2747 2748
		     unsigned long offset, size_t size, struct bio *bio,
		     unsigned long bio_flags)
{
	int ret = 0;
	if (tree->ops && tree->ops->merge_bio_hook)
2749
		ret = tree->ops->merge_bio_hook(rw, page, offset, size, bio,
2750 2751 2752 2753 2754 2755
						bio_flags);
	BUG_ON(ret < 0);
	return ret;

}

2756
static int submit_extent_page(int rw, struct extent_io_tree *tree,
2757
			      struct writeback_control *wbc,
2758 2759 2760 2761 2762
			      struct page *page, sector_t sector,
			      size_t size, unsigned long offset,
			      struct block_device *bdev,
			      struct bio **bio_ret,
			      unsigned long max_pages,
2763
			      bio_end_io_t end_io_func,
C
Chris Mason 已提交
2764 2765
			      int mirror_num,
			      unsigned long prev_bio_flags,
2766 2767
			      unsigned long bio_flags,
			      bool force_bio_submit)
2768 2769 2770
{
	int ret = 0;
	struct bio *bio;
C
Chris Mason 已提交
2771 2772
	int contig = 0;
	int old_compressed = prev_bio_flags & EXTENT_BIO_COMPRESSED;
2773
	size_t page_size = min_t(size_t, size, PAGE_CACHE_SIZE);
2774 2775 2776

	if (bio_ret && *bio_ret) {
		bio = *bio_ret;
C
Chris Mason 已提交
2777
		if (old_compressed)
2778
			contig = bio->bi_iter.bi_sector == sector;
C
Chris Mason 已提交
2779
		else
K
Kent Overstreet 已提交
2780
			contig = bio_end_sector(bio) == sector;
C
Chris Mason 已提交
2781 2782

		if (prev_bio_flags != bio_flags || !contig ||
2783
		    force_bio_submit ||
2784
		    merge_bio(rw, tree, page, offset, page_size, bio, bio_flags) ||
C
Chris Mason 已提交
2785 2786 2787
		    bio_add_page(bio, page, page_size, offset) < page_size) {
			ret = submit_one_bio(rw, bio, mirror_num,
					     prev_bio_flags);
2788 2789
			if (ret < 0) {
				*bio_ret = NULL;
2790
				return ret;
2791
			}
2792 2793
			bio = NULL;
		} else {
2794 2795
			if (wbc)
				wbc_account_io(wbc, page, page_size);
2796 2797 2798
			return 0;
		}
	}
C
Chris Mason 已提交
2799

2800 2801
	bio = btrfs_bio_alloc(bdev, sector, BIO_MAX_PAGES,
			GFP_NOFS | __GFP_HIGH);
2802 2803
	if (!bio)
		return -ENOMEM;
2804

C
Chris Mason 已提交
2805
	bio_add_page(bio, page, page_size, offset);
2806 2807
	bio->bi_end_io = end_io_func;
	bio->bi_private = tree;
2808 2809 2810 2811
	if (wbc) {
		wbc_init_bio(wbc, bio);
		wbc_account_io(wbc, page, page_size);
	}
2812

C
Chris Mason 已提交
2813
	if (bio_ret)
2814
		*bio_ret = bio;
C
Chris Mason 已提交
2815
	else
C
Chris Mason 已提交
2816
		ret = submit_one_bio(rw, bio, mirror_num, bio_flags);
2817 2818 2819 2820

	return ret;
}

2821 2822
static void attach_extent_buffer_page(struct extent_buffer *eb,
				      struct page *page)
2823 2824 2825 2826
{
	if (!PagePrivate(page)) {
		SetPagePrivate(page);
		page_cache_get(page);
J
Josef Bacik 已提交
2827 2828 2829
		set_page_private(page, (unsigned long)eb);
	} else {
		WARN_ON(page->private != (unsigned long)eb);
2830 2831 2832
	}
}

J
Josef Bacik 已提交
2833
void set_page_extent_mapped(struct page *page)
2834
{
J
Josef Bacik 已提交
2835 2836 2837 2838 2839
	if (!PagePrivate(page)) {
		SetPagePrivate(page);
		page_cache_get(page);
		set_page_private(page, EXTENT_PAGE_PRIVATE);
	}
2840 2841
}

2842 2843 2844 2845 2846 2847 2848 2849 2850
static struct extent_map *
__get_extent_map(struct inode *inode, struct page *page, size_t pg_offset,
		 u64 start, u64 len, get_extent_t *get_extent,
		 struct extent_map **em_cached)
{
	struct extent_map *em;

	if (em_cached && *em_cached) {
		em = *em_cached;
2851
		if (extent_map_in_tree(em) && start >= em->start &&
2852 2853 2854 2855 2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868
		    start < extent_map_end(em)) {
			atomic_inc(&em->refs);
			return em;
		}

		free_extent_map(em);
		*em_cached = NULL;
	}

	em = get_extent(inode, page, pg_offset, start, len, 0);
	if (em_cached && !IS_ERR_OR_NULL(em)) {
		BUG_ON(*em_cached);
		atomic_inc(&em->refs);
		*em_cached = em;
	}
	return em;
}
2869 2870 2871 2872
/*
 * basic readpage implementation.  Locked extent state structs are inserted
 * into the tree that are removed when the IO is done (by the end_io
 * handlers)
2873
 * XXX JDM: This needs looking at to ensure proper page locking
2874
 */
2875 2876 2877
static int __do_readpage(struct extent_io_tree *tree,
			 struct page *page,
			 get_extent_t *get_extent,
2878
			 struct extent_map **em_cached,
2879
			 struct bio **bio, int mirror_num,
2880 2881
			 unsigned long *bio_flags, int rw,
			 u64 *prev_em_start)
2882 2883
{
	struct inode *inode = page->mapping->host;
M
Miao Xie 已提交
2884
	u64 start = page_offset(page);
2885 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895 2896
	u64 page_end = start + PAGE_CACHE_SIZE - 1;
	u64 end;
	u64 cur = start;
	u64 extent_offset;
	u64 last_byte = i_size_read(inode);
	u64 block_start;
	u64 cur_end;
	sector_t sector;
	struct extent_map *em;
	struct block_device *bdev;
	int ret;
	int nr = 0;
2897
	int parent_locked = *bio_flags & EXTENT_BIO_PARENT_LOCKED;
2898
	size_t pg_offset = 0;
2899
	size_t iosize;
C
Chris Mason 已提交
2900
	size_t disk_io_size;
2901
	size_t blocksize = inode->i_sb->s_blocksize;
2902
	unsigned long this_bio_flag = *bio_flags & EXTENT_BIO_PARENT_LOCKED;
2903 2904 2905

	set_page_extent_mapped(page);

2906
	end = page_end;
D
Dan Magenheimer 已提交
2907 2908 2909
	if (!PageUptodate(page)) {
		if (cleancache_get_page(page) == 0) {
			BUG_ON(blocksize != PAGE_SIZE);
2910
			unlock_extent(tree, start, end);
D
Dan Magenheimer 已提交
2911 2912 2913 2914
			goto out;
		}
	}

C
Chris Mason 已提交
2915 2916 2917 2918 2919 2920
	if (page->index == last_byte >> PAGE_CACHE_SHIFT) {
		char *userpage;
		size_t zero_offset = last_byte & (PAGE_CACHE_SIZE - 1);

		if (zero_offset) {
			iosize = PAGE_CACHE_SIZE - zero_offset;
2921
			userpage = kmap_atomic(page);
C
Chris Mason 已提交
2922 2923
			memset(userpage + zero_offset, 0, iosize);
			flush_dcache_page(page);
2924
			kunmap_atomic(userpage);
C
Chris Mason 已提交
2925 2926
		}
	}
2927
	while (cur <= end) {
2928
		unsigned long pnr = (last_byte >> PAGE_CACHE_SHIFT) + 1;
2929
		bool force_bio_submit = false;
2930

2931 2932
		if (cur >= last_byte) {
			char *userpage;
2933 2934
			struct extent_state *cached = NULL;

2935
			iosize = PAGE_CACHE_SIZE - pg_offset;
2936
			userpage = kmap_atomic(page);
2937
			memset(userpage + pg_offset, 0, iosize);
2938
			flush_dcache_page(page);
2939
			kunmap_atomic(userpage);
2940
			set_extent_uptodate(tree, cur, cur + iosize - 1,
2941
					    &cached, GFP_NOFS);
2942 2943 2944 2945
			if (!parent_locked)
				unlock_extent_cached(tree, cur,
						     cur + iosize - 1,
						     &cached, GFP_NOFS);
2946 2947
			break;
		}
2948 2949
		em = __get_extent_map(inode, page, pg_offset, cur,
				      end - cur + 1, get_extent, em_cached);
2950
		if (IS_ERR_OR_NULL(em)) {
2951
			SetPageError(page);
2952 2953
			if (!parent_locked)
				unlock_extent(tree, cur, end);
2954 2955 2956 2957 2958 2959
			break;
		}
		extent_offset = cur - em->start;
		BUG_ON(extent_map_end(em) <= cur);
		BUG_ON(end < cur);

2960
		if (test_bit(EXTENT_FLAG_COMPRESSED, &em->flags)) {
2961
			this_bio_flag |= EXTENT_BIO_COMPRESSED;
2962 2963 2964
			extent_set_compress_type(&this_bio_flag,
						 em->compress_type);
		}
C
Chris Mason 已提交
2965

2966 2967
		iosize = min(extent_map_end(em) - cur, end - cur + 1);
		cur_end = min(extent_map_end(em) - 1, end);
2968
		iosize = ALIGN(iosize, blocksize);
C
Chris Mason 已提交
2969 2970 2971 2972 2973 2974 2975
		if (this_bio_flag & EXTENT_BIO_COMPRESSED) {
			disk_io_size = em->block_len;
			sector = em->block_start >> 9;
		} else {
			sector = (em->block_start + extent_offset) >> 9;
			disk_io_size = iosize;
		}
2976 2977
		bdev = em->bdev;
		block_start = em->block_start;
Y
Yan Zheng 已提交
2978 2979
		if (test_bit(EXTENT_FLAG_PREALLOC, &em->flags))
			block_start = EXTENT_MAP_HOLE;
2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012 3013 3014 3015 3016 3017 3018 3019 3020 3021 3022

		/*
		 * If we have a file range that points to a compressed extent
		 * and it's followed by a consecutive file range that points to
		 * to the same compressed extent (possibly with a different
		 * offset and/or length, so it either points to the whole extent
		 * or only part of it), we must make sure we do not submit a
		 * single bio to populate the pages for the 2 ranges because
		 * this makes the compressed extent read zero out the pages
		 * belonging to the 2nd range. Imagine the following scenario:
		 *
		 *  File layout
		 *  [0 - 8K]                     [8K - 24K]
		 *    |                               |
		 *    |                               |
		 * points to extent X,         points to extent X,
		 * offset 4K, length of 8K     offset 0, length 16K
		 *
		 * [extent X, compressed length = 4K uncompressed length = 16K]
		 *
		 * If the bio to read the compressed extent covers both ranges,
		 * it will decompress extent X into the pages belonging to the
		 * first range and then it will stop, zeroing out the remaining
		 * pages that belong to the other range that points to extent X.
		 * So here we make sure we submit 2 bios, one for the first
		 * range and another one for the third range. Both will target
		 * the same physical extent from disk, but we can't currently
		 * make the compressed bio endio callback populate the pages
		 * for both ranges because each compressed bio is tightly
		 * coupled with a single extent map, and each range can have
		 * an extent map with a different offset value relative to the
		 * uncompressed data of our extent and different lengths. This
		 * is a corner case so we prioritize correctness over
		 * non-optimal behavior (submitting 2 bios for the same extent).
		 */
		if (test_bit(EXTENT_FLAG_COMPRESSED, &em->flags) &&
		    prev_em_start && *prev_em_start != (u64)-1 &&
		    *prev_em_start != em->orig_start)
			force_bio_submit = true;

		if (prev_em_start)
			*prev_em_start = em->orig_start;

3023 3024 3025 3026 3027 3028
		free_extent_map(em);
		em = NULL;

		/* we've found a hole, just zero and go on */
		if (block_start == EXTENT_MAP_HOLE) {
			char *userpage;
3029 3030
			struct extent_state *cached = NULL;

3031
			userpage = kmap_atomic(page);
3032
			memset(userpage + pg_offset, 0, iosize);
3033
			flush_dcache_page(page);
3034
			kunmap_atomic(userpage);
3035 3036

			set_extent_uptodate(tree, cur, cur + iosize - 1,
3037
					    &cached, GFP_NOFS);
3038 3039 3040 3041 3042 3043
			if (parent_locked)
				free_extent_state(cached);
			else
				unlock_extent_cached(tree, cur,
						     cur + iosize - 1,
						     &cached, GFP_NOFS);
3044
			cur = cur + iosize;
3045
			pg_offset += iosize;
3046 3047 3048
			continue;
		}
		/* the get_extent function already copied into the page */
3049 3050
		if (test_range_bit(tree, cur, cur_end,
				   EXTENT_UPTODATE, 1, NULL)) {
3051
			check_page_uptodate(tree, page);
3052 3053
			if (!parent_locked)
				unlock_extent(tree, cur, cur + iosize - 1);
3054
			cur = cur + iosize;
3055
			pg_offset += iosize;
3056 3057
			continue;
		}
3058 3059 3060 3061 3062
		/* we have an inline extent but it didn't get marked up
		 * to date.  Error out
		 */
		if (block_start == EXTENT_MAP_INLINE) {
			SetPageError(page);
3063 3064
			if (!parent_locked)
				unlock_extent(tree, cur, cur + iosize - 1);
3065
			cur = cur + iosize;
3066
			pg_offset += iosize;
3067 3068
			continue;
		}
3069

3070
		pnr -= page->index;
3071
		ret = submit_extent_page(rw, tree, NULL, page,
3072
					 sector, disk_io_size, pg_offset,
3073
					 bdev, bio, pnr,
C
Chris Mason 已提交
3074 3075
					 end_bio_extent_readpage, mirror_num,
					 *bio_flags,
3076 3077
					 this_bio_flag,
					 force_bio_submit);
3078 3079 3080 3081
		if (!ret) {
			nr++;
			*bio_flags = this_bio_flag;
		} else {
3082
			SetPageError(page);
3083 3084
			if (!parent_locked)
				unlock_extent(tree, cur, cur + iosize - 1);
3085
		}
3086
		cur = cur + iosize;
3087
		pg_offset += iosize;
3088
	}
D
Dan Magenheimer 已提交
3089
out:
3090 3091 3092 3093 3094 3095 3096 3097
	if (!nr) {
		if (!PageError(page))
			SetPageUptodate(page);
		unlock_page(page);
	}
	return 0;
}

3098 3099 3100 3101
static inline void __do_contiguous_readpages(struct extent_io_tree *tree,
					     struct page *pages[], int nr_pages,
					     u64 start, u64 end,
					     get_extent_t *get_extent,
3102
					     struct extent_map **em_cached,
3103
					     struct bio **bio, int mirror_num,
3104 3105
					     unsigned long *bio_flags, int rw,
					     u64 *prev_em_start)
3106 3107 3108 3109 3110 3111 3112 3113 3114 3115 3116 3117 3118 3119 3120 3121 3122 3123
{
	struct inode *inode;
	struct btrfs_ordered_extent *ordered;
	int index;

	inode = pages[0]->mapping->host;
	while (1) {
		lock_extent(tree, start, end);
		ordered = btrfs_lookup_ordered_range(inode, start,
						     end - start + 1);
		if (!ordered)
			break;
		unlock_extent(tree, start, end);
		btrfs_start_ordered_extent(inode, ordered, 1);
		btrfs_put_ordered_extent(ordered);
	}

	for (index = 0; index < nr_pages; index++) {
3124
		__do_readpage(tree, pages[index], get_extent, em_cached, bio,
3125
			      mirror_num, bio_flags, rw, prev_em_start);
3126 3127 3128 3129 3130 3131 3132
		page_cache_release(pages[index]);
	}
}

static void __extent_readpages(struct extent_io_tree *tree,
			       struct page *pages[],
			       int nr_pages, get_extent_t *get_extent,
3133
			       struct extent_map **em_cached,
3134
			       struct bio **bio, int mirror_num,
3135 3136
			       unsigned long *bio_flags, int rw,
			       u64 *prev_em_start)
3137
{
3138
	u64 start = 0;
3139 3140 3141
	u64 end = 0;
	u64 page_start;
	int index;
3142
	int first_index = 0;
3143 3144 3145 3146 3147 3148 3149 3150 3151 3152 3153 3154

	for (index = 0; index < nr_pages; index++) {
		page_start = page_offset(pages[index]);
		if (!end) {
			start = page_start;
			end = start + PAGE_CACHE_SIZE - 1;
			first_index = index;
		} else if (end + 1 == page_start) {
			end += PAGE_CACHE_SIZE;
		} else {
			__do_contiguous_readpages(tree, &pages[first_index],
						  index - first_index, start,
3155 3156
						  end, get_extent, em_cached,
						  bio, mirror_num, bio_flags,
3157
						  rw, prev_em_start);
3158 3159 3160 3161 3162 3163 3164 3165 3166
			start = page_start;
			end = start + PAGE_CACHE_SIZE - 1;
			first_index = index;
		}
	}

	if (end)
		__do_contiguous_readpages(tree, &pages[first_index],
					  index - first_index, start,
3167
					  end, get_extent, em_cached, bio,
3168 3169
					  mirror_num, bio_flags, rw,
					  prev_em_start);
3170 3171 3172 3173 3174 3175 3176 3177 3178 3179 3180 3181 3182 3183 3184 3185 3186 3187 3188 3189 3190 3191 3192 3193
}

static int __extent_read_full_page(struct extent_io_tree *tree,
				   struct page *page,
				   get_extent_t *get_extent,
				   struct bio **bio, int mirror_num,
				   unsigned long *bio_flags, int rw)
{
	struct inode *inode = page->mapping->host;
	struct btrfs_ordered_extent *ordered;
	u64 start = page_offset(page);
	u64 end = start + PAGE_CACHE_SIZE - 1;
	int ret;

	while (1) {
		lock_extent(tree, start, end);
		ordered = btrfs_lookup_ordered_extent(inode, start);
		if (!ordered)
			break;
		unlock_extent(tree, start, end);
		btrfs_start_ordered_extent(inode, ordered, 1);
		btrfs_put_ordered_extent(ordered);
	}

3194
	ret = __do_readpage(tree, page, get_extent, NULL, bio, mirror_num,
3195
			    bio_flags, rw, NULL);
3196 3197 3198
	return ret;
}

3199
int extent_read_full_page(struct extent_io_tree *tree, struct page *page,
3200
			    get_extent_t *get_extent, int mirror_num)
3201 3202
{
	struct bio *bio = NULL;
C
Chris Mason 已提交
3203
	unsigned long bio_flags = 0;
3204 3205
	int ret;

3206
	ret = __extent_read_full_page(tree, page, get_extent, &bio, mirror_num,
3207
				      &bio_flags, READ);
3208
	if (bio)
3209
		ret = submit_one_bio(READ, bio, mirror_num, bio_flags);
3210 3211 3212
	return ret;
}

3213 3214 3215 3216 3217 3218 3219 3220
int extent_read_full_page_nolock(struct extent_io_tree *tree, struct page *page,
				 get_extent_t *get_extent, int mirror_num)
{
	struct bio *bio = NULL;
	unsigned long bio_flags = EXTENT_BIO_PARENT_LOCKED;
	int ret;

	ret = __do_readpage(tree, page, get_extent, NULL, &bio, mirror_num,
3221
			    &bio_flags, READ, NULL);
3222 3223 3224 3225 3226
	if (bio)
		ret = submit_one_bio(READ, bio, mirror_num, bio_flags);
	return ret;
}

3227 3228 3229 3230 3231 3232 3233 3234 3235 3236
static noinline void update_nr_written(struct page *page,
				      struct writeback_control *wbc,
				      unsigned long nr_written)
{
	wbc->nr_to_write -= nr_written;
	if (wbc->range_cyclic || (wbc->nr_to_write > 0 &&
	    wbc->range_start == 0 && wbc->range_end == LLONG_MAX))
		page->mapping->writeback_index = page->index + nr_written;
}

3237
/*
3238 3239 3240 3241 3242 3243 3244 3245
 * helper for __extent_writepage, doing all of the delayed allocation setup.
 *
 * This returns 1 if our fill_delalloc function did all the work required
 * to write the page (copy into inline extent).  In this case the IO has
 * been started and the page is already unlocked.
 *
 * This returns 0 if all went well (page still locked)
 * This returns < 0 if there were errors (page still locked)
3246
 */
3247 3248 3249 3250 3251 3252 3253 3254 3255 3256 3257 3258 3259 3260 3261 3262 3263 3264 3265 3266 3267 3268
static noinline_for_stack int writepage_delalloc(struct inode *inode,
			      struct page *page, struct writeback_control *wbc,
			      struct extent_page_data *epd,
			      u64 delalloc_start,
			      unsigned long *nr_written)
{
	struct extent_io_tree *tree = epd->tree;
	u64 page_end = delalloc_start + PAGE_CACHE_SIZE - 1;
	u64 nr_delalloc;
	u64 delalloc_to_write = 0;
	u64 delalloc_end = 0;
	int ret;
	int page_started = 0;

	if (epd->extent_locked || !tree->ops || !tree->ops->fill_delalloc)
		return 0;

	while (delalloc_end < page_end) {
		nr_delalloc = find_lock_delalloc_range(inode, tree,
					       page,
					       &delalloc_start,
					       &delalloc_end,
3269
					       BTRFS_MAX_EXTENT_SIZE);
3270 3271 3272 3273 3274 3275 3276 3277 3278 3279 3280 3281 3282 3283 3284 3285 3286 3287 3288 3289 3290 3291 3292 3293 3294 3295 3296 3297 3298 3299 3300 3301 3302 3303 3304 3305 3306 3307 3308 3309 3310 3311 3312 3313 3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325 3326 3327 3328 3329 3330 3331 3332 3333 3334 3335 3336 3337 3338 3339 3340 3341 3342
		if (nr_delalloc == 0) {
			delalloc_start = delalloc_end + 1;
			continue;
		}
		ret = tree->ops->fill_delalloc(inode, page,
					       delalloc_start,
					       delalloc_end,
					       &page_started,
					       nr_written);
		/* File system has been set read-only */
		if (ret) {
			SetPageError(page);
			/* fill_delalloc should be return < 0 for error
			 * but just in case, we use > 0 here meaning the
			 * IO is started, so we don't want to return > 0
			 * unless things are going well.
			 */
			ret = ret < 0 ? ret : -EIO;
			goto done;
		}
		/*
		 * delalloc_end is already one less than the total
		 * length, so we don't subtract one from
		 * PAGE_CACHE_SIZE
		 */
		delalloc_to_write += (delalloc_end - delalloc_start +
				      PAGE_CACHE_SIZE) >>
				      PAGE_CACHE_SHIFT;
		delalloc_start = delalloc_end + 1;
	}
	if (wbc->nr_to_write < delalloc_to_write) {
		int thresh = 8192;

		if (delalloc_to_write < thresh * 2)
			thresh = delalloc_to_write;
		wbc->nr_to_write = min_t(u64, delalloc_to_write,
					 thresh);
	}

	/* did the fill delalloc function already unlock and start
	 * the IO?
	 */
	if (page_started) {
		/*
		 * we've unlocked the page, so we can't update
		 * the mapping's writeback index, just update
		 * nr_to_write.
		 */
		wbc->nr_to_write -= *nr_written;
		return 1;
	}

	ret = 0;

done:
	return ret;
}

/*
 * helper for __extent_writepage.  This calls the writepage start hooks,
 * and does the loop to map the page into extents and bios.
 *
 * We return 1 if the IO is started and the page is unlocked,
 * 0 if all went well (page still locked)
 * < 0 if there were errors (page still locked)
 */
static noinline_for_stack int __extent_writepage_io(struct inode *inode,
				 struct page *page,
				 struct writeback_control *wbc,
				 struct extent_page_data *epd,
				 loff_t i_size,
				 unsigned long nr_written,
				 int write_flags, int *nr_ret)
3343 3344
{
	struct extent_io_tree *tree = epd->tree;
M
Miao Xie 已提交
3345
	u64 start = page_offset(page);
3346 3347 3348 3349 3350 3351 3352
	u64 page_end = start + PAGE_CACHE_SIZE - 1;
	u64 end;
	u64 cur = start;
	u64 extent_offset;
	u64 block_start;
	u64 iosize;
	sector_t sector;
3353
	struct extent_state *cached_state = NULL;
3354 3355
	struct extent_map *em;
	struct block_device *bdev;
3356
	size_t pg_offset = 0;
3357
	size_t blocksize;
3358 3359 3360
	int ret = 0;
	int nr = 0;
	bool compressed;
C
Chris Mason 已提交
3361

3362
	if (tree->ops && tree->ops->writepage_start_hook) {
C
Chris Mason 已提交
3363 3364
		ret = tree->ops->writepage_start_hook(page, start,
						      page_end);
3365 3366 3367 3368 3369 3370
		if (ret) {
			/* Fixup worker will requeue */
			if (ret == -EBUSY)
				wbc->pages_skipped++;
			else
				redirty_page_for_writepage(wbc, page);
3371

3372
			update_nr_written(page, wbc, nr_written);
3373
			unlock_page(page);
3374
			ret = 1;
3375
			goto done_unlocked;
3376 3377 3378
		}
	}

3379 3380 3381 3382 3383
	/*
	 * we don't want to touch the inode after unlocking the page,
	 * so we update the mapping writeback index now
	 */
	update_nr_written(page, wbc, nr_written + 1);
3384

3385
	end = page_end;
3386
	if (i_size <= start) {
3387 3388 3389
		if (tree->ops && tree->ops->writepage_end_io_hook)
			tree->ops->writepage_end_io_hook(page, start,
							 page_end, NULL, 1);
3390 3391 3392 3393 3394 3395
		goto done;
	}

	blocksize = inode->i_sb->s_blocksize;

	while (cur <= end) {
3396 3397
		u64 em_end;
		if (cur >= i_size) {
3398 3399 3400
			if (tree->ops && tree->ops->writepage_end_io_hook)
				tree->ops->writepage_end_io_hook(page, cur,
							 page_end, NULL, 1);
3401 3402
			break;
		}
3403
		em = epd->get_extent(inode, page, pg_offset, cur,
3404
				     end - cur + 1, 1);
3405
		if (IS_ERR_OR_NULL(em)) {
3406
			SetPageError(page);
3407
			ret = PTR_ERR_OR_ZERO(em);
3408 3409 3410 3411
			break;
		}

		extent_offset = cur - em->start;
3412 3413
		em_end = extent_map_end(em);
		BUG_ON(em_end <= cur);
3414
		BUG_ON(end < cur);
3415
		iosize = min(em_end - cur, end - cur + 1);
3416
		iosize = ALIGN(iosize, blocksize);
3417 3418 3419
		sector = (em->block_start + extent_offset) >> 9;
		bdev = em->bdev;
		block_start = em->block_start;
C
Chris Mason 已提交
3420
		compressed = test_bit(EXTENT_FLAG_COMPRESSED, &em->flags);
3421 3422 3423
		free_extent_map(em);
		em = NULL;

C
Chris Mason 已提交
3424 3425 3426 3427 3428
		/*
		 * compressed and inline extents are written through other
		 * paths in the FS
		 */
		if (compressed || block_start == EXTENT_MAP_HOLE ||
3429
		    block_start == EXTENT_MAP_INLINE) {
C
Chris Mason 已提交
3430 3431 3432 3433 3434 3435
			/*
			 * end_io notification does not happen here for
			 * compressed extents
			 */
			if (!compressed && tree->ops &&
			    tree->ops->writepage_end_io_hook)
3436 3437 3438
				tree->ops->writepage_end_io_hook(page, cur,
							 cur + iosize - 1,
							 NULL, 1);
C
Chris Mason 已提交
3439 3440 3441 3442 3443 3444 3445 3446 3447
			else if (compressed) {
				/* we don't want to end_page_writeback on
				 * a compressed extent.  this happens
				 * elsewhere
				 */
				nr++;
			}

			cur += iosize;
3448
			pg_offset += iosize;
3449 3450
			continue;
		}
C
Chris Mason 已提交
3451

3452 3453 3454 3455 3456 3457
		if (tree->ops && tree->ops->writepage_io_hook) {
			ret = tree->ops->writepage_io_hook(page, cur,
						cur + iosize - 1);
		} else {
			ret = 0;
		}
3458
		if (ret) {
3459
			SetPageError(page);
3460
		} else {
3461
			unsigned long max_nr = (i_size >> PAGE_CACHE_SHIFT) + 1;
3462

3463 3464
			set_range_writeback(tree, cur, cur + iosize - 1);
			if (!PageWriteback(page)) {
3465 3466
				btrfs_err(BTRFS_I(inode)->root->fs_info,
					   "page %lu not writeback, cur %llu end %llu",
3467
				       page->index, cur, end);
3468 3469
			}

3470
			ret = submit_extent_page(write_flags, tree, wbc, page,
3471 3472
						 sector, iosize, pg_offset,
						 bdev, &epd->bio, max_nr,
C
Chris Mason 已提交
3473
						 end_bio_extent_writepage,
3474
						 0, 0, 0, false);
3475 3476 3477 3478
			if (ret)
				SetPageError(page);
		}
		cur = cur + iosize;
3479
		pg_offset += iosize;
3480 3481
		nr++;
	}
3482 3483 3484 3485 3486 3487 3488 3489 3490 3491 3492 3493 3494 3495 3496 3497 3498 3499 3500 3501 3502 3503 3504 3505 3506 3507 3508 3509 3510 3511 3512 3513 3514 3515 3516 3517 3518 3519 3520 3521 3522 3523 3524 3525 3526 3527 3528 3529 3530 3531 3532 3533 3534 3535 3536 3537 3538 3539 3540 3541 3542 3543 3544 3545 3546 3547 3548 3549 3550 3551 3552 3553 3554 3555 3556
done:
	*nr_ret = nr;

done_unlocked:

	/* drop our reference on any cached states */
	free_extent_state(cached_state);
	return ret;
}

/*
 * the writepage semantics are similar to regular writepage.  extent
 * records are inserted to lock ranges in the tree, and as dirty areas
 * are found, they are marked writeback.  Then the lock bits are removed
 * and the end_io handler clears the writeback ranges
 */
static int __extent_writepage(struct page *page, struct writeback_control *wbc,
			      void *data)
{
	struct inode *inode = page->mapping->host;
	struct extent_page_data *epd = data;
	u64 start = page_offset(page);
	u64 page_end = start + PAGE_CACHE_SIZE - 1;
	int ret;
	int nr = 0;
	size_t pg_offset = 0;
	loff_t i_size = i_size_read(inode);
	unsigned long end_index = i_size >> PAGE_CACHE_SHIFT;
	int write_flags;
	unsigned long nr_written = 0;

	if (wbc->sync_mode == WB_SYNC_ALL)
		write_flags = WRITE_SYNC;
	else
		write_flags = WRITE;

	trace___extent_writepage(page, inode, wbc);

	WARN_ON(!PageLocked(page));

	ClearPageError(page);

	pg_offset = i_size & (PAGE_CACHE_SIZE - 1);
	if (page->index > end_index ||
	   (page->index == end_index && !pg_offset)) {
		page->mapping->a_ops->invalidatepage(page, 0, PAGE_CACHE_SIZE);
		unlock_page(page);
		return 0;
	}

	if (page->index == end_index) {
		char *userpage;

		userpage = kmap_atomic(page);
		memset(userpage + pg_offset, 0,
		       PAGE_CACHE_SIZE - pg_offset);
		kunmap_atomic(userpage);
		flush_dcache_page(page);
	}

	pg_offset = 0;

	set_page_extent_mapped(page);

	ret = writepage_delalloc(inode, page, wbc, epd, start, &nr_written);
	if (ret == 1)
		goto done_unlocked;
	if (ret)
		goto done;

	ret = __extent_writepage_io(inode, page, wbc, epd,
				    i_size, nr_written, write_flags, &nr);
	if (ret == 1)
		goto done_unlocked;

3557 3558 3559 3560 3561 3562
done:
	if (nr == 0) {
		/* make sure the mapping tag for page dirty gets cleared */
		set_page_writeback(page);
		end_page_writeback(page);
	}
3563 3564 3565 3566
	if (PageError(page)) {
		ret = ret < 0 ? ret : -EIO;
		end_extent_writepage(page, ret, start, page_end);
	}
3567
	unlock_page(page);
3568
	return ret;
3569

3570
done_unlocked:
3571 3572 3573
	return 0;
}

3574
void wait_on_extent_buffer_writeback(struct extent_buffer *eb)
3575
{
3576 3577
	wait_on_bit_io(&eb->bflags, EXTENT_BUFFER_WRITEBACK,
		       TASK_UNINTERRUPTIBLE);
3578 3579
}

3580 3581 3582 3583
static noinline_for_stack int
lock_extent_buffer_for_io(struct extent_buffer *eb,
			  struct btrfs_fs_info *fs_info,
			  struct extent_page_data *epd)
3584 3585 3586 3587 3588 3589 3590 3591 3592 3593 3594 3595 3596 3597 3598 3599 3600 3601 3602
{
	unsigned long i, num_pages;
	int flush = 0;
	int ret = 0;

	if (!btrfs_try_tree_write_lock(eb)) {
		flush = 1;
		flush_write_bio(epd);
		btrfs_tree_lock(eb);
	}

	if (test_bit(EXTENT_BUFFER_WRITEBACK, &eb->bflags)) {
		btrfs_tree_unlock(eb);
		if (!epd->sync_io)
			return 0;
		if (!flush) {
			flush_write_bio(epd);
			flush = 1;
		}
C
Chris Mason 已提交
3603 3604 3605 3606 3607
		while (1) {
			wait_on_extent_buffer_writeback(eb);
			btrfs_tree_lock(eb);
			if (!test_bit(EXTENT_BUFFER_WRITEBACK, &eb->bflags))
				break;
3608 3609 3610 3611
			btrfs_tree_unlock(eb);
		}
	}

3612 3613 3614 3615 3616 3617
	/*
	 * We need to do this to prevent races in people who check if the eb is
	 * under IO since we can end up having no IO bits set for a short period
	 * of time.
	 */
	spin_lock(&eb->refs_lock);
3618 3619
	if (test_and_clear_bit(EXTENT_BUFFER_DIRTY, &eb->bflags)) {
		set_bit(EXTENT_BUFFER_WRITEBACK, &eb->bflags);
3620
		spin_unlock(&eb->refs_lock);
3621
		btrfs_set_header_flag(eb, BTRFS_HEADER_FLAG_WRITTEN);
3622 3623 3624
		__percpu_counter_add(&fs_info->dirty_metadata_bytes,
				     -eb->len,
				     fs_info->dirty_metadata_batch);
3625
		ret = 1;
3626 3627
	} else {
		spin_unlock(&eb->refs_lock);
3628 3629 3630 3631 3632 3633 3634 3635 3636
	}

	btrfs_tree_unlock(eb);

	if (!ret)
		return ret;

	num_pages = num_extent_pages(eb->start, eb->len);
	for (i = 0; i < num_pages; i++) {
3637
		struct page *p = eb->pages[i];
3638 3639 3640 3641 3642 3643 3644 3645 3646 3647 3648 3649 3650 3651 3652 3653

		if (!trylock_page(p)) {
			if (!flush) {
				flush_write_bio(epd);
				flush = 1;
			}
			lock_page(p);
		}
	}

	return ret;
}

static void end_extent_buffer_writeback(struct extent_buffer *eb)
{
	clear_bit(EXTENT_BUFFER_WRITEBACK, &eb->bflags);
3654
	smp_mb__after_atomic();
3655 3656 3657
	wake_up_bit(&eb->bflags, EXTENT_BUFFER_WRITEBACK);
}

3658 3659 3660 3661 3662 3663 3664 3665 3666 3667 3668 3669 3670 3671 3672 3673 3674 3675 3676 3677 3678 3679 3680 3681 3682 3683 3684 3685 3686 3687 3688 3689 3690 3691 3692 3693 3694 3695 3696 3697 3698 3699 3700 3701 3702 3703 3704 3705 3706 3707 3708 3709 3710 3711 3712 3713 3714 3715 3716 3717 3718 3719
static void set_btree_ioerr(struct page *page)
{
	struct extent_buffer *eb = (struct extent_buffer *)page->private;
	struct btrfs_inode *btree_ino = BTRFS_I(eb->fs_info->btree_inode);

	SetPageError(page);
	if (test_and_set_bit(EXTENT_BUFFER_WRITE_ERR, &eb->bflags))
		return;

	/*
	 * If writeback for a btree extent that doesn't belong to a log tree
	 * failed, increment the counter transaction->eb_write_errors.
	 * We do this because while the transaction is running and before it's
	 * committing (when we call filemap_fdata[write|wait]_range against
	 * the btree inode), we might have
	 * btree_inode->i_mapping->a_ops->writepages() called by the VM - if it
	 * returns an error or an error happens during writeback, when we're
	 * committing the transaction we wouldn't know about it, since the pages
	 * can be no longer dirty nor marked anymore for writeback (if a
	 * subsequent modification to the extent buffer didn't happen before the
	 * transaction commit), which makes filemap_fdata[write|wait]_range not
	 * able to find the pages tagged with SetPageError at transaction
	 * commit time. So if this happens we must abort the transaction,
	 * otherwise we commit a super block with btree roots that point to
	 * btree nodes/leafs whose content on disk is invalid - either garbage
	 * or the content of some node/leaf from a past generation that got
	 * cowed or deleted and is no longer valid.
	 *
	 * Note: setting AS_EIO/AS_ENOSPC in the btree inode's i_mapping would
	 * not be enough - we need to distinguish between log tree extents vs
	 * non-log tree extents, and the next filemap_fdatawait_range() call
	 * will catch and clear such errors in the mapping - and that call might
	 * be from a log sync and not from a transaction commit. Also, checking
	 * for the eb flag EXTENT_BUFFER_WRITE_ERR at transaction commit time is
	 * not done and would not be reliable - the eb might have been released
	 * from memory and reading it back again means that flag would not be
	 * set (since it's a runtime flag, not persisted on disk).
	 *
	 * Using the flags below in the btree inode also makes us achieve the
	 * goal of AS_EIO/AS_ENOSPC when writepages() returns success, started
	 * writeback for all dirty pages and before filemap_fdatawait_range()
	 * is called, the writeback for all dirty pages had already finished
	 * with errors - because we were not using AS_EIO/AS_ENOSPC,
	 * filemap_fdatawait_range() would return success, as it could not know
	 * that writeback errors happened (the pages were no longer tagged for
	 * writeback).
	 */
	switch (eb->log_index) {
	case -1:
		set_bit(BTRFS_INODE_BTREE_ERR, &btree_ino->runtime_flags);
		break;
	case 0:
		set_bit(BTRFS_INODE_BTREE_LOG1_ERR, &btree_ino->runtime_flags);
		break;
	case 1:
		set_bit(BTRFS_INODE_BTREE_LOG2_ERR, &btree_ino->runtime_flags);
		break;
	default:
		BUG(); /* unexpected, logic error */
	}
}

3720
static void end_bio_extent_buffer_writepage(struct bio *bio)
3721
{
3722
	struct bio_vec *bvec;
3723
	struct extent_buffer *eb;
3724
	int i, done;
3725

3726
	bio_for_each_segment_all(bvec, bio, i) {
3727 3728 3729 3730 3731 3732
		struct page *page = bvec->bv_page;

		eb = (struct extent_buffer *)page->private;
		BUG_ON(!eb);
		done = atomic_dec_and_test(&eb->io_pages);

3733 3734
		if (bio->bi_error ||
		    test_bit(EXTENT_BUFFER_WRITE_ERR, &eb->bflags)) {
3735
			ClearPageUptodate(page);
3736
			set_btree_ioerr(page);
3737 3738 3739 3740 3741 3742 3743 3744
		}

		end_page_writeback(page);

		if (!done)
			continue;

		end_extent_buffer_writeback(eb);
3745
	}
3746 3747 3748 3749

	bio_put(bio);
}

3750
static noinline_for_stack int write_one_eb(struct extent_buffer *eb,
3751 3752 3753 3754 3755
			struct btrfs_fs_info *fs_info,
			struct writeback_control *wbc,
			struct extent_page_data *epd)
{
	struct block_device *bdev = fs_info->fs_devices->latest_bdev;
3756
	struct extent_io_tree *tree = &BTRFS_I(fs_info->btree_inode)->io_tree;
3757 3758
	u64 offset = eb->start;
	unsigned long i, num_pages;
3759
	unsigned long bio_flags = 0;
3760
	int rw = (epd->sync_io ? WRITE_SYNC : WRITE) | REQ_META;
3761
	int ret = 0;
3762

3763
	clear_bit(EXTENT_BUFFER_WRITE_ERR, &eb->bflags);
3764 3765
	num_pages = num_extent_pages(eb->start, eb->len);
	atomic_set(&eb->io_pages, num_pages);
3766 3767 3768
	if (btrfs_header_owner(eb) == BTRFS_TREE_LOG_OBJECTID)
		bio_flags = EXTENT_BIO_TREE_LOG;

3769
	for (i = 0; i < num_pages; i++) {
3770
		struct page *p = eb->pages[i];
3771 3772 3773

		clear_page_dirty_for_io(p);
		set_page_writeback(p);
3774
		ret = submit_extent_page(rw, tree, wbc, p, offset >> 9,
3775 3776
					 PAGE_CACHE_SIZE, 0, bdev, &epd->bio,
					 -1, end_bio_extent_buffer_writepage,
3777
					 0, epd->bio_flags, bio_flags, false);
3778
		epd->bio_flags = bio_flags;
3779
		if (ret) {
3780
			set_btree_ioerr(p);
3781
			end_page_writeback(p);
3782 3783 3784 3785 3786 3787 3788 3789 3790 3791 3792 3793
			if (atomic_sub_and_test(num_pages - i, &eb->io_pages))
				end_extent_buffer_writeback(eb);
			ret = -EIO;
			break;
		}
		offset += PAGE_CACHE_SIZE;
		update_nr_written(p, wbc, 1);
		unlock_page(p);
	}

	if (unlikely(ret)) {
		for (; i < num_pages; i++) {
3794
			struct page *p = eb->pages[i];
3795
			clear_page_dirty_for_io(p);
3796 3797 3798 3799 3800 3801 3802 3803 3804 3805 3806 3807 3808 3809 3810 3811 3812 3813
			unlock_page(p);
		}
	}

	return ret;
}

int btree_write_cache_pages(struct address_space *mapping,
				   struct writeback_control *wbc)
{
	struct extent_io_tree *tree = &BTRFS_I(mapping->host)->io_tree;
	struct btrfs_fs_info *fs_info = BTRFS_I(mapping->host)->root->fs_info;
	struct extent_buffer *eb, *prev_eb = NULL;
	struct extent_page_data epd = {
		.bio = NULL,
		.tree = tree,
		.extent_locked = 0,
		.sync_io = wbc->sync_mode == WB_SYNC_ALL,
3814
		.bio_flags = 0,
3815 3816 3817 3818 3819 3820 3821 3822 3823 3824 3825 3826 3827 3828 3829 3830 3831 3832 3833 3834 3835 3836 3837 3838 3839 3840 3841 3842 3843 3844 3845 3846 3847 3848 3849 3850 3851 3852 3853 3854 3855 3856 3857 3858
	};
	int ret = 0;
	int done = 0;
	int nr_to_write_done = 0;
	struct pagevec pvec;
	int nr_pages;
	pgoff_t index;
	pgoff_t end;		/* Inclusive */
	int scanned = 0;
	int tag;

	pagevec_init(&pvec, 0);
	if (wbc->range_cyclic) {
		index = mapping->writeback_index; /* Start from prev offset */
		end = -1;
	} else {
		index = wbc->range_start >> PAGE_CACHE_SHIFT;
		end = wbc->range_end >> PAGE_CACHE_SHIFT;
		scanned = 1;
	}
	if (wbc->sync_mode == WB_SYNC_ALL)
		tag = PAGECACHE_TAG_TOWRITE;
	else
		tag = PAGECACHE_TAG_DIRTY;
retry:
	if (wbc->sync_mode == WB_SYNC_ALL)
		tag_pages_for_writeback(mapping, index, end);
	while (!done && !nr_to_write_done && (index <= end) &&
	       (nr_pages = pagevec_lookup_tag(&pvec, mapping, &index, tag,
			min(end - index, (pgoff_t)PAGEVEC_SIZE-1) + 1))) {
		unsigned i;

		scanned = 1;
		for (i = 0; i < nr_pages; i++) {
			struct page *page = pvec.pages[i];

			if (!PagePrivate(page))
				continue;

			if (!wbc->range_cyclic && page->index > end) {
				done = 1;
				break;
			}

3859 3860 3861 3862 3863 3864
			spin_lock(&mapping->private_lock);
			if (!PagePrivate(page)) {
				spin_unlock(&mapping->private_lock);
				continue;
			}

3865
			eb = (struct extent_buffer *)page->private;
3866 3867 3868 3869 3870 3871

			/*
			 * Shouldn't happen and normally this would be a BUG_ON
			 * but no sense in crashing the users box for something
			 * we can survive anyway.
			 */
3872
			if (WARN_ON(!eb)) {
3873
				spin_unlock(&mapping->private_lock);
3874 3875 3876
				continue;
			}

3877 3878
			if (eb == prev_eb) {
				spin_unlock(&mapping->private_lock);
3879
				continue;
3880
			}
3881

3882 3883 3884
			ret = atomic_inc_not_zero(&eb->refs);
			spin_unlock(&mapping->private_lock);
			if (!ret)
3885 3886 3887 3888 3889 3890 3891 3892 3893 3894 3895 3896 3897 3898 3899 3900 3901 3902 3903 3904 3905 3906 3907 3908 3909 3910 3911 3912 3913 3914 3915 3916 3917 3918 3919 3920 3921 3922 3923 3924
				continue;

			prev_eb = eb;
			ret = lock_extent_buffer_for_io(eb, fs_info, &epd);
			if (!ret) {
				free_extent_buffer(eb);
				continue;
			}

			ret = write_one_eb(eb, fs_info, wbc, &epd);
			if (ret) {
				done = 1;
				free_extent_buffer(eb);
				break;
			}
			free_extent_buffer(eb);

			/*
			 * the filesystem may choose to bump up nr_to_write.
			 * We have to make sure to honor the new nr_to_write
			 * at any time
			 */
			nr_to_write_done = wbc->nr_to_write <= 0;
		}
		pagevec_release(&pvec);
		cond_resched();
	}
	if (!scanned && !done) {
		/*
		 * We hit the last page and there is more work to be done: wrap
		 * back to the start of the file
		 */
		scanned = 1;
		index = 0;
		goto retry;
	}
	flush_write_bio(&epd);
	return ret;
}

3925
/**
C
Chris Mason 已提交
3926
 * write_cache_pages - walk the list of dirty pages of the given address space and write all of them.
3927 3928 3929 3930 3931 3932 3933 3934 3935 3936 3937 3938 3939
 * @mapping: address space structure to write
 * @wbc: subtract the number of written pages from *@wbc->nr_to_write
 * @writepage: function called for each page
 * @data: data passed to writepage function
 *
 * If a page is already under I/O, write_cache_pages() skips it, even
 * 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.
 */
3940
static int extent_write_cache_pages(struct extent_io_tree *tree,
C
Chris Mason 已提交
3941 3942
			     struct address_space *mapping,
			     struct writeback_control *wbc,
C
Chris Mason 已提交
3943 3944
			     writepage_t writepage, void *data,
			     void (*flush_fn)(void *))
3945
{
3946
	struct inode *inode = mapping->host;
3947 3948
	int ret = 0;
	int done = 0;
3949
	int err = 0;
3950
	int nr_to_write_done = 0;
3951 3952 3953 3954 3955
	struct pagevec pvec;
	int nr_pages;
	pgoff_t index;
	pgoff_t end;		/* Inclusive */
	int scanned = 0;
3956
	int tag;
3957

3958 3959 3960 3961 3962 3963 3964 3965 3966 3967 3968 3969
	/*
	 * We have to hold onto the inode so that ordered extents can do their
	 * work when the IO finishes.  The alternative to this is failing to add
	 * an ordered extent if the igrab() fails there and that is a huge pain
	 * to deal with, so instead just hold onto the inode throughout the
	 * writepages operation.  If it fails here we are freeing up the inode
	 * anyway and we'd rather not waste our time writing out stuff that is
	 * going to be truncated anyway.
	 */
	if (!igrab(inode))
		return 0;

3970 3971 3972 3973 3974 3975 3976 3977 3978
	pagevec_init(&pvec, 0);
	if (wbc->range_cyclic) {
		index = mapping->writeback_index; /* Start from prev offset */
		end = -1;
	} else {
		index = wbc->range_start >> PAGE_CACHE_SHIFT;
		end = wbc->range_end >> PAGE_CACHE_SHIFT;
		scanned = 1;
	}
3979 3980 3981 3982
	if (wbc->sync_mode == WB_SYNC_ALL)
		tag = PAGECACHE_TAG_TOWRITE;
	else
		tag = PAGECACHE_TAG_DIRTY;
3983
retry:
3984 3985
	if (wbc->sync_mode == WB_SYNC_ALL)
		tag_pages_for_writeback(mapping, index, end);
3986
	while (!done && !nr_to_write_done && (index <= end) &&
3987 3988
	       (nr_pages = pagevec_lookup_tag(&pvec, mapping, &index, tag,
			min(end - index, (pgoff_t)PAGEVEC_SIZE-1) + 1))) {
3989 3990 3991 3992 3993 3994 3995 3996 3997 3998 3999 4000 4001
		unsigned i;

		scanned = 1;
		for (i = 0; i < nr_pages; i++) {
			struct page *page = pvec.pages[i];

			/*
			 * At this point we hold neither mapping->tree_lock nor
			 * lock on the page itself: the page may be truncated or
			 * invalidated (changing page->mapping to NULL), or even
			 * swizzled back from swapper_space to tmpfs file
			 * mapping
			 */
4002 4003 4004
			if (!trylock_page(page)) {
				flush_fn(data);
				lock_page(page);
4005
			}
4006 4007 4008 4009 4010 4011 4012 4013 4014 4015 4016 4017

			if (unlikely(page->mapping != mapping)) {
				unlock_page(page);
				continue;
			}

			if (!wbc->range_cyclic && page->index > end) {
				done = 1;
				unlock_page(page);
				continue;
			}

C
Chris Mason 已提交
4018
			if (wbc->sync_mode != WB_SYNC_NONE) {
4019 4020
				if (PageWriteback(page))
					flush_fn(data);
4021
				wait_on_page_writeback(page);
C
Chris Mason 已提交
4022
			}
4023 4024 4025 4026 4027 4028 4029 4030 4031 4032 4033 4034 4035

			if (PageWriteback(page) ||
			    !clear_page_dirty_for_io(page)) {
				unlock_page(page);
				continue;
			}

			ret = (*writepage)(page, wbc, data);

			if (unlikely(ret == AOP_WRITEPAGE_ACTIVATE)) {
				unlock_page(page);
				ret = 0;
			}
4036 4037
			if (!err && ret < 0)
				err = ret;
4038 4039 4040 4041 4042 4043 4044

			/*
			 * the filesystem may choose to bump up nr_to_write.
			 * We have to make sure to honor the new nr_to_write
			 * at any time
			 */
			nr_to_write_done = wbc->nr_to_write <= 0;
4045 4046 4047 4048
		}
		pagevec_release(&pvec);
		cond_resched();
	}
4049
	if (!scanned && !done && !err) {
4050 4051 4052 4053 4054 4055 4056 4057
		/*
		 * We hit the last page and there is more work to be done: wrap
		 * back to the start of the file
		 */
		scanned = 1;
		index = 0;
		goto retry;
	}
4058
	btrfs_add_delayed_iput(inode);
4059
	return err;
4060 4061
}

4062
static void flush_epd_write_bio(struct extent_page_data *epd)
C
Chris Mason 已提交
4063 4064
{
	if (epd->bio) {
4065 4066 4067
		int rw = WRITE;
		int ret;

4068
		if (epd->sync_io)
4069 4070
			rw = WRITE_SYNC;

4071
		ret = submit_one_bio(rw, epd->bio, 0, epd->bio_flags);
4072
		BUG_ON(ret < 0); /* -ENOMEM */
C
Chris Mason 已提交
4073 4074 4075 4076
		epd->bio = NULL;
	}
}

4077 4078 4079 4080 4081 4082
static noinline void flush_write_bio(void *data)
{
	struct extent_page_data *epd = data;
	flush_epd_write_bio(epd);
}

4083 4084 4085 4086 4087 4088 4089 4090 4091
int extent_write_full_page(struct extent_io_tree *tree, struct page *page,
			  get_extent_t *get_extent,
			  struct writeback_control *wbc)
{
	int ret;
	struct extent_page_data epd = {
		.bio = NULL,
		.tree = tree,
		.get_extent = get_extent,
4092
		.extent_locked = 0,
4093
		.sync_io = wbc->sync_mode == WB_SYNC_ALL,
4094
		.bio_flags = 0,
4095 4096 4097 4098
	};

	ret = __extent_writepage(page, wbc, &epd);

4099
	flush_epd_write_bio(&epd);
4100 4101 4102
	return ret;
}

4103 4104 4105 4106 4107 4108 4109 4110 4111 4112 4113 4114 4115 4116 4117
int extent_write_locked_range(struct extent_io_tree *tree, struct inode *inode,
			      u64 start, u64 end, get_extent_t *get_extent,
			      int mode)
{
	int ret = 0;
	struct address_space *mapping = inode->i_mapping;
	struct page *page;
	unsigned long nr_pages = (end - start + PAGE_CACHE_SIZE) >>
		PAGE_CACHE_SHIFT;

	struct extent_page_data epd = {
		.bio = NULL,
		.tree = tree,
		.get_extent = get_extent,
		.extent_locked = 1,
4118
		.sync_io = mode == WB_SYNC_ALL,
4119
		.bio_flags = 0,
4120 4121 4122 4123 4124 4125 4126 4127
	};
	struct writeback_control wbc_writepages = {
		.sync_mode	= mode,
		.nr_to_write	= nr_pages * 2,
		.range_start	= start,
		.range_end	= end + 1,
	};

C
Chris Mason 已提交
4128
	while (start <= end) {
4129 4130 4131 4132 4133 4134 4135 4136 4137 4138 4139 4140 4141 4142
		page = find_get_page(mapping, start >> PAGE_CACHE_SHIFT);
		if (clear_page_dirty_for_io(page))
			ret = __extent_writepage(page, &wbc_writepages, &epd);
		else {
			if (tree->ops && tree->ops->writepage_end_io_hook)
				tree->ops->writepage_end_io_hook(page, start,
						 start + PAGE_CACHE_SIZE - 1,
						 NULL, 1);
			unlock_page(page);
		}
		page_cache_release(page);
		start += PAGE_CACHE_SIZE;
	}

4143
	flush_epd_write_bio(&epd);
4144 4145
	return ret;
}
4146 4147 4148 4149 4150 4151 4152 4153 4154 4155 4156

int extent_writepages(struct extent_io_tree *tree,
		      struct address_space *mapping,
		      get_extent_t *get_extent,
		      struct writeback_control *wbc)
{
	int ret = 0;
	struct extent_page_data epd = {
		.bio = NULL,
		.tree = tree,
		.get_extent = get_extent,
4157
		.extent_locked = 0,
4158
		.sync_io = wbc->sync_mode == WB_SYNC_ALL,
4159
		.bio_flags = 0,
4160 4161
	};

C
Chris Mason 已提交
4162
	ret = extent_write_cache_pages(tree, mapping, wbc,
C
Chris Mason 已提交
4163 4164
				       __extent_writepage, &epd,
				       flush_write_bio);
4165
	flush_epd_write_bio(&epd);
4166 4167 4168 4169 4170 4171 4172 4173 4174 4175
	return ret;
}

int extent_readpages(struct extent_io_tree *tree,
		     struct address_space *mapping,
		     struct list_head *pages, unsigned nr_pages,
		     get_extent_t get_extent)
{
	struct bio *bio = NULL;
	unsigned page_idx;
C
Chris Mason 已提交
4176
	unsigned long bio_flags = 0;
L
Liu Bo 已提交
4177 4178
	struct page *pagepool[16];
	struct page *page;
4179
	struct extent_map *em_cached = NULL;
L
Liu Bo 已提交
4180
	int nr = 0;
4181
	u64 prev_em_start = (u64)-1;
4182 4183

	for (page_idx = 0; page_idx < nr_pages; page_idx++) {
L
Liu Bo 已提交
4184
		page = list_entry(pages->prev, struct page, lru);
4185 4186 4187

		prefetchw(&page->flags);
		list_del(&page->lru);
L
Liu Bo 已提交
4188
		if (add_to_page_cache_lru(page, mapping,
4189
					page->index, GFP_NOFS)) {
L
Liu Bo 已提交
4190 4191
			page_cache_release(page);
			continue;
4192
		}
L
Liu Bo 已提交
4193 4194 4195 4196

		pagepool[nr++] = page;
		if (nr < ARRAY_SIZE(pagepool))
			continue;
4197
		__extent_readpages(tree, pagepool, nr, get_extent, &em_cached,
4198
				   &bio, 0, &bio_flags, READ, &prev_em_start);
L
Liu Bo 已提交
4199
		nr = 0;
4200
	}
4201
	if (nr)
4202
		__extent_readpages(tree, pagepool, nr, get_extent, &em_cached,
4203
				   &bio, 0, &bio_flags, READ, &prev_em_start);
L
Liu Bo 已提交
4204

4205 4206 4207
	if (em_cached)
		free_extent_map(em_cached);

4208 4209
	BUG_ON(!list_empty(pages));
	if (bio)
4210
		return submit_one_bio(READ, bio, 0, bio_flags);
4211 4212 4213 4214 4215 4216 4217 4218 4219 4220 4221
	return 0;
}

/*
 * basic invalidatepage code, this waits on any locked or writeback
 * ranges corresponding to the page, and then deletes any extent state
 * records from the tree
 */
int extent_invalidatepage(struct extent_io_tree *tree,
			  struct page *page, unsigned long offset)
{
4222
	struct extent_state *cached_state = NULL;
M
Miao Xie 已提交
4223
	u64 start = page_offset(page);
4224 4225 4226
	u64 end = start + PAGE_CACHE_SIZE - 1;
	size_t blocksize = page->mapping->host->i_sb->s_blocksize;

4227
	start += ALIGN(offset, blocksize);
4228 4229 4230
	if (start > end)
		return 0;

4231
	lock_extent_bits(tree, start, end, &cached_state);
4232
	wait_on_page_writeback(page);
4233
	clear_extent_bit(tree, start, end,
4234 4235
			 EXTENT_LOCKED | EXTENT_DIRTY | EXTENT_DELALLOC |
			 EXTENT_DO_ACCOUNTING,
4236
			 1, 1, &cached_state, GFP_NOFS);
4237 4238 4239
	return 0;
}

4240 4241 4242 4243 4244
/*
 * a helper for releasepage, this tests for areas of the page that
 * are locked or under IO and drops the related state bits if it is safe
 * to drop the page.
 */
4245 4246 4247
static int try_release_extent_state(struct extent_map_tree *map,
				    struct extent_io_tree *tree,
				    struct page *page, gfp_t mask)
4248
{
M
Miao Xie 已提交
4249
	u64 start = page_offset(page);
4250 4251 4252
	u64 end = start + PAGE_CACHE_SIZE - 1;
	int ret = 1;

4253
	if (test_range_bit(tree, start, end,
4254
			   EXTENT_IOBITS, 0, NULL))
4255 4256 4257 4258
		ret = 0;
	else {
		if ((mask & GFP_NOFS) == GFP_NOFS)
			mask = GFP_NOFS;
4259 4260 4261 4262
		/*
		 * at this point we can safely clear everything except the
		 * locked bit and the nodatasum bit
		 */
4263
		ret = clear_extent_bit(tree, start, end,
4264 4265
				 ~(EXTENT_LOCKED | EXTENT_NODATASUM),
				 0, 0, NULL, mask);
4266 4267 4268 4269 4270 4271 4272 4273

		/* if clear_extent_bit failed for enomem reasons,
		 * we can't allow the release to continue.
		 */
		if (ret < 0)
			ret = 0;
		else
			ret = 1;
4274 4275 4276 4277
	}
	return ret;
}

4278 4279 4280 4281 4282 4283
/*
 * a helper for releasepage.  As long as there are no locked extents
 * in the range corresponding to the page, both state records and extent
 * map records are removed
 */
int try_release_extent_mapping(struct extent_map_tree *map,
4284 4285
			       struct extent_io_tree *tree, struct page *page,
			       gfp_t mask)
4286 4287
{
	struct extent_map *em;
M
Miao Xie 已提交
4288
	u64 start = page_offset(page);
4289
	u64 end = start + PAGE_CACHE_SIZE - 1;
4290

4291
	if (gfpflags_allow_blocking(mask) &&
4292
	    page->mapping->host->i_size > SZ_16M) {
4293
		u64 len;
4294
		while (start <= end) {
4295
			len = end - start + 1;
4296
			write_lock(&map->lock);
4297
			em = lookup_extent_mapping(map, start, len);
4298
			if (!em) {
4299
				write_unlock(&map->lock);
4300 4301
				break;
			}
4302 4303
			if (test_bit(EXTENT_FLAG_PINNED, &em->flags) ||
			    em->start != start) {
4304
				write_unlock(&map->lock);
4305 4306 4307 4308 4309
				free_extent_map(em);
				break;
			}
			if (!test_range_bit(tree, em->start,
					    extent_map_end(em) - 1,
4310
					    EXTENT_LOCKED | EXTENT_WRITEBACK,
4311
					    0, NULL)) {
4312 4313 4314 4315 4316
				remove_extent_mapping(map, em);
				/* once for the rb tree */
				free_extent_map(em);
			}
			start = extent_map_end(em);
4317
			write_unlock(&map->lock);
4318 4319

			/* once for us */
4320 4321 4322
			free_extent_map(em);
		}
	}
4323
	return try_release_extent_state(map, tree, page, mask);
4324 4325
}

4326 4327 4328 4329 4330 4331 4332 4333 4334 4335 4336 4337 4338 4339 4340 4341
/*
 * helper function for fiemap, which doesn't want to see any holes.
 * This maps until we find something past 'last'
 */
static struct extent_map *get_extent_skip_holes(struct inode *inode,
						u64 offset,
						u64 last,
						get_extent_t *get_extent)
{
	u64 sectorsize = BTRFS_I(inode)->root->sectorsize;
	struct extent_map *em;
	u64 len;

	if (offset >= last)
		return NULL;

4342
	while (1) {
4343 4344 4345
		len = last - offset;
		if (len == 0)
			break;
4346
		len = ALIGN(len, sectorsize);
4347
		em = get_extent(inode, NULL, 0, offset, len, 0);
4348
		if (IS_ERR_OR_NULL(em))
4349 4350 4351 4352 4353 4354 4355 4356 4357 4358 4359 4360 4361 4362 4363 4364 4365
			return em;

		/* if this isn't a hole return it */
		if (!test_bit(EXTENT_FLAG_VACANCY, &em->flags) &&
		    em->block_start != EXTENT_MAP_HOLE) {
			return em;
		}

		/* this is a hole, advance to the next extent */
		offset = extent_map_end(em);
		free_extent_map(em);
		if (offset >= last)
			break;
	}
	return NULL;
}

Y
Yehuda Sadeh 已提交
4366 4367 4368
int extent_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
		__u64 start, __u64 len, get_extent_t *get_extent)
{
J
Josef Bacik 已提交
4369
	int ret = 0;
Y
Yehuda Sadeh 已提交
4370 4371 4372
	u64 off = start;
	u64 max = start + len;
	u32 flags = 0;
J
Josef Bacik 已提交
4373 4374
	u32 found_type;
	u64 last;
4375
	u64 last_for_get_extent = 0;
Y
Yehuda Sadeh 已提交
4376
	u64 disko = 0;
4377
	u64 isize = i_size_read(inode);
J
Josef Bacik 已提交
4378
	struct btrfs_key found_key;
Y
Yehuda Sadeh 已提交
4379
	struct extent_map *em = NULL;
4380
	struct extent_state *cached_state = NULL;
J
Josef Bacik 已提交
4381
	struct btrfs_path *path;
4382
	struct btrfs_root *root = BTRFS_I(inode)->root;
Y
Yehuda Sadeh 已提交
4383
	int end = 0;
4384 4385 4386
	u64 em_start = 0;
	u64 em_len = 0;
	u64 em_end = 0;
Y
Yehuda Sadeh 已提交
4387 4388 4389 4390

	if (len == 0)
		return -EINVAL;

J
Josef Bacik 已提交
4391 4392 4393 4394 4395
	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;
	path->leave_spinning = 1;

4396 4397
	start = round_down(start, BTRFS_I(inode)->root->sectorsize);
	len = round_up(max, BTRFS_I(inode)->root->sectorsize) - start;
4398

4399 4400 4401 4402
	/*
	 * lookup the last file extent.  We're not using i_size here
	 * because there might be preallocation past i_size
	 */
4403 4404
	ret = btrfs_lookup_file_extent(NULL, root, path, btrfs_ino(inode), -1,
				       0);
J
Josef Bacik 已提交
4405 4406 4407 4408 4409 4410 4411
	if (ret < 0) {
		btrfs_free_path(path);
		return ret;
	}
	WARN_ON(!ret);
	path->slots[0]--;
	btrfs_item_key_to_cpu(path->nodes[0], &found_key, path->slots[0]);
4412
	found_type = found_key.type;
J
Josef Bacik 已提交
4413

4414
	/* No extents, but there might be delalloc bits */
L
Li Zefan 已提交
4415
	if (found_key.objectid != btrfs_ino(inode) ||
J
Josef Bacik 已提交
4416
	    found_type != BTRFS_EXTENT_DATA_KEY) {
4417 4418 4419 4420 4421 4422 4423 4424 4425 4426 4427
		/* have to trust i_size as the end */
		last = (u64)-1;
		last_for_get_extent = isize;
	} else {
		/*
		 * remember the start of the last extent.  There are a
		 * bunch of different factors that go into the length of the
		 * extent, so its much less complex to remember where it started
		 */
		last = found_key.offset;
		last_for_get_extent = last + 1;
J
Josef Bacik 已提交
4428
	}
4429
	btrfs_release_path(path);
J
Josef Bacik 已提交
4430

4431 4432 4433 4434 4435 4436 4437 4438 4439 4440
	/*
	 * we might have some extents allocated but more delalloc past those
	 * extents.  so, we trust isize unless the start of the last extent is
	 * beyond isize
	 */
	if (last < isize) {
		last = (u64)-1;
		last_for_get_extent = isize;
	}

4441
	lock_extent_bits(&BTRFS_I(inode)->io_tree, start, start + len - 1,
4442
			 &cached_state);
4443

4444
	em = get_extent_skip_holes(inode, start, last_for_get_extent,
4445
				   get_extent);
Y
Yehuda Sadeh 已提交
4446 4447 4448 4449 4450 4451
	if (!em)
		goto out;
	if (IS_ERR(em)) {
		ret = PTR_ERR(em);
		goto out;
	}
J
Josef Bacik 已提交
4452

Y
Yehuda Sadeh 已提交
4453
	while (!end) {
4454
		u64 offset_in_extent = 0;
4455 4456 4457 4458 4459 4460 4461 4462 4463 4464 4465 4466

		/* break if the extent we found is outside the range */
		if (em->start >= max || extent_map_end(em) < off)
			break;

		/*
		 * get_extent may return an extent that starts before our
		 * requested range.  We have to make sure the ranges
		 * we return to fiemap always move forward and don't
		 * overlap, so adjust the offsets here
		 */
		em_start = max(em->start, off);
Y
Yehuda Sadeh 已提交
4467

4468 4469
		/*
		 * record the offset from the start of the extent
4470 4471 4472
		 * for adjusting the disk offset below.  Only do this if the
		 * extent isn't compressed since our in ram offset may be past
		 * what we have actually allocated on disk.
4473
		 */
4474 4475
		if (!test_bit(EXTENT_FLAG_COMPRESSED, &em->flags))
			offset_in_extent = em_start - em->start;
4476
		em_end = extent_map_end(em);
4477
		em_len = em_end - em_start;
Y
Yehuda Sadeh 已提交
4478 4479 4480
		disko = 0;
		flags = 0;

4481 4482 4483 4484 4485 4486 4487
		/*
		 * bump off for our next call to get_extent
		 */
		off = extent_map_end(em);
		if (off >= max)
			end = 1;

4488
		if (em->block_start == EXTENT_MAP_LAST_BYTE) {
Y
Yehuda Sadeh 已提交
4489 4490
			end = 1;
			flags |= FIEMAP_EXTENT_LAST;
4491
		} else if (em->block_start == EXTENT_MAP_INLINE) {
Y
Yehuda Sadeh 已提交
4492 4493
			flags |= (FIEMAP_EXTENT_DATA_INLINE |
				  FIEMAP_EXTENT_NOT_ALIGNED);
4494
		} else if (em->block_start == EXTENT_MAP_DELALLOC) {
Y
Yehuda Sadeh 已提交
4495 4496
			flags |= (FIEMAP_EXTENT_DELALLOC |
				  FIEMAP_EXTENT_UNKNOWN);
4497 4498 4499
		} else if (fieinfo->fi_extents_max) {
			u64 bytenr = em->block_start -
				(em->start - em->orig_start);
4500

4501
			disko = em->block_start + offset_in_extent;
4502 4503 4504 4505

			/*
			 * As btrfs supports shared space, this information
			 * can be exported to userspace tools via
4506 4507 4508
			 * flag FIEMAP_EXTENT_SHARED.  If fi_extents_max == 0
			 * then we're just getting a count and we can skip the
			 * lookup stuff.
4509
			 */
4510 4511 4512 4513
			ret = btrfs_check_shared(NULL, root->fs_info,
						 root->objectid,
						 btrfs_ino(inode), bytenr);
			if (ret < 0)
4514
				goto out_free;
4515
			if (ret)
4516
				flags |= FIEMAP_EXTENT_SHARED;
4517
			ret = 0;
Y
Yehuda Sadeh 已提交
4518 4519 4520
		}
		if (test_bit(EXTENT_FLAG_COMPRESSED, &em->flags))
			flags |= FIEMAP_EXTENT_ENCODED;
4521 4522
		if (test_bit(EXTENT_FLAG_PREALLOC, &em->flags))
			flags |= FIEMAP_EXTENT_UNWRITTEN;
Y
Yehuda Sadeh 已提交
4523 4524 4525

		free_extent_map(em);
		em = NULL;
4526 4527
		if ((em_start >= last) || em_len == (u64)-1 ||
		   (last == (u64)-1 && isize <= em_end)) {
Y
Yehuda Sadeh 已提交
4528 4529 4530 4531
			flags |= FIEMAP_EXTENT_LAST;
			end = 1;
		}

4532 4533 4534 4535 4536 4537 4538 4539
		/* now scan forward to see if this is really the last extent. */
		em = get_extent_skip_holes(inode, off, last_for_get_extent,
					   get_extent);
		if (IS_ERR(em)) {
			ret = PTR_ERR(em);
			goto out;
		}
		if (!em) {
J
Josef Bacik 已提交
4540 4541 4542
			flags |= FIEMAP_EXTENT_LAST;
			end = 1;
		}
4543 4544
		ret = fiemap_fill_next_extent(fieinfo, em_start, disko,
					      em_len, flags);
4545 4546 4547
		if (ret) {
			if (ret == 1)
				ret = 0;
4548
			goto out_free;
4549
		}
Y
Yehuda Sadeh 已提交
4550 4551 4552 4553
	}
out_free:
	free_extent_map(em);
out:
4554
	btrfs_free_path(path);
L
Liu Bo 已提交
4555
	unlock_extent_cached(&BTRFS_I(inode)->io_tree, start, start + len - 1,
4556
			     &cached_state, GFP_NOFS);
Y
Yehuda Sadeh 已提交
4557 4558 4559
	return ret;
}

4560 4561
static void __free_extent_buffer(struct extent_buffer *eb)
{
4562
	btrfs_leak_debug_del(&eb->leak_list);
4563 4564 4565
	kmem_cache_free(extent_buffer_cache, eb);
}

4566
int extent_buffer_under_io(struct extent_buffer *eb)
4567 4568 4569 4570 4571 4572 4573 4574 4575
{
	return (atomic_read(&eb->io_pages) ||
		test_bit(EXTENT_BUFFER_WRITEBACK, &eb->bflags) ||
		test_bit(EXTENT_BUFFER_DIRTY, &eb->bflags));
}

/*
 * Helper for releasing extent buffer page.
 */
4576
static void btrfs_release_extent_buffer_page(struct extent_buffer *eb)
4577 4578 4579 4580 4581 4582 4583
{
	unsigned long index;
	struct page *page;
	int mapped = !test_bit(EXTENT_BUFFER_DUMMY, &eb->bflags);

	BUG_ON(extent_buffer_under_io(eb));

4584 4585
	index = num_extent_pages(eb->start, eb->len);
	if (index == 0)
4586 4587 4588 4589
		return;

	do {
		index--;
4590
		page = eb->pages[index];
4591 4592 4593
		if (!page)
			continue;
		if (mapped)
4594
			spin_lock(&page->mapping->private_lock);
4595 4596 4597 4598 4599 4600 4601 4602 4603 4604 4605 4606
		/*
		 * We do this since we'll remove the pages after we've
		 * removed the eb from the radix tree, so we could race
		 * and have this page now attached to the new eb.  So
		 * only clear page_private if it's still connected to
		 * this eb.
		 */
		if (PagePrivate(page) &&
		    page->private == (unsigned long)eb) {
			BUG_ON(test_bit(EXTENT_BUFFER_DIRTY, &eb->bflags));
			BUG_ON(PageDirty(page));
			BUG_ON(PageWriteback(page));
4607
			/*
4608 4609
			 * We need to make sure we haven't be attached
			 * to a new eb.
4610
			 */
4611 4612 4613
			ClearPagePrivate(page);
			set_page_private(page, 0);
			/* One for the page private */
4614 4615
			page_cache_release(page);
		}
4616 4617 4618 4619 4620 4621

		if (mapped)
			spin_unlock(&page->mapping->private_lock);

		/* One for when we alloced the page */
		page_cache_release(page);
4622
	} while (index != 0);
4623 4624 4625 4626 4627 4628 4629
}

/*
 * Helper for releasing the extent buffer.
 */
static inline void btrfs_release_extent_buffer(struct extent_buffer *eb)
{
4630
	btrfs_release_extent_buffer_page(eb);
4631 4632 4633
	__free_extent_buffer(eb);
}

4634 4635
static struct extent_buffer *
__alloc_extent_buffer(struct btrfs_fs_info *fs_info, u64 start,
4636
		      unsigned long len)
4637 4638 4639
{
	struct extent_buffer *eb = NULL;

4640
	eb = kmem_cache_zalloc(extent_buffer_cache, GFP_NOFS|__GFP_NOFAIL);
4641 4642
	eb->start = start;
	eb->len = len;
4643
	eb->fs_info = fs_info;
4644
	eb->bflags = 0;
4645 4646 4647 4648 4649 4650 4651
	rwlock_init(&eb->lock);
	atomic_set(&eb->write_locks, 0);
	atomic_set(&eb->read_locks, 0);
	atomic_set(&eb->blocking_readers, 0);
	atomic_set(&eb->blocking_writers, 0);
	atomic_set(&eb->spinning_readers, 0);
	atomic_set(&eb->spinning_writers, 0);
4652
	eb->lock_nested = 0;
4653 4654
	init_waitqueue_head(&eb->write_lock_wq);
	init_waitqueue_head(&eb->read_lock_wq);
4655

4656 4657
	btrfs_leak_debug_add(&eb->leak_list, &buffers);

4658
	spin_lock_init(&eb->refs_lock);
4659
	atomic_set(&eb->refs, 1);
4660
	atomic_set(&eb->io_pages, 0);
4661

4662 4663 4664 4665 4666 4667
	/*
	 * Sanity checks, currently the maximum is 64k covered by 16x 4k pages
	 */
	BUILD_BUG_ON(BTRFS_MAX_METADATA_BLOCKSIZE
		> MAX_INLINE_EXTENT_BUFFER_SIZE);
	BUG_ON(len > MAX_INLINE_EXTENT_BUFFER_SIZE);
4668 4669 4670 4671

	return eb;
}

4672 4673 4674 4675 4676 4677 4678
struct extent_buffer *btrfs_clone_extent_buffer(struct extent_buffer *src)
{
	unsigned long i;
	struct page *p;
	struct extent_buffer *new;
	unsigned long num_pages = num_extent_pages(src->start, src->len);

4679
	new = __alloc_extent_buffer(src->fs_info, src->start, src->len);
4680 4681 4682 4683
	if (new == NULL)
		return NULL;

	for (i = 0; i < num_pages; i++) {
4684
		p = alloc_page(GFP_NOFS);
4685 4686 4687 4688
		if (!p) {
			btrfs_release_extent_buffer(new);
			return NULL;
		}
4689 4690 4691 4692 4693 4694 4695 4696 4697 4698 4699 4700 4701
		attach_extent_buffer_page(new, p);
		WARN_ON(PageDirty(p));
		SetPageUptodate(p);
		new->pages[i] = p;
	}

	copy_extent_buffer(new, src, 0, 0, src->len);
	set_bit(EXTENT_BUFFER_UPTODATE, &new->bflags);
	set_bit(EXTENT_BUFFER_DUMMY, &new->bflags);

	return new;
}

4702 4703
struct extent_buffer *__alloc_dummy_extent_buffer(struct btrfs_fs_info *fs_info,
						  u64 start, unsigned long len)
4704 4705
{
	struct extent_buffer *eb;
4706
	unsigned long num_pages;
4707 4708
	unsigned long i;

4709
	num_pages = num_extent_pages(start, len);
4710 4711

	eb = __alloc_extent_buffer(fs_info, start, len);
4712 4713 4714 4715
	if (!eb)
		return NULL;

	for (i = 0; i < num_pages; i++) {
4716
		eb->pages[i] = alloc_page(GFP_NOFS);
4717 4718 4719 4720 4721 4722 4723 4724 4725
		if (!eb->pages[i])
			goto err;
	}
	set_extent_buffer_uptodate(eb);
	btrfs_set_header_nritems(eb, 0);
	set_bit(EXTENT_BUFFER_DUMMY, &eb->bflags);

	return eb;
err:
4726 4727
	for (; i > 0; i--)
		__free_page(eb->pages[i - 1]);
4728 4729 4730 4731
	__free_extent_buffer(eb);
	return NULL;
}

4732 4733 4734 4735 4736 4737 4738 4739 4740 4741 4742 4743 4744 4745 4746 4747 4748 4749
struct extent_buffer *alloc_dummy_extent_buffer(struct btrfs_fs_info *fs_info,
						u64 start)
{
	unsigned long len;

	if (!fs_info) {
		/*
		 * Called only from tests that don't always have a fs_info
		 * available, but we know that nodesize is 4096
		 */
		len = 4096;
	} else {
		len = fs_info->tree_root->nodesize;
	}

	return __alloc_dummy_extent_buffer(fs_info, start, len);
}

4750 4751
static void check_buffer_tree_ref(struct extent_buffer *eb)
{
4752
	int refs;
4753 4754 4755 4756 4757 4758 4759 4760 4761 4762 4763 4764 4765 4766 4767 4768 4769 4770 4771 4772
	/* the ref bit is tricky.  We have to make sure it is set
	 * if we have the buffer dirty.   Otherwise the
	 * code to free a buffer can end up dropping a dirty
	 * page
	 *
	 * Once the ref bit is set, it won't go away while the
	 * buffer is dirty or in writeback, and it also won't
	 * go away while we have the reference count on the
	 * eb bumped.
	 *
	 * We can't just set the ref bit without bumping the
	 * ref on the eb because free_extent_buffer might
	 * see the ref bit and try to clear it.  If this happens
	 * free_extent_buffer might end up dropping our original
	 * ref by mistake and freeing the page before we are able
	 * to add one more ref.
	 *
	 * So bump the ref count first, then set the bit.  If someone
	 * beat us to it, drop the ref we added.
	 */
4773 4774 4775 4776
	refs = atomic_read(&eb->refs);
	if (refs >= 2 && test_bit(EXTENT_BUFFER_TREE_REF, &eb->bflags))
		return;

4777 4778
	spin_lock(&eb->refs_lock);
	if (!test_and_set_bit(EXTENT_BUFFER_TREE_REF, &eb->bflags))
4779
		atomic_inc(&eb->refs);
4780
	spin_unlock(&eb->refs_lock);
4781 4782
}

4783 4784
static void mark_extent_buffer_accessed(struct extent_buffer *eb,
		struct page *accessed)
4785 4786 4787
{
	unsigned long num_pages, i;

4788 4789
	check_buffer_tree_ref(eb);

4790 4791
	num_pages = num_extent_pages(eb->start, eb->len);
	for (i = 0; i < num_pages; i++) {
4792 4793
		struct page *p = eb->pages[i];

4794 4795
		if (p != accessed)
			mark_page_accessed(p);
4796 4797 4798
	}
}

4799 4800
struct extent_buffer *find_extent_buffer(struct btrfs_fs_info *fs_info,
					 u64 start)
4801 4802 4803 4804
{
	struct extent_buffer *eb;

	rcu_read_lock();
4805 4806
	eb = radix_tree_lookup(&fs_info->buffer_radix,
			       start >> PAGE_CACHE_SHIFT);
4807 4808
	if (eb && atomic_inc_not_zero(&eb->refs)) {
		rcu_read_unlock();
4809 4810 4811 4812 4813 4814 4815 4816 4817 4818 4819 4820 4821 4822 4823 4824 4825 4826 4827
		/*
		 * Lock our eb's refs_lock to avoid races with
		 * free_extent_buffer. When we get our eb it might be flagged
		 * with EXTENT_BUFFER_STALE and another task running
		 * free_extent_buffer might have seen that flag set,
		 * eb->refs == 2, that the buffer isn't under IO (dirty and
		 * writeback flags not set) and it's still in the tree (flag
		 * EXTENT_BUFFER_TREE_REF set), therefore being in the process
		 * of decrementing the extent buffer's reference count twice.
		 * So here we could race and increment the eb's reference count,
		 * clear its stale flag, mark it as dirty and drop our reference
		 * before the other task finishes executing free_extent_buffer,
		 * which would later result in an attempt to free an extent
		 * buffer that is dirty.
		 */
		if (test_bit(EXTENT_BUFFER_STALE, &eb->bflags)) {
			spin_lock(&eb->refs_lock);
			spin_unlock(&eb->refs_lock);
		}
4828
		mark_extent_buffer_accessed(eb, NULL);
4829 4830 4831 4832 4833 4834 4835
		return eb;
	}
	rcu_read_unlock();

	return NULL;
}

4836 4837
#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
struct extent_buffer *alloc_test_extent_buffer(struct btrfs_fs_info *fs_info,
4838
					       u64 start)
4839 4840 4841 4842 4843 4844 4845
{
	struct extent_buffer *eb, *exists = NULL;
	int ret;

	eb = find_extent_buffer(fs_info, start);
	if (eb)
		return eb;
4846
	eb = alloc_dummy_extent_buffer(fs_info, start);
4847 4848 4849 4850 4851 4852 4853 4854 4855 4856 4857 4858 4859 4860 4861 4862 4863 4864 4865 4866 4867 4868 4869 4870 4871 4872 4873 4874 4875 4876 4877 4878 4879 4880 4881 4882
	if (!eb)
		return NULL;
	eb->fs_info = fs_info;
again:
	ret = radix_tree_preload(GFP_NOFS & ~__GFP_HIGHMEM);
	if (ret)
		goto free_eb;
	spin_lock(&fs_info->buffer_lock);
	ret = radix_tree_insert(&fs_info->buffer_radix,
				start >> PAGE_CACHE_SHIFT, eb);
	spin_unlock(&fs_info->buffer_lock);
	radix_tree_preload_end();
	if (ret == -EEXIST) {
		exists = find_extent_buffer(fs_info, start);
		if (exists)
			goto free_eb;
		else
			goto again;
	}
	check_buffer_tree_ref(eb);
	set_bit(EXTENT_BUFFER_IN_TREE, &eb->bflags);

	/*
	 * We will free dummy extent buffer's if they come into
	 * free_extent_buffer with a ref count of 2, but if we are using this we
	 * want the buffers to stay in memory until we're done with them, so
	 * bump the ref count again.
	 */
	atomic_inc(&eb->refs);
	return eb;
free_eb:
	btrfs_release_extent_buffer(eb);
	return exists;
}
#endif

4883
struct extent_buffer *alloc_extent_buffer(struct btrfs_fs_info *fs_info,
4884
					  u64 start)
4885
{
4886
	unsigned long len = fs_info->tree_root->nodesize;
4887 4888 4889 4890
	unsigned long num_pages = num_extent_pages(start, len);
	unsigned long i;
	unsigned long index = start >> PAGE_CACHE_SHIFT;
	struct extent_buffer *eb;
4891
	struct extent_buffer *exists = NULL;
4892
	struct page *p;
4893
	struct address_space *mapping = fs_info->btree_inode->i_mapping;
4894
	int uptodate = 1;
4895
	int ret;
4896

4897
	eb = find_extent_buffer(fs_info, start);
4898
	if (eb)
4899 4900
		return eb;

4901
	eb = __alloc_extent_buffer(fs_info, start, len);
4902
	if (!eb)
4903 4904
		return NULL;

4905
	for (i = 0; i < num_pages; i++, index++) {
4906
		p = find_or_create_page(mapping, index, GFP_NOFS|__GFP_NOFAIL);
4907
		if (!p)
4908
			goto free_eb;
J
Josef Bacik 已提交
4909 4910 4911 4912 4913 4914 4915 4916 4917 4918 4919 4920 4921 4922

		spin_lock(&mapping->private_lock);
		if (PagePrivate(p)) {
			/*
			 * We could have already allocated an eb for this page
			 * and attached one so lets see if we can get a ref on
			 * the existing eb, and if we can we know it's good and
			 * we can just return that one, else we know we can just
			 * overwrite page->private.
			 */
			exists = (struct extent_buffer *)p->private;
			if (atomic_inc_not_zero(&exists->refs)) {
				spin_unlock(&mapping->private_lock);
				unlock_page(p);
4923
				page_cache_release(p);
4924
				mark_extent_buffer_accessed(exists, p);
J
Josef Bacik 已提交
4925 4926
				goto free_eb;
			}
4927
			exists = NULL;
J
Josef Bacik 已提交
4928

4929
			/*
J
Josef Bacik 已提交
4930 4931 4932 4933
			 * Do this so attach doesn't complain and we need to
			 * drop the ref the old guy had.
			 */
			ClearPagePrivate(p);
4934
			WARN_ON(PageDirty(p));
J
Josef Bacik 已提交
4935
			page_cache_release(p);
4936
		}
J
Josef Bacik 已提交
4937 4938
		attach_extent_buffer_page(eb, p);
		spin_unlock(&mapping->private_lock);
4939
		WARN_ON(PageDirty(p));
4940
		eb->pages[i] = p;
4941 4942
		if (!PageUptodate(p))
			uptodate = 0;
C
Chris Mason 已提交
4943 4944 4945 4946 4947

		/*
		 * see below about how we avoid a nasty race with release page
		 * and why we unlock later
		 */
4948 4949
	}
	if (uptodate)
4950
		set_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags);
4951
again:
4952 4953 4954 4955
	ret = radix_tree_preload(GFP_NOFS & ~__GFP_HIGHMEM);
	if (ret)
		goto free_eb;

4956 4957 4958 4959
	spin_lock(&fs_info->buffer_lock);
	ret = radix_tree_insert(&fs_info->buffer_radix,
				start >> PAGE_CACHE_SHIFT, eb);
	spin_unlock(&fs_info->buffer_lock);
4960
	radix_tree_preload_end();
4961
	if (ret == -EEXIST) {
4962
		exists = find_extent_buffer(fs_info, start);
4963 4964 4965
		if (exists)
			goto free_eb;
		else
4966
			goto again;
4967 4968
	}
	/* add one reference for the tree */
4969
	check_buffer_tree_ref(eb);
4970
	set_bit(EXTENT_BUFFER_IN_TREE, &eb->bflags);
C
Chris Mason 已提交
4971 4972 4973 4974 4975 4976 4977 4978 4979 4980

	/*
	 * there is a race where release page may have
	 * tried to find this extent buffer in the radix
	 * but failed.  It will tell the VM it is safe to
	 * reclaim the, and it will clear the page private bit.
	 * We must make sure to set the page private bit properly
	 * after the extent buffer is in the radix tree so
	 * it doesn't get lost
	 */
4981 4982
	SetPageChecked(eb->pages[0]);
	for (i = 1; i < num_pages; i++) {
4983
		p = eb->pages[i];
4984 4985 4986 4987
		ClearPageChecked(p);
		unlock_page(p);
	}
	unlock_page(eb->pages[0]);
4988 4989
	return eb;

4990
free_eb:
4991
	WARN_ON(!atomic_dec_and_test(&eb->refs));
4992 4993 4994 4995
	for (i = 0; i < num_pages; i++) {
		if (eb->pages[i])
			unlock_page(eb->pages[i]);
	}
C
Chris Mason 已提交
4996

4997
	btrfs_release_extent_buffer(eb);
4998
	return exists;
4999 5000
}

5001 5002 5003 5004 5005 5006 5007 5008 5009
static inline void btrfs_release_extent_buffer_rcu(struct rcu_head *head)
{
	struct extent_buffer *eb =
			container_of(head, struct extent_buffer, rcu_head);

	__free_extent_buffer(eb);
}

/* Expects to have eb->eb_lock already held */
5010
static int release_extent_buffer(struct extent_buffer *eb)
5011 5012 5013
{
	WARN_ON(atomic_read(&eb->refs) == 0);
	if (atomic_dec_and_test(&eb->refs)) {
5014
		if (test_and_clear_bit(EXTENT_BUFFER_IN_TREE, &eb->bflags)) {
5015
			struct btrfs_fs_info *fs_info = eb->fs_info;
5016

5017
			spin_unlock(&eb->refs_lock);
5018

5019 5020
			spin_lock(&fs_info->buffer_lock);
			radix_tree_delete(&fs_info->buffer_radix,
5021
					  eb->start >> PAGE_CACHE_SHIFT);
5022
			spin_unlock(&fs_info->buffer_lock);
5023 5024
		} else {
			spin_unlock(&eb->refs_lock);
5025
		}
5026 5027

		/* Should be safe to release our pages at this point */
5028
		btrfs_release_extent_buffer_page(eb);
5029 5030 5031 5032 5033 5034
#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
		if (unlikely(test_bit(EXTENT_BUFFER_DUMMY, &eb->bflags))) {
			__free_extent_buffer(eb);
			return 1;
		}
#endif
5035
		call_rcu(&eb->rcu_head, btrfs_release_extent_buffer_rcu);
5036
		return 1;
5037 5038
	}
	spin_unlock(&eb->refs_lock);
5039 5040

	return 0;
5041 5042
}

5043 5044
void free_extent_buffer(struct extent_buffer *eb)
{
5045 5046
	int refs;
	int old;
5047 5048 5049
	if (!eb)
		return;

5050 5051 5052 5053 5054 5055 5056 5057 5058
	while (1) {
		refs = atomic_read(&eb->refs);
		if (refs <= 3)
			break;
		old = atomic_cmpxchg(&eb->refs, refs, refs - 1);
		if (old == refs)
			return;
	}

5059
	spin_lock(&eb->refs_lock);
5060 5061 5062 5063
	if (atomic_read(&eb->refs) == 2 &&
	    test_bit(EXTENT_BUFFER_DUMMY, &eb->bflags))
		atomic_dec(&eb->refs);

5064 5065
	if (atomic_read(&eb->refs) == 2 &&
	    test_bit(EXTENT_BUFFER_STALE, &eb->bflags) &&
5066
	    !extent_buffer_under_io(eb) &&
5067 5068 5069 5070 5071 5072 5073
	    test_and_clear_bit(EXTENT_BUFFER_TREE_REF, &eb->bflags))
		atomic_dec(&eb->refs);

	/*
	 * I know this is terrible, but it's temporary until we stop tracking
	 * the uptodate bits and such for the extent buffers.
	 */
5074
	release_extent_buffer(eb);
5075 5076 5077 5078 5079
}

void free_extent_buffer_stale(struct extent_buffer *eb)
{
	if (!eb)
5080 5081
		return;

5082 5083 5084
	spin_lock(&eb->refs_lock);
	set_bit(EXTENT_BUFFER_STALE, &eb->bflags);

5085
	if (atomic_read(&eb->refs) == 2 && !extent_buffer_under_io(eb) &&
5086 5087
	    test_and_clear_bit(EXTENT_BUFFER_TREE_REF, &eb->bflags))
		atomic_dec(&eb->refs);
5088
	release_extent_buffer(eb);
5089 5090
}

5091
void clear_extent_buffer_dirty(struct extent_buffer *eb)
5092 5093 5094 5095 5096 5097 5098 5099
{
	unsigned long i;
	unsigned long num_pages;
	struct page *page;

	num_pages = num_extent_pages(eb->start, eb->len);

	for (i = 0; i < num_pages; i++) {
5100
		page = eb->pages[i];
5101
		if (!PageDirty(page))
C
Chris Mason 已提交
5102 5103
			continue;

5104
		lock_page(page);
C
Chris Mason 已提交
5105 5106
		WARN_ON(!PagePrivate(page));

5107
		clear_page_dirty_for_io(page);
5108
		spin_lock_irq(&page->mapping->tree_lock);
5109 5110 5111 5112 5113
		if (!PageDirty(page)) {
			radix_tree_tag_clear(&page->mapping->page_tree,
						page_index(page),
						PAGECACHE_TAG_DIRTY);
		}
5114
		spin_unlock_irq(&page->mapping->tree_lock);
5115
		ClearPageError(page);
5116
		unlock_page(page);
5117
	}
5118
	WARN_ON(atomic_read(&eb->refs) == 0);
5119 5120
}

5121
int set_extent_buffer_dirty(struct extent_buffer *eb)
5122 5123 5124
{
	unsigned long i;
	unsigned long num_pages;
5125
	int was_dirty = 0;
5126

5127 5128
	check_buffer_tree_ref(eb);

5129
	was_dirty = test_and_set_bit(EXTENT_BUFFER_DIRTY, &eb->bflags);
5130

5131
	num_pages = num_extent_pages(eb->start, eb->len);
5132
	WARN_ON(atomic_read(&eb->refs) == 0);
5133 5134
	WARN_ON(!test_bit(EXTENT_BUFFER_TREE_REF, &eb->bflags));

5135
	for (i = 0; i < num_pages; i++)
5136
		set_page_dirty(eb->pages[i]);
5137
	return was_dirty;
5138 5139
}

5140
void clear_extent_buffer_uptodate(struct extent_buffer *eb)
5141 5142 5143 5144 5145
{
	unsigned long i;
	struct page *page;
	unsigned long num_pages;

5146
	clear_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags);
5147
	num_pages = num_extent_pages(eb->start, eb->len);
5148
	for (i = 0; i < num_pages; i++) {
5149
		page = eb->pages[i];
C
Chris Mason 已提交
5150 5151
		if (page)
			ClearPageUptodate(page);
5152 5153 5154
	}
}

5155
void set_extent_buffer_uptodate(struct extent_buffer *eb)
5156 5157 5158 5159 5160
{
	unsigned long i;
	struct page *page;
	unsigned long num_pages;

5161
	set_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags);
5162 5163
	num_pages = num_extent_pages(eb->start, eb->len);
	for (i = 0; i < num_pages; i++) {
5164
		page = eb->pages[i];
5165 5166 5167 5168
		SetPageUptodate(page);
	}
}

5169
int extent_buffer_uptodate(struct extent_buffer *eb)
5170
{
5171
	return test_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags);
5172 5173 5174
}

int read_extent_buffer_pages(struct extent_io_tree *tree,
5175
			     struct extent_buffer *eb, u64 start, int wait,
5176
			     get_extent_t *get_extent, int mirror_num)
5177 5178 5179 5180 5181 5182
{
	unsigned long i;
	unsigned long start_i;
	struct page *page;
	int err;
	int ret = 0;
5183 5184
	int locked_pages = 0;
	int all_uptodate = 1;
5185
	unsigned long num_pages;
5186
	unsigned long num_reads = 0;
5187
	struct bio *bio = NULL;
C
Chris Mason 已提交
5188
	unsigned long bio_flags = 0;
5189

5190
	if (test_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags))
5191 5192 5193 5194 5195 5196 5197 5198 5199 5200 5201 5202
		return 0;

	if (start) {
		WARN_ON(start < eb->start);
		start_i = (start >> PAGE_CACHE_SHIFT) -
			(eb->start >> PAGE_CACHE_SHIFT);
	} else {
		start_i = 0;
	}

	num_pages = num_extent_pages(eb->start, eb->len);
	for (i = start_i; i < num_pages; i++) {
5203
		page = eb->pages[i];
5204
		if (wait == WAIT_NONE) {
5205
			if (!trylock_page(page))
5206
				goto unlock_exit;
5207 5208 5209
		} else {
			lock_page(page);
		}
5210
		locked_pages++;
5211 5212
		if (!PageUptodate(page)) {
			num_reads++;
5213
			all_uptodate = 0;
5214
		}
5215 5216 5217
	}
	if (all_uptodate) {
		if (start_i == 0)
5218
			set_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags);
5219 5220 5221
		goto unlock_exit;
	}

5222
	clear_bit(EXTENT_BUFFER_READ_ERR, &eb->bflags);
5223
	eb->read_mirror = 0;
5224
	atomic_set(&eb->io_pages, num_reads);
5225
	for (i = start_i; i < num_pages; i++) {
5226
		page = eb->pages[i];
5227
		if (!PageUptodate(page)) {
5228
			ClearPageError(page);
5229
			err = __extent_read_full_page(tree, page,
5230
						      get_extent, &bio,
5231 5232
						      mirror_num, &bio_flags,
						      READ | REQ_META);
C
Chris Mason 已提交
5233
			if (err)
5234 5235 5236 5237 5238 5239
				ret = err;
		} else {
			unlock_page(page);
		}
	}

5240
	if (bio) {
5241 5242
		err = submit_one_bio(READ | REQ_META, bio, mirror_num,
				     bio_flags);
5243 5244
		if (err)
			return err;
5245
	}
5246

5247
	if (ret || wait != WAIT_COMPLETE)
5248
		return ret;
C
Chris Mason 已提交
5249

5250
	for (i = start_i; i < num_pages; i++) {
5251
		page = eb->pages[i];
5252
		wait_on_page_locked(page);
C
Chris Mason 已提交
5253
		if (!PageUptodate(page))
5254 5255
			ret = -EIO;
	}
C
Chris Mason 已提交
5256

5257
	return ret;
5258 5259 5260

unlock_exit:
	i = start_i;
C
Chris Mason 已提交
5261
	while (locked_pages > 0) {
5262
		page = eb->pages[i];
5263 5264 5265 5266 5267
		i++;
		unlock_page(page);
		locked_pages--;
	}
	return ret;
5268 5269 5270 5271 5272 5273 5274 5275 5276 5277 5278 5279 5280 5281 5282 5283 5284
}

void read_extent_buffer(struct extent_buffer *eb, void *dstv,
			unsigned long start,
			unsigned long len)
{
	size_t cur;
	size_t offset;
	struct page *page;
	char *kaddr;
	char *dst = (char *)dstv;
	size_t start_offset = eb->start & ((u64)PAGE_CACHE_SIZE - 1);
	unsigned long i = (start_offset + start) >> PAGE_CACHE_SHIFT;

	WARN_ON(start > eb->len);
	WARN_ON(start + len > eb->start + eb->len);

5285
	offset = (start_offset + start) & (PAGE_CACHE_SIZE - 1);
5286

C
Chris Mason 已提交
5287
	while (len > 0) {
5288
		page = eb->pages[i];
5289 5290

		cur = min(len, (PAGE_CACHE_SIZE - offset));
5291
		kaddr = page_address(page);
5292 5293 5294 5295 5296 5297 5298 5299 5300
		memcpy(dst, kaddr + offset, cur);

		dst += cur;
		len -= cur;
		offset = 0;
		i++;
	}
}

5301 5302 5303 5304 5305 5306 5307 5308 5309 5310 5311 5312 5313 5314 5315 5316 5317 5318 5319
int read_extent_buffer_to_user(struct extent_buffer *eb, void __user *dstv,
			unsigned long start,
			unsigned long len)
{
	size_t cur;
	size_t offset;
	struct page *page;
	char *kaddr;
	char __user *dst = (char __user *)dstv;
	size_t start_offset = eb->start & ((u64)PAGE_CACHE_SIZE - 1);
	unsigned long i = (start_offset + start) >> PAGE_CACHE_SHIFT;
	int ret = 0;

	WARN_ON(start > eb->len);
	WARN_ON(start + len > eb->start + eb->len);

	offset = (start_offset + start) & (PAGE_CACHE_SIZE - 1);

	while (len > 0) {
5320
		page = eb->pages[i];
5321 5322 5323 5324 5325 5326 5327 5328 5329 5330 5331 5332 5333 5334 5335 5336 5337

		cur = min(len, (PAGE_CACHE_SIZE - offset));
		kaddr = page_address(page);
		if (copy_to_user(dst, kaddr + offset, cur)) {
			ret = -EFAULT;
			break;
		}

		dst += cur;
		len -= cur;
		offset = 0;
		i++;
	}

	return ret;
}

5338
int map_private_extent_buffer(struct extent_buffer *eb, unsigned long start,
5339
			       unsigned long min_len, char **map,
5340
			       unsigned long *map_start,
5341
			       unsigned long *map_len)
5342 5343 5344 5345 5346 5347 5348 5349 5350 5351 5352 5353 5354 5355 5356 5357 5358 5359 5360
{
	size_t offset = start & (PAGE_CACHE_SIZE - 1);
	char *kaddr;
	struct page *p;
	size_t start_offset = eb->start & ((u64)PAGE_CACHE_SIZE - 1);
	unsigned long i = (start_offset + start) >> PAGE_CACHE_SHIFT;
	unsigned long end_i = (start_offset + start + min_len - 1) >>
		PAGE_CACHE_SHIFT;

	if (i != end_i)
		return -EINVAL;

	if (i == 0) {
		offset = start_offset;
		*map_start = 0;
	} else {
		offset = 0;
		*map_start = ((u64)i << PAGE_CACHE_SHIFT) - start_offset;
	}
C
Chris Mason 已提交
5361

5362
	if (start + min_len > eb->len) {
J
Julia Lawall 已提交
5363
		WARN(1, KERN_ERR "btrfs bad mapping eb start %llu len %lu, "
5364 5365
		       "wanted %lu %lu\n",
		       eb->start, eb->len, start, min_len);
5366
		return -EINVAL;
5367 5368
	}

5369
	p = eb->pages[i];
5370
	kaddr = page_address(p);
5371 5372 5373 5374 5375 5376 5377 5378 5379 5380 5381 5382 5383 5384 5385 5386 5387 5388 5389 5390 5391
	*map = kaddr + offset;
	*map_len = PAGE_CACHE_SIZE - offset;
	return 0;
}

int memcmp_extent_buffer(struct extent_buffer *eb, const void *ptrv,
			  unsigned long start,
			  unsigned long len)
{
	size_t cur;
	size_t offset;
	struct page *page;
	char *kaddr;
	char *ptr = (char *)ptrv;
	size_t start_offset = eb->start & ((u64)PAGE_CACHE_SIZE - 1);
	unsigned long i = (start_offset + start) >> PAGE_CACHE_SHIFT;
	int ret = 0;

	WARN_ON(start > eb->len);
	WARN_ON(start + len > eb->start + eb->len);

5392
	offset = (start_offset + start) & (PAGE_CACHE_SIZE - 1);
5393

C
Chris Mason 已提交
5394
	while (len > 0) {
5395
		page = eb->pages[i];
5396 5397 5398

		cur = min(len, (PAGE_CACHE_SIZE - offset));

5399
		kaddr = page_address(page);
5400 5401 5402 5403 5404 5405 5406 5407 5408 5409 5410 5411 5412 5413 5414 5415 5416 5417 5418 5419 5420 5421 5422 5423 5424 5425
		ret = memcmp(ptr, kaddr + offset, cur);
		if (ret)
			break;

		ptr += cur;
		len -= cur;
		offset = 0;
		i++;
	}
	return ret;
}

void write_extent_buffer(struct extent_buffer *eb, const void *srcv,
			 unsigned long start, unsigned long len)
{
	size_t cur;
	size_t offset;
	struct page *page;
	char *kaddr;
	char *src = (char *)srcv;
	size_t start_offset = eb->start & ((u64)PAGE_CACHE_SIZE - 1);
	unsigned long i = (start_offset + start) >> PAGE_CACHE_SHIFT;

	WARN_ON(start > eb->len);
	WARN_ON(start + len > eb->start + eb->len);

5426
	offset = (start_offset + start) & (PAGE_CACHE_SIZE - 1);
5427

C
Chris Mason 已提交
5428
	while (len > 0) {
5429
		page = eb->pages[i];
5430 5431 5432
		WARN_ON(!PageUptodate(page));

		cur = min(len, PAGE_CACHE_SIZE - offset);
5433
		kaddr = page_address(page);
5434 5435 5436 5437 5438 5439 5440 5441 5442 5443 5444 5445 5446 5447 5448 5449 5450 5451 5452 5453 5454 5455
		memcpy(kaddr + offset, src, cur);

		src += cur;
		len -= cur;
		offset = 0;
		i++;
	}
}

void memset_extent_buffer(struct extent_buffer *eb, char c,
			  unsigned long start, unsigned long len)
{
	size_t cur;
	size_t offset;
	struct page *page;
	char *kaddr;
	size_t start_offset = eb->start & ((u64)PAGE_CACHE_SIZE - 1);
	unsigned long i = (start_offset + start) >> PAGE_CACHE_SHIFT;

	WARN_ON(start > eb->len);
	WARN_ON(start + len > eb->start + eb->len);

5456
	offset = (start_offset + start) & (PAGE_CACHE_SIZE - 1);
5457

C
Chris Mason 已提交
5458
	while (len > 0) {
5459
		page = eb->pages[i];
5460 5461 5462
		WARN_ON(!PageUptodate(page));

		cur = min(len, PAGE_CACHE_SIZE - offset);
5463
		kaddr = page_address(page);
5464 5465 5466 5467 5468 5469 5470 5471 5472 5473 5474 5475 5476 5477 5478 5479 5480 5481 5482 5483 5484 5485 5486
		memset(kaddr + offset, c, cur);

		len -= cur;
		offset = 0;
		i++;
	}
}

void copy_extent_buffer(struct extent_buffer *dst, struct extent_buffer *src,
			unsigned long dst_offset, unsigned long src_offset,
			unsigned long len)
{
	u64 dst_len = dst->len;
	size_t cur;
	size_t offset;
	struct page *page;
	char *kaddr;
	size_t start_offset = dst->start & ((u64)PAGE_CACHE_SIZE - 1);
	unsigned long i = (start_offset + dst_offset) >> PAGE_CACHE_SHIFT;

	WARN_ON(src->len != dst_len);

	offset = (start_offset + dst_offset) &
5487
		(PAGE_CACHE_SIZE - 1);
5488

C
Chris Mason 已提交
5489
	while (len > 0) {
5490
		page = dst->pages[i];
5491 5492 5493 5494
		WARN_ON(!PageUptodate(page));

		cur = min(len, (unsigned long)(PAGE_CACHE_SIZE - offset));

5495
		kaddr = page_address(page);
5496 5497 5498 5499 5500 5501 5502 5503 5504
		read_extent_buffer(src, kaddr + offset, src_offset, cur);

		src_offset += cur;
		len -= cur;
		offset = 0;
		i++;
	}
}

5505 5506 5507 5508 5509 5510 5511 5512 5513 5514 5515 5516 5517 5518 5519 5520 5521 5522 5523 5524 5525 5526 5527 5528 5529 5530 5531 5532 5533 5534 5535 5536 5537 5538 5539 5540 5541 5542 5543 5544 5545 5546 5547 5548 5549 5550 5551 5552 5553 5554 5555 5556 5557 5558 5559 5560 5561 5562 5563 5564 5565 5566 5567 5568 5569 5570 5571 5572 5573 5574 5575 5576 5577 5578 5579 5580 5581 5582 5583 5584 5585 5586 5587 5588 5589 5590 5591 5592 5593 5594 5595 5596 5597 5598 5599 5600 5601 5602 5603 5604 5605 5606 5607 5608 5609 5610 5611 5612 5613 5614 5615 5616 5617 5618 5619 5620 5621 5622 5623 5624 5625 5626 5627 5628 5629 5630 5631 5632 5633 5634 5635 5636 5637 5638 5639 5640 5641 5642 5643 5644 5645 5646 5647 5648 5649 5650 5651 5652 5653
/*
 * The extent buffer bitmap operations are done with byte granularity because
 * bitmap items are not guaranteed to be aligned to a word and therefore a
 * single word in a bitmap may straddle two pages in the extent buffer.
 */
#define BIT_BYTE(nr) ((nr) / BITS_PER_BYTE)
#define BYTE_MASK ((1 << BITS_PER_BYTE) - 1)
#define BITMAP_FIRST_BYTE_MASK(start) \
	((BYTE_MASK << ((start) & (BITS_PER_BYTE - 1))) & BYTE_MASK)
#define BITMAP_LAST_BYTE_MASK(nbits) \
	(BYTE_MASK >> (-(nbits) & (BITS_PER_BYTE - 1)))

/*
 * eb_bitmap_offset() - calculate the page and offset of the byte containing the
 * given bit number
 * @eb: the extent buffer
 * @start: offset of the bitmap item in the extent buffer
 * @nr: bit number
 * @page_index: return index of the page in the extent buffer that contains the
 * given bit number
 * @page_offset: return offset into the page given by page_index
 *
 * This helper hides the ugliness of finding the byte in an extent buffer which
 * contains a given bit.
 */
static inline void eb_bitmap_offset(struct extent_buffer *eb,
				    unsigned long start, unsigned long nr,
				    unsigned long *page_index,
				    size_t *page_offset)
{
	size_t start_offset = eb->start & ((u64)PAGE_CACHE_SIZE - 1);
	size_t byte_offset = BIT_BYTE(nr);
	size_t offset;

	/*
	 * The byte we want is the offset of the extent buffer + the offset of
	 * the bitmap item in the extent buffer + the offset of the byte in the
	 * bitmap item.
	 */
	offset = start_offset + start + byte_offset;

	*page_index = offset >> PAGE_CACHE_SHIFT;
	*page_offset = offset & (PAGE_CACHE_SIZE - 1);
}

/**
 * extent_buffer_test_bit - determine whether a bit in a bitmap item is set
 * @eb: the extent buffer
 * @start: offset of the bitmap item in the extent buffer
 * @nr: bit number to test
 */
int extent_buffer_test_bit(struct extent_buffer *eb, unsigned long start,
			   unsigned long nr)
{
	char *kaddr;
	struct page *page;
	unsigned long i;
	size_t offset;

	eb_bitmap_offset(eb, start, nr, &i, &offset);
	page = eb->pages[i];
	WARN_ON(!PageUptodate(page));
	kaddr = page_address(page);
	return 1U & (kaddr[offset] >> (nr & (BITS_PER_BYTE - 1)));
}

/**
 * extent_buffer_bitmap_set - set an area of a bitmap
 * @eb: the extent buffer
 * @start: offset of the bitmap item in the extent buffer
 * @pos: bit number of the first bit
 * @len: number of bits to set
 */
void extent_buffer_bitmap_set(struct extent_buffer *eb, unsigned long start,
			      unsigned long pos, unsigned long len)
{
	char *kaddr;
	struct page *page;
	unsigned long i;
	size_t offset;
	const unsigned int size = pos + len;
	int bits_to_set = BITS_PER_BYTE - (pos % BITS_PER_BYTE);
	unsigned int mask_to_set = BITMAP_FIRST_BYTE_MASK(pos);

	eb_bitmap_offset(eb, start, pos, &i, &offset);
	page = eb->pages[i];
	WARN_ON(!PageUptodate(page));
	kaddr = page_address(page);

	while (len >= bits_to_set) {
		kaddr[offset] |= mask_to_set;
		len -= bits_to_set;
		bits_to_set = BITS_PER_BYTE;
		mask_to_set = ~0U;
		if (++offset >= PAGE_CACHE_SIZE && len > 0) {
			offset = 0;
			page = eb->pages[++i];
			WARN_ON(!PageUptodate(page));
			kaddr = page_address(page);
		}
	}
	if (len) {
		mask_to_set &= BITMAP_LAST_BYTE_MASK(size);
		kaddr[offset] |= mask_to_set;
	}
}


/**
 * extent_buffer_bitmap_clear - clear an area of a bitmap
 * @eb: the extent buffer
 * @start: offset of the bitmap item in the extent buffer
 * @pos: bit number of the first bit
 * @len: number of bits to clear
 */
void extent_buffer_bitmap_clear(struct extent_buffer *eb, unsigned long start,
				unsigned long pos, unsigned long len)
{
	char *kaddr;
	struct page *page;
	unsigned long i;
	size_t offset;
	const unsigned int size = pos + len;
	int bits_to_clear = BITS_PER_BYTE - (pos % BITS_PER_BYTE);
	unsigned int mask_to_clear = BITMAP_FIRST_BYTE_MASK(pos);

	eb_bitmap_offset(eb, start, pos, &i, &offset);
	page = eb->pages[i];
	WARN_ON(!PageUptodate(page));
	kaddr = page_address(page);

	while (len >= bits_to_clear) {
		kaddr[offset] &= ~mask_to_clear;
		len -= bits_to_clear;
		bits_to_clear = BITS_PER_BYTE;
		mask_to_clear = ~0U;
		if (++offset >= PAGE_CACHE_SIZE && len > 0) {
			offset = 0;
			page = eb->pages[++i];
			WARN_ON(!PageUptodate(page));
			kaddr = page_address(page);
		}
	}
	if (len) {
		mask_to_clear &= BITMAP_LAST_BYTE_MASK(size);
		kaddr[offset] &= ~mask_to_clear;
	}
}

5654 5655 5656 5657 5658 5659
static inline bool areas_overlap(unsigned long src, unsigned long dst, unsigned long len)
{
	unsigned long distance = (src > dst) ? src - dst : dst - src;
	return distance < len;
}

5660 5661 5662 5663
static void copy_pages(struct page *dst_page, struct page *src_page,
		       unsigned long dst_off, unsigned long src_off,
		       unsigned long len)
{
5664
	char *dst_kaddr = page_address(dst_page);
5665
	char *src_kaddr;
5666
	int must_memmove = 0;
5667

5668
	if (dst_page != src_page) {
5669
		src_kaddr = page_address(src_page);
5670
	} else {
5671
		src_kaddr = dst_kaddr;
5672 5673
		if (areas_overlap(src_off, dst_off, len))
			must_memmove = 1;
5674
	}
5675

5676 5677 5678 5679
	if (must_memmove)
		memmove(dst_kaddr + dst_off, src_kaddr + src_off, len);
	else
		memcpy(dst_kaddr + dst_off, src_kaddr + src_off, len);
5680 5681 5682 5683 5684 5685 5686 5687 5688 5689 5690 5691 5692
}

void memcpy_extent_buffer(struct extent_buffer *dst, unsigned long dst_offset,
			   unsigned long src_offset, unsigned long len)
{
	size_t cur;
	size_t dst_off_in_page;
	size_t src_off_in_page;
	size_t start_offset = dst->start & ((u64)PAGE_CACHE_SIZE - 1);
	unsigned long dst_i;
	unsigned long src_i;

	if (src_offset + len > dst->len) {
5693 5694 5695
		btrfs_err(dst->fs_info,
			"memmove bogus src_offset %lu move "
		       "len %lu dst len %lu", src_offset, len, dst->len);
5696 5697 5698
		BUG_ON(1);
	}
	if (dst_offset + len > dst->len) {
5699 5700 5701
		btrfs_err(dst->fs_info,
			"memmove bogus dst_offset %lu move "
		       "len %lu dst len %lu", dst_offset, len, dst->len);
5702 5703 5704
		BUG_ON(1);
	}

C
Chris Mason 已提交
5705
	while (len > 0) {
5706
		dst_off_in_page = (start_offset + dst_offset) &
5707
			(PAGE_CACHE_SIZE - 1);
5708
		src_off_in_page = (start_offset + src_offset) &
5709
			(PAGE_CACHE_SIZE - 1);
5710 5711 5712 5713 5714 5715 5716 5717 5718

		dst_i = (start_offset + dst_offset) >> PAGE_CACHE_SHIFT;
		src_i = (start_offset + src_offset) >> PAGE_CACHE_SHIFT;

		cur = min(len, (unsigned long)(PAGE_CACHE_SIZE -
					       src_off_in_page));
		cur = min_t(unsigned long, cur,
			(unsigned long)(PAGE_CACHE_SIZE - dst_off_in_page));

5719
		copy_pages(dst->pages[dst_i], dst->pages[src_i],
5720 5721 5722 5723 5724 5725 5726 5727 5728 5729 5730 5731 5732 5733 5734 5735 5736 5737 5738 5739 5740
			   dst_off_in_page, src_off_in_page, cur);

		src_offset += cur;
		dst_offset += cur;
		len -= cur;
	}
}

void memmove_extent_buffer(struct extent_buffer *dst, unsigned long dst_offset,
			   unsigned long src_offset, unsigned long len)
{
	size_t cur;
	size_t dst_off_in_page;
	size_t src_off_in_page;
	unsigned long dst_end = dst_offset + len - 1;
	unsigned long src_end = src_offset + len - 1;
	size_t start_offset = dst->start & ((u64)PAGE_CACHE_SIZE - 1);
	unsigned long dst_i;
	unsigned long src_i;

	if (src_offset + len > dst->len) {
5741 5742
		btrfs_err(dst->fs_info, "memmove bogus src_offset %lu move "
		       "len %lu len %lu", src_offset, len, dst->len);
5743 5744 5745
		BUG_ON(1);
	}
	if (dst_offset + len > dst->len) {
5746 5747
		btrfs_err(dst->fs_info, "memmove bogus dst_offset %lu move "
		       "len %lu len %lu", dst_offset, len, dst->len);
5748 5749
		BUG_ON(1);
	}
5750
	if (dst_offset < src_offset) {
5751 5752 5753
		memcpy_extent_buffer(dst, dst_offset, src_offset, len);
		return;
	}
C
Chris Mason 已提交
5754
	while (len > 0) {
5755 5756 5757 5758
		dst_i = (start_offset + dst_end) >> PAGE_CACHE_SHIFT;
		src_i = (start_offset + src_end) >> PAGE_CACHE_SHIFT;

		dst_off_in_page = (start_offset + dst_end) &
5759
			(PAGE_CACHE_SIZE - 1);
5760
		src_off_in_page = (start_offset + src_end) &
5761
			(PAGE_CACHE_SIZE - 1);
5762 5763 5764

		cur = min_t(unsigned long, len, src_off_in_page + 1);
		cur = min(cur, dst_off_in_page + 1);
5765
		copy_pages(dst->pages[dst_i], dst->pages[src_i],
5766 5767 5768 5769 5770 5771 5772 5773
			   dst_off_in_page - cur + 1,
			   src_off_in_page - cur + 1, cur);

		dst_end -= cur;
		src_end -= cur;
		len -= cur;
	}
}
5774

5775
int try_release_extent_buffer(struct page *page)
5776
{
5777 5778
	struct extent_buffer *eb;

5779 5780 5781 5782 5783 5784 5785
	/*
	 * We need to make sure noboody is attaching this page to an eb right
	 * now.
	 */
	spin_lock(&page->mapping->private_lock);
	if (!PagePrivate(page)) {
		spin_unlock(&page->mapping->private_lock);
J
Josef Bacik 已提交
5786
		return 1;
5787
	}
5788

5789 5790
	eb = (struct extent_buffer *)page->private;
	BUG_ON(!eb);
5791 5792

	/*
5793 5794 5795
	 * This is a little awful but should be ok, we need to make sure that
	 * the eb doesn't disappear out from under us while we're looking at
	 * this page.
5796
	 */
5797
	spin_lock(&eb->refs_lock);
5798
	if (atomic_read(&eb->refs) != 1 || extent_buffer_under_io(eb)) {
5799 5800 5801
		spin_unlock(&eb->refs_lock);
		spin_unlock(&page->mapping->private_lock);
		return 0;
5802
	}
5803
	spin_unlock(&page->mapping->private_lock);
5804

5805
	/*
5806 5807
	 * If tree ref isn't set then we know the ref on this eb is a real ref,
	 * so just return, this page will likely be freed soon anyway.
5808
	 */
5809 5810 5811
	if (!test_and_clear_bit(EXTENT_BUFFER_TREE_REF, &eb->bflags)) {
		spin_unlock(&eb->refs_lock);
		return 0;
5812
	}
5813

5814
	return release_extent_buffer(eb);
5815
}