data.c 54.9 KB
Newer Older
J
Jaegeuk Kim 已提交
1
/*
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
 * fs/f2fs/data.c
 *
 * Copyright (c) 2012 Samsung Electronics Co., Ltd.
 *             http://www.samsung.com/
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License version 2 as
 * published by the Free Software Foundation.
 */
#include <linux/fs.h>
#include <linux/f2fs_fs.h>
#include <linux/buffer_head.h>
#include <linux/mpage.h>
#include <linux/writeback.h>
#include <linux/backing-dev.h>
C
Chao Yu 已提交
17
#include <linux/pagevec.h>
18 19
#include <linux/blkdev.h>
#include <linux/bio.h>
20
#include <linux/prefetch.h>
21
#include <linux/uio.h>
22 23
#include <linux/mm.h>
#include <linux/memcontrol.h>
J
Jaegeuk Kim 已提交
24
#include <linux/cleancache.h>
25
#include <linux/sched/signal.h>
26 27 28 29

#include "f2fs.h"
#include "node.h"
#include "segment.h"
J
Jaegeuk Kim 已提交
30
#include "trace.h"
31
#include <trace/events/f2fs.h>
32

33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52
static bool __is_cp_guaranteed(struct page *page)
{
	struct address_space *mapping = page->mapping;
	struct inode *inode;
	struct f2fs_sb_info *sbi;

	if (!mapping)
		return false;

	inode = mapping->host;
	sbi = F2FS_I_SB(inode);

	if (inode->i_ino == F2FS_META_INO(sbi) ||
			inode->i_ino ==  F2FS_NODE_INO(sbi) ||
			S_ISDIR(inode->i_mode) ||
			is_cold_data(page))
		return true;
	return false;
}

53
static void f2fs_read_end_io(struct bio *bio)
54
{
55 56
	struct bio_vec *bvec;
	int i;
57

C
Chao Yu 已提交
58
#ifdef CONFIG_F2FS_FAULT_INJECTION
59 60
	if (time_to_inject(F2FS_P_SB(bio->bi_io_vec->bv_page), FAULT_IO)) {
		f2fs_show_injection_info(FAULT_IO);
61
		bio->bi_status = BLK_STS_IOERR;
62
	}
C
Chao Yu 已提交
63 64
#endif

65
	if (f2fs_bio_encrypted(bio)) {
66
		if (bio->bi_status) {
67
			fscrypt_release_ctx(bio->bi_private);
68
		} else {
69
			fscrypt_decrypt_bio_pages(bio->bi_private, bio);
70 71 72 73
			return;
		}
	}

74 75
	bio_for_each_segment_all(bvec, bio, i) {
		struct page *page = bvec->bv_page;
J
Jaegeuk Kim 已提交
76

77
		if (!bio->bi_status) {
78 79
			if (!PageUptodate(page))
				SetPageUptodate(page);
J
Jaegeuk Kim 已提交
80 81 82 83 84 85 86 87 88
		} else {
			ClearPageUptodate(page);
			SetPageError(page);
		}
		unlock_page(page);
	}
	bio_put(bio);
}

89
static void f2fs_write_end_io(struct bio *bio)
90
{
91
	struct f2fs_sb_info *sbi = bio->bi_private;
92 93
	struct bio_vec *bvec;
	int i;
94

95
	bio_for_each_segment_all(bvec, bio, i) {
96
		struct page *page = bvec->bv_page;
97
		enum count_type type = WB_DATA_TYPE(page);
98

99 100 101 102 103 104
		if (IS_DUMMY_WRITTEN_PAGE(page)) {
			set_page_private(page, (unsigned long)NULL);
			ClearPagePrivate(page);
			unlock_page(page);
			mempool_free(page, sbi->write_io_dummy);

105
			if (unlikely(bio->bi_status))
106 107 108 109
				f2fs_stop_checkpoint(sbi, true);
			continue;
		}

110
		fscrypt_pullback_bio_page(&page, true);
111

112
		if (unlikely(bio->bi_status)) {
113
			mapping_set_error(page->mapping, -EIO);
114 115
			if (type == F2FS_WB_CP_DATA)
				f2fs_stop_checkpoint(sbi, true);
116
		}
117 118
		dec_page_count(sbi, type);
		clear_cold_data(page);
119
		end_page_writeback(page);
120
	}
121
	if (!get_pages(sbi, F2FS_WB_CP_DATA) &&
122
				wq_has_sleeper(&sbi->cp_wait))
123 124 125 126 127
		wake_up(&sbi->cp_wait);

	bio_put(bio);
}

J
Jaegeuk Kim 已提交
128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145
/*
 * Return true, if pre_bio's bdev is same as its target device.
 */
struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi,
				block_t blk_addr, struct bio *bio)
{
	struct block_device *bdev = sbi->sb->s_bdev;
	int i;

	for (i = 0; i < sbi->s_ndevs; i++) {
		if (FDEV(i).start_blk <= blk_addr &&
					FDEV(i).end_blk >= blk_addr) {
			blk_addr -= FDEV(i).start_blk;
			bdev = FDEV(i).bdev;
			break;
		}
	}
	if (bio) {
146
		bio_set_dev(bio, bdev);
J
Jaegeuk Kim 已提交
147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164
		bio->bi_iter.bi_sector = SECTOR_FROM_BLOCK(blk_addr);
	}
	return bdev;
}

int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr)
{
	int i;

	for (i = 0; i < sbi->s_ndevs; i++)
		if (FDEV(i).start_blk <= blkaddr && FDEV(i).end_blk >= blkaddr)
			return i;
	return 0;
}

static bool __same_bdev(struct f2fs_sb_info *sbi,
				block_t blk_addr, struct bio *bio)
{
165 166
	struct block_device *b = f2fs_target_device(sbi, blk_addr, NULL);
	return bio->bi_disk == b->bd_disk && bio->bi_partno == b->bd_partno;
J
Jaegeuk Kim 已提交
167 168
}

169 170 171 172 173 174 175 176
/*
 * Low-level block read/write IO operations.
 */
static struct bio *__bio_alloc(struct f2fs_sb_info *sbi, block_t blk_addr,
				int npages, bool is_read)
{
	struct bio *bio;

177
	bio = f2fs_bio_alloc(sbi, npages, true);
178

J
Jaegeuk Kim 已提交
179
	f2fs_target_device(sbi, blk_addr, bio);
180
	bio->bi_end_io = is_read ? f2fs_read_end_io : f2fs_write_end_io;
181
	bio->bi_private = is_read ? NULL : sbi;
182 183 184 185

	return bio;
}

186 187
static inline void __submit_bio(struct f2fs_sb_info *sbi,
				struct bio *bio, enum page_type type)
188
{
189
	if (!is_read_io(bio_op(bio))) {
190 191
		unsigned int start;

192
		if (f2fs_sb_mounted_blkzoned(sbi->sb) &&
193
			current->plug && (type == DATA || type == NODE))
J
Jaegeuk Kim 已提交
194
			blk_finish_plug(current->plug);
195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223

		if (type != DATA && type != NODE)
			goto submit_io;

		start = bio->bi_iter.bi_size >> F2FS_BLKSIZE_BITS;
		start %= F2FS_IO_SIZE(sbi);

		if (start == 0)
			goto submit_io;

		/* fill dummy pages */
		for (; start < F2FS_IO_SIZE(sbi); start++) {
			struct page *page =
				mempool_alloc(sbi->write_io_dummy,
					GFP_NOIO | __GFP_ZERO | __GFP_NOFAIL);
			f2fs_bug_on(sbi, !page);

			SetPagePrivate(page);
			set_page_private(page, (unsigned long)DUMMY_WRITTEN_PAGE);
			lock_page(page);
			if (bio_add_page(bio, page, PAGE_SIZE, 0) < PAGE_SIZE)
				f2fs_bug_on(sbi, 1);
		}
		/*
		 * In the NODE case, we lose next block address chain. So, we
		 * need to do checkpoint in f2fs_sync_file.
		 */
		if (type == NODE)
			set_sbi_flag(sbi, SBI_NEED_CP);
J
Jaegeuk Kim 已提交
224
	}
225
submit_io:
J
Jaegeuk Kim 已提交
226 227 228 229
	if (is_read_io(bio_op(bio)))
		trace_f2fs_submit_read_bio(sbi->sb, type, bio);
	else
		trace_f2fs_submit_write_bio(sbi->sb, type, bio);
230
	submit_bio(bio);
231 232
}

J
Jaegeuk Kim 已提交
233
static void __submit_merged_bio(struct f2fs_bio_info *io)
234
{
J
Jaegeuk Kim 已提交
235
	struct f2fs_io_info *fio = &io->fio;
236 237 238 239

	if (!io->bio)
		return;

J
Jaegeuk Kim 已提交
240 241
	bio_set_op_attrs(io->bio, fio->op, fio->op_flags);

M
Mike Christie 已提交
242
	if (is_read_io(fio->op))
J
Jaegeuk Kim 已提交
243
		trace_f2fs_prepare_read_bio(io->sbi->sb, fio->type, io->bio);
244
	else
J
Jaegeuk Kim 已提交
245
		trace_f2fs_prepare_write_bio(io->sbi->sb, fio->type, io->bio);
M
Mike Christie 已提交
246

247
	__submit_bio(io->sbi, io->bio, fio->type);
248 249 250
	io->bio = NULL;
}

251 252
static bool __has_merged_page(struct f2fs_bio_info *io,
				struct inode *inode, nid_t ino, pgoff_t idx)
C
Chao Yu 已提交
253 254 255 256 257
{
	struct bio_vec *bvec;
	struct page *target;
	int i;

258
	if (!io->bio)
C
Chao Yu 已提交
259
		return false;
260

261
	if (!inode && !ino)
262
		return true;
C
Chao Yu 已提交
263 264 265

	bio_for_each_segment_all(bvec, io->bio, i) {

266
		if (bvec->bv_page->mapping)
C
Chao Yu 已提交
267
			target = bvec->bv_page;
268 269
		else
			target = fscrypt_control_page(bvec->bv_page);
C
Chao Yu 已提交
270

271 272 273
		if (idx != target->index)
			continue;

274 275 276
		if (inode && inode == target->mapping->host)
			return true;
		if (ino && ino == ino_of_node(target))
C
Chao Yu 已提交
277 278 279 280 281 282
			return true;
	}

	return false;
}

283
static bool has_merged_page(struct f2fs_sb_info *sbi, struct inode *inode,
284
				nid_t ino, pgoff_t idx, enum page_type type)
285 286
{
	enum page_type btype = PAGE_TYPE_OF_BIO(type);
J
Jaegeuk Kim 已提交
287 288 289
	enum temp_type temp;
	struct f2fs_bio_info *io;
	bool ret = false;
290

J
Jaegeuk Kim 已提交
291 292 293 294 295 296
	for (temp = HOT; temp < NR_TEMP_TYPE; temp++) {
		io = sbi->write_io[btype] + temp;

		down_read(&io->io_rwsem);
		ret = __has_merged_page(io, inode, ino, idx);
		up_read(&io->io_rwsem);
297

J
Jaegeuk Kim 已提交
298 299 300 301
		/* TODO: use HOT temp only for meta pages now. */
		if (ret || btype == META)
			break;
	}
302 303 304
	return ret;
}

305
static void __f2fs_submit_merged_write(struct f2fs_sb_info *sbi,
J
Jaegeuk Kim 已提交
306
				enum page_type type, enum temp_type temp)
307 308
{
	enum page_type btype = PAGE_TYPE_OF_BIO(type);
J
Jaegeuk Kim 已提交
309
	struct f2fs_bio_info *io = sbi->write_io[btype] + temp;
310

311
	down_write(&io->io_rwsem);
J
Jaegeuk Kim 已提交
312 313 314 315

	/* change META to META_FLUSH in the checkpoint procedure */
	if (type >= META_FLUSH) {
		io->fio.type = META_FLUSH;
M
Mike Christie 已提交
316
		io->fio.op = REQ_OP_WRITE;
317
		io->fio.op_flags = REQ_META | REQ_PRIO | REQ_SYNC;
318
		if (!test_opt(sbi, NOBARRIER))
319
			io->fio.op_flags |= REQ_PREFLUSH | REQ_FUA;
J
Jaegeuk Kim 已提交
320 321
	}
	__submit_merged_bio(io);
322
	up_write(&io->io_rwsem);
323 324
}

J
Jaegeuk Kim 已提交
325 326 327
static void __submit_merged_write_cond(struct f2fs_sb_info *sbi,
				struct inode *inode, nid_t ino, pgoff_t idx,
				enum page_type type, bool force)
328
{
J
Jaegeuk Kim 已提交
329 330 331 332 333 334 335 336 337 338 339 340 341
	enum temp_type temp;

	if (!force && !has_merged_page(sbi, inode, ino, idx, type))
		return;

	for (temp = HOT; temp < NR_TEMP_TYPE; temp++) {

		__f2fs_submit_merged_write(sbi, type, temp);

		/* TODO: use HOT temp only for meta pages now. */
		if (type >= META)
			break;
	}
342 343
}

344
void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type)
345
{
J
Jaegeuk Kim 已提交
346
	__submit_merged_write_cond(sbi, NULL, 0, 0, type, true);
347 348
}

349
void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi,
350
				struct inode *inode, nid_t ino, pgoff_t idx,
351
				enum page_type type)
352
{
J
Jaegeuk Kim 已提交
353
	__submit_merged_write_cond(sbi, inode, ino, idx, type, false);
354 355
}

356
void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi)
357
{
358 359 360
	f2fs_submit_merged_write(sbi, DATA);
	f2fs_submit_merged_write(sbi, NODE);
	f2fs_submit_merged_write(sbi, META);
361 362
}

363 364
/*
 * Fill the locked page with data located in the block address.
365
 * A caller needs to unlock the page on failure.
366
 */
367
int f2fs_submit_page_bio(struct f2fs_io_info *fio)
368 369
{
	struct bio *bio;
370 371
	struct page *page = fio->encrypted_page ?
			fio->encrypted_page : fio->page;
372

373
	trace_f2fs_submit_page_bio(page, fio);
374
	f2fs_trace_ios(fio, 0);
375 376

	/* Allocate a new bio */
M
Mike Christie 已提交
377
	bio = __bio_alloc(fio->sbi, fio->new_blkaddr, 1, is_read_io(fio->op));
378

379
	if (bio_add_page(bio, page, PAGE_SIZE, 0) < PAGE_SIZE) {
380 381 382
		bio_put(bio);
		return -EFAULT;
	}
M
Mike Christie 已提交
383
	bio_set_op_attrs(bio, fio->op, fio->op_flags);
384

385
	__submit_bio(fio->sbi, bio, fio->type);
386 387 388

	if (!is_read_io(fio->op))
		inc_page_count(fio->sbi, WB_DATA_TYPE(fio->page));
389 390 391
	return 0;
}

392
int f2fs_submit_page_write(struct f2fs_io_info *fio)
393
{
394
	struct f2fs_sb_info *sbi = fio->sbi;
J
Jaegeuk Kim 已提交
395
	enum page_type btype = PAGE_TYPE_OF_BIO(fio->type);
J
Jaegeuk Kim 已提交
396
	struct f2fs_bio_info *io = sbi->write_io[btype] + fio->temp;
397
	struct page *bio_page;
398
	int err = 0;
399

400
	f2fs_bug_on(sbi, is_read_io(fio->op));
401

402 403 404 405 406 407 408 409 410 411 412 413 414
	down_write(&io->io_rwsem);
next:
	if (fio->in_list) {
		spin_lock(&io->io_lock);
		if (list_empty(&io->io_list)) {
			spin_unlock(&io->io_lock);
			goto out_fail;
		}
		fio = list_first_entry(&io->io_list,
						struct f2fs_io_info, list);
		list_del(&fio->list);
		spin_unlock(&io->io_lock);
	}
415

416 417 418
	if (fio->old_blkaddr != NEW_ADDR)
		verify_block_addr(sbi, fio->old_blkaddr);
	verify_block_addr(sbi, fio->new_blkaddr);
419

420 421
	bio_page = fio->encrypted_page ? fio->encrypted_page : fio->page;

422 423
	/* set submitted = true as a return value */
	fio->submitted = true;
424

425
	inc_page_count(sbi, WB_DATA_TYPE(bio_page));
426

427
	if (io->bio && (io->last_block_in_bio != fio->new_blkaddr - 1 ||
J
Jaegeuk Kim 已提交
428 429
	    (io->fio.op != fio->op || io->fio.op_flags != fio->op_flags) ||
			!__same_bdev(sbi, fio->new_blkaddr, io->bio)))
J
Jaegeuk Kim 已提交
430
		__submit_merged_bio(io);
431 432
alloc_new:
	if (io->bio == NULL) {
433 434 435
		if ((fio->type == DATA || fio->type == NODE) &&
				fio->new_blkaddr & F2FS_IO_SIZE_MASK(sbi)) {
			err = -EAGAIN;
436
			dec_page_count(sbi, WB_DATA_TYPE(bio_page));
437 438
			goto out_fail;
		}
439
		io->bio = __bio_alloc(sbi, fio->new_blkaddr,
440
						BIO_MAX_PAGES, false);
J
Jaegeuk Kim 已提交
441
		io->fio = *fio;
442 443
	}

J
Jaegeuk Kim 已提交
444
	if (bio_add_page(io->bio, bio_page, PAGE_SIZE, 0) < PAGE_SIZE) {
J
Jaegeuk Kim 已提交
445
		__submit_merged_bio(io);
446 447 448
		goto alloc_new;
	}

449
	io->last_block_in_bio = fio->new_blkaddr;
450
	f2fs_trace_ios(fio, 0);
451 452 453 454 455

	trace_f2fs_submit_page_write(fio->page, fio);

	if (fio->in_list)
		goto next;
456
out_fail:
457
	up_write(&io->io_rwsem);
458
	return err;
459 460
}

461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476
static struct bio *f2fs_grab_read_bio(struct inode *inode, block_t blkaddr,
							 unsigned nr_pages)
{
	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
	struct fscrypt_ctx *ctx = NULL;
	struct bio *bio;

	if (f2fs_encrypted_file(inode)) {
		ctx = fscrypt_get_ctx(inode, GFP_NOFS);
		if (IS_ERR(ctx))
			return ERR_CAST(ctx);

		/* wait the page to be moved by cleaning */
		f2fs_wait_on_block_writeback(sbi, blkaddr);
	}

477
	bio = f2fs_bio_alloc(sbi, min_t(int, nr_pages, BIO_MAX_PAGES), false);
478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507
	if (!bio) {
		if (ctx)
			fscrypt_release_ctx(ctx);
		return ERR_PTR(-ENOMEM);
	}
	f2fs_target_device(sbi, blkaddr, bio);
	bio->bi_end_io = f2fs_read_end_io;
	bio->bi_private = ctx;
	bio_set_op_attrs(bio, REQ_OP_READ, 0);

	return bio;
}

/* This can handle encryption stuffs */
static int f2fs_submit_page_read(struct inode *inode, struct page *page,
							block_t blkaddr)
{
	struct bio *bio = f2fs_grab_read_bio(inode, blkaddr, 1);

	if (IS_ERR(bio))
		return PTR_ERR(bio);

	if (bio_add_page(bio, page, PAGE_SIZE, 0) < PAGE_SIZE) {
		bio_put(bio);
		return -EFAULT;
	}
	__submit_bio(F2FS_I_SB(inode), bio, DATA);
	return 0;
}

508 509 510 511
static void __set_data_blkaddr(struct dnode_of_data *dn)
{
	struct f2fs_node *rn = F2FS_NODE(dn->node_page);
	__le32 *addr_array;
512 513 514 515
	int base = 0;

	if (IS_INODE(dn->node_page) && f2fs_has_extra_attr(dn->inode))
		base = get_extra_isize(dn->inode);
516 517 518

	/* Get physical address of data block */
	addr_array = blkaddr_in_node(rn);
519
	addr_array[base + dn->ofs_in_node] = cpu_to_le32(dn->data_blkaddr);
520 521
}

J
Jaegeuk Kim 已提交
522
/*
523 524 525 526 527
 * Lock ordering for the change of data block address:
 * ->data_page
 *  ->node_page
 *    update block addresses in the node page
 */
528
void set_data_blkaddr(struct dnode_of_data *dn)
529
{
530 531 532
	f2fs_wait_on_page_writeback(dn->node_page, NODE, true);
	__set_data_blkaddr(dn);
	if (set_page_dirty(dn->node_page))
533
		dn->node_changed = true;
534 535
}

536 537 538 539 540 541 542
void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr)
{
	dn->data_blkaddr = blkaddr;
	set_data_blkaddr(dn);
	f2fs_update_extent_cache(dn);
}

543 544
/* dn->ofs_in_node will be returned with up-to-date last block pointer */
int reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count)
545
{
546
	struct f2fs_sb_info *sbi = F2FS_I_SB(dn->inode);
C
Chao Yu 已提交
547
	int err;
548

549 550 551
	if (!count)
		return 0;

552
	if (unlikely(is_inode_flag_set(dn->inode, FI_NO_ALLOC)))
553
		return -EPERM;
C
Chao Yu 已提交
554 555
	if (unlikely((err = inc_valid_block_count(sbi, dn->inode, &count))))
		return err;
556

557 558 559 560 561 562
	trace_f2fs_reserve_new_blocks(dn->inode, dn->nid,
						dn->ofs_in_node, count);

	f2fs_wait_on_page_writeback(dn->node_page, NODE, true);

	for (; count > 0; dn->ofs_in_node++) {
563 564
		block_t blkaddr = datablock_addr(dn->inode,
					dn->node_page, dn->ofs_in_node);
565 566 567 568 569 570 571 572 573
		if (blkaddr == NULL_ADDR) {
			dn->data_blkaddr = NEW_ADDR;
			__set_data_blkaddr(dn);
			count--;
		}
	}

	if (set_page_dirty(dn->node_page))
		dn->node_changed = true;
574 575 576
	return 0;
}

577 578 579 580 581 582 583 584 585 586 587
/* Should keep dn->ofs_in_node unchanged */
int reserve_new_block(struct dnode_of_data *dn)
{
	unsigned int ofs_in_node = dn->ofs_in_node;
	int ret;

	ret = reserve_new_blocks(dn, 1);
	dn->ofs_in_node = ofs_in_node;
	return ret;
}

588 589 590 591 592 593 594 595
int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index)
{
	bool need_put = dn->inode_page ? false : true;
	int err;

	err = get_dnode_of_data(dn, index, ALLOC_NODE);
	if (err)
		return err;
596

597 598
	if (dn->data_blkaddr == NULL_ADDR)
		err = reserve_new_block(dn);
599
	if (err || need_put)
600 601 602 603
		f2fs_put_dnode(dn);
	return err;
}

604
int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index)
605
{
606
	struct extent_info ei  = {0,0,0};
607
	struct inode *inode = dn->inode;
608

609 610 611
	if (f2fs_lookup_extent_cache(inode, index, &ei)) {
		dn->data_blkaddr = ei.blk + index - ei.fofs;
		return 0;
612
	}
613

614
	return f2fs_reserve_block(dn, index);
615 616
}

617
struct page *get_read_data_page(struct inode *inode, pgoff_t index,
M
Mike Christie 已提交
618
						int op_flags, bool for_write)
619 620 621 622
{
	struct address_space *mapping = inode->i_mapping;
	struct dnode_of_data dn;
	struct page *page;
623
	struct extent_info ei = {0,0,0};
624
	int err;
625

626
	page = f2fs_grab_cache_page(mapping, index, for_write);
627 628 629
	if (!page)
		return ERR_PTR(-ENOMEM);

C
Chao Yu 已提交
630 631 632 633 634
	if (f2fs_lookup_extent_cache(inode, index, &ei)) {
		dn.data_blkaddr = ei.blk + index - ei.fofs;
		goto got_it;
	}

635
	set_new_dnode(&dn, inode, NULL, NULL, 0);
636
	err = get_dnode_of_data(&dn, index, LOOKUP_NODE);
637 638
	if (err)
		goto put_err;
639 640
	f2fs_put_dnode(&dn);

641
	if (unlikely(dn.data_blkaddr == NULL_ADDR)) {
642 643
		err = -ENOENT;
		goto put_err;
644
	}
C
Chao Yu 已提交
645
got_it:
646 647
	if (PageUptodate(page)) {
		unlock_page(page);
648
		return page;
649
	}
650

J
Jaegeuk Kim 已提交
651 652 653 654 655 656 657
	/*
	 * A new dentry page is allocated but not able to be written, since its
	 * new inode page couldn't be allocated due to -ENOSPC.
	 * In such the case, its blkaddr can be remained as NEW_ADDR.
	 * see, f2fs_add_link -> get_new_data_page -> init_inode_metadata.
	 */
	if (dn.data_blkaddr == NEW_ADDR) {
658
		zero_user_segment(page, 0, PAGE_SIZE);
659 660
		if (!PageUptodate(page))
			SetPageUptodate(page);
661
		unlock_page(page);
J
Jaegeuk Kim 已提交
662 663
		return page;
	}
664

665
	err = f2fs_submit_page_read(inode, page, dn.data_blkaddr);
666
	if (err)
667
		goto put_err;
668
	return page;
669 670 671 672

put_err:
	f2fs_put_page(page, 1);
	return ERR_PTR(err);
673 674 675 676 677 678 679 680 681 682 683 684
}

struct page *find_data_page(struct inode *inode, pgoff_t index)
{
	struct address_space *mapping = inode->i_mapping;
	struct page *page;

	page = find_get_page(mapping, index);
	if (page && PageUptodate(page))
		return page;
	f2fs_put_page(page, 0);

685
	page = get_read_data_page(inode, index, 0, false);
686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704
	if (IS_ERR(page))
		return page;

	if (PageUptodate(page))
		return page;

	wait_on_page_locked(page);
	if (unlikely(!PageUptodate(page))) {
		f2fs_put_page(page, 0);
		return ERR_PTR(-EIO);
	}
	return page;
}

/*
 * If it tries to access a hole, return an error.
 * Because, the callers, functions in dir.c and GC, should be able to know
 * whether this page exists or not.
 */
705 706
struct page *get_lock_data_page(struct inode *inode, pgoff_t index,
							bool for_write)
707 708 709 710
{
	struct address_space *mapping = inode->i_mapping;
	struct page *page;
repeat:
711
	page = get_read_data_page(inode, index, 0, for_write);
712 713
	if (IS_ERR(page))
		return page;
714

715
	/* wait for read completion */
716
	lock_page(page);
717
	if (unlikely(page->mapping != mapping)) {
718 719
		f2fs_put_page(page, 1);
		goto repeat;
720
	}
721 722 723 724
	if (unlikely(!PageUptodate(page))) {
		f2fs_put_page(page, 1);
		return ERR_PTR(-EIO);
	}
725 726 727
	return page;
}

J
Jaegeuk Kim 已提交
728
/*
729 730
 * Caller ensures that this data page is never allocated.
 * A new zero-filled data page is allocated in the page cache.
731
 *
C
Chao Yu 已提交
732 733
 * Also, caller should grab and release a rwsem by calling f2fs_lock_op() and
 * f2fs_unlock_op().
734 735
 * Note that, ipage is set only by make_empty_dir, and if any error occur,
 * ipage should be released by this function.
736
 */
737
struct page *get_new_data_page(struct inode *inode,
738
		struct page *ipage, pgoff_t index, bool new_i_size)
739 740 741 742 743
{
	struct address_space *mapping = inode->i_mapping;
	struct page *page;
	struct dnode_of_data dn;
	int err;
744

745
	page = f2fs_grab_cache_page(mapping, index, true);
746 747 748 749 750 751
	if (!page) {
		/*
		 * before exiting, we should make sure ipage will be released
		 * if any error occur.
		 */
		f2fs_put_page(ipage, 1);
752
		return ERR_PTR(-ENOMEM);
753
	}
754

755
	set_new_dnode(&dn, inode, ipage, NULL, 0);
756
	err = f2fs_reserve_block(&dn, index);
757 758
	if (err) {
		f2fs_put_page(page, 1);
759
		return ERR_PTR(err);
760
	}
761 762
	if (!ipage)
		f2fs_put_dnode(&dn);
763 764

	if (PageUptodate(page))
765
		goto got_it;
766 767

	if (dn.data_blkaddr == NEW_ADDR) {
768
		zero_user_segment(page, 0, PAGE_SIZE);
769 770
		if (!PageUptodate(page))
			SetPageUptodate(page);
771
	} else {
772
		f2fs_put_page(page, 1);
773

774 775 776
		/* if ipage exists, blkaddr should be NEW_ADDR */
		f2fs_bug_on(F2FS_I_SB(inode), ipage);
		page = get_lock_data_page(inode, index, true);
777
		if (IS_ERR(page))
778
			return page;
779
	}
780
got_it:
C
Chao Yu 已提交
781
	if (new_i_size && i_size_read(inode) <
782
				((loff_t)(index + 1) << PAGE_SHIFT))
783
		f2fs_i_size_write(inode, ((loff_t)(index + 1) << PAGE_SHIFT));
784 785 786
	return page;
}

787
static int __allocate_data_block(struct dnode_of_data *dn, int seg_type)
788
{
789
	struct f2fs_sb_info *sbi = F2FS_I_SB(dn->inode);
790 791
	struct f2fs_summary sum;
	struct node_info ni;
792
	pgoff_t fofs;
793
	blkcnt_t count = 1;
C
Chao Yu 已提交
794
	int err;
795

796
	if (unlikely(is_inode_flag_set(dn->inode, FI_NO_ALLOC)))
797
		return -EPERM;
798

799 800
	dn->data_blkaddr = datablock_addr(dn->inode,
				dn->node_page, dn->ofs_in_node);
801 802 803
	if (dn->data_blkaddr == NEW_ADDR)
		goto alloc;

C
Chao Yu 已提交
804 805
	if (unlikely((err = inc_valid_block_count(sbi, dn->inode, &count))))
		return err;
806

807
alloc:
808 809 810
	get_node_info(sbi, dn->nid, &ni);
	set_summary(&sum, dn->nid, dn->ofs_in_node, ni.version);

811
	allocate_data_block(sbi, NULL, dn->data_blkaddr, &dn->data_blkaddr,
812
					&sum, seg_type, NULL, false);
813
	set_data_blkaddr(dn);
814

815
	/* update i_size */
816
	fofs = start_bidx_of_node(ofs_of_node(dn->node_page), dn->inode) +
817
							dn->ofs_in_node;
818
	if (i_size_read(dn->inode) < ((loff_t)(fofs + 1) << PAGE_SHIFT))
819
		f2fs_i_size_write(dn->inode,
820
				((loff_t)(fofs + 1) << PAGE_SHIFT));
821 822 823
	return 0;
}

J
Jaegeuk Kim 已提交
824 825
static inline bool __force_buffered_io(struct inode *inode, int rw)
{
826
	return (f2fs_encrypted_file(inode) ||
J
Jaegeuk Kim 已提交
827 828 829 830
			(rw == WRITE && test_opt(F2FS_I_SB(inode), LFS)) ||
			F2FS_I_SB(inode)->s_ndevs);
}

831
int f2fs_preallocate_blocks(struct kiocb *iocb, struct iov_iter *from)
832
{
833
	struct inode *inode = file_inode(iocb->ki_filp);
C
Chao Yu 已提交
834
	struct f2fs_map_blocks map;
835
	int err = 0;
836

837 838 839 840 841 842 843
	/* convert inline data for Direct I/O*/
	if (iocb->ki_flags & IOCB_DIRECT) {
		err = f2fs_convert_inline_inode(inode);
		if (err)
			return err;
	}

844 845 846
	if (is_inode_flag_set(inode, FI_NO_PREALLOC))
		return 0;

847
	map.m_lblk = F2FS_BLK_ALIGN(iocb->ki_pos);
848 849 850 851 852 853
	map.m_len = F2FS_BYTES_TO_BLK(iocb->ki_pos + iov_iter_count(from));
	if (map.m_len > map.m_lblk)
		map.m_len -= map.m_lblk;
	else
		map.m_len = 0;

854
	map.m_next_pgofs = NULL;
855
	map.m_seg_type = NO_CHECK_TYPE;
856

857 858
	if (iocb->ki_flags & IOCB_DIRECT) {
		map.m_seg_type = rw_hint_to_seg_type(iocb->ki_hint);
J
Jaegeuk Kim 已提交
859 860 861 862
		return f2fs_map_blocks(inode, &map, 1,
			__force_buffered_io(inode, WRITE) ?
				F2FS_GET_BLOCK_PRE_AIO :
				F2FS_GET_BLOCK_PRE_DIO);
863
	}
C
Chao Yu 已提交
864
	if (iocb->ki_pos + iov_iter_count(from) > MAX_INLINE_DATA(inode)) {
865 866 867
		err = f2fs_convert_inline_inode(inode);
		if (err)
			return err;
868
	}
869 870 871 872 873 874 875 876
	if (!f2fs_has_inline_data(inode)) {
		err = f2fs_map_blocks(inode, &map, 1, F2FS_GET_BLOCK_PRE_AIO);
		if (map.m_len > 0 && err == -ENOSPC) {
			set_inode_flag(inode, FI_NO_PREALLOC);
			err = 0;
		}
		return err;
	}
877
	return err;
878 879
}

880 881 882 883 884 885 886 887 888 889 890 891 892 893 894
static inline void __do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock)
{
	if (flag == F2FS_GET_BLOCK_PRE_AIO) {
		if (lock)
			down_read(&sbi->node_change);
		else
			up_read(&sbi->node_change);
	} else {
		if (lock)
			f2fs_lock_op(sbi);
		else
			f2fs_unlock_op(sbi);
	}
}

J
Jaegeuk Kim 已提交
895
/*
J
Jaegeuk Kim 已提交
896 897
 * f2fs_map_blocks() now supported readahead/bmap/rw direct_IO with
 * f2fs_map_blocks structure.
C
Chao Yu 已提交
898 899 900 901 902
 * If original data blocks are allocated, then give them to blockdev.
 * Otherwise,
 *     a. preallocate requested block addresses
 *     b. do not use extent cache for better performance
 *     c. give the block addresses to blockdev
903
 */
C
Chao Yu 已提交
904
int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map,
C
Chao Yu 已提交
905
						int create, int flag)
906
{
J
Jaegeuk Kim 已提交
907
	unsigned int maxblocks = map->m_len;
908
	struct dnode_of_data dn;
909
	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
910
	int mode = create ? ALLOC_NODE : LOOKUP_NODE;
911
	pgoff_t pgofs, end_offset, end;
912
	int err = 0, ofs = 1;
913 914
	unsigned int ofs_in_node, last_ofs_in_node;
	blkcnt_t prealloc;
915
	struct extent_info ei = {0,0,0};
916
	block_t blkaddr;
917

918 919 920
	if (!maxblocks)
		return 0;

J
Jaegeuk Kim 已提交
921 922 923 924 925
	map->m_len = 0;
	map->m_flags = 0;

	/* it only supports block size == page size */
	pgofs =	(pgoff_t)map->m_lblk;
926
	end = pgofs + maxblocks;
927

928
	if (!create && f2fs_lookup_extent_cache(inode, pgofs, &ei)) {
J
Jaegeuk Kim 已提交
929 930 931
		map->m_pblk = ei.blk + pgofs - ei.fofs;
		map->m_len = min((pgoff_t)maxblocks, ei.fofs + ei.len - pgofs);
		map->m_flags = F2FS_MAP_MAPPED;
932
		goto out;
933
	}
934

C
Chao Yu 已提交
935
next_dnode:
936
	if (create)
937
		__do_map_lock(sbi, flag, true);
938 939 940

	/* When reading holes, we need its node page */
	set_new_dnode(&dn, inode, NULL, NULL, 0);
941
	err = get_dnode_of_data(&dn, pgofs, mode);
942
	if (err) {
C
Chao Yu 已提交
943 944
		if (flag == F2FS_GET_BLOCK_BMAP)
			map->m_pblk = 0;
945
		if (err == -ENOENT) {
946
			err = 0;
947 948 949 950
			if (map->m_next_pgofs)
				*map->m_next_pgofs =
					get_next_page_offset(&dn, pgofs);
		}
951
		goto unlock_out;
952
	}
C
Chao Yu 已提交
953

954
	prealloc = 0;
955
	last_ofs_in_node = ofs_in_node = dn.ofs_in_node;
956
	end_offset = ADDRS_PER_PAGE(dn.node_page, inode);
C
Chao Yu 已提交
957 958

next_block:
959
	blkaddr = datablock_addr(dn.inode, dn.node_page, dn.ofs_in_node);
C
Chao Yu 已提交
960 961

	if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR) {
C
Chao Yu 已提交
962
		if (create) {
963 964
			if (unlikely(f2fs_cp_error(sbi))) {
				err = -EIO;
C
Chao Yu 已提交
965
				goto sync_out;
966
			}
967
			if (flag == F2FS_GET_BLOCK_PRE_AIO) {
968 969 970 971
				if (blkaddr == NULL_ADDR) {
					prealloc++;
					last_ofs_in_node = dn.ofs_in_node;
				}
972
			} else {
973 974
				err = __allocate_data_block(&dn,
							map->m_seg_type);
975
				if (!err)
976
					set_inode_flag(inode, FI_APPEND_WRITE);
977
			}
C
Chao Yu 已提交
978
			if (err)
C
Chao Yu 已提交
979
				goto sync_out;
980
			map->m_flags |= F2FS_MAP_NEW;
C
Chao Yu 已提交
981
			blkaddr = dn.data_blkaddr;
C
Chao Yu 已提交
982
		} else {
C
Chao Yu 已提交
983 984 985 986
			if (flag == F2FS_GET_BLOCK_BMAP) {
				map->m_pblk = 0;
				goto sync_out;
			}
987 988 989 990
			if (flag == F2FS_GET_BLOCK_FIEMAP &&
						blkaddr == NULL_ADDR) {
				if (map->m_next_pgofs)
					*map->m_next_pgofs = pgofs + 1;
C
Chao Yu 已提交
991
				goto sync_out;
992
			}
C
Chao Yu 已提交
993
			if (flag != F2FS_GET_BLOCK_FIEMAP)
C
Chao Yu 已提交
994
				goto sync_out;
C
Chao Yu 已提交
995 996
		}
	}
997

998 999 1000
	if (flag == F2FS_GET_BLOCK_PRE_AIO)
		goto skip;

C
Chao Yu 已提交
1001 1002 1003 1004 1005 1006 1007 1008 1009 1010
	if (map->m_len == 0) {
		/* preallocated unwritten block should be mapped for fiemap. */
		if (blkaddr == NEW_ADDR)
			map->m_flags |= F2FS_MAP_UNWRITTEN;
		map->m_flags |= F2FS_MAP_MAPPED;

		map->m_pblk = blkaddr;
		map->m_len = 1;
	} else if ((map->m_pblk != NEW_ADDR &&
			blkaddr == (map->m_pblk + ofs)) ||
1011
			(map->m_pblk == NEW_ADDR && blkaddr == NEW_ADDR) ||
1012
			flag == F2FS_GET_BLOCK_PRE_DIO) {
C
Chao Yu 已提交
1013 1014 1015 1016 1017
		ofs++;
		map->m_len++;
	} else {
		goto sync_out;
	}
1018

1019
skip:
1020 1021 1022
	dn.ofs_in_node++;
	pgofs++;

1023 1024 1025
	/* preallocate blocks in batch for one dnode page */
	if (flag == F2FS_GET_BLOCK_PRE_AIO &&
			(pgofs == end || dn.ofs_in_node == end_offset)) {
1026

1027 1028 1029 1030
		dn.ofs_in_node = ofs_in_node;
		err = reserve_new_blocks(&dn, prealloc);
		if (err)
			goto sync_out;
1031

1032 1033 1034 1035
		map->m_len += dn.ofs_in_node - ofs_in_node;
		if (prealloc && dn.ofs_in_node != last_ofs_in_node + 1) {
			err = -ENOSPC;
			goto sync_out;
1036
		}
1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047
		dn.ofs_in_node = end_offset;
	}

	if (pgofs >= end)
		goto sync_out;
	else if (dn.ofs_in_node < end_offset)
		goto next_block;

	f2fs_put_dnode(&dn);

	if (create) {
1048
		__do_map_lock(sbi, flag, false);
1049
		f2fs_balance_fs(sbi, dn.node_changed);
1050
	}
1051
	goto next_dnode;
1052

1053
sync_out:
1054
	f2fs_put_dnode(&dn);
1055
unlock_out:
1056
	if (create) {
1057
		__do_map_lock(sbi, flag, false);
1058
		f2fs_balance_fs(sbi, dn.node_changed);
1059
	}
1060
out:
J
Jaegeuk Kim 已提交
1061
	trace_f2fs_map_blocks(inode, map, err);
1062
	return err;
1063 1064
}

J
Jaegeuk Kim 已提交
1065
static int __get_data_block(struct inode *inode, sector_t iblock,
1066
			struct buffer_head *bh, int create, int flag,
1067
			pgoff_t *next_pgofs, int seg_type)
J
Jaegeuk Kim 已提交
1068 1069
{
	struct f2fs_map_blocks map;
1070
	int err;
J
Jaegeuk Kim 已提交
1071 1072 1073

	map.m_lblk = iblock;
	map.m_len = bh->b_size >> inode->i_blkbits;
1074
	map.m_next_pgofs = next_pgofs;
1075
	map.m_seg_type = seg_type;
J
Jaegeuk Kim 已提交
1076

1077 1078
	err = f2fs_map_blocks(inode, &map, create, flag);
	if (!err) {
J
Jaegeuk Kim 已提交
1079 1080
		map_bh(bh, inode->i_sb, map.m_pblk);
		bh->b_state = (bh->b_state & ~F2FS_MAP_FLAGS) | map.m_flags;
1081
		bh->b_size = (u64)map.m_len << inode->i_blkbits;
J
Jaegeuk Kim 已提交
1082
	}
1083
	return err;
J
Jaegeuk Kim 已提交
1084 1085
}

1086
static int get_data_block(struct inode *inode, sector_t iblock,
1087 1088
			struct buffer_head *bh_result, int create, int flag,
			pgoff_t *next_pgofs)
C
Chao Yu 已提交
1089
{
1090
	return __get_data_block(inode, iblock, bh_result, create,
1091 1092
							flag, next_pgofs,
							NO_CHECK_TYPE);
C
Chao Yu 已提交
1093 1094 1095
}

static int get_data_block_dio(struct inode *inode, sector_t iblock,
1096 1097
			struct buffer_head *bh_result, int create)
{
C
Chao Yu 已提交
1098
	return __get_data_block(inode, iblock, bh_result, create,
1099 1100 1101
						F2FS_GET_BLOCK_DEFAULT, NULL,
						rw_hint_to_seg_type(
							inode->i_write_hint));
1102 1103
}

C
Chao Yu 已提交
1104
static int get_data_block_bmap(struct inode *inode, sector_t iblock,
1105 1106
			struct buffer_head *bh_result, int create)
{
1107
	/* Block number less than F2FS MAX BLOCKS */
C
Chao Yu 已提交
1108
	if (unlikely(iblock >= F2FS_I_SB(inode)->max_file_blocks))
1109 1110
		return -EFBIG;

C
Chao Yu 已提交
1111
	return __get_data_block(inode, iblock, bh_result, create,
1112 1113
						F2FS_GET_BLOCK_BMAP, NULL,
						NO_CHECK_TYPE);
1114 1115
}

1116 1117 1118 1119 1120 1121 1122 1123 1124 1125
static inline sector_t logical_to_blk(struct inode *inode, loff_t offset)
{
	return (offset >> inode->i_blkbits);
}

static inline loff_t blk_to_logical(struct inode *inode, sector_t blk)
{
	return (blk << inode->i_blkbits);
}

J
Jaegeuk Kim 已提交
1126 1127 1128
int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
		u64 start, u64 len)
{
1129 1130
	struct buffer_head map_bh;
	sector_t start_blk, last_blk;
1131
	pgoff_t next_pgofs;
1132 1133 1134 1135 1136 1137 1138 1139
	u64 logical = 0, phys = 0, size = 0;
	u32 flags = 0;
	int ret = 0;

	ret = fiemap_check_flags(fieinfo, FIEMAP_FLAG_SYNC);
	if (ret)
		return ret;

J
Jaegeuk Kim 已提交
1140 1141 1142 1143 1144 1145
	if (f2fs_has_inline_data(inode)) {
		ret = f2fs_inline_data_fiemap(inode, fieinfo, start, len);
		if (ret != -EAGAIN)
			return ret;
	}

A
Al Viro 已提交
1146
	inode_lock(inode);
1147

1148 1149 1150 1151 1152
	if (logical_to_blk(inode, len) == 0)
		len = blk_to_logical(inode, 1);

	start_blk = logical_to_blk(inode, start);
	last_blk = logical_to_blk(inode, start + len - 1);
1153

1154 1155 1156 1157
next:
	memset(&map_bh, 0, sizeof(struct buffer_head));
	map_bh.b_size = len;

C
Chao Yu 已提交
1158
	ret = get_data_block(inode, start_blk, &map_bh, 0,
1159
					F2FS_GET_BLOCK_FIEMAP, &next_pgofs);
1160 1161 1162 1163 1164
	if (ret)
		goto out;

	/* HOLE */
	if (!buffer_mapped(&map_bh)) {
1165
		start_blk = next_pgofs;
1166 1167 1168

		if (blk_to_logical(inode, start_blk) < blk_to_logical(inode,
					F2FS_I_SB(inode)->max_file_blocks))
1169
			goto prep_next;
1170

1171 1172
		flags |= FIEMAP_EXTENT_LAST;
	}
1173

1174 1175 1176 1177
	if (size) {
		if (f2fs_encrypted_inode(inode))
			flags |= FIEMAP_EXTENT_DATA_ENCRYPTED;

1178 1179
		ret = fiemap_fill_next_extent(fieinfo, logical,
				phys, size, flags);
1180
	}
1181

1182 1183
	if (start_blk > last_blk || ret)
		goto out;
1184

1185 1186 1187 1188 1189 1190
	logical = blk_to_logical(inode, start_blk);
	phys = blk_to_logical(inode, map_bh.b_blocknr);
	size = map_bh.b_size;
	flags = 0;
	if (buffer_unwritten(&map_bh))
		flags = FIEMAP_EXTENT_UNWRITTEN;
1191

1192
	start_blk += logical_to_blk(inode, size);
1193

1194
prep_next:
1195 1196 1197 1198 1199 1200 1201 1202 1203
	cond_resched();
	if (fatal_signal_pending(current))
		ret = -EINTR;
	else
		goto next;
out:
	if (ret == 1)
		ret = 0;

A
Al Viro 已提交
1204
	inode_unlock(inode);
1205
	return ret;
J
Jaegeuk Kim 已提交
1206 1207
}

J
Jaegeuk Kim 已提交
1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230
/*
 * This function was originally taken from fs/mpage.c, and customized for f2fs.
 * Major change was from block_size == page_size in f2fs by default.
 */
static int f2fs_mpage_readpages(struct address_space *mapping,
			struct list_head *pages, struct page *page,
			unsigned nr_pages)
{
	struct bio *bio = NULL;
	sector_t last_block_in_bio = 0;
	struct inode *inode = mapping->host;
	const unsigned blkbits = inode->i_blkbits;
	const unsigned blocksize = 1 << blkbits;
	sector_t block_in_file;
	sector_t last_block;
	sector_t last_block_in_file;
	sector_t block_nr;
	struct f2fs_map_blocks map;

	map.m_pblk = 0;
	map.m_lblk = 0;
	map.m_len = 0;
	map.m_flags = 0;
1231
	map.m_next_pgofs = NULL;
1232
	map.m_seg_type = NO_CHECK_TYPE;
J
Jaegeuk Kim 已提交
1233

L
LiFan 已提交
1234
	for (; nr_pages; nr_pages--) {
J
Jaegeuk Kim 已提交
1235
		if (pages) {
1236
			page = list_last_entry(pages, struct page, lru);
1237 1238

			prefetchw(&page->flags);
J
Jaegeuk Kim 已提交
1239 1240
			list_del(&page->lru);
			if (add_to_page_cache_lru(page, mapping,
1241 1242
						  page->index,
						  readahead_gfp_mask(mapping)))
J
Jaegeuk Kim 已提交
1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270
				goto next_page;
		}

		block_in_file = (sector_t)page->index;
		last_block = block_in_file + nr_pages;
		last_block_in_file = (i_size_read(inode) + blocksize - 1) >>
								blkbits;
		if (last_block > last_block_in_file)
			last_block = last_block_in_file;

		/*
		 * Map blocks using the previous result first.
		 */
		if ((map.m_flags & F2FS_MAP_MAPPED) &&
				block_in_file > map.m_lblk &&
				block_in_file < (map.m_lblk + map.m_len))
			goto got_it;

		/*
		 * Then do more f2fs_map_blocks() calls until we are
		 * done with this page.
		 */
		map.m_flags = 0;

		if (block_in_file < last_block) {
			map.m_lblk = block_in_file;
			map.m_len = last_block - block_in_file;

1271
			if (f2fs_map_blocks(inode, &map, 0,
1272
						F2FS_GET_BLOCK_DEFAULT))
J
Jaegeuk Kim 已提交
1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284
				goto set_error_page;
		}
got_it:
		if ((map.m_flags & F2FS_MAP_MAPPED)) {
			block_nr = map.m_pblk + block_in_file - map.m_lblk;
			SetPageMappedToDisk(page);

			if (!PageUptodate(page) && !cleancache_get_page(page)) {
				SetPageUptodate(page);
				goto confused;
			}
		} else {
1285
			zero_user_segment(page, 0, PAGE_SIZE);
1286 1287
			if (!PageUptodate(page))
				SetPageUptodate(page);
J
Jaegeuk Kim 已提交
1288 1289 1290 1291 1292 1293 1294 1295
			unlock_page(page);
			goto next_page;
		}

		/*
		 * This page will go to BIO.  Do we need to send this
		 * BIO off first?
		 */
J
Jaegeuk Kim 已提交
1296 1297
		if (bio && (last_block_in_bio != block_nr - 1 ||
			!__same_bdev(F2FS_I_SB(inode), block_nr, bio))) {
J
Jaegeuk Kim 已提交
1298
submit_and_realloc:
1299
			__submit_bio(F2FS_I_SB(inode), bio, DATA);
J
Jaegeuk Kim 已提交
1300 1301 1302
			bio = NULL;
		}
		if (bio == NULL) {
1303
			bio = f2fs_grab_read_bio(inode, block_nr, nr_pages);
J
Jaegeuk Kim 已提交
1304 1305
			if (IS_ERR(bio)) {
				bio = NULL;
J
Jaegeuk Kim 已提交
1306
				goto set_error_page;
1307
			}
J
Jaegeuk Kim 已提交
1308 1309 1310 1311 1312 1313 1314 1315 1316
		}

		if (bio_add_page(bio, page, blocksize, 0) < blocksize)
			goto submit_and_realloc;

		last_block_in_bio = block_nr;
		goto next_page;
set_error_page:
		SetPageError(page);
1317
		zero_user_segment(page, 0, PAGE_SIZE);
J
Jaegeuk Kim 已提交
1318 1319 1320 1321
		unlock_page(page);
		goto next_page;
confused:
		if (bio) {
1322
			__submit_bio(F2FS_I_SB(inode), bio, DATA);
J
Jaegeuk Kim 已提交
1323 1324 1325 1326 1327
			bio = NULL;
		}
		unlock_page(page);
next_page:
		if (pages)
1328
			put_page(page);
J
Jaegeuk Kim 已提交
1329 1330 1331
	}
	BUG_ON(pages && !list_empty(pages));
	if (bio)
1332
		__submit_bio(F2FS_I_SB(inode), bio, DATA);
J
Jaegeuk Kim 已提交
1333 1334 1335
	return 0;
}

1336 1337
static int f2fs_read_data_page(struct file *file, struct page *page)
{
H
Huajun Li 已提交
1338
	struct inode *inode = page->mapping->host;
1339
	int ret = -EAGAIN;
H
Huajun Li 已提交
1340

1341 1342
	trace_f2fs_readpage(page, DATA);

A
arter97 已提交
1343
	/* If the file has inline data, try to read it directly */
H
Huajun Li 已提交
1344 1345
	if (f2fs_has_inline_data(inode))
		ret = f2fs_read_inline_data(inode, page);
1346
	if (ret == -EAGAIN)
J
Jaegeuk Kim 已提交
1347
		ret = f2fs_mpage_readpages(page->mapping, NULL, page, 1);
H
Huajun Li 已提交
1348
	return ret;
1349 1350 1351 1352 1353 1354
}

static int f2fs_read_data_pages(struct file *file,
			struct address_space *mapping,
			struct list_head *pages, unsigned nr_pages)
{
1355
	struct inode *inode = mapping->host;
1356
	struct page *page = list_last_entry(pages, struct page, lru);
1357 1358

	trace_f2fs_readpages(inode, page, nr_pages);
H
Huajun Li 已提交
1359 1360 1361 1362 1363

	/* If the file has inline data, skip readpages */
	if (f2fs_has_inline_data(inode))
		return 0;

J
Jaegeuk Kim 已提交
1364
	return f2fs_mpage_readpages(mapping, pages, NULL, nr_pages);
1365 1366
}

1367 1368 1369 1370 1371
static int encrypt_one_page(struct f2fs_io_info *fio)
{
	struct inode *inode = fio->page->mapping->host;
	gfp_t gfp_flags = GFP_NOFS;

1372
	if (!f2fs_encrypted_file(inode))
1373 1374 1375
		return 0;

	/* wait for GCed encrypted page writeback */
1376
	f2fs_wait_on_block_writeback(fio->sbi, fio->old_blkaddr);
1377 1378 1379 1380 1381 1382 1383 1384 1385

retry_encrypt:
	fio->encrypted_page = fscrypt_encrypt_page(inode, fio->page,
			PAGE_SIZE, 0, fio->page->index, gfp_flags);
	if (!IS_ERR(fio->encrypted_page))
		return 0;

	/* flush pending IOs and wait for a while in the ENOMEM case */
	if (PTR_ERR(fio->encrypted_page) == -ENOMEM) {
1386
		f2fs_flush_merged_writes(fio->sbi);
1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407
		congestion_wait(BLK_RW_ASYNC, HZ/50);
		gfp_flags |= __GFP_NOFAIL;
		goto retry_encrypt;
	}
	return PTR_ERR(fio->encrypted_page);
}

static inline bool need_inplace_update(struct f2fs_io_info *fio)
{
	struct inode *inode = fio->page->mapping->host;

	if (S_ISDIR(inode->i_mode) || f2fs_is_atomic_file(inode))
		return false;
	if (is_cold_data(fio->page))
		return false;
	if (IS_ATOMIC_WRITTEN_PAGE(fio->page))
		return false;

	return need_inplace_update_policy(inode, fio);
}

1408 1409 1410 1411 1412 1413 1414 1415 1416
static inline bool valid_ipu_blkaddr(struct f2fs_io_info *fio)
{
	if (fio->old_blkaddr == NEW_ADDR)
		return false;
	if (fio->old_blkaddr == NULL_ADDR)
		return false;
	return true;
}

1417
int do_write_data_page(struct f2fs_io_info *fio)
1418
{
1419
	struct page *page = fio->page;
1420 1421
	struct inode *inode = page->mapping->host;
	struct dnode_of_data dn;
1422 1423
	struct extent_info ei = {0,0,0};
	bool ipu_force = false;
1424 1425 1426
	int err = 0;

	set_new_dnode(&dn, inode, NULL, NULL, 0);
1427 1428 1429
	if (need_inplace_update(fio) &&
			f2fs_lookup_extent_cache(inode, page->index, &ei)) {
		fio->old_blkaddr = ei.blk + page->index - ei.fofs;
1430 1431

		if (valid_ipu_blkaddr(fio)) {
1432
			ipu_force = true;
1433
			fio->need_lock = LOCK_DONE;
1434 1435 1436
			goto got_it;
		}
	}
1437

1438 1439 1440
	/* Deadlock due to between page->lock and f2fs_lock_op */
	if (fio->need_lock == LOCK_REQ && !f2fs_trylock_op(fio->sbi))
		return -EAGAIN;
1441

1442
	err = get_dnode_of_data(&dn, page->index, LOOKUP_NODE);
1443
	if (err)
1444
		goto out;
1445

1446
	fio->old_blkaddr = dn.data_blkaddr;
1447 1448

	/* This page is already truncated */
1449
	if (fio->old_blkaddr == NULL_ADDR) {
1450
		ClearPageUptodate(page);
1451
		goto out_writepage;
1452
	}
1453
got_it:
1454 1455 1456 1457
	/*
	 * If current allocation needs SSR,
	 * it had better in-place writes for updated data.
	 */
1458
	if (ipu_force || (valid_ipu_blkaddr(fio) && need_inplace_update(fio))) {
1459 1460 1461 1462 1463
		err = encrypt_one_page(fio);
		if (err)
			goto out_writepage;

		set_page_writeback(page);
1464
		f2fs_put_dnode(&dn);
1465
		if (fio->need_lock == LOCK_REQ)
1466
			f2fs_unlock_op(fio->sbi);
1467
		err = rewrite_data_page(fio);
1468
		trace_f2fs_do_write_data_page(fio->page, IPU);
1469
		set_inode_flag(inode, FI_UPDATE_WRITE);
1470
		return err;
1471
	}
1472

1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486
	if (fio->need_lock == LOCK_RETRY) {
		if (!f2fs_trylock_op(fio->sbi)) {
			err = -EAGAIN;
			goto out_writepage;
		}
		fio->need_lock = LOCK_REQ;
	}

	err = encrypt_one_page(fio);
	if (err)
		goto out_writepage;

	set_page_writeback(page);

1487 1488 1489 1490 1491 1492
	/* LFS mode write path */
	write_data_page(&dn, fio);
	trace_f2fs_do_write_data_page(page, OPU);
	set_inode_flag(inode, FI_APPEND_WRITE);
	if (page->index == 0)
		set_inode_flag(inode, FI_FIRST_BLOCK_WRITTEN);
1493 1494
out_writepage:
	f2fs_put_dnode(&dn);
1495
out:
1496
	if (fio->need_lock == LOCK_REQ)
1497
		f2fs_unlock_op(fio->sbi);
1498 1499 1500
	return err;
}

1501
static int __write_data_page(struct page *page, bool *submitted,
C
Chao Yu 已提交
1502 1503
				struct writeback_control *wbc,
				enum iostat_type io_type)
1504 1505
{
	struct inode *inode = page->mapping->host;
1506
	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1507 1508
	loff_t i_size = i_size_read(inode);
	const pgoff_t end_index = ((unsigned long long) i_size)
1509
							>> PAGE_SHIFT;
1510
	loff_t psize = (page->index + 1) << PAGE_SHIFT;
H
Huajun Li 已提交
1511
	unsigned offset = 0;
1512
	bool need_balance_fs = false;
1513
	int err = 0;
J
Jaegeuk Kim 已提交
1514
	struct f2fs_io_info fio = {
1515
		.sbi = sbi,
C
Chao Yu 已提交
1516
		.ino = inode->i_ino,
J
Jaegeuk Kim 已提交
1517
		.type = DATA,
M
Mike Christie 已提交
1518
		.op = REQ_OP_WRITE,
J
Jens Axboe 已提交
1519
		.op_flags = wbc_to_write_flags(wbc),
1520
		.old_blkaddr = NULL_ADDR,
1521
		.page = page,
1522
		.encrypted_page = NULL,
1523
		.submitted = false,
1524
		.need_lock = LOCK_RETRY,
C
Chao Yu 已提交
1525
		.io_type = io_type,
J
Jaegeuk Kim 已提交
1526
	};
1527

1528 1529
	trace_f2fs_writepage(page, DATA);

1530 1531 1532
	if (unlikely(is_sbi_flag_set(sbi, SBI_POR_DOING)))
		goto redirty_out;

1533
	if (page->index < end_index)
1534
		goto write;
1535 1536 1537 1538 1539

	/*
	 * If the offset is out-of-range of file size,
	 * this page does not have to be written to disk.
	 */
1540
	offset = i_size & (PAGE_SIZE - 1);
1541
	if ((page->index >= end_index + 1) || !offset)
1542
		goto out;
1543

1544
	zero_user_segment(page, offset, PAGE_SIZE);
1545
write:
1546 1547
	if (f2fs_is_drop_cache(inode))
		goto out;
1548 1549 1550 1551
	/* we should not write 0'th page having journal header */
	if (f2fs_is_volatile_file(inode) && (!page->index ||
			(!wbc->for_reclaim &&
			available_free_memory(sbi, BASE_CHECK))))
1552
		goto redirty_out;
1553

1554 1555
	/* we should bypass data pages to proceed the kworkder jobs */
	if (unlikely(f2fs_cp_error(sbi))) {
J
Jaegeuk Kim 已提交
1556
		mapping_set_error(page->mapping, -EIO);
1557
		goto out;
1558 1559
	}

1560
	/* Dentry blocks are controlled by checkpoint */
1561
	if (S_ISDIR(inode->i_mode)) {
1562
		fio.need_lock = LOCK_DONE;
1563
		err = do_write_data_page(&fio);
1564 1565
		goto done;
	}
H
Huajun Li 已提交
1566

1567
	if (!wbc->for_reclaim)
1568
		need_balance_fs = true;
1569
	else if (has_not_enough_free_secs(sbi, 0, 0))
1570
		goto redirty_out;
1571 1572
	else
		set_inode_flag(inode, FI_HOT_DATA);
1573

1574
	err = -EAGAIN;
1575
	if (f2fs_has_inline_data(inode)) {
1576
		err = f2fs_write_inline_data(inode, page);
1577 1578 1579
		if (!err)
			goto out;
	}
1580

1581
	if (err == -EAGAIN) {
1582
		err = do_write_data_page(&fio);
1583 1584 1585 1586 1587
		if (err == -EAGAIN) {
			fio.need_lock = LOCK_REQ;
			err = do_write_data_page(&fio);
		}
	}
1588 1589

	down_write(&F2FS_I(inode)->i_sem);
1590 1591
	if (F2FS_I(inode)->last_disk_size < psize)
		F2FS_I(inode)->last_disk_size = psize;
1592
	up_write(&F2FS_I(inode)->i_sem);
1593

1594 1595 1596
done:
	if (err && err != -ENOENT)
		goto redirty_out;
1597

1598
out:
1599
	inode_dec_dirty_pages(inode);
1600 1601
	if (err)
		ClearPageUptodate(page);
1602 1603

	if (wbc->for_reclaim) {
1604
		f2fs_submit_merged_write_cond(sbi, inode, 0, page->index, DATA);
1605
		clear_inode_flag(inode, FI_HOT_DATA);
1606
		remove_dirty_inode(inode);
1607
		submitted = NULL;
1608 1609
	}

1610
	unlock_page(page);
J
Jaegeuk Kim 已提交
1611 1612
	if (!S_ISDIR(inode->i_mode))
		f2fs_balance_fs(sbi, need_balance_fs);
1613

1614
	if (unlikely(f2fs_cp_error(sbi))) {
1615
		f2fs_submit_merged_write(sbi, DATA);
1616 1617 1618 1619 1620
		submitted = NULL;
	}

	if (submitted)
		*submitted = fio.submitted;
1621

1622 1623 1624
	return 0;

redirty_out:
1625
	redirty_page_for_writepage(wbc, page);
1626 1627
	if (!err)
		return AOP_WRITEPAGE_ACTIVATE;
J
Jaegeuk Kim 已提交
1628 1629
	unlock_page(page);
	return err;
1630 1631
}

1632 1633 1634
static int f2fs_write_data_page(struct page *page,
					struct writeback_control *wbc)
{
C
Chao Yu 已提交
1635
	return __write_data_page(page, NULL, wbc, FS_DATA_IO);
1636 1637
}

C
Chao Yu 已提交
1638 1639 1640 1641 1642 1643
/*
 * This function was copied from write_cche_pages from mm/page-writeback.c.
 * The major change is making write step of cold data page separately from
 * warm/hot data page.
 */
static int f2fs_write_cache_pages(struct address_space *mapping,
C
Chao Yu 已提交
1644 1645
					struct writeback_control *wbc,
					enum iostat_type io_type)
C
Chao Yu 已提交
1646 1647 1648 1649 1650 1651 1652 1653 1654
{
	int ret = 0;
	int done = 0;
	struct pagevec pvec;
	int nr_pages;
	pgoff_t uninitialized_var(writeback_index);
	pgoff_t index;
	pgoff_t end;		/* Inclusive */
	pgoff_t done_index;
1655
	pgoff_t last_idx = ULONG_MAX;
C
Chao Yu 已提交
1656 1657 1658 1659
	int cycled;
	int range_whole = 0;
	int tag;

1660
	pagevec_init(&pvec);
1661

1662 1663 1664 1665 1666 1667
	if (get_dirty_pages(mapping->host) <=
				SM_I(F2FS_M_SB(mapping))->min_hot_blocks)
		set_inode_flag(mapping->host, FI_HOT_DATA);
	else
		clear_inode_flag(mapping->host, FI_HOT_DATA);

C
Chao Yu 已提交
1668 1669 1670 1671 1672 1673 1674 1675 1676
	if (wbc->range_cyclic) {
		writeback_index = mapping->writeback_index; /* prev offset */
		index = writeback_index;
		if (index == 0)
			cycled = 1;
		else
			cycled = 0;
		end = -1;
	} else {
1677 1678
		index = wbc->range_start >> PAGE_SHIFT;
		end = wbc->range_end >> PAGE_SHIFT;
C
Chao Yu 已提交
1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693
		if (wbc->range_start == 0 && wbc->range_end == LLONG_MAX)
			range_whole = 1;
		cycled = 1; /* ignore range_cyclic tests */
	}
	if (wbc->sync_mode == WB_SYNC_ALL || wbc->tagged_writepages)
		tag = PAGECACHE_TAG_TOWRITE;
	else
		tag = PAGECACHE_TAG_DIRTY;
retry:
	if (wbc->sync_mode == WB_SYNC_ALL || wbc->tagged_writepages)
		tag_pages_for_writeback(mapping, index, end);
	done_index = index;
	while (!done && (index <= end)) {
		int i;

J
Jan Kara 已提交
1694
		nr_pages = pagevec_lookup_range_tag(&pvec, mapping, &index, end,
1695
				tag);
C
Chao Yu 已提交
1696 1697 1698 1699 1700
		if (nr_pages == 0)
			break;

		for (i = 0; i < nr_pages; i++) {
			struct page *page = pvec.pages[i];
1701
			bool submitted = false;
C
Chao Yu 已提交
1702 1703

			done_index = page->index;
1704
retry_write:
C
Chao Yu 已提交
1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719
			lock_page(page);

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

			if (!PageDirty(page)) {
				/* someone wrote it for us */
				goto continue_unlock;
			}

			if (PageWriteback(page)) {
				if (wbc->sync_mode != WB_SYNC_NONE)
1720 1721
					f2fs_wait_on_page_writeback(page,
								DATA, true);
C
Chao Yu 已提交
1722 1723 1724 1725 1726 1727 1728 1729
				else
					goto continue_unlock;
			}

			BUG_ON(PageWriteback(page));
			if (!clear_page_dirty_for_io(page))
				goto continue_unlock;

C
Chao Yu 已提交
1730
			ret = __write_data_page(page, &submitted, wbc, io_type);
C
Chao Yu 已提交
1731
			if (unlikely(ret)) {
1732 1733 1734 1735 1736 1737 1738 1739
				/*
				 * keep nr_to_write, since vfs uses this to
				 * get # of written pages.
				 */
				if (ret == AOP_WRITEPAGE_ACTIVATE) {
					unlock_page(page);
					ret = 0;
					continue;
1740 1741 1742 1743 1744 1745 1746 1747 1748
				} else if (ret == -EAGAIN) {
					ret = 0;
					if (wbc->sync_mode == WB_SYNC_ALL) {
						cond_resched();
						congestion_wait(BLK_RW_ASYNC,
									HZ/50);
						goto retry_write;
					}
					continue;
1749
				}
J
Jaegeuk Kim 已提交
1750 1751 1752
				done_index = page->index + 1;
				done = 1;
				break;
1753
			} else if (submitted) {
1754
				last_idx = page->index;
C
Chao Yu 已提交
1755 1756
			}

1757 1758 1759 1760
			/* give a priority to WB_SYNC threads */
			if ((atomic_read(&F2FS_M_SB(mapping)->wb_sync_req) ||
					--wbc->nr_to_write <= 0) &&
					wbc->sync_mode == WB_SYNC_NONE) {
C
Chao Yu 已提交
1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777
				done = 1;
				break;
			}
		}
		pagevec_release(&pvec);
		cond_resched();
	}

	if (!cycled && !done) {
		cycled = 1;
		index = 0;
		end = writeback_index - 1;
		goto retry;
	}
	if (wbc->range_cyclic || (range_whole && wbc->nr_to_write > 0))
		mapping->writeback_index = done_index;

1778
	if (last_idx != ULONG_MAX)
1779 1780
		f2fs_submit_merged_write_cond(F2FS_M_SB(mapping), mapping->host,
						0, last_idx, DATA);
C
Chao Yu 已提交
1781

C
Chao Yu 已提交
1782 1783 1784
	return ret;
}

C
Chao Yu 已提交
1785 1786 1787
int __f2fs_write_data_pages(struct address_space *mapping,
						struct writeback_control *wbc,
						enum iostat_type io_type)
1788 1789
{
	struct inode *inode = mapping->host;
1790
	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1791
	struct blk_plug plug;
1792 1793
	int ret;

P
P J P 已提交
1794 1795 1796 1797
	/* deal with chardevs and other special file */
	if (!mapping->a_ops->writepage)
		return 0;

1798 1799 1800 1801
	/* skip writing if there is no dirty page in this inode */
	if (!get_dirty_pages(inode) && wbc->sync_mode == WB_SYNC_NONE)
		return 0;

1802 1803 1804 1805
	/* during POR, we don't need to trigger writepage at all. */
	if (unlikely(is_sbi_flag_set(sbi, SBI_POR_DOING)))
		goto skip_write;

1806 1807 1808 1809 1810
	if (S_ISDIR(inode->i_mode) && wbc->sync_mode == WB_SYNC_NONE &&
			get_dirty_pages(inode) < nr_pages_to_skip(sbi, DATA) &&
			available_free_memory(sbi, DIRTY_DENTS))
		goto skip_write;

C
Chao Yu 已提交
1811
	/* skip writing during file defragment */
1812
	if (is_inode_flag_set(inode, FI_DO_DEFRAG))
C
Chao Yu 已提交
1813 1814
		goto skip_write;

Y
Yunlei He 已提交
1815 1816
	trace_f2fs_writepages(mapping->host, wbc, DATA);

1817 1818 1819 1820 1821 1822
	/* to avoid spliting IOs due to mixed WB_SYNC_ALL and WB_SYNC_NONE */
	if (wbc->sync_mode == WB_SYNC_ALL)
		atomic_inc(&sbi->wb_sync_req);
	else if (atomic_read(&sbi->wb_sync_req))
		goto skip_write;

1823
	blk_start_plug(&plug);
C
Chao Yu 已提交
1824
	ret = f2fs_write_cache_pages(mapping, wbc, io_type);
1825
	blk_finish_plug(&plug);
1826 1827 1828

	if (wbc->sync_mode == WB_SYNC_ALL)
		atomic_dec(&sbi->wb_sync_req);
1829 1830 1831 1832
	/*
	 * if some pages were truncated, we cannot guarantee its mapping->host
	 * to detect pending bios.
	 */
J
Jaegeuk Kim 已提交
1833

1834
	remove_dirty_inode(inode);
1835
	return ret;
1836 1837

skip_write:
1838
	wbc->pages_skipped += get_dirty_pages(inode);
Y
Yunlei He 已提交
1839
	trace_f2fs_writepages(mapping->host, wbc, DATA);
1840
	return 0;
1841 1842
}

C
Chao Yu 已提交
1843 1844 1845 1846 1847 1848 1849 1850 1851 1852
static int f2fs_write_data_pages(struct address_space *mapping,
			    struct writeback_control *wbc)
{
	struct inode *inode = mapping->host;

	return __f2fs_write_data_pages(mapping, wbc,
			F2FS_I(inode)->cp_task == current ?
			FS_CP_DATA_IO : FS_DATA_IO);
}

1853 1854 1855
static void f2fs_write_failed(struct address_space *mapping, loff_t to)
{
	struct inode *inode = mapping->host;
J
Jaegeuk Kim 已提交
1856
	loff_t i_size = i_size_read(inode);
1857

J
Jaegeuk Kim 已提交
1858
	if (to > i_size) {
1859
		down_write(&F2FS_I(inode)->i_mmap_sem);
J
Jaegeuk Kim 已提交
1860 1861
		truncate_pagecache(inode, i_size);
		truncate_blocks(inode, i_size, true);
1862
		up_write(&F2FS_I(inode)->i_mmap_sem);
1863 1864 1865
	}
}

1866 1867 1868 1869 1870 1871 1872 1873
static int prepare_write_begin(struct f2fs_sb_info *sbi,
			struct page *page, loff_t pos, unsigned len,
			block_t *blk_addr, bool *node_changed)
{
	struct inode *inode = page->mapping->host;
	pgoff_t index = page->index;
	struct dnode_of_data dn;
	struct page *ipage;
1874
	bool locked = false;
1875
	struct extent_info ei = {0,0,0};
1876 1877
	int err = 0;

1878 1879 1880 1881
	/*
	 * we already allocated all the blocks, so we don't need to get
	 * the block addresses when there is no need to fill the page.
	 */
1882 1883
	if (!f2fs_has_inline_data(inode) && len == PAGE_SIZE &&
			!is_inode_flag_set(inode, FI_NO_PREALLOC))
1884 1885
		return 0;

1886
	if (f2fs_has_inline_data(inode) ||
1887
			(pos & PAGE_MASK) >= i_size_read(inode)) {
1888
		__do_map_lock(sbi, F2FS_GET_BLOCK_PRE_AIO, true);
1889 1890 1891
		locked = true;
	}
restart:
1892 1893 1894 1895 1896 1897 1898 1899 1900 1901
	/* check inline_data */
	ipage = get_node_page(sbi, inode->i_ino);
	if (IS_ERR(ipage)) {
		err = PTR_ERR(ipage);
		goto unlock_out;
	}

	set_new_dnode(&dn, inode, ipage, ipage, 0);

	if (f2fs_has_inline_data(inode)) {
C
Chao Yu 已提交
1902
		if (pos + len <= MAX_INLINE_DATA(inode)) {
1903
			read_inline_data(page, ipage);
1904
			set_inode_flag(inode, FI_DATA_EXIST);
1905 1906
			if (inode->i_nlink)
				set_inline_node(ipage);
1907 1908 1909
		} else {
			err = f2fs_convert_inline_page(&dn, page);
			if (err)
1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921
				goto out;
			if (dn.data_blkaddr == NULL_ADDR)
				err = f2fs_get_block(&dn, index);
		}
	} else if (locked) {
		err = f2fs_get_block(&dn, index);
	} else {
		if (f2fs_lookup_extent_cache(inode, index, &ei)) {
			dn.data_blkaddr = ei.blk + index - ei.fofs;
		} else {
			/* hole case */
			err = get_dnode_of_data(&dn, index, LOOKUP_NODE);
1922
			if (err || dn.data_blkaddr == NULL_ADDR) {
1923
				f2fs_put_dnode(&dn);
1924 1925
				__do_map_lock(sbi, F2FS_GET_BLOCK_PRE_AIO,
								true);
1926 1927 1928
				locked = true;
				goto restart;
			}
1929 1930
		}
	}
1931

1932 1933 1934
	/* convert_inline_page can make node_changed */
	*blk_addr = dn.data_blkaddr;
	*node_changed = dn.node_changed;
1935
out:
1936 1937
	f2fs_put_dnode(&dn);
unlock_out:
1938
	if (locked)
1939
		__do_map_lock(sbi, F2FS_GET_BLOCK_PRE_AIO, false);
1940 1941 1942
	return err;
}

1943 1944 1945 1946 1947
static int f2fs_write_begin(struct file *file, struct address_space *mapping,
		loff_t pos, unsigned len, unsigned flags,
		struct page **pagep, void **fsdata)
{
	struct inode *inode = mapping->host;
1948
	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1949
	struct page *page = NULL;
1950
	pgoff_t index = ((unsigned long long) pos) >> PAGE_SHIFT;
1951 1952
	bool need_balance = false;
	block_t blkaddr = NULL_ADDR;
1953 1954
	int err = 0;

1955 1956
	trace_f2fs_write_begin(inode, pos, len, flags);

J
Jaegeuk Kim 已提交
1957 1958 1959 1960 1961 1962
	if (f2fs_is_atomic_file(inode) &&
			!available_free_memory(sbi, INMEM_PAGES)) {
		err = -ENOMEM;
		goto fail;
	}

1963 1964 1965 1966 1967 1968 1969 1970 1971 1972
	/*
	 * We should check this at this moment to avoid deadlock on inode page
	 * and #0 page. The locking rule for inline_data conversion should be:
	 * lock_page(page #0) -> lock_page(inode_page)
	 */
	if (index != 0) {
		err = f2fs_convert_inline_inode(inode);
		if (err)
			goto fail;
	}
1973
repeat:
1974 1975 1976 1977
	/*
	 * Do not use grab_cache_page_write_begin() to avoid deadlock due to
	 * wait_for_stable_page. Will wait that below with our IO control.
	 */
C
Chao Yu 已提交
1978
	page = f2fs_pagecache_get_page(mapping, index,
1979
				FGP_LOCK | FGP_WRITE | FGP_CREAT, GFP_NOFS);
1980 1981 1982 1983
	if (!page) {
		err = -ENOMEM;
		goto fail;
	}
1984

1985 1986
	*pagep = page;

1987 1988
	err = prepare_write_begin(sbi, page, pos, len,
					&blkaddr, &need_balance);
1989
	if (err)
1990
		goto fail;
1991

1992
	if (need_balance && has_not_enough_free_secs(sbi, 0, 0)) {
1993
		unlock_page(page);
J
Jaegeuk Kim 已提交
1994
		f2fs_balance_fs(sbi, true);
1995 1996 1997 1998 1999 2000 2001 2002
		lock_page(page);
		if (page->mapping != mapping) {
			/* The page got truncated from under us */
			f2fs_put_page(page, 1);
			goto repeat;
		}
	}

2003
	f2fs_wait_on_page_writeback(page, DATA, false);
2004

2005
	/* wait for GCed encrypted page writeback */
2006
	if (f2fs_encrypted_file(inode))
2007
		f2fs_wait_on_block_writeback(sbi, blkaddr);
2008

2009 2010
	if (len == PAGE_SIZE || PageUptodate(page))
		return 0;
2011

2012 2013 2014 2015 2016
	if (!(pos & (PAGE_SIZE - 1)) && (pos + len) >= i_size_read(inode)) {
		zero_user_segment(page, len, PAGE_SIZE);
		return 0;
	}

2017
	if (blkaddr == NEW_ADDR) {
2018
		zero_user_segment(page, 0, PAGE_SIZE);
2019
		SetPageUptodate(page);
2020
	} else {
2021 2022
		err = f2fs_submit_page_read(inode, page, blkaddr);
		if (err)
2023
			goto fail;
2024

2025
		lock_page(page);
2026
		if (unlikely(page->mapping != mapping)) {
2027 2028
			f2fs_put_page(page, 1);
			goto repeat;
2029
		}
2030 2031 2032
		if (unlikely(!PageUptodate(page))) {
			err = -EIO;
			goto fail;
2033
		}
2034 2035
	}
	return 0;
2036

2037
fail:
2038
	f2fs_put_page(page, 1);
2039
	f2fs_write_failed(mapping, pos + len);
J
Jaegeuk Kim 已提交
2040 2041
	if (f2fs_is_atomic_file(inode))
		drop_inmem_pages_all(sbi);
2042
	return err;
2043 2044
}

2045 2046 2047 2048 2049 2050 2051
static int f2fs_write_end(struct file *file,
			struct address_space *mapping,
			loff_t pos, unsigned len, unsigned copied,
			struct page *page, void *fsdata)
{
	struct inode *inode = page->mapping->host;

2052 2053
	trace_f2fs_write_end(inode, pos, len, copied);

2054 2055 2056 2057 2058 2059
	/*
	 * This should be come from len == PAGE_SIZE, and we expect copied
	 * should be PAGE_SIZE. Otherwise, we treat it with zero copied and
	 * let generic_perform_write() try to copy data again through copied=0.
	 */
	if (!PageUptodate(page)) {
2060
		if (unlikely(copied != len))
2061 2062 2063 2064 2065 2066 2067
			copied = 0;
		else
			SetPageUptodate(page);
	}
	if (!copied)
		goto unlock_out;

2068
	set_page_dirty(page);
2069

2070 2071
	if (pos + copied > i_size_read(inode))
		f2fs_i_size_write(inode, pos + copied);
2072
unlock_out:
2073
	f2fs_put_page(page, 1);
2074
	f2fs_update_time(F2FS_I_SB(inode), REQ_TIME);
2075 2076 2077
	return copied;
}

2078 2079
static int check_direct_IO(struct inode *inode, struct iov_iter *iter,
			   loff_t offset)
2080 2081 2082 2083 2084 2085
{
	unsigned blocksize_mask = inode->i_sb->s_blocksize - 1;

	if (offset & blocksize_mask)
		return -EINVAL;

A
Al Viro 已提交
2086 2087 2088
	if (iov_iter_alignment(iter) & blocksize_mask)
		return -EINVAL;

2089 2090 2091
	return 0;
}

2092
static ssize_t f2fs_direct_IO(struct kiocb *iocb, struct iov_iter *iter)
2093
{
2094
	struct address_space *mapping = iocb->ki_filp->f_mapping;
2095 2096
	struct inode *inode = mapping->host;
	size_t count = iov_iter_count(iter);
2097
	loff_t offset = iocb->ki_pos;
2098
	int rw = iov_iter_rw(iter);
2099
	int err;
2100

2101
	err = check_direct_IO(inode, iter, offset);
2102 2103
	if (err)
		return err;
H
Huajun Li 已提交
2104

J
Jaegeuk Kim 已提交
2105
	if (__force_buffered_io(inode, rw))
2106
		return 0;
2107

2108
	trace_f2fs_direct_IO_enter(inode, offset, count, rw);
2109

2110
	down_read(&F2FS_I(inode)->dio_rwsem[rw]);
2111
	err = blockdev_direct_IO(iocb, inode, iter, get_data_block_dio);
2112 2113 2114
	up_read(&F2FS_I(inode)->dio_rwsem[rw]);

	if (rw == WRITE) {
C
Chao Yu 已提交
2115 2116 2117
		if (err > 0) {
			f2fs_update_iostat(F2FS_I_SB(inode), APP_DIRECT_IO,
									err);
2118
			set_inode_flag(inode, FI_UPDATE_WRITE);
C
Chao Yu 已提交
2119
		} else if (err < 0) {
2120
			f2fs_write_failed(mapping, offset + count);
C
Chao Yu 已提交
2121
		}
2122
	}
2123

2124
	trace_f2fs_direct_IO_exit(inode, offset, count, rw, err);
2125

2126
	return err;
2127 2128
}

2129 2130
void f2fs_invalidate_page(struct page *page, unsigned int offset,
							unsigned int length)
2131 2132
{
	struct inode *inode = page->mapping->host;
2133
	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
2134

2135
	if (inode->i_ino >= F2FS_ROOT_INO(sbi) &&
2136
		(offset % PAGE_SIZE || length != PAGE_SIZE))
2137 2138
		return;

2139
	if (PageDirty(page)) {
2140
		if (inode->i_ino == F2FS_META_INO(sbi)) {
2141
			dec_page_count(sbi, F2FS_DIRTY_META);
2142
		} else if (inode->i_ino == F2FS_NODE_INO(sbi)) {
2143
			dec_page_count(sbi, F2FS_DIRTY_NODES);
2144
		} else {
2145
			inode_dec_dirty_pages(inode);
2146 2147
			remove_dirty_inode(inode);
		}
2148
	}
C
Chao Yu 已提交
2149 2150 2151

	/* This is atomic written page, keep Private */
	if (IS_ATOMIC_WRITTEN_PAGE(page))
2152
		return drop_inmem_page(inode, page);
C
Chao Yu 已提交
2153

2154
	set_page_private(page, 0);
2155 2156 2157
	ClearPagePrivate(page);
}

2158
int f2fs_release_page(struct page *page, gfp_t wait)
2159
{
2160 2161 2162 2163
	/* If this is dirty page, keep PagePrivate */
	if (PageDirty(page))
		return 0;

C
Chao Yu 已提交
2164 2165 2166 2167
	/* This is atomic written page, keep Private */
	if (IS_ATOMIC_WRITTEN_PAGE(page))
		return 0;

2168
	set_page_private(page, 0);
2169
	ClearPagePrivate(page);
2170
	return 1;
2171 2172
}

2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199 2200 2201
/*
 * This was copied from __set_page_dirty_buffers which gives higher performance
 * in very high speed storages. (e.g., pmem)
 */
void f2fs_set_page_dirty_nobuffers(struct page *page)
{
	struct address_space *mapping = page->mapping;
	unsigned long flags;

	if (unlikely(!mapping))
		return;

	spin_lock(&mapping->private_lock);
	lock_page_memcg(page);
	SetPageDirty(page);
	spin_unlock(&mapping->private_lock);

	spin_lock_irqsave(&mapping->tree_lock, flags);
	WARN_ON_ONCE(!PageUptodate(page));
	account_page_dirtied(page, mapping);
	radix_tree_tag_set(&mapping->page_tree,
			page_index(page), PAGECACHE_TAG_DIRTY);
	spin_unlock_irqrestore(&mapping->tree_lock, flags);
	unlock_page_memcg(page);

	__mark_inode_dirty(mapping->host, I_DIRTY_PAGES);
	return;
}

2202 2203 2204 2205 2206
static int f2fs_set_data_page_dirty(struct page *page)
{
	struct address_space *mapping = page->mapping;
	struct inode *inode = mapping->host;

2207 2208
	trace_f2fs_set_page_dirty(page, DATA);

2209 2210
	if (!PageUptodate(page))
		SetPageUptodate(page);
2211

C
Chao Yu 已提交
2212
	if (f2fs_is_atomic_file(inode) && !f2fs_is_commit_atomic_write(inode)) {
C
Chao Yu 已提交
2213 2214 2215 2216 2217 2218 2219 2220 2221
		if (!IS_ATOMIC_WRITTEN_PAGE(page)) {
			register_inmem_page(inode, page);
			return 1;
		}
		/*
		 * Previously, this page has been registered, we just
		 * return here.
		 */
		return 0;
2222 2223
	}

2224
	if (!PageDirty(page)) {
2225
		f2fs_set_page_dirty_nobuffers(page);
2226
		update_dirty_page(inode, page);
2227 2228 2229 2230 2231
		return 1;
	}
	return 0;
}

J
Jaegeuk Kim 已提交
2232 2233
static sector_t f2fs_bmap(struct address_space *mapping, sector_t block)
{
2234 2235
	struct inode *inode = mapping->host;

J
Jaegeuk Kim 已提交
2236 2237 2238 2239 2240 2241 2242
	if (f2fs_has_inline_data(inode))
		return 0;

	/* make sure allocating whole blocks */
	if (mapping_tagged(mapping, PAGECACHE_TAG_DIRTY))
		filemap_write_and_wait(mapping);

C
Chao Yu 已提交
2243
	return generic_block_bmap(mapping, block, get_data_block_bmap);
2244 2245
}

2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258
#ifdef CONFIG_MIGRATION
#include <linux/migrate.h>

int f2fs_migrate_page(struct address_space *mapping,
		struct page *newpage, struct page *page, enum migrate_mode mode)
{
	int rc, extra_count;
	struct f2fs_inode_info *fi = F2FS_I(mapping->host);
	bool atomic_written = IS_ATOMIC_WRITTEN_PAGE(page);

	BUG_ON(PageWriteback(page));

	/* migrating an atomic written page is safe with the inmem_lock hold */
2259 2260 2261 2262 2263 2264
	if (atomic_written) {
		if (mode != MIGRATE_SYNC)
			return -EBUSY;
		if (!mutex_trylock(&fi->inmem_lock))
			return -EAGAIN;
	}
2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295

	/*
	 * A reference is expected if PagePrivate set when move mapping,
	 * however F2FS breaks this for maintaining dirty page counts when
	 * truncating pages. So here adjusting the 'extra_count' make it work.
	 */
	extra_count = (atomic_written ? 1 : 0) - page_has_private(page);
	rc = migrate_page_move_mapping(mapping, newpage,
				page, NULL, mode, extra_count);
	if (rc != MIGRATEPAGE_SUCCESS) {
		if (atomic_written)
			mutex_unlock(&fi->inmem_lock);
		return rc;
	}

	if (atomic_written) {
		struct inmem_pages *cur;
		list_for_each_entry(cur, &fi->inmem_pages, list)
			if (cur->page == page) {
				cur->page = newpage;
				break;
			}
		mutex_unlock(&fi->inmem_lock);
		put_page(page);
		get_page(newpage);
	}

	if (PagePrivate(page))
		SetPagePrivate(newpage);
	set_page_private(newpage, page_private(page));

2296 2297 2298 2299
	if (mode != MIGRATE_SYNC_NO_COPY)
		migrate_page_copy(newpage, page);
	else
		migrate_page_states(newpage, page);
2300 2301 2302 2303 2304

	return MIGRATEPAGE_SUCCESS;
}
#endif

2305 2306 2307 2308 2309 2310
const struct address_space_operations f2fs_dblock_aops = {
	.readpage	= f2fs_read_data_page,
	.readpages	= f2fs_read_data_pages,
	.writepage	= f2fs_write_data_page,
	.writepages	= f2fs_write_data_pages,
	.write_begin	= f2fs_write_begin,
2311
	.write_end	= f2fs_write_end,
2312
	.set_page_dirty	= f2fs_set_data_page_dirty,
2313 2314
	.invalidatepage	= f2fs_invalidate_page,
	.releasepage	= f2fs_release_page,
2315
	.direct_IO	= f2fs_direct_IO,
J
Jaegeuk Kim 已提交
2316
	.bmap		= f2fs_bmap,
2317 2318 2319
#ifdef CONFIG_MIGRATION
	.migratepage    = f2fs_migrate_page,
#endif
2320
};