checkpoint.c 31.6 KB
Newer Older
J
Jaegeuk Kim 已提交
1
/*
J
Jaegeuk Kim 已提交
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
 * fs/f2fs/checkpoint.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/bio.h>
#include <linux/mpage.h>
#include <linux/writeback.h>
#include <linux/blkdev.h>
#include <linux/f2fs_fs.h>
#include <linux/pagevec.h>
#include <linux/swap.h>

#include "f2fs.h"
#include "node.h"
#include "segment.h"
J
Jaegeuk Kim 已提交
23
#include "trace.h"
24
#include <trace/events/f2fs.h>
J
Jaegeuk Kim 已提交
25

J
Jaegeuk Kim 已提交
26
static struct kmem_cache *ino_entry_slab;
27
struct kmem_cache *inode_entry_slab;
J
Jaegeuk Kim 已提交
28

29 30
void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io)
{
31
	set_ckpt_flags(sbi, CP_ERROR_FLAG);
32 33 34 35 36
	sbi->sb->s_flags |= MS_RDONLY;
	if (!end_io)
		f2fs_flush_merged_bios(sbi);
}

J
Jaegeuk Kim 已提交
37
/*
J
Jaegeuk Kim 已提交
38 39 40 41
 * We guarantee no failure on the returned page.
 */
struct page *grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index)
{
G
Gu Zheng 已提交
42
	struct address_space *mapping = META_MAPPING(sbi);
J
Jaegeuk Kim 已提交
43 44
	struct page *page = NULL;
repeat:
45
	page = f2fs_grab_cache_page(mapping, index, false);
J
Jaegeuk Kim 已提交
46 47 48 49
	if (!page) {
		cond_resched();
		goto repeat;
	}
50
	f2fs_wait_on_page_writeback(page, META, true);
51 52
	if (!PageUptodate(page))
		SetPageUptodate(page);
J
Jaegeuk Kim 已提交
53 54 55
	return page;
}

J
Jaegeuk Kim 已提交
56
/*
J
Jaegeuk Kim 已提交
57 58
 * We guarantee no failure on the returned page.
 */
59 60
static struct page *__get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index,
							bool is_meta)
J
Jaegeuk Kim 已提交
61
{
G
Gu Zheng 已提交
62
	struct address_space *mapping = META_MAPPING(sbi);
J
Jaegeuk Kim 已提交
63
	struct page *page;
64
	struct f2fs_io_info fio = {
65
		.sbi = sbi,
66
		.type = META,
M
Mike Christie 已提交
67 68
		.op = REQ_OP_READ,
		.op_flags = READ_SYNC | REQ_META | REQ_PRIO,
69 70
		.old_blkaddr = index,
		.new_blkaddr = index,
71
		.encrypted_page = NULL,
72
	};
73 74

	if (unlikely(!is_meta))
M
Mike Christie 已提交
75
		fio.op_flags &= ~REQ_META;
J
Jaegeuk Kim 已提交
76
repeat:
77
	page = f2fs_grab_cache_page(mapping, index, false);
J
Jaegeuk Kim 已提交
78 79 80 81
	if (!page) {
		cond_resched();
		goto repeat;
	}
82 83 84
	if (PageUptodate(page))
		goto out;

85 86
	fio.page = page;

87 88
	if (f2fs_submit_page_bio(&fio)) {
		f2fs_put_page(page, 1);
J
Jaegeuk Kim 已提交
89
		goto repeat;
90
	}
J
Jaegeuk Kim 已提交
91

92
	lock_page(page);
93
	if (unlikely(page->mapping != mapping)) {
94 95 96
		f2fs_put_page(page, 1);
		goto repeat;
	}
97 98 99 100 101 102 103

	/*
	 * if there is any IO error when accessing device, make our filesystem
	 * readonly and make sure do not write checkpoint with non-uptodate
	 * meta page.
	 */
	if (unlikely(!PageUptodate(page)))
104
		f2fs_stop_checkpoint(sbi, false);
105
out:
J
Jaegeuk Kim 已提交
106 107 108
	return page;
}

109 110 111 112 113 114 115 116 117 118 119
struct page *get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index)
{
	return __get_meta_page(sbi, index, true);
}

/* for POR only */
struct page *get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index)
{
	return __get_meta_page(sbi, index, false);
}

120
bool is_valid_blkaddr(struct f2fs_sb_info *sbi, block_t blkaddr, int type)
121 122 123
{
	switch (type) {
	case META_NAT:
124
		break;
125
	case META_SIT:
126 127 128
		if (unlikely(blkaddr >= SIT_BLK_CNT(sbi)))
			return false;
		break;
129
	case META_SSA:
130 131 132 133
		if (unlikely(blkaddr >= MAIN_BLKADDR(sbi) ||
			blkaddr < SM_I(sbi)->ssa_blkaddr))
			return false;
		break;
134
	case META_CP:
135 136 137 138
		if (unlikely(blkaddr >= SIT_I(sbi)->sit_base_addr ||
			blkaddr < __start_cp_addr(sbi)))
			return false;
		break;
139
	case META_POR:
140 141 142 143
		if (unlikely(blkaddr >= MAX_BLKADDR(sbi) ||
			blkaddr < MAIN_BLKADDR(sbi)))
			return false;
		break;
144 145 146
	default:
		BUG();
	}
147 148

	return true;
149 150 151
}

/*
152
 * Readahead CP/NAT/SIT/SSA pages
153
 */
154 155
int ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages,
							int type, bool sync)
156 157
{
	struct page *page;
158
	block_t blkno = start;
159
	struct f2fs_io_info fio = {
160
		.sbi = sbi,
161
		.type = META,
M
Mike Christie 已提交
162
		.op = REQ_OP_READ,
163
		.op_flags = sync ? (READ_SYNC | REQ_META | REQ_PRIO) : REQ_RAHEAD,
164
		.encrypted_page = NULL,
165
	};
C
Chao Yu 已提交
166
	struct blk_plug plug;
167

168
	if (unlikely(type == META_POR))
M
Mike Christie 已提交
169
		fio.op_flags &= ~REQ_META;
170

C
Chao Yu 已提交
171
	blk_start_plug(&plug);
172 173
	for (; nrpages-- > 0; blkno++) {

174 175 176
		if (!is_valid_blkaddr(sbi, blkno, type))
			goto out;

177 178
		switch (type) {
		case META_NAT:
179 180
			if (unlikely(blkno >=
					NAT_BLOCK_OFFSET(NM_I(sbi)->max_nid)))
181
				blkno = 0;
182
			/* get nat block addr */
183
			fio.new_blkaddr = current_nat_addr(sbi,
184 185 186 187
					blkno * NAT_ENTRY_PER_BLOCK);
			break;
		case META_SIT:
			/* get sit block addr */
188
			fio.new_blkaddr = current_sit_addr(sbi,
189 190
					blkno * SIT_ENTRY_PER_BLOCK);
			break;
191
		case META_SSA:
192
		case META_CP:
193
		case META_POR:
194
			fio.new_blkaddr = blkno;
195 196 197 198 199
			break;
		default:
			BUG();
		}

200 201
		page = f2fs_grab_cache_page(META_MAPPING(sbi),
						fio.new_blkaddr, false);
202 203 204 205 206 207 208
		if (!page)
			continue;
		if (PageUptodate(page)) {
			f2fs_put_page(page, 1);
			continue;
		}

209
		fio.page = page;
210
		fio.old_blkaddr = fio.new_blkaddr;
211
		f2fs_submit_page_mbio(&fio);
212 213 214 215
		f2fs_put_page(page, 0);
	}
out:
	f2fs_submit_merged_bio(sbi, META, READ);
C
Chao Yu 已提交
216
	blk_finish_plug(&plug);
217 218 219
	return blkno - start;
}

220 221 222 223 224 225
void ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index)
{
	struct page *page;
	bool readahead = false;

	page = find_get_page(META_MAPPING(sbi), index);
226
	if (!page || !PageUptodate(page))
227 228 229 230
		readahead = true;
	f2fs_put_page(page, 0);

	if (readahead)
231
		ra_meta_pages(sbi, index, MAX_BIO_BLOCKS(sbi), META_POR, true);
232 233
}

J
Jaegeuk Kim 已提交
234 235 236
static int f2fs_write_meta_page(struct page *page,
				struct writeback_control *wbc)
{
237
	struct f2fs_sb_info *sbi = F2FS_P_SB(page);
J
Jaegeuk Kim 已提交
238

239 240
	trace_f2fs_writepage(page, META);

241
	if (unlikely(is_sbi_flag_set(sbi, SBI_POR_DOING)))
242
		goto redirty_out;
243
	if (wbc->for_reclaim && page->index < GET_SUM_BLOCK(sbi, 0))
244
		goto redirty_out;
245
	if (unlikely(f2fs_cp_error(sbi)))
246
		goto redirty_out;
J
Jaegeuk Kim 已提交
247

248 249
	write_meta_page(sbi, page);
	dec_page_count(sbi, F2FS_DIRTY_META);
250 251 252 253

	if (wbc->for_reclaim)
		f2fs_submit_merged_bio_cond(sbi, NULL, page, 0, META, WRITE);

254
	unlock_page(page);
255

256
	if (unlikely(f2fs_cp_error(sbi)))
257
		f2fs_submit_merged_bio(sbi, META, WRITE);
258

259
	return 0;
260 261

redirty_out:
262
	redirty_page_for_writepage(wbc, page);
263
	return AOP_WRITEPAGE_ACTIVATE;
J
Jaegeuk Kim 已提交
264 265 266 267 268
}

static int f2fs_write_meta_pages(struct address_space *mapping,
				struct writeback_control *wbc)
{
269
	struct f2fs_sb_info *sbi = F2FS_M_SB(mapping);
270
	struct blk_plug plug;
271
	long diff, written;
J
Jaegeuk Kim 已提交
272

273
	/* collect a number of dirty meta pages and write together */
274 275
	if (wbc->for_kupdate ||
		get_pages(sbi, F2FS_DIRTY_META) < nr_pages_to_skip(sbi, META))
276
		goto skip_write;
J
Jaegeuk Kim 已提交
277

Y
Yunlei He 已提交
278 279
	trace_f2fs_writepages(mapping->host, wbc, META);

J
Jaegeuk Kim 已提交
280 281
	/* if mounting is failed, skip writing node pages */
	mutex_lock(&sbi->cp_mutex);
282
	diff = nr_pages_to_write(sbi, META, wbc);
283
	blk_start_plug(&plug);
284
	written = sync_meta_pages(sbi, META, wbc->nr_to_write);
285
	blk_finish_plug(&plug);
J
Jaegeuk Kim 已提交
286
	mutex_unlock(&sbi->cp_mutex);
287
	wbc->nr_to_write = max((long)0, wbc->nr_to_write - written - diff);
J
Jaegeuk Kim 已提交
288
	return 0;
289 290 291

skip_write:
	wbc->pages_skipped += get_pages(sbi, F2FS_DIRTY_META);
Y
Yunlei He 已提交
292
	trace_f2fs_writepages(mapping->host, wbc, META);
293
	return 0;
J
Jaegeuk Kim 已提交
294 295 296 297 298
}

long sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type,
						long nr_to_write)
{
G
Gu Zheng 已提交
299
	struct address_space *mapping = META_MAPPING(sbi);
300
	pgoff_t index = 0, end = ULONG_MAX, prev = ULONG_MAX;
J
Jaegeuk Kim 已提交
301 302 303 304 305
	struct pagevec pvec;
	long nwritten = 0;
	struct writeback_control wbc = {
		.for_reclaim = 0,
	};
C
Chao Yu 已提交
306
	struct blk_plug plug;
J
Jaegeuk Kim 已提交
307 308 309

	pagevec_init(&pvec, 0);

C
Chao Yu 已提交
310 311
	blk_start_plug(&plug);

J
Jaegeuk Kim 已提交
312 313 314 315 316
	while (index <= end) {
		int i, nr_pages;
		nr_pages = pagevec_lookup_tag(&pvec, mapping, &index,
				PAGECACHE_TAG_DIRTY,
				min(end - index, (pgoff_t)PAGEVEC_SIZE-1) + 1);
317
		if (unlikely(nr_pages == 0))
J
Jaegeuk Kim 已提交
318 319 320 321
			break;

		for (i = 0; i < nr_pages; i++) {
			struct page *page = pvec.pages[i];
322

323
			if (prev == ULONG_MAX)
324 325 326 327 328 329
				prev = page->index - 1;
			if (nr_to_write != LONG_MAX && page->index != prev + 1) {
				pagevec_release(&pvec);
				goto stop;
			}

J
Jaegeuk Kim 已提交
330
			lock_page(page);
331 332 333 334 335 336 337 338 339 340 341

			if (unlikely(page->mapping != mapping)) {
continue_unlock:
				unlock_page(page);
				continue;
			}
			if (!PageDirty(page)) {
				/* someone wrote it for us */
				goto continue_unlock;
			}

342 343 344
			f2fs_wait_on_page_writeback(page, META, true);

			BUG_ON(PageWriteback(page));
345 346 347
			if (!clear_page_dirty_for_io(page))
				goto continue_unlock;

348
			if (mapping->a_ops->writepage(page, &wbc)) {
349 350 351
				unlock_page(page);
				break;
			}
352
			nwritten++;
353
			prev = page->index;
354
			if (unlikely(nwritten >= nr_to_write))
J
Jaegeuk Kim 已提交
355 356 357 358 359
				break;
		}
		pagevec_release(&pvec);
		cond_resched();
	}
360
stop:
J
Jaegeuk Kim 已提交
361
	if (nwritten)
J
Jaegeuk Kim 已提交
362
		f2fs_submit_merged_bio(sbi, type, WRITE);
J
Jaegeuk Kim 已提交
363

C
Chao Yu 已提交
364 365
	blk_finish_plug(&plug);

J
Jaegeuk Kim 已提交
366 367 368 369 370
	return nwritten;
}

static int f2fs_set_meta_page_dirty(struct page *page)
{
371 372
	trace_f2fs_set_page_dirty(page, META);

373 374
	if (!PageUptodate(page))
		SetPageUptodate(page);
J
Jaegeuk Kim 已提交
375
	if (!PageDirty(page)) {
376
		f2fs_set_page_dirty_nobuffers(page);
377
		inc_page_count(F2FS_P_SB(page), F2FS_DIRTY_META);
378
		SetPagePrivate(page);
J
Jaegeuk Kim 已提交
379
		f2fs_trace_pid(page);
J
Jaegeuk Kim 已提交
380 381 382 383 384 385 386 387 388
		return 1;
	}
	return 0;
}

const struct address_space_operations f2fs_meta_aops = {
	.writepage	= f2fs_write_meta_page,
	.writepages	= f2fs_write_meta_pages,
	.set_page_dirty	= f2fs_set_meta_page_dirty,
389 390
	.invalidatepage = f2fs_invalidate_page,
	.releasepage	= f2fs_release_page,
391 392 393
#ifdef CONFIG_MIGRATION
	.migratepage    = f2fs_migrate_page,
#endif
J
Jaegeuk Kim 已提交
394 395
};

J
Jaegeuk Kim 已提交
396
static void __add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type)
397
{
398
	struct inode_management *im = &sbi->im[type];
399 400 401
	struct ino_entry *e, *tmp;

	tmp = f2fs_kmem_cache_alloc(ino_entry_slab, GFP_NOFS);
402
retry:
403
	radix_tree_preload(GFP_NOFS | __GFP_NOFAIL);
404

405 406
	spin_lock(&im->ino_lock);
	e = radix_tree_lookup(&im->ino_root, ino);
407
	if (!e) {
408
		e = tmp;
409 410
		if (radix_tree_insert(&im->ino_root, ino, e)) {
			spin_unlock(&im->ino_lock);
411
			radix_tree_preload_end();
412 413 414 415
			goto retry;
		}
		memset(e, 0, sizeof(struct ino_entry));
		e->ino = ino;
416

417
		list_add_tail(&e->list, &im->ino_list);
418
		if (type != ORPHAN_INO)
419
			im->ino_num++;
420
	}
421
	spin_unlock(&im->ino_lock);
422
	radix_tree_preload_end();
423 424 425

	if (e != tmp)
		kmem_cache_free(ino_entry_slab, tmp);
426 427
}

J
Jaegeuk Kim 已提交
428
static void __remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type)
429
{
430
	struct inode_management *im = &sbi->im[type];
J
Jaegeuk Kim 已提交
431
	struct ino_entry *e;
432

433 434
	spin_lock(&im->ino_lock);
	e = radix_tree_lookup(&im->ino_root, ino);
435 436
	if (e) {
		list_del(&e->list);
437 438 439
		radix_tree_delete(&im->ino_root, ino);
		im->ino_num--;
		spin_unlock(&im->ino_lock);
440 441
		kmem_cache_free(ino_entry_slab, e);
		return;
442
	}
443
	spin_unlock(&im->ino_lock);
444 445
}

446
void add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type)
447 448 449 450 451
{
	/* add new dirty ino entry into list */
	__add_ino_entry(sbi, ino, type);
}

452
void remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type)
453 454 455 456 457 458 459 460
{
	/* remove dirty ino entry from list */
	__remove_ino_entry(sbi, ino, type);
}

/* mode should be APPEND_INO or UPDATE_INO */
bool exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode)
{
461
	struct inode_management *im = &sbi->im[mode];
462
	struct ino_entry *e;
463 464 465 466

	spin_lock(&im->ino_lock);
	e = radix_tree_lookup(&im->ino_root, ino);
	spin_unlock(&im->ino_lock);
467 468 469
	return e ? true : false;
}

470
void release_ino_entry(struct f2fs_sb_info *sbi, bool all)
471 472 473 474
{
	struct ino_entry *e, *tmp;
	int i;

475
	for (i = all ? ORPHAN_INO: APPEND_INO; i <= UPDATE_INO; i++) {
476 477 478 479
		struct inode_management *im = &sbi->im[i];

		spin_lock(&im->ino_lock);
		list_for_each_entry_safe(e, tmp, &im->ino_list, list) {
480
			list_del(&e->list);
481
			radix_tree_delete(&im->ino_root, e->ino);
482
			kmem_cache_free(ino_entry_slab, e);
483
			im->ino_num--;
484
		}
485
		spin_unlock(&im->ino_lock);
486 487 488
	}
}

J
Jaegeuk Kim 已提交
489
int acquire_orphan_inode(struct f2fs_sb_info *sbi)
J
Jaegeuk Kim 已提交
490
{
491
	struct inode_management *im = &sbi->im[ORPHAN_INO];
J
Jaegeuk Kim 已提交
492 493
	int err = 0;

494
	spin_lock(&im->ino_lock);
J
Jaegeuk Kim 已提交
495 496 497 498 499 500 501

#ifdef CONFIG_F2FS_FAULT_INJECTION
	if (time_to_inject(FAULT_ORPHAN)) {
		spin_unlock(&im->ino_lock);
		return -ENOSPC;
	}
#endif
502
	if (unlikely(im->ino_num >= sbi->max_orphans))
J
Jaegeuk Kim 已提交
503
		err = -ENOSPC;
J
Jaegeuk Kim 已提交
504
	else
505 506
		im->ino_num++;
	spin_unlock(&im->ino_lock);
507

J
Jaegeuk Kim 已提交
508 509 510
	return err;
}

J
Jaegeuk Kim 已提交
511 512
void release_orphan_inode(struct f2fs_sb_info *sbi)
{
513 514 515 516 517 518
	struct inode_management *im = &sbi->im[ORPHAN_INO];

	spin_lock(&im->ino_lock);
	f2fs_bug_on(sbi, im->ino_num == 0);
	im->ino_num--;
	spin_unlock(&im->ino_lock);
J
Jaegeuk Kim 已提交
519 520
}

521
void add_orphan_inode(struct inode *inode)
J
Jaegeuk Kim 已提交
522
{
523
	/* add new orphan ino entry into list */
524 525
	__add_ino_entry(F2FS_I_SB(inode), inode->i_ino, ORPHAN_INO);
	update_inode_page(inode);
J
Jaegeuk Kim 已提交
526 527 528 529
}

void remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino)
{
530
	/* remove orphan entry from orphan list */
J
Jaegeuk Kim 已提交
531
	__remove_ino_entry(sbi, ino, ORPHAN_INO);
J
Jaegeuk Kim 已提交
532 533
}

534
static int recover_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino)
J
Jaegeuk Kim 已提交
535
{
536
	struct inode *inode;
537
	struct node_info ni;
538 539 540 541 542 543 544 545 546 547 548
	int err = acquire_orphan_inode(sbi);

	if (err) {
		set_sbi_flag(sbi, SBI_NEED_FSCK);
		f2fs_msg(sbi->sb, KERN_WARNING,
				"%s: orphan failed (ino=%x), run fsck to fix.",
				__func__, ino);
		return err;
	}

	__add_ino_entry(sbi, ino, ORPHAN_INO);
549

550
	inode = f2fs_iget_retry(sbi->sb, ino);
551 552 553 554 555 556 557 558 559
	if (IS_ERR(inode)) {
		/*
		 * there should be a bug that we can't find the entry
		 * to orphan inode.
		 */
		f2fs_bug_on(sbi, PTR_ERR(inode) == -ENOENT);
		return PTR_ERR(inode);
	}

J
Jaegeuk Kim 已提交
560 561 562 563
	clear_nlink(inode);

	/* truncate all the data during iput */
	iput(inode);
564 565 566 567 568

	get_node_info(sbi, ino, &ni);

	/* ENOMEM was fully retried in f2fs_evict_inode. */
	if (ni.blk_addr != NULL_ADDR) {
569 570 571 572 573
		set_sbi_flag(sbi, SBI_NEED_FSCK);
		f2fs_msg(sbi->sb, KERN_WARNING,
			"%s: orphan failed (ino=%x), run fsck to fix.",
				__func__, ino);
		return -EIO;
574
	}
575
	__remove_ino_entry(sbi, ino, ORPHAN_INO);
576
	return 0;
J
Jaegeuk Kim 已提交
577 578
}

579
int recover_orphan_inodes(struct f2fs_sb_info *sbi)
J
Jaegeuk Kim 已提交
580
{
581
	block_t start_blk, orphan_blocks, i, j;
582
	int err;
J
Jaegeuk Kim 已提交
583

584
	if (!is_set_ckpt_flags(sbi, CP_ORPHAN_PRESENT_FLAG))
585
		return 0;
J
Jaegeuk Kim 已提交
586

W
Wanpeng Li 已提交
587
	start_blk = __start_cp_addr(sbi) + 1 + __cp_payload(sbi);
588
	orphan_blocks = __start_sum_addr(sbi) - 1 - __cp_payload(sbi);
J
Jaegeuk Kim 已提交
589

590
	ra_meta_pages(sbi, start_blk, orphan_blocks, META_CP, true);
591

592
	for (i = 0; i < orphan_blocks; i++) {
J
Jaegeuk Kim 已提交
593 594 595 596 597 598
		struct page *page = get_meta_page(sbi, start_blk + i);
		struct f2fs_orphan_block *orphan_blk;

		orphan_blk = (struct f2fs_orphan_block *)page_address(page);
		for (j = 0; j < le32_to_cpu(orphan_blk->entry_count); j++) {
			nid_t ino = le32_to_cpu(orphan_blk->ino[j]);
599 600 601 602 603
			err = recover_orphan_inode(sbi, ino);
			if (err) {
				f2fs_put_page(page, 1);
				return err;
			}
J
Jaegeuk Kim 已提交
604 605 606 607
		}
		f2fs_put_page(page, 1);
	}
	/* clear Orphan Flag */
608
	clear_ckpt_flags(sbi, CP_ORPHAN_PRESENT_FLAG);
609
	return 0;
J
Jaegeuk Kim 已提交
610 611 612 613
}

static void write_orphan_inodes(struct f2fs_sb_info *sbi, block_t start_blk)
{
614
	struct list_head *head;
J
Jaegeuk Kim 已提交
615 616
	struct f2fs_orphan_block *orphan_blk = NULL;
	unsigned int nentries = 0;
C
Chao Yu 已提交
617
	unsigned short index = 1;
618
	unsigned short orphan_blocks;
619
	struct page *page = NULL;
J
Jaegeuk Kim 已提交
620
	struct ino_entry *orphan = NULL;
621
	struct inode_management *im = &sbi->im[ORPHAN_INO];
J
Jaegeuk Kim 已提交
622

623
	orphan_blocks = GET_ORPHAN_BLOCKS(im->ino_num);
624

625 626 627 628 629
	/*
	 * we don't need to do spin_lock(&im->ino_lock) here, since all the
	 * orphan inode operations are covered under f2fs_lock_op().
	 * And, spin_lock should be avoided due to page operations below.
	 */
630
	head = &im->ino_list;
J
Jaegeuk Kim 已提交
631 632

	/* loop for each orphan inode entry and write them in Jornal block */
633 634
	list_for_each_entry(orphan, head, list) {
		if (!page) {
C
Chao Yu 已提交
635
			page = grab_meta_page(sbi, start_blk++);
636 637 638 639
			orphan_blk =
				(struct f2fs_orphan_block *)page_address(page);
			memset(orphan_blk, 0, sizeof(*orphan_blk));
		}
J
Jaegeuk Kim 已提交
640

641
		orphan_blk->ino[nentries++] = cpu_to_le32(orphan->ino);
J
Jaegeuk Kim 已提交
642

643
		if (nentries == F2FS_ORPHANS_PER_BLOCK) {
J
Jaegeuk Kim 已提交
644 645 646 647 648 649 650 651 652 653 654 655 656 657
			/*
			 * an orphan block is full of 1020 entries,
			 * then we need to flush current orphan blocks
			 * and bring another one in memory
			 */
			orphan_blk->blk_addr = cpu_to_le16(index);
			orphan_blk->blk_count = cpu_to_le16(orphan_blocks);
			orphan_blk->entry_count = cpu_to_le32(nentries);
			set_page_dirty(page);
			f2fs_put_page(page, 1);
			index++;
			nentries = 0;
			page = NULL;
		}
658
	}
J
Jaegeuk Kim 已提交
659

660 661 662 663 664 665
	if (page) {
		orphan_blk->blk_addr = cpu_to_le16(index);
		orphan_blk->blk_count = cpu_to_le16(orphan_blocks);
		orphan_blk->entry_count = cpu_to_le32(nentries);
		set_page_dirty(page);
		f2fs_put_page(page, 1);
J
Jaegeuk Kim 已提交
666 667 668 669 670 671 672 673 674 675 676
	}
}

static struct page *validate_checkpoint(struct f2fs_sb_info *sbi,
				block_t cp_addr, unsigned long long *version)
{
	struct page *cp_page_1, *cp_page_2 = NULL;
	unsigned long blk_size = sbi->blocksize;
	struct f2fs_checkpoint *cp_block;
	unsigned long long cur_version = 0, pre_version = 0;
	size_t crc_offset;
J
Jaegeuk Kim 已提交
677
	__u32 crc = 0;
J
Jaegeuk Kim 已提交
678 679 680 681 682 683 684 685 686 687

	/* Read the 1st cp block in this CP pack */
	cp_page_1 = get_meta_page(sbi, cp_addr);

	/* get the version number */
	cp_block = (struct f2fs_checkpoint *)page_address(cp_page_1);
	crc_offset = le32_to_cpu(cp_block->checksum_offset);
	if (crc_offset >= blk_size)
		goto invalid_cp1;

J
Jaegeuk Kim 已提交
688
	crc = le32_to_cpu(*((__le32 *)((unsigned char *)cp_block + crc_offset)));
K
Keith Mok 已提交
689
	if (!f2fs_crc_valid(sbi, crc, cp_block, crc_offset))
J
Jaegeuk Kim 已提交
690 691
		goto invalid_cp1;

692
	pre_version = cur_cp_version(cp_block);
J
Jaegeuk Kim 已提交
693 694

	/* Read the 2nd cp block in this CP pack */
695
	cp_addr += le32_to_cpu(cp_block->cp_pack_total_block_count) - 1;
J
Jaegeuk Kim 已提交
696 697 698 699 700 701 702
	cp_page_2 = get_meta_page(sbi, cp_addr);

	cp_block = (struct f2fs_checkpoint *)page_address(cp_page_2);
	crc_offset = le32_to_cpu(cp_block->checksum_offset);
	if (crc_offset >= blk_size)
		goto invalid_cp2;

J
Jaegeuk Kim 已提交
703
	crc = le32_to_cpu(*((__le32 *)((unsigned char *)cp_block + crc_offset)));
K
Keith Mok 已提交
704
	if (!f2fs_crc_valid(sbi, crc, cp_block, crc_offset))
J
Jaegeuk Kim 已提交
705 706
		goto invalid_cp2;

707
	cur_version = cur_cp_version(cp_block);
J
Jaegeuk Kim 已提交
708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728

	if (cur_version == pre_version) {
		*version = cur_version;
		f2fs_put_page(cp_page_2, 1);
		return cp_page_1;
	}
invalid_cp2:
	f2fs_put_page(cp_page_2, 1);
invalid_cp1:
	f2fs_put_page(cp_page_1, 1);
	return NULL;
}

int get_valid_checkpoint(struct f2fs_sb_info *sbi)
{
	struct f2fs_checkpoint *cp_block;
	struct f2fs_super_block *fsb = sbi->raw_super;
	struct page *cp1, *cp2, *cur_page;
	unsigned long blk_size = sbi->blocksize;
	unsigned long long cp1_version = 0, cp2_version = 0;
	unsigned long long cp_start_blk_no;
W
Wanpeng Li 已提交
729
	unsigned int cp_blks = 1 + __cp_payload(sbi);
C
Changman Lee 已提交
730 731
	block_t cp_blk_no;
	int i;
J
Jaegeuk Kim 已提交
732

C
Changman Lee 已提交
733
	sbi->ckpt = kzalloc(cp_blks * blk_size, GFP_KERNEL);
J
Jaegeuk Kim 已提交
734 735 736 737 738 739 740 741 742 743
	if (!sbi->ckpt)
		return -ENOMEM;
	/*
	 * Finding out valid cp block involves read both
	 * sets( cp pack1 and cp pack 2)
	 */
	cp_start_blk_no = le32_to_cpu(fsb->cp_blkaddr);
	cp1 = validate_checkpoint(sbi, cp_start_blk_no, &cp1_version);

	/* The second checkpoint pack should start at the next segment */
744 745
	cp_start_blk_no += ((unsigned long long)1) <<
				le32_to_cpu(fsb->log_blocks_per_seg);
J
Jaegeuk Kim 已提交
746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763
	cp2 = validate_checkpoint(sbi, cp_start_blk_no, &cp2_version);

	if (cp1 && cp2) {
		if (ver_after(cp2_version, cp1_version))
			cur_page = cp2;
		else
			cur_page = cp1;
	} else if (cp1) {
		cur_page = cp1;
	} else if (cp2) {
		cur_page = cp2;
	} else {
		goto fail_no_cp;
	}

	cp_block = (struct f2fs_checkpoint *)page_address(cur_page);
	memcpy(sbi->ckpt, cp_block, blk_size);

764 765 766 767
	/* Sanity checking of checkpoint */
	if (sanity_check_ckpt(sbi))
		goto fail_no_cp;

C
Changman Lee 已提交
768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784
	if (cp_blks <= 1)
		goto done;

	cp_blk_no = le32_to_cpu(fsb->cp_blkaddr);
	if (cur_page == cp2)
		cp_blk_no += 1 << le32_to_cpu(fsb->log_blocks_per_seg);

	for (i = 1; i < cp_blks; i++) {
		void *sit_bitmap_ptr;
		unsigned char *ckpt = (unsigned char *)sbi->ckpt;

		cur_page = get_meta_page(sbi, cp_blk_no + i);
		sit_bitmap_ptr = page_address(cur_page);
		memcpy(ckpt + i * blk_size, sit_bitmap_ptr, blk_size);
		f2fs_put_page(cur_page, 1);
	}
done:
J
Jaegeuk Kim 已提交
785 786 787 788 789 790 791 792 793
	f2fs_put_page(cp1, 1);
	f2fs_put_page(cp2, 1);
	return 0;

fail_no_cp:
	kfree(sbi->ckpt);
	return -EINVAL;
}

794
static void __add_dirty_inode(struct inode *inode, enum inode_type type)
J
Jaegeuk Kim 已提交
795
{
796
	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
797
	int flag = (type == DIR_INODE) ? FI_DIRTY_DIR : FI_DIRTY_FILE;
J
Jaegeuk Kim 已提交
798

799
	if (is_inode_flag_set(inode, flag))
800
		return;
801

802 803
	set_inode_flag(inode, flag);
	list_add_tail(&F2FS_I(inode)->dirty_list, &sbi->inode_list[type]);
C
Chao Yu 已提交
804
	stat_inc_dirty_inode(sbi, type);
805 806
}

807
static void __remove_dirty_inode(struct inode *inode, enum inode_type type)
C
Chao Yu 已提交
808
{
809
	int flag = (type == DIR_INODE) ? FI_DIRTY_DIR : FI_DIRTY_FILE;
C
Chao Yu 已提交
810

811
	if (get_dirty_pages(inode) || !is_inode_flag_set(inode, flag))
C
Chao Yu 已提交
812 813
		return;

814 815
	list_del_init(&F2FS_I(inode)->dirty_list);
	clear_inode_flag(inode, flag);
C
Chao Yu 已提交
816
	stat_dec_dirty_inode(F2FS_I_SB(inode), type);
C
Chao Yu 已提交
817 818
}

819
void update_dirty_page(struct inode *inode, struct page *page)
820
{
821
	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
822
	enum inode_type type = S_ISDIR(inode->i_mode) ? DIR_INODE : FILE_INODE;
823

824 825
	if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) &&
			!S_ISLNK(inode->i_mode))
J
Jaegeuk Kim 已提交
826
		return;
827

828 829
	spin_lock(&sbi->inode_lock[type]);
	if (type != FILE_INODE || test_opt(sbi, DATA_FLUSH))
830
		__add_dirty_inode(inode, type);
831
	inode_inc_dirty_pages(inode);
832 833
	spin_unlock(&sbi->inode_lock[type]);

834
	SetPagePrivate(page);
J
Jaegeuk Kim 已提交
835
	f2fs_trace_pid(page);
836 837
}

838
void remove_dirty_inode(struct inode *inode)
J
Jaegeuk Kim 已提交
839
{
840
	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
841
	enum inode_type type = S_ISDIR(inode->i_mode) ? DIR_INODE : FILE_INODE;
J
Jaegeuk Kim 已提交
842

843 844
	if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) &&
			!S_ISLNK(inode->i_mode))
J
Jaegeuk Kim 已提交
845 846
		return;

847 848 849
	if (type == FILE_INODE && !test_opt(sbi, DATA_FLUSH))
		return;

850 851 852
	spin_lock(&sbi->inode_lock[type]);
	__remove_dirty_inode(inode, type);
	spin_unlock(&sbi->inode_lock[type]);
853 854
}

C
Chao Yu 已提交
855
int sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type)
J
Jaegeuk Kim 已提交
856
{
857
	struct list_head *head;
J
Jaegeuk Kim 已提交
858
	struct inode *inode;
859
	struct f2fs_inode_info *fi;
860 861 862 863 864
	bool is_dir = (type == DIR_INODE);

	trace_f2fs_sync_dirty_inodes_enter(sbi->sb, is_dir,
				get_pages(sbi, is_dir ?
				F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA));
J
Jaegeuk Kim 已提交
865
retry:
866
	if (unlikely(f2fs_cp_error(sbi)))
C
Chao Yu 已提交
867
		return -EIO;
868

869
	spin_lock(&sbi->inode_lock[type]);
870

871
	head = &sbi->inode_list[type];
J
Jaegeuk Kim 已提交
872
	if (list_empty(head)) {
873
		spin_unlock(&sbi->inode_lock[type]);
874 875 876
		trace_f2fs_sync_dirty_inodes_exit(sbi->sb, is_dir,
				get_pages(sbi, is_dir ?
				F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA));
C
Chao Yu 已提交
877
		return 0;
J
Jaegeuk Kim 已提交
878
	}
879 880
	fi = list_entry(head->next, struct f2fs_inode_info, dirty_list);
	inode = igrab(&fi->vfs_inode);
881
	spin_unlock(&sbi->inode_lock[type]);
J
Jaegeuk Kim 已提交
882
	if (inode) {
883
		filemap_fdatawrite(inode->i_mapping);
J
Jaegeuk Kim 已提交
884 885 886 887 888 889
		iput(inode);
	} else {
		/*
		 * We should submit bio, since it exists several
		 * wribacking dentry pages in the freeing inode.
		 */
J
Jaegeuk Kim 已提交
890
		f2fs_submit_merged_bio(sbi, DATA, WRITE);
891
		cond_resched();
J
Jaegeuk Kim 已提交
892 893 894 895
	}
	goto retry;
}

896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923
int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi)
{
	struct list_head *head = &sbi->inode_list[DIRTY_META];
	struct inode *inode;
	struct f2fs_inode_info *fi;
	s64 total = get_pages(sbi, F2FS_DIRTY_IMETA);

	while (total--) {
		if (unlikely(f2fs_cp_error(sbi)))
			return -EIO;

		spin_lock(&sbi->inode_lock[DIRTY_META]);
		if (list_empty(head)) {
			spin_unlock(&sbi->inode_lock[DIRTY_META]);
			return 0;
		}
		fi = list_entry(head->next, struct f2fs_inode_info,
							gdirty_list);
		inode = igrab(&fi->vfs_inode);
		spin_unlock(&sbi->inode_lock[DIRTY_META]);
		if (inode) {
			update_inode_page(inode);
			iput(inode);
		}
	};
	return 0;
}

J
Jaegeuk Kim 已提交
924
/*
J
Jaegeuk Kim 已提交
925 926
 * Freeze all the FS-operations for checkpoint.
 */
927
static int block_operations(struct f2fs_sb_info *sbi)
J
Jaegeuk Kim 已提交
928 929 930 931 932 933
{
	struct writeback_control wbc = {
		.sync_mode = WB_SYNC_ALL,
		.nr_to_write = LONG_MAX,
		.for_reclaim = 0,
	};
934
	struct blk_plug plug;
935
	int err = 0;
936 937 938

	blk_start_plug(&plug);

939
retry_flush_dents:
940
	f2fs_lock_all(sbi);
J
Jaegeuk Kim 已提交
941 942
	/* write all the dirty dentry pages */
	if (get_pages(sbi, F2FS_DIRTY_DENTS)) {
943
		f2fs_unlock_all(sbi);
C
Chao Yu 已提交
944 945
		err = sync_dirty_inodes(sbi, DIR_INODE);
		if (err)
946
			goto out;
947
		goto retry_flush_dents;
J
Jaegeuk Kim 已提交
948 949
	}

950 951 952 953 954 955 956 957
	if (get_pages(sbi, F2FS_DIRTY_IMETA)) {
		f2fs_unlock_all(sbi);
		err = f2fs_sync_inode_meta(sbi);
		if (err)
			goto out;
		goto retry_flush_dents;
	}

J
Jaegeuk Kim 已提交
958
	/*
A
arter97 已提交
959
	 * POR: we should ensure that there are no dirty node pages
J
Jaegeuk Kim 已提交
960 961
	 * until finishing nat/sit flush.
	 */
962
retry_flush_nodes:
963
	down_write(&sbi->node_write);
J
Jaegeuk Kim 已提交
964 965

	if (get_pages(sbi, F2FS_DIRTY_NODES)) {
966
		up_write(&sbi->node_write);
967
		err = sync_node_pages(sbi, &wbc);
C
Chao Yu 已提交
968
		if (err) {
969 970 971
			f2fs_unlock_all(sbi);
			goto out;
		}
972
		goto retry_flush_nodes;
J
Jaegeuk Kim 已提交
973
	}
974
out:
975
	blk_finish_plug(&plug);
976
	return err;
J
Jaegeuk Kim 已提交
977 978 979 980
}

static void unblock_operations(struct f2fs_sb_info *sbi)
{
981
	up_write(&sbi->node_write);
982 983

	build_free_nids(sbi);
984
	f2fs_unlock_all(sbi);
J
Jaegeuk Kim 已提交
985 986
}

987 988 989 990 991 992 993
static void wait_on_all_pages_writeback(struct f2fs_sb_info *sbi)
{
	DEFINE_WAIT(wait);

	for (;;) {
		prepare_to_wait(&sbi->cp_wait, &wait, TASK_UNINTERRUPTIBLE);

994
		if (!atomic_read(&sbi->nr_wb_bios))
995 996
			break;

997
		io_schedule_timeout(5*HZ);
998 999 1000 1001
	}
	finish_wait(&sbi->cp_wait, &wait);
}

C
Chao Yu 已提交
1002
static int do_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc)
J
Jaegeuk Kim 已提交
1003 1004
{
	struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
1005
	struct f2fs_nm_info *nm_i = NM_I(sbi);
1006
	unsigned long orphan_num = sbi->im[ORPHAN_INO].ino_num;
1007
	nid_t last_nid = nm_i->next_scan_nid;
J
Jaegeuk Kim 已提交
1008 1009
	block_t start_blk;
	unsigned int data_sum_blocks, orphan_blocks;
J
Jaegeuk Kim 已提交
1010
	__u32 crc32 = 0;
J
Jaegeuk Kim 已提交
1011
	int i;
W
Wanpeng Li 已提交
1012
	int cp_payload_blks = __cp_payload(sbi);
1013 1014 1015
	struct super_block *sb = sbi->sb;
	struct curseg_info *seg_i = CURSEG_I(sbi, CURSEG_HOT_NODE);
	u64 kbytes_written;
J
Jaegeuk Kim 已提交
1016 1017

	/* Flush all the NAT/SIT pages */
1018
	while (get_pages(sbi, F2FS_DIRTY_META)) {
J
Jaegeuk Kim 已提交
1019
		sync_meta_pages(sbi, META, LONG_MAX);
1020
		if (unlikely(f2fs_cp_error(sbi)))
C
Chao Yu 已提交
1021
			return -EIO;
1022
	}
J
Jaegeuk Kim 已提交
1023 1024 1025 1026 1027 1028 1029 1030 1031 1032

	next_free_nid(sbi, &last_nid);

	/*
	 * modify checkpoint
	 * version number is already updated
	 */
	ckpt->elapsed_time = cpu_to_le64(get_mtime(sbi));
	ckpt->valid_block_count = cpu_to_le64(valid_user_blocks(sbi));
	ckpt->free_segment_count = cpu_to_le32(free_segments(sbi));
C
Chao Yu 已提交
1033
	for (i = 0; i < NR_CURSEG_NODE_TYPE; i++) {
J
Jaegeuk Kim 已提交
1034 1035 1036 1037 1038 1039 1040
		ckpt->cur_node_segno[i] =
			cpu_to_le32(curseg_segno(sbi, i + CURSEG_HOT_NODE));
		ckpt->cur_node_blkoff[i] =
			cpu_to_le16(curseg_blkoff(sbi, i + CURSEG_HOT_NODE));
		ckpt->alloc_type[i + CURSEG_HOT_NODE] =
				curseg_alloc_type(sbi, i + CURSEG_HOT_NODE);
	}
C
Chao Yu 已提交
1041
	for (i = 0; i < NR_CURSEG_DATA_TYPE; i++) {
J
Jaegeuk Kim 已提交
1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054
		ckpt->cur_data_segno[i] =
			cpu_to_le32(curseg_segno(sbi, i + CURSEG_HOT_DATA));
		ckpt->cur_data_blkoff[i] =
			cpu_to_le16(curseg_blkoff(sbi, i + CURSEG_HOT_DATA));
		ckpt->alloc_type[i + CURSEG_HOT_DATA] =
				curseg_alloc_type(sbi, i + CURSEG_HOT_DATA);
	}

	ckpt->valid_node_count = cpu_to_le32(valid_node_count(sbi));
	ckpt->valid_inode_count = cpu_to_le32(valid_inode_count(sbi));
	ckpt->next_free_nid = cpu_to_le32(last_nid);

	/* 2 cp  + n data seg summary + orphan inode blocks */
1055
	data_sum_blocks = npages_for_summary_flush(sbi, false);
1056
	spin_lock(&sbi->cp_lock);
C
Chao Yu 已提交
1057
	if (data_sum_blocks < NR_CURSEG_DATA_TYPE)
1058
		__set_ckpt_flags(ckpt, CP_COMPACT_SUM_FLAG);
J
Jaegeuk Kim 已提交
1059
	else
1060 1061
		__clear_ckpt_flags(ckpt, CP_COMPACT_SUM_FLAG);
	spin_unlock(&sbi->cp_lock);
J
Jaegeuk Kim 已提交
1062

1063
	orphan_blocks = GET_ORPHAN_BLOCKS(orphan_num);
C
Changman Lee 已提交
1064 1065
	ckpt->cp_pack_start_sum = cpu_to_le32(1 + cp_payload_blks +
			orphan_blocks);
J
Jaegeuk Kim 已提交
1066

1067
	if (__remain_node_summaries(cpc->reason))
C
Chao Yu 已提交
1068
		ckpt->cp_pack_total_block_count = cpu_to_le32(F2FS_CP_PACKS+
C
Changman Lee 已提交
1069 1070
				cp_payload_blks + data_sum_blocks +
				orphan_blocks + NR_CURSEG_NODE_TYPE);
1071
	else
C
Chao Yu 已提交
1072
		ckpt->cp_pack_total_block_count = cpu_to_le32(F2FS_CP_PACKS +
C
Changman Lee 已提交
1073 1074
				cp_payload_blks + data_sum_blocks +
				orphan_blocks);
1075

1076
	spin_lock(&sbi->cp_lock);
1077
	if (cpc->reason == CP_UMOUNT)
1078
		__set_ckpt_flags(ckpt, CP_UMOUNT_FLAG);
1079
	else
1080
		__clear_ckpt_flags(ckpt, CP_UMOUNT_FLAG);
1081 1082

	if (cpc->reason == CP_FASTBOOT)
1083
		__set_ckpt_flags(ckpt, CP_FASTBOOT_FLAG);
1084
	else
1085
		__clear_ckpt_flags(ckpt, CP_FASTBOOT_FLAG);
J
Jaegeuk Kim 已提交
1086

1087
	if (orphan_num)
1088
		__set_ckpt_flags(ckpt, CP_ORPHAN_PRESENT_FLAG);
J
Jaegeuk Kim 已提交
1089
	else
1090
		__clear_ckpt_flags(ckpt, CP_ORPHAN_PRESENT_FLAG);
J
Jaegeuk Kim 已提交
1091

1092
	if (is_sbi_flag_set(sbi, SBI_NEED_FSCK))
1093
		__set_ckpt_flags(ckpt, CP_FSCK_FLAG);
1094

1095
	/* set this flag to activate crc|cp_ver for recovery */
1096 1097 1098
	__set_ckpt_flags(ckpt, CP_CRC_RECOVERY_FLAG);

	spin_unlock(&sbi->cp_lock);
1099

J
Jaegeuk Kim 已提交
1100 1101 1102 1103
	/* update SIT/NAT bitmap */
	get_sit_bitmap(sbi, __bitmap_ptr(sbi, SIT_BITMAP));
	get_nat_bitmap(sbi, __bitmap_ptr(sbi, NAT_BITMAP));

K
Keith Mok 已提交
1104
	crc32 = f2fs_crc32(sbi, ckpt, le32_to_cpu(ckpt->checksum_offset));
J
Jaegeuk Kim 已提交
1105 1106
	*((__le32 *)((unsigned char *)ckpt +
				le32_to_cpu(ckpt->checksum_offset)))
J
Jaegeuk Kim 已提交
1107 1108 1109 1110
				= cpu_to_le32(crc32);

	start_blk = __start_cp_addr(sbi);

1111 1112 1113
	/* need to wait for end_io results */
	wait_on_all_pages_writeback(sbi);
	if (unlikely(f2fs_cp_error(sbi)))
C
Chao Yu 已提交
1114
		return -EIO;
1115

J
Jaegeuk Kim 已提交
1116
	/* write out checkpoint buffer at block 0 */
C
Chao Yu 已提交
1117 1118 1119 1120 1121
	update_meta_page(sbi, ckpt, start_blk++);

	for (i = 1; i < 1 + cp_payload_blks; i++)
		update_meta_page(sbi, (char *)ckpt + i * F2FS_BLKSIZE,
							start_blk++);
C
Changman Lee 已提交
1122

1123
	if (orphan_num) {
J
Jaegeuk Kim 已提交
1124 1125 1126 1127 1128 1129
		write_orphan_inodes(sbi, start_blk);
		start_blk += orphan_blocks;
	}

	write_data_summaries(sbi, start_blk);
	start_blk += data_sum_blocks;
1130 1131 1132 1133 1134 1135

	/* Record write statistics in the hot node summary */
	kbytes_written = sbi->kbytes_written;
	if (sb->s_bdev->bd_part)
		kbytes_written += BD_PART_WRITTEN(sbi);

1136
	seg_i->journal->info.kbytes_written = cpu_to_le64(kbytes_written);
1137

1138
	if (__remain_node_summaries(cpc->reason)) {
J
Jaegeuk Kim 已提交
1139 1140 1141 1142 1143
		write_node_summaries(sbi, start_blk);
		start_blk += NR_CURSEG_NODE_TYPE;
	}

	/* writeout checkpoint block */
C
Chao Yu 已提交
1144
	update_meta_page(sbi, ckpt, start_blk);
J
Jaegeuk Kim 已提交
1145 1146

	/* wait for previous submitted node/meta pages writeback */
1147
	wait_on_all_pages_writeback(sbi);
J
Jaegeuk Kim 已提交
1148

1149
	if (unlikely(f2fs_cp_error(sbi)))
C
Chao Yu 已提交
1150
		return -EIO;
1151

1152 1153
	filemap_fdatawait_range(NODE_MAPPING(sbi), 0, LLONG_MAX);
	filemap_fdatawait_range(META_MAPPING(sbi), 0, LLONG_MAX);
J
Jaegeuk Kim 已提交
1154 1155 1156

	/* update user_block_counts */
	sbi->last_valid_block_count = sbi->total_valid_block_count;
1157
	percpu_counter_set(&sbi->alloc_valid_block_count, 0);
J
Jaegeuk Kim 已提交
1158 1159

	/* Here, we only have one bio having CP pack */
1160
	sync_meta_pages(sbi, META_FLUSH, LONG_MAX);
J
Jaegeuk Kim 已提交
1161

1162 1163 1164
	/* wait for previous submitted meta pages writeback */
	wait_on_all_pages_writeback(sbi);

1165
	release_ino_entry(sbi, false);
1166 1167

	if (unlikely(f2fs_cp_error(sbi)))
C
Chao Yu 已提交
1168
		return -EIO;
1169

1170
	clear_prefree_segments(sbi, cpc);
1171
	clear_sbi_flag(sbi, SBI_IS_DIRTY);
1172
	clear_sbi_flag(sbi, SBI_NEED_CP);
C
Chao Yu 已提交
1173

1174 1175 1176 1177 1178 1179 1180 1181 1182 1183
	/*
	 * redirty superblock if metadata like node page or inode cache is
	 * updated during writing checkpoint.
	 */
	if (get_pages(sbi, F2FS_DIRTY_NODES) ||
			get_pages(sbi, F2FS_DIRTY_IMETA))
		set_sbi_flag(sbi, SBI_IS_DIRTY);

	f2fs_bug_on(sbi, get_pages(sbi, F2FS_DIRTY_DENTS));

C
Chao Yu 已提交
1184
	return 0;
J
Jaegeuk Kim 已提交
1185 1186
}

J
Jaegeuk Kim 已提交
1187
/*
A
arter97 已提交
1188
 * We guarantee that this checkpoint procedure will not fail.
J
Jaegeuk Kim 已提交
1189
 */
C
Chao Yu 已提交
1190
int write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc)
J
Jaegeuk Kim 已提交
1191 1192 1193
{
	struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
	unsigned long long ckpt_ver;
C
Chao Yu 已提交
1194
	int err = 0;
J
Jaegeuk Kim 已提交
1195

1196
	mutex_lock(&sbi->cp_mutex);
J
Jaegeuk Kim 已提交
1197

1198
	if (!is_sbi_flag_set(sbi, SBI_IS_DIRTY) &&
1199 1200
		(cpc->reason == CP_FASTBOOT || cpc->reason == CP_SYNC ||
		(cpc->reason == CP_DISCARD && !sbi->discard_blks)))
J
Jaegeuk Kim 已提交
1201
		goto out;
C
Chao Yu 已提交
1202 1203
	if (unlikely(f2fs_cp_error(sbi))) {
		err = -EIO;
1204
		goto out;
C
Chao Yu 已提交
1205 1206 1207
	}
	if (f2fs_readonly(sbi->sb)) {
		err = -EROFS;
1208
		goto out;
C
Chao Yu 已提交
1209
	}
W
Wanpeng Li 已提交
1210 1211 1212

	trace_f2fs_write_checkpoint(sbi->sb, cpc->reason, "start block_ops");

C
Chao Yu 已提交
1213 1214
	err = block_operations(sbi);
	if (err)
1215
		goto out;
J
Jaegeuk Kim 已提交
1216

1217
	trace_f2fs_write_checkpoint(sbi->sb, cpc->reason, "finish block_ops");
1218

1219
	f2fs_flush_merged_bios(sbi);
J
Jaegeuk Kim 已提交
1220

1221 1222 1223 1224 1225 1226 1227
	/* this is the case of multiple fstrims without any changes */
	if (cpc->reason == CP_DISCARD && !is_sbi_flag_set(sbi, SBI_IS_DIRTY)) {
		f2fs_bug_on(sbi, NM_I(sbi)->dirty_nat_cnt);
		f2fs_bug_on(sbi, SIT_I(sbi)->dirty_sentries);
		f2fs_bug_on(sbi, prefree_segments(sbi));
		flush_sit_entries(sbi, cpc);
		clear_prefree_segments(sbi, cpc);
C
Chao Yu 已提交
1228
		f2fs_wait_all_discard_bio(sbi);
1229 1230 1231 1232
		unblock_operations(sbi);
		goto out;
	}

J
Jaegeuk Kim 已提交
1233 1234 1235 1236 1237
	/*
	 * update checkpoint pack index
	 * Increase the version number so that
	 * SIT entries and seg summaries are written at correct place
	 */
1238
	ckpt_ver = cur_cp_version(ckpt);
J
Jaegeuk Kim 已提交
1239 1240 1241 1242
	ckpt->checkpoint_ver = cpu_to_le64(++ckpt_ver);

	/* write cached NAT/SIT entries to NAT/SIT area */
	flush_nat_entries(sbi);
1243
	flush_sit_entries(sbi, cpc);
J
Jaegeuk Kim 已提交
1244 1245

	/* unlock all the fs_lock[] in do_checkpoint() */
C
Chao Yu 已提交
1246
	err = do_checkpoint(sbi, cpc);
J
Jaegeuk Kim 已提交
1247

C
Chao Yu 已提交
1248 1249
	f2fs_wait_all_discard_bio(sbi);

J
Jaegeuk Kim 已提交
1250
	unblock_operations(sbi);
1251
	stat_inc_cp_count(sbi->stat_info);
1252 1253 1254 1255

	if (cpc->reason == CP_RECOVERY)
		f2fs_msg(sbi->sb, KERN_NOTICE,
			"checkpoint: version = %llx", ckpt_ver);
1256 1257

	/* do checkpoint periodically */
1258
	f2fs_update_time(sbi, CP_TIME);
1259
	trace_f2fs_write_checkpoint(sbi->sb, cpc->reason, "finish checkpoint");
J
Jaegeuk Kim 已提交
1260 1261
out:
	mutex_unlock(&sbi->cp_mutex);
C
Chao Yu 已提交
1262
	return err;
J
Jaegeuk Kim 已提交
1263 1264
}

J
Jaegeuk Kim 已提交
1265
void init_ino_entry_info(struct f2fs_sb_info *sbi)
J
Jaegeuk Kim 已提交
1266
{
J
Jaegeuk Kim 已提交
1267 1268 1269
	int i;

	for (i = 0; i < MAX_INO_ENTRY; i++) {
1270 1271 1272 1273 1274 1275
		struct inode_management *im = &sbi->im[i];

		INIT_RADIX_TREE(&im->ino_root, GFP_ATOMIC);
		spin_lock_init(&im->ino_lock);
		INIT_LIST_HEAD(&im->ino_list);
		im->ino_num = 0;
J
Jaegeuk Kim 已提交
1276 1277
	}

C
Chao Yu 已提交
1278
	sbi->max_orphans = (sbi->blocks_per_seg - F2FS_CP_PACKS -
1279 1280
			NR_CURSEG_TYPE - __cp_payload(sbi)) *
				F2FS_ORPHANS_PER_BLOCK;
J
Jaegeuk Kim 已提交
1281 1282
}

1283
int __init create_checkpoint_caches(void)
J
Jaegeuk Kim 已提交
1284
{
J
Jaegeuk Kim 已提交
1285 1286 1287
	ino_entry_slab = f2fs_kmem_cache_create("f2fs_ino_entry",
			sizeof(struct ino_entry));
	if (!ino_entry_slab)
J
Jaegeuk Kim 已提交
1288
		return -ENOMEM;
1289 1290
	inode_entry_slab = f2fs_kmem_cache_create("f2fs_inode_entry",
			sizeof(struct inode_entry));
1291
	if (!inode_entry_slab) {
J
Jaegeuk Kim 已提交
1292
		kmem_cache_destroy(ino_entry_slab);
J
Jaegeuk Kim 已提交
1293 1294 1295 1296 1297 1298 1299
		return -ENOMEM;
	}
	return 0;
}

void destroy_checkpoint_caches(void)
{
J
Jaegeuk Kim 已提交
1300
	kmem_cache_destroy(ino_entry_slab);
J
Jaegeuk Kim 已提交
1301 1302
	kmem_cache_destroy(inode_entry_slab);
}