checkpoint.c 29.6 KB
Newer Older
Jaegeuk Kim's avatar
Jaegeuk Kim committed
1
/*
Jaegeuk Kim's avatar
Jaegeuk Kim committed
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"
Jaegeuk Kim's avatar
Jaegeuk Kim committed
23
#include "trace.h"
24
#include <trace/events/f2fs.h>
Jaegeuk Kim's avatar
Jaegeuk Kim committed
25

26
static struct kmem_cache *ino_entry_slab;
27
struct kmem_cache *inode_entry_slab;
Jaegeuk Kim's avatar
Jaegeuk Kim committed
28

Jaegeuk Kim's avatar
Jaegeuk Kim committed
29
/*
Jaegeuk Kim's avatar
Jaegeuk Kim committed
30 31 32 33
 * We guarantee no failure on the returned page.
 */
struct page *grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index)
{
34
	struct address_space *mapping = META_MAPPING(sbi);
Jaegeuk Kim's avatar
Jaegeuk Kim committed
35 36
	struct page *page = NULL;
repeat:
37
	page = grab_cache_page(mapping, index);
Jaegeuk Kim's avatar
Jaegeuk Kim committed
38 39 40 41
	if (!page) {
		cond_resched();
		goto repeat;
	}
42
	f2fs_wait_on_page_writeback(page, META, true);
Jaegeuk Kim's avatar
Jaegeuk Kim committed
43 44 45 46
	SetPageUptodate(page);
	return page;
}

Jaegeuk Kim's avatar
Jaegeuk Kim committed
47
/*
Jaegeuk Kim's avatar
Jaegeuk Kim committed
48 49
 * We guarantee no failure on the returned page.
 */
50 51
static struct page *__get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index,
							bool is_meta)
Jaegeuk Kim's avatar
Jaegeuk Kim committed
52
{
53
	struct address_space *mapping = META_MAPPING(sbi);
Jaegeuk Kim's avatar
Jaegeuk Kim committed
54
	struct page *page;
55
	struct f2fs_io_info fio = {
56
		.sbi = sbi,
57 58
		.type = META,
		.rw = READ_SYNC | REQ_META | REQ_PRIO,
59 60
		.old_blkaddr = index,
		.new_blkaddr = index,
61
		.encrypted_page = NULL,
62
	};
63 64 65

	if (unlikely(!is_meta))
		fio.rw &= ~REQ_META;
Jaegeuk Kim's avatar
Jaegeuk Kim committed
66 67 68 69 70 71
repeat:
	page = grab_cache_page(mapping, index);
	if (!page) {
		cond_resched();
		goto repeat;
	}
72 73 74
	if (PageUptodate(page))
		goto out;

75 76
	fio.page = page;

77 78
	if (f2fs_submit_page_bio(&fio)) {
		f2fs_put_page(page, 1);
Jaegeuk Kim's avatar
Jaegeuk Kim committed
79
		goto repeat;
80
	}
Jaegeuk Kim's avatar
Jaegeuk Kim committed
81

82
	lock_page(page);
83
	if (unlikely(page->mapping != mapping)) {
84 85 86
		f2fs_put_page(page, 1);
		goto repeat;
	}
87 88 89 90 91 92 93 94

	/*
	 * 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)))
		f2fs_stop_checkpoint(sbi);
95
out:
Jaegeuk Kim's avatar
Jaegeuk Kim committed
96 97 98
	return page;
}

99 100 101 102 103 104 105 106 107 108 109
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);
}

110
bool is_valid_blkaddr(struct f2fs_sb_info *sbi, block_t blkaddr, int type)
111 112 113
{
	switch (type) {
	case META_NAT:
114
		break;
115
	case META_SIT:
116 117 118
		if (unlikely(blkaddr >= SIT_BLK_CNT(sbi)))
			return false;
		break;
119
	case META_SSA:
120 121 122 123
		if (unlikely(blkaddr >= MAIN_BLKADDR(sbi) ||
			blkaddr < SM_I(sbi)->ssa_blkaddr))
			return false;
		break;
124
	case META_CP:
125 126 127 128
		if (unlikely(blkaddr >= SIT_I(sbi)->sit_base_addr ||
			blkaddr < __start_cp_addr(sbi)))
			return false;
		break;
129
	case META_POR:
130 131 132 133
		if (unlikely(blkaddr >= MAX_BLKADDR(sbi) ||
			blkaddr < MAIN_BLKADDR(sbi)))
			return false;
		break;
134 135 136
	default:
		BUG();
	}
137 138

	return true;
139 140 141
}

/*
142
 * Readahead CP/NAT/SIT/SSA pages
143
 */
144 145
int ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages,
							int type, bool sync)
146 147
{
	struct page *page;
148
	block_t blkno = start;
149
	struct f2fs_io_info fio = {
150
		.sbi = sbi,
151
		.type = META,
152
		.rw = sync ? (READ_SYNC | REQ_META | REQ_PRIO) : READA,
153
		.encrypted_page = NULL,
154
	};
155
	struct blk_plug plug;
156

157 158 159
	if (unlikely(type == META_POR))
		fio.rw &= ~REQ_META;

160
	blk_start_plug(&plug);
161 162
	for (; nrpages-- > 0; blkno++) {

163 164 165
		if (!is_valid_blkaddr(sbi, blkno, type))
			goto out;

166 167
		switch (type) {
		case META_NAT:
168 169
			if (unlikely(blkno >=
					NAT_BLOCK_OFFSET(NM_I(sbi)->max_nid)))
170
				blkno = 0;
171
			/* get nat block addr */
172
			fio.new_blkaddr = current_nat_addr(sbi,
173 174 175 176
					blkno * NAT_ENTRY_PER_BLOCK);
			break;
		case META_SIT:
			/* get sit block addr */
177
			fio.new_blkaddr = current_sit_addr(sbi,
178 179
					blkno * SIT_ENTRY_PER_BLOCK);
			break;
180
		case META_SSA:
181
		case META_CP:
182
		case META_POR:
183
			fio.new_blkaddr = blkno;
184 185 186 187 188
			break;
		default:
			BUG();
		}

189
		page = grab_cache_page(META_MAPPING(sbi), fio.new_blkaddr);
190 191 192 193 194 195 196
		if (!page)
			continue;
		if (PageUptodate(page)) {
			f2fs_put_page(page, 1);
			continue;
		}

197
		fio.page = page;
198
		fio.old_blkaddr = fio.new_blkaddr;
199
		f2fs_submit_page_mbio(&fio);
200 201 202 203
		f2fs_put_page(page, 0);
	}
out:
	f2fs_submit_merged_bio(sbi, META, READ);
204
	blk_finish_plug(&plug);
205 206 207
	return blkno - start;
}

208 209 210 211 212 213 214 215 216 217 218
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);
	if (!page || (page && !PageUptodate(page)))
		readahead = true;
	f2fs_put_page(page, 0);

	if (readahead)
219
		ra_meta_pages(sbi, index, MAX_BIO_BLOCKS(sbi), META_POR, true);
220 221
}

Jaegeuk Kim's avatar
Jaegeuk Kim committed
222 223 224
static int f2fs_write_meta_page(struct page *page,
				struct writeback_control *wbc)
{
225
	struct f2fs_sb_info *sbi = F2FS_P_SB(page);
Jaegeuk Kim's avatar
Jaegeuk Kim committed
226

227 228
	trace_f2fs_writepage(page, META);

229
	if (unlikely(is_sbi_flag_set(sbi, SBI_POR_DOING)))
230
		goto redirty_out;
231
	if (wbc->for_reclaim && page->index < GET_SUM_BLOCK(sbi, 0))
232
		goto redirty_out;
233
	if (unlikely(f2fs_cp_error(sbi)))
234
		goto redirty_out;
Jaegeuk Kim's avatar
Jaegeuk Kim committed
235

236 237
	write_meta_page(sbi, page);
	dec_page_count(sbi, F2FS_DIRTY_META);
238 239 240 241

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

242
	unlock_page(page);
243

244
	if (unlikely(f2fs_cp_error(sbi)))
245
		f2fs_submit_merged_bio(sbi, META, WRITE);
246

247
	return 0;
248 249

redirty_out:
250
	redirty_page_for_writepage(wbc, page);
251
	return AOP_WRITEPAGE_ACTIVATE;
Jaegeuk Kim's avatar
Jaegeuk Kim committed
252 253 254 255 256
}

static int f2fs_write_meta_pages(struct address_space *mapping,
				struct writeback_control *wbc)
{
257
	struct f2fs_sb_info *sbi = F2FS_M_SB(mapping);
258
	long diff, written;
Jaegeuk Kim's avatar
Jaegeuk Kim committed
259

260
	/* collect a number of dirty meta pages and write together */
261 262
	if (wbc->for_kupdate ||
		get_pages(sbi, F2FS_DIRTY_META) < nr_pages_to_skip(sbi, META))
263
		goto skip_write;
Jaegeuk Kim's avatar
Jaegeuk Kim committed
264

Yunlei He's avatar
Yunlei He committed
265 266
	trace_f2fs_writepages(mapping->host, wbc, META);

Jaegeuk Kim's avatar
Jaegeuk Kim committed
267 268
	/* if mounting is failed, skip writing node pages */
	mutex_lock(&sbi->cp_mutex);
269 270
	diff = nr_pages_to_write(sbi, META, wbc);
	written = sync_meta_pages(sbi, META, wbc->nr_to_write);
Jaegeuk Kim's avatar
Jaegeuk Kim committed
271
	mutex_unlock(&sbi->cp_mutex);
272
	wbc->nr_to_write = max((long)0, wbc->nr_to_write - written - diff);
Jaegeuk Kim's avatar
Jaegeuk Kim committed
273
	return 0;
274 275 276

skip_write:
	wbc->pages_skipped += get_pages(sbi, F2FS_DIRTY_META);
Yunlei He's avatar
Yunlei He committed
277
	trace_f2fs_writepages(mapping->host, wbc, META);
278
	return 0;
Jaegeuk Kim's avatar
Jaegeuk Kim committed
279 280 281 282 283
}

long sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type,
						long nr_to_write)
{
284
	struct address_space *mapping = META_MAPPING(sbi);
285
	pgoff_t index = 0, end = ULONG_MAX, prev = ULONG_MAX;
Jaegeuk Kim's avatar
Jaegeuk Kim committed
286 287 288 289 290
	struct pagevec pvec;
	long nwritten = 0;
	struct writeback_control wbc = {
		.for_reclaim = 0,
	};
291
	struct blk_plug plug;
Jaegeuk Kim's avatar
Jaegeuk Kim committed
292 293 294

	pagevec_init(&pvec, 0);

295 296
	blk_start_plug(&plug);

Jaegeuk Kim's avatar
Jaegeuk Kim committed
297 298 299 300 301
	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);
302
		if (unlikely(nr_pages == 0))
Jaegeuk Kim's avatar
Jaegeuk Kim committed
303 304 305 306
			break;

		for (i = 0; i < nr_pages; i++) {
			struct page *page = pvec.pages[i];
307

308
			if (prev == ULONG_MAX)
309 310 311 312 313 314
				prev = page->index - 1;
			if (nr_to_write != LONG_MAX && page->index != prev + 1) {
				pagevec_release(&pvec);
				goto stop;
			}

Jaegeuk Kim's avatar
Jaegeuk Kim committed
315
			lock_page(page);
316 317 318 319 320 321 322 323 324 325 326

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

327 328 329
			f2fs_wait_on_page_writeback(page, META, true);

			BUG_ON(PageWriteback(page));
330 331 332
			if (!clear_page_dirty_for_io(page))
				goto continue_unlock;

333
			if (mapping->a_ops->writepage(page, &wbc)) {
334 335 336
				unlock_page(page);
				break;
			}
337
			nwritten++;
338
			prev = page->index;
339
			if (unlikely(nwritten >= nr_to_write))
Jaegeuk Kim's avatar
Jaegeuk Kim committed
340 341 342 343 344
				break;
		}
		pagevec_release(&pvec);
		cond_resched();
	}
345
stop:
Jaegeuk Kim's avatar
Jaegeuk Kim committed
346
	if (nwritten)
Jaegeuk Kim's avatar
Jaegeuk Kim committed
347
		f2fs_submit_merged_bio(sbi, type, WRITE);
Jaegeuk Kim's avatar
Jaegeuk Kim committed
348

349 350
	blk_finish_plug(&plug);

Jaegeuk Kim's avatar
Jaegeuk Kim committed
351 352 353 354 355
	return nwritten;
}

static int f2fs_set_meta_page_dirty(struct page *page)
{
356 357
	trace_f2fs_set_page_dirty(page, META);

Jaegeuk Kim's avatar
Jaegeuk Kim committed
358 359 360
	SetPageUptodate(page);
	if (!PageDirty(page)) {
		__set_page_dirty_nobuffers(page);
361
		inc_page_count(F2FS_P_SB(page), F2FS_DIRTY_META);
362
		SetPagePrivate(page);
Jaegeuk Kim's avatar
Jaegeuk Kim committed
363
		f2fs_trace_pid(page);
Jaegeuk Kim's avatar
Jaegeuk Kim committed
364 365 366 367 368 369 370 371 372
		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,
373 374
	.invalidatepage = f2fs_invalidate_page,
	.releasepage	= f2fs_release_page,
Jaegeuk Kim's avatar
Jaegeuk Kim committed
375 376
};

377
static void __add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type)
378
{
379
	struct inode_management *im = &sbi->im[type];
380 381 382
	struct ino_entry *e, *tmp;

	tmp = f2fs_kmem_cache_alloc(ino_entry_slab, GFP_NOFS);
383
retry:
384
	radix_tree_preload(GFP_NOFS | __GFP_NOFAIL);
385

386 387
	spin_lock(&im->ino_lock);
	e = radix_tree_lookup(&im->ino_root, ino);
388
	if (!e) {
389
		e = tmp;
390 391
		if (radix_tree_insert(&im->ino_root, ino, e)) {
			spin_unlock(&im->ino_lock);
392
			radix_tree_preload_end();
393 394 395 396
			goto retry;
		}
		memset(e, 0, sizeof(struct ino_entry));
		e->ino = ino;
397

398
		list_add_tail(&e->list, &im->ino_list);
399
		if (type != ORPHAN_INO)
400
			im->ino_num++;
401
	}
402
	spin_unlock(&im->ino_lock);
403
	radix_tree_preload_end();
404 405 406

	if (e != tmp)
		kmem_cache_free(ino_entry_slab, tmp);
407 408
}

409
static void __remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type)
410
{
411
	struct inode_management *im = &sbi->im[type];
412
	struct ino_entry *e;
413

414 415
	spin_lock(&im->ino_lock);
	e = radix_tree_lookup(&im->ino_root, ino);
416 417
	if (e) {
		list_del(&e->list);
418 419 420
		radix_tree_delete(&im->ino_root, ino);
		im->ino_num--;
		spin_unlock(&im->ino_lock);
421 422
		kmem_cache_free(ino_entry_slab, e);
		return;
423
	}
424
	spin_unlock(&im->ino_lock);
425 426
}

427
void add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type)
428 429 430 431 432
{
	/* add new dirty ino entry into list */
	__add_ino_entry(sbi, ino, type);
}

433
void remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type)
434 435 436 437 438 439 440 441
{
	/* 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)
{
442
	struct inode_management *im = &sbi->im[mode];
443
	struct ino_entry *e;
444 445 446 447

	spin_lock(&im->ino_lock);
	e = radix_tree_lookup(&im->ino_root, ino);
	spin_unlock(&im->ino_lock);
448 449 450
	return e ? true : false;
}

451
void release_ino_entry(struct f2fs_sb_info *sbi)
452 453 454 455 456
{
	struct ino_entry *e, *tmp;
	int i;

	for (i = APPEND_INO; i <= UPDATE_INO; i++) {
457 458 459 460
		struct inode_management *im = &sbi->im[i];

		spin_lock(&im->ino_lock);
		list_for_each_entry_safe(e, tmp, &im->ino_list, list) {
461
			list_del(&e->list);
462
			radix_tree_delete(&im->ino_root, e->ino);
463
			kmem_cache_free(ino_entry_slab, e);
464
			im->ino_num--;
465
		}
466
		spin_unlock(&im->ino_lock);
467 468 469
	}
}

Jaegeuk Kim's avatar
Jaegeuk Kim committed
470
int acquire_orphan_inode(struct f2fs_sb_info *sbi)
Jaegeuk Kim's avatar
Jaegeuk Kim committed
471
{
472
	struct inode_management *im = &sbi->im[ORPHAN_INO];
Jaegeuk Kim's avatar
Jaegeuk Kim committed
473 474
	int err = 0;

475 476
	spin_lock(&im->ino_lock);
	if (unlikely(im->ino_num >= sbi->max_orphans))
Jaegeuk Kim's avatar
Jaegeuk Kim committed
477
		err = -ENOSPC;
Jaegeuk Kim's avatar
Jaegeuk Kim committed
478
	else
479 480
		im->ino_num++;
	spin_unlock(&im->ino_lock);
481

Jaegeuk Kim's avatar
Jaegeuk Kim committed
482 483 484
	return err;
}

Jaegeuk Kim's avatar
Jaegeuk Kim committed
485 486
void release_orphan_inode(struct f2fs_sb_info *sbi)
{
487 488 489 490 491 492
	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);
Jaegeuk Kim's avatar
Jaegeuk Kim committed
493 494
}

Jaegeuk Kim's avatar
Jaegeuk Kim committed
495 496
void add_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino)
{
497
	/* add new orphan ino entry into list */
498
	__add_ino_entry(sbi, ino, ORPHAN_INO);
Jaegeuk Kim's avatar
Jaegeuk Kim committed
499 500 501 502
}

void remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino)
{
503
	/* remove orphan entry from orphan list */
504
	__remove_ino_entry(sbi, ino, ORPHAN_INO);
Jaegeuk Kim's avatar
Jaegeuk Kim committed
505 506
}

507
static int recover_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino)
Jaegeuk Kim's avatar
Jaegeuk Kim committed
508
{
509 510 511 512 513 514 515 516 517 518 519 520
	struct inode *inode;

	inode = f2fs_iget(sbi->sb, ino);
	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);
	}

Jaegeuk Kim's avatar
Jaegeuk Kim committed
521 522 523 524
	clear_nlink(inode);

	/* truncate all the data during iput */
	iput(inode);
525
	return 0;
Jaegeuk Kim's avatar
Jaegeuk Kim committed
526 527
}

528
int recover_orphan_inodes(struct f2fs_sb_info *sbi)
Jaegeuk Kim's avatar
Jaegeuk Kim committed
529
{
530
	block_t start_blk, orphan_blocks, i, j;
531
	int err;
Jaegeuk Kim's avatar
Jaegeuk Kim committed
532

533
	if (!is_set_ckpt_flags(F2FS_CKPT(sbi), CP_ORPHAN_PRESENT_FLAG))
534
		return 0;
Jaegeuk Kim's avatar
Jaegeuk Kim committed
535

536
	start_blk = __start_cp_addr(sbi) + 1 + __cp_payload(sbi);
537
	orphan_blocks = __start_sum_addr(sbi) - 1 - __cp_payload(sbi);
Jaegeuk Kim's avatar
Jaegeuk Kim committed
538

539
	ra_meta_pages(sbi, start_blk, orphan_blocks, META_CP, true);
540

541
	for (i = 0; i < orphan_blocks; i++) {
Jaegeuk Kim's avatar
Jaegeuk Kim committed
542 543 544 545 546 547
		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]);
548 549 550 551 552
			err = recover_orphan_inode(sbi, ino);
			if (err) {
				f2fs_put_page(page, 1);
				return err;
			}
Jaegeuk Kim's avatar
Jaegeuk Kim committed
553 554 555 556
		}
		f2fs_put_page(page, 1);
	}
	/* clear Orphan Flag */
557
	clear_ckpt_flags(F2FS_CKPT(sbi), CP_ORPHAN_PRESENT_FLAG);
558
	return 0;
Jaegeuk Kim's avatar
Jaegeuk Kim committed
559 560 561 562
}

static void write_orphan_inodes(struct f2fs_sb_info *sbi, block_t start_blk)
{
563
	struct list_head *head;
Jaegeuk Kim's avatar
Jaegeuk Kim committed
564 565
	struct f2fs_orphan_block *orphan_blk = NULL;
	unsigned int nentries = 0;
Chao Yu's avatar
Chao Yu committed
566
	unsigned short index = 1;
567
	unsigned short orphan_blocks;
568
	struct page *page = NULL;
569
	struct ino_entry *orphan = NULL;
570
	struct inode_management *im = &sbi->im[ORPHAN_INO];
Jaegeuk Kim's avatar
Jaegeuk Kim committed
571

572
	orphan_blocks = GET_ORPHAN_BLOCKS(im->ino_num);
573

574 575 576 577 578
	/*
	 * 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.
	 */
579
	head = &im->ino_list;
Jaegeuk Kim's avatar
Jaegeuk Kim committed
580 581

	/* loop for each orphan inode entry and write them in Jornal block */
582 583
	list_for_each_entry(orphan, head, list) {
		if (!page) {
Chao Yu's avatar
Chao Yu committed
584
			page = grab_meta_page(sbi, start_blk++);
585 586 587 588
			orphan_blk =
				(struct f2fs_orphan_block *)page_address(page);
			memset(orphan_blk, 0, sizeof(*orphan_blk));
		}
Jaegeuk Kim's avatar
Jaegeuk Kim committed
589

590
		orphan_blk->ino[nentries++] = cpu_to_le32(orphan->ino);
Jaegeuk Kim's avatar
Jaegeuk Kim committed
591

592
		if (nentries == F2FS_ORPHANS_PER_BLOCK) {
Jaegeuk Kim's avatar
Jaegeuk Kim committed
593 594 595 596 597 598 599 600 601 602 603 604 605 606
			/*
			 * 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;
		}
607
	}
Jaegeuk Kim's avatar
Jaegeuk Kim committed
608

609 610 611 612 613 614
	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);
Jaegeuk Kim's avatar
Jaegeuk Kim committed
615 616 617 618 619 620 621 622 623 624 625
	}
}

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;
Jaegeuk Kim's avatar
Jaegeuk Kim committed
626
	__u32 crc = 0;
Jaegeuk Kim's avatar
Jaegeuk Kim committed
627 628 629 630 631 632 633 634 635 636

	/* 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;

Jaegeuk Kim's avatar
Jaegeuk Kim committed
637
	crc = le32_to_cpu(*((__le32 *)((unsigned char *)cp_block + crc_offset)));
Jaegeuk Kim's avatar
Jaegeuk Kim committed
638 639 640
	if (!f2fs_crc_valid(crc, cp_block, crc_offset))
		goto invalid_cp1;

641
	pre_version = cur_cp_version(cp_block);
Jaegeuk Kim's avatar
Jaegeuk Kim committed
642 643

	/* Read the 2nd cp block in this CP pack */
644
	cp_addr += le32_to_cpu(cp_block->cp_pack_total_block_count) - 1;
Jaegeuk Kim's avatar
Jaegeuk Kim committed
645 646 647 648 649 650 651
	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;

Jaegeuk Kim's avatar
Jaegeuk Kim committed
652
	crc = le32_to_cpu(*((__le32 *)((unsigned char *)cp_block + crc_offset)));
Jaegeuk Kim's avatar
Jaegeuk Kim committed
653 654 655
	if (!f2fs_crc_valid(crc, cp_block, crc_offset))
		goto invalid_cp2;

656
	cur_version = cur_cp_version(cp_block);
Jaegeuk Kim's avatar
Jaegeuk Kim committed
657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677

	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;
678
	unsigned int cp_blks = 1 + __cp_payload(sbi);
Changman Lee's avatar
Changman Lee committed
679 680
	block_t cp_blk_no;
	int i;
Jaegeuk Kim's avatar
Jaegeuk Kim committed
681

Changman Lee's avatar
Changman Lee committed
682
	sbi->ckpt = kzalloc(cp_blks * blk_size, GFP_KERNEL);
Jaegeuk Kim's avatar
Jaegeuk Kim committed
683 684 685 686 687 688 689 690 691 692
	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 */
693 694
	cp_start_blk_no += ((unsigned long long)1) <<
				le32_to_cpu(fsb->log_blocks_per_seg);
Jaegeuk Kim's avatar
Jaegeuk Kim committed
695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712
	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);

713 714 715 716
	/* Sanity checking of checkpoint */
	if (sanity_check_ckpt(sbi))
		goto fail_no_cp;

Changman Lee's avatar
Changman Lee committed
717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733
	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:
Jaegeuk Kim's avatar
Jaegeuk Kim committed
734 735 736 737 738 739 740 741 742
	f2fs_put_page(cp1, 1);
	f2fs_put_page(cp2, 1);
	return 0;

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

743
static void __add_dirty_inode(struct inode *inode, enum inode_type type)
Jaegeuk Kim's avatar
Jaegeuk Kim committed
744
{
745
	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
746
	struct f2fs_inode_info *fi = F2FS_I(inode);
747
	int flag = (type == DIR_INODE) ? FI_DIRTY_DIR : FI_DIRTY_FILE;
Jaegeuk Kim's avatar
Jaegeuk Kim committed
748

749
	if (is_inode_flag_set(fi, flag))
750
		return;
751

752 753
	set_inode_flag(fi, flag);
	list_add_tail(&fi->dirty_list, &sbi->inode_list[type]);
754
	stat_inc_dirty_inode(sbi, type);
755 756
}

757
static void __remove_dirty_inode(struct inode *inode, enum inode_type type)
Chao Yu's avatar
Chao Yu committed
758 759
{
	struct f2fs_inode_info *fi = F2FS_I(inode);
760
	int flag = (type == DIR_INODE) ? FI_DIRTY_DIR : FI_DIRTY_FILE;
Chao Yu's avatar
Chao Yu committed
761 762

	if (get_dirty_pages(inode) ||
763
			!is_inode_flag_set(F2FS_I(inode), flag))
Chao Yu's avatar
Chao Yu committed
764 765 766
		return;

	list_del_init(&fi->dirty_list);
767
	clear_inode_flag(fi, flag);
768
	stat_dec_dirty_inode(F2FS_I_SB(inode), type);
Chao Yu's avatar
Chao Yu committed
769 770
}

771
void update_dirty_page(struct inode *inode, struct page *page)
772
{
773
	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
774
	enum inode_type type = S_ISDIR(inode->i_mode) ? DIR_INODE : FILE_INODE;
775

776 777
	if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) &&
			!S_ISLNK(inode->i_mode))
Jaegeuk Kim's avatar
Jaegeuk Kim committed
778
		return;
779

780 781
	spin_lock(&sbi->inode_lock[type]);
	__add_dirty_inode(inode, type);
782
	inode_inc_dirty_pages(inode);
783
	spin_unlock(&sbi->inode_lock[type]);
784

785
	SetPagePrivate(page);
Jaegeuk Kim's avatar
Jaegeuk Kim committed
786
	f2fs_trace_pid(page);
787 788 789 790
}

void add_dirty_dir_inode(struct inode *inode)
{
791
	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
Jaegeuk Kim's avatar
Jaegeuk Kim committed
792

793 794 795
	spin_lock(&sbi->inode_lock[DIR_INODE]);
	__add_dirty_inode(inode, DIR_INODE);
	spin_unlock(&sbi->inode_lock[DIR_INODE]);
Jaegeuk Kim's avatar
Jaegeuk Kim committed
796 797
}

798
void remove_dirty_inode(struct inode *inode)
Jaegeuk Kim's avatar
Jaegeuk Kim committed
799
{
800
	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
801
	struct f2fs_inode_info *fi = F2FS_I(inode);
802
	enum inode_type type = S_ISDIR(inode->i_mode) ? DIR_INODE : FILE_INODE;
Jaegeuk Kim's avatar
Jaegeuk Kim committed
803

804 805
	if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) &&
			!S_ISLNK(inode->i_mode))
Jaegeuk Kim's avatar
Jaegeuk Kim committed
806 807
		return;

808 809 810
	spin_lock(&sbi->inode_lock[type]);
	__remove_dirty_inode(inode, type);
	spin_unlock(&sbi->inode_lock[type]);
811 812

	/* Only from the recovery routine */
813 814
	if (is_inode_flag_set(fi, FI_DELAY_IPUT)) {
		clear_inode_flag(fi, FI_DELAY_IPUT);
815
		iput(inode);
816
	}
817 818
}

819
int sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type)
Jaegeuk Kim's avatar
Jaegeuk Kim committed
820
{
821
	struct list_head *head;
Jaegeuk Kim's avatar
Jaegeuk Kim committed
822
	struct inode *inode;
823
	struct f2fs_inode_info *fi;
824 825 826 827 828
	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));
Jaegeuk Kim's avatar
Jaegeuk Kim committed
829
retry:
830
	if (unlikely(f2fs_cp_error(sbi)))
831
		return -EIO;
832

833
	spin_lock(&sbi->inode_lock[type]);
834

835
	head = &sbi->inode_list[type];
Jaegeuk Kim's avatar
Jaegeuk Kim committed
836
	if (list_empty(head)) {
837
		spin_unlock(&sbi->inode_lock[type]);
838 839 840
		trace_f2fs_sync_dirty_inodes_exit(sbi->sb, is_dir,
				get_pages(sbi, is_dir ?
				F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA));
841
		return 0;
Jaegeuk Kim's avatar
Jaegeuk Kim committed
842
	}
843 844
	fi = list_entry(head->next, struct f2fs_inode_info, dirty_list);
	inode = igrab(&fi->vfs_inode);
845
	spin_unlock(&sbi->inode_lock[type]);
Jaegeuk Kim's avatar
Jaegeuk Kim committed
846
	if (inode) {