[deliverable/linux.git] / fs / f2fs / recovery.c

/*
 * fs/f2fs/recovery.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 "f2fs.h"
#include "node.h"
#include "segment.h"

static struct kmem_cache *fsync_entry_slab;

bool space_for_roll_forward(struct f2fs_sb_info *sbi)
{
	if (sbi->last_valid_block_count + sbi->alloc_valid_block_count
			> sbi->user_block_count)
		return false;
	return true;
}

static struct fsync_inode_entry *get_fsync_inode(struct list_head *head,
								nid_t ino)
{
	struct list_head *this;
	struct fsync_inode_entry *entry;

	list_for_each(this, head) {
		entry = list_entry(this, struct fsync_inode_entry, list);
		if (entry->inode->i_ino == ino)
			return entry;
	}
	return NULL;
}

static int recover_dentry(struct page *ipage, struct inode *inode)
{
	void *kaddr = page_address(ipage);
	struct f2fs_node *raw_node = (struct f2fs_node *)kaddr;
	struct f2fs_inode *raw_inode = &(raw_node->i);
	nid_t pino = le32_to_cpu(raw_inode->i_pino);
	struct f2fs_dir_entry *de;
	struct qstr name;
	struct page *page;
	struct inode *dir, *einode;
	int err = 0;

	dir = check_dirty_dir_inode(F2FS_SB(inode->i_sb), pino);
	if (!dir) {
		dir = f2fs_iget(inode->i_sb, pino);
		if (IS_ERR(dir)) {
			err = PTR_ERR(dir);
			goto out;
		}
		set_inode_flag(F2FS_I(dir), FI_DELAY_IPUT);
		add_dirty_dir_inode(dir);
	}

	name.len = le32_to_cpu(raw_inode->i_namelen);
	name.name = raw_inode->i_name;
retry:
	de = f2fs_find_entry(dir, &name, &page);
	if (de && inode->i_ino == le32_to_cpu(de->ino)) {
		kunmap(page);
		f2fs_put_page(page, 0);
		goto out;
	}
	if (de) {
		einode = f2fs_iget(inode->i_sb, le32_to_cpu(de->ino));
		if (IS_ERR(einode)) {
			WARN_ON(1);
			if (PTR_ERR(einode) == -ENOENT)
				err = -EEXIST;
			goto out;
		}
		f2fs_delete_entry(de, page, einode);
		iput(einode);
		goto retry;
	}
	err = __f2fs_add_link(dir, &name, inode);
out:
	f2fs_msg(inode->i_sb, KERN_NOTICE, "recover_inode and its dentry: "
			"ino = %x, name = %s, dir = %lx, err = %d",
			ino_of_node(ipage), raw_inode->i_name,
			IS_ERR(dir) ? 0 : dir->i_ino, err);
	return err;
}

static int recover_inode(struct inode *inode, struct page *node_page)
{
	void *kaddr = page_address(node_page);
	struct f2fs_node *raw_node = (struct f2fs_node *)kaddr;
	struct f2fs_inode *raw_inode = &(raw_node->i);

	if (!IS_INODE(node_page))
		return 0;

	inode->i_mode = le16_to_cpu(raw_inode->i_mode);
	i_size_write(inode, le64_to_cpu(raw_inode->i_size));
	inode->i_atime.tv_sec = le64_to_cpu(raw_inode->i_mtime);
	inode->i_ctime.tv_sec = le64_to_cpu(raw_inode->i_ctime);
	inode->i_mtime.tv_sec = le64_to_cpu(raw_inode->i_mtime);
	inode->i_atime.tv_nsec = le32_to_cpu(raw_inode->i_mtime_nsec);
	inode->i_ctime.tv_nsec = le32_to_cpu(raw_inode->i_ctime_nsec);
	inode->i_mtime.tv_nsec = le32_to_cpu(raw_inode->i_mtime_nsec);

	if (is_dent_dnode(node_page))
		return recover_dentry(node_page, inode);

	f2fs_msg(inode->i_sb, KERN_NOTICE, "recover_inode: ino = %x, name = %s",
			ino_of_node(node_page), raw_inode->i_name);
	return 0;
}

static int find_fsync_dnodes(struct f2fs_sb_info *sbi, struct list_head *head)
{
	unsigned long long cp_ver = le64_to_cpu(sbi->ckpt->checkpoint_ver);
	struct curseg_info *curseg;
	struct page *page;
	block_t blkaddr;
	int err = 0;

	/* get node pages in the current segment */
	curseg = CURSEG_I(sbi, CURSEG_WARM_NODE);
	blkaddr = START_BLOCK(sbi, curseg->segno) + curseg->next_blkoff;

	/* read node page */
	page = alloc_page(GFP_F2FS_ZERO);
	if (IS_ERR(page))
		return PTR_ERR(page);
	lock_page(page);

	while (1) {
		struct fsync_inode_entry *entry;

		err = f2fs_readpage(sbi, page, blkaddr, READ_SYNC);
		if (err)
			goto out;

		lock_page(page);

		if (cp_ver != cpver_of_node(page))
			break;

		if (!is_fsync_dnode(page))
			goto next;

		entry = get_fsync_inode(head, ino_of_node(page));
		if (entry) {
			if (IS_INODE(page) && is_dent_dnode(page))
				set_inode_flag(F2FS_I(entry->inode),
							FI_INC_LINK);
		} else {
			if (IS_INODE(page) && is_dent_dnode(page)) {
				err = recover_inode_page(sbi, page);
				if (err)
					break;
			}

			/* add this fsync inode to the list */
			entry = kmem_cache_alloc(fsync_entry_slab, GFP_NOFS);
			if (!entry) {
				err = -ENOMEM;
				break;
			}

			entry->inode = f2fs_iget(sbi->sb, ino_of_node(page));
			if (IS_ERR(entry->inode)) {
				err = PTR_ERR(entry->inode);
				kmem_cache_free(fsync_entry_slab, entry);
				break;
			}
			list_add_tail(&entry->list, head);
		}
		entry->blkaddr = blkaddr;

		err = recover_inode(entry->inode, page);
		if (err && err != -ENOENT)
			break;
next:
		/* check next segment */
		blkaddr = next_blkaddr_of_node(page);
	}
	unlock_page(page);
out:
	__free_pages(page, 0);
	return err;
}

static void destroy_fsync_dnodes(struct f2fs_sb_info *sbi,
					struct list_head *head)
{
	struct fsync_inode_entry *entry, *tmp;

	list_for_each_entry_safe(entry, tmp, head, list) {
		iput(entry->inode);
		list_del(&entry->list);
		kmem_cache_free(fsync_entry_slab, entry);
	}
}

static int check_index_in_prev_nodes(struct f2fs_sb_info *sbi,
			block_t blkaddr, struct dnode_of_data *dn)
{
	struct seg_entry *sentry;
	unsigned int segno = GET_SEGNO(sbi, blkaddr);
	unsigned short blkoff = GET_SEGOFF_FROM_SEG0(sbi, blkaddr) &
					(sbi->blocks_per_seg - 1);
	struct f2fs_summary sum;
	nid_t ino, nid;
	void *kaddr;
	struct inode *inode;
	struct page *node_page;
	block_t bidx;
	int i;

	sentry = get_seg_entry(sbi, segno);
	if (!f2fs_test_bit(blkoff, sentry->cur_valid_map))
		return 0;

	/* Get the previous summary */
	for (i = CURSEG_WARM_DATA; i <= CURSEG_COLD_DATA; i++) {
		struct curseg_info *curseg = CURSEG_I(sbi, i);
		if (curseg->segno == segno) {
			sum = curseg->sum_blk->entries[blkoff];
			break;
		}
	}
	if (i > CURSEG_COLD_DATA) {
		struct page *sum_page = get_sum_page(sbi, segno);
		struct f2fs_summary_block *sum_node;
		kaddr = page_address(sum_page);
		sum_node = (struct f2fs_summary_block *)kaddr;
		sum = sum_node->entries[blkoff];
		f2fs_put_page(sum_page, 1);
	}

	/* Use the locked dnode page and inode */
	nid = le32_to_cpu(sum.nid);
	if (dn->inode->i_ino == nid) {
		struct dnode_of_data tdn = *dn;
		tdn.nid = nid;
		tdn.node_page = dn->inode_page;
		tdn.ofs_in_node = le16_to_cpu(sum.ofs_in_node);
		truncate_data_blocks_range(&tdn, 1);
		return 0;
	} else if (dn->nid == nid) {
		struct dnode_of_data tdn = *dn;
		tdn.ofs_in_node = le16_to_cpu(sum.ofs_in_node);
		truncate_data_blocks_range(&tdn, 1);
		return 0;
	}

	/* Get the node page */
	node_page = get_node_page(sbi, nid);
	if (IS_ERR(node_page))
		return PTR_ERR(node_page);
	bidx = start_bidx_of_node(ofs_of_node(node_page)) +
					le16_to_cpu(sum.ofs_in_node);
	ino = ino_of_node(node_page);
	f2fs_put_page(node_page, 1);

	/* Deallocate previous index in the node page */
	inode = f2fs_iget(sbi->sb, ino);
	if (IS_ERR(inode))
		return PTR_ERR(inode);

	truncate_hole(inode, bidx, bidx + 1);
	iput(inode);
	return 0;
}

static int do_recover_data(struct f2fs_sb_info *sbi, struct inode *inode,
					struct page *page, block_t blkaddr)
{
	unsigned int start, end;
	struct dnode_of_data dn;
	struct f2fs_summary sum;
	struct node_info ni;
	int err = 0, recovered = 0;
	int ilock;

	start = start_bidx_of_node(ofs_of_node(page));
	if (IS_INODE(page))
		end = start + ADDRS_PER_INODE;
	else
		end = start + ADDRS_PER_BLOCK;

	ilock = mutex_lock_op(sbi);
	set_new_dnode(&dn, inode, NULL, NULL, 0);

	err = get_dnode_of_data(&dn, start, ALLOC_NODE);
	if (err) {
		mutex_unlock_op(sbi, ilock);
		return err;
	}

	wait_on_page_writeback(dn.node_page);

	get_node_info(sbi, dn.nid, &ni);
	BUG_ON(ni.ino != ino_of_node(page));
	BUG_ON(ofs_of_node(dn.node_page) != ofs_of_node(page));

	for (; start < end; start++) {
		block_t src, dest;

		src = datablock_addr(dn.node_page, dn.ofs_in_node);
		dest = datablock_addr(page, dn.ofs_in_node);

		if (src != dest && dest != NEW_ADDR && dest != NULL_ADDR) {
			if (src == NULL_ADDR) {
				int err = reserve_new_block(&dn);
				/* We should not get -ENOSPC */
				BUG_ON(err);
			}

			/* Check the previous node page having this index */
			err = check_index_in_prev_nodes(sbi, dest, &dn);
			if (err)
				goto err;

			set_summary(&sum, dn.nid, dn.ofs_in_node, ni.version);

			/* write dummy data page */
			recover_data_page(sbi, NULL, &sum, src, dest);
			update_extent_cache(dest, &dn);
			recovered++;
		}
		dn.ofs_in_node++;
	}

	/* write node page in place */
	set_summary(&sum, dn.nid, 0, 0);
	if (IS_INODE(dn.node_page))
		sync_inode_page(&dn);

	copy_node_footer(dn.node_page, page);
	fill_node_footer(dn.node_page, dn.nid, ni.ino,
					ofs_of_node(page), false);
	set_page_dirty(dn.node_page);

	recover_node_page(sbi, dn.node_page, &sum, &ni, blkaddr);
err:
	f2fs_put_dnode(&dn);
	mutex_unlock_op(sbi, ilock);

	f2fs_msg(sbi->sb, KERN_NOTICE, "recover_data: ino = %lx, "
			"recovered_data = %d blocks, err = %d",
			inode->i_ino, recovered, err);
	return err;
}

static int recover_data(struct f2fs_sb_info *sbi,
				struct list_head *head, int type)
{
	unsigned long long cp_ver = le64_to_cpu(sbi->ckpt->checkpoint_ver);
	struct curseg_info *curseg;
	struct page *page;
	int err = 0;
	block_t blkaddr;

	/* get node pages in the current segment */
	curseg = CURSEG_I(sbi, type);
	blkaddr = NEXT_FREE_BLKADDR(sbi, curseg);

	/* read node page */
	page = alloc_page(GFP_NOFS | __GFP_ZERO);
	if (IS_ERR(page))
		return -ENOMEM;

	lock_page(page);

	while (1) {
		struct fsync_inode_entry *entry;

		err = f2fs_readpage(sbi, page, blkaddr, READ_SYNC);
		if (err)
			goto out;

		lock_page(page);

		if (cp_ver != cpver_of_node(page))
			break;

		entry = get_fsync_inode(head, ino_of_node(page));
		if (!entry)
			goto next;

		err = do_recover_data(sbi, entry->inode, page, blkaddr);
		if (err)
			break;

		if (entry->blkaddr == blkaddr) {
			iput(entry->inode);
			list_del(&entry->list);
			kmem_cache_free(fsync_entry_slab, entry);
		}
next:
		/* check next segment */
		blkaddr = next_blkaddr_of_node(page);
	}
	unlock_page(page);
out:
	__free_pages(page, 0);

	if (!err)
		allocate_new_segments(sbi);
	return err;
}

int recover_fsync_data(struct f2fs_sb_info *sbi)
{
	struct list_head inode_list;
	int err;

	fsync_entry_slab = f2fs_kmem_cache_create("f2fs_fsync_inode_entry",
			sizeof(struct fsync_inode_entry), NULL);
	if (unlikely(!fsync_entry_slab))
		return -ENOMEM;

	INIT_LIST_HEAD(&inode_list);

	/* step #1: find fsynced inode numbers */
	sbi->por_doing = 1;
	err = find_fsync_dnodes(sbi, &inode_list);
	if (err)
		goto out;

	if (list_empty(&inode_list))
		goto out;

	/* step #2: recover data */
	err = recover_data(sbi, &inode_list, CURSEG_WARM_NODE);
	BUG_ON(!list_empty(&inode_list));
out:
	destroy_fsync_dnodes(sbi, &inode_list);
	kmem_cache_destroy(fsync_entry_slab);
	sbi->por_doing = 0;
	if (!err)
		write_checkpoint(sbi, false);
	return err;
}
Commit	Line	Data
0a8165d7	1	/*
d624c96f JK	2	* fs/f2fs/recovery.c
	3	*
	4	* Copyright (c) 2012 Samsung Electronics Co., Ltd.
	5	* http://www.samsung.com/
	6	*
	7	* This program is free software; you can redistribute it and/or modify
	8	* it under the terms of the GNU General Public License version 2 as
	9	* published by the Free Software Foundation.
	10	*/
	11	#include <linux/fs.h>
	12	#include <linux/f2fs_fs.h>
	13	#include "f2fs.h"
	14	#include "node.h"
	15	#include "segment.h"
	16
	17	static struct kmem_cache *fsync_entry_slab;
	18
	19	bool space_for_roll_forward(struct f2fs_sb_info *sbi)
	20	{
	21	if (sbi->last_valid_block_count + sbi->alloc_valid_block_count
	22	> sbi->user_block_count)
	23	return false;
	24	return true;
	25	}
	26
	27	static struct fsync_inode_entry get_fsync_inode(struct list_head head,
	28	nid_t ino)
	29	{
	30	struct list_head *this;
	31	struct fsync_inode_entry *entry;
	32
	33	list_for_each(this, head) {
	34	entry = list_entry(this, struct fsync_inode_entry, list);
	35	if (entry->inode->i_ino == ino)
	36	return entry;
	37	}
	38	return NULL;
	39	}
	40
	41	static int recover_dentry(struct page ipage, struct inode inode)
	42	{
9a55ed65 JK	43	void *kaddr = page_address(ipage);
9a55ed65 JK	44	struct f2fs_node raw_node = (struct f2fs_node )kaddr;
d624c96f	45	struct f2fs_inode *raw_inode = &(raw_node->i);
74d0b917	46	nid_t pino = le32_to_cpu(raw_inode->i_pino);
6b8213d9	47	struct f2fs_dir_entry *de;
b7f7a5e0	48	struct qstr name;
d624c96f	49	struct page *page;
6b8213d9	50	struct inode dir, einode;
d624c96f JK	51	int err = 0;
d624c96f JK	52
74d0b917 JK	53	dir = check_dirty_dir_inode(F2FS_SB(inode->i_sb), pino);
	54	if (!dir) {
	55	dir = f2fs_iget(inode->i_sb, pino);
	56	if (IS_ERR(dir)) {
	57	err = PTR_ERR(dir);
	58	goto out;
	59	}
	60	set_inode_flag(F2FS_I(dir), FI_DELAY_IPUT);
5deb8267	61	add_dirty_dir_inode(dir);
d624c96f JK	62	}
d624c96f JK	63
b7f7a5e0 AV	64	name.len = le32_to_cpu(raw_inode->i_namelen);
b7f7a5e0 AV	65	name.name = raw_inode->i_name;
6b8213d9 JK	66	retry:
	67	de = f2fs_find_entry(dir, &name, &page);
	68	if (de && inode->i_ino == le32_to_cpu(de->ino)) {
d624c96f JK	69	kunmap(page);
d624c96f JK	70	f2fs_put_page(page, 0);
6b8213d9 JK	71	goto out;
	72	}
	73	if (de) {
	74	einode = f2fs_iget(inode->i_sb, le32_to_cpu(de->ino));
	75	if (IS_ERR(einode)) {
	76	WARN_ON(1);
	77	if (PTR_ERR(einode) == -ENOENT)
	78	err = -EEXIST;
	79	goto out;
	80	}
	81	f2fs_delete_entry(de, page, einode);
	82	iput(einode);
	83	goto retry;
d624c96f	84	}
6b8213d9	85	err = __f2fs_add_link(dir, &name, inode);
d624c96f	86	out:
f356fe0c JK	87	f2fs_msg(inode->i_sb, KERN_NOTICE, "recover_inode and its dentry: "
f356fe0c JK	88	"ino = %x, name = %s, dir = %lx, err = %d",
f28c06fa DC	89	ino_of_node(ipage), raw_inode->i_name,
f28c06fa DC	90	IS_ERR(dir) ? 0 : dir->i_ino, err);
d624c96f JK	91	return err;
	92	}
	93
	94	static int recover_inode(struct inode inode, struct page node_page)
	95	{
	96	void *kaddr = page_address(node_page);
	97	struct f2fs_node raw_node = (struct f2fs_node )kaddr;
	98	struct f2fs_inode *raw_inode = &(raw_node->i);
	99
f356fe0c JK	100	if (!IS_INODE(node_page))
	101	return 0;
	102
25ca923b	103	inode->i_mode = le16_to_cpu(raw_inode->i_mode);
d624c96f JK	104	i_size_write(inode, le64_to_cpu(raw_inode->i_size));
	105	inode->i_atime.tv_sec = le64_to_cpu(raw_inode->i_mtime);
	106	inode->i_ctime.tv_sec = le64_to_cpu(raw_inode->i_ctime);
	107	inode->i_mtime.tv_sec = le64_to_cpu(raw_inode->i_mtime);
	108	inode->i_atime.tv_nsec = le32_to_cpu(raw_inode->i_mtime_nsec);
	109	inode->i_ctime.tv_nsec = le32_to_cpu(raw_inode->i_ctime_nsec);
	110	inode->i_mtime.tv_nsec = le32_to_cpu(raw_inode->i_mtime_nsec);
	111
f356fe0c JK	112	if (is_dent_dnode(node_page))
	113	return recover_dentry(node_page, inode);
	114
	115	f2fs_msg(inode->i_sb, KERN_NOTICE, "recover_inode: ino = %x, name = %s",
	116	ino_of_node(node_page), raw_inode->i_name);
	117	return 0;
d624c96f JK	118	}
	119
	120	static int find_fsync_dnodes(struct f2fs_sb_info sbi, struct list_head head)
	121	{
	122	unsigned long long cp_ver = le64_to_cpu(sbi->ckpt->checkpoint_ver);
	123	struct curseg_info *curseg;
	124	struct page *page;
	125	block_t blkaddr;
	126	int err = 0;
	127
	128	/* get node pages in the current segment */
	129	curseg = CURSEG_I(sbi, CURSEG_WARM_NODE);
	130	blkaddr = START_BLOCK(sbi, curseg->segno) + curseg->next_blkoff;
	131
	132	/* read node page */
	133	page = alloc_page(GFP_F2FS_ZERO);
	134	if (IS_ERR(page))
	135	return PTR_ERR(page);
	136	lock_page(page);
	137
	138	while (1) {
	139	struct fsync_inode_entry *entry;
	140
393ff91f JK	141	err = f2fs_readpage(sbi, page, blkaddr, READ_SYNC);
393ff91f JK	142	if (err)
d624c96f JK	143	goto out;
d624c96f JK	144
393ff91f JK	145	lock_page(page);
393ff91f JK	146
6ead1142	147	if (cp_ver != cpver_of_node(page))
f356fe0c	148	break;
d624c96f JK	149
	150	if (!is_fsync_dnode(page))
	151	goto next;
	152
	153	entry = get_fsync_inode(head, ino_of_node(page));
	154	if (entry) {
d624c96f JK	155	if (IS_INODE(page) && is_dent_dnode(page))
	156	set_inode_flag(F2FS_I(entry->inode),
	157	FI_INC_LINK);
	158	} else {
	159	if (IS_INODE(page) && is_dent_dnode(page)) {
6ead1142 JK	160	err = recover_inode_page(sbi, page);
6ead1142 JK	161	if (err)
f356fe0c	162	break;
d624c96f JK	163	}
	164
	165	/* add this fsync inode to the list */
	166	entry = kmem_cache_alloc(fsync_entry_slab, GFP_NOFS);
	167	if (!entry) {
	168	err = -ENOMEM;
f356fe0c	169	break;
d624c96f JK	170	}
d624c96f JK	171
d624c96f JK	172	entry->inode = f2fs_iget(sbi->sb, ino_of_node(page));
	173	if (IS_ERR(entry->inode)) {
	174	err = PTR_ERR(entry->inode);
fd8bb65f	175	kmem_cache_free(fsync_entry_slab, entry);
f356fe0c	176	break;
d624c96f	177	}
fd8bb65f	178	list_add_tail(&entry->list, head);
d624c96f	179	}
addbe45b JK	180	entry->blkaddr = blkaddr;
addbe45b JK	181
f356fe0c JK	182	err = recover_inode(entry->inode, page);
	183	if (err && err != -ENOENT)
	184	break;
d624c96f JK	185	next:
	186	/* check next segment */
	187	blkaddr = next_blkaddr_of_node(page);
d624c96f	188	}
d624c96f	189	unlock_page(page);
393ff91f	190	out:
d624c96f JK	191	__free_pages(page, 0);
	192	return err;
	193	}
	194
	195	static void destroy_fsync_dnodes(struct f2fs_sb_info *sbi,
	196	struct list_head *head)
	197	{
d8b79b2f DC	198	struct fsync_inode_entry entry, tmp;
	199
	200	list_for_each_entry_safe(entry, tmp, head, list) {
d624c96f JK	201	iput(entry->inode);
	202	list_del(&entry->list);
	203	kmem_cache_free(fsync_entry_slab, entry);
	204	}
	205	}
	206
39cf72cf	207	static int check_index_in_prev_nodes(struct f2fs_sb_info *sbi,
b292dcab	208	block_t blkaddr, struct dnode_of_data *dn)
d624c96f JK	209	{
	210	struct seg_entry *sentry;
	211	unsigned int segno = GET_SEGNO(sbi, blkaddr);
	212	unsigned short blkoff = GET_SEGOFF_FROM_SEG0(sbi, blkaddr) &
	213	(sbi->blocks_per_seg - 1);
	214	struct f2fs_summary sum;
b292dcab	215	nid_t ino, nid;
d624c96f JK	216	void *kaddr;
	217	struct inode *inode;
	218	struct page *node_page;
	219	block_t bidx;
	220	int i;
	221
	222	sentry = get_seg_entry(sbi, segno);
	223	if (!f2fs_test_bit(blkoff, sentry->cur_valid_map))
39cf72cf	224	return 0;
d624c96f JK	225
	226	/* Get the previous summary */
	227	for (i = CURSEG_WARM_DATA; i <= CURSEG_COLD_DATA; i++) {
	228	struct curseg_info *curseg = CURSEG_I(sbi, i);
	229	if (curseg->segno == segno) {
	230	sum = curseg->sum_blk->entries[blkoff];
	231	break;
	232	}
	233	}
	234	if (i > CURSEG_COLD_DATA) {
	235	struct page *sum_page = get_sum_page(sbi, segno);
	236	struct f2fs_summary_block *sum_node;
	237	kaddr = page_address(sum_page);
	238	sum_node = (struct f2fs_summary_block *)kaddr;
	239	sum = sum_node->entries[blkoff];
	240	f2fs_put_page(sum_page, 1);
	241	}
	242
b292dcab JK	243	/* Use the locked dnode page and inode */
	244	nid = le32_to_cpu(sum.nid);
	245	if (dn->inode->i_ino == nid) {
	246	struct dnode_of_data tdn = *dn;
	247	tdn.nid = nid;
	248	tdn.node_page = dn->inode_page;
060dd67b	249	tdn.ofs_in_node = le16_to_cpu(sum.ofs_in_node);
b292dcab	250	truncate_data_blocks_range(&tdn, 1);
39cf72cf	251	return 0;
b292dcab JK	252	} else if (dn->nid == nid) {
b292dcab JK	253	struct dnode_of_data tdn = *dn;
060dd67b	254	tdn.ofs_in_node = le16_to_cpu(sum.ofs_in_node);
b292dcab	255	truncate_data_blocks_range(&tdn, 1);
39cf72cf	256	return 0;
b292dcab JK	257	}
b292dcab JK	258
d624c96f	259	/* Get the node page */
b292dcab	260	node_page = get_node_page(sbi, nid);
39cf72cf JK	261	if (IS_ERR(node_page))
39cf72cf JK	262	return PTR_ERR(node_page);
d624c96f	263	bidx = start_bidx_of_node(ofs_of_node(node_page)) +
b292dcab	264	le16_to_cpu(sum.ofs_in_node);
d624c96f JK	265	ino = ino_of_node(node_page);
	266	f2fs_put_page(node_page, 1);
	267
	268	/* Deallocate previous index in the node page */
d4686d56	269	inode = f2fs_iget(sbi->sb, ino);
06025f4d	270	if (IS_ERR(inode))
39cf72cf	271	return PTR_ERR(inode);
06025f4d	272
d624c96f JK	273	truncate_hole(inode, bidx, bidx + 1);
d624c96f JK	274	iput(inode);
39cf72cf	275	return 0;
d624c96f JK	276	}
d624c96f JK	277
6ead1142	278	static int do_recover_data(struct f2fs_sb_info sbi, struct inode inode,
d624c96f JK	279	struct page *page, block_t blkaddr)
	280	{
	281	unsigned int start, end;
	282	struct dnode_of_data dn;
	283	struct f2fs_summary sum;
	284	struct node_info ni;
f356fe0c	285	int err = 0, recovered = 0;
39936837	286	int ilock;
d624c96f JK	287
	288	start = start_bidx_of_node(ofs_of_node(page));
	289	if (IS_INODE(page))
	290	end = start + ADDRS_PER_INODE;
	291	else
	292	end = start + ADDRS_PER_BLOCK;
	293
39936837	294	ilock = mutex_lock_op(sbi);
d624c96f	295	set_new_dnode(&dn, inode, NULL, NULL, 0);
39936837	296
6ead1142	297	err = get_dnode_of_data(&dn, start, ALLOC_NODE);
39936837 JK	298	if (err) {
39936837 JK	299	mutex_unlock_op(sbi, ilock);
6ead1142	300	return err;
39936837	301	}
d624c96f JK	302
	303	wait_on_page_writeback(dn.node_page);
	304
	305	get_node_info(sbi, dn.nid, &ni);
	306	BUG_ON(ni.ino != ino_of_node(page));
	307	BUG_ON(ofs_of_node(dn.node_page) != ofs_of_node(page));
	308
	309	for (; start < end; start++) {
	310	block_t src, dest;
	311
	312	src = datablock_addr(dn.node_page, dn.ofs_in_node);
	313	dest = datablock_addr(page, dn.ofs_in_node);
	314
	315	if (src != dest && dest != NEW_ADDR && dest != NULL_ADDR) {
	316	if (src == NULL_ADDR) {
	317	int err = reserve_new_block(&dn);
	318	/* We should not get -ENOSPC */
	319	BUG_ON(err);
	320	}
	321
	322	/* Check the previous node page having this index */
39cf72cf JK	323	err = check_index_in_prev_nodes(sbi, dest, &dn);
	324	if (err)
	325	goto err;
d624c96f JK	326
	327	set_summary(&sum, dn.nid, dn.ofs_in_node, ni.version);
	328
	329	/* write dummy data page */
	330	recover_data_page(sbi, NULL, &sum, src, dest);
	331	update_extent_cache(dest, &dn);
f356fe0c	332	recovered++;
d624c96f JK	333	}
	334	dn.ofs_in_node++;
	335	}
	336
	337	/* write node page in place */
	338	set_summary(&sum, dn.nid, 0, 0);
	339	if (IS_INODE(dn.node_page))
	340	sync_inode_page(&dn);
	341
	342	copy_node_footer(dn.node_page, page);
	343	fill_node_footer(dn.node_page, dn.nid, ni.ino,
	344	ofs_of_node(page), false);
	345	set_page_dirty(dn.node_page);
	346
	347	recover_node_page(sbi, dn.node_page, &sum, &ni, blkaddr);
39cf72cf	348	err:
d624c96f	349	f2fs_put_dnode(&dn);
39936837	350	mutex_unlock_op(sbi, ilock);
f356fe0c JK	351
f356fe0c JK	352	f2fs_msg(sbi->sb, KERN_NOTICE, "recover_data: ino = %lx, "
39cf72cf JK	353	"recovered_data = %d blocks, err = %d",
	354	inode->i_ino, recovered, err);
	355	return err;
d624c96f JK	356	}
d624c96f JK	357
6ead1142	358	static int recover_data(struct f2fs_sb_info *sbi,
d624c96f JK	359	struct list_head *head, int type)
	360	{
	361	unsigned long long cp_ver = le64_to_cpu(sbi->ckpt->checkpoint_ver);
	362	struct curseg_info *curseg;
	363	struct page *page;
6ead1142	364	int err = 0;
d624c96f JK	365	block_t blkaddr;
	366
	367	/* get node pages in the current segment */
	368	curseg = CURSEG_I(sbi, type);
	369	blkaddr = NEXT_FREE_BLKADDR(sbi, curseg);
	370
	371	/* read node page */
	372	page = alloc_page(GFP_NOFS \| __GFP_ZERO);
	373	if (IS_ERR(page))
6ead1142 JK	374	return -ENOMEM;
6ead1142 JK	375
d624c96f JK	376	lock_page(page);
	377
	378	while (1) {
	379	struct fsync_inode_entry *entry;
	380
6ead1142 JK	381	err = f2fs_readpage(sbi, page, blkaddr, READ_SYNC);
6ead1142 JK	382	if (err)
d624c96f JK	383	goto out;
d624c96f JK	384
393ff91f JK	385	lock_page(page);
393ff91f JK	386
d624c96f	387	if (cp_ver != cpver_of_node(page))
45856aff	388	break;
d624c96f JK	389
	390	entry = get_fsync_inode(head, ino_of_node(page));
	391	if (!entry)
	392	goto next;
	393
6ead1142 JK	394	err = do_recover_data(sbi, entry->inode, page, blkaddr);
6ead1142 JK	395	if (err)
45856aff	396	break;
d624c96f JK	397
	398	if (entry->blkaddr == blkaddr) {
	399	iput(entry->inode);
	400	list_del(&entry->list);
	401	kmem_cache_free(fsync_entry_slab, entry);
	402	}
	403	next:
	404	/* check next segment */
	405	blkaddr = next_blkaddr_of_node(page);
d624c96f	406	}
d624c96f	407	unlock_page(page);
393ff91f	408	out:
d624c96f JK	409	__free_pages(page, 0);
d624c96f JK	410
6ead1142 JK	411	if (!err)
	412	allocate_new_segments(sbi);
	413	return err;
d624c96f JK	414	}
d624c96f JK	415
6ead1142	416	int recover_fsync_data(struct f2fs_sb_info *sbi)
d624c96f JK	417	{
d624c96f JK	418	struct list_head inode_list;
6ead1142	419	int err;
d624c96f JK	420
	421	fsync_entry_slab = f2fs_kmem_cache_create("f2fs_fsync_inode_entry",
	422	sizeof(struct fsync_inode_entry), NULL);
	423	if (unlikely(!fsync_entry_slab))
6ead1142	424	return -ENOMEM;
d624c96f JK	425
	426	INIT_LIST_HEAD(&inode_list);
	427
	428	/* step #1: find fsynced inode numbers */
8c26d7d5	429	sbi->por_doing = 1;
6ead1142 JK	430	err = find_fsync_dnodes(sbi, &inode_list);
6ead1142 JK	431	if (err)
d624c96f JK	432	goto out;
	433
	434	if (list_empty(&inode_list))
	435	goto out;
	436
	437	/* step #2: recover data */
6ead1142	438	err = recover_data(sbi, &inode_list, CURSEG_WARM_NODE);
d624c96f JK	439	BUG_ON(!list_empty(&inode_list));
	440	out:
	441	destroy_fsync_dnodes(sbi, &inode_list);
	442	kmem_cache_destroy(fsync_entry_slab);
8c26d7d5	443	sbi->por_doing = 0;
2c2c149f JK	444	if (!err)
2c2c149f JK	445	write_checkpoint(sbi, false);
6ead1142	446	return err;
d624c96f	447	}