[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"

/*
 * Roll forward recovery scenarios.
 *
 * [Term] F: fsync_mark, D: dentry_mark
 *
 * 1. inode(x) | CP | inode(x) | dnode(F)
 * -> Update the latest inode(x).
 *
 * 2. inode(x) | CP | inode(F) | dnode(F)
 * -> No problem.
 *
 * 3. inode(x) | CP | dnode(F) | inode(x)
 * -> Recover to the latest dnode(F), and drop the last inode(x)
 *
 * 4. inode(x) | CP | dnode(F) | inode(F)
 * -> No problem.
 *
 * 5. CP | inode(x) | dnode(F)
 * -> The inode(DF) was missing. Should drop this dnode(F).
 *
 * 6. CP | inode(DF) | dnode(F)
 * -> No problem.
 *
 * 7. CP | dnode(F) | inode(DF)
 * -> If f2fs_iget fails, then goto next to find inode(DF).
 *
 * 8. CP | dnode(F) | inode(x)
 * -> If f2fs_iget fails, then goto next to find inode(DF).
 *    But it will fail due to no inode(DF).
 */

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 fsync_inode_entry *entry;

	list_for_each_entry(entry, head, list)
		if (entry->inode->i_ino == ino)
			return entry;

	return NULL;
}

static int recover_dentry(struct inode *inode, struct page *ipage)
{
	struct f2fs_inode *raw_inode = F2FS_INODE(ipage);
	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 = f2fs_iget(inode->i_sb, pino);
	if (IS_ERR(dir)) {
		err = PTR_ERR(dir);
		goto out;
	}

	if (file_enc_name(inode)) {
		iput(dir);
		return 0;
	}

	name.len = le32_to_cpu(raw_inode->i_namelen);
	name.name = raw_inode->i_name;

	if (unlikely(name.len > F2FS_NAME_LEN)) {
		WARN_ON(1);
		err = -ENAMETOOLONG;
		goto out_err;
	}
retry:
	de = f2fs_find_entry(dir, &name, &page);
	if (de && inode->i_ino == le32_to_cpu(de->ino))
		goto out_unmap_put;

	if (de) {
		einode = f2fs_iget(inode->i_sb, le32_to_cpu(de->ino));
		if (IS_ERR(einode)) {
			WARN_ON(1);
			err = PTR_ERR(einode);
			if (err == -ENOENT)
				err = -EEXIST;
			goto out_unmap_put;
		}
		err = acquire_orphan_inode(F2FS_I_SB(inode));
		if (err) {
			iput(einode);
			goto out_unmap_put;
		}
		f2fs_delete_entry(de, page, dir, einode);
		iput(einode);
		goto retry;
	}
	err = __f2fs_add_link(dir, &name, inode, inode->i_ino, inode->i_mode);
	if (err)
		goto out_err;

	if (is_inode_flag_set(F2FS_I(dir), FI_DELAY_IPUT)) {
		iput(dir);
	} else {
		add_dirty_dir_inode(dir);
		set_inode_flag(F2FS_I(dir), FI_DELAY_IPUT);
	}

	goto out;

out_unmap_put:
	f2fs_dentry_kunmap(dir, page);
	f2fs_put_page(page, 0);
out_err:
	iput(dir);
out:
	f2fs_msg(inode->i_sb, KERN_NOTICE,
			"%s: ino = %x, name = %s, dir = %lx, err = %d",
			__func__, ino_of_node(ipage), raw_inode->i_name,
			IS_ERR(dir) ? 0 : dir->i_ino, err);
	return err;
}

static void recover_inode(struct inode *inode, struct page *page)
{
	struct f2fs_inode *raw = F2FS_INODE(page);
	char *name;

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

	if (file_enc_name(inode))
		name = "<encrypted>";
	else
		name = F2FS_INODE(page)->i_name;

	f2fs_msg(inode->i_sb, KERN_NOTICE, "recover_inode: ino = %x, name = %s",
			ino_of_node(page), name);
}

static int find_fsync_dnodes(struct f2fs_sb_info *sbi, struct list_head *head)
{
	unsigned long long cp_ver = cur_cp_version(F2FS_CKPT(sbi));
	struct curseg_info *curseg;
	struct page *page = NULL;
	block_t blkaddr;
	int err = 0;

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

	ra_meta_pages(sbi, blkaddr, 1, META_POR);

	while (1) {
		struct fsync_inode_entry *entry;

		if (!is_valid_blkaddr(sbi, blkaddr, META_POR))
			return 0;

		page = get_meta_page(sbi, blkaddr);

		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)) {
				err = recover_inode_page(sbi, page);
				if (err)
					break;
			}

			/* add this fsync inode to the list */
			entry = kmem_cache_alloc(fsync_entry_slab, GFP_F2FS_ZERO);
			if (!entry) {
				err = -ENOMEM;
				break;
			}
			/*
			 * CP | dnode(F) | inode(DF)
			 * For this case, we should not give up now.
			 */
			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);
				if (err == -ENOENT) {
					err = 0;
					goto next;
				}
				break;
			}
			list_add_tail(&entry->list, head);
		}
		entry->blkaddr = blkaddr;

		if (IS_INODE(page)) {
			entry->last_inode = blkaddr;
			if (is_dent_dnode(page))
				entry->last_dentry = blkaddr;
		}
next:
		/* check next segment */
		blkaddr = next_blkaddr_of_node(page);
		f2fs_put_page(page, 1);

		ra_meta_pages_cond(sbi, blkaddr);
	}
	f2fs_put_page(page, 1);
	return err;
}

static void destroy_fsync_dnodes(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_BLKOFF_FROM_SEG0(sbi, blkaddr);
	struct f2fs_summary_block *sum_node;
	struct f2fs_summary sum;
	struct page *sum_page, *node_page;
	struct dnode_of_data tdn = *dn;
	nid_t ino, nid;
	struct inode *inode;
	unsigned int offset;
	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];
			goto got_it;
		}
	}

	sum_page = get_sum_page(sbi, segno);
	sum_node = (struct f2fs_summary_block *)page_address(sum_page);
	sum = sum_node->entries[blkoff];
	f2fs_put_page(sum_page, 1);
got_it:
	/* Use the locked dnode page and inode */
	nid = le32_to_cpu(sum.nid);
	if (dn->inode->i_ino == nid) {
		tdn.nid = nid;
		if (!dn->inode_page_locked)
			lock_page(dn->inode_page);
		tdn.node_page = dn->inode_page;
		tdn.ofs_in_node = le16_to_cpu(sum.ofs_in_node);
		goto truncate_out;
	} else if (dn->nid == nid) {
		tdn.ofs_in_node = le16_to_cpu(sum.ofs_in_node);
		goto truncate_out;
	}

	/* Get the node page */
	node_page = get_node_page(sbi, nid);
	if (IS_ERR(node_page))
		return PTR_ERR(node_page);

	offset = ofs_of_node(node_page);
	ino = ino_of_node(node_page);
	f2fs_put_page(node_page, 1);

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

	bidx = start_bidx_of_node(offset, F2FS_I(inode)) +
			le16_to_cpu(sum.ofs_in_node);

	/*
	 * if inode page is locked, unlock temporarily, but its reference
	 * count keeps alive.
	 */
	if (ino == dn->inode->i_ino && dn->inode_page_locked)
		unlock_page(dn->inode_page);

	set_new_dnode(&tdn, inode, NULL, NULL, 0);
	if (get_dnode_of_data(&tdn, bidx, LOOKUP_NODE))
		goto out;

	if (tdn.data_blkaddr == blkaddr)
		truncate_data_blocks_range(&tdn, 1);

	f2fs_put_dnode(&tdn);
out:
	if (ino != dn->inode->i_ino)
		iput(inode);
	else if (dn->inode_page_locked)
		lock_page(dn->inode_page);
	return 0;

truncate_out:
	if (datablock_addr(tdn.node_page, tdn.ofs_in_node) == blkaddr)
		truncate_data_blocks_range(&tdn, 1);
	if (dn->inode->i_ino == nid && !dn->inode_page_locked)
		unlock_page(dn->inode_page);
	return 0;
}

static int do_recover_data(struct f2fs_sb_info *sbi, struct inode *inode,
					struct page *page, block_t blkaddr)
{
	struct f2fs_inode_info *fi = F2FS_I(inode);
	unsigned int start, end;
	struct dnode_of_data dn;
	struct node_info ni;
	int err = 0, recovered = 0;

	/* step 1: recover xattr */
	if (IS_INODE(page)) {
		recover_inline_xattr(inode, page);
	} else if (f2fs_has_xattr_block(ofs_of_node(page))) {
		/*
		 * Deprecated; xattr blocks should be found from cold log.
		 * But, we should remain this for backward compatibility.
		 */
		recover_xattr_data(inode, page, blkaddr);
		goto out;
	}

	/* step 2: recover inline data */
	if (recover_inline_data(inode, page))
		goto out;

	/* step 3: recover data indices */
	start = start_bidx_of_node(ofs_of_node(page), fi);
	end = start + ADDRS_PER_PAGE(page, fi);

	f2fs_lock_op(sbi);

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

	err = get_dnode_of_data(&dn, start, ALLOC_NODE);
	if (err) {
		f2fs_unlock_op(sbi);
		goto out;
	}

	f2fs_wait_on_page_writeback(dn.node_page, NODE);

	get_node_info(sbi, dn.nid, &ni);
	f2fs_bug_on(sbi, ni.ino != ino_of_node(page));
	f2fs_bug_on(sbi, ofs_of_node(dn.node_page) != ofs_of_node(page));

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

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

		/* skip recovering if dest is the same as src */
		if (src == dest)
			continue;

		/* dest is invalid, just invalidate src block */
		if (dest == NULL_ADDR) {
			truncate_data_blocks_range(&dn, 1);
			continue;
		}

		/*
		 * dest is reserved block, invalidate src block
		 * and then reserve one new block in dnode page.
		 */
		if (dest == NEW_ADDR) {
			truncate_data_blocks_range(&dn, 1);
			err = reserve_new_block(&dn);
			f2fs_bug_on(sbi, err);
			continue;
		}

		/* dest is valid block, try to recover from src to dest */
		if (is_valid_blkaddr(sbi, dest, META_POR)) {

			if (src == NULL_ADDR) {
				err = reserve_new_block(&dn);
				/* We should not get -ENOSPC */
				f2fs_bug_on(sbi, err);
			}

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

			/* write dummy data page */
			f2fs_replace_block(sbi, &dn, src, dest,
							ni.version, false);
			recovered++;
		}
	}

	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);
err:
	f2fs_put_dnode(&dn);
	f2fs_unlock_op(sbi);
out:
	f2fs_msg(sbi->sb, KERN_NOTICE,
		"recover_data: ino = %lx, recovered = %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 = cur_cp_version(F2FS_CKPT(sbi));
	struct curseg_info *curseg;
	struct page *page = NULL;
	int err = 0;
	block_t blkaddr;

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

	while (1) {
		struct fsync_inode_entry *entry;

		if (!is_valid_blkaddr(sbi, blkaddr, META_POR))
			break;

		ra_meta_pages_cond(sbi, blkaddr);

		page = get_meta_page(sbi, blkaddr);

		if (cp_ver != cpver_of_node(page)) {
			f2fs_put_page(page, 1);
			break;
		}

		entry = get_fsync_inode(head, ino_of_node(page));
		if (!entry)
			goto next;
		/*
		 * inode(x) | CP | inode(x) | dnode(F)
		 * In this case, we can lose the latest inode(x).
		 * So, call recover_inode for the inode update.
		 */
		if (entry->last_inode == blkaddr)
			recover_inode(entry->inode, page);
		if (entry->last_dentry == blkaddr) {
			err = recover_dentry(entry->inode, page);
			if (err) {
				f2fs_put_page(page, 1);
				break;
			}
		}
		err = do_recover_data(sbi, entry->inode, page, blkaddr);
		if (err) {
			f2fs_put_page(page, 1);
			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);
		f2fs_put_page(page, 1);
	}
	if (!err)
		allocate_new_segments(sbi);
	return err;
}

int recover_fsync_data(struct f2fs_sb_info *sbi)
{
	struct curseg_info *curseg = CURSEG_I(sbi, CURSEG_WARM_NODE);
	struct list_head inode_list;
	block_t blkaddr;
	int err;
	bool need_writecp = false;

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

	INIT_LIST_HEAD(&inode_list);

	/* prevent checkpoint */
	mutex_lock(&sbi->cp_mutex);

	blkaddr = NEXT_FREE_BLKADDR(sbi, curseg);

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

	if (list_empty(&inode_list))
		goto out;

	need_writecp = true;

	/* step #2: recover data */
	err = recover_data(sbi, &inode_list, CURSEG_WARM_NODE);
	if (!err)
		f2fs_bug_on(sbi, !list_empty(&inode_list));
out:
	destroy_fsync_dnodes(&inode_list);
	kmem_cache_destroy(fsync_entry_slab);

	/* truncate meta pages to be used by the recovery */
	truncate_inode_pages_range(META_MAPPING(sbi),
			MAIN_BLKADDR(sbi) << PAGE_CACHE_SHIFT, -1);

	if (err) {
		truncate_inode_pages_final(NODE_MAPPING(sbi));
		truncate_inode_pages_final(META_MAPPING(sbi));
	}

	clear_sbi_flag(sbi, SBI_POR_DOING);
	if (err) {
		bool invalidate = false;

		if (discard_next_dnode(sbi, blkaddr))
			invalidate = true;

		/* Flush all the NAT/SIT pages */
		while (get_pages(sbi, F2FS_DIRTY_META))
			sync_meta_pages(sbi, META, LONG_MAX);

		/* invalidate temporary meta page */
		if (invalidate)
			invalidate_mapping_pages(META_MAPPING(sbi),
							blkaddr, blkaddr);

		set_ckpt_flags(sbi->ckpt, CP_ERROR_FLAG);
		mutex_unlock(&sbi->cp_mutex);
	} else if (need_writecp) {
		struct cp_control cpc = {
			.reason = CP_RECOVERY,
		};
		mutex_unlock(&sbi->cp_mutex);
		write_checkpoint(sbi, &cpc);
	} else {
		mutex_unlock(&sbi->cp_mutex);
	}
	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
441ac5cb JK	17	/*
	18	* Roll forward recovery scenarios.
	19	*
	20	* [Term] F: fsync_mark, D: dentry_mark
	21	*
	22	* 1. inode(x) \| CP \| inode(x) \| dnode(F)
	23	* -> Update the latest inode(x).
	24	*
	25	* 2. inode(x) \| CP \| inode(F) \| dnode(F)
	26	* -> No problem.
	27	*
	28	* 3. inode(x) \| CP \| dnode(F) \| inode(x)
	29	* -> Recover to the latest dnode(F), and drop the last inode(x)
	30	*
	31	* 4. inode(x) \| CP \| dnode(F) \| inode(F)
	32	* -> No problem.
	33	*
	34	* 5. CP \| inode(x) \| dnode(F)
	35	* -> The inode(DF) was missing. Should drop this dnode(F).
	36	*
	37	* 6. CP \| inode(DF) \| dnode(F)
	38	* -> No problem.
	39	*
	40	* 7. CP \| dnode(F) \| inode(DF)
	41	* -> If f2fs_iget fails, then goto next to find inode(DF).
	42	*
	43	* 8. CP \| dnode(F) \| inode(x)
	44	* -> If f2fs_iget fails, then goto next to find inode(DF).
	45	* But it will fail due to no inode(DF).
	46	*/
	47
d624c96f JK	48	static struct kmem_cache *fsync_entry_slab;
	49
	50	bool space_for_roll_forward(struct f2fs_sb_info *sbi)
	51	{
	52	if (sbi->last_valid_block_count + sbi->alloc_valid_block_count
	53	> sbi->user_block_count)
	54	return false;
	55	return true;
	56	}
	57
	58	static struct fsync_inode_entry get_fsync_inode(struct list_head head,
	59	nid_t ino)
	60	{
d624c96f JK	61	struct fsync_inode_entry *entry;
d624c96f JK	62
2d7b822a	63	list_for_each_entry(entry, head, list)
d624c96f JK	64	if (entry->inode->i_ino == ino)
d624c96f JK	65	return entry;
2d7b822a	66
d624c96f JK	67	return NULL;
	68	}
	69
c52e1b10	70	static int recover_dentry(struct inode inode, struct page ipage)
d624c96f	71	{
58bfaf44	72	struct f2fs_inode *raw_inode = F2FS_INODE(ipage);
74d0b917	73	nid_t pino = le32_to_cpu(raw_inode->i_pino);
6b8213d9	74	struct f2fs_dir_entry *de;
b7f7a5e0	75	struct qstr name;
d624c96f	76	struct page *page;
6b8213d9	77	struct inode dir, einode;
d624c96f JK	78	int err = 0;
d624c96f JK	79
ed57c27f JK	80	dir = f2fs_iget(inode->i_sb, pino);
	81	if (IS_ERR(dir)) {
	82	err = PTR_ERR(dir);
	83	goto out;
	84	}
	85
e7d55452 JK	86	if (file_enc_name(inode)) {
	87	iput(dir);
	88	return 0;
	89	}
	90
b7f7a5e0 AV	91	name.len = le32_to_cpu(raw_inode->i_namelen);
b7f7a5e0 AV	92	name.name = raw_inode->i_name;
d96b1431 CY	93
	94	if (unlikely(name.len > F2FS_NAME_LEN)) {
	95	WARN_ON(1);
	96	err = -ENAMETOOLONG;
86928f98	97	goto out_err;
d96b1431	98	}
6b8213d9 JK	99	retry:
6b8213d9 JK	100	de = f2fs_find_entry(dir, &name, &page);
418f6c27	101	if (de && inode->i_ino == le32_to_cpu(de->ino))
2e5558f4	102	goto out_unmap_put;
418f6c27	103
6b8213d9 JK	104	if (de) {
	105	einode = f2fs_iget(inode->i_sb, le32_to_cpu(de->ino));
	106	if (IS_ERR(einode)) {
	107	WARN_ON(1);
5c1f9927 CY	108	err = PTR_ERR(einode);
5c1f9927 CY	109	if (err == -ENOENT)
6b8213d9	110	err = -EEXIST;
2e5558f4 RK	111	goto out_unmap_put;
2e5558f4 RK	112	}
4081363f	113	err = acquire_orphan_inode(F2FS_I_SB(inode));
2e5558f4 RK	114	if (err) {
	115	iput(einode);
	116	goto out_unmap_put;
6b8213d9	117	}
dbeacf02	118	f2fs_delete_entry(de, page, dir, einode);
6b8213d9 JK	119	iput(einode);
6b8213d9 JK	120	goto retry;
d624c96f	121	}
510022a8	122	err = __f2fs_add_link(dir, &name, inode, inode->i_ino, inode->i_mode);
86928f98 JK	123	if (err)
	124	goto out_err;
	125
	126	if (is_inode_flag_set(F2FS_I(dir), FI_DELAY_IPUT)) {
	127	iput(dir);
	128	} else {
	129	add_dirty_dir_inode(dir);
	130	set_inode_flag(F2FS_I(dir), FI_DELAY_IPUT);
	131	}
	132
2e5558f4 RK	133	goto out;
	134
	135	out_unmap_put:
9486ba44	136	f2fs_dentry_kunmap(dir, page);
2e5558f4	137	f2fs_put_page(page, 0);
86928f98 JK	138	out_err:
86928f98 JK	139	iput(dir);
d624c96f	140	out:
6c311ec6 CF	141	f2fs_msg(inode->i_sb, KERN_NOTICE,
	142	"%s: ino = %x, name = %s, dir = %lx, err = %d",
	143	__func__, ino_of_node(ipage), raw_inode->i_name,
f28c06fa	144	IS_ERR(dir) ? 0 : dir->i_ino, err);
d624c96f JK	145	return err;
	146	}
	147
c52e1b10	148	static void recover_inode(struct inode inode, struct page page)
d624c96f	149	{
441ac5cb	150	struct f2fs_inode *raw = F2FS_INODE(page);
e7d55452	151	char *name;
441ac5cb JK	152
	153	inode->i_mode = le16_to_cpu(raw->i_mode);
	154	i_size_write(inode, le64_to_cpu(raw->i_size));
	155	inode->i_atime.tv_sec = le64_to_cpu(raw->i_mtime);
	156	inode->i_ctime.tv_sec = le64_to_cpu(raw->i_ctime);
	157	inode->i_mtime.tv_sec = le64_to_cpu(raw->i_mtime);
	158	inode->i_atime.tv_nsec = le32_to_cpu(raw->i_mtime_nsec);
	159	inode->i_ctime.tv_nsec = le32_to_cpu(raw->i_ctime_nsec);
	160	inode->i_mtime.tv_nsec = le32_to_cpu(raw->i_mtime_nsec);
f356fe0c	161
e7d55452 JK	162	if (file_enc_name(inode))
	163	name = "<encrypted>";
	164	else
	165	name = F2FS_INODE(page)->i_name;
	166
f356fe0c	167	f2fs_msg(inode->i_sb, KERN_NOTICE, "recover_inode: ino = %x, name = %s",
e7d55452	168	ino_of_node(page), name);
d624c96f JK	169	}
	170
	171	static int find_fsync_dnodes(struct f2fs_sb_info sbi, struct list_head head)
	172	{
d71b5564	173	unsigned long long cp_ver = cur_cp_version(F2FS_CKPT(sbi));
d624c96f	174	struct curseg_info *curseg;
4c521f49	175	struct page *page = NULL;
d624c96f JK	176	block_t blkaddr;
	177	int err = 0;
	178
	179	/* get node pages in the current segment */
	180	curseg = CURSEG_I(sbi, CURSEG_WARM_NODE);
695fd1ed	181	blkaddr = NEXT_FREE_BLKADDR(sbi, curseg);
d624c96f	182
635aee1f CY	183	ra_meta_pages(sbi, blkaddr, 1, META_POR);
635aee1f CY	184
d624c96f JK	185	while (1) {
	186	struct fsync_inode_entry *entry;
	187
f0c9cada	188	if (!is_valid_blkaddr(sbi, blkaddr, META_POR))
4c521f49	189	return 0;
d624c96f	190
635aee1f	191	page = get_meta_page(sbi, blkaddr);
393ff91f	192
6ead1142	193	if (cp_ver != cpver_of_node(page))
f356fe0c	194	break;
d624c96f JK	195
	196	if (!is_fsync_dnode(page))
	197	goto next;
	198
	199	entry = get_fsync_inode(head, ino_of_node(page));
418f6c27	200	if (!entry) {
d624c96f	201	if (IS_INODE(page) && is_dent_dnode(page)) {
6ead1142 JK	202	err = recover_inode_page(sbi, page);
6ead1142 JK	203	if (err)
f356fe0c	204	break;
d624c96f JK	205	}
	206
	207	/* add this fsync inode to the list */
c52e1b10	208	entry = kmem_cache_alloc(fsync_entry_slab, GFP_F2FS_ZERO);
d624c96f JK	209	if (!entry) {
d624c96f JK	210	err = -ENOMEM;
f356fe0c	211	break;
d624c96f	212	}
441ac5cb JK	213	/*
	214	* CP \| dnode(F) \| inode(DF)
	215	* For this case, we should not give up now.
	216	*/
d624c96f JK	217	entry->inode = f2fs_iget(sbi->sb, ino_of_node(page));
	218	if (IS_ERR(entry->inode)) {
	219	err = PTR_ERR(entry->inode);
fd8bb65f	220	kmem_cache_free(fsync_entry_slab, entry);
8fbc418f JK	221	if (err == -ENOENT) {
8fbc418f JK	222	err = 0;
441ac5cb	223	goto next;
8fbc418f	224	}
f356fe0c	225	break;
d624c96f	226	}
fd8bb65f	227	list_add_tail(&entry->list, head);
d624c96f	228	}
addbe45b JK	229	entry->blkaddr = blkaddr;
addbe45b JK	230
c52e1b10 JK	231	if (IS_INODE(page)) {
	232	entry->last_inode = blkaddr;
	233	if (is_dent_dnode(page))
	234	entry->last_dentry = blkaddr;
	235	}
d624c96f JK	236	next:
	237	/* check next segment */
	238	blkaddr = next_blkaddr_of_node(page);
4c521f49	239	f2fs_put_page(page, 1);
635aee1f CY	240
635aee1f CY	241	ra_meta_pages_cond(sbi, blkaddr);
d624c96f	242	}
4c521f49	243	f2fs_put_page(page, 1);
d624c96f JK	244	return err;
	245	}
	246
5ebefc5b	247	static void destroy_fsync_dnodes(struct list_head *head)
d624c96f	248	{
d8b79b2f DC	249	struct fsync_inode_entry entry, tmp;
	250
	251	list_for_each_entry_safe(entry, tmp, head, list) {
d624c96f JK	252	iput(entry->inode);
	253	list_del(&entry->list);
	254	kmem_cache_free(fsync_entry_slab, entry);
	255	}
	256	}
	257
39cf72cf	258	static int check_index_in_prev_nodes(struct f2fs_sb_info *sbi,
b292dcab	259	block_t blkaddr, struct dnode_of_data *dn)
d624c96f JK	260	{
	261	struct seg_entry *sentry;
	262	unsigned int segno = GET_SEGNO(sbi, blkaddr);
491c0854	263	unsigned short blkoff = GET_BLKOFF_FROM_SEG0(sbi, blkaddr);
f6517cfc	264	struct f2fs_summary_block *sum_node;
d624c96f	265	struct f2fs_summary sum;
f6517cfc	266	struct page sum_page, node_page;
c9ef4810	267	struct dnode_of_data tdn = *dn;
b292dcab	268	nid_t ino, nid;
d624c96f	269	struct inode *inode;
de93653f	270	unsigned int offset;
d624c96f JK	271	block_t bidx;
	272	int i;
	273
	274	sentry = get_seg_entry(sbi, segno);
	275	if (!f2fs_test_bit(blkoff, sentry->cur_valid_map))
39cf72cf	276	return 0;
d624c96f JK	277
	278	/* Get the previous summary */
	279	for (i = CURSEG_WARM_DATA; i <= CURSEG_COLD_DATA; i++) {
	280	struct curseg_info *curseg = CURSEG_I(sbi, i);
	281	if (curseg->segno == segno) {
	282	sum = curseg->sum_blk->entries[blkoff];
f6517cfc	283	goto got_it;
d624c96f JK	284	}
d624c96f JK	285	}
d624c96f	286
f6517cfc JK	287	sum_page = get_sum_page(sbi, segno);
	288	sum_node = (struct f2fs_summary_block *)page_address(sum_page);
	289	sum = sum_node->entries[blkoff];
	290	f2fs_put_page(sum_page, 1);
	291	got_it:
b292dcab JK	292	/* Use the locked dnode page and inode */
	293	nid = le32_to_cpu(sum.nid);
	294	if (dn->inode->i_ino == nid) {
b292dcab	295	tdn.nid = nid;
c9ef4810 JK	296	if (!dn->inode_page_locked)
c9ef4810 JK	297	lock_page(dn->inode_page);
b292dcab	298	tdn.node_page = dn->inode_page;
060dd67b	299	tdn.ofs_in_node = le16_to_cpu(sum.ofs_in_node);
c9ef4810	300	goto truncate_out;
b292dcab	301	} else if (dn->nid == nid) {
060dd67b	302	tdn.ofs_in_node = le16_to_cpu(sum.ofs_in_node);
c9ef4810	303	goto truncate_out;
b292dcab JK	304	}
b292dcab JK	305
d624c96f	306	/* Get the node page */
b292dcab	307	node_page = get_node_page(sbi, nid);
39cf72cf JK	308	if (IS_ERR(node_page))
39cf72cf JK	309	return PTR_ERR(node_page);
de93653f JK	310
de93653f JK	311	offset = ofs_of_node(node_page);
d624c96f JK	312	ino = ino_of_node(node_page);
	313	f2fs_put_page(node_page, 1);
	314
60979115 JK	315	if (ino != dn->inode->i_ino) {
	316	/* Deallocate previous index in the node page */
	317	inode = f2fs_iget(sbi->sb, ino);
	318	if (IS_ERR(inode))
	319	return PTR_ERR(inode);
	320	} else {
	321	inode = dn->inode;
	322	}
06025f4d	323
de93653f	324	bidx = start_bidx_of_node(offset, F2FS_I(inode)) +
60979115	325	le16_to_cpu(sum.ofs_in_node);
de93653f	326
c9ef4810 JK	327	/*
	328	* if inode page is locked, unlock temporarily, but its reference
	329	* count keeps alive.
	330	*/
	331	if (ino == dn->inode->i_ino && dn->inode_page_locked)
	332	unlock_page(dn->inode_page);
	333
	334	set_new_dnode(&tdn, inode, NULL, NULL, 0);
	335	if (get_dnode_of_data(&tdn, bidx, LOOKUP_NODE))
	336	goto out;
	337
	338	if (tdn.data_blkaddr == blkaddr)
	339	truncate_data_blocks_range(&tdn, 1);
	340
	341	f2fs_put_dnode(&tdn);
	342	out:
	343	if (ino != dn->inode->i_ino)
60979115	344	iput(inode);
c9ef4810 JK	345	else if (dn->inode_page_locked)
	346	lock_page(dn->inode_page);
	347	return 0;
	348
	349	truncate_out:
	350	if (datablock_addr(tdn.node_page, tdn.ofs_in_node) == blkaddr)
	351	truncate_data_blocks_range(&tdn, 1);
	352	if (dn->inode->i_ino == nid && !dn->inode_page_locked)
	353	unlock_page(dn->inode_page);
39cf72cf	354	return 0;
d624c96f JK	355	}
d624c96f JK	356
6ead1142	357	static int do_recover_data(struct f2fs_sb_info sbi, struct inode inode,
d624c96f JK	358	struct page *page, block_t blkaddr)
d624c96f JK	359	{
de93653f	360	struct f2fs_inode_info *fi = F2FS_I(inode);
d624c96f JK	361	unsigned int start, end;
d624c96f JK	362	struct dnode_of_data dn;
d624c96f	363	struct node_info ni;
f356fe0c	364	int err = 0, recovered = 0;
d624c96f	365
1c35a90e JK	366	/* step 1: recover xattr */
	367	if (IS_INODE(page)) {
	368	recover_inline_xattr(inode, page);
	369	} else if (f2fs_has_xattr_block(ofs_of_node(page))) {
bc4a1f87 JK	370	/*
	371	* Deprecated; xattr blocks should be found from cold log.
	372	* But, we should remain this for backward compatibility.
	373	*/
1c35a90e	374	recover_xattr_data(inode, page, blkaddr);
1e1bb4ba	375	goto out;
1c35a90e	376	}
1e1bb4ba	377
1c35a90e JK	378	/* step 2: recover inline data */
1c35a90e JK	379	if (recover_inline_data(inode, page))
abb2366c JK	380	goto out;
abb2366c JK	381
1c35a90e	382	/* step 3: recover data indices */
de93653f	383	start = start_bidx_of_node(ofs_of_node(page), fi);
6403eb1f	384	end = start + ADDRS_PER_PAGE(page, fi);
d624c96f	385
e479556b	386	f2fs_lock_op(sbi);
1e1bb4ba	387
d624c96f	388	set_new_dnode(&dn, inode, NULL, NULL, 0);
39936837	389
6ead1142	390	err = get_dnode_of_data(&dn, start, ALLOC_NODE);
39936837	391	if (err) {
e479556b	392	f2fs_unlock_op(sbi);
1e1bb4ba	393	goto out;
39936837	394	}
d624c96f	395
3cb5ad15	396	f2fs_wait_on_page_writeback(dn.node_page, NODE);
d624c96f JK	397
d624c96f JK	398	get_node_info(sbi, dn.nid, &ni);
9850cf4a JK	399	f2fs_bug_on(sbi, ni.ino != ino_of_node(page));
9850cf4a JK	400	f2fs_bug_on(sbi, ofs_of_node(dn.node_page) != ofs_of_node(page));
d624c96f	401
12a8343e	402	for (; start < end; start++, dn.ofs_in_node++) {
d624c96f JK	403	block_t src, dest;
	404
	405	src = datablock_addr(dn.node_page, dn.ofs_in_node);
	406	dest = datablock_addr(page, dn.ofs_in_node);
	407
12a8343e CY	408	/* skip recovering if dest is the same as src */
	409	if (src == dest)
	410	continue;
	411
	412	/* dest is invalid, just invalidate src block */
	413	if (dest == NULL_ADDR) {
	414	truncate_data_blocks_range(&dn, 1);
	415	continue;
	416	}
	417
	418	/*
	419	* dest is reserved block, invalidate src block
	420	* and then reserve one new block in dnode page.
	421	*/
	422	if (dest == NEW_ADDR) {
	423	truncate_data_blocks_range(&dn, 1);
	424	err = reserve_new_block(&dn);
	425	f2fs_bug_on(sbi, err);
	426	continue;
	427	}
	428
	429	/* dest is valid block, try to recover from src to dest */
	430	if (is_valid_blkaddr(sbi, dest, META_POR)) {
e03b07d9	431
d624c96f	432	if (src == NULL_ADDR) {
5d56b671	433	err = reserve_new_block(&dn);
d624c96f	434	/* We should not get -ENOSPC */
9850cf4a	435	f2fs_bug_on(sbi, err);
d624c96f JK	436	}
	437
	438	/* Check the previous node page having this index */
39cf72cf JK	439	err = check_index_in_prev_nodes(sbi, dest, &dn);
	440	if (err)
	441	goto err;
d624c96f	442
d624c96f	443	/* write dummy data page */
528e3459 CY	444	f2fs_replace_block(sbi, &dn, src, dest,
528e3459 CY	445	ni.version, false);
f356fe0c	446	recovered++;
d624c96f	447	}
d624c96f JK	448	}
d624c96f JK	449
d624c96f JK	450	if (IS_INODE(dn.node_page))
	451	sync_inode_page(&dn);
	452
	453	copy_node_footer(dn.node_page, page);
	454	fill_node_footer(dn.node_page, dn.nid, ni.ino,
	455	ofs_of_node(page), false);
	456	set_page_dirty(dn.node_page);
39cf72cf	457	err:
d624c96f	458	f2fs_put_dnode(&dn);
e479556b	459	f2fs_unlock_op(sbi);
1e1bb4ba	460	out:
6c311ec6 CF	461	f2fs_msg(sbi->sb, KERN_NOTICE,
	462	"recover_data: ino = %lx, recovered = %d blocks, err = %d",
	463	inode->i_ino, recovered, err);
39cf72cf	464	return err;
d624c96f JK	465	}
d624c96f JK	466
6ead1142	467	static int recover_data(struct f2fs_sb_info *sbi,
d624c96f JK	468	struct list_head *head, int type)
d624c96f JK	469	{
d71b5564	470	unsigned long long cp_ver = cur_cp_version(F2FS_CKPT(sbi));
d624c96f	471	struct curseg_info *curseg;
4c521f49	472	struct page *page = NULL;
6ead1142	473	int err = 0;
d624c96f JK	474	block_t blkaddr;
	475
	476	/* get node pages in the current segment */
	477	curseg = CURSEG_I(sbi, type);
	478	blkaddr = NEXT_FREE_BLKADDR(sbi, curseg);
	479
d624c96f JK	480	while (1) {
	481	struct fsync_inode_entry *entry;
	482
f0c9cada	483	if (!is_valid_blkaddr(sbi, blkaddr, META_POR))
4c521f49	484	break;
d624c96f	485
635aee1f CY	486	ra_meta_pages_cond(sbi, blkaddr);
	487
	488	page = get_meta_page(sbi, blkaddr);
393ff91f	489
4c521f49 JK	490	if (cp_ver != cpver_of_node(page)) {
4c521f49 JK	491	f2fs_put_page(page, 1);
45856aff	492	break;
4c521f49	493	}
d624c96f JK	494
	495	entry = get_fsync_inode(head, ino_of_node(page));
	496	if (!entry)
	497	goto next;
441ac5cb JK	498	/*
	499	* inode(x) \| CP \| inode(x) \| dnode(F)
	500	* In this case, we can lose the latest inode(x).
c52e1b10	501	* So, call recover_inode for the inode update.
441ac5cb	502	*/
c52e1b10 JK	503	if (entry->last_inode == blkaddr)
	504	recover_inode(entry->inode, page);
	505	if (entry->last_dentry == blkaddr) {
	506	err = recover_dentry(entry->inode, page);
	507	if (err) {
	508	f2fs_put_page(page, 1);
	509	break;
	510	}
	511	}
6ead1142	512	err = do_recover_data(sbi, entry->inode, page, blkaddr);
4c521f49 JK	513	if (err) {
4c521f49 JK	514	f2fs_put_page(page, 1);
45856aff	515	break;
4c521f49	516	}
d624c96f JK	517
	518	if (entry->blkaddr == blkaddr) {
	519	iput(entry->inode);
	520	list_del(&entry->list);
	521	kmem_cache_free(fsync_entry_slab, entry);
	522	}
	523	next:
	524	/* check next segment */
	525	blkaddr = next_blkaddr_of_node(page);
4c521f49	526	f2fs_put_page(page, 1);
d624c96f	527	}
6ead1142 JK	528	if (!err)
	529	allocate_new_segments(sbi);
	530	return err;
d624c96f JK	531	}
d624c96f JK	532
6ead1142	533	int recover_fsync_data(struct f2fs_sb_info *sbi)
d624c96f	534	{
cf2271e7	535	struct curseg_info *curseg = CURSEG_I(sbi, CURSEG_WARM_NODE);
d624c96f	536	struct list_head inode_list;
cf2271e7	537	block_t blkaddr;
6ead1142	538	int err;
aabe5136	539	bool need_writecp = false;
d624c96f JK	540
d624c96f JK	541	fsync_entry_slab = f2fs_kmem_cache_create("f2fs_fsync_inode_entry",
e8512d2e	542	sizeof(struct fsync_inode_entry));
6bacf52f	543	if (!fsync_entry_slab)
6ead1142	544	return -ENOMEM;
d624c96f JK	545
	546	INIT_LIST_HEAD(&inode_list);
	547
14f4e690 JK	548	/* prevent checkpoint */
	549	mutex_lock(&sbi->cp_mutex);
	550
cf2271e7 JK	551	blkaddr = NEXT_FREE_BLKADDR(sbi, curseg);
cf2271e7 JK	552
315df839	553	/* step #1: find fsynced inode numbers */
6ead1142 JK	554	err = find_fsync_dnodes(sbi, &inode_list);
6ead1142 JK	555	if (err)
d624c96f JK	556	goto out;
	557
	558	if (list_empty(&inode_list))
	559	goto out;
	560
aabe5136	561	need_writecp = true;
691c6fd2	562
d624c96f	563	/* step #2: recover data */
6ead1142	564	err = recover_data(sbi, &inode_list, CURSEG_WARM_NODE);
b307384e	565	if (!err)
9850cf4a	566	f2fs_bug_on(sbi, !list_empty(&inode_list));
d624c96f	567	out:
5ebefc5b	568	destroy_fsync_dnodes(&inode_list);
d624c96f	569	kmem_cache_destroy(fsync_entry_slab);
cf2271e7	570
4c521f49 JK	571	/* truncate meta pages to be used by the recovery */
4c521f49 JK	572	truncate_inode_pages_range(META_MAPPING(sbi),
7cd8558b	573	MAIN_BLKADDR(sbi) << PAGE_CACHE_SHIFT, -1);
4c521f49	574
cf2271e7 JK	575	if (err) {
	576	truncate_inode_pages_final(NODE_MAPPING(sbi));
	577	truncate_inode_pages_final(META_MAPPING(sbi));
	578	}
	579
caf0047e	580	clear_sbi_flag(sbi, SBI_POR_DOING);
cf2271e7	581	if (err) {
e90c2d28 CY	582	bool invalidate = false;
	583
	584	if (discard_next_dnode(sbi, blkaddr))
	585	invalidate = true;
cf2271e7 JK	586
	587	/* Flush all the NAT/SIT pages */
	588	while (get_pages(sbi, F2FS_DIRTY_META))
	589	sync_meta_pages(sbi, META, LONG_MAX);
e90c2d28 CY	590
	591	/* invalidate temporary meta page */
	592	if (invalidate)
	593	invalidate_mapping_pages(META_MAPPING(sbi),
	594	blkaddr, blkaddr);
	595
14f4e690 JK	596	set_ckpt_flags(sbi->ckpt, CP_ERROR_FLAG);
14f4e690 JK	597	mutex_unlock(&sbi->cp_mutex);
cf2271e7	598	} else if (need_writecp) {
75ab4cb8	599	struct cp_control cpc = {
10027551	600	.reason = CP_RECOVERY,
75ab4cb8	601	};
14f4e690	602	mutex_unlock(&sbi->cp_mutex);
75ab4cb8	603	write_checkpoint(sbi, &cpc);
14f4e690 JK	604	} else {
14f4e690 JK	605	mutex_unlock(&sbi->cp_mutex);
cf2271e7	606	}
6ead1142	607	return err;
d624c96f	608	}