[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"

/*
 * 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 struct fsync_inode_entry *add_fsync_inode(struct list_head *head,
							struct inode *inode)
{
	struct fsync_inode_entry *entry;

	entry = kmem_cache_alloc(fsync_entry_slab, GFP_F2FS_ZERO);
	if (!entry)
		return NULL;

	entry->inode = inode;
	list_add_tail(&entry->list, head);

	return entry;
}

static void del_fsync_inode(struct fsync_inode_entry *entry)
{
	iput(entry->inode);
	list_del(&entry->list);
	kmem_cache_free(fsync_entry_slab, entry);
}

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 bool is_same_inode(struct inode *inode, struct page *ipage)
{
	struct f2fs_inode *ri = F2FS_INODE(ipage);
	struct timespec disk;

	if (!IS_INODE(ipage))
		return true;

	disk.tv_sec = le64_to_cpu(ri->i_ctime);
	disk.tv_nsec = le32_to_cpu(ri->i_ctime_nsec);
	if (timespec_compare(&inode->i_ctime, &disk) > 0)
		return false;

	disk.tv_sec = le64_to_cpu(ri->i_atime);
	disk.tv_nsec = le32_to_cpu(ri->i_atime_nsec);
	if (timespec_compare(&inode->i_atime, &disk) > 0)
		return false;

	disk.tv_sec = le64_to_cpu(ri->i_mtime);
	disk.tv_nsec = le32_to_cpu(ri->i_mtime_nsec);
	if (timespec_compare(&inode->i_mtime, &disk) > 0)
		return false;

	return true;
}

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 inode *inode;
	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);

	while (1) {
		struct fsync_inode_entry *entry;

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

		page = get_tmp_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_same_inode(entry->inode, page))
				goto next;
		} else {
			if (IS_INODE(page) && is_dent_dnode(page)) {
				err = recover_inode_page(sbi, page);
				if (err)
					break;
			}

			/*
			 * CP | dnode(F) | inode(DF)
			 * For this case, we should not give up now.
			 */
			inode = f2fs_iget(sbi->sb, ino_of_node(page));
			if (IS_ERR(inode)) {
				err = PTR_ERR(inode);
				if (err == -ENOENT) {
					err = 0;
					goto next;
				}
				break;
			}

			/* add this fsync inode to the list */
			entry = add_fsync_inode(head, inode);
			if (!entry) {
				err = -ENOMEM;
				iput(inode);
				break;
			}
		}
		entry->blkaddr = blkaddr;

		if (IS_INODE(page) && 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)
		del_fsync_inode(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, 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 dnode_of_data dn;
	struct node_info ni;
	unsigned int start, end;
	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), inode);
	end = start + ADDRS_PER_PAGE(page, inode);

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

	err = get_dnode_of_data(&dn, start, ALLOC_NODE);
	if (err)
		goto out;

	f2fs_wait_on_page_writeback(dn.node_page, NODE, true);

	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, 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);
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)
{
	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, CURSEG_WARM_NODE);
	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_tmp_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 (IS_INODE(page))
			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)
			del_fsync_inode(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, bool check_only)
{
	struct curseg_info *curseg = CURSEG_I(sbi, CURSEG_WARM_NODE);
	struct list_head inode_list;
	block_t blkaddr;
	int err;
	int ret = 0;
	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 || list_empty(&inode_list))
		goto out;

	if (check_only) {
		ret = 1;
		goto out;
	}

	need_writecp = true;

	/* step #2: recover data */
	err = recover_data(sbi, &inode_list);
	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),
			(loff_t)MAIN_BLKADDR(sbi) << PAGE_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);
		err = write_checkpoint(sbi, &cpc);
	} else {
		mutex_unlock(&sbi->cp_mutex);
	}
	return ret ? ret: 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
3f8ab270 CY	70	static struct fsync_inode_entry add_fsync_inode(struct list_head head,
	71	struct inode *inode)
	72	{
	73	struct fsync_inode_entry *entry;
	74
	75	entry = kmem_cache_alloc(fsync_entry_slab, GFP_F2FS_ZERO);
	76	if (!entry)
	77	return NULL;
	78
	79	entry->inode = inode;
	80	list_add_tail(&entry->list, head);
	81
	82	return entry;
	83	}
	84
	85	static void del_fsync_inode(struct fsync_inode_entry *entry)
	86	{
	87	iput(entry->inode);
	88	list_del(&entry->list);
	89	kmem_cache_free(fsync_entry_slab, entry);
	90	}
	91
c52e1b10	92	static int recover_dentry(struct inode inode, struct page ipage)
d624c96f	93	{
58bfaf44	94	struct f2fs_inode *raw_inode = F2FS_INODE(ipage);
74d0b917	95	nid_t pino = le32_to_cpu(raw_inode->i_pino);
6b8213d9	96	struct f2fs_dir_entry *de;
b7f7a5e0	97	struct qstr name;
d624c96f	98	struct page *page;
6b8213d9	99	struct inode dir, einode;
d624c96f JK	100	int err = 0;
d624c96f JK	101
ed57c27f JK	102	dir = f2fs_iget(inode->i_sb, pino);
	103	if (IS_ERR(dir)) {
	104	err = PTR_ERR(dir);
	105	goto out;
	106	}
	107
e7d55452 JK	108	if (file_enc_name(inode)) {
	109	iput(dir);
	110	return 0;
	111	}
	112
b7f7a5e0 AV	113	name.len = le32_to_cpu(raw_inode->i_namelen);
b7f7a5e0 AV	114	name.name = raw_inode->i_name;
d96b1431 CY	115
	116	if (unlikely(name.len > F2FS_NAME_LEN)) {
	117	WARN_ON(1);
	118	err = -ENAMETOOLONG;
86928f98	119	goto out_err;
d96b1431	120	}
6b8213d9 JK	121	retry:
6b8213d9 JK	122	de = f2fs_find_entry(dir, &name, &page);
418f6c27	123	if (de && inode->i_ino == le32_to_cpu(de->ino))
2e5558f4	124	goto out_unmap_put;
418f6c27	125
6b8213d9 JK	126	if (de) {
	127	einode = f2fs_iget(inode->i_sb, le32_to_cpu(de->ino));
	128	if (IS_ERR(einode)) {
	129	WARN_ON(1);
5c1f9927 CY	130	err = PTR_ERR(einode);
5c1f9927 CY	131	if (err == -ENOENT)
6b8213d9	132	err = -EEXIST;
2e5558f4 RK	133	goto out_unmap_put;
2e5558f4 RK	134	}
4081363f	135	err = acquire_orphan_inode(F2FS_I_SB(inode));
2e5558f4 RK	136	if (err) {
	137	iput(einode);
	138	goto out_unmap_put;
6b8213d9	139	}
dbeacf02	140	f2fs_delete_entry(de, page, dir, einode);
6b8213d9 JK	141	iput(einode);
6b8213d9 JK	142	goto retry;
d624c96f	143	}
510022a8	144	err = __f2fs_add_link(dir, &name, inode, inode->i_ino, inode->i_mode);
86928f98 JK	145	if (err)
	146	goto out_err;
	147
	148	if (is_inode_flag_set(F2FS_I(dir), FI_DELAY_IPUT)) {
	149	iput(dir);
	150	} else {
	151	add_dirty_dir_inode(dir);
	152	set_inode_flag(F2FS_I(dir), FI_DELAY_IPUT);
	153	}
	154
2e5558f4 RK	155	goto out;
	156
	157	out_unmap_put:
9486ba44	158	f2fs_dentry_kunmap(dir, page);
2e5558f4	159	f2fs_put_page(page, 0);
86928f98 JK	160	out_err:
86928f98 JK	161	iput(dir);
d624c96f	162	out:
6c311ec6 CF	163	f2fs_msg(inode->i_sb, KERN_NOTICE,
	164	"%s: ino = %x, name = %s, dir = %lx, err = %d",
	165	__func__, ino_of_node(ipage), raw_inode->i_name,
f28c06fa	166	IS_ERR(dir) ? 0 : dir->i_ino, err);
d624c96f JK	167	return err;
	168	}
	169
c52e1b10	170	static void recover_inode(struct inode inode, struct page page)
d624c96f	171	{
441ac5cb	172	struct f2fs_inode *raw = F2FS_INODE(page);
e7d55452	173	char *name;
441ac5cb JK	174
	175	inode->i_mode = le16_to_cpu(raw->i_mode);
	176	i_size_write(inode, le64_to_cpu(raw->i_size));
	177	inode->i_atime.tv_sec = le64_to_cpu(raw->i_mtime);
	178	inode->i_ctime.tv_sec = le64_to_cpu(raw->i_ctime);
	179	inode->i_mtime.tv_sec = le64_to_cpu(raw->i_mtime);
	180	inode->i_atime.tv_nsec = le32_to_cpu(raw->i_mtime_nsec);
	181	inode->i_ctime.tv_nsec = le32_to_cpu(raw->i_ctime_nsec);
	182	inode->i_mtime.tv_nsec = le32_to_cpu(raw->i_mtime_nsec);
f356fe0c	183
e7d55452 JK	184	if (file_enc_name(inode))
	185	name = "<encrypted>";
	186	else
	187	name = F2FS_INODE(page)->i_name;
	188
f356fe0c	189	f2fs_msg(inode->i_sb, KERN_NOTICE, "recover_inode: ino = %x, name = %s",
e7d55452	190	ino_of_node(page), name);
d624c96f JK	191	}
d624c96f JK	192
807b1e1c JK	193	static bool is_same_inode(struct inode inode, struct page ipage)
	194	{
	195	struct f2fs_inode *ri = F2FS_INODE(ipage);
	196	struct timespec disk;
	197
	198	if (!IS_INODE(ipage))
	199	return true;
	200
	201	disk.tv_sec = le64_to_cpu(ri->i_ctime);
	202	disk.tv_nsec = le32_to_cpu(ri->i_ctime_nsec);
	203	if (timespec_compare(&inode->i_ctime, &disk) > 0)
	204	return false;
	205
	206	disk.tv_sec = le64_to_cpu(ri->i_atime);
	207	disk.tv_nsec = le32_to_cpu(ri->i_atime_nsec);
	208	if (timespec_compare(&inode->i_atime, &disk) > 0)
	209	return false;
	210
	211	disk.tv_sec = le64_to_cpu(ri->i_mtime);
	212	disk.tv_nsec = le32_to_cpu(ri->i_mtime_nsec);
	213	if (timespec_compare(&inode->i_mtime, &disk) > 0)
	214	return false;
	215
	216	return true;
	217	}
	218
d624c96f JK	219	static int find_fsync_dnodes(struct f2fs_sb_info sbi, struct list_head head)
d624c96f JK	220	{
d71b5564	221	unsigned long long cp_ver = cur_cp_version(F2FS_CKPT(sbi));
d624c96f	222	struct curseg_info *curseg;
3f8ab270	223	struct inode *inode;
4c521f49	224	struct page *page = NULL;
d624c96f JK	225	block_t blkaddr;
	226	int err = 0;
	227
	228	/* get node pages in the current segment */
	229	curseg = CURSEG_I(sbi, CURSEG_WARM_NODE);
695fd1ed	230	blkaddr = NEXT_FREE_BLKADDR(sbi, curseg);
d624c96f	231
d624c96f JK	232	while (1) {
	233	struct fsync_inode_entry *entry;
	234
f0c9cada	235	if (!is_valid_blkaddr(sbi, blkaddr, META_POR))
4c521f49	236	return 0;
d624c96f	237
2b947003	238	page = get_tmp_page(sbi, blkaddr);
393ff91f	239
6ead1142	240	if (cp_ver != cpver_of_node(page))
f356fe0c	241	break;
d624c96f JK	242
	243	if (!is_fsync_dnode(page))
	244	goto next;
	245
	246	entry = get_fsync_inode(head, ino_of_node(page));
807b1e1c JK	247	if (entry) {
	248	if (!is_same_inode(entry->inode, page))
	249	goto next;
	250	} else {
d624c96f	251	if (IS_INODE(page) && is_dent_dnode(page)) {
6ead1142 JK	252	err = recover_inode_page(sbi, page);
6ead1142 JK	253	if (err)
f356fe0c	254	break;
d624c96f JK	255	}
d624c96f JK	256
441ac5cb JK	257	/*
	258	* CP \| dnode(F) \| inode(DF)
	259	* For this case, we should not give up now.
	260	*/
3f8ab270 CY	261	inode = f2fs_iget(sbi->sb, ino_of_node(page));
	262	if (IS_ERR(inode)) {
	263	err = PTR_ERR(inode);
8fbc418f JK	264	if (err == -ENOENT) {
8fbc418f JK	265	err = 0;
441ac5cb	266	goto next;
8fbc418f	267	}
f356fe0c	268	break;
d624c96f	269	}
3f8ab270 CY	270
	271	/* add this fsync inode to the list */
	272	entry = add_fsync_inode(head, inode);
	273	if (!entry) {
	274	err = -ENOMEM;
	275	iput(inode);
	276	break;
	277	}
d624c96f	278	}
addbe45b JK	279	entry->blkaddr = blkaddr;
addbe45b JK	280
608514de JK	281	if (IS_INODE(page) && is_dent_dnode(page))
608514de JK	282	entry->last_dentry = blkaddr;
d624c96f JK	283	next:
	284	/* check next segment */
	285	blkaddr = next_blkaddr_of_node(page);
4c521f49	286	f2fs_put_page(page, 1);
635aee1f CY	287
635aee1f CY	288	ra_meta_pages_cond(sbi, blkaddr);
d624c96f	289	}
4c521f49	290	f2fs_put_page(page, 1);
d624c96f JK	291	return err;
	292	}
	293
5ebefc5b	294	static void destroy_fsync_dnodes(struct list_head *head)
d624c96f	295	{
d8b79b2f DC	296	struct fsync_inode_entry entry, tmp;
d8b79b2f DC	297
3f8ab270 CY	298	list_for_each_entry_safe(entry, tmp, head, list)
3f8ab270 CY	299	del_fsync_inode(entry);
d624c96f JK	300	}
d624c96f JK	301
39cf72cf	302	static int check_index_in_prev_nodes(struct f2fs_sb_info *sbi,
b292dcab	303	block_t blkaddr, struct dnode_of_data *dn)
d624c96f JK	304	{
	305	struct seg_entry *sentry;
	306	unsigned int segno = GET_SEGNO(sbi, blkaddr);
491c0854	307	unsigned short blkoff = GET_BLKOFF_FROM_SEG0(sbi, blkaddr);
f6517cfc	308	struct f2fs_summary_block *sum_node;
d624c96f	309	struct f2fs_summary sum;
f6517cfc	310	struct page sum_page, node_page;
c9ef4810	311	struct dnode_of_data tdn = *dn;
b292dcab	312	nid_t ino, nid;
d624c96f	313	struct inode *inode;
de93653f	314	unsigned int offset;
d624c96f JK	315	block_t bidx;
	316	int i;
	317
	318	sentry = get_seg_entry(sbi, segno);
	319	if (!f2fs_test_bit(blkoff, sentry->cur_valid_map))
39cf72cf	320	return 0;
d624c96f JK	321
	322	/* Get the previous summary */
	323	for (i = CURSEG_WARM_DATA; i <= CURSEG_COLD_DATA; i++) {
	324	struct curseg_info *curseg = CURSEG_I(sbi, i);
	325	if (curseg->segno == segno) {
	326	sum = curseg->sum_blk->entries[blkoff];
f6517cfc	327	goto got_it;
d624c96f JK	328	}
d624c96f JK	329	}
d624c96f	330
f6517cfc JK	331	sum_page = get_sum_page(sbi, segno);
	332	sum_node = (struct f2fs_summary_block *)page_address(sum_page);
	333	sum = sum_node->entries[blkoff];
	334	f2fs_put_page(sum_page, 1);
	335	got_it:
b292dcab JK	336	/* Use the locked dnode page and inode */
	337	nid = le32_to_cpu(sum.nid);
	338	if (dn->inode->i_ino == nid) {
b292dcab	339	tdn.nid = nid;
c9ef4810 JK	340	if (!dn->inode_page_locked)
c9ef4810 JK	341	lock_page(dn->inode_page);
b292dcab	342	tdn.node_page = dn->inode_page;
060dd67b	343	tdn.ofs_in_node = le16_to_cpu(sum.ofs_in_node);
c9ef4810	344	goto truncate_out;
b292dcab	345	} else if (dn->nid == nid) {
060dd67b	346	tdn.ofs_in_node = le16_to_cpu(sum.ofs_in_node);
c9ef4810	347	goto truncate_out;
b292dcab JK	348	}
b292dcab JK	349
d624c96f	350	/* Get the node page */
b292dcab	351	node_page = get_node_page(sbi, nid);
39cf72cf JK	352	if (IS_ERR(node_page))
39cf72cf JK	353	return PTR_ERR(node_page);
de93653f JK	354
de93653f JK	355	offset = ofs_of_node(node_page);
d624c96f JK	356	ino = ino_of_node(node_page);
	357	f2fs_put_page(node_page, 1);
	358
60979115 JK	359	if (ino != dn->inode->i_ino) {
	360	/* Deallocate previous index in the node page */
	361	inode = f2fs_iget(sbi->sb, ino);
	362	if (IS_ERR(inode))
	363	return PTR_ERR(inode);
	364	} else {
	365	inode = dn->inode;
	366	}
06025f4d	367
81ca7350	368	bidx = start_bidx_of_node(offset, inode) + le16_to_cpu(sum.ofs_in_node);
de93653f	369
c9ef4810 JK	370	/*
	371	* if inode page is locked, unlock temporarily, but its reference
	372	* count keeps alive.
	373	*/
	374	if (ino == dn->inode->i_ino && dn->inode_page_locked)
	375	unlock_page(dn->inode_page);
	376
	377	set_new_dnode(&tdn, inode, NULL, NULL, 0);
	378	if (get_dnode_of_data(&tdn, bidx, LOOKUP_NODE))
	379	goto out;
	380
	381	if (tdn.data_blkaddr == blkaddr)
	382	truncate_data_blocks_range(&tdn, 1);
	383
	384	f2fs_put_dnode(&tdn);
	385	out:
	386	if (ino != dn->inode->i_ino)
60979115	387	iput(inode);
c9ef4810 JK	388	else if (dn->inode_page_locked)
	389	lock_page(dn->inode_page);
	390	return 0;
	391
	392	truncate_out:
	393	if (datablock_addr(tdn.node_page, tdn.ofs_in_node) == blkaddr)
	394	truncate_data_blocks_range(&tdn, 1);
	395	if (dn->inode->i_ino == nid && !dn->inode_page_locked)
	396	unlock_page(dn->inode_page);
39cf72cf	397	return 0;
d624c96f JK	398	}
d624c96f JK	399
6ead1142	400	static int do_recover_data(struct f2fs_sb_info sbi, struct inode inode,
d624c96f JK	401	struct page *page, block_t blkaddr)
d624c96f JK	402	{
d624c96f	403	struct dnode_of_data dn;
d624c96f	404	struct node_info ni;
81ca7350	405	unsigned int start, end;
f356fe0c	406	int err = 0, recovered = 0;
d624c96f	407
1c35a90e JK	408	/* step 1: recover xattr */
	409	if (IS_INODE(page)) {
	410	recover_inline_xattr(inode, page);
	411	} else if (f2fs_has_xattr_block(ofs_of_node(page))) {
bc4a1f87 JK	412	/*
	413	* Deprecated; xattr blocks should be found from cold log.
	414	* But, we should remain this for backward compatibility.
	415	*/
1c35a90e	416	recover_xattr_data(inode, page, blkaddr);
1e1bb4ba	417	goto out;
1c35a90e	418	}
1e1bb4ba	419
1c35a90e JK	420	/* step 2: recover inline data */
1c35a90e JK	421	if (recover_inline_data(inode, page))
abb2366c JK	422	goto out;
abb2366c JK	423
1c35a90e	424	/* step 3: recover data indices */
81ca7350 CY	425	start = start_bidx_of_node(ofs_of_node(page), inode);
81ca7350 CY	426	end = start + ADDRS_PER_PAGE(page, inode);
d624c96f JK	427
d624c96f JK	428	set_new_dnode(&dn, inode, NULL, NULL, 0);
39936837	429
6ead1142	430	err = get_dnode_of_data(&dn, start, ALLOC_NODE);
72235541	431	if (err)
1e1bb4ba	432	goto out;
d624c96f	433
fec1d657	434	f2fs_wait_on_page_writeback(dn.node_page, NODE, true);
d624c96f JK	435
d624c96f JK	436	get_node_info(sbi, dn.nid, &ni);
9850cf4a JK	437	f2fs_bug_on(sbi, ni.ino != ino_of_node(page));
9850cf4a JK	438	f2fs_bug_on(sbi, ofs_of_node(dn.node_page) != ofs_of_node(page));
d624c96f	439
12a8343e	440	for (; start < end; start++, dn.ofs_in_node++) {
d624c96f JK	441	block_t src, dest;
	442
	443	src = datablock_addr(dn.node_page, dn.ofs_in_node);
	444	dest = datablock_addr(page, dn.ofs_in_node);
	445
12a8343e CY	446	/* skip recovering if dest is the same as src */
	447	if (src == dest)
	448	continue;
	449
	450	/* dest is invalid, just invalidate src block */
	451	if (dest == NULL_ADDR) {
	452	truncate_data_blocks_range(&dn, 1);
	453	continue;
	454	}
	455
	456	/*
	457	* dest is reserved block, invalidate src block
	458	* and then reserve one new block in dnode page.
	459	*/
	460	if (dest == NEW_ADDR) {
	461	truncate_data_blocks_range(&dn, 1);
	462	err = reserve_new_block(&dn);
	463	f2fs_bug_on(sbi, err);
	464	continue;
	465	}
	466
	467	/* dest is valid block, try to recover from src to dest */
	468	if (is_valid_blkaddr(sbi, dest, META_POR)) {
e03b07d9	469
d624c96f	470	if (src == NULL_ADDR) {
5d56b671	471	err = reserve_new_block(&dn);
d624c96f	472	/* We should not get -ENOSPC */
9850cf4a	473	f2fs_bug_on(sbi, err);
d624c96f JK	474	}
	475
	476	/* Check the previous node page having this index */
39cf72cf JK	477	err = check_index_in_prev_nodes(sbi, dest, &dn);
	478	if (err)
	479	goto err;
d624c96f	480
d624c96f	481	/* write dummy data page */
528e3459	482	f2fs_replace_block(sbi, &dn, src, dest,
28bc106b	483	ni.version, false, false);
f356fe0c	484	recovered++;
d624c96f	485	}
d624c96f JK	486	}
d624c96f JK	487
d624c96f JK	488	if (IS_INODE(dn.node_page))
	489	sync_inode_page(&dn);
	490
	491	copy_node_footer(dn.node_page, page);
	492	fill_node_footer(dn.node_page, dn.nid, ni.ino,
	493	ofs_of_node(page), false);
	494	set_page_dirty(dn.node_page);
39cf72cf	495	err:
d624c96f	496	f2fs_put_dnode(&dn);
1e1bb4ba	497	out:
6c311ec6 CF	498	f2fs_msg(sbi->sb, KERN_NOTICE,
	499	"recover_data: ino = %lx, recovered = %d blocks, err = %d",
	500	inode->i_ino, recovered, err);
39cf72cf	501	return err;
d624c96f JK	502	}
d624c96f JK	503
b7973f23	504	static int recover_data(struct f2fs_sb_info sbi, struct list_head head)
d624c96f	505	{
d71b5564	506	unsigned long long cp_ver = cur_cp_version(F2FS_CKPT(sbi));
d624c96f	507	struct curseg_info *curseg;
4c521f49	508	struct page *page = NULL;
6ead1142	509	int err = 0;
d624c96f JK	510	block_t blkaddr;
	511
	512	/* get node pages in the current segment */
b7973f23	513	curseg = CURSEG_I(sbi, CURSEG_WARM_NODE);
d624c96f JK	514	blkaddr = NEXT_FREE_BLKADDR(sbi, curseg);
d624c96f JK	515
d624c96f JK	516	while (1) {
	517	struct fsync_inode_entry *entry;
	518
f0c9cada	519	if (!is_valid_blkaddr(sbi, blkaddr, META_POR))
4c521f49	520	break;
d624c96f	521
635aee1f CY	522	ra_meta_pages_cond(sbi, blkaddr);
635aee1f CY	523
2b947003	524	page = get_tmp_page(sbi, blkaddr);
393ff91f	525
4c521f49 JK	526	if (cp_ver != cpver_of_node(page)) {
4c521f49 JK	527	f2fs_put_page(page, 1);
45856aff	528	break;
4c521f49	529	}
d624c96f JK	530
	531	entry = get_fsync_inode(head, ino_of_node(page));
	532	if (!entry)
	533	goto next;
441ac5cb JK	534	/*
	535	* inode(x) \| CP \| inode(x) \| dnode(F)
	536	* In this case, we can lose the latest inode(x).
c52e1b10	537	* So, call recover_inode for the inode update.
441ac5cb	538	*/
608514de	539	if (IS_INODE(page))
c52e1b10 JK	540	recover_inode(entry->inode, page);
	541	if (entry->last_dentry == blkaddr) {
	542	err = recover_dentry(entry->inode, page);
	543	if (err) {
	544	f2fs_put_page(page, 1);
	545	break;
	546	}
	547	}
6ead1142	548	err = do_recover_data(sbi, entry->inode, page, blkaddr);
4c521f49 JK	549	if (err) {
4c521f49 JK	550	f2fs_put_page(page, 1);
45856aff	551	break;
4c521f49	552	}
d624c96f	553
3f8ab270 CY	554	if (entry->blkaddr == blkaddr)
3f8ab270 CY	555	del_fsync_inode(entry);
d624c96f JK	556	next:
	557	/* check next segment */
	558	blkaddr = next_blkaddr_of_node(page);
4c521f49	559	f2fs_put_page(page, 1);
d624c96f	560	}
6ead1142 JK	561	if (!err)
	562	allocate_new_segments(sbi);
	563	return err;
d624c96f JK	564	}
d624c96f JK	565
6781eabb	566	int recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only)
d624c96f	567	{
cf2271e7	568	struct curseg_info *curseg = CURSEG_I(sbi, CURSEG_WARM_NODE);
d624c96f	569	struct list_head inode_list;
cf2271e7	570	block_t blkaddr;
6ead1142	571	int err;
6781eabb	572	int ret = 0;
aabe5136	573	bool need_writecp = false;
d624c96f JK	574
d624c96f JK	575	fsync_entry_slab = f2fs_kmem_cache_create("f2fs_fsync_inode_entry",
e8512d2e	576	sizeof(struct fsync_inode_entry));
6bacf52f	577	if (!fsync_entry_slab)
6ead1142	578	return -ENOMEM;
d624c96f JK	579
	580	INIT_LIST_HEAD(&inode_list);
	581
14f4e690 JK	582	/* prevent checkpoint */
	583	mutex_lock(&sbi->cp_mutex);
	584
cf2271e7 JK	585	blkaddr = NEXT_FREE_BLKADDR(sbi, curseg);
cf2271e7 JK	586
315df839	587	/* step #1: find fsynced inode numbers */
6ead1142	588	err = find_fsync_dnodes(sbi, &inode_list);
6781eabb	589	if (err \|\| list_empty(&inode_list))
d624c96f JK	590	goto out;
d624c96f JK	591
6781eabb JK	592	if (check_only) {
6781eabb JK	593	ret = 1;
d624c96f	594	goto out;
6781eabb	595	}
d624c96f	596
aabe5136	597	need_writecp = true;
691c6fd2	598
d624c96f	599	/* step #2: recover data */
b7973f23	600	err = recover_data(sbi, &inode_list);
b307384e	601	if (!err)
9850cf4a	602	f2fs_bug_on(sbi, !list_empty(&inode_list));
d624c96f	603	out:
5ebefc5b	604	destroy_fsync_dnodes(&inode_list);
d624c96f	605	kmem_cache_destroy(fsync_entry_slab);
cf2271e7	606
4c521f49 JK	607	/* truncate meta pages to be used by the recovery */
4c521f49 JK	608	truncate_inode_pages_range(META_MAPPING(sbi),
09cbfeaf	609	(loff_t)MAIN_BLKADDR(sbi) << PAGE_SHIFT, -1);
4c521f49	610
cf2271e7 JK	611	if (err) {
	612	truncate_inode_pages_final(NODE_MAPPING(sbi));
	613	truncate_inode_pages_final(META_MAPPING(sbi));
	614	}
	615
caf0047e	616	clear_sbi_flag(sbi, SBI_POR_DOING);
cf2271e7	617	if (err) {
e90c2d28 CY	618	bool invalidate = false;
	619
	620	if (discard_next_dnode(sbi, blkaddr))
	621	invalidate = true;
cf2271e7 JK	622
	623	/* Flush all the NAT/SIT pages */
	624	while (get_pages(sbi, F2FS_DIRTY_META))
	625	sync_meta_pages(sbi, META, LONG_MAX);
e90c2d28 CY	626
	627	/* invalidate temporary meta page */
	628	if (invalidate)
	629	invalidate_mapping_pages(META_MAPPING(sbi),
	630	blkaddr, blkaddr);
	631
14f4e690 JK	632	set_ckpt_flags(sbi->ckpt, CP_ERROR_FLAG);
14f4e690 JK	633	mutex_unlock(&sbi->cp_mutex);
cf2271e7	634	} else if (need_writecp) {
75ab4cb8	635	struct cp_control cpc = {
10027551	636	.reason = CP_RECOVERY,
75ab4cb8	637	};
14f4e690	638	mutex_unlock(&sbi->cp_mutex);
c34f42e2	639	err = write_checkpoint(sbi, &cpc);
14f4e690 JK	640	} else {
14f4e690 JK	641	mutex_unlock(&sbi->cp_mutex);
cf2271e7	642	}
6781eabb	643	return ret ? ret: err;
d624c96f	644	}