Linux-f2fs-devel Archive on lore.kernel.org
 help / color / mirror / Atom feed
From: liujinbao <jinbaoliu365@gmail.com>
To: jaegeuk@kernel.org
Cc: Yongpeng Yang <yangyongpeng@xiaomi.com>,
	Sheng Yong <shengyong1@xiaomi.com>,
	liujinbao1 <liujinbao1@xiaomi.com>,
	linux-f2fs-devel@lists.sourceforge.net
Subject: Re: [f2fs-dev] [PATCH v3] fsck.f2fs: Check and clear invalid dent flag during recovery
Date: Tue, 20 Jan 2026 10:59:47 +0800	[thread overview]
Message-ID: <9ad2f6f9-476a-4f6a-ae55-e5b8c024afb3@gmail.com> (raw)
In-Reply-To: <20260114083705.944024-1-jinbaoliu365@gmail.com>

ping

On 2026/1/14 16:37, liujinbao1 wrote:
> From: liujinbao1 <liujinbao1@xiaomi.com>
>
> During SPO tests, when mounting F2FS, an -EINVAL error was returned.
> The issue originates from the f2fs_recover_inode_page function's
> check, where old_ni.blk_addr != NULL_ADDR under the conditions of
> IS_INODE(folio) && is_dent_dnode(folio).
> Clear dent flag of the node block to fix this issue.
> Test Step:
> 1.Modify files and induce SPO to generate fsync inode list.
>   blkaddr: 0x15828, ino: 1378, is_inode: 1, is_fsync: 2, is_dent: 0
>
> 2.Use inject.f2fs to set dent flag for a fsync inode.
>   inject.f2fs --cp 0 --mb flag --idx 2 --val 7 /dev/vda
>   Info: inject blkaddr[2] flag of cp 2: 0x3 -> 0x7
>
> 3.Boot verification, System fails to mount during recovery:
>   f2fs_recover_inode_page:ino: 1378, ni.blkaddr: 0x1581a,
>   old_ni.blk_addr != NULL_ADDR
>   F2FS-fs (vda): Cannot recover all fsync data errno=-22
> 4.Run fsck
>   fsck.f2fs -f /dev/vda
>   [ASSERT] (f2fs_find_fsync_inode:3924) --> Invalid dent flag:
>   blkaddr: 0x15828, ino: 1378, is_dent: 4, nat entry blkaddr: 0x1581a
>   [FIX] (f2fs_find_fsync_inode:3931) --> Clear dent flag: blkaddr: 0x15828, ino: 1378
> After this fix, the system boots normally.
>
> Signed-off-by: Sheng Yong <shengyong1@xiaomi.com>
> Signed-off-by: Yongpeng Yang <yangyongpeng@xiaomi.com>
> Signed-off-by: Jinbao Liu <liujinbao1@xiaomi.com>
> ---
> v3:
> - Add test steps to verify the effect.
> ---
> v2:
> - Clear the node_blk dent flag, proceed with recovery
> of this and subsequent data.
> ---
> v1:
> - Clear the node_blk fsync flag, set next_blkaddr=NULL,
> skip recovery of this and later data.
> ---
>   fsck/mount.c | 19 ++++++++++++++++++-
>   fsck/node.h  | 13 +++++++++++++
>   2 files changed, 31 insertions(+), 1 deletion(-)
>
> diff --git a/fsck/mount.c b/fsck/mount.c
> index 6f640a0..3385489 100644
> --- a/fsck/mount.c
> +++ b/fsck/mount.c
> @@ -3878,6 +3878,7 @@ int f2fs_find_fsync_inode(struct f2fs_sb_info *sbi, struct list_head *head)
>   
>   	while (1) {
>   		struct fsync_inode_entry *entry;
> +		struct f2fs_nat_entry nat_entry;
>   
>   		if (!f2fs_is_valid_blkaddr(sbi, blkaddr, META_POR))
>   			break;
> @@ -3902,8 +3903,24 @@ int f2fs_find_fsync_inode(struct f2fs_sb_info *sbi, struct list_head *head)
>   		}
>   		entry->blkaddr = blkaddr;
>   
> -		if (IS_INODE(node_blk) && is_dent_dnode(node_blk))
> +		if (IS_INODE(node_blk) && is_dent_dnode(node_blk)) {
> +			get_nat_entry(sbi, ino_of_node(node_blk), &nat_entry);
> +			if (is_valid_data_blkaddr(nat_entry.block_addr)) {
> +				ASSERT_MSG("Invalid dent flag: blkaddr: 0x%x, "
> +					"ino: %u, is_dent: %d, nat entry blkaddr: 0x%x\n",
> +					blkaddr, ino_of_node(node_blk), is_dent_dnode(node_blk),
> +					nat_entry.block_addr);
> +				if (c.fix_on && f2fs_dev_is_writable()) {
> +					FIX_MSG("Clear dent flag: blkaddr: 0x%x, ino: %u\n",
> +						blkaddr, ino_of_node(node_blk));
> +					set_dentry_mark(node_blk, 0);
> +					err = update_inode(sbi, node_blk, &blkaddr);
> +					ASSERT(err >= 0);
> +					goto next;
> +				}
> +			}
>   			entry->last_dentry = blkaddr;
> +		}
>   next:
>   		blkaddr = next_blkaddr_of_node(node_blk);
>   
> diff --git a/fsck/node.h b/fsck/node.h
> index 19f1e57..f3e2c46 100644
> --- a/fsck/node.h
> +++ b/fsck/node.h
> @@ -173,6 +173,19 @@ static inline void set_cold_node(struct f2fs_node *rn, bool is_dir)
>   	F2FS_NODE_FOOTER(rn)->flag = cpu_to_le32(flag);
>   }
>   
> +static inline void set_mark(struct f2fs_node *rn, int mark, int type)
> +{
> +	unsigned int flag = le32_to_cpu(F2FS_NODE_FOOTER(rn)->flag);
> +
> +	if (mark)
> +		flag |= (1 << type);
> +	else
> +		flag &= ~(1 << type);
> +	F2FS_NODE_FOOTER(rn)->flag = cpu_to_le32(flag);
> +}
> +
> +#define set_dentry_mark(page, mark)	set_mark(page, mark, DENT_BIT_SHIFT)
> +
>   #define is_fsync_dnode(node_blk)	is_node(node_blk, FSYNC_BIT_SHIFT)
>   #define is_dent_dnode(node_blk)		is_node(node_blk, DENT_BIT_SHIFT)
>   


_______________________________________________
Linux-f2fs-devel mailing list
Linux-f2fs-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel

  reply	other threads:[~2026-01-20  3:00 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-01-14  8:37 [f2fs-dev] [PATCH v3] fsck.f2fs: Check and clear invalid dent flag during recovery liujinbao1
2026-01-20  2:59 ` liujinbao [this message]
2026-01-20 12:26 ` Chao Yu via Linux-f2fs-devel

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=9ad2f6f9-476a-4f6a-ae55-e5b8c024afb3@gmail.com \
    --to=jinbaoliu365@gmail.com \
    --cc=jaegeuk@kernel.org \
    --cc=linux-f2fs-devel@lists.sourceforge.net \
    --cc=liujinbao1@xiaomi.com \
    --cc=shengyong1@xiaomi.com \
    --cc=yangyongpeng@xiaomi.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox