linux-kernel - Re: [PATCH v2 2/2] f2fs: Add sanity checks before unlinking and loading inodes

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <1f519357-a489-41fe-8159-a8e319aedd17@kernel.org>
Date: Mon, 3 Nov 2025 16:12:15 +0800
From: Chao Yu <chao@...nel.org>
To: "Nikola Z. Ivanov" <zlatistiv@...il.com>
Cc: chao@...nel.org, jaegeuk@...nel.org,
 linux-f2fs-devel@...ts.sourceforge.net, linux-kernel@...r.kernel.org,
 skhan@...uxfoundation.org, david.hunter.linux@...il.com,
 linux-kernel-mentees@...ts.linuxfoundation.org, khalid@...nel.org,
 syzbot+c07d47c7bc68f47b9083@...kaller.appspotmail.com
Subject: Re: [PATCH v2 2/2] f2fs: Add sanity checks before unlinking and
 loading inodes

On 11/3/25 15:55, Nikola Z. Ivanov wrote:
> On Mon, Nov 03, 2025 at 10:35:17AM +0800, Chao Yu wrote:
>> On 11/1/25 20:56, Nikola Z. Ivanov wrote:
>>> Add check for inode->i_nlink == 1 for directories during unlink,
>>> as their value is decremented twice, which can trigger a warning in
>>> drop_nlink. In such case mark the filesystem as corrupted and return
>>> from the function call with the relevant failure return value.
>>>
>>> Additionally add the 2 checks for i_nlink == 0 and i_nlink == 1 in
>>> sanity_check_inode in order to detect on-disk corruption early.
>>>
>>> Reported-by: syzbot+c07d47c7bc68f47b9083@...kaller.appspotmail.com
>>> Closes: https://syzkaller.appspot.com/bug?extid=c07d47c7bc68f47b9083
>>> Tested-by: syzbot+c07d47c7bc68f47b9083@...kaller.appspotmail.com
>>> Signed-off-by: Nikola Z. Ivanov <zlatistiv@...il.com>
>>> ---
>>>  fs/f2fs/inode.c | 10 ++++++++++
>>>  fs/f2fs/namei.c | 15 +++++++++++----
>>>  2 files changed, 21 insertions(+), 4 deletions(-)
>>>
>>> diff --git a/fs/f2fs/inode.c b/fs/f2fs/inode.c
>>> index 8c4eafe9ffac..089cbf3646f0 100644
>>> --- a/fs/f2fs/inode.c
>>> +++ b/fs/f2fs/inode.c
>>> @@ -294,6 +294,16 @@ static bool sanity_check_inode(struct inode *inode, struct folio *node_folio)
>>>  		return false;
>>>  	}
>>>  
>>> +	if (unlikely(inode->i_nlink == 0)) {
>>
>> This is a possible case, as an orphan inode may exist in filesystem after sudden
>> power-cut.
>>
>> Thanks,
>>
> 
> Hi Chao,
> 
> Do you suggest that it should not be wrapped in unlikely()?

Nikola,

No, I think we should not add this sanity check "inode->i_nlink == 0"
into sanity_check_inode(), as for an orphan inode, its i_nlink is zero.
We expect to get the inode w/o failure in recover_orphan_inode().

> 
> I also now realise that I intended to wrap the "else if" case
> as well but I've missed it in the final patch.

Looks fine to add unlike for the "dir->i_nlink == 1" case.

Thanks,

> 
> Should I resend the patch with both cases wrapped in "unlikely()"
> or would you suggest otherwise?
> 
> 
>>> +		f2fs_warn(F2FS_I_SB(inode), "%s: inode (ino=%lx) has zero i_nlink",
>>> +			  __func__, inode->i_ino);
>>> +		return false;
>>> +	} else if (S_ISDIR(inode->i_mode) && inode->i_nlink == 1) {
>>> +		f2fs_warn(F2FS_I_SB(inode), "%s: directory inode (ino=%lx) has a single i_nlink",
>>> +			  __func__, inode->i_ino);
>>> +		return false;
>>> +	}
>>> +
>>>  	if (f2fs_has_extra_attr(inode)) {
>>>  		if (!f2fs_sb_has_extra_attr(sbi)) {
>>>  			f2fs_warn(sbi, "%s: inode (ino=%lx) is with extra_attr, but extra_attr feature is off",
>>> diff --git a/fs/f2fs/namei.c b/fs/f2fs/namei.c
>>> index 40cf80fd9d9a..d13077bad482 100644
>>> --- a/fs/f2fs/namei.c
>>> +++ b/fs/f2fs/namei.c
>>> @@ -572,10 +572,11 @@ static int f2fs_unlink(struct inode *dir, struct dentry *dentry)
>>>  	if (unlikely(inode->i_nlink == 0)) {
>>>  		f2fs_warn(F2FS_I_SB(inode), "%s: inode (ino=%lx) has zero i_nlink",
>>>  			  __func__, inode->i_ino);
>>> -		err = -EFSCORRUPTED;
>>> -		set_sbi_flag(F2FS_I_SB(inode), SBI_NEED_FSCK);
>>> -		f2fs_folio_put(folio, false);
>>> -		goto out;
>>> +		goto corrupted;
>>> +	} else if (S_ISDIR(inode->i_mode) && inode->i_nlink == 1) {
>>> +		f2fs_warn(F2FS_I_SB(inode), "%s: directory inode (ino=%lx) has a single i_nlink",
>>> +			  __func__, inode->i_ino);
>>> +		goto corrupted;
>>>  	}
>>>  
>>>  	f2fs_balance_fs(sbi, true);
>>> @@ -601,6 +602,12 @@ static int f2fs_unlink(struct inode *dir, struct dentry *dentry)
>>>  
>>>  	if (IS_DIRSYNC(dir))
>>>  		f2fs_sync_fs(sbi->sb, 1);
>>> +
>>> +	goto out;
>>> +corrupted:
>>> +	err = -EFSCORRUPTED;
>>> +	set_sbi_flag(F2FS_I_SB(inode), SBI_NEED_FSCK);
>>> +	f2fs_folio_put(folio, false);
>>>  out:
>>>  	trace_f2fs_unlink_exit(inode, err);
>>>  	return err;
>>