Re: [RFC RFT PATCH] ocfs2: Invalidate inode if i_mode is zero after block read

[Joseph Qi <joseph.qi@linux.alibaba.com>]

On 2025/10/23 06:27, Ahmet Eray Karadag wrote:
> A panic occurs in ocfs2_unlink due to WARN_ON(inode->i_nlink == 0) when
> handling a corrupted inode with i_mode=0 and i_nlink=0 in memory.
> 
> This "zombie" inode is created because ocfs2_read_locked_inode proceeds
> even after ocfs2_validate_inode_block successfully validates a block
> that structurally looks okay (passes checksum, signature etc.) but
> contains semantically invalid data (specifically i_mode=0). The current
> validation function doesn't check for i_mode being zero.
> 
> This results in an in-memory inode with i_mode=0 being added to the VFS
> cache, which later triggers the panic during unlink.
> 
> Prevent this by adding an explicit check for i_mode == 0 within
> ocfs2_validate_inode_block. If i_mode is zero, return -EFSCORRUPTED to signal
> corruption. This causes the caller (ocfs2_read_locked_inode) to invoke
> make_bad_inode(), correctly preventing the zombie inode from entering
> the cache.
> 
> ---
> [RFC]:
> The current fix handles i_mode=0 corruption detected during inode read
> by returning -EFSCORRUPTED from ocfs2_validate_inode_block, which leads to
> make_bad_inode() being called, preventing the corrupted inode from
> entering the cache. This approach avoids immediately forcing the entire
> filesystem read-only, assuming the corruption might be localized to
> this inode.
> 
> Is this less aggressive error handling strategy appropriate for i_mode=0
> corruption? Or is this condition considered severe enough that we *should*
> explicitly call ocfs2_error() within the validation function to guarantee
> the filesystem is marked read-only immediately upon detection?
> Feedback and testing on the correct severity assessment and error
> handling for this type of corruption would be appreciated.
> ---
> 
> Reported-by: syzbot+55c40ae8a0e5f3659f2b@syzkaller.appspotmail.com
> Fixes: https://syzkaller.appspot.com/bug?extid=55c40ae8a0e5f3659f2b
> Co-developed-by: Albin Babu Varghese <albinbabuvarghese20@gmail.com>
> Signed-off-by: Albin Babu Varghese <albinbabuvarghese20@gmail.com>
> Signed-off-by: Ahmet Eray Karadag <eraykrdg1@gmail.com>
> ---
>  fs/ocfs2/inode.c | 6 ++++++
>  1 file changed, 6 insertions(+)
> 
> diff --git a/fs/ocfs2/inode.c b/fs/ocfs2/inode.c
> index 14bf440ea4df..d4142ff9ce65 100644
> --- a/fs/ocfs2/inode.c
> +++ b/fs/ocfs2/inode.c
> @@ -1456,6 +1456,12 @@ int ocfs2_validate_inode_block(struct super_block *sb,
>  		goto bail;
>  	}
>  
> +	if (unlikely(le16_to_cpu(di->i_mode) == 0)) {

It seems the buggy image is carefully crafted so that it can pass the
OCFS2_VALID_FL check.
Checking i_mode here looks wried. Could we check i_links_count instead?

Thanks,
Joseph

> +		mlog(ML_ERROR, "Invalid dinode #%llu: i_mode is zero!\n",
> +                           (unsigned long long)bh->b_blocknr);
> +		rc = -EFSCORRUPTED;
> +		goto bail;
> +	}
>  	/*
>  	 * Errors after here are fatal.
>  	 */