Re: [PATCH] ext4: fix WARNING in lock_two_nondirectories

[Matthew Wilcox]

On Sun, Dec 24, 2023 at 07:53:00PM +0800, Edward Adam Davis wrote:
> If inode is the ext4 boot loader inode, then when it is a directory, the inode
> should also be set to bad inode.

... what?  Are you saying that syzbot has randomly created a filesystem
where the boot loader inode is a directory?  If so, I'd suggest just
rejecting the filesystem with EFSCORRUPT.

> Reported-and-tested-by: syzbot+2c4a3b922a860084cc7f@xxxxxxxxxxxxxxxxxxxxxxxxx
> Signed-off-by: Edward Adam Davis <eadavis@xxxxxx>
> ---
>  fs/ext4/inode.c | 8 ++++++--
>  1 file changed, 6 insertions(+), 2 deletions(-)
> 
> diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
> index 61277f7f8722..b311f610f008 100644
> --- a/fs/ext4/inode.c
> +++ b/fs/ext4/inode.c
> @@ -4944,8 +4944,12 @@ struct inode *__ext4_iget(struct super_block *sb, unsigned long ino,
>  		inode->i_fop = &ext4_file_operations;
>  		ext4_set_aops(inode);
>  	} else if (S_ISDIR(inode->i_mode)) {
> -		inode->i_op = &ext4_dir_inode_operations;
> -		inode->i_fop = &ext4_dir_operations;
> +		if (ino == EXT4_BOOT_LOADER_INO)
> +			make_bad_inode(inode);
> +		else {
> +			inode->i_op = &ext4_dir_inode_operations;
> +			inode->i_fop = &ext4_dir_operations;
> +		}
>  	} else if (S_ISLNK(inode->i_mode)) {
>  		/* VFS does not allow setting these so must be corruption */
>  		if (IS_APPEND(inode) || IS_IMMUTABLE(inode)) {
> -- 
> 2.43.0
> 
>