Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Mon, 15 Aug 2022 19:27:09 +0100
From:   Al Viro <viro@zeniv.linux.org.uk>
To:     Ryusuke Konishi <konishi.ryusuke@gmail.com>
Cc:     Andrew Morton <akpm@linux-foundation.org>,
        linux-nilfs <linux-nilfs@vger.kernel.org>,
        LKML <linux-kernel@vger.kernel.org>,
        Jiacheng Xu <stitch@zju.edu.cn>,
        Mudong Liang <mudongliangabcd@gmail.com>
Subject: Re: [PATCH] nilfs2: fix use-after-free bug in nilfs_mdt_destroy()
Message-ID: <YvqP/f2P2YgIIO9U@ZenIV>
References: <20220815175114.23576-1-konishi.ryusuke@gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20220815175114.23576-1-konishi.ryusuke@gmail.com>
Sender: Al Viro <viro@ftp.linux.org.uk>
Precedence: bulk

On Tue, Aug 16, 2022 at 02:51:14AM +0900, Ryusuke Konishi wrote:
> In alloc_inode(), inode_init_always() could return -ENOMEM if
> security_inode_alloc() fails.  If this happens for nilfs2,
> nilfs_free_inode() is called without initializing inode->i_private and
> nilfs_free_inode() wrongly calls nilfs_mdt_destroy(), which frees
> uninitialized inode->i_private and can trigger a crash.
> 
> Fix this bug by initializing inode->i_private in nilfs_alloc_inode().
> 
> Link: https://lkml.kernel.org/r/CAFcO6XOcf1Jj2SeGt=jJV59wmhESeSKpfR0omdFRq+J9nD1vfQ@mail.gmail.com
> Link: https://lkml.kernel.org/r/20211011030956.2459172-1-mudongliangabcd@gmail.com
> Reported-by: butt3rflyh4ck <butterflyhuangxx@gmail.com>
> Reported-by: Hao Sun <sunhao.th@gmail.com>
> Reported-by: Jiacheng Xu <stitch@zju.edu.cn>
> Reported-by: Mudong Liang <mudongliangabcd@gmail.com>
> Signed-off-by: Ryusuke Konishi <konishi.ryusuke@gmail.com>
> Cc: Al Viro <viro@zeniv.linux.org.uk>
> ---
>  fs/nilfs2/super.c | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/fs/nilfs2/super.c b/fs/nilfs2/super.c
> index ba108f915391..aca5614f1b44 100644
> --- a/fs/nilfs2/super.c
> +++ b/fs/nilfs2/super.c
> @@ -159,6 +159,7 @@ struct inode *nilfs_alloc_inode(struct super_block *sb)
>  	ii->i_cno = 0;
>  	ii->i_assoc_inode = NULL;
>  	ii->i_bmap = &ii->i_bmap_data;
> +	ii->vfs_inode.i_private = NULL;
>  	return &ii->vfs_inode;
>  }

FWIW, I think it's better to deal with that in inode_init_always(), but
not just moving ->i_private initialization up - we ought to move
security_inode_alloc() to the very end.  No sense playing whack-a-mole
with further possible bugs of that sort...