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;
From:   "Eric W. Biederman" <ebiederm@xmission.com>
To:     Yongzhi Liu <lyz_cs@pku.edu.cn>
Cc:     serge@hallyn.com, jmorris@namei.org, viro@zeniv.linux.org.uk,
        dhowells@redhat.com, linux-security-module@vger.kernel.org,
        linux-kernel@vger.kernel.org
References: <1652722802-66170-1-git-send-email-lyz_cs@pku.edu.cn>
Date:   Mon, 16 May 2022 13:14:03 -0500
In-Reply-To: <1652722802-66170-1-git-send-email-lyz_cs@pku.edu.cn> (Yongzhi
        Liu's message of "Mon, 16 May 2022 10:40:02 -0700")
Message-ID: <87y1z1bb10.fsf@email.froward.int.ebiederm.org>
User-Agent: Gnus/5.13 (Gnus v5.13) Emacs/27.1 (gnu/linux)
MIME-Version: 1.0
Content-Type: text/plain
Subject: Re: [PATCH] commoncap: check return value to avoid null pointer
 dereference
Precedence: bulk

Yongzhi Liu <lyz_cs@pku.edu.cn> writes:

> The pointer inode is dereferenced before a null pointer
> check on inode, hence if inode is actually null we will
> get a null pointer dereference. Fix this by only dereferencing
> inode after the null pointer check on inode.
>
> Fixes: c6f493d631c ("VFS: security/: d_backing_inode() annotations")
> Fixes: 8db6c34 ("Introduce v3 namespaced file capabilities")

I don't see how this fixes anything.

The dentry should be non-negative so d_backing_inode should always
return true.


> Signed-off-by: Yongzhi Liu <lyz_cs@pku.edu.cn>
> ---
>  security/commoncap.c | 8 ++++++--
>  1 file changed, 6 insertions(+), 2 deletions(-)
>
> diff --git a/security/commoncap.c b/security/commoncap.c
> index 5fc8986..978f011 100644
> --- a/security/commoncap.c
> +++ b/security/commoncap.c
> @@ -298,6 +298,8 @@ int cap_inode_need_killpriv(struct dentry *dentry)
>  	struct inode *inode = d_backing_inode(dentry);
>  	int error;
>  
> +	if (!inode)
> +		return 0;

How can  dentry->d_inode be valid and d_backing_inode not be valid?

That would seem to be a bug elsewhere in the code if it actually happens.

>  	error = __vfs_getxattr(dentry, inode, XATTR_NAME_CAPS, NULL, 0);
>  	return error > 0;
>  }
> @@ -545,11 +547,13 @@ int cap_convert_nscap(struct user_namespace *mnt_userns, struct dentry *dentry,
>  	const struct vfs_cap_data *cap = *ivalue;
>  	__u32 magic, nsmagic;
>  	struct inode *inode = d_backing_inode(dentry);
> -	struct user_namespace *task_ns = current_user_ns(),
> -		*fs_ns = inode->i_sb->s_user_ns;
> +	struct user_namespace *task_ns = current_user_ns(), *fs_ns;
>  	kuid_t rootid;
>  	size_t newsize;
>  
> +	if (!inode)
> +		return -EINVAL;
> +	fs_ns = inode->i_sb->s_user_ns;
>  	if (!*ivalue)
>  		return -EINVAL;
>  	if (!validheader(size, cap))

Same with this one.  Short of a negative dentry I don't see how
d_backing_inode can be NULL, and we are talking about negative
dentries that should have been handled long before these two functions
are called.

Eric