Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
MIME-Version: 1.0
References: <20210514215209.GA33310@embeddedor>
In-Reply-To: <20210514215209.GA33310@embeddedor>
From:   Ilya Dryomov <idryomov@gmail.com>
Date:   Sat, 15 May 2021 12:42:03 +0200
Message-ID: <CAOi1vP8NARpXVsK2AVOZ4_m58gXMKVQSi_okZVcrLsew1nLizg@mail.gmail.com>
Subject: Re: [PATCH][next] ceph: Replace zero-length array with flexible array member
To:     "Gustavo A. R. Silva" <gustavoars@kernel.org>
Cc:     Jeff Layton <jlayton@kernel.org>,
        "David S. Miller" <davem@davemloft.net>,
        Jakub Kicinski <kuba@kernel.org>,
        Ceph Development <ceph-devel@vger.kernel.org>,
        netdev <netdev@vger.kernel.org>,
        LKML <linux-kernel@vger.kernel.org>,
        linux-hardening@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

On Fri, May 14, 2021 at 11:51 PM Gustavo A. R. Silva
<gustavoars@kernel.org> wrote:
>
> There is a regular need in the kernel to provide a way to declare
> having a dynamically sized set of trailing elements in a structure.
> Kernel code should always use =E2=80=9Cflexible array members=E2=80=9D[1]=
 for these
> cases. The older style of one-element or zero-length arrays should
> no longer be used[2].
>
> Notice that, in this case, sizeof(au->reply_buf) translates to zero,
> becase in the original code reply_buf is a zero-length array. Now that
> reply_buf is transformed into a flexible array, the mentioned line of
> code is now replaced by a literal 0.
>
> Also, as a safeguard, explicitly assign NULL to
> auth->authorizer_reply_buf, as no heap is allocated for it, therefore
> it should not be accessible.
>
> [1] https://en.wikipedia.org/wiki/Flexible_array_member
> [2] https://www.kernel.org/doc/html/v5.10/process/deprecated.html#zero-le=
ngth-and-one-element-arrays
>
> Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org>
> ---
>  net/ceph/auth_none.c | 4 ++--
>  net/ceph/auth_none.h | 2 +-
>  2 files changed, 3 insertions(+), 3 deletions(-)
>
> diff --git a/net/ceph/auth_none.c b/net/ceph/auth_none.c
> index 70e86e462250..10ee16d2cbf0 100644
> --- a/net/ceph/auth_none.c
> +++ b/net/ceph/auth_none.c
> @@ -111,8 +111,8 @@ static int ceph_auth_none_create_authorizer(
>         auth->authorizer =3D (struct ceph_authorizer *) au;
>         auth->authorizer_buf =3D au->buf;
>         auth->authorizer_buf_len =3D au->buf_len;
> -       auth->authorizer_reply_buf =3D au->reply_buf;
> -       auth->authorizer_reply_buf_len =3D sizeof (au->reply_buf);
> +       auth->authorizer_reply_buf_len =3D 0;
> +       auth->authorizer_reply_buf =3D NULL;
>
>         return 0;
>  }
> diff --git a/net/ceph/auth_none.h b/net/ceph/auth_none.h
> index 4158f064302e..3c68c0ee3dab 100644
> --- a/net/ceph/auth_none.h
> +++ b/net/ceph/auth_none.h
> @@ -16,7 +16,7 @@ struct ceph_none_authorizer {
>         struct ceph_authorizer base;
>         char buf[128];
>         int buf_len;
> -       char reply_buf[0];
> +       char reply_buf[];
>  };
>
>  struct ceph_auth_none_info {

Hi Gustavo,

I went ahead and removed reply_buf.  We never receive authorizer
replies in auth_none mode, so patching it to be a flexible array
is rather pointless.

Thanks,

                Ilya