Received-SPF: pass (google.com: domain of linux-kernel+bounces-71747-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.48.161 as permitted sender) client-ip=147.75.48.161;
Precedence: bulk
MIME-Version: 1.0
References: <20240215-upstream-net-20240215-misc-fixes-v1-0-8c01a55d8f6a@kernel.org>
 <20240215-upstream-net-20240215-misc-fixes-v1-3-8c01a55d8f6a@kernel.org>
In-Reply-To: <20240215-upstream-net-20240215-misc-fixes-v1-3-8c01a55d8f6a@kernel.org>
From: Eric Dumazet <edumazet@google.com>
Date: Mon, 19 Feb 2024 18:21:52 +0100
Message-ID: <CANn89iJ=Oecw6OZDwmSYc9HJKQ_G32uN11L+oUcMu+TOD5Xiaw@mail.gmail.com>
Subject: Re: [PATCH net 03/13] mptcp: fix lockless access in subflow ULP diag
To: "Matthieu Baerts (NGI0)" <matttbe@kernel.org>
Cc: mptcp@lists.linux.dev, Mat Martineau <martineau@kernel.org>, 
	Geliang Tang <geliang@kernel.org>, "David S. Miller" <davem@davemloft.net>, 
	Jakub Kicinski <kuba@kernel.org>, Paolo Abeni <pabeni@redhat.com>, 
	Davide Caratti <dcaratti@redhat.com>, Shuah Khan <shuah@kernel.org>, netdev@vger.kernel.org, 
	linux-kernel@vger.kernel.org, linux-kselftest@vger.kernel.org, 
	stable@vger.kernel.org, Boris Pismenny <borisp@nvidia.com>, 
	John Fastabend <john.fastabend@gmail.com>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

On Thu, Feb 15, 2024 at 7:25=E2=80=AFPM Matthieu Baerts (NGI0)
<matttbe@kernel.org> wrote:
>
> From: Paolo Abeni <pabeni@redhat.com>
>
> Since the introduction of the subflow ULP diag interface, the
> dump callback accessed all the subflow data with lockless.
>
> We need either to annotate all the read and write operation accordingly,
> or acquire the subflow socket lock. Let's do latter, even if slower, to
> avoid a diffstat havoc.
>
> Fixes: 5147dfb50832 ("mptcp: allow dumping subflow context to userspace")
> Cc: stable@vger.kernel.org
> Signed-off-by: Paolo Abeni <pabeni@redhat.com>
> Reviewed-by: Mat Martineau <martineau@kernel.org>
> Signed-off-by: Matthieu Baerts (NGI0) <matttbe@kernel.org>
> ---
> Notes:
>   - This patch modifies the existing ULP API. No better solutions have
>     been found for -net, and there is some similar prior art, see
>     commit 0df48c26d841 ("tcp: add tcpi_bytes_acked to tcp_info").
>
>     Please also note that TLS ULP Diag has likely the same issue.
> To: Boris Pismenny <borisp@nvidia.com>
> To: John Fastabend <john.fastabend@gmail.com>
> ---
>  include/net/tcp.h  | 2 +-
>  net/mptcp/diag.c   | 6 +++++-
>  net/tls/tls_main.c | 2 +-
>  3 files changed, 7 insertions(+), 3 deletions(-)
>
> diff --git a/include/net/tcp.h b/include/net/tcp.h
> index dd78a1181031..f6eba9652d01 100644
> --- a/include/net/tcp.h
> +++ b/include/net/tcp.h
> @@ -2506,7 +2506,7 @@ struct tcp_ulp_ops {
>         /* cleanup ulp */
>         void (*release)(struct sock *sk);
>         /* diagnostic */
> -       int (*get_info)(const struct sock *sk, struct sk_buff *skb);
> +       int (*get_info)(struct sock *sk, struct sk_buff *skb);
>         size_t (*get_info_size)(const struct sock *sk);
>         /* clone ulp */
>         void (*clone)(const struct request_sock *req, struct sock *newsk,
> diff --git a/net/mptcp/diag.c b/net/mptcp/diag.c
> index a536586742f2..e57c5f47f035 100644
> --- a/net/mptcp/diag.c
> +++ b/net/mptcp/diag.c
> @@ -13,17 +13,19 @@
>  #include <uapi/linux/mptcp.h>
>  #include "protocol.h"
>
> -static int subflow_get_info(const struct sock *sk, struct sk_buff *skb)
> +static int subflow_get_info(struct sock *sk, struct sk_buff *skb)
>  {
>         struct mptcp_subflow_context *sf;
>         struct nlattr *start;
>         u32 flags =3D 0;
> +       bool slow;
>         int err;
>
>         start =3D nla_nest_start_noflag(skb, INET_ULP_INFO_MPTCP);
>         if (!start)
>                 return -EMSGSIZE;
>
> +       slow =3D lock_sock_fast(sk);
>         rcu_read_lock();

I am afraid lockdep is not happy with this change.

Paolo, we probably need the READ_ONCE() annotations after all.