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;
References: <1587819040-38793-1-git-send-email-xiyuyang19@fudan.edu.cn>
User-agent: mu4e 1.1.0; emacs 26.3
From:   Jakub Sitnicki <jakub@cloudflare.com>
To:     Xiyu Yang <xiyuyang19@fudan.edu.cn>
Cc:     John Fastabend <john.fastabend@gmail.com>,
        Daniel Borkmann <daniel@iogearbox.net>,
        Lorenz Bauer <lmb@cloudflare.com>,
        Eric Dumazet <edumazet@google.com>,
        "David S. Miller" <davem@davemloft.net>,
        Alexey Kuznetsov <kuznet@ms2.inr.ac.ru>,
        Hideaki YOSHIFUJI <yoshfuji@linux-ipv6.org>,
        Jakub Kicinski <kuba@kernel.org>,
        Alexei Starovoitov <ast@kernel.org>,
        Martin KaFai Lau <kafai@fb.com>,
        Song Liu <songliubraving@fb.com>, Yonghong Song <yhs@fb.com>,
        Andrii Nakryiko <andriin@fb.com>,
        KP Singh <kpsingh@chromium.org>, netdev@vger.kernel.org,
        bpf@vger.kernel.org, linux-kernel@vger.kernel.org,
        yuanxzhang@fudan.edu.cn, kjlu@umn.edu,
        Xin Tan <tanxin.ctf@gmail.com>
Subject: Re: [PATCH] bpf: Fix sk_psock refcnt leak when receiving message
In-reply-to: <1587819040-38793-1-git-send-email-xiyuyang19@fudan.edu.cn>
Date:   Sat, 25 Apr 2020 16:24:44 +0200
Message-ID: <87lfmjve1f.fsf@cloudflare.com>
MIME-Version: 1.0
Content-Type: text/plain
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Sat, Apr 25, 2020 at 02:50 PM CEST, Xiyu Yang wrote:
> tcp_bpf_recvmsg() invokes sk_psock_get(), which returns a reference of
> the specified sk_psock object to "psock" with increased refcnt.
>
> When tcp_bpf_recvmsg() returns, local variable "psock" becomes invalid,
> so the refcount should be decreased to keep refcount balanced.
>
> The reference counting issue happens in several exception handling paths
> of tcp_bpf_recvmsg(). When those error scenarios occur such as "flags"
> includes MSG_ERRQUEUE, the function forgets to decrease the refcnt
> increased by sk_psock_get(), causing a refcnt leak.
>
> Fix this issue by calling sk_psock_put() when those error scenarios
> occur.
>
> Signed-off-by: Xiyu Yang <xiyuyang19@fudan.edu.cn>
> Signed-off-by: Xin Tan <tanxin.ctf@gmail.com>
> ---
>  net/ipv4/tcp_bpf.c | 8 ++++++--
>  1 file changed, 6 insertions(+), 2 deletions(-)
>
> diff --git a/net/ipv4/tcp_bpf.c b/net/ipv4/tcp_bpf.c
> index 5a05327f97c1..feb6b90672c1 100644
> --- a/net/ipv4/tcp_bpf.c
> +++ b/net/ipv4/tcp_bpf.c
> @@ -265,11 +265,15 @@ static int tcp_bpf_recvmsg(struct sock *sk, struct msghdr *msg, size_t len,
>  	psock = sk_psock_get(sk);
>  	if (unlikely(!psock))
>  		return tcp_recvmsg(sk, msg, len, nonblock, flags, addr_len);
> -	if (unlikely(flags & MSG_ERRQUEUE))
> +	if (unlikely(flags & MSG_ERRQUEUE)) {
> +		sk_psock_put(sk, psock);
>  		return inet_recv_error(sk, msg, len, addr_len);
> +	}
>  	if (!skb_queue_empty(&sk->sk_receive_queue) &&
> -	    sk_psock_queue_empty(psock))
> +	    sk_psock_queue_empty(psock)) {
> +		sk_psock_put(sk, psock);
>  		return tcp_recvmsg(sk, msg, len, nonblock, flags, addr_len);
> +	}
>  	lock_sock(sk);
>  msg_bytes_ready:
>  	copied = __tcp_bpf_recvmsg(sk, psock, msg, len, flags);

Thanks for the fix.

We can pull up the error queue read handling, that is the `flags &
MSG_ERRQUEUE` branch, so that it happens before we grab a psock ref.

The effect is the same because now, if we hit the !psock branch,
tcp_recvmsg will first check if user wants to read the error queue
anyway.

That would translate to something like below, in addition to your
changes.

WDYT?

---8<---

diff --git a/net/ipv4/tcp_bpf.c b/net/ipv4/tcp_bpf.c
index 5a05327f97c1..99aa57bd1901 100644
--- a/net/ipv4/tcp_bpf.c
+++ b/net/ipv4/tcp_bpf.c
@@ -262,14 +262,17 @@ static int tcp_bpf_recvmsg(struct sock *sk, struct msghdr *msg, size_t len,
 	struct sk_psock *psock;
 	int copied, ret;

+	if (unlikely(flags & MSG_ERRQUEUE))
+		return inet_recv_error(sk, msg, len, addr_len);
+
 	psock = sk_psock_get(sk);
 	if (unlikely(!psock))
 		return tcp_recvmsg(sk, msg, len, nonblock, flags, addr_len);
-	if (unlikely(flags & MSG_ERRQUEUE))
-		return inet_recv_error(sk, msg, len, addr_len);
 	if (!skb_queue_empty(&sk->sk_receive_queue) &&
 	    sk_psock_queue_empty(psock))
		return tcp_recvmsg(sk, msg, len, nonblock, flags, addr_len);