From nobody Sat May 4 07:53:01 2024 Received: from us-smtp-delivery-124.mimecast.com (us-smtp-delivery-124.mimecast.com [170.10.133.124]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 4E9C03C32 for ; Tue, 11 Oct 2022 17:40:51 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=redhat.com; s=mimecast20190719; t=1665510050; h=from:from:reply-to:subject:subject:date:date:message-id:message-id: to:to:cc:mime-version:mime-version:content-type:content-type: content-transfer-encoding:content-transfer-encoding; bh=CK3nBHCByVkW/XjTaL1M0B4m+lRChvMPwYy2OA7B7rY=; b=ZwZ/7x5J5mTDKSa5gTM3+ZCEk/Rp+eLPO87328Z6Yomas0xEolehQ2SoCCj4erUmID8sY1 xny1UPwoWIMGgW1g4+MNbJM6Yk5XeROk0HjEYTmJxmBOtQbjO+knGSVrOCzkFYsWmAs9ET QsIXEpE4sGnFG4X8Opauclpg5MGilyY= Received: from mimecast-mx02.redhat.com (mx3-rdu2.redhat.com [66.187.233.73]) by relay.mimecast.com with ESMTP with STARTTLS (version=TLSv1.2, cipher=TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384) id us-mta-150-Cxmuw3C8OA-CUUUjbtjBmQ-1; Tue, 11 Oct 2022 13:40:49 -0400 X-MC-Unique: Cxmuw3C8OA-CUUUjbtjBmQ-1 Received: from smtp.corp.redhat.com (int-mx10.intmail.prod.int.rdu2.redhat.com [10.11.54.10]) (using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits)) (No client certificate requested) by mimecast-mx02.redhat.com (Postfix) with ESMTPS id D116F1C05AEB for ; Tue, 11 Oct 2022 17:40:48 +0000 (UTC) Received: from gerbillo.redhat.com (unknown [10.39.195.27]) by smtp.corp.redhat.com (Postfix) with ESMTP id 60F3D492B05 for ; Tue, 11 Oct 2022 17:40:48 +0000 (UTC) From: Paolo Abeni To: mptcp@lists.linux.dev Subject: [PATCH mptcp-net v2] mptcp: fix abba deadlock on fastopen Date: Tue, 11 Oct 2022 19:40:42 +0200 Message-Id: Precedence: bulk X-Mailing-List: mptcp@lists.linux.dev List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 X-Scanned-By: MIMEDefang 3.1 on 10.11.54.10 X-Mimecast-Spam-Score: 0 X-Mimecast-Originator: redhat.com Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8"; x-default="true" Our CI reported lockdep splat in the fastopen code: =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D WARNING: possible circular locking dependency detected 6.0.0.mptcp_f5e8bfe9878d+ #1558 Not tainted ------------------------------------------------------ packetdrill/1071 is trying to acquire lock: ffff8881bd198140 (sk_lock-AF_INET){+.+.}-{0:0}, at: inet_wait_for_connect+= 0x19c/0x310 but task is already holding lock: ffff8881b8346540 (k-sk_lock-AF_INET){+.+.}-{0:0}, at: mptcp_sendmsg+0xfdf/= 0x1740 which lock already depends on the new lock. the existing dependency chain (in reverse order) is: -> #1 (k-sk_lock-AF_INET){+.+.}-{0:0}: __lock_acquire+0xb6d/0x1860 lock_acquire+0x1d8/0x620 lock_sock_nested+0x37/0xd0 inet_stream_connect+0x3f/0xa0 mptcp_connect+0x411/0x800 __inet_stream_connect+0x3ab/0x800 mptcp_stream_connect+0xac/0x110 __sys_connect+0x101/0x130 __x64_sys_connect+0x6e/0xb0 do_syscall_64+0x59/0x90 entry_SYSCALL_64_after_hwframe+0x63/0xcd -> #0 (sk_lock-AF_INET){+.+.}-{0:0}: check_prev_add+0x15e/0x2110 validate_chain+0xace/0xdf0 __lock_acquire+0xb6d/0x1860 lock_acquire+0x1d8/0x620 lock_sock_nested+0x37/0xd0 inet_wait_for_connect+0x19c/0x310 __inet_stream_connect+0x26c/0x800 tcp_sendmsg_fastopen+0x341/0x650 mptcp_sendmsg+0x109d/0x1740 sock_sendmsg+0xe1/0x120 __sys_sendto+0x1c7/0x2a0 __x64_sys_sendto+0xdc/0x1b0 do_syscall_64+0x59/0x90 entry_SYSCALL_64_after_hwframe+0x63/0xcd other info that might help us debug this: Possible unsafe locking scenario: CPU0 CPU1 ---- ---- lock(k-sk_lock-AF_INET); lock(sk_lock-AF_INET); lock(k-sk_lock-AF_INET); lock(sk_lock-AF_INET); *** DEADLOCK *** 1 lock held by packetdrill/1071: #0: ffff8881b8346540 (k-sk_lock-AF_INET){+.+.}-{0:0}, at: mptcp_sendmsg+0= xfdf/0x1740 =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D The problem is caused by the blocking inet_wait_for_connect() releasing and re-acquiring the msk socket lock while the subflow socket lock is still held and the MPTCP socket requires that the msk socket lock must be acquired before the subflow socket lock. Address the issue always invoking tcp_sendmsg_fastopen() in an unblocking manner, and later eventually complete the blocking __inet_stream_connect() as needed. Fixes: d98a82a6afc7 ("mptcp: handle defer connect in mptcp_sendmsg") Signed-off-by: Paolo Abeni --- v1 -> v2: - really invoke fastopen in unblocking manner - clear copied_syn on completion errors commit ("mptcp: factor out mptcp_connect()") will need some mangling on top of this one. A possible alternative would be targeting even the above to -net. Not strictly required, but it will also make more striaght-forward/ legit the __inet_stream_connect() invocation here. --- net/mptcp/protocol.c | 43 ++++++++++++++++++++++++++++++------------- 1 file changed, 30 insertions(+), 13 deletions(-) diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c index f599ad44ed24..21f5676a60e8 100644 --- a/net/mptcp/protocol.c +++ b/net/mptcp/protocol.c @@ -1673,6 +1673,31 @@ static void mptcp_set_nospace(struct sock *sk) set_bit(MPTCP_NOSPACE, &mptcp_sk(sk)->flags); } =20 +static int mptcp_sendmsg_fastopen(struct sock *sk, struct sock *ssk, struc= t msghdr *msg, + size_t len, int *copied_syn) +{ + unsigned int saved_flags =3D msg->msg_flags; + int ret; + + lock_sock(ssk); + msg->msg_flags |=3D MSG_DONTWAIT; + ret =3D tcp_sendmsg_fastopen(ssk, msg, copied_syn, len, NULL); + msg->msg_flags =3D saved_flags; + release_sock(ssk); + + /* do the blocking bits of inet_stream_connect outside the ssk socket loc= k*/ + if (ret =3D=3D -EINPROGRESS && !(msg->msg_flags & MSG_DONTWAIT)) { + ret =3D __inet_stream_connect(sk->sk_socket, msg->msg_name, + msg->msg_namelen, msg->msg_flags, 1); + + /* On connect hard error fastopen no bytes actually sent. */ + if (ret && ret !=3D -EINPROGRESS && ret !=3D -ERESTARTSYS && ret !=3D -E= INTR) + *copied_syn =3D 0; + } + + return ret; +} + static int mptcp_sendmsg(struct sock *sk, struct msghdr *msg, size_t len) { struct mptcp_sock *msk =3D mptcp_sk(sk); @@ -1692,24 +1717,16 @@ static int mptcp_sendmsg(struct sock *sk, struct ms= ghdr *msg, size_t len) lock_sock(sk); =20 ssock =3D __mptcp_nmpc_socket(msk); - if (unlikely(ssock && inet_sk(ssock->sk)->defer_connect)) { - struct sock *ssk =3D ssock->sk; + if (unlikely(ssock && (inet_sk(ssock->sk)->defer_connect || + msg->msg_flags & MSG_FASTOPEN))) { int copied_syn =3D 0; =20 - lock_sock(ssk); - - ret =3D tcp_sendmsg_fastopen(ssk, msg, &copied_syn, len, NULL); + ret =3D mptcp_sendmsg_fastopen(sk, ssock->sk, msg, len, &copied_syn); copied +=3D copied_syn; - if (ret =3D=3D -EINPROGRESS && copied_syn > 0) { - /* reflect the new state on the MPTCP socket */ - inet_sk_state_store(sk, inet_sk_state_load(ssk)); - release_sock(ssk); + if (ret =3D=3D -EINPROGRESS && copied_syn > 0) goto out; - } else if (ret) { - release_sock(ssk); + else if (ret) goto do_error; - } - release_sock(ssk); } =20 timeo =3D sock_sndtimeo(sk, msg->msg_flags & MSG_DONTWAIT); --=20 2.37.3