[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20200122225752.caxnk5sylskbrzsg@kafai-mbp.dhcp.thefacebook.com>
Date: Wed, 22 Jan 2020 22:57:55 +0000
From: Martin Lau <kafai@...com>
To: Jakub Sitnicki <jakub@...udflare.com>
CC: "bpf@...r.kernel.org" <bpf@...r.kernel.org>,
"netdev@...r.kernel.org" <netdev@...r.kernel.org>,
"kernel-team@...udflare.com" <kernel-team@...udflare.com>,
Alexei Starovoitov <alexei.starovoitov@...il.com>,
Daniel Borkmann <daniel@...earbox.net>,
"John Fastabend" <john.fastabend@...il.com>,
Lorenz Bauer <lmb@...udflare.com>
Subject: Re: [PATCH bpf-next v3 02/12] net, sk_msg: Annotate lockless access
to sk_prot on clone
On Wed, Jan 22, 2020 at 02:05:39PM +0100, Jakub Sitnicki wrote:
> sk_msg and ULP frameworks override protocol callbacks pointer in
> sk->sk_prot, while tcp accesses it locklessly when cloning the listening
> socket, that is with neither sk_lock nor sk_callback_lock held.
>
> Once we enable use of listening sockets with sockmap (and hence sk_msg),
> there will be shared access to sk->sk_prot if socket is getting cloned
> while being inserted/deleted to/from the sockmap from another CPU:
>
> Read side:
>
> tcp_v4_rcv
> sk = __inet_lookup_skb(...)
> tcp_check_req(sk)
> inet_csk(sk)->icsk_af_ops->syn_recv_sock
> tcp_v4_syn_recv_sock
> tcp_create_openreq_child
> inet_csk_clone_lock
> sk_clone_lock
> READ_ONCE(sk->sk_prot)
>
> Write side:
>
> sock_map_ops->map_update_elem
> sock_map_update_elem
> sock_map_update_common
> sock_map_link_no_progs
> tcp_bpf_init
> tcp_bpf_update_sk_prot
> sk_psock_update_proto
> WRITE_ONCE(sk->sk_prot, ops)
>
> sock_map_ops->map_delete_elem
> sock_map_delete_elem
> __sock_map_delete
> sock_map_unref
> sk_psock_put
> sk_psock_drop
> sk_psock_restore_proto
> tcp_update_ulp
> WRITE_ONCE(sk->sk_prot, proto)
>
> Mark the shared access with READ_ONCE/WRITE_ONCE annotations.
Acked-by: Martin KaFai Lau <kafai@...com>
Powered by blists - more mailing lists