diff options
| author | Jakub Sitnicki <jakub@cloudflare.com> | 2020-02-18 17:10:13 +0000 |
|---|---|---|
| committer | Daniel Borkmann <daniel@iogearbox.net> | 2020-02-21 22:29:45 +0100 |
| commit | b8e202d1d1d0f182f01062804efb523ea9a9008c (patch) | |
| tree | 57a1c1d075e2d470fd9c1487744286c2f3bed747 /net/ipv4/tcp_bpf.c | |
| parent | e42da4c62abb547d9c9138e0e7fcd1f36057b5e8 (diff) | |
net, sk_msg: Annotate lockless access to sk_prot on clone
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.
Signed-off-by: Jakub Sitnicki <jakub@cloudflare.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20200218171023.844439-2-jakub@cloudflare.com
Diffstat (limited to 'net/ipv4/tcp_bpf.c')
| -rw-r--r-- | net/ipv4/tcp_bpf.c | 4 |
1 files changed, 3 insertions, 1 deletions
diff --git a/net/ipv4/tcp_bpf.c b/net/ipv4/tcp_bpf.c index 8a01428f80c1..dd183b050642 100644 --- a/net/ipv4/tcp_bpf.c +++ b/net/ipv4/tcp_bpf.c @@ -645,8 +645,10 @@ static void tcp_bpf_reinit_sk_prot(struct sock *sk, struct sk_psock *psock) /* Reinit occurs when program types change e.g. TCP_BPF_TX is removed * or added requiring sk_prot hook updates. We keep original saved * hooks in this case. + * + * Pairs with lockless read in sk_clone_lock(). */ - sk->sk_prot = &tcp_bpf_prots[family][config]; + WRITE_ONCE(sk->sk_prot, &tcp_bpf_prots[family][config]); } static int tcp_bpf_assert_proto_ops(struct proto *ops) |