diff options
Diffstat (limited to 'net/mptcp/protocol.c')
-rw-r--r-- | net/mptcp/protocol.c | 177 |
1 files changed, 90 insertions, 87 deletions
diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c index a0b8356cd8c5..3ed4709a7509 100644 --- a/net/mptcp/protocol.c +++ b/net/mptcp/protocol.c @@ -55,28 +55,14 @@ static u64 mptcp_wnd_end(const struct mptcp_sock *msk) return READ_ONCE(msk->wnd_end); } -static bool mptcp_is_tcpsk(struct sock *sk) +static const struct proto_ops *mptcp_fallback_tcp_ops(const struct sock *sk) { - struct socket *sock = sk->sk_socket; - - if (unlikely(sk->sk_prot == &tcp_prot)) { - /* we are being invoked after mptcp_accept() has - * accepted a non-mp-capable flow: sk is a tcp_sk, - * not an mptcp one. - * - * Hand the socket over to tcp so all further socket ops - * bypass mptcp. - */ - WRITE_ONCE(sock->ops, &inet_stream_ops); - return true; #if IS_ENABLED(CONFIG_MPTCP_IPV6) - } else if (unlikely(sk->sk_prot == &tcpv6_prot)) { - WRITE_ONCE(sock->ops, &inet6_stream_ops); - return true; + if (sk->sk_prot == &tcpv6_prot) + return &inet6_stream_ops; #endif - } - - return false; + WARN_ON_ONCE(sk->sk_prot != &tcp_prot); + return &inet_stream_ops; } static int __mptcp_socket_create(struct mptcp_sock *msk) @@ -443,11 +429,11 @@ static void mptcp_check_data_fin_ack(struct sock *sk) switch (sk->sk_state) { case TCP_FIN_WAIT1: - inet_sk_state_store(sk, TCP_FIN_WAIT2); + mptcp_set_state(sk, TCP_FIN_WAIT2); break; case TCP_CLOSING: case TCP_LAST_ACK: - inet_sk_state_store(sk, TCP_CLOSE); + mptcp_set_state(sk, TCP_CLOSE); break; } @@ -608,13 +594,13 @@ static bool mptcp_check_data_fin(struct sock *sk) switch (sk->sk_state) { case TCP_ESTABLISHED: - inet_sk_state_store(sk, TCP_CLOSE_WAIT); + mptcp_set_state(sk, TCP_CLOSE_WAIT); break; case TCP_FIN_WAIT1: - inet_sk_state_store(sk, TCP_CLOSING); + mptcp_set_state(sk, TCP_CLOSING); break; case TCP_FIN_WAIT2: - inet_sk_state_store(sk, TCP_CLOSE); + mptcp_set_state(sk, TCP_CLOSE); break; default: /* Other states not expected */ @@ -789,7 +775,7 @@ static bool __mptcp_subflow_error_report(struct sock *sk, struct sock *ssk) */ ssk_state = inet_sk_state_load(ssk); if (ssk_state == TCP_CLOSE && !sock_flag(sk, SOCK_DEAD)) - inet_sk_state_store(sk, ssk_state); + mptcp_set_state(sk, ssk_state); WRITE_ONCE(sk->sk_err, -err); /* This barrier is coupled with smp_rmb() in mptcp_poll() */ @@ -1230,6 +1216,8 @@ static void mptcp_update_infinite_map(struct mptcp_sock *msk, mptcp_do_fallback(ssk); } +#define MPTCP_MAX_GSO_SIZE (GSO_LEGACY_MAX_SIZE - (MAX_TCP_HEADER + 1)) + static int mptcp_sendmsg_frag(struct sock *sk, struct sock *ssk, struct mptcp_data_frag *dfrag, struct mptcp_sendmsg_info *info) @@ -1256,6 +1244,8 @@ static int mptcp_sendmsg_frag(struct sock *sk, struct sock *ssk, return -EAGAIN; /* compute send limit */ + if (unlikely(ssk->sk_gso_max_size > MPTCP_MAX_GSO_SIZE)) + ssk->sk_gso_max_size = MPTCP_MAX_GSO_SIZE; info->mss_now = tcp_send_mss(ssk, &info->size_goal, info->flags); copy = info->size_goal; @@ -2473,7 +2463,7 @@ out: inet_sk_state_load(msk->first) == TCP_CLOSE) { if (sk->sk_state != TCP_ESTABLISHED || msk->in_accept_queue || sock_flag(sk, SOCK_DEAD)) { - inet_sk_state_store(sk, TCP_CLOSE); + mptcp_set_state(sk, TCP_CLOSE); mptcp_close_wake_up(sk); } else { mptcp_start_tout_timer(sk); @@ -2568,7 +2558,7 @@ static void mptcp_check_fastclose(struct mptcp_sock *msk) WRITE_ONCE(sk->sk_err, ECONNRESET); } - inet_sk_state_store(sk, TCP_CLOSE); + mptcp_set_state(sk, TCP_CLOSE); WRITE_ONCE(sk->sk_shutdown, SHUTDOWN_MASK); smp_mb__before_atomic(); /* SHUTDOWN must be visible first */ set_bit(MPTCP_WORK_CLOSE_SUBFLOW, &msk->flags); @@ -2703,7 +2693,7 @@ static void mptcp_do_fastclose(struct sock *sk) struct mptcp_subflow_context *subflow, *tmp; struct mptcp_sock *msk = mptcp_sk(sk); - inet_sk_state_store(sk, TCP_CLOSE); + mptcp_set_state(sk, TCP_CLOSE); mptcp_for_each_subflow_safe(msk, subflow, tmp) __mptcp_close_ssk(sk, mptcp_subflow_tcp_sock(subflow), subflow, MPTCP_CF_FASTCLOSE); @@ -2881,6 +2871,24 @@ void mptcp_subflow_shutdown(struct sock *sk, struct sock *ssk, int how) release_sock(ssk); } +void mptcp_set_state(struct sock *sk, int state) +{ + int oldstate = sk->sk_state; + + switch (state) { + case TCP_ESTABLISHED: + if (oldstate != TCP_ESTABLISHED) + MPTCP_INC_STATS(sock_net(sk), MPTCP_MIB_CURRESTAB); + break; + + default: + if (oldstate == TCP_ESTABLISHED) + MPTCP_DEC_STATS(sock_net(sk), MPTCP_MIB_CURRESTAB); + } + + inet_sk_state_store(sk, state); +} + static const unsigned char new_state[16] = { /* current state: new state: action: */ [0 /* (Invalid) */] = TCP_CLOSE, @@ -2903,7 +2911,7 @@ static int mptcp_close_state(struct sock *sk) int next = (int)new_state[sk->sk_state]; int ns = next & TCP_STATE_MASK; - inet_sk_state_store(sk, ns); + mptcp_set_state(sk, ns); return next & TCP_ACTION_FIN; } @@ -3014,7 +3022,7 @@ bool __mptcp_close(struct sock *sk, long timeout) if ((1 << sk->sk_state) & (TCPF_LISTEN | TCPF_CLOSE)) { mptcp_check_listen_stop(sk); - inet_sk_state_store(sk, TCP_CLOSE); + mptcp_set_state(sk, TCP_CLOSE); goto cleanup; } @@ -3057,7 +3065,7 @@ cleanup: * state, let's not keep resources busy for no reasons */ if (subflows_alive == 0) - inet_sk_state_store(sk, TCP_CLOSE); + mptcp_set_state(sk, TCP_CLOSE); sock_hold(sk); pr_debug("msk=%p state=%d", sk, sk->sk_state); @@ -3123,7 +3131,7 @@ static int mptcp_disconnect(struct sock *sk, int flags) return -EBUSY; mptcp_check_listen_stop(sk); - inet_sk_state_store(sk, TCP_CLOSE); + mptcp_set_state(sk, TCP_CLOSE); mptcp_stop_rtx_timer(sk); mptcp_stop_tout_timer(sk); @@ -3211,7 +3219,7 @@ struct sock *mptcp_sk_clone_init(const struct sock *sk, /* this can't race with mptcp_close(), as the msk is * not yet exposted to user-space */ - inet_sk_state_store(nsk, TCP_ESTABLISHED); + mptcp_set_state(nsk, TCP_ESTABLISHED); /* The msk maintain a ref to each subflow in the connections list */ WRITE_ONCE(msk->first, ssk); @@ -3254,44 +3262,6 @@ void mptcp_rcv_space_init(struct mptcp_sock *msk, const struct sock *ssk) WRITE_ONCE(msk->wnd_end, msk->snd_nxt + tcp_sk(ssk)->snd_wnd); } -static struct sock *mptcp_accept(struct sock *ssk, int flags, int *err, - bool kern) -{ - struct sock *newsk; - - pr_debug("ssk=%p, listener=%p", ssk, mptcp_subflow_ctx(ssk)); - newsk = inet_csk_accept(ssk, flags, err, kern); - if (!newsk) - return NULL; - - pr_debug("newsk=%p, subflow is mptcp=%d", newsk, sk_is_mptcp(newsk)); - if (sk_is_mptcp(newsk)) { - struct mptcp_subflow_context *subflow; - struct sock *new_mptcp_sock; - - subflow = mptcp_subflow_ctx(newsk); - new_mptcp_sock = subflow->conn; - - /* is_mptcp should be false if subflow->conn is missing, see - * subflow_syn_recv_sock() - */ - if (WARN_ON_ONCE(!new_mptcp_sock)) { - tcp_sk(newsk)->is_mptcp = 0; - goto out; - } - - newsk = new_mptcp_sock; - MPTCP_INC_STATS(sock_net(ssk), MPTCP_MIB_MPCAPABLEPASSIVEACK); - } else { - MPTCP_INC_STATS(sock_net(ssk), - MPTCP_MIB_MPCAPABLEPASSIVEFALLBACK); - } - -out: - newsk->sk_kern_sock = kern; - return newsk; -} - void mptcp_destroy_common(struct mptcp_sock *msk, unsigned int flags) { struct mptcp_subflow_context *subflow, *tmp; @@ -3398,11 +3368,12 @@ static void mptcp_release_cb(struct sock *sk) if (__test_and_clear_bit(MPTCP_CLEAN_UNA, &msk->cb_flags)) __mptcp_clean_una_wakeup(sk); if (unlikely(msk->cb_flags)) { - /* be sure to set the current sk state before tacking actions - * depending on sk_state, that is processing MPTCP_ERROR_REPORT + /* be sure to sync the msk state before taking actions + * depending on sk_state (MPTCP_ERROR_REPORT) + * On sk release avoid actions depending on the first subflow */ - if (__test_and_clear_bit(MPTCP_CONNECTED, &msk->cb_flags)) - __mptcp_set_connected(sk); + if (__test_and_clear_bit(MPTCP_SYNC_STATE, &msk->cb_flags) && msk->first) + __mptcp_sync_state(sk, msk->pending_state); if (__test_and_clear_bit(MPTCP_ERROR_REPORT, &msk->cb_flags)) __mptcp_error_report(sk); if (__test_and_clear_bit(MPTCP_SYNC_SNDBUF, &msk->cb_flags)) @@ -3669,7 +3640,7 @@ static int mptcp_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len) if (IS_ERR(ssk)) return PTR_ERR(ssk); - inet_sk_state_store(sk, TCP_SYN_SENT); + mptcp_set_state(sk, TCP_SYN_SENT); subflow = mptcp_subflow_ctx(ssk); #ifdef CONFIG_TCP_MD5SIG /* no MPTCP if MD5SIG is enabled on this socket or we may run out of @@ -3719,7 +3690,7 @@ out: if (unlikely(err)) { /* avoid leaving a dangling token in an unconnected socket */ mptcp_token_destroy(msk); - inet_sk_state_store(sk, TCP_CLOSE); + mptcp_set_state(sk, TCP_CLOSE); return err; } @@ -3734,7 +3705,6 @@ static struct proto mptcp_prot = { .connect = mptcp_connect, .disconnect = mptcp_disconnect, .close = mptcp_close, - .accept = mptcp_accept, .setsockopt = mptcp_setsockopt, .getsockopt = mptcp_getsockopt, .shutdown = mptcp_shutdown, @@ -3809,13 +3779,13 @@ static int mptcp_listen(struct socket *sock, int backlog) goto unlock; } - inet_sk_state_store(sk, TCP_LISTEN); + mptcp_set_state(sk, TCP_LISTEN); sock_set_flag(sk, SOCK_RCU_FREE); lock_sock(ssk); err = __inet_listen_sk(ssk, backlog); release_sock(ssk); - inet_sk_state_store(sk, inet_sk_state_load(ssk)); + mptcp_set_state(sk, inet_sk_state_load(ssk)); if (!err) { sock_prot_inuse_add(sock_net(sk), sk->sk_prot, 1); @@ -3844,18 +3814,36 @@ static int mptcp_stream_accept(struct socket *sock, struct socket *newsock, if (!ssk) return -EINVAL; - newsk = mptcp_accept(ssk, flags, &err, kern); + pr_debug("ssk=%p, listener=%p", ssk, mptcp_subflow_ctx(ssk)); + newsk = inet_csk_accept(ssk, flags, &err, kern); if (!newsk) return err; - lock_sock(newsk); - - __inet_accept(sock, newsock, newsk); - if (!mptcp_is_tcpsk(newsock->sk)) { - struct mptcp_sock *msk = mptcp_sk(newsk); + pr_debug("newsk=%p, subflow is mptcp=%d", newsk, sk_is_mptcp(newsk)); + if (sk_is_mptcp(newsk)) { struct mptcp_subflow_context *subflow; + struct sock *new_mptcp_sock; + + subflow = mptcp_subflow_ctx(newsk); + new_mptcp_sock = subflow->conn; + + /* is_mptcp should be false if subflow->conn is missing, see + * subflow_syn_recv_sock() + */ + if (WARN_ON_ONCE(!new_mptcp_sock)) { + tcp_sk(newsk)->is_mptcp = 0; + goto tcpfallback; + } + + newsk = new_mptcp_sock; + MPTCP_INC_STATS(sock_net(ssk), MPTCP_MIB_MPCAPABLEPASSIVEACK); + + newsk->sk_kern_sock = kern; + lock_sock(newsk); + __inet_accept(sock, newsock, newsk); set_bit(SOCK_CUSTOM_SOCKOPT, &newsock->flags); + msk = mptcp_sk(newsk); msk->in_accept_queue = 0; /* set ssk->sk_socket of accept()ed flows to mptcp socket. @@ -3875,8 +3863,23 @@ static int mptcp_stream_accept(struct socket *sock, struct socket *newsock, __mptcp_close_ssk(newsk, msk->first, mptcp_subflow_ctx(msk->first), 0); if (unlikely(list_is_singular(&msk->conn_list))) - inet_sk_state_store(newsk, TCP_CLOSE); + mptcp_set_state(newsk, TCP_CLOSE); } + } else { + MPTCP_INC_STATS(sock_net(ssk), + MPTCP_MIB_MPCAPABLEPASSIVEFALLBACK); +tcpfallback: + newsk->sk_kern_sock = kern; + lock_sock(newsk); + __inet_accept(sock, newsock, newsk); + /* we are being invoked after accepting a non-mp-capable + * flow: sk is a tcp_sk, not an mptcp one. + * + * Hand the socket over to tcp so all further socket ops + * bypass mptcp. + */ + WRITE_ONCE(newsock->sk->sk_socket->ops, + mptcp_fallback_tcp_ops(newsock->sk)); } release_sock(newsk); |