From 6bec041147a2a64a490d1f813e8a004443061b38 Mon Sep 17 00:00:00 2001 From: Paolo Abeni Date: Sat, 16 Sep 2023 12:52:45 +0200 Subject: [PATCH 1/5] mptcp: fix bogus receive window shrinkage with multiple subflows In case multiple subflows race to update the mptcp-level receive window, the subflow losing the race should use the window value provided by the "winning" subflow to update it's own tcp-level rcv_wnd. To such goal, the current code bogusly uses the mptcp-level rcv_wnd value as observed before the update attempt. On unlucky circumstances that may lead to TCP-level window shrinkage, and stall the other end. Address the issue feeding to the rcv wnd update the correct value. Fixes: f3589be0c420 ("mptcp: never shrink offered window") Cc: stable@vger.kernel.org Closes: https://github.com/multipath-tcp/mptcp_net-next/issues/427 Signed-off-by: Paolo Abeni Reviewed-by: Mat Martineau Signed-off-by: Matthieu Baerts Signed-off-by: David S. Miller --- net/mptcp/options.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/net/mptcp/options.c b/net/mptcp/options.c index c254accb14de..cd15ec73073e 100644 --- a/net/mptcp/options.c +++ b/net/mptcp/options.c @@ -1269,12 +1269,13 @@ static void mptcp_set_rwin(struct tcp_sock *tp, struct tcphdr *th) if (rcv_wnd == rcv_wnd_old) break; - if (before64(rcv_wnd_new, rcv_wnd)) { + + rcv_wnd_old = rcv_wnd; + if (before64(rcv_wnd_new, rcv_wnd_old)) { MPTCP_INC_STATS(sock_net(ssk), MPTCP_MIB_RCVWNDCONFLICTUPDATE); goto raise_win; } MPTCP_INC_STATS(sock_net(ssk), MPTCP_MIB_RCVWNDCONFLICT); - rcv_wnd_old = rcv_wnd; } return; } From d5fbeff1ab812b6c473b6924bee8748469462e2c Mon Sep 17 00:00:00 2001 From: Paolo Abeni Date: Sat, 16 Sep 2023 12:52:46 +0200 Subject: [PATCH 2/5] mptcp: move __mptcp_error_report in protocol.c This will simplify the next patch ("mptcp: process pending subflow error on close"). No functional change intended. Cc: stable@vger.kernel.org # v5.12+ Signed-off-by: Paolo Abeni Reviewed-by: Mat Martineau Signed-off-by: Matthieu Baerts Signed-off-by: David S. Miller --- net/mptcp/protocol.c | 36 ++++++++++++++++++++++++++++++++++++ net/mptcp/subflow.c | 36 ------------------------------------ 2 files changed, 36 insertions(+), 36 deletions(-) diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c index a7fc16f5175d..915860027b1a 100644 --- a/net/mptcp/protocol.c +++ b/net/mptcp/protocol.c @@ -770,6 +770,42 @@ static bool __mptcp_ofo_queue(struct mptcp_sock *msk) return moved; } +void __mptcp_error_report(struct sock *sk) +{ + struct mptcp_subflow_context *subflow; + struct mptcp_sock *msk = mptcp_sk(sk); + + mptcp_for_each_subflow(msk, subflow) { + struct sock *ssk = mptcp_subflow_tcp_sock(subflow); + int err = sock_error(ssk); + int ssk_state; + + if (!err) + continue; + + /* only propagate errors on fallen-back sockets or + * on MPC connect + */ + if (sk->sk_state != TCP_SYN_SENT && !__mptcp_check_fallback(msk)) + continue; + + /* We need to propagate only transition to CLOSE state. + * Orphaned socket will see such state change via + * subflow_sched_work_if_closed() and that path will properly + * destroy the msk as needed. + */ + ssk_state = inet_sk_state_load(ssk); + if (ssk_state == TCP_CLOSE && !sock_flag(sk, SOCK_DEAD)) + inet_sk_state_store(sk, ssk_state); + WRITE_ONCE(sk->sk_err, -err); + + /* This barrier is coupled with smp_rmb() in mptcp_poll() */ + smp_wmb(); + sk_error_report(sk); + break; + } +} + /* In most cases we will be able to lock the mptcp socket. If its already * owned, we need to defer to the work queue to avoid ABBA deadlock. */ diff --git a/net/mptcp/subflow.c b/net/mptcp/subflow.c index 9bf3c7bc1762..2f40c23fdb0d 100644 --- a/net/mptcp/subflow.c +++ b/net/mptcp/subflow.c @@ -1362,42 +1362,6 @@ void mptcp_space(const struct sock *ssk, int *space, int *full_space) *full_space = mptcp_win_from_space(sk, READ_ONCE(sk->sk_rcvbuf)); } -void __mptcp_error_report(struct sock *sk) -{ - struct mptcp_subflow_context *subflow; - struct mptcp_sock *msk = mptcp_sk(sk); - - mptcp_for_each_subflow(msk, subflow) { - struct sock *ssk = mptcp_subflow_tcp_sock(subflow); - int err = sock_error(ssk); - int ssk_state; - - if (!err) - continue; - - /* only propagate errors on fallen-back sockets or - * on MPC connect - */ - if (sk->sk_state != TCP_SYN_SENT && !__mptcp_check_fallback(msk)) - continue; - - /* We need to propagate only transition to CLOSE state. - * Orphaned socket will see such state change via - * subflow_sched_work_if_closed() and that path will properly - * destroy the msk as needed. - */ - ssk_state = inet_sk_state_load(ssk); - if (ssk_state == TCP_CLOSE && !sock_flag(sk, SOCK_DEAD)) - inet_sk_state_store(sk, ssk_state); - WRITE_ONCE(sk->sk_err, -err); - - /* This barrier is coupled with smp_rmb() in mptcp_poll() */ - smp_wmb(); - sk_error_report(sk); - break; - } -} - static void subflow_error_report(struct sock *ssk) { struct sock *sk = mptcp_subflow_ctx(ssk)->conn; From 9f1a98813b4b686482e5ef3c9d998581cace0ba6 Mon Sep 17 00:00:00 2001 From: Paolo Abeni Date: Sat, 16 Sep 2023 12:52:47 +0200 Subject: [PATCH 3/5] mptcp: process pending subflow error on close On incoming TCP reset, subflow closing could happen before error propagation. That in turn could cause the socket error being ignored, and a missing socket state transition, as reported by Daire-Byrne. Address the issues explicitly checking for subflow socket error at close time. To avoid code duplication, factor-out of __mptcp_error_report() a new helper implementing the relevant bits. Closes: https://github.com/multipath-tcp/mptcp_net-next/issues/429 Fixes: 15cc10453398 ("mptcp: deliver ssk errors to msk") Cc: stable@vger.kernel.org Signed-off-by: Paolo Abeni Reviewed-by: Mat Martineau Signed-off-by: Matthieu Baerts Signed-off-by: David S. Miller --- net/mptcp/protocol.c | 63 ++++++++++++++++++++++++-------------------- 1 file changed, 34 insertions(+), 29 deletions(-) diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c index 915860027b1a..1c96b8da71df 100644 --- a/net/mptcp/protocol.c +++ b/net/mptcp/protocol.c @@ -770,40 +770,44 @@ static bool __mptcp_ofo_queue(struct mptcp_sock *msk) return moved; } +static bool __mptcp_subflow_error_report(struct sock *sk, struct sock *ssk) +{ + int err = sock_error(ssk); + int ssk_state; + + if (!err) + return false; + + /* only propagate errors on fallen-back sockets or + * on MPC connect + */ + if (sk->sk_state != TCP_SYN_SENT && !__mptcp_check_fallback(mptcp_sk(sk))) + return false; + + /* We need to propagate only transition to CLOSE state. + * Orphaned socket will see such state change via + * subflow_sched_work_if_closed() and that path will properly + * destroy the msk as needed. + */ + ssk_state = inet_sk_state_load(ssk); + if (ssk_state == TCP_CLOSE && !sock_flag(sk, SOCK_DEAD)) + inet_sk_state_store(sk, ssk_state); + WRITE_ONCE(sk->sk_err, -err); + + /* This barrier is coupled with smp_rmb() in mptcp_poll() */ + smp_wmb(); + sk_error_report(sk); + return true; +} + void __mptcp_error_report(struct sock *sk) { struct mptcp_subflow_context *subflow; struct mptcp_sock *msk = mptcp_sk(sk); - mptcp_for_each_subflow(msk, subflow) { - struct sock *ssk = mptcp_subflow_tcp_sock(subflow); - int err = sock_error(ssk); - int ssk_state; - - if (!err) - continue; - - /* only propagate errors on fallen-back sockets or - * on MPC connect - */ - if (sk->sk_state != TCP_SYN_SENT && !__mptcp_check_fallback(msk)) - continue; - - /* We need to propagate only transition to CLOSE state. - * Orphaned socket will see such state change via - * subflow_sched_work_if_closed() and that path will properly - * destroy the msk as needed. - */ - ssk_state = inet_sk_state_load(ssk); - if (ssk_state == TCP_CLOSE && !sock_flag(sk, SOCK_DEAD)) - inet_sk_state_store(sk, ssk_state); - WRITE_ONCE(sk->sk_err, -err); - - /* This barrier is coupled with smp_rmb() in mptcp_poll() */ - smp_wmb(); - sk_error_report(sk); - break; - } + mptcp_for_each_subflow(msk, subflow) + if (__mptcp_subflow_error_report(sk, mptcp_subflow_tcp_sock(subflow))) + break; } /* In most cases we will be able to lock the mptcp socket. If its already @@ -2428,6 +2432,7 @@ static void __mptcp_close_ssk(struct sock *sk, struct sock *ssk, } out_release: + __mptcp_subflow_error_report(sk, ssk); release_sock(ssk); sock_put(ssk); From f6909dc1c1f4452879278128012da6c76bc186a5 Mon Sep 17 00:00:00 2001 From: Paolo Abeni Date: Sat, 16 Sep 2023 12:52:48 +0200 Subject: [PATCH 4/5] mptcp: rename timer related helper to less confusing names The msk socket uses to different timeout to track close related events and retransmissions. The existing helpers do not indicate clearly which timer they actually touch, making the related code quite confusing. Change the existing helpers name to avoid such confusion. No functional change intended. This patch is linked to the next one ("mptcp: fix dangling connection hang-up"). The two patches are supposed to be backported together. Cc: stable@vger.kernel.org # v5.11+ Signed-off-by: Paolo Abeni Reviewed-by: Matthieu Baerts Reviewed-by: Mat Martineau Signed-off-by: Matthieu Baerts Signed-off-by: David S. Miller --- net/mptcp/protocol.c | 42 +++++++++++++++++++++--------------------- net/mptcp/protocol.h | 2 +- net/mptcp/subflow.c | 2 +- 3 files changed, 23 insertions(+), 23 deletions(-) diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c index 1c96b8da71df..c8f38f303a90 100644 --- a/net/mptcp/protocol.c +++ b/net/mptcp/protocol.c @@ -405,7 +405,7 @@ static bool __mptcp_move_skb(struct mptcp_sock *msk, struct sock *ssk, return false; } -static void mptcp_stop_timer(struct sock *sk) +static void mptcp_stop_rtx_timer(struct sock *sk) { struct inet_connection_sock *icsk = inet_csk(sk); @@ -911,12 +911,12 @@ static void __mptcp_flush_join_list(struct sock *sk, struct list_head *join_list } } -static bool mptcp_timer_pending(struct sock *sk) +static bool mptcp_rtx_timer_pending(struct sock *sk) { return timer_pending(&inet_csk(sk)->icsk_retransmit_timer); } -static void mptcp_reset_timer(struct sock *sk) +static void mptcp_reset_rtx_timer(struct sock *sk) { struct inet_connection_sock *icsk = inet_csk(sk); unsigned long tout; @@ -1050,10 +1050,10 @@ static void __mptcp_clean_una(struct sock *sk) out: if (snd_una == READ_ONCE(msk->snd_nxt) && snd_una == READ_ONCE(msk->write_seq)) { - if (mptcp_timer_pending(sk) && !mptcp_data_fin_enabled(msk)) - mptcp_stop_timer(sk); + if (mptcp_rtx_timer_pending(sk) && !mptcp_data_fin_enabled(msk)) + mptcp_stop_rtx_timer(sk); } else { - mptcp_reset_timer(sk); + mptcp_reset_rtx_timer(sk); } } @@ -1626,8 +1626,8 @@ void __mptcp_push_pending(struct sock *sk, unsigned int flags) mptcp_push_release(ssk, &info); /* ensure the rtx timer is running */ - if (!mptcp_timer_pending(sk)) - mptcp_reset_timer(sk); + if (!mptcp_rtx_timer_pending(sk)) + mptcp_reset_rtx_timer(sk); if (do_check_data_fin) mptcp_check_send_data_fin(sk); } @@ -1690,8 +1690,8 @@ static void __mptcp_subflow_push_pending(struct sock *sk, struct sock *ssk, bool if (copied) { tcp_push(ssk, 0, info.mss_now, tcp_sk(ssk)->nonagle, info.size_goal); - if (!mptcp_timer_pending(sk)) - mptcp_reset_timer(sk); + if (!mptcp_rtx_timer_pending(sk)) + mptcp_reset_rtx_timer(sk); if (msk->snd_data_fin_enable && msk->snd_nxt + 1 == msk->write_seq) @@ -2260,7 +2260,7 @@ static void mptcp_retransmit_timer(struct timer_list *t) sock_put(sk); } -static void mptcp_timeout_timer(struct timer_list *t) +static void mptcp_tout_timer(struct timer_list *t) { struct sock *sk = from_timer(sk, t, sk_timer); @@ -2629,14 +2629,14 @@ static void __mptcp_retrans(struct sock *sk) reset_timer: mptcp_check_and_set_pending(sk); - if (!mptcp_timer_pending(sk)) - mptcp_reset_timer(sk); + if (!mptcp_rtx_timer_pending(sk)) + mptcp_reset_rtx_timer(sk); } /* schedule the timeout timer for the relevant event: either close timeout * or mp_fail timeout. The close timeout takes precedence on the mp_fail one */ -void mptcp_reset_timeout(struct mptcp_sock *msk, unsigned long fail_tout) +void mptcp_reset_tout_timer(struct mptcp_sock *msk, unsigned long fail_tout) { struct sock *sk = (struct sock *)msk; unsigned long timeout, close_timeout; @@ -2669,7 +2669,7 @@ static void mptcp_mp_fail_no_response(struct mptcp_sock *msk) WRITE_ONCE(mptcp_subflow_ctx(ssk)->fail_tout, 0); unlock_sock_fast(ssk, slow); - mptcp_reset_timeout(msk, 0); + mptcp_reset_tout_timer(msk, 0); } static void mptcp_do_fastclose(struct sock *sk) @@ -2758,7 +2758,7 @@ static void __mptcp_init_sock(struct sock *sk) /* re-use the csk retrans timer for MPTCP-level retrans */ timer_setup(&msk->sk.icsk_retransmit_timer, mptcp_retransmit_timer, 0); - timer_setup(&sk->sk_timer, mptcp_timeout_timer, 0); + timer_setup(&sk->sk_timer, mptcp_tout_timer, 0); } static void mptcp_ca_reset(struct sock *sk) @@ -2849,8 +2849,8 @@ void mptcp_subflow_shutdown(struct sock *sk, struct sock *ssk, int how) } else { pr_debug("Sending DATA_FIN on subflow %p", ssk); tcp_send_ack(ssk); - if (!mptcp_timer_pending(sk)) - mptcp_reset_timer(sk); + if (!mptcp_rtx_timer_pending(sk)) + mptcp_reset_rtx_timer(sk); } break; } @@ -2933,7 +2933,7 @@ static void __mptcp_destroy_sock(struct sock *sk) might_sleep(); - mptcp_stop_timer(sk); + mptcp_stop_rtx_timer(sk); sk_stop_timer(sk, &sk->sk_timer); msk->pm.status = 0; mptcp_release_sched(msk); @@ -3053,7 +3053,7 @@ bool __mptcp_close(struct sock *sk, long timeout) __mptcp_destroy_sock(sk); do_cancel_work = true; } else { - mptcp_reset_timeout(msk, 0); + mptcp_reset_tout_timer(msk, 0); } return do_cancel_work; @@ -3116,7 +3116,7 @@ static int mptcp_disconnect(struct sock *sk, int flags) mptcp_check_listen_stop(sk); inet_sk_state_store(sk, TCP_CLOSE); - mptcp_stop_timer(sk); + mptcp_stop_rtx_timer(sk); sk_stop_timer(sk, &sk->sk_timer); if (msk->token) diff --git a/net/mptcp/protocol.h b/net/mptcp/protocol.h index 7254b3562575..5e2026815c8e 100644 --- a/net/mptcp/protocol.h +++ b/net/mptcp/protocol.h @@ -718,7 +718,7 @@ void mptcp_get_options(const struct sk_buff *skb, void mptcp_finish_connect(struct sock *sk); void __mptcp_set_connected(struct sock *sk); -void mptcp_reset_timeout(struct mptcp_sock *msk, unsigned long fail_tout); +void mptcp_reset_tout_timer(struct mptcp_sock *msk, unsigned long fail_tout); static inline bool mptcp_is_fully_established(struct sock *sk) { return inet_sk_state_load(sk) == TCP_ESTABLISHED && diff --git a/net/mptcp/subflow.c b/net/mptcp/subflow.c index 2f40c23fdb0d..433f290984c8 100644 --- a/net/mptcp/subflow.c +++ b/net/mptcp/subflow.c @@ -1226,7 +1226,7 @@ static void mptcp_subflow_fail(struct mptcp_sock *msk, struct sock *ssk) WRITE_ONCE(subflow->fail_tout, fail_tout); tcp_send_ack(ssk); - mptcp_reset_timeout(msk, subflow->fail_tout); + mptcp_reset_tout_timer(msk, subflow->fail_tout); } static bool subflow_check_data_avail(struct sock *ssk) From 27e5ccc2d5a50ed61bb73153edb1066104b108b3 Mon Sep 17 00:00:00 2001 From: Paolo Abeni Date: Sat, 16 Sep 2023 12:52:49 +0200 Subject: [PATCH 5/5] mptcp: fix dangling connection hang-up According to RFC 8684 section 3.3: A connection is not closed unless [...] or an implementation-specific connection-level send timeout. Currently the MPTCP protocol does not implement such timeout, and connection timing-out at the TCP-level never move to close state. Introduces a catch-up condition at subflow close time to move the MPTCP socket to close, too. That additionally allows removing similar existing inside the worker. Finally, allow some additional timeout for plain ESTABLISHED mptcp sockets, as the protocol allows creating new subflows even at that point and making the connection functional again. This issue is actually present since the beginning, but it is basically impossible to solve without a long chain of functional pre-requisites topped by commit bbd49d114d57 ("mptcp: consolidate transition to TCP_CLOSE in mptcp_do_fastclose()"). When backporting this current patch, please also backport this other commit as well. Closes: https://github.com/multipath-tcp/mptcp_net-next/issues/430 Fixes: e16163b6e2b7 ("mptcp: refactor shutdown and close") Cc: stable@vger.kernel.org Signed-off-by: Paolo Abeni Reviewed-by: Matthieu Baerts Reviewed-by: Mat Martineau Signed-off-by: Matthieu Baerts Signed-off-by: David S. Miller --- net/mptcp/protocol.c | 86 ++++++++++++++++++++++---------------------- net/mptcp/protocol.h | 22 ++++++++++++ net/mptcp/subflow.c | 1 + 3 files changed, 65 insertions(+), 44 deletions(-) diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c index c8f38f303a90..e252539b1e19 100644 --- a/net/mptcp/protocol.c +++ b/net/mptcp/protocol.c @@ -892,6 +892,7 @@ static bool __mptcp_finish_join(struct mptcp_sock *msk, struct sock *ssk) mptcp_subflow_ctx(ssk)->subflow_id = msk->subflow_id++; mptcp_sockopt_sync_locked(msk, ssk); mptcp_subflow_joined(msk, ssk); + mptcp_stop_tout_timer(sk); return true; } @@ -2369,18 +2370,14 @@ static void __mptcp_close_ssk(struct sock *sk, struct sock *ssk, bool dispose_it, need_push = false; /* If the first subflow moved to a close state before accept, e.g. due - * to an incoming reset, mptcp either: - * - if either the subflow or the msk are dead, destroy the context - * (the subflow socket is deleted by inet_child_forget) and the msk - * - otherwise do nothing at the moment and take action at accept and/or - * listener shutdown - user-space must be able to accept() the closed - * socket. + * to an incoming reset or listener shutdown, the subflow socket is + * already deleted by inet_child_forget() and the mptcp socket can't + * survive too. */ - if (msk->in_accept_queue && msk->first == ssk) { - if (!sock_flag(sk, SOCK_DEAD) && !sock_flag(ssk, SOCK_DEAD)) - return; - + if (msk->in_accept_queue && msk->first == ssk && + (sock_flag(sk, SOCK_DEAD) || sock_flag(ssk, SOCK_DEAD))) { /* ensure later check in mptcp_worker() will dispose the msk */ + mptcp_set_close_tout(sk, tcp_jiffies32 - (TCP_TIMEWAIT_LEN + 1)); sock_set_flag(sk, SOCK_DEAD); lock_sock_nested(ssk, SINGLE_DEPTH_NESTING); mptcp_subflow_drop_ctx(ssk); @@ -2443,6 +2440,22 @@ static void __mptcp_close_ssk(struct sock *sk, struct sock *ssk, out: if (need_push) __mptcp_push_pending(sk, 0); + + /* Catch every 'all subflows closed' scenario, including peers silently + * closing them, e.g. due to timeout. + * For established sockets, allow an additional timeout before closing, + * as the protocol can still create more subflows. + */ + if (list_is_singular(&msk->conn_list) && msk->first && + 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_close_wake_up(sk); + } else { + mptcp_start_tout_timer(sk); + } + } } void mptcp_close_ssk(struct sock *sk, struct sock *ssk, @@ -2486,23 +2499,14 @@ static void __mptcp_close_subflow(struct sock *sk) } -static bool mptcp_should_close(const struct sock *sk) +static bool mptcp_close_tout_expired(const struct sock *sk) { - s32 delta = tcp_jiffies32 - inet_csk(sk)->icsk_mtup.probe_timestamp; - struct mptcp_subflow_context *subflow; + if (!inet_csk(sk)->icsk_mtup.probe_timestamp || + sk->sk_state == TCP_CLOSE) + return false; - if (delta >= TCP_TIMEWAIT_LEN || mptcp_sk(sk)->in_accept_queue) - return true; - - /* if all subflows are in closed status don't bother with additional - * timeout - */ - mptcp_for_each_subflow(mptcp_sk(sk), subflow) { - if (inet_sk_state_load(mptcp_subflow_tcp_sock(subflow)) != - TCP_CLOSE) - return false; - } - return true; + return time_after32(tcp_jiffies32, + inet_csk(sk)->icsk_mtup.probe_timestamp + TCP_TIMEWAIT_LEN); } static void mptcp_check_fastclose(struct mptcp_sock *msk) @@ -2641,15 +2645,16 @@ void mptcp_reset_tout_timer(struct mptcp_sock *msk, unsigned long fail_tout) struct sock *sk = (struct sock *)msk; unsigned long timeout, close_timeout; - if (!fail_tout && !sock_flag(sk, SOCK_DEAD)) + if (!fail_tout && !inet_csk(sk)->icsk_mtup.probe_timestamp) return; - close_timeout = inet_csk(sk)->icsk_mtup.probe_timestamp - tcp_jiffies32 + jiffies + TCP_TIMEWAIT_LEN; + close_timeout = inet_csk(sk)->icsk_mtup.probe_timestamp - tcp_jiffies32 + jiffies + + TCP_TIMEWAIT_LEN; /* the close timeout takes precedence on the fail one, and here at least one of * them is active */ - timeout = sock_flag(sk, SOCK_DEAD) ? close_timeout : fail_tout; + timeout = inet_csk(sk)->icsk_mtup.probe_timestamp ? close_timeout : fail_tout; sk_reset_timer(sk, &sk->sk_timer, timeout); } @@ -2668,8 +2673,6 @@ static void mptcp_mp_fail_no_response(struct mptcp_sock *msk) mptcp_subflow_reset(ssk); WRITE_ONCE(mptcp_subflow_ctx(ssk)->fail_tout, 0); unlock_sock_fast(ssk, slow); - - mptcp_reset_tout_timer(msk, 0); } static void mptcp_do_fastclose(struct sock *sk) @@ -2706,18 +2709,14 @@ static void mptcp_worker(struct work_struct *work) if (test_and_clear_bit(MPTCP_WORK_CLOSE_SUBFLOW, &msk->flags)) __mptcp_close_subflow(sk); - /* There is no point in keeping around an orphaned sk timedout or - * closed, but we need the msk around to reply to incoming DATA_FIN, - * even if it is orphaned and in FIN_WAIT2 state - */ - if (sock_flag(sk, SOCK_DEAD)) { - if (mptcp_should_close(sk)) - mptcp_do_fastclose(sk); + if (mptcp_close_tout_expired(sk)) { + mptcp_do_fastclose(sk); + mptcp_close_wake_up(sk); + } - if (sk->sk_state == TCP_CLOSE) { - __mptcp_destroy_sock(sk); - goto unlock; - } + if (sock_flag(sk, SOCK_DEAD) && sk->sk_state == TCP_CLOSE) { + __mptcp_destroy_sock(sk); + goto unlock; } if (test_and_clear_bit(MPTCP_WORK_RTX, &msk->flags)) @@ -3016,7 +3015,6 @@ bool __mptcp_close(struct sock *sk, long timeout) cleanup: /* orphan all the subflows */ - inet_csk(sk)->icsk_mtup.probe_timestamp = tcp_jiffies32; mptcp_for_each_subflow(msk, subflow) { struct sock *ssk = mptcp_subflow_tcp_sock(subflow); bool slow = lock_sock_fast_nested(ssk); @@ -3053,7 +3051,7 @@ bool __mptcp_close(struct sock *sk, long timeout) __mptcp_destroy_sock(sk); do_cancel_work = true; } else { - mptcp_reset_tout_timer(msk, 0); + mptcp_start_tout_timer(sk); } return do_cancel_work; @@ -3117,7 +3115,7 @@ static int mptcp_disconnect(struct sock *sk, int flags) inet_sk_state_store(sk, TCP_CLOSE); mptcp_stop_rtx_timer(sk); - sk_stop_timer(sk, &sk->sk_timer); + mptcp_stop_tout_timer(sk); if (msk->token) mptcp_event(MPTCP_EVENT_CLOSED, msk, NULL, GFP_KERNEL); diff --git a/net/mptcp/protocol.h b/net/mptcp/protocol.h index 5e2026815c8e..ed61d6850cce 100644 --- a/net/mptcp/protocol.h +++ b/net/mptcp/protocol.h @@ -719,6 +719,28 @@ void mptcp_get_options(const struct sk_buff *skb, void mptcp_finish_connect(struct sock *sk); void __mptcp_set_connected(struct sock *sk); void mptcp_reset_tout_timer(struct mptcp_sock *msk, unsigned long fail_tout); + +static inline void mptcp_stop_tout_timer(struct sock *sk) +{ + if (!inet_csk(sk)->icsk_mtup.probe_timestamp) + return; + + sk_stop_timer(sk, &sk->sk_timer); + inet_csk(sk)->icsk_mtup.probe_timestamp = 0; +} + +static inline void mptcp_set_close_tout(struct sock *sk, unsigned long tout) +{ + /* avoid 0 timestamp, as that means no close timeout */ + inet_csk(sk)->icsk_mtup.probe_timestamp = tout ? : 1; +} + +static inline void mptcp_start_tout_timer(struct sock *sk) +{ + mptcp_set_close_tout(sk, tcp_jiffies32); + mptcp_reset_tout_timer(mptcp_sk(sk), 0); +} + static inline bool mptcp_is_fully_established(struct sock *sk) { return inet_sk_state_load(sk) == TCP_ESTABLISHED && diff --git a/net/mptcp/subflow.c b/net/mptcp/subflow.c index 433f290984c8..918c1a235790 100644 --- a/net/mptcp/subflow.c +++ b/net/mptcp/subflow.c @@ -1552,6 +1552,7 @@ int __mptcp_subflow_connect(struct sock *sk, const struct mptcp_addr_info *loc, mptcp_sock_graft(ssk, sk->sk_socket); iput(SOCK_INODE(sf)); WRITE_ONCE(msk->allow_infinite_fallback, false); + mptcp_stop_tout_timer(sk); return 0; failed_unlink: