From a158bdd3247b9656df36ba133235fff702e9fdc3 Mon Sep 17 00:00:00 2001 From: David Howells Date: Fri, 24 Nov 2017 10:18:41 +0000 Subject: rxrpc: Fix call timeouts Fix the rxrpc call expiration timeouts and make them settable from userspace. By analogy with other rx implementations, there should be three timeouts: (1) "Normal timeout" This is set for all calls and is triggered if we haven't received any packets from the peer in a while. It is measured from the last time we received any packet on that call. This is not reset by any connection packets (such as CHALLENGE/RESPONSE packets). If a service operation takes a long time, the server should generate PING ACKs at a duration that's substantially less than the normal timeout so is to keep both sides alive. This is set at 1/6 of normal timeout. (2) "Idle timeout" This is set only for a service call and is triggered if we stop receiving the DATA packets that comprise the request data. It is measured from the last time we received a DATA packet. (3) "Hard timeout" This can be set for a call and specified the maximum lifetime of that call. It should not be specified by default. Some operations (such as volume transfer) take a long time. Allow userspace to set/change the timeouts on a call with sendmsg, using a control message: RXRPC_SET_CALL_TIMEOUTS The data to the message is a number of 32-bit words, not all of which need be given: u32 hard_timeout; /* sec from first packet */ u32 idle_timeout; /* msec from packet Rx */ u32 normal_timeout; /* msec from data Rx */ This can be set in combination with any other sendmsg() that affects a call. Signed-off-by: David Howells --- net/rxrpc/call_event.c | 179 +++++++++++++++++++++++-------------------------- 1 file changed, 84 insertions(+), 95 deletions(-) (limited to 'net/rxrpc/call_event.c') diff --git a/net/rxrpc/call_event.c b/net/rxrpc/call_event.c index 3574508baf9a..c14395d5ad8c 100644 --- a/net/rxrpc/call_event.c +++ b/net/rxrpc/call_event.c @@ -21,80 +21,6 @@ #include #include "ar-internal.h" -/* - * Set the timer - */ -void __rxrpc_set_timer(struct rxrpc_call *call, enum rxrpc_timer_trace why, - ktime_t now) -{ - unsigned long t_j, now_j = jiffies; - ktime_t t; - bool queue = false; - - if (call->state < RXRPC_CALL_COMPLETE) { - t = call->expire_at; - if (!ktime_after(t, now)) { - trace_rxrpc_timer(call, why, now, now_j); - queue = true; - goto out; - } - - if (!ktime_after(call->resend_at, now)) { - call->resend_at = call->expire_at; - if (!test_and_set_bit(RXRPC_CALL_EV_RESEND, &call->events)) - queue = true; - } else if (ktime_before(call->resend_at, t)) { - t = call->resend_at; - } - - if (!ktime_after(call->ack_at, now)) { - call->ack_at = call->expire_at; - if (!test_and_set_bit(RXRPC_CALL_EV_ACK, &call->events)) - queue = true; - } else if (ktime_before(call->ack_at, t)) { - t = call->ack_at; - } - - if (!ktime_after(call->ping_at, now)) { - call->ping_at = call->expire_at; - if (!test_and_set_bit(RXRPC_CALL_EV_PING, &call->events)) - queue = true; - } else if (ktime_before(call->ping_at, t)) { - t = call->ping_at; - } - - t_j = nsecs_to_jiffies(ktime_to_ns(ktime_sub(t, now))); - t_j += jiffies; - - /* We have to make sure that the calculated jiffies value falls - * at or after the nsec value, or we may loop ceaselessly - * because the timer times out, but we haven't reached the nsec - * timeout yet. - */ - t_j++; - - if (call->timer.expires != t_j || !timer_pending(&call->timer)) { - mod_timer(&call->timer, t_j); - trace_rxrpc_timer(call, why, now, now_j); - } - } - -out: - if (queue) - rxrpc_queue_call(call); -} - -/* - * Set the timer - */ -void rxrpc_set_timer(struct rxrpc_call *call, enum rxrpc_timer_trace why, - ktime_t now) -{ - read_lock_bh(&call->state_lock); - __rxrpc_set_timer(call, why, now); - read_unlock_bh(&call->state_lock); -} - /* * Propose a PING ACK be sent. */ @@ -106,12 +32,13 @@ static void rxrpc_propose_ping(struct rxrpc_call *call, !test_and_set_bit(RXRPC_CALL_EV_PING, &call->events)) rxrpc_queue_call(call); } else { - ktime_t now = ktime_get_real(); - ktime_t ping_at = ktime_add_ms(now, rxrpc_idle_ack_delay); + unsigned long now = jiffies; + unsigned long ping_at = now + rxrpc_idle_ack_delay; - if (ktime_before(ping_at, call->ping_at)) { - call->ping_at = ping_at; - rxrpc_set_timer(call, rxrpc_timer_set_for_ping, now); + if (time_before(ping_at, call->ping_at)) { + WRITE_ONCE(call->ping_at, ping_at); + rxrpc_reduce_call_timer(call, ping_at, now, + rxrpc_timer_set_for_ping); } } } @@ -125,8 +52,7 @@ static void __rxrpc_propose_ACK(struct rxrpc_call *call, u8 ack_reason, enum rxrpc_propose_ack_trace why) { enum rxrpc_propose_ack_outcome outcome = rxrpc_propose_ack_use; - unsigned int expiry = rxrpc_soft_ack_delay; - ktime_t now, ack_at; + unsigned long now, ack_at, expiry = rxrpc_soft_ack_delay; s8 prior = rxrpc_ack_priority[ack_reason]; /* Pings are handled specially because we don't want to accidentally @@ -190,11 +116,12 @@ static void __rxrpc_propose_ACK(struct rxrpc_call *call, u8 ack_reason, background) rxrpc_queue_call(call); } else { - now = ktime_get_real(); - ack_at = ktime_add_ms(now, expiry); - if (ktime_before(ack_at, call->ack_at)) { - call->ack_at = ack_at; - rxrpc_set_timer(call, rxrpc_timer_set_for_ack, now); + now = jiffies; + ack_at = jiffies + expiry; + if (time_before(ack_at, call->ack_at)) { + WRITE_ONCE(call->ack_at, ack_at); + rxrpc_reduce_call_timer(call, ack_at, now, + rxrpc_timer_set_for_ack); } } @@ -227,18 +154,20 @@ static void rxrpc_congestion_timeout(struct rxrpc_call *call) /* * Perform retransmission of NAK'd and unack'd packets. */ -static void rxrpc_resend(struct rxrpc_call *call, ktime_t now) +static void rxrpc_resend(struct rxrpc_call *call, unsigned long now_j) { struct rxrpc_skb_priv *sp; struct sk_buff *skb; + unsigned long resend_at; rxrpc_seq_t cursor, seq, top; - ktime_t max_age, oldest, ack_ts; + ktime_t now, max_age, oldest, ack_ts; int ix; u8 annotation, anno_type, retrans = 0, unacked = 0; _enter("{%d,%d}", call->tx_hard_ack, call->tx_top); - max_age = ktime_sub_ms(now, rxrpc_resend_timeout); + now = ktime_get_real(); + max_age = ktime_sub_ms(now, rxrpc_resend_timeout * 1000 / HZ); spin_lock_bh(&call->lock); @@ -282,7 +211,9 @@ static void rxrpc_resend(struct rxrpc_call *call, ktime_t now) ktime_to_ns(ktime_sub(skb->tstamp, max_age))); } - call->resend_at = ktime_add_ms(oldest, rxrpc_resend_timeout); + resend_at = nsecs_to_jiffies(ktime_to_ns(ktime_sub(oldest, now))); + resend_at += jiffies + rxrpc_resend_timeout; + WRITE_ONCE(call->resend_at, resend_at); if (unacked) rxrpc_congestion_timeout(call); @@ -292,7 +223,8 @@ static void rxrpc_resend(struct rxrpc_call *call, ktime_t now) * retransmitting data. */ if (!retrans) { - rxrpc_set_timer(call, rxrpc_timer_set_for_resend, now); + rxrpc_reduce_call_timer(call, resend_at, now, + rxrpc_timer_set_for_resend); spin_unlock_bh(&call->lock); ack_ts = ktime_sub(now, call->acks_latest_ts); if (ktime_to_ns(ack_ts) < call->peer->rtt) @@ -364,7 +296,7 @@ void rxrpc_process_call(struct work_struct *work) { struct rxrpc_call *call = container_of(work, struct rxrpc_call, processor); - ktime_t now; + unsigned long now, next, t; rxrpc_see_call(call); @@ -384,8 +316,50 @@ recheck_state: goto out_put; } - now = ktime_get_real(); - if (ktime_before(call->expire_at, now)) { + /* Work out if any timeouts tripped */ + now = jiffies; + t = READ_ONCE(call->expect_rx_by); + if (time_after_eq(now, t)) { + trace_rxrpc_timer(call, rxrpc_timer_exp_normal, now); + set_bit(RXRPC_CALL_EV_EXPIRED, &call->events); + } + + t = READ_ONCE(call->expect_req_by); + if (call->state == RXRPC_CALL_SERVER_RECV_REQUEST && + time_after_eq(now, t)) { + trace_rxrpc_timer(call, rxrpc_timer_exp_idle, now); + set_bit(RXRPC_CALL_EV_EXPIRED, &call->events); + } + + t = READ_ONCE(call->expect_term_by); + if (time_after_eq(now, t)) { + trace_rxrpc_timer(call, rxrpc_timer_exp_hard, now); + set_bit(RXRPC_CALL_EV_EXPIRED, &call->events); + } + + t = READ_ONCE(call->ack_at); + if (time_after_eq(now, t)) { + trace_rxrpc_timer(call, rxrpc_timer_exp_ack, now); + cmpxchg(&call->ack_at, t, now + MAX_JIFFY_OFFSET); + set_bit(RXRPC_CALL_EV_ACK, &call->events); + } + + t = READ_ONCE(call->ping_at); + if (time_after_eq(now, t)) { + trace_rxrpc_timer(call, rxrpc_timer_exp_ping, now); + cmpxchg(&call->ping_at, t, now + MAX_JIFFY_OFFSET); + set_bit(RXRPC_CALL_EV_PING, &call->events); + } + + t = READ_ONCE(call->resend_at); + if (time_after_eq(now, t)) { + trace_rxrpc_timer(call, rxrpc_timer_exp_resend, now); + cmpxchg(&call->resend_at, t, now + MAX_JIFFY_OFFSET); + set_bit(RXRPC_CALL_EV_RESEND, &call->events); + } + + /* Process events */ + if (test_and_clear_bit(RXRPC_CALL_EV_EXPIRED, &call->events)) { rxrpc_abort_call("EXP", call, 0, RX_USER_ABORT, -ETIME); set_bit(RXRPC_CALL_EV_ABORT, &call->events); goto recheck_state; @@ -408,7 +382,22 @@ recheck_state: goto recheck_state; } - rxrpc_set_timer(call, rxrpc_timer_set_for_resend, now); + /* Make sure the timer is restarted */ + next = call->expect_rx_by; + +#define set(T) { t = READ_ONCE(T); if (time_before(t, next)) next = t; } + + set(call->expect_req_by); + set(call->expect_term_by); + set(call->ack_at); + set(call->resend_at); + set(call->ping_at); + + now = jiffies; + if (time_after_eq(now, next)) + goto recheck_state; + + rxrpc_reduce_call_timer(call, next, now, rxrpc_timer_restart); /* other events may have been raised since we started checking */ if (call->events && call->state < RXRPC_CALL_COMPLETE) { -- cgit From beb8e5e4f38cc3e4c2839cfc143e0312bf53d0e0 Mon Sep 17 00:00:00 2001 From: David Howells Date: Fri, 24 Nov 2017 10:18:41 +0000 Subject: rxrpc: Express protocol timeouts in terms of RTT Express protocol timeouts for data retransmission and deferred ack generation in terms on RTT rather than specified timeouts once we have sufficient RTT samples. For the moment, this requires just one RTT sample to be able to use this for ack deferral and two for data retransmission. The data retransmission timeout is set at RTT*1.5 and the ACK deferral timeout is set at RTT. Note that the calculated timeout is limited to a minimum of 4ns to make sure it doesn't happen too quickly. Signed-off-by: David Howells --- net/rxrpc/call_event.c | 22 ++++++++++++++++++---- 1 file changed, 18 insertions(+), 4 deletions(-) (limited to 'net/rxrpc/call_event.c') diff --git a/net/rxrpc/call_event.c b/net/rxrpc/call_event.c index c14395d5ad8c..da91f16ac77c 100644 --- a/net/rxrpc/call_event.c +++ b/net/rxrpc/call_event.c @@ -52,7 +52,7 @@ static void __rxrpc_propose_ACK(struct rxrpc_call *call, u8 ack_reason, enum rxrpc_propose_ack_trace why) { enum rxrpc_propose_ack_outcome outcome = rxrpc_propose_ack_use; - unsigned long now, ack_at, expiry = rxrpc_soft_ack_delay; + unsigned long expiry = rxrpc_soft_ack_delay; s8 prior = rxrpc_ack_priority[ack_reason]; /* Pings are handled specially because we don't want to accidentally @@ -116,7 +116,13 @@ static void __rxrpc_propose_ACK(struct rxrpc_call *call, u8 ack_reason, background) rxrpc_queue_call(call); } else { - now = jiffies; + unsigned long now = jiffies, ack_at; + + if (call->peer->rtt_usage > 0) + ack_at = nsecs_to_jiffies(call->peer->rtt); + else + ack_at = expiry; + ack_at = jiffies + expiry; if (time_before(ack_at, call->ack_at)) { WRITE_ONCE(call->ack_at, ack_at); @@ -160,14 +166,22 @@ static void rxrpc_resend(struct rxrpc_call *call, unsigned long now_j) struct sk_buff *skb; unsigned long resend_at; rxrpc_seq_t cursor, seq, top; - ktime_t now, max_age, oldest, ack_ts; + ktime_t now, max_age, oldest, ack_ts, timeout, min_timeo; int ix; u8 annotation, anno_type, retrans = 0, unacked = 0; _enter("{%d,%d}", call->tx_hard_ack, call->tx_top); + if (call->peer->rtt_usage > 1) + timeout = ns_to_ktime(call->peer->rtt * 3 / 2); + else + timeout = ms_to_ktime(rxrpc_resend_timeout); + min_timeo = ns_to_ktime((1000000000 / HZ) * 4); + if (ktime_before(timeout, min_timeo)) + timeout = min_timeo; + now = ktime_get_real(); - max_age = ktime_sub_ms(now, rxrpc_resend_timeout * 1000 / HZ); + max_age = ktime_sub(now, timeout); spin_lock_bh(&call->lock); -- cgit From bd1fdf8cfdf3fdbccd2b21c33ec649ebd7429af7 Mon Sep 17 00:00:00 2001 From: David Howells Date: Fri, 24 Nov 2017 10:18:42 +0000 Subject: rxrpc: Add a timeout for detecting lost ACKs/lost DATA Add an extra timeout that is set/updated when we send a DATA packet that has the request-ack flag set. This allows us to detect if we don't get an ACK in response to the latest flagged packet. The ACK packet is adjudged to have been lost if it doesn't turn up within 2*RTT of the transmission. If the timeout occurs, we schedule the sending of a PING ACK to find out the state of the other side. If a new DATA packet is ready to go sooner, we cancel the sending of the ping and set the request-ack flag on that instead. If we get back a PING-RESPONSE ACK that indicates a lower tx_top than what we had at the time of the ping transmission, we adjudge all the DATA packets sent between the response tx_top and the ping-time tx_top to have been lost and retransmit immediately. Rather than sending a PING ACK, we could just pick a DATA packet and speculatively retransmit that with request-ack set. It should result in either a REQUESTED ACK or a DUPLICATE ACK which we can then use in lieu the a PING-RESPONSE ACK mentioned above. Signed-off-by: David Howells --- net/rxrpc/call_event.c | 26 ++++++++++++++++++++++---- 1 file changed, 22 insertions(+), 4 deletions(-) (limited to 'net/rxrpc/call_event.c') diff --git a/net/rxrpc/call_event.c b/net/rxrpc/call_event.c index da91f16ac77c..c65666b2f39e 100644 --- a/net/rxrpc/call_event.c +++ b/net/rxrpc/call_event.c @@ -245,7 +245,7 @@ static void rxrpc_resend(struct rxrpc_call *call, unsigned long now_j) goto out; rxrpc_propose_ACK(call, RXRPC_ACK_PING, 0, 0, true, false, rxrpc_propose_ack_ping_for_lost_ack); - rxrpc_send_ack_packet(call, true); + rxrpc_send_ack_packet(call, true, NULL); goto out; } @@ -310,6 +310,7 @@ void rxrpc_process_call(struct work_struct *work) { struct rxrpc_call *call = container_of(work, struct rxrpc_call, processor); + rxrpc_serial_t *send_ack; unsigned long now, next, t; rxrpc_see_call(call); @@ -358,6 +359,13 @@ recheck_state: set_bit(RXRPC_CALL_EV_ACK, &call->events); } + t = READ_ONCE(call->ack_lost_at); + if (time_after_eq(now, t)) { + trace_rxrpc_timer(call, rxrpc_timer_exp_lost_ack, now); + cmpxchg(&call->ack_lost_at, t, now + MAX_JIFFY_OFFSET); + set_bit(RXRPC_CALL_EV_ACK_LOST, &call->events); + } + t = READ_ONCE(call->ping_at); if (time_after_eq(now, t)) { trace_rxrpc_timer(call, rxrpc_timer_exp_ping, now); @@ -379,15 +387,24 @@ recheck_state: goto recheck_state; } - if (test_and_clear_bit(RXRPC_CALL_EV_ACK, &call->events)) { + send_ack = NULL; + if (test_and_clear_bit(RXRPC_CALL_EV_ACK_LOST, &call->events)) { + call->acks_lost_top = call->tx_top; + rxrpc_propose_ACK(call, RXRPC_ACK_PING, 0, 0, true, false, + rxrpc_propose_ack_ping_for_lost_ack); + send_ack = &call->acks_lost_ping; + } + + if (test_and_clear_bit(RXRPC_CALL_EV_ACK, &call->events) || + send_ack) { if (call->ackr_reason) { - rxrpc_send_ack_packet(call, false); + rxrpc_send_ack_packet(call, false, send_ack); goto recheck_state; } } if (test_and_clear_bit(RXRPC_CALL_EV_PING, &call->events)) { - rxrpc_send_ack_packet(call, true); + rxrpc_send_ack_packet(call, true, NULL); goto recheck_state; } @@ -404,6 +421,7 @@ recheck_state: set(call->expect_req_by); set(call->expect_term_by); set(call->ack_at); + set(call->ack_lost_at); set(call->resend_at); set(call->ping_at); -- cgit From 415f44e43282a16ec0808c7ccfd401762e587437 Mon Sep 17 00:00:00 2001 From: David Howells Date: Fri, 24 Nov 2017 10:18:42 +0000 Subject: rxrpc: Add keepalive for a call We need to transmit a packet every so often to act as a keepalive for the peer (which has a timeout from the last time it received a packet) and also to prevent any intervening firewalls from closing the route. Do this by resetting a timer every time we transmit a packet. If the timer ever expires, we transmit a PING ACK packet and thereby also elicit a PING RESPONSE ACK from the other side - which prevents our last-rx timeout from expiring. The timer is set to 1/6 of the last-rx timeout so that we can detect the other side going away if it misses 6 replies in a row. This is particularly necessary for servers where the processing of the service function may take a significant amount of time. Signed-off-by: David Howells --- net/rxrpc/call_event.c | 10 ++++++++++ 1 file changed, 10 insertions(+) (limited to 'net/rxrpc/call_event.c') diff --git a/net/rxrpc/call_event.c b/net/rxrpc/call_event.c index c65666b2f39e..bda952ffe6a6 100644 --- a/net/rxrpc/call_event.c +++ b/net/rxrpc/call_event.c @@ -366,6 +366,15 @@ recheck_state: set_bit(RXRPC_CALL_EV_ACK_LOST, &call->events); } + t = READ_ONCE(call->keepalive_at); + if (time_after_eq(now, t)) { + trace_rxrpc_timer(call, rxrpc_timer_exp_keepalive, now); + cmpxchg(&call->keepalive_at, t, now + MAX_JIFFY_OFFSET); + rxrpc_propose_ACK(call, RXRPC_ACK_PING, 0, 0, true, true, + rxrpc_propose_ack_ping_for_keepalive); + set_bit(RXRPC_CALL_EV_PING, &call->events); + } + t = READ_ONCE(call->ping_at); if (time_after_eq(now, t)) { trace_rxrpc_timer(call, rxrpc_timer_exp_ping, now); @@ -423,6 +432,7 @@ recheck_state: set(call->ack_at); set(call->ack_lost_at); set(call->resend_at); + set(call->keepalive_at); set(call->ping_at); now = jiffies; -- cgit