summaryrefslogtreecommitdiff
path: root/net/ipv4
diff options
context:
space:
mode:
authorYuchung Cheng <ycheng@google.com>2017-07-20 01:41:26 +0300
committerDavid S. Miller <davem@davemloft.net>2017-07-20 02:14:10 +0300
commitbb4d991a28cc86a2dfbeefeff32911ca9f779c18 (patch)
tree6fd415a7d0b54a9e639067375013fe8f2c3a48b7 /net/ipv4
parentc4b2bf6b4a35348fe6d1eb06928eb68d7b9d99a9 (diff)
downloadlinux-bb4d991a28cc86a2dfbeefeff32911ca9f779c18.tar.xz
tcp: adjust tail loss probe timeout
This patch adjusts the timeout formula to schedule the TCP loss probe (TLP). The previous formula uses 2*SRTT or 1.5*RTT + DelayACKMax if only one packet is in flight. It keeps a lower bound of 10 msec which is too large for short RTT connections (e.g. within a data-center). The new formula = 2*RTT + (inflight == 1 ? 200ms : 2ticks) which performs better for short and fast connections. Signed-off-by: Yuchung Cheng <ycheng@google.com> Signed-off-by: Neal Cardwell <ncardwell@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Diffstat (limited to 'net/ipv4')
-rw-r--r--net/ipv4/tcp_output.c17
-rw-r--r--net/ipv4/tcp_recovery.c2
2 files changed, 11 insertions, 8 deletions
diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c
index 4e985dea1dd2..886d874775df 100644
--- a/net/ipv4/tcp_output.c
+++ b/net/ipv4/tcp_output.c
@@ -2377,7 +2377,6 @@ bool tcp_schedule_loss_probe(struct sock *sk)
struct inet_connection_sock *icsk = inet_csk(sk);
struct tcp_sock *tp = tcp_sk(sk);
u32 timeout, tlp_time_stamp, rto_time_stamp;
- u32 rtt = usecs_to_jiffies(tp->srtt_us >> 3);
/* No consecutive loss probes. */
if (WARN_ON(icsk->icsk_pending == ICSK_TIME_LOSS_PROBE)) {
@@ -2406,15 +2405,19 @@ bool tcp_schedule_loss_probe(struct sock *sk)
tcp_send_head(sk))
return false;
- /* Probe timeout is at least 1.5*rtt + TCP_DELACK_MAX to account
+ /* Probe timeout is 2*rtt. Add minimum RTO to account
* for delayed ack when there's one outstanding packet. If no RTT
* sample is available then probe after TCP_TIMEOUT_INIT.
*/
- timeout = rtt << 1 ? : TCP_TIMEOUT_INIT;
- if (tp->packets_out == 1)
- timeout = max_t(u32, timeout,
- (rtt + (rtt >> 1) + TCP_DELACK_MAX));
- timeout = max_t(u32, timeout, msecs_to_jiffies(10));
+ if (tp->srtt_us) {
+ timeout = usecs_to_jiffies(tp->srtt_us >> 2);
+ if (tp->packets_out == 1)
+ timeout += TCP_RTO_MIN;
+ else
+ timeout += TCP_TIMEOUT_MIN;
+ } else {
+ timeout = TCP_TIMEOUT_INIT;
+ }
/* If RTO is shorter, just schedule TLP in its place. */
tlp_time_stamp = tcp_jiffies32 + timeout;
diff --git a/net/ipv4/tcp_recovery.c b/net/ipv4/tcp_recovery.c
index fe9a493d0208..449cd914d58e 100644
--- a/net/ipv4/tcp_recovery.c
+++ b/net/ipv4/tcp_recovery.c
@@ -113,7 +113,7 @@ void tcp_rack_mark_lost(struct sock *sk)
tp->rack.advanced = 0;
tcp_rack_detect_loss(sk, &timeout);
if (timeout) {
- timeout = usecs_to_jiffies(timeout + TCP_REO_TIMEOUT_MIN);
+ timeout = usecs_to_jiffies(timeout) + TCP_TIMEOUT_MIN;
inet_csk_reset_xmit_timer(sk, ICSK_TIME_REO_TIMEOUT,
timeout, inet_csk(sk)->icsk_rto);
}