summaryrefslogtreecommitdiffstats
path: root/net/ipv4/tcp_cong.c
diff options
context:
space:
mode:
authorYuchung Cheng <ycheng@google.com>2013-10-31 11:07:31 -0700
committerDavid S. Miller <davem@davemloft.net>2013-11-04 19:57:59 -0500
commit9f9843a751d0a2057f9f3d313886e7e5e6ebaac9 (patch)
treea89df5cc0c5f5280b2cfffba7f6933e4db20736f /net/ipv4/tcp_cong.c
parent0d41cca490c274352211efac50e9598d39a9dc80 (diff)
downloadlinux-9f9843a751d0a2057f9f3d313886e7e5e6ebaac9.tar.gz
linux-9f9843a751d0a2057f9f3d313886e7e5e6ebaac9.tar.bz2
linux-9f9843a751d0a2057f9f3d313886e7e5e6ebaac9.zip
tcp: properly handle stretch acks in slow start
Slow start now increases cwnd by 1 if an ACK acknowledges some packets, regardless the number of packets. Consequently slow start performance is highly dependent on the degree of the stretch ACKs caused by receiver or network ACK compression mechanisms (e.g., delayed-ACK, GRO, etc). But slow start algorithm is to send twice the amount of packets of packets left so it should process a stretch ACK of degree N as if N ACKs of degree 1, then exits when cwnd exceeds ssthresh. A follow up patch will use the remainder of the N (if greater than 1) to adjust cwnd in the congestion avoidance phase. In addition this patch retires the experimental limited slow start (LSS) feature. LSS has multiple drawbacks but questionable benefit. The fractional cwnd increase in LSS requires a loop in slow start even though it's rarely used. Configuring such an increase step via a global sysctl on different BDPS seems hard. Finally and most importantly the slow start overshoot concern is now better covered by the Hybrid slow start (hystart) enabled by default. Signed-off-by: Yuchung Cheng <ycheng@google.com> Signed-off-by: Neal Cardwell <ncardwell@google.com> Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Diffstat (limited to 'net/ipv4/tcp_cong.c')
-rw-r--r--net/ipv4/tcp_cong.c47
1 files changed, 17 insertions, 30 deletions
diff --git a/net/ipv4/tcp_cong.c b/net/ipv4/tcp_cong.c
index 019c2389a341..ad37bf18ae4b 100644
--- a/net/ipv4/tcp_cong.c
+++ b/net/ipv4/tcp_cong.c
@@ -15,8 +15,6 @@
#include <linux/gfp.h>
#include <net/tcp.h>
-int sysctl_tcp_max_ssthresh = 0;
-
static DEFINE_SPINLOCK(tcp_cong_list_lock);
static LIST_HEAD(tcp_cong_list);
@@ -299,35 +297,24 @@ bool tcp_is_cwnd_limited(const struct sock *sk, u32 in_flight)
}
EXPORT_SYMBOL_GPL(tcp_is_cwnd_limited);
-/*
- * Slow start is used when congestion window is less than slow start
- * threshold. This version implements the basic RFC2581 version
- * and optionally supports:
- * RFC3742 Limited Slow Start - growth limited to max_ssthresh
- * RFC3465 Appropriate Byte Counting - growth limited by bytes acknowledged
+/* Slow start is used when congestion window is no greater than the slow start
+ * threshold. We base on RFC2581 and also handle stretch ACKs properly.
+ * We do not implement RFC3465 Appropriate Byte Counting (ABC) per se but
+ * something better;) a packet is only considered (s)acked in its entirety to
+ * defend the ACK attacks described in the RFC. Slow start processes a stretch
+ * ACK of degree N as if N acks of degree 1 are received back to back except
+ * ABC caps N to 2. Slow start exits when cwnd grows over ssthresh and
+ * returns the leftover acks to adjust cwnd in congestion avoidance mode.
*/
-void tcp_slow_start(struct tcp_sock *tp)
+int tcp_slow_start(struct tcp_sock *tp, u32 acked)
{
- int cnt; /* increase in packets */
- unsigned int delta = 0;
- u32 snd_cwnd = tp->snd_cwnd;
-
- if (unlikely(!snd_cwnd)) {
- pr_err_once("snd_cwnd is nul, please report this bug.\n");
- snd_cwnd = 1U;
- }
+ u32 cwnd = tp->snd_cwnd + acked;
- if (sysctl_tcp_max_ssthresh > 0 && tp->snd_cwnd > sysctl_tcp_max_ssthresh)
- cnt = sysctl_tcp_max_ssthresh >> 1; /* limited slow start */
- else
- cnt = snd_cwnd; /* exponential increase */
-
- tp->snd_cwnd_cnt += cnt;
- while (tp->snd_cwnd_cnt >= snd_cwnd) {
- tp->snd_cwnd_cnt -= snd_cwnd;
- delta++;
- }
- tp->snd_cwnd = min(snd_cwnd + delta, tp->snd_cwnd_clamp);
+ if (cwnd > tp->snd_ssthresh)
+ cwnd = tp->snd_ssthresh + 1;
+ acked -= cwnd - tp->snd_cwnd;
+ tp->snd_cwnd = min(cwnd, tp->snd_cwnd_clamp);
+ return acked;
}
EXPORT_SYMBOL_GPL(tcp_slow_start);
@@ -351,7 +338,7 @@ EXPORT_SYMBOL_GPL(tcp_cong_avoid_ai);
/* This is Jacobson's slow start and congestion avoidance.
* SIGCOMM '88, p. 328.
*/
-void tcp_reno_cong_avoid(struct sock *sk, u32 ack, u32 in_flight)
+void tcp_reno_cong_avoid(struct sock *sk, u32 ack, u32 acked, u32 in_flight)
{
struct tcp_sock *tp = tcp_sk(sk);
@@ -360,7 +347,7 @@ void tcp_reno_cong_avoid(struct sock *sk, u32 ack, u32 in_flight)
/* In "safe" area, increase. */
if (tp->snd_cwnd <= tp->snd_ssthresh)
- tcp_slow_start(tp);
+ tcp_slow_start(tp, acked);
/* In dangerous area, increase slowly. */
else
tcp_cong_avoid_ai(tp, tp->snd_cwnd);