summaryrefslogtreecommitdiffstats
path: root/include/net/tcp.h
diff options
context:
space:
mode:
authorEric Dumazet <edumazet@google.com>2017-05-16 14:00:14 -0700
committerDavid S. Miller <davem@davemloft.net>2017-05-17 16:06:01 -0400
commit9a568de4818dea9a05af141046bd3e589245ab83 (patch)
tree6f1502edf55ecb7205660d62bd683ebcf912cfea /include/net/tcp.h
parentac9517fcf310327fa3e3b0d8366e4b11236b1b4b (diff)
downloadlinux-9a568de4818dea9a05af141046bd3e589245ab83.tar.bz2
tcp: switch TCP TS option (RFC 7323) to 1ms clock
TCP Timestamps option is defined in RFC 7323 Traditionally on linux, it has been tied to the internal 'jiffies' variable, because it had been a cheap and good enough generator. For TCP flows on the Internet, 1 ms resolution would be much better than 4ms or 10ms (HZ=250 or HZ=100 respectively) For TCP flows in the DC, Google has used usec resolution for more than two years with great success [1] Receive size autotuning (DRS) is indeed more precise and converges faster to optimal window size. This patch converts tp->tcp_mstamp to a plain u64 value storing a 1 usec TCP clock. This choice will allow us to upstream the 1 usec TS option as discussed in IETF 97. [1] https://www.ietf.org/proceedings/97/slides/slides-97-tcpm-tcp-options-for-low-latency-00.pdf Signed-off-by: Eric Dumazet <edumazet@google.com> Acked-by: Soheil Hassas Yeganeh <soheil@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Diffstat (limited to 'include/net/tcp.h')
-rw-r--r--include/net/tcp.h59
1 files changed, 50 insertions, 9 deletions
diff --git a/include/net/tcp.h b/include/net/tcp.h
index 5b2932b8363f..82462db97183 100644
--- a/include/net/tcp.h
+++ b/include/net/tcp.h
@@ -519,7 +519,7 @@ static inline u32 tcp_cookie_time(void)
u32 __cookie_v4_init_sequence(const struct iphdr *iph, const struct tcphdr *th,
u16 *mssp);
__u32 cookie_v4_init_sequence(const struct sk_buff *skb, __u16 *mss);
-__u32 cookie_init_timestamp(struct request_sock *req);
+u64 cookie_init_timestamp(struct request_sock *req);
bool cookie_timestamp_decode(struct tcp_options_received *opt);
bool cookie_ecn_ok(const struct tcp_options_received *opt,
const struct net *net, const struct dst_entry *dst);
@@ -706,14 +706,55 @@ void tcp_send_window_probe(struct sock *sk);
*/
#define tcp_jiffies32 ((u32)jiffies)
-/* Generator for TCP TS option (RFC 7323)
- * Currently tied to 'jiffies' but will soon be driven by 1 ms clock.
+/*
+ * Deliver a 32bit value for TCP timestamp option (RFC 7323)
+ * It is no longer tied to jiffies, but to 1 ms clock.
+ * Note: double check if you want to use tcp_jiffies32 instead of this.
+ */
+#define TCP_TS_HZ 1000
+
+static inline u64 tcp_clock_ns(void)
+{
+ return local_clock();
+}
+
+static inline u64 tcp_clock_us(void)
+{
+ return div_u64(tcp_clock_ns(), NSEC_PER_USEC);
+}
+
+/* This should only be used in contexts where tp->tcp_mstamp is up to date */
+static inline u32 tcp_time_stamp(const struct tcp_sock *tp)
+{
+ return div_u64(tp->tcp_mstamp, USEC_PER_SEC / TCP_TS_HZ);
+}
+
+/* Could use tcp_clock_us() / 1000, but this version uses a single divide */
+static inline u32 tcp_time_stamp_raw(void)
+{
+ return div_u64(tcp_clock_ns(), NSEC_PER_SEC / TCP_TS_HZ);
+}
+
+
+/* Refresh 1us clock of a TCP socket,
+ * ensuring monotically increasing values.
*/
-#define tcp_time_stamp ((__u32)(jiffies))
+static inline void tcp_mstamp_refresh(struct tcp_sock *tp)
+{
+ u64 val = tcp_clock_us();
+
+ if (val > tp->tcp_mstamp)
+ tp->tcp_mstamp = val;
+}
+
+static inline u32 tcp_stamp_us_delta(u64 t1, u64 t0)
+{
+ return max_t(s64, t1 - t0, 0);
+}
static inline u32 tcp_skb_timestamp(const struct sk_buff *skb)
{
- return skb->skb_mstamp.stamp_jiffies;
+ return div_u64(skb->skb_mstamp, USEC_PER_SEC / TCP_TS_HZ);
}
@@ -778,9 +819,9 @@ struct tcp_skb_cb {
/* pkts S/ACKed so far upon tx of skb, incl retrans: */
__u32 delivered;
/* start of send pipeline phase */
- struct skb_mstamp first_tx_mstamp;
+ u64 first_tx_mstamp;
/* when we reached the "delivered" count */
- struct skb_mstamp delivered_mstamp;
+ u64 delivered_mstamp;
} tx; /* only used for outgoing skbs */
union {
struct inet_skb_parm h4;
@@ -896,7 +937,7 @@ struct ack_sample {
* A sample is invalid if "delivered" or "interval_us" is negative.
*/
struct rate_sample {
- struct skb_mstamp prior_mstamp; /* starting timestamp for interval */
+ u64 prior_mstamp; /* starting timestamp for interval */
u32 prior_delivered; /* tp->delivered at "prior_mstamp" */
s32 delivered; /* number of packets delivered over interval */
long interval_us; /* time for tp->delivered to incr "delivered" */
@@ -1862,7 +1903,7 @@ void tcp_init(void);
/* tcp_recovery.c */
extern void tcp_rack_mark_lost(struct sock *sk);
extern void tcp_rack_advance(struct tcp_sock *tp, u8 sacked, u32 end_seq,
- const struct skb_mstamp *xmit_time);
+ u64 xmit_time);
extern void tcp_rack_reo_timeout(struct sock *sk);
/*