[bpf-next] bpf: add mrtt and srtt as BPF_SOCK_OPS_RTT_CB args

Commit Message

Philo Lu April 21, 2024, 4:24 a.m. UTC

Two important arguments in RTT estimation, mrtt and srtt, are passed to
tcp_bpf_rtt(), so that bpf programs get more information about RTT
computation in BPF_SOCK_OPS_RTT_CB.

The difference between bpf_sock_ops->srtt_us and the srtt here is: the
former is an old rtt before update, while srtt passed by tcp_bpf_rtt()
is that after update.

Signed-off-by: Philo Lu <[e-mail]>
---
 include/net/tcp.h              | 4 ++--
 include/uapi/linux/bpf.h       | 2 ++
 net/ipv4/tcp_input.c           | 4 ++--
 tools/include/uapi/linux/bpf.h | 2 ++
 4 files changed, 8 insertions(+), 4 deletions(-)

Comments

Stanislav Fomichev April 22, 2024, 11:09 p.m. UTC | #1

On 04/21, Philo Lu wrote:
> Two important arguments in RTT estimation, mrtt and srtt, are passed to
> tcp_bpf_rtt(), so that bpf programs get more information about RTT
> computation in BPF_SOCK_OPS_RTT_CB.
> 
> The difference between bpf_sock_ops->srtt_us and the srtt here is: the
> former is an old rtt before update, while srtt passed by tcp_bpf_rtt()
> is that after update.

Can you also extend the rtt selftest to exercise there new numbers?
Something simple like making sure they are non-zero should be enough.

Philo Lu April 23, 2024, 12:58 a.m. UTC | #2

On 2024/4/23 07:09, Stanislav Fomichev wrote:
> On 04/21, Philo Lu wrote:
>> Two important arguments in RTT estimation, mrtt and srtt, are passed to
>> tcp_bpf_rtt(), so that bpf programs get more information about RTT
>> computation in BPF_SOCK_OPS_RTT_CB.
>>
>> The difference between bpf_sock_ops->srtt_us and the srtt here is: the
>> former is an old rtt before update, while srtt passed by tcp_bpf_rtt()
>> is that after update.
> 
> Can you also extend the rtt selftest to exercise there new numbers?
> Something simple like making sure they are non-zero should be enough.

Of course. I will add it in next version.

Thansk.

Patch

diff --git a/include/net/tcp.h b/include/net/tcp.h
index 6ae35199d3b3c..0f75d03287c25 100644
--- a/include/net/tcp.h
+++ b/include/net/tcp.h
@@ -2706,10 +2706,10 @@  static inline bool tcp_bpf_ca_needs_ecn(struct sock *sk)
 	return (tcp_call_bpf(sk, BPF_SOCK_OPS_NEEDS_ECN, 0, NULL) == 1);
 }
 
-static inline void tcp_bpf_rtt(struct sock *sk)
+static inline void tcp_bpf_rtt(struct sock *sk, long mrtt, u32 srtt)
 {
 	if (BPF_SOCK_OPS_TEST_FLAG(tcp_sk(sk), BPF_SOCK_OPS_RTT_CB_FLAG))
-		tcp_call_bpf(sk, BPF_SOCK_OPS_RTT_CB, 0, NULL);
+		tcp_call_bpf_2arg(sk, BPF_SOCK_OPS_RTT_CB, mrtt, srtt);
 }
 
 #if IS_ENABLED(CONFIG_SMC)
diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h
index cee0a7915c08a..d80bef9bbdc15 100644
--- a/include/uapi/linux/bpf.h
+++ b/include/uapi/linux/bpf.h
@@ -6947,6 +6947,8 @@  enum {
 					 * socket transition to LISTEN state.
 					 */
 	BPF_SOCK_OPS_RTT_CB,		/* Called on every RTT.
+					 * Arg1: measured RTT input (mrtt)
+					 * Arg2: updated srtt
 					 */
 	BPF_SOCK_OPS_PARSE_HDR_OPT_CB,	/* Parse the header option.
 					 * It will be called to handle
diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index 5d874817a78db..d1115d7c3936a 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -911,7 +911,7 @@  static void tcp_rtt_estimator(struct sock *sk, long mrtt_us)
 			tp->rtt_seq = tp->snd_nxt;
 			tp->mdev_max_us = tcp_rto_min_us(sk);
 
-			tcp_bpf_rtt(sk);
+			tcp_bpf_rtt(sk, mrtt_us, srtt);
 		}
 	} else {
 		/* no previous measure. */
@@ -921,7 +921,7 @@  static void tcp_rtt_estimator(struct sock *sk, long mrtt_us)
 		tp->mdev_max_us = tp->rttvar_us;
 		tp->rtt_seq = tp->snd_nxt;
 
-		tcp_bpf_rtt(sk);
+		tcp_bpf_rtt(sk, mrtt_us, srtt);
 	}
 	tp->srtt_us = max(1U, srtt);
 }
diff --git a/tools/include/uapi/linux/bpf.h b/tools/include/uapi/linux/bpf.h
index cee0a7915c08a..d80bef9bbdc15 100644
--- a/tools/include/uapi/linux/bpf.h
+++ b/tools/include/uapi/linux/bpf.h
@@ -6947,6 +6947,8 @@  enum {
 					 * socket transition to LISTEN state.
 					 */
 	BPF_SOCK_OPS_RTT_CB,		/* Called on every RTT.
+					 * Arg1: measured RTT input (mrtt)
+					 * Arg2: updated srtt
 					 */
 	BPF_SOCK_OPS_PARSE_HDR_OPT_CB,	/* Parse the header option.
 					 * It will be called to handle