Re: [PATCH net-next v4 1/4] tcp: export CE count in rate_sample

From: Eric Dumazet

Date: Thu Nov 19 2026 - 13:46:56 EST


Le jeu. 19 nov. 2026 à 19:01, <chia-yu.chang@xxxxxxxxxxxxxxxxxxx> a écrit :
>
> From: Ilpo Järvinen <ij@xxxxxxxxxx>
>
> Store the number of newly ACKed/SACKed CE-marked packets in rate_sample.
> This will be used by upcoming congestion control algorithms, e.g.,
> TCP Prague, that need access to AccECN-derived congestion info.
>
> For RFC3168 ECN mode, acked_sacked_ce equals delivered when FLAG_ECE
> is present. For AccECN mode, it carries the derived CE-marked packets
> from AccECN options or ACE field based on RFC9768.
>
> Signed-off-by: Ilpo Järvinen <ij@xxxxxxxxxx>
> Co-developed-by: Olivier Tilmans <olivier.tilmans@xxxxxxxxx>
> Signed-off-by: Olivier Tilmans <olivier.tilmans@xxxxxxxxx>
> Co-developed-by: Chia-Yu Chang <chia-yu.chang@xxxxxxxxxxxxxxxxxxx>
> Signed-off-by: Chia-Yu Chang <chia-yu.chang@xxxxxxxxxxxxxxxxxxx>
>
> --
> v3:
> - Improve commit message
> - Rename ece_delta to acked_sacked_ce
> - Remove redundant rate_sample initializer
> ---
> include/net/tcp.h | 1 +
> net/ipv4/tcp_input.c | 40 +++++++++++++++++++---------------------
> 2 files changed, 20 insertions(+), 21 deletions(-)
>
> diff --git a/include/net/tcp.h b/include/net/tcp.h
> index 25d11fab1b03..4b1159e6c125 100644
> --- a/include/net/tcp.h
> +++ b/include/net/tcp.h
> @@ -1353,6 +1353,7 @@ struct rate_sample {
> long rtt_us; /* RTT of last (S)ACKed packet (or -1) */
> int losses; /* number of packets marked lost upon ACK */
> u32 acked_sacked; /* number of packets newly (S)ACKed upon ACK */
> + u32 acked_sacked_ce; /* number of packets ECE-marked upon ACK */

I presume you meant "CE-marked" instead of "ECE-marked"

> u32 prior_in_flight; /* in flight before this ACK */
> u32 last_end_seq; /* end_seq of most recently ACKed packet */
> bool is_app_limited; /* is sample from packet with bubble in pipe? */
> diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
> index 67e4aa73bd6a..72826cafa246 100644
> --- a/net/ipv4/tcp_input.c
> +++ b/net/ipv4/tcp_input.c
> @@ -616,9 +616,10 @@ static u32 __tcp_accecn_process(struct sock *sk, const struct sk_buff *skb,
> return safe_delta;
> }
>
> -static u32 tcp_accecn_process(struct sock *sk, const struct sk_buff *skb,
> - u32 delivered_pkts, u32 delivered_bytes,
> - u64 prior_bytes_acked, int *flag)
> +static void tcp_accecn_process(struct sock *sk, struct rate_sample *rs,
> + const struct sk_buff *skb, u32 delivered_pkts,
> + u32 delivered_bytes, u64 prior_bytes_acked,
> + int *flag)
> {
> struct tcp_sock *tp = tcp_sk(sk);
> u32 delta;
> @@ -628,11 +629,11 @@ static u32 tcp_accecn_process(struct sock *sk, const struct sk_buff *skb,
> if (delta > 0) {
> tcp_count_delivered_ce(tp, delta);
> *flag |= FLAG_ECE;
> + rs->acked_sacked_ce = delta;
> /* Recalculate header predictor */
> if (tp->pred_flags)
> tcp_fast_path_on(tp);
> }
> - return delta;
> }
>
> /* Buffer size and advertised window tuning.
> @@ -4248,8 +4249,8 @@ static void tcp_xmit_recovery(struct sock *sk, int rexmit)
> }
>
> /* Returns the number of packets newly acked or sacked by the current ACK */
> -static u32 tcp_newly_delivered(struct sock *sk, u32 prior_delivered,
> - u32 ecn_count, int flag)
> +static u32 tcp_newly_delivered(struct sock *sk, struct rate_sample *rs,
> + u32 prior_delivered, int flag)
> {
> const struct net *net = sock_net(sk);
> struct tcp_sock *tp = tcp_sk(sk);
> @@ -4260,8 +4261,9 @@ static u32 tcp_newly_delivered(struct sock *sk, u32 prior_delivered,
>
> if (flag & FLAG_ECE) {
> if (tcp_ecn_mode_rfc3168(tp))
> - ecn_count = delivered;
> - NET_ADD_STATS(net, LINUX_MIB_TCPDELIVEREDCE, ecn_count);
> + rs->acked_sacked_ce = delivered;
> + NET_ADD_STATS(net, LINUX_MIB_TCPDELIVEREDCE,
> + rs->acked_sacked_ce);

In RFC 3168 mode this is every packet delivered by an ACK carrying
ECE, not the number of CE-marked packets. This was fine for a MIB
counter, but once exposed to congestion control modules it should be
documented in the field comment, especially since rate_sample already
has delivered_ce with a different meaning.

> }
>
> return delivered;
> @@ -4327,7 +4329,6 @@ static int tcp_ack(struct sock *sk, const struct sk_buff *skb, int flag)
> u32 delivered = tp->delivered;
> u32 lost = tp->lost;
> int rexmit = REXMIT_NONE; /* Flag to (re)transmit to recover losses */
> - u32 ecn_count = 0; /* Did we receive ECE/an AccECN ACE update? */
> u32 prior_fack;
>
> sack_state.first_sackt = 0;
> @@ -4441,10 +4442,9 @@ static int tcp_ack(struct sock *sk, const struct sk_buff *skb, int flag)
> tcp_rack_update_reo_wnd(sk, &rs);
>
> if (tcp_ecn_mode_accecn(tp))
> - ecn_count = tcp_accecn_process(sk, skb,
> - tp->delivered - delivered,
> - sack_state.delivered_bytes,
> - prior_bytes_acked, &flag);
> + tcp_accecn_process(sk, &rs, skb, tp->delivered - delivered,
> + sack_state.delivered_bytes,
> + prior_bytes_acked, &flag);
>
> tcp_in_ack_event(sk, flag);
>
> @@ -4470,7 +4470,7 @@ static int tcp_ack(struct sock *sk, const struct sk_buff *skb, int flag)
> if ((flag & FLAG_FORWARD_PROGRESS) || !(flag & FLAG_NOT_DUP))
> sk_dst_confirm(sk);
>
> - delivered = tcp_newly_delivered(sk, delivered, ecn_count, flag);
> + delivered = tcp_newly_delivered(sk, &rs, delivered, flag);
>
> lost = tp->lost - lost; /* freshly marked lost */
> rs.is_ack_delayed = !!(flag & FLAG_ACK_MAYBE_DELAYED);
> @@ -4481,17 +4481,15 @@ static int tcp_ack(struct sock *sk, const struct sk_buff *skb, int flag)
>
> no_queue:
> if (tcp_ecn_mode_accecn(tp))
> - ecn_count = tcp_accecn_process(sk, skb,
> - tp->delivered - delivered,
> - sack_state.delivered_bytes,
> - prior_bytes_acked, &flag);
> -
> + tcp_accecn_process(sk, &rs, skb, tp->delivered - delivered,
> + sack_state.delivered_bytes,
> + prior_bytes_acked, &flag);
> tcp_in_ack_event(sk, flag);
> /* If data was DSACKed, see if we can undo a cwnd reduction. */
> if (flag & FLAG_DSACKING_ACK) {
> tcp_fastretrans_alert(sk, prior_snd_una, num_dupack, &flag,
> &rexmit);
> - tcp_newly_delivered(sk, delivered, ecn_count, flag);
> + tcp_newly_delivered(sk, &rs, delivered, flag);
> }
> /* If this ack opens up a zero window, clear backoff. It was
> * being used to time the probes, and is probably far higher than
> @@ -4518,7 +4516,7 @@ static int tcp_ack(struct sock *sk, const struct sk_buff *skb, int flag)
> &sack_state);
> tcp_fastretrans_alert(sk, prior_snd_una, num_dupack, &flag,
> &rexmit);
> - tcp_newly_delivered(sk, delivered, ecn_count, flag);
> + tcp_newly_delivered(sk, &rs, delivered, flag);
> tcp_xmit_recovery(sk, rexmit);
> }
>
> --
> 2.34.1
>