* [PATCH v7 net-next 01/13] tcp: try to avoid safer when ACKs are thinned
2025-12-01 16:37 [PATCH v7 net-next 00/13] AccECN protocol case handling series chia-yu.chang
@ 2025-12-01 16:37 ` chia-yu.chang
2025-12-01 16:37 ` [PATCH v7 net-next 02/13] gro: flushing when CWR is set negatively affects AccECN chia-yu.chang
` (12 subsequent siblings)
13 siblings, 0 replies; 17+ messages in thread
From: chia-yu.chang @ 2025-12-01 16:37 UTC (permalink / raw)
To: pabeni, edumazet, parav, linux-doc, corbet, horms, dsahern,
kuniyu, bpf, netdev, dave.taht, jhs, kuba, stephen,
xiyou.wangcong, jiri, davem, andrew+netdev, donald.hunter, ast,
liuhangbin, shuah, linux-kselftest, ij, ncardwell,
koen.de_schepper, g.white, ingemar.s.johansson, mirja.kuehlewind,
cheshire, rs.ietf, Jason_Livingood, vidhi_goel
Cc: Chia-Yu Chang
From: Ilpo Järvinen <ij@kernel.org>
Add newly acked pkts EWMA. When ACK thinning occurs, select
between safer and unsafe cep delta in AccECN processing based
on it. If the packets ACKed per ACK tends to be large, don't
conservatively assume ACE field overflow.
This patch uses the existing 2-byte holes in the rx group for new
u16 variables withtout creating more holes. Below are the pahole
outcomes before and after this patch:
[BEFORE THIS PATCH]
struct tcp_sock {
[...]
u32 delivered_ecn_bytes[3]; /* 2744 12 */
/* XXX 4 bytes hole, try to pack */
[...]
__cacheline_group_end__tcp_sock_write_rx[0]; /* 2816 0 */
[...]
/* size: 3264, cachelines: 51, members: 177 */
}
[AFTER THIS PATCH]
struct tcp_sock {
[...]
u32 delivered_ecn_bytes[3]; /* 2744 12 */
u16 pkts_acked_ewma; /* 2756 2 */
/* XXX 2 bytes hole, try to pack */
[...]
__cacheline_group_end__tcp_sock_write_rx[0]; /* 2816 0 */
[...]
/* size: 3264, cachelines: 51, members: 178 */
}
Signed-off-by: Ilpo Järvinen <ij@kernel.org>
Co-developed-by: Chia-Yu Chang <chia-yu.chang@nokia-bell-labs.com>
Signed-off-by: Chia-Yu Chang <chia-yu.chang@nokia-bell-labs.com>
Acked-by: Paolo Abeni <pabeni@redhat.com>
---
v3:
- Add additional min() check if pkts_acked_ewma is not initialized.
---
.../networking/net_cachelines/tcp_sock.rst | 1 +
include/linux/tcp.h | 1 +
net/ipv4/tcp.c | 2 ++
net/ipv4/tcp_input.c | 20 ++++++++++++++++++-
4 files changed, 23 insertions(+), 1 deletion(-)
diff --git a/Documentation/networking/net_cachelines/tcp_sock.rst b/Documentation/networking/net_cachelines/tcp_sock.rst
index 26f32dbcf6ec..563daea10d6c 100644
--- a/Documentation/networking/net_cachelines/tcp_sock.rst
+++ b/Documentation/networking/net_cachelines/tcp_sock.rst
@@ -105,6 +105,7 @@ u32 received_ce read_mostly read_w
u32[3] received_ecn_bytes read_mostly read_write
u8:4 received_ce_pending read_mostly read_write
u32[3] delivered_ecn_bytes read_write
+u16 pkts_acked_ewma read_write
u8:2 syn_ect_snt write_mostly read_write
u8:2 syn_ect_rcv read_mostly read_write
u8:2 accecn_minlen write_mostly read_write
diff --git a/include/linux/tcp.h b/include/linux/tcp.h
index 20b8c6e21fef..683f38362977 100644
--- a/include/linux/tcp.h
+++ b/include/linux/tcp.h
@@ -345,6 +345,7 @@ struct tcp_sock {
u32 rate_interval_us; /* saved rate sample: time elapsed */
u32 rcv_rtt_last_tsecr;
u32 delivered_ecn_bytes[3];
+ u16 pkts_acked_ewma;/* Pkts acked EWMA for AccECN cep heuristic */
u64 first_tx_mstamp; /* start of window send phase */
u64 delivered_mstamp; /* time we reached "delivered" */
u64 bytes_acked; /* RFC4898 tcpEStatsAppHCThruOctetsAcked
diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c
index f035440c475a..cd25fb50e81b 100644
--- a/net/ipv4/tcp.c
+++ b/net/ipv4/tcp.c
@@ -3420,6 +3420,7 @@ int tcp_disconnect(struct sock *sk, int flags)
tcp_accecn_init_counters(tp);
tp->prev_ecnfield = 0;
tp->accecn_opt_tstamp = 0;
+ tp->pkts_acked_ewma = 0;
if (icsk->icsk_ca_initialized && icsk->icsk_ca_ops->release)
icsk->icsk_ca_ops->release(sk);
memset(icsk->icsk_ca_priv, 0, sizeof(icsk->icsk_ca_priv));
@@ -5193,6 +5194,7 @@ static void __init tcp_struct_check(void)
CACHELINE_ASSERT_GROUP_MEMBER(struct tcp_sock, tcp_sock_write_rx, rate_interval_us);
CACHELINE_ASSERT_GROUP_MEMBER(struct tcp_sock, tcp_sock_write_rx, rcv_rtt_last_tsecr);
CACHELINE_ASSERT_GROUP_MEMBER(struct tcp_sock, tcp_sock_write_rx, delivered_ecn_bytes);
+ CACHELINE_ASSERT_GROUP_MEMBER(struct tcp_sock, tcp_sock_write_rx, pkts_acked_ewma);
CACHELINE_ASSERT_GROUP_MEMBER(struct tcp_sock, tcp_sock_write_rx, first_tx_mstamp);
CACHELINE_ASSERT_GROUP_MEMBER(struct tcp_sock, tcp_sock_write_rx, delivered_mstamp);
CACHELINE_ASSERT_GROUP_MEMBER(struct tcp_sock, tcp_sock_write_rx, bytes_acked);
diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index 198f8a0d37be..8e95a4e302f4 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -488,6 +488,10 @@ static void tcp_count_delivered(struct tcp_sock *tp, u32 delivered,
tcp_count_delivered_ce(tp, delivered);
}
+#define PKTS_ACKED_WEIGHT 6
+#define PKTS_ACKED_PREC 6
+#define ACK_COMP_THRESH 4
+
/* Returns the ECN CE delta */
static u32 __tcp_accecn_process(struct sock *sk, const struct sk_buff *skb,
u32 delivered_pkts, u32 delivered_bytes,
@@ -499,6 +503,7 @@ static u32 __tcp_accecn_process(struct sock *sk, const struct sk_buff *skb,
u32 delta, safe_delta, d_ceb;
bool opt_deltas_valid;
u32 corrected_ace;
+ u32 ewma;
/* Reordered ACK or uncertain due to lack of data to send and ts */
if (!(flag & (FLAG_FORWARD_PROGRESS | FLAG_TS_PROGRESS)))
@@ -507,6 +512,18 @@ static u32 __tcp_accecn_process(struct sock *sk, const struct sk_buff *skb,
opt_deltas_valid = tcp_accecn_process_option(tp, skb,
delivered_bytes, flag);
+ if (delivered_pkts) {
+ if (!tp->pkts_acked_ewma) {
+ ewma = delivered_pkts << PKTS_ACKED_PREC;
+ } else {
+ ewma = tp->pkts_acked_ewma;
+ ewma = (((ewma << PKTS_ACKED_WEIGHT) - ewma) +
+ (delivered_pkts << PKTS_ACKED_PREC)) >>
+ PKTS_ACKED_WEIGHT;
+ }
+ tp->pkts_acked_ewma = min_t(u32, ewma, 0xFFFFU);
+ }
+
if (!(flag & FLAG_SLOWPATH)) {
/* AccECN counter might overflow on large ACKs */
if (delivered_pkts <= TCP_ACCECN_CEP_ACE_MASK)
@@ -555,7 +572,8 @@ static u32 __tcp_accecn_process(struct sock *sk, const struct sk_buff *skb,
if (d_ceb <
safe_delta * tp->mss_cache >> TCP_ACCECN_SAFETY_SHIFT)
return delta;
- }
+ } else if (tp->pkts_acked_ewma > (ACK_COMP_THRESH << PKTS_ACKED_PREC))
+ return delta;
return safe_delta;
}
--
2.34.1
^ permalink raw reply related [flat|nested] 17+ messages in thread* [PATCH v7 net-next 02/13] gro: flushing when CWR is set negatively affects AccECN
2025-12-01 16:37 [PATCH v7 net-next 00/13] AccECN protocol case handling series chia-yu.chang
2025-12-01 16:37 ` [PATCH v7 net-next 01/13] tcp: try to avoid safer when ACKs are thinned chia-yu.chang
@ 2025-12-01 16:37 ` chia-yu.chang
2025-12-01 16:37 ` [PATCH v7 net-next 03/13] selftests/net: gro: add self-test for TCP CWR flag chia-yu.chang
` (11 subsequent siblings)
13 siblings, 0 replies; 17+ messages in thread
From: chia-yu.chang @ 2025-12-01 16:37 UTC (permalink / raw)
To: pabeni, edumazet, parav, linux-doc, corbet, horms, dsahern,
kuniyu, bpf, netdev, dave.taht, jhs, kuba, stephen,
xiyou.wangcong, jiri, davem, andrew+netdev, donald.hunter, ast,
liuhangbin, shuah, linux-kselftest, ij, ncardwell,
koen.de_schepper, g.white, ingemar.s.johansson, mirja.kuehlewind,
cheshire, rs.ietf, Jason_Livingood, vidhi_goel
Cc: Chia-Yu Chang
From: Ilpo Järvinen <ij@kernel.org>
As AccECN may keep CWR bit asserted due to different
interpretation of the bit, flushing with GRO because of
CWR may effectively disable GRO until AccECN counter
field changes such that CWR-bit becomes 0.
There is no harm done from not immediately forwarding the
CWR'ed segment with RFC3168 ECN.
Signed-off-by: Ilpo Järvinen <ij@kernel.org>
Signed-off-by: Chia-Yu Chang <chia-yu.chang@nokia-bell-labs.com>
---
net/ipv4/tcp_offload.c | 3 +--
1 file changed, 1 insertion(+), 2 deletions(-)
diff --git a/net/ipv4/tcp_offload.c b/net/ipv4/tcp_offload.c
index fdda18b1abda..9bd710c7bc95 100644
--- a/net/ipv4/tcp_offload.c
+++ b/net/ipv4/tcp_offload.c
@@ -303,8 +303,7 @@ struct sk_buff *tcp_gro_receive(struct list_head *head, struct sk_buff *skb,
goto out_check_final;
th2 = tcp_hdr(p);
- flush = (__force int)(flags & TCP_FLAG_CWR);
- flush |= (__force int)((flags ^ tcp_flag_word(th2)) &
+ flush = (__force int)((flags ^ tcp_flag_word(th2)) &
~(TCP_FLAG_FIN | TCP_FLAG_PSH));
flush |= (__force int)(th->ack_seq ^ th2->ack_seq);
for (i = sizeof(*th); i < thlen; i += 4)
--
2.34.1
^ permalink raw reply related [flat|nested] 17+ messages in thread* [PATCH v7 net-next 03/13] selftests/net: gro: add self-test for TCP CWR flag
2025-12-01 16:37 [PATCH v7 net-next 00/13] AccECN protocol case handling series chia-yu.chang
2025-12-01 16:37 ` [PATCH v7 net-next 01/13] tcp: try to avoid safer when ACKs are thinned chia-yu.chang
2025-12-01 16:37 ` [PATCH v7 net-next 02/13] gro: flushing when CWR is set negatively affects AccECN chia-yu.chang
@ 2025-12-01 16:37 ` chia-yu.chang
2025-12-01 16:37 ` [PATCH v7 net-next 04/13] tcp: ECT_1_NEGOTIATION and NEEDS_ACCECN identifiers chia-yu.chang
` (10 subsequent siblings)
13 siblings, 0 replies; 17+ messages in thread
From: chia-yu.chang @ 2025-12-01 16:37 UTC (permalink / raw)
To: pabeni, edumazet, parav, linux-doc, corbet, horms, dsahern,
kuniyu, bpf, netdev, dave.taht, jhs, kuba, stephen,
xiyou.wangcong, jiri, davem, andrew+netdev, donald.hunter, ast,
liuhangbin, shuah, linux-kselftest, ij, ncardwell,
koen.de_schepper, g.white, ingemar.s.johansson, mirja.kuehlewind,
cheshire, rs.ietf, Jason_Livingood, vidhi_goel
Cc: Chia-Yu Chang
From: Chia-Yu Chang <chia-yu.chang@nokia-bell-labs.com>
Currently, GRO does not flush packets when the CWR bit is set.
A corresponding self-test is being added, in which the CWR flag
is set for two consecutive packets, but the first packet with the
CWR flag set will not be flushed immediately.
+===================+==========+===============+===========+
| Packet id | CWR flag | Payload | Flushing? |
+===================+==========+===============+===========+
| 0 | 0 | PAYLOAD_LEN | 0 |
| ... | 0 | PAYLOAD_LEN | 1 |
+-------------------+----------+---------------+-----------+
| NUM_PACKETS/2 - 1 | 1 | payload_len | 0 |
| NUM_PACKETS/2 | 1 | payload_len | 1 |
+-------------------+----------+---------------+-----------+
| ... | 0 | PAYLOAD_LEN | 0 |
| NUM_PACKETS | 0 | PAYLOAD_LEN | 1 |
+===================+==========+===============+===========+
Signed-off-by: Chia-Yu Chang <chia-yu.chang@nokia-bell-labs.com>
Acked-by: Paolo Abeni <pabeni@redhat.com>
---
v7:
- Update comments
---
tools/testing/selftests/drivers/net/gro.c | 81 ++++++++++++++++-------
1 file changed, 58 insertions(+), 23 deletions(-)
diff --git a/tools/testing/selftests/drivers/net/gro.c b/tools/testing/selftests/drivers/net/gro.c
index 995b492f5bcb..00fef326efdf 100644
--- a/tools/testing/selftests/drivers/net/gro.c
+++ b/tools/testing/selftests/drivers/net/gro.c
@@ -11,8 +11,8 @@
* 2.ack
* Pure ACK does not coalesce.
* 3.flags
- * Specific test cases: no packets with PSH, SYN, URG, RST set will
- * be coalesced.
+ * Specific test cases: no packets with PSH, SYN, URG, RST, CWR set
+ * will be coalesced.
* 4.tcp
* Packets with incorrect checksum, non-consecutive seqno and
* different TCP header options shouldn't coalesce. Nit: given that
@@ -333,32 +333,58 @@ static void create_packet(void *buf, int seq_offset, int ack_offset,
fill_datalinklayer(buf);
}
-/* send one extra flag, not first and not last pkt */
-static void send_flags(int fd, struct sockaddr_ll *daddr, int psh, int syn,
- int rst, int urg)
+#ifndef TH_CWR
+#define TH_CWR 0x80
+#endif
+static void set_flags(struct tcphdr *tcph, int payload_len, int psh, int syn,
+ int rst, int urg, int cwr)
{
- static char flag_buf[MAX_HDR_LEN + PAYLOAD_LEN];
- static char buf[MAX_HDR_LEN + PAYLOAD_LEN];
- int payload_len, pkt_size, flag, i;
- struct tcphdr *tcph;
-
- payload_len = PAYLOAD_LEN * psh;
- pkt_size = total_hdr_len + payload_len;
- flag = NUM_PACKETS / 2;
-
- create_packet(flag_buf, flag * payload_len, 0, payload_len, 0);
-
- tcph = (struct tcphdr *)(flag_buf + tcp_offset);
tcph->psh = psh;
tcph->syn = syn;
tcph->rst = rst;
tcph->urg = urg;
+ if (cwr)
+ tcph->th_flags |= TH_CWR;
+ else
+ tcph->th_flags &= ~TH_CWR;
tcph->check = 0;
tcph->check = tcp_checksum(tcph, payload_len);
+}
+
+/* send extra flags of the (NUM_PACKETS / 2) and (NUM_PACKETS / 2 - 1)
+ * pkts, not first and not last pkt
+ */
+static void send_flags(int fd, struct sockaddr_ll *daddr, int psh, int syn,
+ int rst, int urg, int cwr)
+{
+ static char flag_buf[2][MAX_HDR_LEN + PAYLOAD_LEN];
+ static char buf[MAX_HDR_LEN + PAYLOAD_LEN];
+ int payload_len, pkt_size, i;
+ struct tcphdr *tcph;
+ int flag[2];
+
+ payload_len = PAYLOAD_LEN * (psh || cwr);
+ pkt_size = total_hdr_len + payload_len;
+ flag[0] = NUM_PACKETS / 2;
+ flag[1] = NUM_PACKETS / 2 - 1;
+
+ /* Create and configure packets with flags
+ */
+ for (i = 0; i < 2; i++) {
+ if (flag[i] > 0) {
+ create_packet(flag_buf[i], flag[i] * payload_len, 0,
+ payload_len, 0);
+ tcph = (struct tcphdr *)(flag_buf[i] + tcp_offset);
+ set_flags(tcph, payload_len, psh, syn, rst, urg, cwr);
+ }
+ }
for (i = 0; i < NUM_PACKETS + 1; i++) {
- if (i == flag) {
- write_packet(fd, flag_buf, pkt_size, daddr);
+ if (i == flag[0]) {
+ write_packet(fd, flag_buf[0], pkt_size, daddr);
+ continue;
+ } else if (i == flag[1] && cwr) {
+ write_packet(fd, flag_buf[1], pkt_size, daddr);
continue;
}
create_packet(buf, i * PAYLOAD_LEN, 0, PAYLOAD_LEN, 0);
@@ -1021,16 +1047,19 @@ static void gro_sender(void)
send_ack(txfd, &daddr);
write_packet(txfd, fin_pkt, total_hdr_len, &daddr);
} else if (strcmp(testname, "flags") == 0) {
- send_flags(txfd, &daddr, 1, 0, 0, 0);
+ send_flags(txfd, &daddr, 1, 0, 0, 0, 0);
write_packet(txfd, fin_pkt, total_hdr_len, &daddr);
- send_flags(txfd, &daddr, 0, 1, 0, 0);
+ send_flags(txfd, &daddr, 0, 1, 0, 0, 0);
write_packet(txfd, fin_pkt, total_hdr_len, &daddr);
- send_flags(txfd, &daddr, 0, 0, 1, 0);
+ send_flags(txfd, &daddr, 0, 0, 1, 0, 0);
write_packet(txfd, fin_pkt, total_hdr_len, &daddr);
- send_flags(txfd, &daddr, 0, 0, 0, 1);
+ send_flags(txfd, &daddr, 0, 0, 0, 1, 0);
+ write_packet(txfd, fin_pkt, total_hdr_len, &daddr);
+
+ send_flags(txfd, &daddr, 0, 0, 0, 0, 1);
write_packet(txfd, fin_pkt, total_hdr_len, &daddr);
} else if (strcmp(testname, "tcp") == 0) {
send_changed_checksum(txfd, &daddr);
@@ -1166,6 +1195,12 @@ static void gro_receiver(void)
printf("urg flag ends coalescing: ");
check_recv_pkts(rxfd, correct_payload, 3);
+
+ correct_payload[0] = PAYLOAD_LEN;
+ correct_payload[1] = PAYLOAD_LEN * 2;
+ correct_payload[2] = PAYLOAD_LEN * 2;
+ printf("cwr flag ends coalescing: ");
+ check_recv_pkts(rxfd, correct_payload, 3);
} else if (strcmp(testname, "tcp") == 0) {
correct_payload[0] = PAYLOAD_LEN;
correct_payload[1] = PAYLOAD_LEN;
--
2.34.1
^ permalink raw reply related [flat|nested] 17+ messages in thread* [PATCH v7 net-next 04/13] tcp: ECT_1_NEGOTIATION and NEEDS_ACCECN identifiers
2025-12-01 16:37 [PATCH v7 net-next 00/13] AccECN protocol case handling series chia-yu.chang
` (2 preceding siblings ...)
2025-12-01 16:37 ` [PATCH v7 net-next 03/13] selftests/net: gro: add self-test for TCP CWR flag chia-yu.chang
@ 2025-12-01 16:37 ` chia-yu.chang
2025-12-01 16:37 ` [PATCH v7 net-next 05/13] tcp: disable RFC3168 fallback identifier for CC modules chia-yu.chang
` (9 subsequent siblings)
13 siblings, 0 replies; 17+ messages in thread
From: chia-yu.chang @ 2025-12-01 16:37 UTC (permalink / raw)
To: pabeni, edumazet, parav, linux-doc, corbet, horms, dsahern,
kuniyu, bpf, netdev, dave.taht, jhs, kuba, stephen,
xiyou.wangcong, jiri, davem, andrew+netdev, donald.hunter, ast,
liuhangbin, shuah, linux-kselftest, ij, ncardwell,
koen.de_schepper, g.white, ingemar.s.johansson, mirja.kuehlewind,
cheshire, rs.ietf, Jason_Livingood, vidhi_goel
Cc: Chia-Yu Chang, Olivier Tilmans
From: Chia-Yu Chang <chia-yu.chang@nokia-bell-labs.com>
Two CA module flags are added in this patch related to AccECN negotiation.
First, a new CA module flag (TCP_CONG_NEEDS_ACCECN) defines that the CA
expects to negotiate AccECN functionality using the ECE, CWR and AE flags
in the TCP header.
Second, during ECN negotiation, ECT(0) in the IP header is used. This patch
enables CA to control whether ECT(0) or ECT(1) should be used on a per-segment
basis. A new flag (TCP_CONG_ECT_1_NEGOTIATION) defines the expected ECT value
in the IP header by the CA when not-yet initialized for the connection.
The detailed AccECN negotiaotn during the 3WHS can be found in the AccECN spec:
https://tools.ietf.org/id/draft-ietf-tcpm-accurate-ecn-28.txt
Co-developed-by: Olivier Tilmans <olivier.tilmans@nokia.com>
Signed-off-by: Olivier Tilmans <olivier.tilmans@nokia.com>
Signed-off-by: Ilpo Järvinen <ij@kernel.org>
Signed-off-by: Chia-Yu Chang <chia-yu.chang@nokia-bell-labs.com>
Acked-by: Paolo Abeni <pabeni@redhat.com>
---
v6:
- Rename TCP_CONG_WANTS_ECT_1 to TCP_CONG_ECT_1_NEGOTIATION to distinguish
it from TCP_CONG_ECT_1_ESTABLISH.
- Move TCP_CONG_ECT_1_ESTABLISH to latter TCP Prague patch series.
v3:
- Change TCP_CONG_WANTS_ECT_1 into individual flag.
- Add helper function INET_ECN_xmit_wants_ect_1().
---
include/net/inet_ecn.h | 20 +++++++++++++++++---
include/net/tcp.h | 21 ++++++++++++++++++++-
include/net/tcp_ecn.h | 13 ++++++++++---
net/ipv4/tcp_cong.c | 5 +++--
net/ipv4/tcp_input.c | 3 ++-
5 files changed, 52 insertions(+), 10 deletions(-)
diff --git a/include/net/inet_ecn.h b/include/net/inet_ecn.h
index ea32393464a2..827b87a95dab 100644
--- a/include/net/inet_ecn.h
+++ b/include/net/inet_ecn.h
@@ -51,11 +51,25 @@ static inline __u8 INET_ECN_encapsulate(__u8 outer, __u8 inner)
return outer;
}
+/* Apply either ECT(0) or ECT(1) */
+static inline void __INET_ECN_xmit(struct sock *sk, bool use_ect_1)
+{
+ __u8 ect = use_ect_1 ? INET_ECN_ECT_1 : INET_ECN_ECT_0;
+
+ /* Mask the complete byte in case the connection alternates between
+ * ECT(0) and ECT(1).
+ */
+ inet_sk(sk)->tos &= ~INET_ECN_MASK;
+ inet_sk(sk)->tos |= ect;
+ if (inet6_sk(sk)) {
+ inet6_sk(sk)->tclass &= ~INET_ECN_MASK;
+ inet6_sk(sk)->tclass |= ect;
+ }
+}
+
static inline void INET_ECN_xmit(struct sock *sk)
{
- inet_sk(sk)->tos |= INET_ECN_ECT_0;
- if (inet6_sk(sk) != NULL)
- inet6_sk(sk)->tclass |= INET_ECN_ECT_0;
+ __INET_ECN_xmit(sk, false);
}
static inline void INET_ECN_dontxmit(struct sock *sk)
diff --git a/include/net/tcp.h b/include/net/tcp.h
index 0deb5e9dd911..c4c25f2e0c0d 100644
--- a/include/net/tcp.h
+++ b/include/net/tcp.h
@@ -1203,7 +1203,12 @@ enum tcp_ca_ack_event_flags {
#define TCP_CONG_NON_RESTRICTED BIT(0)
/* Requires ECN/ECT set on all packets */
#define TCP_CONG_NEEDS_ECN BIT(1)
-#define TCP_CONG_MASK (TCP_CONG_NON_RESTRICTED | TCP_CONG_NEEDS_ECN)
+/* Require successfully negotiated AccECN capability */
+#define TCP_CONG_NEEDS_ACCECN BIT(2)
+/* Use ECT(1) instead of ECT(0) while the CA is uninitialized */
+#define TCP_CONG_ECT_1_NEGOTIATION BIT(3)
+#define TCP_CONG_MASK (TCP_CONG_NON_RESTRICTED | TCP_CONG_NEEDS_ECN | \
+ TCP_CONG_NEEDS_ACCECN | TCP_CONG_ECT_1_NEGOTIATION)
union tcp_cc_info;
@@ -1335,6 +1340,20 @@ static inline bool tcp_ca_needs_ecn(const struct sock *sk)
return icsk->icsk_ca_ops->flags & TCP_CONG_NEEDS_ECN;
}
+static inline bool tcp_ca_needs_accecn(const struct sock *sk)
+{
+ const struct inet_connection_sock *icsk = inet_csk(sk);
+
+ return icsk->icsk_ca_ops->flags & TCP_CONG_NEEDS_ACCECN;
+}
+
+static inline bool tcp_ca_ect_1_negotiation(const struct sock *sk)
+{
+ const struct inet_connection_sock *icsk = inet_csk(sk);
+
+ return icsk->icsk_ca_ops->flags & TCP_CONG_ECT_1_NEGOTIATION;
+}
+
static inline void tcp_ca_event(struct sock *sk, const enum tcp_ca_event event)
{
const struct inet_connection_sock *icsk = inet_csk(sk);
diff --git a/include/net/tcp_ecn.h b/include/net/tcp_ecn.h
index f13e5cd2b1ac..fdde1c342b35 100644
--- a/include/net/tcp_ecn.h
+++ b/include/net/tcp_ecn.h
@@ -31,6 +31,12 @@ enum tcp_accecn_option {
TCP_ACCECN_OPTION_FULL = 2,
};
+/* Apply either ECT(0) or ECT(1) based on TCP_CONG_ECT_1_NEGOTIATION flag */
+static inline void INET_ECN_xmit_ect_1_negotiation(struct sock *sk)
+{
+ __INET_ECN_xmit(sk, tcp_ca_ect_1_negotiation(sk));
+}
+
static inline void tcp_ecn_queue_cwr(struct tcp_sock *tp)
{
/* Do not set CWR if in AccECN mode! */
@@ -561,7 +567,7 @@ static inline void tcp_ecn_send_synack(struct sock *sk, struct sk_buff *skb)
TCP_SKB_CB(skb)->tcp_flags &= ~TCPHDR_ECE;
else if (tcp_ca_needs_ecn(sk) ||
tcp_bpf_ca_needs_ecn(sk))
- INET_ECN_xmit(sk);
+ INET_ECN_xmit_ect_1_negotiation(sk);
if (tp->ecn_flags & TCP_ECN_MODE_ACCECN) {
TCP_SKB_CB(skb)->tcp_flags &= ~TCPHDR_ACE;
@@ -579,7 +585,8 @@ static inline void tcp_ecn_send_syn(struct sock *sk, struct sk_buff *skb)
bool use_ecn, use_accecn;
u8 tcp_ecn = READ_ONCE(sock_net(sk)->ipv4.sysctl_tcp_ecn);
- use_accecn = tcp_ecn == TCP_ECN_IN_ACCECN_OUT_ACCECN;
+ use_accecn = tcp_ecn == TCP_ECN_IN_ACCECN_OUT_ACCECN ||
+ tcp_ca_needs_accecn(sk);
use_ecn = tcp_ecn == TCP_ECN_IN_ECN_OUT_ECN ||
tcp_ecn == TCP_ECN_IN_ACCECN_OUT_ECN ||
tcp_ca_needs_ecn(sk) || bpf_needs_ecn || use_accecn;
@@ -595,7 +602,7 @@ static inline void tcp_ecn_send_syn(struct sock *sk, struct sk_buff *skb)
if (use_ecn) {
if (tcp_ca_needs_ecn(sk) || bpf_needs_ecn)
- INET_ECN_xmit(sk);
+ INET_ECN_xmit_ect_1_negotiation(sk);
TCP_SKB_CB(skb)->tcp_flags |= TCPHDR_ECE | TCPHDR_CWR;
if (use_accecn) {
diff --git a/net/ipv4/tcp_cong.c b/net/ipv4/tcp_cong.c
index df758adbb445..e9f6c77e0631 100644
--- a/net/ipv4/tcp_cong.c
+++ b/net/ipv4/tcp_cong.c
@@ -16,6 +16,7 @@
#include <linux/gfp.h>
#include <linux/jhash.h>
#include <net/tcp.h>
+#include <net/tcp_ecn.h>
#include <trace/events/tcp.h>
static DEFINE_SPINLOCK(tcp_cong_list_lock);
@@ -227,7 +228,7 @@ void tcp_assign_congestion_control(struct sock *sk)
memset(icsk->icsk_ca_priv, 0, sizeof(icsk->icsk_ca_priv));
if (ca->flags & TCP_CONG_NEEDS_ECN)
- INET_ECN_xmit(sk);
+ INET_ECN_xmit_ect_1_negotiation(sk);
else
INET_ECN_dontxmit(sk);
}
@@ -257,7 +258,7 @@ static void tcp_reinit_congestion_control(struct sock *sk,
memset(icsk->icsk_ca_priv, 0, sizeof(icsk->icsk_ca_priv));
if (ca->flags & TCP_CONG_NEEDS_ECN)
- INET_ECN_xmit(sk);
+ INET_ECN_xmit_ect_1_negotiation(sk);
else
INET_ECN_dontxmit(sk);
diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index 8e95a4e302f4..ccbab5569680 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -7266,7 +7266,8 @@ static void tcp_ecn_create_request(struct request_sock *req,
u32 ecn_ok_dst;
if (tcp_accecn_syn_requested(th) &&
- READ_ONCE(net->ipv4.sysctl_tcp_ecn) >= 3) {
+ (READ_ONCE(net->ipv4.sysctl_tcp_ecn) >= 3 ||
+ tcp_ca_needs_accecn(listen_sk))) {
inet_rsk(req)->ecn_ok = 1;
tcp_rsk(req)->accecn_ok = 1;
tcp_rsk(req)->syn_ect_rcv = TCP_SKB_CB(skb)->ip_dsfield &
--
2.34.1
^ permalink raw reply related [flat|nested] 17+ messages in thread* [PATCH v7 net-next 05/13] tcp: disable RFC3168 fallback identifier for CC modules
2025-12-01 16:37 [PATCH v7 net-next 00/13] AccECN protocol case handling series chia-yu.chang
` (3 preceding siblings ...)
2025-12-01 16:37 ` [PATCH v7 net-next 04/13] tcp: ECT_1_NEGOTIATION and NEEDS_ACCECN identifiers chia-yu.chang
@ 2025-12-01 16:37 ` chia-yu.chang
2025-12-01 16:37 ` [PATCH v7 net-next 06/13] tcp: accecn: handle unexpected AccECN negotiation feedback chia-yu.chang
` (8 subsequent siblings)
13 siblings, 0 replies; 17+ messages in thread
From: chia-yu.chang @ 2025-12-01 16:37 UTC (permalink / raw)
To: pabeni, edumazet, parav, linux-doc, corbet, horms, dsahern,
kuniyu, bpf, netdev, dave.taht, jhs, kuba, stephen,
xiyou.wangcong, jiri, davem, andrew+netdev, donald.hunter, ast,
liuhangbin, shuah, linux-kselftest, ij, ncardwell,
koen.de_schepper, g.white, ingemar.s.johansson, mirja.kuehlewind,
cheshire, rs.ietf, Jason_Livingood, vidhi_goel
Cc: Chia-Yu Chang
From: Chia-Yu Chang <chia-yu.chang@nokia-bell-labs.com>
When AccECN is not successfully negociated for a TCP flow, it defaults
fallback to classic ECN (RFC3168). However, L4S service will fallback
to non-ECN.
This patch enables congestion control module to control whether it
should not fallback to classic ECN after unsuccessful AccECN negotiation.
A new CA module flag (TCP_CONG_NO_FALLBACK_RFC3168) identifies this
behavior expected by the CA.
Signed-off-by: Chia-Yu Chang <chia-yu.chang@nokia-bell-labs.com>
Acked-by: Paolo Abeni <pabeni@redhat.com>
---
v3:
- Add empty line between variable declarations and code.
---
include/net/tcp.h | 12 +++++++++++-
include/net/tcp_ecn.h | 11 ++++++++---
net/ipv4/tcp_input.c | 2 +-
net/ipv4/tcp_minisocks.c | 7 ++++---
4 files changed, 24 insertions(+), 8 deletions(-)
diff --git a/include/net/tcp.h b/include/net/tcp.h
index c4c25f2e0c0d..e0a5cf2f7818 100644
--- a/include/net/tcp.h
+++ b/include/net/tcp.h
@@ -1207,8 +1207,11 @@ enum tcp_ca_ack_event_flags {
#define TCP_CONG_NEEDS_ACCECN BIT(2)
/* Use ECT(1) instead of ECT(0) while the CA is uninitialized */
#define TCP_CONG_ECT_1_NEGOTIATION BIT(3)
+/* Cannot fallback to RFC3168 during AccECN negotiation */
+#define TCP_CONG_NO_FALLBACK_RFC3168 BIT(4)
#define TCP_CONG_MASK (TCP_CONG_NON_RESTRICTED | TCP_CONG_NEEDS_ECN | \
- TCP_CONG_NEEDS_ACCECN | TCP_CONG_ECT_1_NEGOTIATION)
+ TCP_CONG_NEEDS_ACCECN | TCP_CONG_ECT_1_NEGOTIATION | \
+ TCP_CONG_NO_FALLBACK_RFC3168)
union tcp_cc_info;
@@ -1354,6 +1357,13 @@ static inline bool tcp_ca_ect_1_negotiation(const struct sock *sk)
return icsk->icsk_ca_ops->flags & TCP_CONG_ECT_1_NEGOTIATION;
}
+static inline bool tcp_ca_no_fallback_rfc3168(const struct sock *sk)
+{
+ const struct inet_connection_sock *icsk = inet_csk(sk);
+
+ return icsk->icsk_ca_ops->flags & TCP_CONG_NO_FALLBACK_RFC3168;
+}
+
static inline void tcp_ca_event(struct sock *sk, const enum tcp_ca_event event)
{
const struct inet_connection_sock *icsk = inet_csk(sk);
diff --git a/include/net/tcp_ecn.h b/include/net/tcp_ecn.h
index fdde1c342b35..2e1637edf1d3 100644
--- a/include/net/tcp_ecn.h
+++ b/include/net/tcp_ecn.h
@@ -507,7 +507,9 @@ static inline void tcp_ecn_rcv_synack(struct sock *sk, const struct sk_buff *skb
* | ECN | AccECN | 0 0 1 | Classic ECN |
* +========+========+============+=============+
*/
- if (tcp_ecn_mode_pending(tp))
+ if (tcp_ca_no_fallback_rfc3168(sk))
+ tcp_ecn_mode_set(tp, TCP_ECN_DISABLED);
+ else if (tcp_ecn_mode_pending(tp))
/* Downgrade from AccECN, or requested initially */
tcp_ecn_mode_set(tp, TCP_ECN_MODE_RFC3168);
break;
@@ -531,9 +533,11 @@ static inline void tcp_ecn_rcv_synack(struct sock *sk, const struct sk_buff *skb
}
}
-static inline void tcp_ecn_rcv_syn(struct tcp_sock *tp, const struct tcphdr *th,
+static inline void tcp_ecn_rcv_syn(struct sock *sk, const struct tcphdr *th,
const struct sk_buff *skb)
{
+ struct tcp_sock *tp = tcp_sk(sk);
+
if (tcp_ecn_mode_pending(tp)) {
if (!tcp_accecn_syn_requested(th)) {
/* Downgrade to classic ECN feedback */
@@ -545,7 +549,8 @@ static inline void tcp_ecn_rcv_syn(struct tcp_sock *tp, const struct tcphdr *th,
tcp_ecn_mode_set(tp, TCP_ECN_MODE_ACCECN);
}
}
- if (tcp_ecn_mode_rfc3168(tp) && (!th->ece || !th->cwr))
+ if (tcp_ecn_mode_rfc3168(tp) &&
+ (!th->ece || !th->cwr || tcp_ca_no_fallback_rfc3168(sk)))
tcp_ecn_mode_set(tp, TCP_ECN_DISABLED);
}
diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index ccbab5569680..e5c9cf586437 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -6861,7 +6861,7 @@ static int tcp_rcv_synsent_state_process(struct sock *sk, struct sk_buff *skb,
tp->snd_wl1 = TCP_SKB_CB(skb)->seq;
tp->max_window = tp->snd_wnd;
- tcp_ecn_rcv_syn(tp, th, skb);
+ tcp_ecn_rcv_syn(sk, th, skb);
tcp_mtup_init(sk);
tcp_sync_mss(sk, icsk->icsk_pmtu_cookie);
diff --git a/net/ipv4/tcp_minisocks.c b/net/ipv4/tcp_minisocks.c
index bd5462154f97..9776c921d1bb 100644
--- a/net/ipv4/tcp_minisocks.c
+++ b/net/ipv4/tcp_minisocks.c
@@ -485,9 +485,10 @@ static void tcp_ecn_openreq_child(struct sock *sk,
tp->accecn_opt_demand = 1;
tcp_ecn_received_counters_payload(sk, skb);
} else {
- tcp_ecn_mode_set(tp, inet_rsk(req)->ecn_ok ?
- TCP_ECN_MODE_RFC3168 :
- TCP_ECN_DISABLED);
+ if (inet_rsk(req)->ecn_ok && !tcp_ca_no_fallback_rfc3168(sk))
+ tcp_ecn_mode_set(tp, TCP_ECN_MODE_RFC3168);
+ else
+ tcp_ecn_mode_set(tp, TCP_ECN_DISABLED);
}
}
--
2.34.1
^ permalink raw reply related [flat|nested] 17+ messages in thread* [PATCH v7 net-next 06/13] tcp: accecn: handle unexpected AccECN negotiation feedback
2025-12-01 16:37 [PATCH v7 net-next 00/13] AccECN protocol case handling series chia-yu.chang
` (4 preceding siblings ...)
2025-12-01 16:37 ` [PATCH v7 net-next 05/13] tcp: disable RFC3168 fallback identifier for CC modules chia-yu.chang
@ 2025-12-01 16:37 ` chia-yu.chang
2025-12-01 16:37 ` [PATCH v7 net-next 07/13] tcp: accecn: retransmit downgraded SYN in AccECN negotiation chia-yu.chang
` (7 subsequent siblings)
13 siblings, 0 replies; 17+ messages in thread
From: chia-yu.chang @ 2025-12-01 16:37 UTC (permalink / raw)
To: pabeni, edumazet, parav, linux-doc, corbet, horms, dsahern,
kuniyu, bpf, netdev, dave.taht, jhs, kuba, stephen,
xiyou.wangcong, jiri, davem, andrew+netdev, donald.hunter, ast,
liuhangbin, shuah, linux-kselftest, ij, ncardwell,
koen.de_schepper, g.white, ingemar.s.johansson, mirja.kuehlewind,
cheshire, rs.ietf, Jason_Livingood, vidhi_goel
Cc: Chia-Yu Chang
From: Chia-Yu Chang <chia-yu.chang@nokia-bell-labs.com>
According to Section 3.1.2 of AccECN spec (RFC9768), if a TCP Client
has sent a SYN requesting AccECN feedback with (AE,CWR,ECE) = (1,1,1)
then receives a SYN/ACK with the currently reserved combination
(AE,CWR,ECE) = (1,0,1) but it does not have logic specific to such a
combination, the Client MUST enable AccECN mode as if the SYN/ACK
confirmed that the Server supported AccECN and as if it fed back that
the IP-ECN field on the SYN had arrived unchanged.
Fixes: 3cae34274c79 ("tcp: accecn: AccECN negotiation").
Signed-off-by: Chia-Yu Chang <chia-yu.chang@nokia-bell-labs.com>
Acked-by: Paolo Abeni <pabeni@redhat.com>
---
v5:
- Add "Fixes" tag.
v3:
- Update commit message to fix old AccECN commits.
---
include/net/tcp_ecn.h | 44 ++++++++++++++++++++++++++++++-------------
1 file changed, 31 insertions(+), 13 deletions(-)
diff --git a/include/net/tcp_ecn.h b/include/net/tcp_ecn.h
index 2e1637edf1d3..a709fb1756eb 100644
--- a/include/net/tcp_ecn.h
+++ b/include/net/tcp_ecn.h
@@ -473,6 +473,26 @@ static inline u8 tcp_accecn_option_init(const struct sk_buff *skb,
return TCP_ACCECN_OPT_COUNTER_SEEN;
}
+static inline void tcp_ecn_rcv_synack_accecn(struct sock *sk,
+ const struct sk_buff *skb, u8 dsf)
+{
+ struct tcp_sock *tp = tcp_sk(sk);
+
+ tcp_ecn_mode_set(tp, TCP_ECN_MODE_ACCECN);
+ tp->syn_ect_rcv = dsf & INET_ECN_MASK;
+ /* Demand Accurate ECN option in response to the SYN on the SYN/ACK
+ * and the TCP server will try to send one more packet with an AccECN
+ * Option at a later point during the connection.
+ */
+ if (tp->rx_opt.accecn &&
+ tp->saw_accecn_opt < TCP_ACCECN_OPT_COUNTER_SEEN) {
+ u8 saw_opt = tcp_accecn_option_init(skb, tp->rx_opt.accecn);
+
+ tcp_accecn_saw_opt_fail_recv(tp, saw_opt);
+ tp->accecn_opt_demand = 2;
+ }
+}
+
/* See Table 2 of the AccECN draft */
static inline void tcp_ecn_rcv_synack(struct sock *sk, const struct sk_buff *skb,
const struct tcphdr *th, u8 ip_dsfield)
@@ -495,13 +515,11 @@ static inline void tcp_ecn_rcv_synack(struct sock *sk, const struct sk_buff *skb
tcp_ecn_mode_set(tp, TCP_ECN_DISABLED);
break;
case 0x1:
- case 0x5:
/* +========+========+============+=============+
* | A | B | SYN/ACK | Feedback |
* | | | B->A | Mode of A |
* | | | AE CWR ECE | |
* +========+========+============+=============+
- * | AccECN | Nonce | 1 0 1 | (Reserved) |
* | AccECN | ECN | 0 0 1 | Classic ECN |
* | Nonce | AccECN | 0 0 1 | Classic ECN |
* | ECN | AccECN | 0 0 1 | Classic ECN |
@@ -509,20 +527,20 @@ static inline void tcp_ecn_rcv_synack(struct sock *sk, const struct sk_buff *skb
*/
if (tcp_ca_no_fallback_rfc3168(sk))
tcp_ecn_mode_set(tp, TCP_ECN_DISABLED);
- else if (tcp_ecn_mode_pending(tp))
- /* Downgrade from AccECN, or requested initially */
+ else
tcp_ecn_mode_set(tp, TCP_ECN_MODE_RFC3168);
break;
- default:
- tcp_ecn_mode_set(tp, TCP_ECN_MODE_ACCECN);
- tp->syn_ect_rcv = ip_dsfield & INET_ECN_MASK;
- if (tp->rx_opt.accecn &&
- tp->saw_accecn_opt < TCP_ACCECN_OPT_COUNTER_SEEN) {
- u8 saw_opt = tcp_accecn_option_init(skb, tp->rx_opt.accecn);
-
- tcp_accecn_saw_opt_fail_recv(tp, saw_opt);
- tp->accecn_opt_demand = 2;
+ case 0x5:
+ if (tcp_ecn_mode_pending(tp)) {
+ tcp_ecn_rcv_synack_accecn(sk, skb, ip_dsfield);
+ if (INET_ECN_is_ce(ip_dsfield)) {
+ tp->received_ce++;
+ tp->received_ce_pending++;
+ }
}
+ break;
+ default:
+ tcp_ecn_rcv_synack_accecn(sk, skb, ip_dsfield);
if (INET_ECN_is_ce(ip_dsfield) &&
tcp_accecn_validate_syn_feedback(sk, ace,
tp->syn_ect_snt)) {
--
2.34.1
^ permalink raw reply related [flat|nested] 17+ messages in thread* [PATCH v7 net-next 07/13] tcp: accecn: retransmit downgraded SYN in AccECN negotiation
2025-12-01 16:37 [PATCH v7 net-next 00/13] AccECN protocol case handling series chia-yu.chang
` (5 preceding siblings ...)
2025-12-01 16:37 ` [PATCH v7 net-next 06/13] tcp: accecn: handle unexpected AccECN negotiation feedback chia-yu.chang
@ 2025-12-01 16:37 ` chia-yu.chang
2025-12-01 16:37 ` [PATCH v7 net-next 08/13] tcp: add TCP_SYNACK_RETRANS synack_type chia-yu.chang
` (6 subsequent siblings)
13 siblings, 0 replies; 17+ messages in thread
From: chia-yu.chang @ 2025-12-01 16:37 UTC (permalink / raw)
To: pabeni, edumazet, parav, linux-doc, corbet, horms, dsahern,
kuniyu, bpf, netdev, dave.taht, jhs, kuba, stephen,
xiyou.wangcong, jiri, davem, andrew+netdev, donald.hunter, ast,
liuhangbin, shuah, linux-kselftest, ij, ncardwell,
koen.de_schepper, g.white, ingemar.s.johansson, mirja.kuehlewind,
cheshire, rs.ietf, Jason_Livingood, vidhi_goel
Cc: Chia-Yu Chang
From: Chia-Yu Chang <chia-yu.chang@nokia-bell-labs.com>
Based on AccECN spec (RFC9768), if the sender of an AccECN SYN
(the TCP Client) times out before receiving the SYN/ACK, it SHOULD
attempt to negotiate the use of AccECN at least one more time by
continuing to set all three TCP ECN flags (AE,CWR,ECE) = (1,1,1) on
the first retransmitted SYN (using the usual retransmission time-outs).
If this first retransmission also fails to be acknowledged, in
deployment scenarios where AccECN path traversal might be problematic,
the TCP Client SHOULD send subsequent retransmissions of the SYN with
the three TCP-ECN flags cleared (AE,CWR,ECE) = (0,0,0).
Signed-off-by: Chia-Yu Chang <chia-yu.chang@nokia-bell-labs.com>
Acked-by: Paolo Abeni <pabeni@redhat.com>
---
v5:
- Update commit message and the if condition statement.
---
net/ipv4/tcp_output.c | 15 +++++++++------
1 file changed, 9 insertions(+), 6 deletions(-)
diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c
index 479afb714bdf..8039c726d235 100644
--- a/net/ipv4/tcp_output.c
+++ b/net/ipv4/tcp_output.c
@@ -3571,12 +3571,15 @@ int __tcp_retransmit_skb(struct sock *sk, struct sk_buff *skb, int segs)
tcp_retrans_try_collapse(sk, skb, avail_wnd);
}
- /* RFC3168, section 6.1.1.1. ECN fallback
- * As AccECN uses the same SYN flags (+ AE), this check covers both
- * cases.
- */
- if ((TCP_SKB_CB(skb)->tcp_flags & TCPHDR_SYN_ECN) == TCPHDR_SYN_ECN)
- tcp_ecn_clear_syn(sk, skb);
+ if (!tcp_ecn_mode_pending(tp) || icsk->icsk_retransmits > 1) {
+ /* RFC3168, section 6.1.1.1. ECN fallback
+ * As AccECN uses the same SYN flags (+ AE), this check
+ * covers both cases.
+ */
+ if ((TCP_SKB_CB(skb)->tcp_flags & TCPHDR_SYN_ECN) ==
+ TCPHDR_SYN_ECN)
+ tcp_ecn_clear_syn(sk, skb);
+ }
/* Update global and local TCP statistics. */
segs = tcp_skb_pcount(skb);
--
2.34.1
^ permalink raw reply related [flat|nested] 17+ messages in thread* [PATCH v7 net-next 08/13] tcp: add TCP_SYNACK_RETRANS synack_type
2025-12-01 16:37 [PATCH v7 net-next 00/13] AccECN protocol case handling series chia-yu.chang
` (6 preceding siblings ...)
2025-12-01 16:37 ` [PATCH v7 net-next 07/13] tcp: accecn: retransmit downgraded SYN in AccECN negotiation chia-yu.chang
@ 2025-12-01 16:37 ` chia-yu.chang
2025-12-01 16:37 ` [PATCH v7 net-next 09/13] tcp: accecn: retransmit SYN/ACK without AccECN option or non-AccECN SYN/ACK chia-yu.chang
` (5 subsequent siblings)
13 siblings, 0 replies; 17+ messages in thread
From: chia-yu.chang @ 2025-12-01 16:37 UTC (permalink / raw)
To: pabeni, edumazet, parav, linux-doc, corbet, horms, dsahern,
kuniyu, bpf, netdev, dave.taht, jhs, kuba, stephen,
xiyou.wangcong, jiri, davem, andrew+netdev, donald.hunter, ast,
liuhangbin, shuah, linux-kselftest, ij, ncardwell,
koen.de_schepper, g.white, ingemar.s.johansson, mirja.kuehlewind,
cheshire, rs.ietf, Jason_Livingood, vidhi_goel
Cc: Chia-Yu Chang
From: Chia-Yu Chang <chia-yu.chang@nokia-bell-labs.com>
Before this patch, retransmitted SYN/ACK did not have a specific synack_type;
however, the upcoming patch needs to distinguish between retransmitted and
non-retransmitted SYN/ACK for AccECN negotiation to transmit the fallback
SYN/ACK during AccECN negotiation. Therefore, this patch introduces a new
synack_type (TCP_SYNACK_RETRANS).
Signed-off-by: Chia-Yu Chang <chia-yu.chang@nokia-bell-labs.com>
Acked-by: Paolo Abeni <pabeni@redhat.com>
---
v6:
- Add new synack_type instead of moving the increment of num_retran.
---
include/net/tcp.h | 1 +
net/ipv4/tcp_output.c | 3 ++-
2 files changed, 3 insertions(+), 1 deletion(-)
diff --git a/include/net/tcp.h b/include/net/tcp.h
index e0a5cf2f7818..607a4cb4984c 100644
--- a/include/net/tcp.h
+++ b/include/net/tcp.h
@@ -541,6 +541,7 @@ enum tcp_synack_type {
TCP_SYNACK_NORMAL,
TCP_SYNACK_FASTOPEN,
TCP_SYNACK_COOKIE,
+ TCP_SYNACK_RETRANS,
};
struct sk_buff *tcp_make_synack(const struct sock *sk, struct dst_entry *dst,
struct request_sock *req,
diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c
index 8039c726d235..5fa14a73d03f 100644
--- a/net/ipv4/tcp_output.c
+++ b/net/ipv4/tcp_output.c
@@ -3921,6 +3921,7 @@ struct sk_buff *tcp_make_synack(const struct sock *sk, struct dst_entry *dst,
switch (synack_type) {
case TCP_SYNACK_NORMAL:
+ case TCP_SYNACK_RETRANS:
skb_set_owner_edemux(skb, req_to_sk(req));
break;
case TCP_SYNACK_COOKIE:
@@ -4606,7 +4607,7 @@ int tcp_rtx_synack(const struct sock *sk, struct request_sock *req)
/* Paired with WRITE_ONCE() in sock_setsockopt() */
if (READ_ONCE(sk->sk_txrehash) == SOCK_TXREHASH_ENABLED)
WRITE_ONCE(tcp_rsk(req)->txhash, net_tx_rndhash());
- res = af_ops->send_synack(sk, NULL, &fl, req, NULL, TCP_SYNACK_NORMAL,
+ res = af_ops->send_synack(sk, NULL, &fl, req, NULL, TCP_SYNACK_RETRANS,
NULL);
if (!res) {
TCP_INC_STATS(sock_net(sk), TCP_MIB_RETRANSSEGS);
--
2.34.1
^ permalink raw reply related [flat|nested] 17+ messages in thread* [PATCH v7 net-next 09/13] tcp: accecn: retransmit SYN/ACK without AccECN option or non-AccECN SYN/ACK
2025-12-01 16:37 [PATCH v7 net-next 00/13] AccECN protocol case handling series chia-yu.chang
` (7 preceding siblings ...)
2025-12-01 16:37 ` [PATCH v7 net-next 08/13] tcp: add TCP_SYNACK_RETRANS synack_type chia-yu.chang
@ 2025-12-01 16:37 ` chia-yu.chang
2025-12-01 16:37 ` [PATCH v7 net-next 10/13] tcp: accecn: unset ECT if receive or send ACE=0 in AccECN negotiaion chia-yu.chang
` (4 subsequent siblings)
13 siblings, 0 replies; 17+ messages in thread
From: chia-yu.chang @ 2025-12-01 16:37 UTC (permalink / raw)
To: pabeni, edumazet, parav, linux-doc, corbet, horms, dsahern,
kuniyu, bpf, netdev, dave.taht, jhs, kuba, stephen,
xiyou.wangcong, jiri, davem, andrew+netdev, donald.hunter, ast,
liuhangbin, shuah, linux-kselftest, ij, ncardwell,
koen.de_schepper, g.white, ingemar.s.johansson, mirja.kuehlewind,
cheshire, rs.ietf, Jason_Livingood, vidhi_goel
Cc: Chia-Yu Chang
From: Chia-Yu Chang <chia-yu.chang@nokia-bell-labs.com>
For Accurate ECN, the first SYN/ACK sent by the TCP server shall set the
ACE flag (see Table 1 of RFC9768) and the AccECN option to complete the
capability negotiation. However, if the TCP server needs to retransmit such
a SYN/ACK (for example, because it did not receive an ACK acknowledging its
SYN/ACK, or received a second SYN requesting AccECN support), the TCP server
retransmits the SYN/ACK without the AccECN option. This is because the
SYN/ACK may be lost due to congestion, or a middlebox may block the AccECN
option. Furthermore, if this retransmission also times out, to expedite
connection establishment, the TCP server should retransmit the SYN/ACK with
(AE,CWR,ECE) = (0,0,0) and without the AccECN option, while maintaining
AccECN feedback mode.
This complies with Section 3.2.3.2.2 of the AccECN specification (RFC9768).
Signed-off-by: Chia-Yu Chang <chia-yu.chang@nokia-bell-labs.com>
---
v7:
- Update comments and use synack_type TCP_SYNACK_RETRANS and num_timeout.
v6:
- Use new synack_type TCP_SYNACK_RETRANS and num_retrans.
---
include/net/tcp_ecn.h | 20 +++++++++++++++-----
net/ipv4/tcp_output.c | 4 ++--
2 files changed, 17 insertions(+), 7 deletions(-)
diff --git a/include/net/tcp_ecn.h b/include/net/tcp_ecn.h
index a709fb1756eb..796c613b5ef3 100644
--- a/include/net/tcp_ecn.h
+++ b/include/net/tcp_ecn.h
@@ -649,12 +649,22 @@ static inline void tcp_ecn_clear_syn(struct sock *sk, struct sk_buff *skb)
}
static inline void
-tcp_ecn_make_synack(const struct request_sock *req, struct tcphdr *th)
+tcp_ecn_make_synack(const struct request_sock *req, struct tcphdr *th,
+ enum tcp_synack_type synack_type)
{
- if (tcp_rsk(req)->accecn_ok)
- tcp_accecn_echo_syn_ect(th, tcp_rsk(req)->syn_ect_rcv);
- else if (inet_rsk(req)->ecn_ok)
- th->ece = 1;
+ /* Accurate ECN shall retransmit SYN/ACK with ACE=0 if the
+ * previously retransmitted SYN/ACK also times out.
+ */
+ if (!req->num_timeout || synack_type != TCP_SYNACK_RETRANS) {
+ if (tcp_rsk(req)->accecn_ok)
+ tcp_accecn_echo_syn_ect(th, tcp_rsk(req)->syn_ect_rcv);
+ else if (inet_rsk(req)->ecn_ok)
+ th->ece = 1;
+ } else if (tcp_rsk(req)->accecn_ok) {
+ th->ae = 0;
+ th->cwr = 0;
+ th->ece = 0;
+ }
}
static inline bool tcp_accecn_option_beacon_check(const struct sock *sk)
diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c
index 5fa14a73d03f..c6754854ad09 100644
--- a/net/ipv4/tcp_output.c
+++ b/net/ipv4/tcp_output.c
@@ -1106,7 +1106,7 @@ static unsigned int tcp_synack_options(const struct sock *sk,
if (treq->accecn_ok &&
READ_ONCE(sock_net(sk)->ipv4.sysctl_tcp_ecn_option) &&
- req->num_timeout < 1 && remaining >= TCPOLEN_ACCECN_BASE) {
+ synack_type != TCP_SYNACK_RETRANS && remaining >= TCPOLEN_ACCECN_BASE) {
opts->use_synack_ecn_bytes = 1;
remaining -= tcp_options_fit_accecn(opts, 0, remaining);
}
@@ -4004,7 +4004,7 @@ struct sk_buff *tcp_make_synack(const struct sock *sk, struct dst_entry *dst,
memset(th, 0, sizeof(struct tcphdr));
th->syn = 1;
th->ack = 1;
- tcp_ecn_make_synack(req, th);
+ tcp_ecn_make_synack(req, th, synack_type);
th->source = htons(ireq->ir_num);
th->dest = ireq->ir_rmt_port;
skb->mark = ireq->ir_mark;
--
2.34.1
^ permalink raw reply related [flat|nested] 17+ messages in thread* [PATCH v7 net-next 10/13] tcp: accecn: unset ECT if receive or send ACE=0 in AccECN negotiaion
2025-12-01 16:37 [PATCH v7 net-next 00/13] AccECN protocol case handling series chia-yu.chang
` (8 preceding siblings ...)
2025-12-01 16:37 ` [PATCH v7 net-next 09/13] tcp: accecn: retransmit SYN/ACK without AccECN option or non-AccECN SYN/ACK chia-yu.chang
@ 2025-12-01 16:37 ` chia-yu.chang
2025-12-01 16:37 ` [PATCH v7 net-next 11/13] tcp: accecn: fallback outgoing half link to non-AccECN chia-yu.chang
` (3 subsequent siblings)
13 siblings, 0 replies; 17+ messages in thread
From: chia-yu.chang @ 2025-12-01 16:37 UTC (permalink / raw)
To: pabeni, edumazet, parav, linux-doc, corbet, horms, dsahern,
kuniyu, bpf, netdev, dave.taht, jhs, kuba, stephen,
xiyou.wangcong, jiri, davem, andrew+netdev, donald.hunter, ast,
liuhangbin, shuah, linux-kselftest, ij, ncardwell,
koen.de_schepper, g.white, ingemar.s.johansson, mirja.kuehlewind,
cheshire, rs.ietf, Jason_Livingood, vidhi_goel
Cc: Chia-Yu Chang
From: Chia-Yu Chang <chia-yu.chang@nokia-bell-labs.com>
Based on specification:
https://tools.ietf.org/id/draft-ietf-tcpm-accurate-ecn-28.txt
Based on Section 3.1.5 of AccECN spec (RFC9768), a TCP Server in
AccECN mode MUST NOT set ECT on any packet for the rest of the connection,
if it has received or sent at least one valid SYN or Acceptable SYN/ACK
with (AE,CWR,ECE) = (0,0,0) during the handshake.
In addition, a host in AccECN mode that is feeding back the IP-ECN
field on a SYN or SYN/ACK MUST feed back the IP-ECN field on the
latest valid SYN or acceptable SYN/ACK to arrive.
Signed-off-by: Chia-Yu Chang <chia-yu.chang@nokia-bell-labs.com>
---
v6:
- Do not cast const struct request_sock into struct request_sock
- Set tcp_accecn_fail_mode after calling tcp_rtx_synack().
---
net/ipv4/inet_connection_sock.c | 4 ++++
net/ipv4/tcp_input.c | 2 ++
net/ipv4/tcp_minisocks.c | 39 ++++++++++++++++++++++++---------
net/ipv4/tcp_output.c | 3 ++-
net/ipv4/tcp_timer.c | 3 +++
5 files changed, 40 insertions(+), 11 deletions(-)
diff --git a/net/ipv4/inet_connection_sock.c b/net/ipv4/inet_connection_sock.c
index 97d57c52b9ad..c994a00c83d2 100644
--- a/net/ipv4/inet_connection_sock.c
+++ b/net/ipv4/inet_connection_sock.c
@@ -20,6 +20,7 @@
#include <net/tcp_states.h>
#include <net/xfrm.h>
#include <net/tcp.h>
+#include <net/tcp_ecn.h>
#include <net/sock_reuseport.h>
#include <net/addrconf.h>
@@ -1103,6 +1104,9 @@ static void reqsk_timer_handler(struct timer_list *t)
(!resend ||
!tcp_rtx_synack(sk_listener, req) ||
inet_rsk(req)->acked)) {
+ if (req->num_retrans > 1 && tcp_rsk(req)->accecn_ok)
+ tcp_accecn_fail_mode_set(tcp_sk(sk_listener),
+ TCP_ACCECN_ACE_FAIL_SEND);
if (req->num_timeout++ == 0)
atomic_dec(&queue->young);
mod_timer(&req->rsk_timer, jiffies + tcp_reqsk_timeout(req));
diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index e5c9cf586437..db361daebff8 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -6240,6 +6240,8 @@ static bool tcp_validate_incoming(struct sock *sk, struct sk_buff *skb,
if (th->syn) {
if (tcp_ecn_mode_accecn(tp)) {
accecn_reflector = true;
+ tp->syn_ect_rcv = TCP_SKB_CB(skb)->ip_dsfield &
+ INET_ECN_MASK;
if (tp->rx_opt.accecn &&
tp->saw_accecn_opt < TCP_ACCECN_OPT_COUNTER_SEEN) {
u8 saw_opt = tcp_accecn_option_init(skb, tp->rx_opt.accecn);
diff --git a/net/ipv4/tcp_minisocks.c b/net/ipv4/tcp_minisocks.c
index 9776c921d1bb..ff573fad465f 100644
--- a/net/ipv4/tcp_minisocks.c
+++ b/net/ipv4/tcp_minisocks.c
@@ -749,16 +749,35 @@ struct sock *tcp_check_req(struct sock *sk, struct sk_buff *skb,
*/
if (!tcp_oow_rate_limited(sock_net(sk), skb,
LINUX_MIB_TCPACKSKIPPEDSYNRECV,
- &tcp_rsk(req)->last_oow_ack_time) &&
-
- !tcp_rtx_synack(sk, req)) {
- unsigned long expires = jiffies;
-
- expires += tcp_reqsk_timeout(req);
- if (!fastopen)
- mod_timer_pending(&req->rsk_timer, expires);
- else
- req->rsk_timer.expires = expires;
+ &tcp_rsk(req)->last_oow_ack_time)) {
+ if (tcp_rsk(req)->accecn_ok) {
+ u8 ect_rcv = TCP_SKB_CB(skb)->ip_dsfield &
+ INET_ECN_MASK;
+
+ tcp_rsk(req)->syn_ect_rcv = ect_rcv;
+ if (tcp_accecn_ace(tcp_hdr(skb)) == 0x0) {
+ u8 fail_mode = TCP_ACCECN_ACE_FAIL_RECV;
+
+ tcp_accecn_fail_mode_set(tcp_sk(sk),
+ fail_mode);
+ }
+ }
+ if (!tcp_rtx_synack(sk, req)) {
+ u8 fail_mode = TCP_ACCECN_ACE_FAIL_SEND;
+ unsigned long expires = jiffies;
+
+ if (req->num_retrans > 1 &&
+ tcp_rsk(req)->accecn_ok)
+ tcp_accecn_fail_mode_set(tcp_sk(sk),
+ fail_mode);
+
+ expires += tcp_reqsk_timeout(req);
+ if (!fastopen)
+ mod_timer_pending(&req->rsk_timer,
+ expires);
+ else
+ req->rsk_timer.expires = expires;
+ }
}
return NULL;
}
diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c
index c6754854ad09..9489cda7322e 100644
--- a/net/ipv4/tcp_output.c
+++ b/net/ipv4/tcp_output.c
@@ -334,7 +334,8 @@ static void tcp_ecn_send(struct sock *sk, struct sk_buff *skb,
return;
if (tcp_ecn_mode_accecn(tp)) {
- if (!tcp_accecn_ace_fail_recv(tp))
+ if (!tcp_accecn_ace_fail_recv(tp) &&
+ !tcp_accecn_ace_fail_send(tp))
INET_ECN_xmit(sk);
tcp_accecn_set_ace(tp, skb, th);
skb_shinfo(skb)->gso_type |= SKB_GSO_TCP_ACCECN;
diff --git a/net/ipv4/tcp_timer.c b/net/ipv4/tcp_timer.c
index 160080c9021d..8ef2f105cc21 100644
--- a/net/ipv4/tcp_timer.c
+++ b/net/ipv4/tcp_timer.c
@@ -22,6 +22,7 @@
#include <linux/module.h>
#include <linux/gfp.h>
#include <net/tcp.h>
+#include <net/tcp_ecn.h>
#include <net/rstreason.h>
static u32 tcp_clamp_rto_to_user_timeout(const struct sock *sk)
@@ -479,6 +480,8 @@ static void tcp_fastopen_synack_timer(struct sock *sk, struct request_sock *req)
* it's not good to give up too easily.
*/
tcp_rtx_synack(sk, req);
+ if (req->num_retrans > 1 && tcp_rsk(req)->accecn_ok)
+ tcp_accecn_fail_mode_set(tcp_sk(sk), TCP_ACCECN_ACE_FAIL_SEND);
req->num_timeout++;
tcp_update_rto_stats(sk);
if (!tp->retrans_stamp)
--
2.34.1
^ permalink raw reply related [flat|nested] 17+ messages in thread* [PATCH v7 net-next 11/13] tcp: accecn: fallback outgoing half link to non-AccECN
2025-12-01 16:37 [PATCH v7 net-next 00/13] AccECN protocol case handling series chia-yu.chang
` (9 preceding siblings ...)
2025-12-01 16:37 ` [PATCH v7 net-next 10/13] tcp: accecn: unset ECT if receive or send ACE=0 in AccECN negotiaion chia-yu.chang
@ 2025-12-01 16:37 ` chia-yu.chang
2025-12-01 16:37 ` [PATCH v7 net-next 12/13] tcp: accecn: detect loss ACK w/ AccECN option and add TCP_ACCECN_OPTION_PERSIST chia-yu.chang
` (2 subsequent siblings)
13 siblings, 0 replies; 17+ messages in thread
From: chia-yu.chang @ 2025-12-01 16:37 UTC (permalink / raw)
To: pabeni, edumazet, parav, linux-doc, corbet, horms, dsahern,
kuniyu, bpf, netdev, dave.taht, jhs, kuba, stephen,
xiyou.wangcong, jiri, davem, andrew+netdev, donald.hunter, ast,
liuhangbin, shuah, linux-kselftest, ij, ncardwell,
koen.de_schepper, g.white, ingemar.s.johansson, mirja.kuehlewind,
cheshire, rs.ietf, Jason_Livingood, vidhi_goel
Cc: Chia-Yu Chang
From: Chia-Yu Chang <chia-yu.chang@nokia-bell-labs.com>
According to Section 3.2.2.1 of AccECN spec (RFC9768), if the Server
is in AccECN mode and in SYN-RCVD state, and if it receives a value of
zero on a pure ACK with SYN=0 and no SACK blocks, for the rest of the
connection the Server MUST NOT set ECT on outgoing packets and MUST
NOT respond to AccECN feedback. Nonetheless, as a Data Receiver it
MUST NOT disable AccECN feedback.
Signed-off-by: Chia-Yu Chang <chia-yu.chang@nokia-bell-labs.com>
Acked-by: Paolo Abeni <pabeni@redhat.com>
---
v3:
- Remove unnecessary brackets.
---
include/net/tcp_ecn.h | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/include/net/tcp_ecn.h b/include/net/tcp_ecn.h
index 796c613b5ef3..49e0b865fe02 100644
--- a/include/net/tcp_ecn.h
+++ b/include/net/tcp_ecn.h
@@ -175,7 +175,9 @@ static inline void tcp_accecn_third_ack(struct sock *sk,
switch (ace) {
case 0x0:
/* Invalid value */
- tcp_accecn_fail_mode_set(tp, TCP_ACCECN_ACE_FAIL_RECV);
+ if (!TCP_SKB_CB(skb)->sacked)
+ tcp_accecn_fail_mode_set(tp, TCP_ACCECN_ACE_FAIL_RECV |
+ TCP_ACCECN_OPT_FAIL_RECV);
break;
case 0x7:
case 0x5:
--
2.34.1
^ permalink raw reply related [flat|nested] 17+ messages in thread* [PATCH v7 net-next 12/13] tcp: accecn: detect loss ACK w/ AccECN option and add TCP_ACCECN_OPTION_PERSIST
2025-12-01 16:37 [PATCH v7 net-next 00/13] AccECN protocol case handling series chia-yu.chang
` (10 preceding siblings ...)
2025-12-01 16:37 ` [PATCH v7 net-next 11/13] tcp: accecn: fallback outgoing half link to non-AccECN chia-yu.chang
@ 2025-12-01 16:37 ` chia-yu.chang
2025-12-01 16:38 ` [PATCH v7 net-next 13/13] tcp: accecn: enable AccECN chia-yu.chang
2025-12-01 23:05 ` [PATCH v7 net-next 00/13] AccECN protocol case handling series Jakub Kicinski
13 siblings, 0 replies; 17+ messages in thread
From: chia-yu.chang @ 2025-12-01 16:37 UTC (permalink / raw)
To: pabeni, edumazet, parav, linux-doc, corbet, horms, dsahern,
kuniyu, bpf, netdev, dave.taht, jhs, kuba, stephen,
xiyou.wangcong, jiri, davem, andrew+netdev, donald.hunter, ast,
liuhangbin, shuah, linux-kselftest, ij, ncardwell,
koen.de_schepper, g.white, ingemar.s.johansson, mirja.kuehlewind,
cheshire, rs.ietf, Jason_Livingood, vidhi_goel
Cc: Chia-Yu Chang
From: Chia-Yu Chang <chia-yu.chang@nokia-bell-labs.com>
Detect spurious retransmission of a previously sent ACK carrying the
AccECN option after the second retransmission. Since this might be caused
by the middlebox dropping ACK with options it does not recognize, disable
the sending of the AccECN option in all subsequent ACKs. This patch
follows Section 3.2.3.2.2 of AccECN spec (RFC9768).
Also, a new AccECN option sending mode is added to tcp_ecn_option sysctl:
(TCP_ECN_OPTION_PERSIST), which ignores the AccECN fallback policy and
persistently sends AccECN option once it fits into TCP option space.
Signed-off-by: Chia-Yu Chang <chia-yu.chang@nokia-bell-labs.com>
Acked-by: Paolo Abeni <pabeni@redhat.com>
---
v5:
- Add empty line between variable declarations and code
---
Documentation/networking/ip-sysctl.rst | 4 +++-
include/linux/tcp.h | 3 ++-
include/net/tcp_ecn.h | 2 ++
net/ipv4/sysctl_net_ipv4.c | 2 +-
net/ipv4/tcp_input.c | 10 ++++++++++
net/ipv4/tcp_output.c | 7 ++++++-
6 files changed, 24 insertions(+), 4 deletions(-)
diff --git a/Documentation/networking/ip-sysctl.rst b/Documentation/networking/ip-sysctl.rst
index bc9a01606daf..28c7e4f5ecf9 100644
--- a/Documentation/networking/ip-sysctl.rst
+++ b/Documentation/networking/ip-sysctl.rst
@@ -482,7 +482,9 @@ tcp_ecn_option - INTEGER
1 Send AccECN option sparingly according to the minimum option
rules outlined in draft-ietf-tcpm-accurate-ecn.
2 Send AccECN option on every packet whenever it fits into TCP
- option space.
+ option space except when AccECN fallback is triggered.
+ 3 Send AccECN option on every packet whenever it fits into TCP
+ option space even when AccECN fallback is triggered.
= ============================================================
Default: 2
diff --git a/include/linux/tcp.h b/include/linux/tcp.h
index 683f38362977..32b031d09294 100644
--- a/include/linux/tcp.h
+++ b/include/linux/tcp.h
@@ -294,7 +294,8 @@ struct tcp_sock {
u8 nonagle : 4,/* Disable Nagle algorithm? */
rate_app_limited:1; /* rate_{delivered,interval_us} limited? */
u8 received_ce_pending:4, /* Not yet transmit cnt of received_ce */
- unused2:4;
+ accecn_opt_sent:1,/* Sent AccECN option in previous ACK */
+ unused2:3;
u8 accecn_minlen:2,/* Minimum length of AccECN option sent */
est_ecnfield:2,/* ECN field for AccECN delivered estimates */
accecn_opt_demand:2,/* Demand AccECN option for n next ACKs */
diff --git a/include/net/tcp_ecn.h b/include/net/tcp_ecn.h
index 49e0b865fe02..bbf29f3e4fc5 100644
--- a/include/net/tcp_ecn.h
+++ b/include/net/tcp_ecn.h
@@ -29,6 +29,7 @@ enum tcp_accecn_option {
TCP_ACCECN_OPTION_DISABLED = 0,
TCP_ACCECN_OPTION_MINIMUM = 1,
TCP_ACCECN_OPTION_FULL = 2,
+ TCP_ACCECN_OPTION_PERSIST = 3,
};
/* Apply either ECT(0) or ECT(1) based on TCP_CONG_ECT_1_NEGOTIATION flag */
@@ -406,6 +407,7 @@ static inline void tcp_accecn_init_counters(struct tcp_sock *tp)
tp->received_ce_pending = 0;
__tcp_accecn_init_bytes_counters(tp->received_ecn_bytes);
__tcp_accecn_init_bytes_counters(tp->delivered_ecn_bytes);
+ tp->accecn_opt_sent = 0;
tp->accecn_minlen = 0;
tp->accecn_opt_demand = 0;
tp->est_ecnfield = 0;
diff --git a/net/ipv4/sysctl_net_ipv4.c b/net/ipv4/sysctl_net_ipv4.c
index a1a50a5c80dc..385b5b986d23 100644
--- a/net/ipv4/sysctl_net_ipv4.c
+++ b/net/ipv4/sysctl_net_ipv4.c
@@ -749,7 +749,7 @@ static struct ctl_table ipv4_net_table[] = {
.mode = 0644,
.proc_handler = proc_dou8vec_minmax,
.extra1 = SYSCTL_ZERO,
- .extra2 = SYSCTL_TWO,
+ .extra2 = SYSCTL_THREE,
},
{
.procname = "tcp_ecn_option_beacon",
diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index db361daebff8..2aae397b9d66 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -4819,6 +4819,8 @@ static void tcp_dsack_extend(struct sock *sk, u32 seq, u32 end_seq)
static void tcp_rcv_spurious_retrans(struct sock *sk, const struct sk_buff *skb)
{
+ struct tcp_sock *tp = tcp_sk(sk);
+
/* When the ACK path fails or drops most ACKs, the sender would
* timeout and spuriously retransmit the same segment repeatedly.
* If it seems our ACKs are not reaching the other side,
@@ -4838,6 +4840,14 @@ static void tcp_rcv_spurious_retrans(struct sock *sk, const struct sk_buff *skb)
/* Save last flowlabel after a spurious retrans. */
tcp_save_lrcv_flowlabel(sk, skb);
#endif
+ /* Check DSACK info to detect that the previous ACK carrying the
+ * AccECN option was lost after the second retransmision, and then
+ * stop sending AccECN option in all subsequent ACKs.
+ */
+ if (tcp_ecn_mode_accecn(tp) &&
+ TCP_SKB_CB(skb)->seq == tp->duplicate_sack[0].start_seq &&
+ tp->accecn_opt_sent)
+ tcp_accecn_fail_mode_set(tp, TCP_ACCECN_OPT_FAIL_SEND);
}
static void tcp_send_dupack(struct sock *sk, const struct sk_buff *skb)
diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c
index 9489cda7322e..63c7f448037a 100644
--- a/net/ipv4/tcp_output.c
+++ b/net/ipv4/tcp_output.c
@@ -713,9 +713,12 @@ static void tcp_options_write(struct tcphdr *th, struct tcp_sock *tp,
if (tp) {
tp->accecn_minlen = 0;
tp->accecn_opt_tstamp = tp->tcp_mstamp;
+ tp->accecn_opt_sent = 1;
if (tp->accecn_opt_demand)
tp->accecn_opt_demand--;
}
+ } else if (tp) {
+ tp->accecn_opt_sent = 0;
}
if (unlikely(OPTION_SACK_ADVERTISE & options)) {
@@ -1187,7 +1190,9 @@ static unsigned int tcp_established_options(struct sock *sk, struct sk_buff *skb
if (tcp_ecn_mode_accecn(tp)) {
int ecn_opt = READ_ONCE(sock_net(sk)->ipv4.sysctl_tcp_ecn_option);
- if (ecn_opt && tp->saw_accecn_opt && !tcp_accecn_opt_fail_send(tp) &&
+ if (ecn_opt && tp->saw_accecn_opt &&
+ (ecn_opt >= TCP_ACCECN_OPTION_PERSIST ||
+ !tcp_accecn_opt_fail_send(tp)) &&
(ecn_opt >= TCP_ACCECN_OPTION_FULL || tp->accecn_opt_demand ||
tcp_accecn_option_beacon_check(sk))) {
opts->use_synack_ecn_bytes = 0;
--
2.34.1
^ permalink raw reply related [flat|nested] 17+ messages in thread* [PATCH v7 net-next 13/13] tcp: accecn: enable AccECN
2025-12-01 16:37 [PATCH v7 net-next 00/13] AccECN protocol case handling series chia-yu.chang
` (11 preceding siblings ...)
2025-12-01 16:37 ` [PATCH v7 net-next 12/13] tcp: accecn: detect loss ACK w/ AccECN option and add TCP_ACCECN_OPTION_PERSIST chia-yu.chang
@ 2025-12-01 16:38 ` chia-yu.chang
2025-12-01 23:05 ` [PATCH v7 net-next 00/13] AccECN protocol case handling series Jakub Kicinski
13 siblings, 0 replies; 17+ messages in thread
From: chia-yu.chang @ 2025-12-01 16:38 UTC (permalink / raw)
To: pabeni, edumazet, parav, linux-doc, corbet, horms, dsahern,
kuniyu, bpf, netdev, dave.taht, jhs, kuba, stephen,
xiyou.wangcong, jiri, davem, andrew+netdev, donald.hunter, ast,
liuhangbin, shuah, linux-kselftest, ij, ncardwell,
koen.de_schepper, g.white, ingemar.s.johansson, mirja.kuehlewind,
cheshire, rs.ietf, Jason_Livingood, vidhi_goel
Cc: Chia-Yu Chang
From: Chia-Yu Chang <chia-yu.chang@nokia-bell-labs.com>
Enable Accurate ECN negotiation and request for incoming and
outgoing connection by setting sysctl_tcp_ecn:
+==============+===========================================+
| | Highest ECN variant (Accurate ECN, ECN, |
| tcp_ecn | or no ECN) to be negotiated & requested |
| +---------------------+---------------------+
| | Incoming connection | Outgoing connection |
+==============+=====================+=====================+
| 0 | No ECN | No ECN |
| 1 | ECN | ECN |
| 2 | ECN | No ECN |
+--------------+---------------------+---------------------+
| 3 | Accurate ECN | Accurate ECN |
| 4 | Accurate ECN | ECN |
| 5 | Accurate ECN | No ECN |
+==============+=====================+=====================+
Refer Documentation/networking/ip-sysctl.rst for more details.
Signed-off-by: Chia-Yu Chang <chia-yu.chang@nokia-bell-labs.com>
Acked-by: Paolo Abeni <pabeni@redhat.com>
---
net/ipv4/sysctl_net_ipv4.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/net/ipv4/sysctl_net_ipv4.c b/net/ipv4/sysctl_net_ipv4.c
index 385b5b986d23..643763bc2142 100644
--- a/net/ipv4/sysctl_net_ipv4.c
+++ b/net/ipv4/sysctl_net_ipv4.c
@@ -47,7 +47,7 @@ static unsigned int udp_child_hash_entries_max = UDP_HTABLE_SIZE_MAX;
static int tcp_plb_max_rounds = 31;
static int tcp_plb_max_cong_thresh = 256;
static unsigned int tcp_tw_reuse_delay_max = TCP_PAWS_MSL * MSEC_PER_SEC;
-static int tcp_ecn_mode_max = 2;
+static int tcp_ecn_mode_max = 5;
static u32 icmp_errors_extension_mask_all =
GENMASK_U8(ICMP_ERR_EXT_COUNT - 1, 0);
--
2.34.1
^ permalink raw reply related [flat|nested] 17+ messages in thread* Re: [PATCH v7 net-next 00/13] AccECN protocol case handling series
2025-12-01 16:37 [PATCH v7 net-next 00/13] AccECN protocol case handling series chia-yu.chang
` (12 preceding siblings ...)
2025-12-01 16:38 ` [PATCH v7 net-next 13/13] tcp: accecn: enable AccECN chia-yu.chang
@ 2025-12-01 23:05 ` Jakub Kicinski
2025-12-03 16:58 ` Neal Cardwell
13 siblings, 1 reply; 17+ messages in thread
From: Jakub Kicinski @ 2025-12-01 23:05 UTC (permalink / raw)
To: chia-yu.chang
Cc: pabeni, edumazet, parav, linux-doc, corbet, horms, dsahern,
kuniyu, bpf, netdev, dave.taht, jhs, stephen, xiyou.wangcong,
jiri, davem, andrew+netdev, donald.hunter, ast, liuhangbin, shuah,
linux-kselftest, ij, ncardwell, koen.de_schepper, g.white,
ingemar.s.johansson, mirja.kuehlewind, cheshire, rs.ietf,
Jason_Livingood, vidhi_goel
On Mon, 1 Dec 2025 17:37:47 +0100 chia-yu.chang@nokia-bell-labs.com
wrote:
> Plesae find the v7 AccECN case handling patch series, which covers
> several excpetional case handling of Accurate ECN spec (RFC9768),
> adds new identifiers to be used by CC modules, adds ecn_delta into
> rate_sample, and keeps the ACE counter for computation, etc.
>
> This patch series is part of the full AccECN patch series, which is available at
> https://github.com/L4STeam/linux-net-next/commits/upstream_l4steam/
Linux tagged 6.18 final, so net-next is closed for new code submissions
per the announcement at
https://lore.kernel.org/20251130174502.3908e3ee@kernel.org
--
pw-bot: defer
^ permalink raw reply [flat|nested] 17+ messages in thread* Re: [PATCH v7 net-next 00/13] AccECN protocol case handling series
2025-12-01 23:05 ` [PATCH v7 net-next 00/13] AccECN protocol case handling series Jakub Kicinski
@ 2025-12-03 16:58 ` Neal Cardwell
2025-12-11 15:54 ` Chia-Yu Chang (Nokia)
0 siblings, 1 reply; 17+ messages in thread
From: Neal Cardwell @ 2025-12-03 16:58 UTC (permalink / raw)
To: Jakub Kicinski
Cc: chia-yu.chang, pabeni, edumazet, parav, linux-doc, corbet, horms,
dsahern, kuniyu, bpf, netdev, dave.taht, jhs, stephen,
xiyou.wangcong, jiri, davem, andrew+netdev, donald.hunter, ast,
liuhangbin, shuah, linux-kselftest, ij, koen.de_schepper, g.white,
ingemar.s.johansson, mirja.kuehlewind, cheshire, rs.ietf,
Jason_Livingood, vidhi_goel
On Mon, Dec 1, 2025 at 6:05 PM Jakub Kicinski <kuba@kernel.org> wrote:
>
> On Mon, 1 Dec 2025 17:37:47 +0100 chia-yu.chang@nokia-bell-labs.com
> wrote:
> > Plesae find the v7 AccECN case handling patch series, which covers
> > several excpetional case handling of Accurate ECN spec (RFC9768),
> > adds new identifiers to be used by CC modules, adds ecn_delta into
> > rate_sample, and keeps the ACE counter for computation, etc.
> >
> > This patch series is part of the full AccECN patch series, which is available at
> > https://github.com/L4STeam/linux-net-next/commits/upstream_l4steam/
Hi Chia-Yu,
My understanding is that you still have a set of packetdrill tests you
have been using to test this AccECN patch series. For the Linux
networking stack, the recent best practice for a significant patch
series like this is to add packetdrill tests to the
tools/testing/selftests/net/packetdrill/ directory as a separate
commit in the patch series.
For a recent example, see:
selftest: packetdrill: Add max RTO test for SYN+ACK.
https://git.kernel.org/pub/scm/linux/kernel/git/netdev/net-next.git/commit/?id=ffc56c90819e86d3a8c4eff6f831317d1c1476b6
When you next post the AccECN patch series for review, can you please
include a patch at the end of the series that posts your packetdrill
tests in the tools/testing/selftests/net/packetdrill/ directory? In
the commit description for that patch, please include a mention of the
packetdrill SHA1 you are using and a link to the packetdrill branch
you are using, somewhere on github or similar. Then I will look into
merging any packetdrill tool changes that you are depending on, if
there are packetdrill commits that you depend on that I have not
merged into packetdrill yet.
Thanks!
neal
^ permalink raw reply [flat|nested] 17+ messages in thread
* RE: [PATCH v7 net-next 00/13] AccECN protocol case handling series
2025-12-03 16:58 ` Neal Cardwell
@ 2025-12-11 15:54 ` Chia-Yu Chang (Nokia)
0 siblings, 0 replies; 17+ messages in thread
From: Chia-Yu Chang (Nokia) @ 2025-12-11 15:54 UTC (permalink / raw)
To: Neal Cardwell, Jakub Kicinski
Cc: pabeni@redhat.com, edumazet@google.com, parav@nvidia.com,
linux-doc@vger.kernel.org, corbet@lwn.net, horms@kernel.org,
dsahern@kernel.org, kuniyu@google.com, bpf@vger.kernel.org,
netdev@vger.kernel.org, dave.taht@gmail.com, jhs@mojatatu.com,
stephen@networkplumber.org, xiyou.wangcong@gmail.com,
jiri@resnulli.us, davem@davemloft.net, andrew+netdev@lunn.ch,
donald.hunter@gmail.com, ast@fiberby.net, liuhangbin@gmail.com,
shuah@kernel.org, linux-kselftest@vger.kernel.org, ij@kernel.org,
Koen De Schepper (Nokia), g.white@cablelabs.com,
ingemar.s.johansson@ericsson.com, mirja.kuehlewind@ericsson.com,
cheshire, rs.ietf@gmx.at, Jason_Livingood@comcast.com, Vidhi Goel
> -----Original Message-----
> From: Neal Cardwell <ncardwell@google.com>
> Sent: Wednesday, December 3, 2025 5:59 PM
> To: Jakub Kicinski <kuba@kernel.org>
> Cc: Chia-Yu Chang (Nokia) <chia-yu.chang@nokia-bell-labs.com>; pabeni@redhat.com; edumazet@google.com; parav@nvidia.com; linux-doc@vger.kernel.org; corbet@lwn.net; horms@kernel.org; dsahern@kernel.org; kuniyu@google.com; bpf@vger.kernel.org; netdev@vger.kernel.org; dave.taht@gmail.com; jhs@mojatatu.com; stephen@networkplumber.org; xiyou.wangcong@gmail.com; jiri@resnulli.us; davem@davemloft.net; andrew+netdev@lunn.ch; donald.hunter@gmail.com; ast@fiberby.net; liuhangbin@gmail.com; shuah@kernel.org; linux-kselftest@vger.kernel.org; ij@kernel.org; Koen De Schepper (Nokia) <koen.de_schepper@nokia-bell-labs.com>; g.white@cablelabs.com; ingemar.s.johansson@ericsson.com; mirja.kuehlewind@ericsson.com; cheshire <cheshire@apple.com>; rs.ietf@gmx.at; Jason_Livingood@comcast.com; Vidhi Goel <vidhi_goel@apple.com>
> Subject: Re: [PATCH v7 net-next 00/13] AccECN protocol case handling series
>
>
> CAUTION: This is an external email. Please be very careful when clicking links or opening attachments. See the URL nok.it/ext for additional information.
>
>
>
> On Mon, Dec 1, 2025 at 6:05 PM Jakub Kicinski <kuba@kernel.org> wrote:
> >
> > On Mon, 1 Dec 2025 17:37:47 +0100 chia-yu.chang@nokia-bell-labs.com
> > wrote:
> > > Plesae find the v7 AccECN case handling patch series, which covers
> > > several excpetional case handling of Accurate ECN spec (RFC9768),
> > > adds new identifiers to be used by CC modules, adds ecn_delta into
> > > rate_sample, and keeps the ACE counter for computation, etc.
> > >
> > > This patch series is part of the full AccECN patch series, which is
> > > available at
> > > https://gi/
> > > thub.com%2FL4STeam%2Flinux-net-next%2Fcommits%2Fupstream_l4steam%2F&
> > > data=05%7C02%7Cchia-yu.chang%40nokia-bell-labs.com%7C31949326090d451
> > > 8264708de328d3cef%7C5d4717519675428d917b70f44f9630b0%7C0%7C0%7C63900
> > > 3779346421209%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOi
> > > IwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7
> > > C%7C%7C&sdata=J%2FR5DZJ1XFpY9rRlZOygChTspEEm68zYnaA1ZVYHhrc%3D&reser
> > > ved=0
>
> Hi Chia-Yu,
>
> My understanding is that you still have a set of packetdrill tests you have been using to test this AccECN patch series. For the Linux networking stack, the recent best practice for a significant patch series like this is to add packetdrill tests to the tools/testing/selftests/net/packetdrill/ directory as a separate commit in the patch series.
>
> For a recent example, see:
>
> selftest: packetdrill: Add max RTO test for SYN+ACK.
> https://git.kernel.org/pub/scm/linux/kernel/git/netdev/net-next.git/commit/?id=ffc56c90819e86d3a8c4eff6f831317d1c1476b6
>
> When you next post the AccECN patch series for review, can you please include a patch at the end of the series that posts your packetdrill tests in the tools/testing/selftests/net/packetdrill/ directory? In the commit description for that patch, please include a mention of the packetdrill SHA1 you are using and a link to the packetdrill branch you are using, somewhere on github or similar. Then I will look into merging any packetdrill tool changes that you are depending on, if there are packetdrill commits that you depend on that I have not merged into packetdrill yet.
>
> Thanks!
> neal
Hi Neal,
Sure, I will include packetdrill in a separated patch.
But do you know when net-next will be open for submission? Thanks.
Chia-Yu
^ permalink raw reply [flat|nested] 17+ messages in thread