By default, the Linux TCP implementation does not shrink the
advertised window (RFC 7323 calls this "window retraction") with the
following exceptions:
- When an incoming segment cannot be added due to the receive buffer
running out of memory. Since commit 8c670bdfa5 ("tcp: correct
handling of extreme memory squeeze") a zero window will be
advertised in this case. It turns out that reaching the required
memory pressure is easy when window scaling is in use. In the
simplest case, sending a sufficient number of segments smaller than
the scale factor to a receiver that does not read data is enough.
- Commit b650d953cd ("tcp: enforce receive buffer memory limits by
allowing the tcp window to shrink") addressed the "eating memory"
problem by introducing a sysctl knob that allows shrinking the
window before running out of memory.
However, RFC 7323 does not only state that shrinking the window is
necessary in some cases, it also formulates requirements for TCP
implementations when doing so (Section 2.4).
This commit addresses the receiver-side requirements: After retracting
the window, the peer may have a snd_nxt that lies within a previously
advertised window but is now beyond the retracted window. This means
that all incoming segments (including pure ACKs) will be rejected
until the application happens to read enough data to let the peer's
snd_nxt be in window again (which may be never).
To comply with RFC 7323, the receiver MUST honor any segment that
would have been in window for any ACK sent by the receiver and, when
window scaling is in effect, SHOULD track the maximum window sequence
number it has advertised. This patch tracks that maximum window
sequence number rcv_mwnd_seq throughout the connection and uses it in
tcp_sequence() when deciding whether a segment is acceptable.
rcv_mwnd_seq is updated together with rcv_wup and rcv_wnd in
tcp_select_window(). If we count tcp_sequence() as fast path, it is
read in the fast path. Therefore, rcv_mwnd_seq is put into rcv_wnd's
cacheline group.
The logic for handling received data in tcp_data_queue() is already
sufficient and does not need to be updated.
Signed-off-by: Simon Baatz <gmbnomis@gmail.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://patch.msgid.link/20260309-tcp_rfc7323_retract_wnd_rfc-v3-1-4c7f96b1ec69@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Majority of tcp_release_cb() calls do nothing at all.
Provide tcp_release_cb_cond() helper so that release_sock()
can avoid these calls.
Also hint the compiler that __release_sock() and wake_up()
are rarely called.
$ scripts/bloat-o-meter -t vmlinux.old vmlinux.new
add/remove: 0/0 grow/shrink: 0/1 up/down: 0/-77 (-77)
Function old new delta
release_sock 258 181 -77
Total: Before=25235790, After=25235713, chg -0.00%
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Kuniyuki Iwashima <kuniyu@google.com>
Link: https://patch.msgid.link/20260310124451.2280968-1-edumazet@google.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
chrono_type is currently in tcp_sock_read_txrx group, which
is supposed to hold read-mostly fields.
But chrono_type is mostly written in tx path, it should
be moved to tcp_sock_write_tx group, close to other
chrono fields (chrono_stat[], chrono_start).
Note this adds holes, but data locality is far more important.
Use a full u8 for the time being, compiler can generate
more efficient code.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Neal Cardwell <ncardwell@google.com>
Link: https://patch.msgid.link/20260308122302.2895067-1-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
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), and a new field
(accecn_opt_sent_w_dsack) is added to indicate that an AccECN option was
sent with duplicate SACK info.
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>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://patch.msgid.link/20260131222515.8485-13-chia-yu.chang@nokia-bell-labs.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
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>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://patch.msgid.link/20260131222515.8485-2-chia-yu.chang@nokia-bell-labs.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
AccECN option may fail in various way, handle these:
- Attempt to negotiate the use of AccECN on the 1st retransmitted SYN
- From the 2nd retransmitted SYN, stop AccECN negotiation
- Remove option from SYN/ACK rexmits to handle blackholes
- If no option arrives in SYN/ACK, assume Option is not usable
- If an option arrives later, re-enabled
- If option is zeroed, disable AccECN option processing
This patch use existing padding bits in tcp_request_sock and
holes in tcp_sock without increasing the size.
Signed-off-by: Ilpo Järvinen <ij@kernel.org>
Signed-off-by: Chia-Yu Chang <chia-yu.chang@nokia-bell-labs.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://patch.msgid.link/20250916082434.100722-9-chia-yu.chang@nokia-bell-labs.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Instead of sending the option in every ACK, limit sending to
those ACKs where the option is necessary:
- Handshake
- "Change-triggered ACK" + the ACK following it. The
2nd ACK is necessary to unambiguously indicate which
of the ECN byte counters in increasing. The first
ACK has two counters increasing due to the ecnfield
edge.
- ACKs with CE to allow CEP delta validations to take
advantage of the option.
- Force option to be sent every at least once per 2^22
bytes. The check is done using the bit edges of the
byte counters (avoids need for extra variables).
- AccECN option beacon to send a few times per RTT even if
nothing in the ECN state requires that. The default is 3
times per RTT, and its period can be set via
sysctl_tcp_ecn_option_beacon.
Below are the pahole outcomes before and after this patch,
in which the group size of tcp_sock_write_tx is increased
from 89 to 97 due to the new u64 accecn_opt_tstamp member:
[BEFORE THIS PATCH]
struct tcp_sock {
[...]
u64 tcp_wstamp_ns; /* 2488 8 */
struct list_head tsorted_sent_queue; /* 2496 16 */
[...]
__cacheline_group_end__tcp_sock_write_tx[0]; /* 2521 0 */
__cacheline_group_begin__tcp_sock_write_txrx[0]; /* 2521 0 */
u8 nonagle:4; /* 2521: 0 1 */
u8 rate_app_limited:1; /* 2521: 4 1 */
/* XXX 3 bits hole, try to pack */
/* Force alignment to the next boundary: */
u8 :0;
u8 received_ce_pending:4;/* 2522: 0 1 */
u8 unused2:4; /* 2522: 4 1 */
u8 accecn_minlen:2; /* 2523: 0 1 */
u8 est_ecnfield:2; /* 2523: 2 1 */
u8 unused3:4; /* 2523: 4 1 */
[...]
__cacheline_group_end__tcp_sock_write_txrx[0]; /* 2628 0 */
[...]
/* size: 3200, cachelines: 50, members: 171 */
}
[AFTER THIS PATCH]
struct tcp_sock {
[...]
u64 tcp_wstamp_ns; /* 2488 8 */
u64 accecn_opt_tstamp; /* 2596 8 */
struct list_head tsorted_sent_queue; /* 2504 16 */
[...]
__cacheline_group_end__tcp_sock_write_tx[0]; /* 2529 0 */
__cacheline_group_begin__tcp_sock_write_txrx[0]; /* 2529 0 */
u8 nonagle:4; /* 2529: 0 1 */
u8 rate_app_limited:1; /* 2529: 4 1 */
/* XXX 3 bits hole, try to pack */
/* Force alignment to the next boundary: */
u8 :0;
u8 received_ce_pending:4;/* 2530: 0 1 */
u8 unused2:4; /* 2530: 4 1 */
u8 accecn_minlen:2; /* 2531: 0 1 */
u8 est_ecnfield:2; /* 2531: 2 1 */
u8 accecn_opt_demand:2; /* 2531: 4 1 */
u8 prev_ecnfield:2; /* 2531: 6 1 */
[...]
__cacheline_group_end__tcp_sock_write_txrx[0]; /* 2636 0 */
[...]
/* size: 3200, cachelines: 50, members: 173 */
}
Signed-off-by: Chia-Yu Chang <chia-yu.chang@nokia-bell-labs.com>
Co-developed-by: Ilpo Järvinen <ij@kernel.org>
Signed-off-by: Ilpo Järvinen <ij@kernel.org>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://patch.msgid.link/20250916082434.100722-8-chia-yu.chang@nokia-bell-labs.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
The Accurate ECN allows echoing back the sum of bytes for
each IP ECN field value in the received packets using
AccECN option. This change implements AccECN option tx & rx
side processing without option send control related features
that are added by a later change.
Based on specification:
https://tools.ietf.org/id/draft-ietf-tcpm-accurate-ecn-28.txt
(Some features of the spec will be added in the later changes
rather than in this one).
A full-length AccECN option is always attempted but if it does
not fit, the minimum length is selected based on the counters
that have changed since the last update. The AccECN option
(with 24-bit fields) often ends in odd sizes so the option
write code tries to take advantage of some nop used to pad
the other TCP options.
The delivered_ecn_bytes pairs with received_ecn_bytes similar
to how delivered_ce pairs with received_ce. In contrast to
ACE field, however, the option is not always available to update
delivered_ecn_bytes. For ACK w/o AccECN option, the delivered
bytes calculated based on the cumulative ACK+SACK information
are assigned to one of the counters using an estimation
heuristic to select the most likely ECN byte counter. Any
estimation error is corrected when the next AccECN option
arrives. It may occur that the heuristic gets too confused
when there are enough different byte counter deltas between
ACKs with the AccECN option in which case the heuristic just
gives up on updating the counters for a while.
tcp_ecn_option sysctl can be used to select option sending
mode for AccECN: TCP_ECN_OPTION_DISABLED, TCP_ECN_OPTION_MINIMUM,
and TCP_ECN_OPTION_FULL.
This patch increases the size of tcp_info struct, as there is
no existing holes for new u32 variables. Below are the pahole
outcomes before and after this patch:
[BEFORE THIS PATCH]
struct tcp_info {
[...]
__u32 tcpi_total_rto_time; /* 244 4 */
/* size: 248, cachelines: 4, members: 61 */
}
[AFTER THIS PATCH]
struct tcp_info {
[...]
__u32 tcpi_total_rto_time; /* 244 4 */
__u32 tcpi_received_ce; /* 248 4 */
__u32 tcpi_delivered_e1_bytes; /* 252 4 */
__u32 tcpi_delivered_e0_bytes; /* 256 4 */
__u32 tcpi_delivered_ce_bytes; /* 260 4 */
__u32 tcpi_received_e1_bytes; /* 264 4 */
__u32 tcpi_received_e0_bytes; /* 268 4 */
__u32 tcpi_received_ce_bytes; /* 272 4 */
/* size: 280, cachelines: 5, members: 68 */
}
This patch uses the existing 1-byte holes in the tcp_sock_write_txrx
group for new u8 members, but adds a 4-byte hole in tcp_sock_write_rx
group after the new u32 delivered_ecn_bytes[3] member. Therefore, the
group size of tcp_sock_write_rx is increased from 96 to 112. Below
are the pahole outcomes before and after this patch:
[BEFORE THIS PATCH]
struct tcp_sock {
[...]
u8 received_ce_pending:4; /* 2522: 0 1 */
u8 unused2:4; /* 2522: 4 1 */
/* XXX 1 byte hole, try to pack */
[...]
u32 rcv_rtt_last_tsecr; /* 2668 4 */
[...]
__cacheline_group_end__tcp_sock_write_rx[0]; /* 2728 0 */
[...]
/* size: 3200, cachelines: 50, members: 167 */
}
[AFTER THIS PATCH]
struct tcp_sock {
[...]
u8 received_ce_pending:4;/* 2522: 0 1 */
u8 unused2:4; /* 2522: 4 1 */
u8 accecn_minlen:2; /* 2523: 0 1 */
u8 est_ecnfield:2; /* 2523: 2 1 */
u8 unused3:4; /* 2523: 4 1 */
[...]
u32 rcv_rtt_last_tsecr; /* 2668 4 */
u32 delivered_ecn_bytes[3];/* 2672 12 */
/* XXX 4 bytes hole, try to pack */
[...]
__cacheline_group_end__tcp_sock_write_rx[0]; /* 2744 0 */
[...]
/* size: 3200, cachelines: 50, members: 171 */
}
Signed-off-by: Ilpo Järvinen <ij@kernel.org>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
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>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://patch.msgid.link/20250916082434.100722-7-chia-yu.chang@nokia-bell-labs.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
These three byte counters track IP ECN field payload byte sums for
all arriving (acceptable) packets for ECT0, ECT1, and CE. The
AccECN option (added by a later patch in the series) echoes these
counters back to sender side; therefore, it is placed within the
group of tcp_sock_write_txrx.
Below are the pahole outcomes before and after this patch, in which
the group size of tcp_sock_write_txrx is increased from 95 + 4 to
107 + 4 and an extra 4-byte hole is created but will be exploited
in later patches:
[BEFORE THIS PATCH]
struct tcp_sock {
[...]
u32 delivered_ce; /* 2576 4 */
u32 received_ce; /* 2580 4 */
u32 app_limited; /* 2584 4 */
u32 rcv_wnd; /* 2588 4 */
struct tcp_options_received rx_opt; /* 2592 24 */
__cacheline_group_end__tcp_sock_write_txrx[0]; /* 2616 0 */
[...]
/* size: 3200, cachelines: 50, members: 166 */
}
[AFTER THIS PATCH]
struct tcp_sock {
[...]
u32 delivered_ce; /* 2576 4 */
u32 received_ce; /* 2580 4 */
u32 received_ecn_bytes[3];/* 2584 12 */
u32 app_limited; /* 2596 4 */
u32 rcv_wnd; /* 2600 4 */
struct tcp_options_received rx_opt; /* 2604 24 */
__cacheline_group_end__tcp_sock_write_txrx[0]; /* 2628 0 */
/* XXX 4 bytes hole, try to pack */
[...]
/* size: 3200, cachelines: 50, members: 167 */
}
Signed-off-by: Ilpo Järvinen <ij@kernel.org>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
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>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://patch.msgid.link/20250916082434.100722-4-chia-yu.chang@nokia-bell-labs.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Accurate ECN negotiation parts based on the specification:
https://tools.ietf.org/id/draft-ietf-tcpm-accurate-ecn-28.txt
Accurate ECN is negotiated using ECE, CWR and AE flags in the
TCP header. TCP falls back into using RFC3168 ECN if one of the
ends supports only RFC3168-style ECN.
The AccECN negotiation includes reflecting IP ECN field value
seen in SYN and SYNACK back using the same bits as negotiation
to allow responding to SYN CE marks and to detect ECN field
mangling. CE marks should not occur currently because SYN=1
segments are sent with Non-ECT in IP ECN field (but proposal
exists to remove this restriction).
Reflecting SYN IP ECN field in SYNACK is relatively simple.
Reflecting SYNACK IP ECN field in the final/third ACK of
the handshake is more challenging. Linux TCP code is not well
prepared for using the final/third ACK a signalling channel
which makes things somewhat complicated here.
tcp_ecn sysctl can be used to select the highest ECN variant
(Accurate ECN, ECN, No ECN) that is attemped to be negotiated and
requested for incoming connection and outgoing connection:
TCP_ECN_IN_NOECN_OUT_NOECN, TCP_ECN_IN_ECN_OUT_ECN,
TCP_ECN_IN_ECN_OUT_NOECN, TCP_ECN_IN_ACCECN_OUT_ACCECN,
TCP_ECN_IN_ACCECN_OUT_ECN, and TCP_ECN_IN_ACCECN_OUT_NOECN.
After this patch, the size of tcp_request_sock remains unchanged
and no new holes are added. Below are the pahole outcomes before
and after this patch:
[BEFORE THIS PATCH]
struct tcp_request_sock {
[...]
u32 rcv_nxt; /* 352 4 */
u8 syn_tos; /* 356 1 */
/* size: 360, cachelines: 6, members: 16 */
}
[AFTER THIS PATCH]
struct tcp_request_sock {
[...]
u32 rcv_nxt; /* 352 4 */
u8 syn_tos; /* 356 1 */
bool accecn_ok; /* 357 1 */
u8 syn_ect_snt:2; /* 358: 0 1 */
u8 syn_ect_rcv:2; /* 358: 2 1 */
u8 accecn_fail_mode:4; /* 358: 4 1 */
/* size: 360, cachelines: 6, members: 20 */
}
After this patch, the size of tcp_sock remains unchanged and no new
holes are added. Also, 4 bits of the existing 2-byte hole are exploited.
Below are the pahole outcomes before and after this patch:
[BEFORE THIS PATCH]
struct tcp_sock {
[...]
u8 dup_ack_counter:2; /* 2761: 0 1 */
u8 tlp_retrans:1; /* 2761: 2 1 */
u8 unused:5; /* 2761: 3 1 */
u8 thin_lto:1; /* 2762: 0 1 */
u8 fastopen_connect:1; /* 2762: 1 1 */
u8 fastopen_no_cookie:1; /* 2762: 2 1 */
u8 fastopen_client_fail:2; /* 2762: 3 1 */
u8 frto:1; /* 2762: 5 1 */
/* XXX 2 bits hole, try to pack */
[...]
u8 keepalive_probes; /* 2765 1 */
/* XXX 2 bytes hole, try to pack */
[...]
/* size: 3200, cachelines: 50, members: 164 */
}
[AFTER THIS PATCH]
struct tcp_sock {
[...]
u8 dup_ack_counter:2; /* 2761: 0 1 */
u8 tlp_retrans:1; /* 2761: 2 1 */
u8 syn_ect_snt:2; /* 2761: 3 1 */
u8 syn_ect_rcv:2; /* 2761: 5 1 */
u8 thin_lto:1; /* 2761: 7 1 */
u8 fastopen_connect:1; /* 2762: 0 1 */
u8 fastopen_no_cookie:1; /* 2762: 1 1 */
u8 fastopen_client_fail:2; /* 2762: 2 1 */
u8 frto:1; /* 2762: 4 1 */
/* XXX 3 bits hole, try to pack */
[...]
u8 keepalive_probes; /* 2765 1 */
u8 accecn_fail_mode:4; /* 2766: 0 1 */
/* XXX 4 bits hole, try to pack */
/* XXX 1 byte hole, try to pack */
[...]
/* size: 3200, cachelines: 50, members: 166 */
}
Signed-off-by: Ilpo Järvinen <ij@kernel.org>
Co-developed-by: Olivier Tilmans <olivier.tilmans@nokia.com>
Signed-off-by: Olivier Tilmans <olivier.tilmans@nokia.com>
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>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://patch.msgid.link/20250916082434.100722-3-chia-yu.chang@nokia-bell-labs.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
This change implements Accurate ECN without negotiation and
AccECN Option (that will be added by later changes). Based on
AccECN specifications:
https://tools.ietf.org/id/draft-ietf-tcpm-accurate-ecn-28.txt
Accurate ECN allows feeding back the number of CE (congestion
experienced) marks accurately to the sender in contrast to
RFC3168 ECN that can only signal one marks-seen-yes/no per RTT.
Congestion control algorithms can take advantage of the accurate
ECN information to fine-tune their congestion response to avoid
drastic rate reduction when only mild congestion is encountered.
With Accurate ECN, tp->received_ce (r.cep in AccECN spec) keeps
track of how many segments have arrived with a CE mark. Accurate
ECN uses ACE field (ECE, CWR, AE) to communicate the value back
to the sender which updates tp->delivered_ce (s.cep) based on the
feedback. This signalling channel is lossy when ACE field overflow
occurs.
Conservative strategy is selected here to deal with the ACE
overflow, however, some strategies using the AccECN option later
in the overall patchset mitigate against false overflows detected.
The ACE field values on the wire are offset by
TCP_ACCECN_CEP_INIT_OFFSET. Delivered_ce/received_ce count the
real CE marks rather than forcing all downstream users to adapt
to the wire offset.
This patch uses the first 1-byte hole and the last 4-byte hole of
the tcp_sock_write_txrx for 'received_ce_pending' and 'received_ce'.
Also, the group size of tcp_sock_write_txrx is increased from
91 + 4 to 95 + 4 due to the new u32 received_ce member. Below are
the trimmed pahole outcomes before and after this patch.
[BEFORE THIS PATCH]
struct tcp_sock {
[...]
__cacheline_group_begin__tcp_sock_write_txrx[0]; /* 2521 0 */
u8 nonagle:4; /* 2521: 0 1 */
u8 rate_app_limited:1; /* 2521: 4 1 */
/* XXX 3 bits hole, try to pack */
/* XXX 2 bytes hole, try to pack */
[...]
u32 delivered_ce; /* 2576 4 */
u32 app_limited; /* 2580 4 */
u32 rcv_wnd; /* 2684 4 */
struct tcp_options_received rx_opt; /* 2688 24 */
__cacheline_group_end__tcp_sock_write_txrx[0]; /* 2612 0 */
/* XXX 4 bytes hole, try to pack */
[...]
/* size: 3200, cachelines: 50, members: 161 */
}
[AFTER THIS PATCH]
struct tcp_sock {
[...]
__cacheline_group_begin__tcp_sock_write_txrx[0]; /* 2521 0 */
u8 nonagle:4; /* 2521: 0 1 */
u8 rate_app_limited:1; /* 2521: 4 1 */
/* XXX 3 bits hole, try to pack */
/* Force alignment to the next boundary: */
u8 :0;
u8 received_ce_pending:4;/* 2522: 0 1 */
u8 unused2:4; /* 2522: 4 1 */
/* XXX 1 byte hole, try to pack */
[...]
u32 delivered_ce; /* 2576 4 */
u32 received_ce; /* 2580 4 */
u32 app_limited; /* 2584 4 */
u32 rcv_wnd; /* 2588 4 */
struct tcp_options_received rx_opt; /* 2592 24 */
__cacheline_group_end__tcp_sock_write_txrx[0]; /* 2616 0 */
[...]
/* size: 3200, cachelines: 50, members: 164 */
}
Signed-off-by: Ilpo Järvinen <ij@kernel.org>
Co-developed-by: Olivier Tilmans <olivier.tilmans@nokia.com>
Signed-off-by: Olivier Tilmans <olivier.tilmans@nokia.com>
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>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://patch.msgid.link/20250916082434.100722-2-chia-yu.chang@nokia-bell-labs.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Now that obsolete RFC3517/RFC6675 TCP loss detection has been removed,
we can remove the somewhat complex and intrusive code to maintain its
hint state: lost_skb_hint and lost_cnt_hint.
This commit makes tcp_clear_retrans_hints_partial() empty. We will
remove tcp_clear_retrans_hints_partial() and its call sites in the
next commit.
Suggested-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Reviewed-by: Yuchung Cheng <ycheng@google.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://patch.msgid.link/20250615001435.2390793-3-ncardwell.sw@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
If the application can not drain fast enough a TCP socket queue,
tcp_rcv_space_adjust() can overestimate tp->rcvq_space.space.
Then sk->sk_rcvbuf can grow and hit tcp_rmem[2] for no good reason.
Fix this by taking into acount the number of available bytes.
Keeping sk->sk_rcvbuf at the right size allows better cache efficiency.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Wei Wang <weiwan@google.com>
Link: https://patch.msgid.link/20250513193919.1089692-5-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
As a followup of my presentation in Zagreb for netdev 0x19:
icsk_clean_acked is only used by TCP when/if CONFIG_TLS_DEVICE
is enabled from tcp_ack().
Rename it to tcp_clean_acked, move it to tcp_sock structure
in the tcp_sock_read_rx for better cache locality in TCP
fast path.
Define this field only when CONFIG_TLS_DEVICE is enabled
saving 8 bytes on configs not using it.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Neal Cardwell <ncardwell@google.com>
Reviewed-by: Sabrina Dubroca <sd@queasysnail.net>
Reviewed-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Link: https://patch.msgid.link/20250317085313.2023214-1-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Yong-Hao Zou mentioned that linux was not strict as other OS in 3WHS,
for flows using TCP TS option (RFC 7323)
As hinted by an old comment in tcp_check_req(),
we can check the TSEcr value in the incoming packet corresponds
to one of the SYNACK TSval values we have sent.
In this patch, I record the oldest and most recent values
that SYNACK packets have used.
Send a challenge ACK if we receive a TSEcr outside
of this range, and increase a new SNMP counter.
nstat -az | grep TSEcrRejected
TcpExtTSEcrRejected 0 0.0
Due to TCP fastopen implementation, do not apply yet these checks
for fastopen flows.
v2: No longer use req->num_timeout, but treq->snt_tsval_first
to detect when first SYNACK is prepared. This means
we make sure to not send an initial zero TSval.
Make sure MPTCP and TCP selftests are passing.
Change MIB name to TcpExtTSEcrRejected
v1: https://lore.kernel.org/netdev/CADVnQykD8i4ArpSZaPKaoNxLJ2if2ts9m4As+=Jvdkrgx1qMHw@mail.gmail.com/T/
Reported-by: Yong-Hao Zou <yonghaoz1994@gmail.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Matthieu Baerts (NGI0) <matttbe@kernel.org>
Reviewed-by: Neal Cardwell <ncardwell@google.com>
Reviewed-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Link: https://patch.msgid.link/20250225171048.3105061-1-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
tp->recvmsg_inq is used from tcp recvmsg() thus should
be in tcp_sock_read_rx group.
tp->tcp_clock_cache and tp->tcp_mstamp are written
both in rx and tx paths, thus are better placed
in tcp_sock_write_txrx group.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
By moving some fields around, this patch shrinks
holes size from 56 to 32, saving 24 bytes on 64bit arches.
After the patch pahole gives the following for 'struct tcp_sock':
/* size: 2304, cachelines: 36, members: 162 */
/* sum members: 2234, holes: 6, sum holes: 32 */
/* sum bitfield members: 34 bits, bit holes: 5, sum bit holes: 14 bits */
/* padding: 32 */
/* paddings: 3, sum paddings: 10 */
/* forced alignments: 1, forced holes: 1, sum forced holes: 12 */
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Jiri Pirko <jiri@nvidia.com>
Link: https://lore.kernel.org/r/20240227192721.3558982-1-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
This extra check doesn't work for a handshake when SYN segment has
(current_key.maclen != rnext_key.maclen). It could be amended to
preserve rnext_key.maclen instead of current_key.maclen, but that
requires a lookup on listen socket.
Originally, this extra maclen check was introduced just because it was
cheap. Drop it and convert tcp_request_sock::maclen into boolean
tcp_request_sock::used_tcp_ao.
Fixes: 06b22ef295 ("net/tcp: Wire TCP-AO to request sockets")
Signed-off-by: Dmitry Safonov <dima@arista.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
The variables are organized according in the following way:
- TX read-mostly hotpath cache lines
- TXRX read-mostly hotpath cache lines
- RX read-mostly hotpath cache lines
- TX read-write hotpath cache line
- TXRX read-write hotpath cache line
- RX read-write hotpath cache line
Fastpath cachelines end after rcvq_space.
Cache line boundaries are enforced only between read-mostly and
read-write. That is, if read-mostly tx cachelines bleed into
read-mostly txrx cachelines, we do not care. We care about the
boundaries between read and write cachelines because we want
to prevent false sharing.
Fast path variables span cache lines before change: 12
Fast path variables span cache lines after change: 8
Suggested-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Wei Wang <weiwan@google.com>
Signed-off-by: Coco Li <lixiaoyan@google.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: David Ahern <dsahern@kernel.org>
Link: https://lore.kernel.org/r/20231204201232.520025-3-lixiaoyan@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
After blamed commit, TFO client-ack-dropped-then-recovery-ms-timestamps
packetdrill test failed.
David Morley and Neal Cardwell started investigating and Neal pointed
that we had :
tcp_conn_request()
tcp_try_fastopen()
-> tcp_fastopen_create_child
-> child = inet_csk(sk)->icsk_af_ops->syn_recv_sock()
-> tcp_create_openreq_child()
-> copy req_usec_ts from req:
newtp->tcp_usec_ts = treq->req_usec_ts;
// now the new TFO server socket always does usec TS, no matter
// what the route options are...
send_synack()
-> tcp_make_synack()
// disable tcp_rsk(req)->req_usec_ts if route option is not present:
if (tcp_rsk(req)->req_usec_ts < 0)
tcp_rsk(req)->req_usec_ts = dst_tcp_usec_ts(dst);
tcp_conn_request() has the initial dst, we can initialize
tcp_rsk(req)->req_usec_ts there instead of later in send_synack();
This means tcp_rsk(req)->req_usec_ts can be a boolean.
Many thanks to David an Neal for their help.
Fixes: 614e8316aa ("tcp: add support for usec resolution in TCP TS values")
Reported-by: kernel test robot <oliver.sang@intel.com>
Closes: https://lore.kernel.org/oe-lkp/202310302216.f79d78bc-oliver.sang@intel.com
Suggested-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: David Morley <morleyd@google.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Now when the new request socket is created from the listening socket,
it's recorded what MKT was used by the peer. tcp_rsk_used_ao() is
a new helper for checking if TCP-AO option was used to create the
request socket.
tcp_ao_copy_all_matching() will copy all keys that match the peer on the
request socket, as well as preparing them for the usage (creating
traffic keys).
Co-developed-by: Francesco Ruggeri <fruggeri@arista.com>
Signed-off-by: Francesco Ruggeri <fruggeri@arista.com>
Co-developed-by: Salam Noureddine <noureddine@arista.com>
Signed-off-by: Salam Noureddine <noureddine@arista.com>
Signed-off-by: Dmitry Safonov <dima@arista.com>
Acked-by: David Ahern <dsahern@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add support for sockets in time-wait state.
ao_info as well as all keys are inherited on transition to time-wait
socket. The lifetime of ao_info is now protected by ref counter, so
that tcp_ao_destroy_sock() will destruct it only when the last user is
gone.
Co-developed-by: Francesco Ruggeri <fruggeri@arista.com>
Signed-off-by: Francesco Ruggeri <fruggeri@arista.com>
Co-developed-by: Salam Noureddine <noureddine@arista.com>
Signed-off-by: Salam Noureddine <noureddine@arista.com>
Signed-off-by: Dmitry Safonov <dima@arista.com>
Acked-by: David Ahern <dsahern@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Back in 2015, Van Jacobson suggested to use usec resolution in TCP TS values.
This has been implemented in our private kernels.
Goals were :
1) better observability of delays in networking stacks.
2) better disambiguation of events based on TSval/ecr values.
3) building block for congestion control modules needing usec resolution.
Back then we implemented a schem based on private SYN options
to negotiate the feature.
For upstream submission, we chose to use a route attribute,
because this feature is probably going to be used in private
networks [1] [2].
ip route add 10/8 ... features tcp_usec_ts
Note that RFC 7323 recommends a
"timestamp clock frequency in the range 1 ms to 1 sec per tick.",
but also mentions
"the maximum acceptable clock frequency is one tick every 59 ns."
[1] Unfortunately RFC 7323 5.5 (Outdated Timestamps) suggests
to invalidate TS.Recent values after a flow was idle for more
than 24 days. This is the part making usec_ts a problem
for peers following this recommendation for long living
idle flows.
[2] Attempts to standardize usec ts went nowhere:
https://www.ietf.org/proceedings/97/slides/slides-97-tcpm-tcp-options-for-low-latency-00.pdfhttps://datatracker.ietf.org/doc/draft-wang-tcpm-low-latency-opt/
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This new dst feature flag will be used to allow TCP to use usec
based timestamps instead of msec ones.
ip route .... feature tcp_usec_ts
Also document that RTAX_FEATURE_SACK and RTAX_FEATURE_TIMESTAMP
are unused.
RTAX_FEATURE_ALLFRAG is also going away soon.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The 2023 SIGCOMM paper "Improving Network Availability with Protective
ReRoute" has indicated Linux TCP's RTO-triggered txhash rehashing can
effectively reduce application disruption during outages. To better
measure the efficacy of this feature, this patch adds three more
detailed stats during RTO recovery and exports via TCP_INFO.
Applications and monitoring systems can leverage this data to measure
the network path diversity and end-to-end repair latency during network
outages to improve their network infrastructure.
The following counters are added to tcp_sock in order to track RTO
events over the lifetime of a TCP socket.
1. u16 total_rto - Counts the total number of RTO timeouts.
2. u16 total_rto_recoveries - Counts the total number of RTO recoveries.
3. u32 total_rto_time - Counts the total time spent (ms) in RTO
recoveries. (time spent in CA_Loss and
CA_Recovery states)
To compute total_rto_time, we add a new u32 rto_stamp field to
tcp_sock. rto_stamp records the start timestamp (ms) of the last RTO
recovery (CA_Loss).
Corresponding fields are also added to the tcp_info struct.
Signed-off-by: Aananth V <aananthv@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This idea came after a particular workload requested
the quickack attribute set on routes, and a performance
drop was noticed for large bulk transfers.
For high throughput flows, it is best to use one cpu
running the user thread issuing socket system calls,
and a separate cpu to process incoming packets from BH context.
(With TSO/GRO, bottleneck is usually the 'user' cpu)
Problem is the user thread can spend a lot of time while holding
the socket lock, forcing BH handler to queue most of incoming
packets in the socket backlog.
Whenever the user thread releases the socket lock, it must first
process all accumulated packets in the backlog, potentially
adding latency spikes. Due to flood mitigation, having too many
packets in the backlog increases chance of unexpected drops.
Backlog processing unfortunately shifts a fair amount of cpu cycles
from the BH cpu to the 'user' cpu, thus reducing max throughput.
This patch takes advantage of the backlog processing,
and the fact that ACK are mostly cumulative.
The idea is to detect we are in the backlog processing
and defer all eligible ACK into a single one,
sent from tcp_release_cb().
This saves cpu cycles on both sides, and network resources.
Performance of a single TCP flow on a 200Gbit NIC:
- Throughput is increased by 20% (100Gbit -> 120Gbit).
- Number of generated ACK per second shrinks from 240,000 to 40,000.
- Number of backlog drops per second shrinks from 230 to 0.
Benchmark context:
- Regular netperf TCP_STREAM (no zerocopy)
- Intel(R) Xeon(R) Platinum 8481C (Saphire Rapids)
- MAX_SKB_FRAGS = 17 (~60KB per GRO packet)
This feature is guarded by a new sysctl, and enabled by default:
/proc/sys/net/ipv4/tcp_backlog_ack_defer
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Yuchung Cheng <ycheng@google.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Acked-by: Soheil Hassas Yeganeh <soheil@google.com>
Acked-by: Dave Taht <dave.taht@gmail.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
icsk->icsk_user_timeout can be set locklessly,
if all read sides use READ_ONCE().
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>
With modern NIC drivers shifting to full page allocations per
received frame, we face the following issue:
TCP has one per-netns sysctl used to tweak how to translate
a memory use into an expected payload (RWIN), in RX path.
tcp_win_from_space() implementation is limited to few cases.
For hosts dealing with various MSS, we either under estimate
or over estimate the RWIN we send to the remote peers.
For instance with the default sysctl_tcp_adv_win_scale value,
we expect to store 50% of payload per allocated chunk of memory.
For the typical use of MTU=1500 traffic, and order-0 pages allocations
by NIC drivers, we are sending too big RWIN, leading to potential
tcp collapse operations, which are extremely expensive and source
of latency spikes.
This patch makes sysctl_tcp_adv_win_scale obsolete, and instead
uses a per socket scaling factor, so that we can precisely
adjust the RWIN based on effective skb->len/skb->truesize ratio.
This patch alone can double TCP receive performance when receivers
are too slow to drain their receive queue, or by allowing
a bigger RWIN when MSS is close to PAGE_SIZE.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Soheil Hassas Yeganeh <soheil@google.com>
Link: https://lore.kernel.org/r/20230717152917.751987-1-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
We can change tcp_sk() to propagate its argument const qualifier,
thanks to container_of_const().
We have two places where a const sock pointer has to be upgraded
to a write one. We have been using const qualifier for lockless
listeners to clearly identify points where writes could happen.
Add tcp_sk_rw() helper to better document these.
tcp_inbound_md5_hash(), __tcp_grow_window(), tcp_reset_check()
and tcp_rack_reo_wnd() get an additional const qualififer
for their @tp local variables.
smc_check_reset_syn_req() also needs a similar change.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Simon Horman <simon.horman@corigine.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
A u32 counter is added to tcp_sock for counting the number of PLB
triggered rehashes for a TCP connection. An SNMP counter is also
added to count overall PLB triggered rehash events for a host. These
counters are hooked up to PLB implementation for DCTCP.
TCP_NLA_REHASH is added to SCM_TIMESTAMPING_OPT_STATS that reports
the rehash attempts triggered due to PLB or timeouts. This gives
a historical view of sustained congestion or timeouts experienced
by the TCP connection.
Signed-off-by: Mubashir Adnan Qureshi <mubashirq@google.com>
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This commit fixes a bug in the tracking of max_packets_out and
is_cwnd_limited. This bug can cause the connection to fail to remember
that is_cwnd_limited is true, causing the connection to fail to grow
cwnd when it should, causing throughput to be lower than it should be.
The following event sequence is an example that triggers the bug:
(a) The connection is cwnd_limited, but packets_out is not at its
peak due to TSO deferral deciding not to send another skb yet.
In such cases the connection can advance max_packets_seq and set
tp->is_cwnd_limited to true and max_packets_out to a small
number.
(b) Then later in the round trip the connection is pacing-limited (not
cwnd-limited), and packets_out is larger. In such cases the
connection would raise max_packets_out to a bigger number but
(unexpectedly) flip tp->is_cwnd_limited from true to false.
This commit fixes that bug.
One straightforward fix would be to separately track (a) the next
window after max_packets_out reaches a maximum, and (b) the next
window after tp->is_cwnd_limited is set to true. But this would
require consuming an extra u32 sequence number.
Instead, to save space we track only the most important
information. Specifically, we track the strongest available signal of
the degree to which the cwnd is fully utilized:
(1) If the connection is cwnd-limited then we remember that fact for
the current window.
(2) If the connection not cwnd-limited then we track the maximum
number of outstanding packets in the current window.
In particular, note that the new logic cannot trigger the buggy
(a)/(b) sequence above because with the new logic a condition where
tp->packets_out > tp->max_packets_out can only trigger an update of
tp->is_cwnd_limited if tp->is_cwnd_limited is false.
This first showed up in a testing of a BBRv2 dev branch, but this
buggy behavior highlighted a general issue with the
tcp_cwnd_validate() logic that can cause cwnd to fail to increase at
the proper rate for any TCP congestion control, including Reno or
CUBIC.
Fixes: ca8a226343 ("tcp: make cwnd-limited checks measurement-based, and gentler")
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Kevin(Yudong) Yang <yyd@google.com>
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When a bad bpf prog '.init' calls
bpf_setsockopt(TCP_CONGESTION, "itself"), it will trigger this loop:
.init => bpf_setsockopt(tcp_cc) => .init => bpf_setsockopt(tcp_cc) ...
... => .init => bpf_setsockopt(tcp_cc).
It was prevented by the prog->active counter before but the prog->active
detection cannot be used in struct_ops as explained in the earlier
patch of the set.
In this patch, the second bpf_setsockopt(tcp_cc) is not allowed
in order to break the loop. This is done by using a bit of
an existing 1 byte hole in tcp_sock to check if there is
on-going bpf_setsockopt(TCP_CONGESTION) in this tcp_sock.
Note that this essentially limits only the first '.init' can
call bpf_setsockopt(TCP_CONGESTION) to pick a fallback cc (eg. peer
does not support ECN) and the second '.init' cannot fallback to
another cc. This applies even the second
bpf_setsockopt(TCP_CONGESTION) will not cause a loop.
Signed-off-by: Martin KaFai Lau <martin.lau@kernel.org>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://lore.kernel.org/r/20220929070407.965581-5-martin.lau@linux.dev
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Most drivers use "skb_transport_offset(skb) + tcp_hdrlen(skb)"
to compute headers length for a TCP packet, but others
use more convoluted (but equivalent) ways.
Add skb_tcp_all_headers() and skb_inner_tcp_all_headers()
helpers to harmonize this a bit.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Simon Baatz
Eric Dumazet
Chia-Yu Chang
Ilpo Järvinen
Geliang Tang
Neal Cardwell
Jeremy Harris
Menglong Dong
Jakub Kicinski
Dmitry Safonov
Coco Li
Aananth V
Mubashir Adnan Qureshi
Martin KaFai Lau