fix(proto): tail-loss probes should always be ack-eliciting

[flub]

Description

The PacketBuilder gets is_ack_eliciting from the SendableFrames, but
that is empty if data had to be queued especially for the probe. This
would mean that the PacketBuilder::finish_and_track did not note the
correct number of in-flight bytes for the congestion controller and
also did not update the
PacketNumberSpace::time_of_last_ack_eliciting_packet. That field in
turn is used by the PTO calculation as time from which to arm the next
PTO timeout.

To fix this we add information about all frames on whether they are
ack-eliciting or not. And if any ack-eliciting frame is written by the
PacketBuilder it keeps track of this internally.

Tail-loss probes should also not happen for the Data spaces as long as
the handshake has not yet been confirmed. During this time the Initial
and Handshake spaces' tail-loss probes are the ones that drive the
connection forward. And sending tail-loss probes on the Data space already
consumes congestion controller window and pacing tokens. Also you
don't even know if the peer has keys to decrypt them.

Breaking Changes

n/a

Notes & open questions

n/a

[github-actions]

Documentation for this PR has been generated and is available at: https://n0-computer.github.io/noq/pr/561/docs/noq/

Last updated: 2026-04-08T16:08:40Z

[github-actions]

Performance Comparison Report

cdce1f8837fffff61503a935e31466b885e987ea - artifacts

Raw Benchmarks (localhost)

Scenario	noq	upstream	Delta	CPU (avg/max)
large-single	5644.6 Mbps	8287.8 Mbps	-31.9%	96.2% / 107.0%
medium-concurrent	5823.9 Mbps	7987.7 Mbps	-27.1%	95.7% / 106.0%
medium-single	4022.1 Mbps	4716.9 Mbps	-14.7%	98.4% / 163.0%
small-concurrent	3902.1 Mbps	5296.1 Mbps	-26.3%	93.8% / 108.0%
small-single	3635.7 Mbps	4676.4 Mbps	-22.3%	93.7% / 108.0%

Netsim Benchmarks (network simulation)

Condition	noq	upstream	Delta
ideal	3184.4 Mbps	4009.6 Mbps	-20.6%
lan	782.4 Mbps	810.3 Mbps	-3.4%
lossy	69.8 Mbps	55.9 Mbps	+25.0%
wan	83.8 Mbps	83.8 Mbps	~0%

Summary

noq is 24.4% slower on average

---

c187201cb12f3629419d35a9cd7ff3d8d94a767b - artifacts

Raw Benchmarks (localhost)

Scenario	noq	upstream	Delta	CPU (avg/max)
large-single	5785.7 Mbps	8088.1 Mbps	-28.5%	98.6% / 159.0%
medium-concurrent	5715.7 Mbps	7290.7 Mbps	-21.6%	97.9% / 159.0%
medium-single	3827.5 Mbps	4625.2 Mbps	-17.2%	93.6% / 108.0%
small-concurrent	3889.4 Mbps	5274.4 Mbps	-26.3%	95.5% / 109.0%
small-single	3656.9 Mbps	4650.8 Mbps	-21.4%	90.5% / 97.9%

Netsim Benchmarks (network simulation)

Condition	noq	upstream	Delta
ideal	3063.1 Mbps	3605.7 Mbps	-15.0%
lan	782.0 Mbps	796.4 Mbps	-1.8%
lossy	69.8 Mbps	55.9 Mbps	+25.0%
wan	83.8 Mbps	83.8 Mbps	~0%

Summary

noq is 22.0% slower on average

---

d8c93053a98715db0206a87615d9451a87931178 - artifacts

No results available

---

545501f11aef8691411f8516cd01febc3e7c1855 - artifacts

Raw Benchmarks (localhost)

Scenario	noq	upstream	Delta	CPU (avg/max)
large-single	5416.1 Mbps	7952.1 Mbps	-31.9%	90.7% / 96.8%
medium-concurrent	5373.4 Mbps	7839.9 Mbps	-31.5%	93.9% / 99.4%
medium-single	4032.2 Mbps	4749.3 Mbps	-15.1%	96.0% / 129.0%
small-concurrent	3884.5 Mbps	5338.5 Mbps	-27.2%	88.3% / 98.2%
small-single	3427.8 Mbps	4819.3 Mbps	-28.9%	92.5% / 101.0%

Netsim Benchmarks (network simulation)

Condition	noq	upstream	Delta
ideal	N/A	3922.2 Mbps	N/A
lan	N/A	803.1 Mbps	N/A
lossy	N/A	55.9 Mbps	N/A
wan	N/A	83.8 Mbps	N/A

Summary

noq is 27.9% slower on average

[flub]

This is ready for review now.

[matheus23]

really happy with the PacketBuilder changes.
Unfortunate that this requires another bugfix (the loss during handshake thing), and that's a lot harder to understand (otherwise I'd have given the 👍 long ago 😅).

[divagant-martian]

LGTM

[matheus23]

Please merge!