refactor: split poll_transmit into per-path and per-space parts by flub · Pull Request #338 · n0-computer/noq

flub · 2026-01-17T18:14:35Z

Description

This starts to pull the poll_transmit loop apart. Makes it easier to make/follow decisions on some levels. There is now a clean separation between paths and on and off-path packet building.

Surfaces some pre-existing bugs. This PR is not the time to fix them.

Closes #170

Breaking Changes

n/a

Notes & open questions

Follow up work:

Fix packet scheduling bug
Fix other bugs
Pull the remainder of the loop further apart.
Improve TransmitBuf::start_new_datagram (see refactor: split up poll_transmit into per_path and per_space parts #255 (comment))

github-actions · 2026-01-18T08:46:09Z

Documentation for this PR has been generated and is available at: https://n0-computer.github.io/quinn/pr/338/docs/iroh_quinn/

Last updated: 2026-01-26T14:04:13Z

codecov-commenter · 2026-01-18T08:47:06Z

Codecov Report

❌ Patch coverage is 94.28571% with 16 lines in your changes missing coverage. Please review.
✅ Project coverage is 76.76%. Comparing base (e424c7f) to head (4570684).
⚠️ Report is 3 commits behind head on main.

Files with missing lines	Patch %	Lines
quinn-proto/src/connection/mod.rs	94.20%	16 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #338      +/-   ##
==========================================
- Coverage   76.90%   76.76%   -0.15%     
==========================================
  Files          81       81              
  Lines       22828    22907      +79     
==========================================
+ Hits        17557    17584      +27     
- Misses       5271     5323      +52

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

github-actions · 2026-01-18T09:25:42Z

Performance Comparison Report

e82043fa8a4eab55cbd871a9b0f6e85e1602185d - artifacts

Raw Benchmarks (localhost)

Scenario	iroh-quinn	upstream	Delta	CPU (avg/max)
large-single	N/A	7572.8 Mbps	N/A	N/A
medium-concurrent	N/A	7581.3 Mbps	N/A	N/A
medium-single	N/A	4236.9 Mbps	N/A	N/A
small-concurrent	N/A	4866.5 Mbps	N/A	N/A
small-single	N/A	4185.0 Mbps	N/A	N/A

Netsim Benchmarks (network simulation)

Condition	iroh-quinn	upstream	Delta
ideal	1410.1 Mbps	3533.5 Mbps	-60.1%
lan	684.6 Mbps	796.5 Mbps	-14.0%
lossy	69.8 Mbps	69.8 Mbps	~0%
wan	83.8 Mbps	83.8 Mbps	~0%

Summary

iroh-quinn is 49.9% slower on average

---

fa271d48a7db763d53f83661443623bca5f14e4c - artifacts

Raw Benchmarks (localhost)

Scenario	iroh-quinn	upstream	Delta	CPU (avg/max)
large-single	5347.4 Mbps	7544.7 Mbps	-29.1%	97.2% / 162.0%
medium-concurrent	5388.5 Mbps	6931.3 Mbps	-22.3%	95.4% / 111.0%
medium-single	3919.9 Mbps	5279.0 Mbps	-25.7%	92.5% / 111.0%
small-concurrent	3808.9 Mbps	5014.7 Mbps	-24.0%	95.3% / 114.0%
small-single	3358.5 Mbps	4299.9 Mbps	-21.9%	91.0% / 98.4%

Netsim Benchmarks (network simulation)

Condition	iroh-quinn	upstream	Delta
ideal	2833.2 Mbps	3881.2 Mbps	-27.0%
lan	768.5 Mbps	799.7 Mbps	-3.9%
lossy	69.8 Mbps	55.9 Mbps	+25.0%
wan	83.8 Mbps	83.8 Mbps	~0%

Summary

iroh-quinn is 24.5% slower on average

---

f7301a8764a12ae5c7c8cf219b2a6d34147b88c4 - artifacts

Raw Benchmarks (localhost)

Scenario	iroh-quinn	upstream	Delta	CPU (avg/max)
large-single	5325.4 Mbps	7789.2 Mbps	-31.6%	93.4% / 98.1%
medium-concurrent	5353.5 Mbps	7712.4 Mbps	-30.6%	95.7% / 111.0%
medium-single	3713.1 Mbps	4575.2 Mbps	-18.8%	93.6% / 110.0%
small-concurrent	3854.3 Mbps	5189.3 Mbps	-25.7%	96.7% / 113.0%
small-single	3260.3 Mbps	4659.6 Mbps	-30.0%	89.1% / 96.9%

Netsim Benchmarks (network simulation)

Condition	iroh-quinn	upstream	Delta
ideal	2891.1 Mbps	3605.2 Mbps	-19.8%
lan	768.5 Mbps	796.4 Mbps	-3.5%
lossy	69.8 Mbps	55.9 Mbps	+25.0%
wan	83.8 Mbps	83.8 Mbps	~0%

Summary

iroh-quinn is 26.5% slower on average

---

33b9d34acbfa3140f089a1ed8c919bb52bf77025 - artifacts

Raw Benchmarks (localhost)

Scenario	iroh-quinn	upstream	Delta	CPU (avg/max)
large-single	5362.5 Mbps	N/A	N/A	94.3% / 117.0%
medium-concurrent	5511.3 Mbps	N/A	N/A	92.9% / 97.7%
medium-single	4041.8 Mbps	N/A	N/A	90.2% / 97.3%
small-concurrent	3856.8 Mbps	N/A	N/A	102.3% / 179.0%
small-single	3481.0 Mbps	N/A	N/A	89.8% / 97.3%

Netsim Benchmarks (network simulation)

Condition	iroh-quinn	upstream	Delta
ideal	2793.2 Mbps	3561.8 Mbps	-21.6%
lan	777.3 Mbps	796.5 Mbps	-2.4%
wan	83.8 Mbps	83.8 Mbps	~0%

Summary

iroh-quinn is 17.7% slower on average

---

6d461ef36fa45ae09cb79753ce2f700e3131728b - artifacts

Raw Benchmarks (localhost)

Scenario	iroh-quinn	upstream	Delta	CPU (avg/max)
large-single	5393.5 Mbps	7142.6 Mbps	-24.5%	95.5% / 118.0%
medium-concurrent	5356.0 Mbps	6824.3 Mbps	-21.5%	88.9% / 95.8%
medium-single	4049.3 Mbps	4177.6 Mbps	-3.1%	88.7% / 96.8%
small-concurrent	3668.1 Mbps	4874.8 Mbps	-24.8%	87.9% / 97.9%
small-single	3327.1 Mbps	4076.9 Mbps	-18.4%	90.2% / 97.4%

Netsim Benchmarks (network simulation)

Condition	iroh-quinn	upstream	Delta
ideal	2891.4 Mbps	3868.9 Mbps	-25.3%
lan	782.4 Mbps	796.4 Mbps	-1.8%
lossy	69.8 Mbps	69.8 Mbps	~0%
wan	83.8 Mbps	83.8 Mbps	~0%

Summary

iroh-quinn is 19.7% slower on average

---

b12e786e96eb0d5b32df972572c78b180ae9c65b - artifacts

Raw Benchmarks (localhost)

Scenario	iroh-quinn	upstream	Delta	CPU (avg/max)
large-single	5421.9 Mbps	8006.5 Mbps	-32.3%	94.6% / 111.0%
medium-concurrent	5320.5 Mbps	7898.4 Mbps	-32.6%	91.9% / 96.9%
medium-single	4140.8 Mbps	4470.4 Mbps	-7.4%	95.9% / 112.0%
small-concurrent	3724.7 Mbps	5100.7 Mbps	-27.0%	88.8% / 98.1%
small-single	3307.7 Mbps	4654.8 Mbps	-28.9%	95.5% / 114.0%

Netsim Benchmarks (network simulation)

Condition	iroh-quinn	upstream	Delta
ideal	2849.2 Mbps	3897.3 Mbps	-26.9%
lan	769.7 Mbps	810.3 Mbps	-5.0%
lossy	69.9 Mbps	69.8 Mbps	~0%
wan	83.8 Mbps	83.8 Mbps	~0%

Summary

iroh-quinn is 26.6% slower on average

---

7af282faafdbc1008e0c17f9a7a885d39e3f654c - artifacts

No results available

---

ff91de9c3029db16ce6f1e3f5f45260442b4e29d - artifacts

No results available

---

08751b8ea7c96c36fa3264d12e8bcfbab085e36f - artifacts

Raw Benchmarks (localhost)

Scenario	iroh-quinn	upstream	Delta	CPU (avg/max)
large-single	5293.7 Mbps	7600.8 Mbps	-30.4%	92.5% / 97.2%
medium-concurrent	5433.5 Mbps	7243.1 Mbps	-25.0%	93.9% / 117.0%
medium-single	3800.9 Mbps	4146.4 Mbps	-8.3%	88.1% / 96.3%
small-concurrent	3829.3 Mbps	5112.1 Mbps	-25.1%	94.7% / 121.0%
small-single	3492.2 Mbps	4391.6 Mbps	-20.5%	87.0% / 95.7%

Netsim Benchmarks (network simulation)

Condition	iroh-quinn	upstream	Delta
ideal	2852.7 Mbps	3659.8 Mbps	-22.1%
lan	782.5 Mbps	796.4 Mbps	-1.7%
lossy	69.8 Mbps	55.9 Mbps	+25.0%
wan	83.8 Mbps	83.8 Mbps	~0%

Summary

iroh-quinn is 22.5% slower on average

---

ae89bc51d1d9faf2861d10bd98f049eb5396a2e3 - artifacts

Raw Benchmarks (localhost)

Scenario	iroh-quinn	upstream	Delta	CPU (avg/max)
large-single	5516.6 Mbps	7908.4 Mbps	-30.2%	95.8% / 119.0%
medium-concurrent	5501.6 Mbps	7577.7 Mbps	-27.4%	91.6% / 97.0%
medium-single	4120.1 Mbps	4404.0 Mbps	-6.4%	92.0% / 100.0%
small-concurrent	3792.4 Mbps	5058.6 Mbps	-25.0%	89.1% / 98.1%
small-single	3474.5 Mbps	4518.0 Mbps	-23.1%	90.9% / 98.2%

Netsim Benchmarks (network simulation)

Condition	iroh-quinn	upstream	Delta
ideal	2820.3 Mbps	3603.8 Mbps	-21.7%
lan	782.5 Mbps	796.4 Mbps	-1.7%
lossy	69.8 Mbps	69.8 Mbps	~0%
wan	83.8 Mbps	83.8 Mbps	~0%

Summary

iroh-quinn is 23.1% slower on average

---

b8238c46cd46b0c893a5500539233b52ab491851 - artifacts

No results available

---

3c9d82ca94f623d2a6038f36ea801aca87f8088c - artifacts

Raw Benchmarks (localhost)

Scenario	iroh-quinn	upstream	Delta	CPU (avg/max)
large-single	5307.7 Mbps	7943.2 Mbps	-33.2%	90.3% / 96.8%
medium-concurrent	5460.3 Mbps	7679.4 Mbps	-28.9%	91.7% / 97.0%
medium-single	3846.9 Mbps	4470.6 Mbps	-13.9%	93.3% / 116.0%
small-concurrent	3788.7 Mbps	5109.0 Mbps	-25.8%	91.2% / 99.0%
small-single	3480.4 Mbps	4685.3 Mbps	-25.7%	93.8% / 120.0%

Netsim Benchmarks (network simulation)

Condition	iroh-quinn	upstream	Delta
ideal	2897.8 Mbps	3611.2 Mbps	-19.8%
lan	782.5 Mbps	796.5 Mbps	-1.8%
lossy	69.9 Mbps	55.9 Mbps	+25.0%
wan	83.8 Mbps	83.8 Mbps	~0%

Summary

iroh-quinn is 25.3% slower on average

---

e81119cdfa80b4a063d325b36835352b7b16dac2 - artifacts

No results available

---

54ffa24b6f636f5ef8e52d2af27a38746a23fbf5 - artifacts

Raw Benchmarks (localhost)

Scenario	iroh-quinn	upstream	Delta	CPU (avg/max)
large-single	5403.6 Mbps	7329.1 Mbps	-26.3%	94.4% / 111.0%
medium-concurrent	5419.4 Mbps	6842.9 Mbps	-20.8%	94.4% / 109.0%
medium-single	3709.3 Mbps	3911.2 Mbps	-5.2%	98.6% / 163.0%
small-concurrent	3758.8 Mbps	5004.7 Mbps	-24.9%	92.0% / 99.5%
small-single	3357.4 Mbps	4293.9 Mbps	-21.8%	95.1% / 114.0%

Netsim Benchmarks (network simulation)

Condition	iroh-quinn	upstream	Delta
ideal	2864.8 Mbps	3954.8 Mbps	-27.6%
lan	782.3 Mbps	807.0 Mbps	-3.1%
lossy	69.8 Mbps	55.9 Mbps	+25.0%
wan	83.8 Mbps	83.8 Mbps	~0%

Summary

iroh-quinn is 21.2% slower on average

---

70f27512fa4c8098dc3dadfe2c7c0bb7b55bac15 - artifacts

Raw Benchmarks (localhost)

Scenario	iroh-quinn	upstream	Delta	CPU (avg/max)
large-single	5330.9 Mbps	7473.5 Mbps	-28.7%	97.3% / 119.0%
medium-concurrent	5391.8 Mbps	6749.8 Mbps	-20.1%	96.8% / 122.0%
medium-single	3572.7 Mbps	3914.6 Mbps	-8.7%	91.2% / 98.4%
small-concurrent	3701.1 Mbps	4941.1 Mbps	-25.1%	96.7% / 122.0%
small-single	3488.8 Mbps	4324.1 Mbps	-19.3%	91.0% / 98.1%

Netsim Benchmarks (network simulation)

Condition	iroh-quinn	upstream	Delta
ideal	3058.8 Mbps	3577.0 Mbps	-14.5%
lan	782.4 Mbps	796.5 Mbps	-1.8%
lossy	69.8 Mbps	55.9 Mbps	+24.9%
wan	83.8 Mbps	83.8 Mbps	~0%

Summary

iroh-quinn is 20.2% slower on average

---

c8720bc1861ed3210cb096e8a1452169dbca9b36 - artifacts

Raw Benchmarks (localhost)

Scenario	iroh-quinn	upstream	Delta	CPU (avg/max)
large-single	5407.5 Mbps	7867.4 Mbps	-31.3%	94.7% / 108.0%
medium-concurrent	5416.3 Mbps	7635.7 Mbps	-29.1%	93.1% / 98.0%
medium-single	3532.2 Mbps	4554.3 Mbps	-22.4%	92.4% / 109.0%
small-concurrent	3698.6 Mbps	5201.1 Mbps	-28.9%	91.4% / 99.1%
small-single	3303.1 Mbps	4738.2 Mbps	-30.3%	88.3% / 97.2%

Netsim Benchmarks (network simulation)

Condition	iroh-quinn	upstream	Delta
ideal	2891.0 Mbps	3687.0 Mbps	-21.6%
lan	782.4 Mbps	796.4 Mbps	-1.8%
lossy	69.8 Mbps	55.9 Mbps	+25.0%
wan	83.8 Mbps	83.8 Mbps	~0%

Summary

iroh-quinn is 27.3% slower on average

---

45706842d3eac39b8140dc22f298f13e03c9e0b1 - artifacts

Raw Benchmarks (localhost)

Scenario	iroh-quinn	upstream	Delta	CPU (avg/max)
large-single	5409.2 Mbps	N/A	N/A	93.5% / 98.1%
medium-concurrent	5457.0 Mbps	N/A	N/A	91.5% / 97.2%
medium-single	3782.0 Mbps	N/A	N/A	91.4% / 100.0%
small-concurrent	3668.6 Mbps	N/A	N/A	101.5% / 182.0%
small-single	3379.8 Mbps	N/A	N/A	94.9% / 122.0%

Netsim Benchmarks (network simulation)

Condition	iroh-quinn	upstream	Delta
ideal	2975.1 Mbps	N/A	N/A
lan	782.5 Mbps	N/A	N/A

---

ce19b48e1bf39d6baab714112d4f78db74bf5597 - artifacts

Raw Benchmarks (localhost)

Scenario	iroh-quinn	upstream	Delta	CPU (avg/max)
large-single	5263.4 Mbps	7893.8 Mbps	-33.3%	93.7% / 98.2%
medium-concurrent	5465.4 Mbps	8050.9 Mbps	-32.1%	92.0% / 97.4%
medium-single	3803.1 Mbps	4749.0 Mbps	-19.9%	94.2% / 121.0%
small-concurrent	3725.8 Mbps	5341.2 Mbps	-30.2%	97.8% / 121.0%
small-single	3477.0 Mbps	4830.9 Mbps	-28.0%	93.7% / 120.0%

Netsim Benchmarks (network simulation)

Condition	iroh-quinn	upstream	Delta
ideal	2868.7 Mbps	3723.8 Mbps	-23.0%
lan	769.4 Mbps	796.5 Mbps	-3.4%
lossy	55.9 Mbps	55.9 Mbps	~0%
wan	83.8 Mbps	83.8 Mbps	~0%

Summary

iroh-quinn is 28.2% slower on average

quinn-proto/src/connection/mod.rs

matheus23 · 2026-01-19T10:56:35Z

Should we merge #339 before doing all of the poll transmit refactors? It might help with all of the Send vs. SendTransmit hacks.

And collecting some functions together. Split off from #338.

And collecting some functions together. Split off from #338. Co-authored-by: Diva Martínez <26765164+divagant-martian@users.noreply.github.com>

…l-transmit-ohno

quinn-proto/src/connection/mod.rs

dignifiedquire · 2026-01-26T12:05:09Z

once this merges, we should create issues for the todos, to keep track of them

dignifiedquire · 2026-01-26T12:10:38Z

quinn-proto/src/connection/mod.rs

+    }
+
+    /// poll_transmit logic for off-path data.
+    fn poll_transmit_off_path(


these method names make me happy

divagant-martian

🐸

dignifiedquire and others added 12 commits January 14, 2026 18:04

split up poll_transmit per path and per space

f6d3eda

refactor: return last_packet_number

d5f8386

cleanup status enums

382ad3d

refactor: streamline mtu probe building

9daa159

refactor: extract remote cid exhaustion handling into function

6e3106d

refactor: move remote_cid check per path

c4d87c9

refactor: simplify send space check

f3c8a40

fixup

3aac39d

docs(proto): describe enums

b5c2cce

Merge branch 'main' into refactor-poll-transmit-path

73839ad

wip

b7acb73

friday was too short. make it work, maybe

d2cad6d

n0bot bot added this to iroh Jan 17, 2026

github-project-automation bot moved this to 🏗 In progress in iroh Jan 17, 2026

Merge branch 'main' into flub/poll-transmit-ohno

e82043f

flub added 5 commits January 18, 2026 15:19

Do coalescing right

fa271d4

tweak comments, logs

07b5423

remove redundant size-setting

f7301a8

fix path scheduling description, add bug description

33b9d34

fixup path stats

6d461ef

flub commented Jan 18, 2026

View reviewed changes

quinn-proto/src/connection/mod.rs Show resolved Hide resolved

flub marked this pull request as ready for review January 18, 2026 16:37

flub requested review from Frando, dignifiedquire and divagant-martian January 18, 2026 16:37

wording, wording

b12e786

flub mentioned this pull request Jan 21, 2026

use PATH_CHALLENGE for hole punching #231

Closed

divagant-martian added 2 commits January 21, 2026 10:38

Merge remote-tracking branch 'n0/main' into flub/poll-transmit-ohno

e81119c

fix merge

54ffa24

flub added a commit that referenced this pull request Jan 21, 2026

refactor: Some much needed renaming of variables mostly

cdd378e

And collecting some functions together. Split off from #338.

flub mentioned this pull request Jan 21, 2026

refactor: Some much needed renaming of variables mostly #354

Merged

Pass around a Vec while no transmit has been started yet

94554c9

divagant-martian added 2 commits January 21, 2026 13:22

Merge commit '37524202b6fcd598a94d02ea9009a97f1347b4f0' into flub/pol…

03efa76

…l-transmit-ohno

Merge remote-tracking branch 'n0/main' into flub/poll-transmit-ohno

70f2751

divagant-martian reviewed Jan 21, 2026

View reviewed changes

quinn-proto/src/connection/mod.rs Outdated Show resolved Hide resolved

Merge branch 'main' into flub/poll-transmit-ohno

c8720bc

flub mentioned this pull request Jan 22, 2026

fix: Do not send off-path data when closing #358

Merged

flub linked an issue Jan 23, 2026 that may be closed by this pull request

Improve poll_transmit loop logic #170

Closed

dignifiedquire reviewed Jan 26, 2026

View reviewed changes

quinn-proto/src/connection/mod.rs Show resolved Hide resolved

dignifiedquire reviewed Jan 26, 2026

View reviewed changes

quinn-proto/src/connection/mod.rs Outdated Show resolved Hide resolved

dignifiedquire reviewed Jan 26, 2026

View reviewed changes

flub added 2 commits January 26, 2026 14:53

expand on what is meant here, link to new issue

4570684

pointless return

ce19b48

dignifiedquire approved these changes Jan 26, 2026

View reviewed changes

flub self-assigned this Jan 26, 2026

flub added the refactor label Jan 26, 2026

flub added this to the quinn: iroh v0.96 milestone Jan 26, 2026

divagant-martian approved these changes Jan 26, 2026

View reviewed changes

divagant-martian added this pull request to the merge queue Jan 26, 2026

Merged via the queue into main with commit 2cc515a Jan 26, 2026
59 checks passed

github-project-automation bot moved this from 🏗 In progress to ✅ Done in iroh Jan 26, 2026

divagant-martian deleted the flub/poll-transmit-ohno branch January 26, 2026 15:42

divagant-martian mentioned this pull request Jan 27, 2026

fix: send a fresh challenge on old paths for the current hp round #375

Closed

Conversation

flub commented Jan 17, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Description

Breaking Changes

Notes & open questions

Uh oh!

github-actions bot commented Jan 18, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

codecov-commenter commented Jan 18, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

github-actions bot commented Jan 18, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Performance Comparison Report

Raw Benchmarks (localhost)

Netsim Benchmarks (network simulation)

Summary

Raw Benchmarks (localhost)

Netsim Benchmarks (network simulation)

Summary

Raw Benchmarks (localhost)

Netsim Benchmarks (network simulation)

Summary

Raw Benchmarks (localhost)

Netsim Benchmarks (network simulation)

Summary

Raw Benchmarks (localhost)

Netsim Benchmarks (network simulation)

Summary

Raw Benchmarks (localhost)

Netsim Benchmarks (network simulation)

Summary

Raw Benchmarks (localhost)

Netsim Benchmarks (network simulation)

Summary

Raw Benchmarks (localhost)

Netsim Benchmarks (network simulation)

Summary

Raw Benchmarks (localhost)

Netsim Benchmarks (network simulation)

Summary

Raw Benchmarks (localhost)

Netsim Benchmarks (network simulation)

Summary

Raw Benchmarks (localhost)

Netsim Benchmarks (network simulation)

Summary

Raw Benchmarks (localhost)

Netsim Benchmarks (network simulation)

Summary

Raw Benchmarks (localhost)

Netsim Benchmarks (network simulation)

Raw Benchmarks (localhost)

Netsim Benchmarks (network simulation)

Summary

Uh oh!

Uh oh!

matheus23 commented Jan 19, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

dignifiedquire commented Jan 26, 2026

Uh oh!

dignifiedquire Jan 26, 2026

Choose a reason for hiding this comment

Uh oh!

divagant-martian left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

flub commented Jan 17, 2026 •

edited

Loading

github-actions bot commented Jan 18, 2026 •

edited

Loading

codecov-commenter commented Jan 18, 2026 •

edited

Loading

github-actions bot commented Jan 18, 2026 •

edited

Loading

matheus23 commented Jan 19, 2026 •

edited

Loading