Autobahn migrated from sei-v3 by pompon0 · Pull Request #2791 · sei-protocol/sei-chain

pompon0 · 2026-02-03T13:11:03Z

Migrated consensus,avail,data,producer and types modules. Migrated corresponding tests. Replaced gRPC usage with custom rpc layer on top of mux. GigaRouter is still hardcoded to be disabled.

NOTE: there is a lot of missing parts and coverage, but this pr will allow us to stop development of autobahn in sei-v3, so that we can start making it production-ready in sei-chain.

…hain into gprusak-proto

github-actions · 2026-02-03T13:19:21Z

The latest Buf updates on your PR. Results from workflow Buf / buf (pull_request).

Build	Format	Lint	Breaking	Updated (UTC)
`✅ passed`	`✅ passed`	`✅ passed`	`✅ passed`	Feb 5, 2026, 1:34 PM

codecov · 2026-02-03T13:29:32Z

Codecov Report

❌ Patch coverage is 69.72763% with 778 lines in your changes missing coverage. Please review.
✅ Project coverage is 56.79%. Comparing base (ec9e52e) to head (50ea1d9).

Files with missing lines	Patch %	Lines
sei-tendermint/internal/autobahn/types/proposal.go	66.66%	48 Missing and 37 partials ⚠️
sei-tendermint/internal/autobahn/data/state.go	63.91%	52 Missing and 18 partials ⚠️
sei-tendermint/internal/autobahn/avail/state.go	73.22%	41 Missing and 23 partials ⚠️
sei-tendermint/internal/p2p/giga/avail.go	59.87%	32 Missing and 31 partials ⚠️
sei-tendermint/internal/autobahn/types/timeout.go	60.95%	40 Missing and 17 partials ⚠️
sei-tendermint/internal/autobahn/producer/state.go	0.00%	51 Missing ⚠️
sei-tendermint/internal/autobahn/types/msg.go	68.70%	31 Missing and 15 partials ⚠️
sei-tendermint/internal/autobahn/data/testonly.go	58.82%	41 Missing and 1 partial ⚠️
sei-tendermint/internal/autobahn/types/block.go	65.21%	27 Missing and 13 partials ⚠️
sei-tendermint/internal/p2p/giga/consensus.go	46.57%	29 Missing and 10 partials ⚠️
... and 26 more

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #2791      +/-   ##
==========================================
+ Coverage   46.94%   56.79%   +9.85%     
==========================================
  Files        1965     2068     +103     
  Lines      160525   167802    +7277     
==========================================
+ Hits        75352    95297   +19945     
+ Misses      78643    64056   -14587     
- Partials     6530     8449    +1919

Flag	Coverage Δ
sei-chain	`40.83% <0.00%> (-0.69%)`	⬇️
sei-cosmos	`48.13% <ø> (+<0.01%)`	⬆️
sei-db	`68.72% <ø> (ø)`
sei-tendermint	`58.68% <69.72%> (?)`

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
sei-tendermint/internal/autobahn/avail/queue.go	`100.00% <100.00%> (ø)`
sei-tendermint/internal/autobahn/types/opt.go	`100.00% <100.00%> (ø)`
sei-tendermint/internal/p2p/giga/service.go	`100.00% <100.00%> (ø)`
sei-tendermint/internal/p2p/mux/mux.go	`76.88% <100.00%> (+76.88%)`	⬆️
sei-tendermint/internal/p2p/rpc/rpc.go	`94.82% <100.00%> (+82.54%)`	⬆️
...-tendermint/internal/autobahn/avail/block_votes.go	`85.71% <85.71%> (ø)`
sei-tendermint/internal/autobahn/types/app_vote.go	`81.81% <81.81%> (ø)`
...-tendermint/internal/autobahn/types/commit_vote.go	`81.81% <81.81%> (ø)`
...ei-tendermint/internal/autobahn/types/committee.go	`95.74% <95.74%> (ø)`
...endermint/internal/autobahn/types/lane_proposal.go	`84.61% <84.61%> (ø)`
... and 31 more

... and 317 files with indirect coverage changes

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

sei-tendermint/internal/autobahn/avail/inner.go

+	for _, vs := range i.votes[lane].q[n].byHeader {
+		if len(vs) >= c.LaneQuorum() {
+			return types.NewLaneQC(vs[:c.LaneQuorum()]), true
+		}
+	}


sei-tendermint/internal/autobahn/avail/inner.go

+	for lane := range i.votes {
+		lr := commitQC.LaneRange(lane)
+		i.votes[lr.Lane()].prune(lr.First())
+		i.blocks[lr.Lane()].prune(lr.First())
+	}


sei-tendermint/internal/autobahn/avail/subscriptions.go

+			for lane, bq := range inner.blocks {
+				for i := max(bq.first, r.next[lane]); i < bq.next; i++ {
+					batch = append(batch, bq.q[i].Msg().Block().Header())
+				}
+				r.next[lane] = bq.next
+			}


sei-tendermint/internal/autobahn/consensus/commit_votes.go

+	for _, v := range cv.byHash[h] {
+		votes = append(votes, v)
+	}


sei-tendermint/internal/autobahn/consensus/prepare_votes.go

+	for _, v := range pv.byHash[h] {
+		votes = append(votes, v)
+	}


sei-tendermint/internal/autobahn/types/proposal.go

+		for _, qc := range m.laneQCs {
+			laneQCs = append(laneQCs, qc)
+		}


sei-tendermint/internal/autobahn/types/testonly.go

+
+// GenProposal generates a random Proposal.
+func GenProposal(rng utils.Rng) *Proposal {
+	return newProposal(GenView(rng), time.Now(), utils.GenSlice(rng, GenLaneRange), utils.Some(GenAppProposal(rng)))


sei-tendermint/internal/autobahn/types/timeout.go

+	for _, l := range p.laneRanges {
+		laneRanges = append(laneRanges, l)
+	}


sei-tendermint/internal/p2p/giga/pb/api.pb.go

+	pb "github.com/tendermint/tendermint/internal/autobahn/pb"
+	protoreflect "google.golang.org/protobuf/reflect/protoreflect"
+	protoimpl "google.golang.org/protobuf/runtime/protoimpl"
+	reflect "reflect"


sei-tendermint/internal/p2p/giga/pb/api.pb.go

+	protoimpl "google.golang.org/protobuf/runtime/protoimpl"
+	reflect "reflect"
+	sync "sync"
+	unsafe "unsafe"


wen-coding · 2026-02-03T23:32:45Z

sei-tendermint/light/example_test.go

-		}
-		time.Sleep(time.Second)
+	t.Logf("wait for the blocks used in test")
+	utils.OrPanic1(waitForBlock(ctx, primary, 3))


Just curious, why do we need to change this test?

The test is flaky, it is not related to the autobahn in any sense.

wen-coding · 2026-02-03T23:55:35Z

sei-tendermint/internal/autobahn/autobahn.proto

 // Equivalent of `google.protobuf.Timestamp` but supports canonical encoding.
 // See `google.protobuf.Timestamp` for more detailed specification.
 message Timestamp {
  option (hashable.hashable) = true;


nit: comment what this does?

this is documented on the hashable package.

wen-coding · 2026-02-04T00:50:18Z

sei-tendermint/internal/p2p/mux/mux.go

 					}
 				}
 			}
+			// TODO(gprusak): this is counterintuitive asymmetric behavior:


Sorry I'm confused, didn't find this in v3 code, Is this migrated from some other repo?

No this is my own code. I've found a bug when plugging the mux into autobahn, which is a result of this inconsistency.

wen-coding · 2026-02-04T00:56:15Z

sei-tendermint/internal/p2p/giga/consensus.go

+
+// Ping implements pb.StreamAPIServer.
+// Note that we use streaming RPC, because unary RPC apparently causes 10ms extra delay on avg (empirically tested).
+func (x *Service) serverPing(ctx context.Context, server rpc.Server[API]) error {


And from here on is from stream/consensus/server.go?

Why don't we keep the original client/server separation?

the client and server implementation are tightly coupled, given that we use streaming RPCs. However, the way I've distributed the code across the files is kinda arbitrary, we can cleanup it later. A lot of it will need to be moved around anyway.

wen-coding · 2026-02-04T00:58:01Z

sei-tendermint/internal/p2p/giga/avail.go

+	"github.com/tendermint/tendermint/internal/p2p/rpc"
+)
+
+func (x *Service) serverStreamLaneProposals(ctx context.Context, server rpc.Server[API]) error {


So from here on it's stream/consensus/avail/server.go?

wen-coding · 2026-02-04T01:00:59Z

sei-tendermint/internal/p2p/giga/avail.go

+	})
+}
+
+func (x *Service) clientStreamLaneProposals(ctx context.Context, c rpc.Client[API]) error {


nit: If we have to put client and server in the same file, can we always do client first or always do server first maybe?

sure, this is a work-in-progress layout, definitely will need a cleanup.

wen-coding · 2026-02-04T23:00:23Z

sei-tendermint/internal/autobahn/avail/queue.go

@@ -0,0 +1,33 @@
+package avail
+


Since queue sounds like a general enough data structure, can you maybe comment why we need it defined in avail/, not at some more generic location?

my rule of thumb is to write general enough abstractions to avoid tangling logic. The question of visibility (where the code should be located, which packages should have access to it) is a totally separate question. So far this struct was under a lot of code churn, that's why it is in the package that is making use of it.

wen-coding · 2026-02-04T23:01:57Z

sei-tendermint/internal/autobahn/avail/subscriptions.go

@@ -0,0 +1,103 @@
+package avail
+


I believe this file is new, can you maybe explain a little bit why it's organized this way? How many receiver instances do we plan to have for each kind?

There is 1 receiver per connection currently. I wanted to separate autobahn from the transport layer. Autobahn api surface is way too large atm (number of receivers, custom receivers for each consensus message type, proper buffer sizes in rpcs, etc), it will require some work to simplify it. Or we can fallback to merge back the transport layer into autobahn (which is currently under p2p/giga). This is TBD.

wen-coding · 2026-02-04T23:25:11Z

sei-tendermint/internal/autobahn/consensus/commit_votes.go

+	for _, v := range cv.byHash[h] {
+		votes = append(votes, v)
+	}
+	cv.qc.Store(utils.Some(types.NewCommitQC(votes)))


Why are we changing from cv.qc.Update to cv.qc.Load and cv.qc.Store now?

thank for reminding me. The AtomicWatch has evolved since I implemented it in sei-v3. Now it is separated into AtomicSend and AtomicRecv and does not provide atomic updated by default - you need to wrap AtomicSend into a mutex - it simplifies the AtomicSend api AND allows AtomicSend to share mutex with other data, which is usually very convenient.

I've left a bunch of TODOs, because this change introduces some race conditions. Let me address them.

wen-coding · 2026-02-04T23:41:50Z

sei-tendermint/internal/autobahn/data/testonly.go

+	return types.NewLaneQC(votes)
+}
+
+func TestCommitQC(


Nice, I like that the helpers are all moved here.

wen-coding · 2026-02-04T23:43:54Z

sei-tendermint/internal/autobahn/producer/state.go

+	MaxGasPerBlock   uint64
+	MaxTxsPerBlock   uint64
+	MaxTxsPerSecond  utils.Option[uint64]
+	MempoolSize      uint64


Do we still have a mempool in Giga?

the block production pipeline in giga needs to be designed and implemented. I've copied over the producer for reference.

wen-coding · 2026-02-04T23:45:01Z

sei-tendermint/internal/p2p/giga/consensus.go

+)
+
+// Sends a consensus message to the peer whenever atomic watch is updated.
+func sendUpdates[T interface {


Okay, this one is moved from stream/consensus/client.go?

wen-coding · 2026-02-05T22:54:55Z

sei-tendermint/internal/autobahn/consensus/inner.go

-	if qc.Proposal().Index() < types.NextIndexOpt(i.viewSpec.CommitQC) {
-		return nil
+	for i := range s.inner.Lock() {
+		if qc.Proposal().Index() < types.NextIndexOpt(i.Load().viewSpec.CommitQC) {


nit: is this the same check as the one on line 26:
i.viewSpec.View().Index > qc.Proposal().Index()

Would help readability if we use the same expression to show that we are just rechecking the same condition.

wen-coding · 2026-02-05T22:59:27Z

sei-tendermint/internal/autobahn/consensus/state.go

+	inner := utils.Alloc(utils.NewAtomicSend(inner{}))
 	return &State{
 		cfg: cfg,
 		// metrics: NewMetrics(),


We do plan to add metrics back right?

pompon0 added 30 commits December 10, 2025 18:30

buf plugin draft

56fda78

plugin has been placed

9e34cae

fmt

e4ecde6

WIP

29364aa

separate module

e74e728

added canonical encoding

3f5798b

IsHashable

f8753b7

fix and test

510c3b9

test updates

d1b30ff

everything in place

9ad06cf

go fmt

117563c

Merge remote-tracking branch 'origin/main' into gprusak-proto

c27f7e6

Merge branch 'main' into gprusak-proto

77dc3a7

changed location of the plugin

4c7826a

Merge branch 'gprusak-proto' of https://github.com/sei-protocol/sei-c…

24a2e25

…hain into gprusak-proto

nits

118d566

panic msg

0189653

moved all hashable stuff together

c0a0ec5

moved testonly proto to internal

15faaa5

fmt

9794896

moved under internal

a82ac92

comment fix

50ae626

Merge branch 'main' into gprusak-proto

6160058

ed25519 cleanup wip

51c83d6

missing changes

e571075

compilation fix

bf6e999

Merge branch 'gprusak-proto' of https://github.com/sei-protocol/sei-c…

f435bee

…hain into gprusak-proto

Merge branch 'gprusak-proto' into gprusak-validator-proto

ac44467

wip - what about non-validator mode?

9a3edcb

wip

28bd678

pompon0 added 4 commits February 3, 2026 13:41

bugfixes

b656520

fmt

bb2296b

moved consensus state to GigaRouterConfig

4c16d0b

Merge branch 'main' into gprusak-router

60eb973

pompon0 changed the base branch from main to gprusak-router February 3, 2026 13:15

pompon0 added 2 commits February 3, 2026 14:18

Merge branch 'gprusak-router' into gprusak-autobahn

655a3d5

reverted mock gen

3bea1d6

pompon0 requested review from stevenlanders and wen-coding February 3, 2026 13:21

pompon0 added the non-app-hash-breaking label Feb 3, 2026

pompon0 marked this pull request as ready for review February 3, 2026 13:22

test fix

9e11e60

github-advanced-security bot found potential problems Feb 3, 2026

View reviewed changes

pompon0 added 5 commits February 3, 2026 14:47

lint

c4b8a65

more reliable test

0ae1466

Merge branch 'gprusak-router' into gprusak-autobahn

2d3dadc

test fix

6e2461f

Merge branch 'gprusak-router' into gprusak-autobahn

49f60bb

Base automatically changed from gprusak-router to main February 3, 2026 14:51

pompon0 added 3 commits February 3, 2026 15:52

Merge remote-tracking branch 'origin/main' into gprusak-router

8f53488

Merge branch 'gprusak-router' into gprusak-autobahn

f2f5c0b

test fix

9656b11

wen-coding reviewed Feb 4, 2026

View reviewed changes

pompon0 added 2 commits February 5, 2026 14:23

made the consensus transitions atomic again

62314b6

Merge branch 'main' into gprusak-autobahn

50ea1d9

pompon0 requested a review from wen-coding February 5, 2026 13:34

wen-coding approved these changes Feb 5, 2026

View reviewed changes

Conversation

pompon0 commented Feb 3, 2026

Uh oh!

github-actions bot commented Feb 3, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

codecov bot commented Feb 3, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

Check warning

Check warning

Check warning

Check warning

Check warning

Check warning

Check warning

Check warning

Check notice

Check notice

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

github-actions bot commented Feb 3, 2026 •

edited

Loading

codecov bot commented Feb 3, 2026 •

edited

Loading