Skip to content

Reapply "fix: use the correct l2_chain_id (#782)"#107

Merged
karlb merged 6 commits intodevelopfrom
karlb/revert-revert
Feb 10, 2026
Merged

Reapply "fix: use the correct l2_chain_id (#782)"#107
karlb merged 6 commits intodevelopfrom
karlb/revert-revert

Conversation

@karlb
Copy link

@karlb karlb commented Feb 3, 2026

This reverts commit 18c6583 and makes it work including the previously reverted upstream commit.

Hokulea used kona-registry for L2 chain configs while we used celo-registry in other places. The solution is to pass the correct information to hokulea, so that it doesn't have to look it up itself.

See also the upstream PR to accept the BootInfo in hokulea.

Closes https://github.com/celo-org/celo-blockchain-planning/issues/1333

@seolaoh
Copy link
Collaborator

seolaoh commented Feb 3, 2026

The issue occurred while executing the program in range proof generation. The reason why you didn't get the error when running locally is that when you don't set these vars SP1_PROVER=mock OP_SUCCINCT_MOCK=true, the cost estimator skips the local execution simulation and just tries to request the range proof generation to the prover network. So you might be able to reproduce the error with either way:

  • run SP1_PROVER=mock OP_SUCCINCT_MOCK=true RUST_LOG=info cargo run --bin cost-estimator --release --features eigenda -- --rolling
  • set the proper NETWORK_PRIVATE_KEY and run RUST_LOG=info cargo run --bin cost-estimator --release --features eigenda -- --rolling, you can see the proof generation fails in the prover explorer.

@karlb
Copy link
Author

karlb commented Feb 3, 2026

Yes, I'm able to reproduce it with SP1_PROVER=mock OP_SUCCINCT_MOCK=true. The problem might be that hokulea relies on kona-registry instead of celo-registry for the L2 definitions. I'm trying to verify this theory.

@karlb
Copy link
Author

karlb commented Feb 3, 2026

I found a way to make this work with a small change in Hokulea and updated the PR description accordingly.

@github-actions
Copy link

github-actions bot commented Feb 3, 2026

Performance Comparison (ELF: eigenda-range-elf-embedded)

Range 17449485~17449490

Metric Base Branch Current PR Diff (%)
Total Instructions 2325364144 2311278808 -0.61%
Oracle Verify Cycles 113038701 113038645 -0.00%
Derivation Cycles 919613698 900663952 -2.06%
Block Execution Cycles 93458193 93421002 -0.04%
Blob Verification Cycles 728 728 0.00%
Total SP1 Gas 3024459535 3011341099 -0.43%
Cycles per Block 465072828 462255761 -0.61%
Cycles per Transaction 387560690 385213134 -0.61%
BN Pair Cycles 0 0 0.00%
BN Add Cycles 0 0 0.00%
BN Mul Cycles 0 0 0.00%
KZG Eval Cycles 0 0 0.00%
EC Recover Cycles 0 0 0.00%
P256 Verify Cycles 0 0 0.00%

@karlb karlb marked this pull request as ready for review February 3, 2026 16:17
@karlb
Copy link
Author

karlb commented Feb 4, 2026

As mentioned in Layr-Labs/hokulea#301 (comment), another way to make this work is to remove Celo Sepolia from celo_registry until it is also available in the superchain_registry used by kona.

Copy link
Collaborator

@seolaoh seolaoh left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Great to find the root cause! Why don't we remove hokulea patch and use upstream version directly since your PR is merged in upstream?

This reverts commit 18c6583.

The next commits should resolve the problems that made it necessary to
skip that commit.
so that hokulea does not have to rely on kona-registry.
Layr-Labs/hokulea@bf561cc
is merged in upstream and provides what we need, so we can stop
depending on our own hokulea fork.
@karlb karlb merged commit b156f48 into develop Feb 10, 2026
14 checks passed
@karlb karlb deleted the karlb/revert-revert branch February 10, 2026 15:35
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants