Repository navigation
Conversation
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
Contributor
|
Important Review skippedToo many files! This PR contains 112 files, which is 12 over the limit of 100. To get a review, reduce the PR to 100 files or fewer by splitting it into smaller PRs or changing its base branch. Upgrade to a paid plan to raise the limit. CodeRabbit couldn't verify your organization's usage billing state, so no usage-priced review was started. Retry after the billing service recovers.
|
ctrlc03
force-pushed
the
large/params-n45
branch
from
November 20, 2026 09:48
2072bd1 to
9365aa1
Compare
ctrlc03
force-pushed
the
large/params-n45
branch
from
November 20, 2026 17:21
9365aa1 to
f205b56
Compare
ctrlc03
force-pushed
the
large/params-n45
branch
from
November 21, 2026 15:26
f205b56 to
f177b67
Compare
ctrlc03
force-pushed
the
large/params-n45
branch
from
November 21, 2026 16:27
f177b67 to
52156dd
Compare
ctrlc03
changed the base branch from
large/c4-per-limb
to
large/nodes-fold-three
November 21, 2026 16:29
Split from #2398. Raise the secure-8192 and insecure-64 search to n = 45 (rerun of search_params, the insecure one with lambda 2 and 12-bit primes), so one parameter set serves every committee up to the large one. The moduli are unchanged; the threshold error1 variance grows (secure B_Enc 518814677073081139200, insecure 460800), so the smudging and user-encryption e0 bounds grow, and the insecure e0 bound gains one bit. Correctness margins at n = 45: secure 2.07 bits, insecure 2.52 bits. Both presets' parameter hashes and every configuration identity change (BfvProtocol.sol, crypto-config.json and the independently recomputed ethers fixtures). Compiled artifacts, the VK catalogue and the BFV verifiers are regenerated separately.
- upgrades.md replaces the unresolved registry-index gap with the cutover procedure: register index 3 (secure) and index 4 (insecure, testnets) in the paused governance batch, check the exact bytes before resume, and keep indices 0, 1 and 2 as historical registry reads. - The developer pages, the committee-selection note and the SDK README name indices 3 and 4.
…er batch - `upgrade:v19 prepare` registered only the secure index 3, and `validate` checked only that index. Sepolia requests default to the insecure index 4, so a resumed Sepolia would reject them with "BFV param set not registered". - `prepare` now adds one `setParamSet` call for each parameter set of the chain that the registry lacks, and stops if an index holds other bytes. - `validate`, and `resume` through it, check every parameter set of the chain. - The plan records `paramSetsRegisteredByBatch` in place of `paramSet` and `paramSetRegisteredByBatch`. - The cutover test uses index 3 and covers a batch that registers two sets.
…ameter-set branch
…set 3 sync-secure-8192 takes the SS secure threshold preset, so it follows that preset's move to registry index 3, as sync-insecure-64 already follows index 4.
…y the rebase onto main
Pre-cutover E3s at slots 0 and 2 now resolve through the registry bytes rather than a local preset. The fake Interfold gains setParamSet, and a backfill test registers the historical insecure v5 tuple at slot 0: a request under its threshold-only ID is stored with those exact bytes, and a request under another ID at the slot is skipped.
Sepolia index 0 holds the degree-64 v5 tuple of a later redeployment, not the v0.19.x tuple; a fresh Mainnet deployment registers index 3 only; an unregistered historical index fails as not registered before the crypto-config check; clients derive the index from bfvConfiguration in @interfold/contracts/bfv, since the SDK does not export ParamSet; and a rollback leaves indices 3 and 4 registered but inert.
Define a fourth committee size beside minimum, micro and small. It is wire variant 3 of CiphernodesCommitteeSize and IInterfold.CommitteeSize, so the published enum values stay. The shared committee owner carries N = 45, H = 34 and T = 22, and the generated Noir, Solidity and TypeScript bindings follow it. - Both SS profiles admit the committee, so each gains a configuration ID (SS_INSECURE_64_LARGE_CONFIG_ID, SS_SECURE_8192_LARGE_CONFIG_ID). ActiveCryptoConfig accepts the size and returns those IDs. - build-pairs.json builds insecure-64/large and secure-8192/large. The release matrix and the chain deployment routes do not change here. - The SDK, the hardhat request task, the circuit and committee checks, the Rust enum consumers and the docs list the fourth size. Breaking: nodes that predate the variant cannot decode a Large request.
Three node proofs per step stay within 2^22 at n = 45 (3.71M gates).
…over its thresholds
The CRISP server rejects E3_COMMITTEE_SIZE above 2 at startup (theinterfold/CRISP server/src/config.rs), so the settings table must not offer 3 for the large committee.
The DKG aggregation shape test and the C1/C2 commitment width test name every canonical committee, and they stopped at small.
The dashboard hard-coded T for Minimum, Micro and Small, so a Large E3 showed 0 of 45 shares required. It now reads COMMITTEES from @interfold/contracts/bfv. The mainnet notes in ActiveCryptoConfig.sol and committee-selection.md say the contract accepts any size whose threshold is set and the configuration keeps mainnet to minimum/micro/small; the circuit-artifacts rewrap no longer splits 'provenance/ publication'.
…meters and the large committee From the matrix build of stamp 5c2ddf13094ffd62 (published as circuit-artifacts 86a25e3): the named VK catalogue for the eight build pairs; the Honk verifiers, including insecure-64/large and secure-8192/large; the layout locks for CiphernodesCommitteeSize::Large on top of the circuits PR's rows; the BFV VK-binding fixture; and both synchronized Minimum candidate manifests, requalified after the 45-party search (each has a new C6 pin and configuration ID; the C7 and outer pins are unchanged). The benchmark note records the secure C6 gate count after the wider noise checks.
committee-selection.md said Sepolia and local chains set Large; the routes and thresholds that do so come in the next PR. The BfvProtocol spec's invalid-committee case passed parameter set 0, which no longer exists, so it reverted on the parameter set and never reached the committee check; it now uses ss-insecure-64's set. The CommitteeSize encoding test records that the hidden __Invalid sentinel moves from 3 to 4.
ctrlc03
force-pushed
the
large/params-n45
branch
from
November 21, 2026 16:59
52156dd to
0afbe68
Compare
ctrlc03
force-pushed
the
large/params-n45
branch
from
November 21, 2026 19:51
0afbe68 to
d77c814
Compare
This branch was successfully deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What and why
Split from #2398 (Zara's work). Stacked on #2531.
Adds the Large committee (N = 45, H = 34, T = 22). The BFV parameter search was sized for 19 parties,
so this PR first resizes it for 45 parties, registers the new parameter sets at new registry indices,
and then adds the committee.
BFV parameters for 45 parties
SEARCH_Nis 45 for the secure and insecure searches. The error-1 variance constants and thebuilder test vectors follow the new result.
BfvProtocol.solandcrypto-config.jsoncarry the new parameterhashes (insecure
0x2898f8a9…a844, secure0x4f402598…08c3) and the config IDs derived from them.bfv_protocol.jsonandbfv_profiles.jsonare recomputed independently with ethers.too. sync-secure-8192 gets a new configuration ID (
0xe3996b37…) and moves to parameter set 3.generator now emits
C6_NOISE_BIT(145) andC6_SHIFTED_NOISE_BIT(146), and the statement takesthem as parameters, with static asserts. A unit test keeps the bound within 145 bits.
Parameter sets at registry indices 3 and 4
setParamSetis append-only, and live registries already hold earlier tuples at indices 0–2. Thenew tuples take index 3 (secure-8192) and index 4 (insecure-64); the old ones stay readable.
registry bytes under their frozen IDs, so pre-cutover E3s still backfill.
ParamSetenum,fake_interfold.soland the contract tests use the new indices.upgrade:v19 prepareregisters every parameter set the chain lacks (3 on Mainnet; 3 and 4 onSepolia) and stops if an index holds other bytes.
validatechecks them all.docs/architecture/upgrades.mdgives the cutover procedure.Large committee
CiphernodesCommitteeSize::LargeandIInterfold.CommitteeSizevariant 3, appended after Small.Both SS profiles admit it, each with a new configuration ID.
ActiveCryptoConfigaccepts the size;Interfold.MAX_COMMITTEE_SIZErises from 19 to 45. TheSDK,
bfv.ts, the harness, the hardhat request task, the EVM enum and the dashboard follow.build-pairs.jsonbuildsinsecure-64/largeandsecure-8192/large. Deployment routes, therelease matrix and the mainnet configuration are unchanged here; feat(contracts)!: route the large committee on testnets and release it #2537 adds them.
Regenerated outputs from one matrix build (stamp
5c2ddf13094ffd62, published tocircuit-artifacts): the VK catalogue, the Honk verifiers for the two Large pairs (the sixexisting pairs' Honk verifiers regenerate unchanged), the layout-lock rows for
Large, the BFVVK-binding fixture, and both synchronized Minimum candidate manifests.
Verification
cargo check,cargo fmt --check,nargo checkwith the Large committee selectedfor both presets,
scripts/circuits/lint.sh,scripts/checks/committee.sh, blueprint check,crypto-config.mjs --check --sources,check:crypto-config --full,pnpm check:bindings,prettier,
tsc --noEmit.26.6 GiB peak). The VK-binding check passes.
paramSetRegistrybytes (read-only
eth_call); indices 3 and 4 are empty on both chains.Compatibility
Breaking within the unreleased protocol-9 cutover (no separate counter change).
cryptoConfigIdchange, and with them the VK catalogue anchorsand the BFV verifiers deployed for each route.
getE3Quoteandrequestreject 0–2, and nodes skiprequest events that name them. Clients that hard-code an index, such as the CRISP server's
E3_PARAM_SET, move to 3 (secure) or 4 (insecure).setParamSet(3, …)on Mainnet and Sepolia andsetParamSet(4, …)onSepolia. Before resume, check
keccak256(paramSetRegistry(3))=0x4f402598…08c3andkeccak256(paramSetRegistry(4))=0x2898f8a9…a844.paramSet/paramSetRegisteredByBatchbecomesparamSetsRegisteredByBatch: number[].mainnet routes are unchanged.