[smtmfft]

Summary

• bump the taiko-geth replacement from the Mainnet Unzen activation revision to the canonical taiko branch merge commit 4e001283f48841068c28fd5335dd784d4ed96bca
• consume taiko-geth's exact 7-byte Shasta extraData validation, Masaya removal, and merged EIP-4788 sealing/import consistency fix from taiko-geth #601
• keep Gaiko2 proving, zkGas, configuration, and release behavior unchanged

Compatibility

• the merged revision resolves as v1.18.1-0.20260922032335-4e001283f488
• the only taiko-geth delta from the previously reviewed fe4f45c83f7a pin is merged PR #601
• Gaiko2 already validates the exact Shasta manifest extraData, requires an Unzen parent beacon root to be present and zero, and executes the EIP-4788 system call before transaction replay; no Gaiko2 source migration is required
• this PR does not build, publish, register, or deploy a new enclave image

Verification

• go list -m -json github.com/ethereum/go-ethereum
• go mod verify
• go mod tidy -diff
• go test ./...
• go test -race ./...
• go vet ./...
• go build ./...
• focused Unzen parent-beacon-root and shared Shasta fixture replay tests
• git diff --check origin/main

Independent review and behavioral verification found no material compatibility or module-provenance issue.

[cursor]

Stale comment

Approved: Cursor Bugbot completed successfully with no findings that need human review, and no approval policy requires extra review for this dependency pin. Reviewers were not assigned.

  

_{Sent by Cursor Approval Agent: Pull Request Router}

Dismissed the prior approval: Cursor Bugbot is still pending after the wait window, so this head no longer meets the approval criteria. Human review is needed; reviewers were not assigned because two are already requested.

[davidtaikocha]

Approving. I reviewed the taiko-geth changes and how gaiko2 uses them.

What the bump pulls in (compared with main's 8f5abc6d0636)

• Four taiko-geth PRs: #591, #577 (release 2.6.0), #598 and #601.
• 4e001283f488 is the current tip of taiko and builds directly on the old pin.
• taiko-geth's go.mod is byte-identical, so no other dependencies change.
• The go.sum hash matches sum.golang.org.

Effect on gaiko2

• Of the changed packages, gaiko2 only links params, core and core/taiko_genesis, test binaries included. Their only change is the Masaya removal. gaiko2 references none of it, and chain 167011 is still rejected by chainConfigFor.
• #591 and #601 land in consensus/taiko and miner, which gaiko2 does not link. gaiko2 already enforces the same rules itself:
  • Shasta extraData must be exactly 7 bytes;
  • Unzen parentBeaconRoot must be present and zero;
  • EIP-4788 then EIP-2935 run before transactions.
• A throwaway side-by-side test with the 4788/2935 contracts deployed matched taiko-geth import under both pins.
• The tx-error classifier and zk-gas schedule that gaiko2 copies from taiko-geth are unchanged upstream.

Verified locally at 15b9e58: go mod verify, go mod tidy -diff, gofmt, build, vet, go test ./... and the race subset all pass.

Non-blocking notes

• PR description wording. It says gaiko2 "consumes" #591 and #601, but gaiko2 doesn't link those packages; its own existing checks provide the same protection. "The only delta is #601" is true relative to this PR's first commit; compared with main it is the four PRs above.
• No Unzen replay test. No checked-in test executes an Unzen block: processUnzenReplayBlock has 0% coverage and the mainnet fixture predates Unzen. So CI can't catch an Unzen execution change from a future bump. Mainnet has been on Unzen since 2026-08-06, so a post-Unzen fixture can be captured now.
• Untagged pin. The pin is v2.6.0 plus #598 and #601, and has no tag. Worth re-pinning to the v2.7.0 release commit once taikoxyz/taiko-geth#599 lands.