Lines of code report

Total lines added: 817
Total lines removed: 8
Total lines changed: 825

+------------------------------------------------------------------------+-------+------+
| File                                                                   | Lines | Diff |
+------------------------------------------------------------------------+-------+------+
| ethrex/crates/blockchain/constants.rs                                  | 18    | +2   |
+------------------------------------------------------------------------+-------+------+
| ethrex/crates/blockchain/error.rs                                      | 149   | +2   |
+------------------------------------------------------------------------+-------+------+
| ethrex/crates/blockchain/fork_choice.rs                                | 169   | +21  |
+------------------------------------------------------------------------+-------+------+
| ethrex/crates/blockchain/mempool.rs                                    | 464   | +11  |
+------------------------------------------------------------------------+-------+------+
| ethrex/crates/blockchain/payload.rs                                    | 841   | +28  |
+------------------------------------------------------------------------+-------+------+
| ethrex/crates/common/types/block_access_list.rs                        | 1107  | +15  |
+------------------------------------------------------------------------+-------+------+
| ethrex/crates/common/types/genesis.rs                                  | 1021  | +3   |
+------------------------------------------------------------------------+-------+------+
| ethrex/crates/common/validation.rs                                     | 243   | -2   |
+------------------------------------------------------------------------+-------+------+
| ethrex/crates/l2/sequencer/block_producer/payload_builder.rs           | 271   | +37  |
+------------------------------------------------------------------------+-------+------+
| ethrex/crates/networking/rpc/engine/fork_choice.rs                     | 565   | +13  |
+------------------------------------------------------------------------+-------+------+
| ethrex/crates/networking/rpc/rpc.rs                                    | 1189  | +1   |
+------------------------------------------------------------------------+-------+------+
| ethrex/crates/networking/rpc/utils.rs                                  | 335   | +7   |
+------------------------------------------------------------------------+-------+------+
| ethrex/crates/vm/backends/levm/mod.rs                                  | 2357  | +147 |
+------------------------------------------------------------------------+-------+------+
| ethrex/crates/vm/levm/src/call_frame.rs                                | 390   | +16  |
+------------------------------------------------------------------------+-------+------+
| ethrex/crates/vm/levm/src/environment.rs                               | 98    | +1   |
+------------------------------------------------------------------------+-------+------+
| ethrex/crates/vm/levm/src/errors.rs                                    | 244   | +3   |
+------------------------------------------------------------------------+-------+------+
| ethrex/crates/vm/levm/src/gas_cost.rs                                  | 875   | +19  |
+------------------------------------------------------------------------+-------+------+
| ethrex/crates/vm/levm/src/hooks/default_hook.rs                        | 536   | +78  |
+------------------------------------------------------------------------+-------+------+
| ethrex/crates/vm/levm/src/hooks/l2_hook.rs                             | 669   | -6   |
+------------------------------------------------------------------------+-------+------+
| ethrex/crates/vm/levm/src/opcode_handlers/stack_memory_storage_flow.rs | 327   | +1   |
+------------------------------------------------------------------------+-------+------+
| ethrex/crates/vm/levm/src/opcode_handlers/system.rs                    | 1063  | +114 |
+------------------------------------------------------------------------+-------+------+
| ethrex/crates/vm/levm/src/utils.rs                                     | 604   | +144 |
+------------------------------------------------------------------------+-------+------+
| ethrex/crates/vm/levm/src/vm.rs                                        | 709   | +151 |
+------------------------------------------------------------------------+-------+------+
| ethrex/crates/vm/lib.rs                                                | 16    | +3   |
+------------------------------------------------------------------------+-------+------+

Benchmark Results Comparison

No significant difference was registered for any benchmark run.

⚠️ Known Issues — intentionally skipped tests

Source: docs/known_issues.md

Known Issues

Tests intentionally excluded from CI. Source of truth for the Known
Issues section the L1 workflow appends to each ef-tests job summary
and posts as a sticky PR comment.

Hive — bal-devnet-6 Amsterdam consume-engine tests — 32 functions / 54 cases

Same root cause as the blockchain-runner skip list (see EF Tests —
Blockchain below): snobal-devnet-6 fixtures expect bal-devnet-6 spec
semantics, but our impl runs ahead due to the bal-devnet-7-prep
set_delegation SELFDESTRUCT-style refund subtraction. These fixtures
are routed through hive's eels/consume-engine simulator and produce
the same failures. Excluded via KNOWN_EXCLUDED_TESTS (substring
match on test_<name>[fork_Amsterdam, anchoring to the Amsterdam fork
so legacy Prague/Osaka variants still run).

EF Tests — Stateless coverage narrowed to EIP-8025 optional-proofs

make -C tooling/ef_tests/blockchain test calls test-stateless-zkevm
instead of test-stateless. The zkevm@v0.3.3 fixtures are filled against
bal@v5.6.1, out of sync with current bal spec; the broad target trips ~549
fixtures. Re-broaden once the zkevm bundle is regenerated.

EF Tests — Blockchain bal-devnet-6 (Amsterdam fork) — 74 tests

snobal-devnet-6 fixtures expect bal-devnet-6 spec semantics, but our impl
runs ahead due to the bal-devnet-7-prep set_delegation SELFDESTRUCT-style
refund subtraction. Skipped in
tooling/ef_tests/blockchain/tests/all.rs::SKIPPED_BASE, anchored on
[fork_Amsterdam so legacy Prague / Osaka variants still run.

🤖 Kimi Code Review

⚠️ Warning: Diff was truncated from 10193 to 10000 lines. Some changes were not reviewed.

This is a substantial PR implementing the Amsterdam fork (EIP-7928 Block Access Lists, EIP-8037 Multidimensional Gas, EIP-7976/7981 Floor Cost adjustments) with extensive testing. Below is the review feedback organized by category.

Critical Consensus & Safety

1. Fork Choice Reorg Depth Validation

File: crates/blockchain/fork_choice.rs
Lines: 73-81, 95-128

The implementation of execution-apis PR #786 correctly distinguishes between:

• Spec check: Rejecting reorgs that replace finalized blocks (canonical_link_height < stored_finalized)
• Implementation cap: The 128-block state-history limit (REORG_DEPTH_LIMIT)

Issue: The error variant TooDeepReorg is used for both cases but with different limit values (spec limit vs. 128). This could confuse operators debugging reorg rejections. Consider splitting into two error variants or clarifying the message.

// Line 118-122: Spec violation
return Err(InvalidForkChoice::TooDeepReorg {
    reorg_depth,
    limit: latest.saturating_sub(stored_finalized), // This is the spec limit
});

// Line 127-131: Implementation limit  
return Err(InvalidForkChoice::TooDeepReorg {
    reorg_depth,
    limit: REORG_DEPTH_LIMIT, // This is the implementation cap
});

2. BAL Index Type Widening (u16 → u32)

File: crates/common/types/block_access_list.rs
Status: ✅ Correctly implemented

The change from u16 to u32 for BlockAccessIndex is properly propagated through:

• RLP encoding/decoding
• Storage change tracking
• Validation logic (validate_bal_indices)
• Test assertions

The unwrap_or(u32::MAX) fallbacks in payload building (lines 464, 651) are acceptable since >4 billion transactions per block is physically impossible, but consider expect with a message for clarity.

VM & Gas Accounting

3. EIP-8037 Clamp-and-Spill Refund Logic

File: crates/vm/levm/src/vm.rs
Lines: 653-710 (credit_state_gas_refund)

The refund mechanism splitting credits between applied_to_spill and applied_to_drain is subtle. The debug assertions are good, but verify the invariant:

// Line 691-699: This math must exactly match EELS fork.py
let frame_outstanding_delta = self
    .state_gas_spill_outstanding
    .saturating_sub(self.current_call_frame.state_gas_spill_outstanding_snapshot);
let applied_to_spill = clamped.min(frame_outstanding_delta);

Risk: If state_gas_spill_outstanding_snapshot is not correctly captured at frame entry, the split will be wrong, causing consensus failures on contracts with nested CREATE/SSTORE patterns.

4. Intrinsic Gas Parity

File: crates/vm/levm/src/utils.rs (lines 737-890)
File: crates/blockchain/mempool.rs (lines 516-538)

The standalone intrinsic_gas_dimensions function must stay byte-for-byte compatible with VM::get_intrinsic_gas. The parity tests in test_intrinsic_parity_* are essential—ensure they run in CI for every fork.

Note: The Amsterdam CREATE intrinsic split (REGULAR_GAS_CREATE + state portion) is correctly implemented in both paths.

5. System Call Gas Limit

File: crates/vm/backends/levm/mod.rs (line 2542)

System calls now use the actual block_gas_limit instead of i64::MAX for EIP-8037 CPSB calculation:

block_gas_limit: block_header.gas_limit,  // Was i64::MAX
is_system_call: true,

This prevents OOG on state-gas-heavy system calls (EIP-2935, EIP-4788) when cost_per_state_byte is large. ✅ Correct.

Block Building & Validation

6. Builder/Validator Parity

File: test/tests/blockchain/builder_validator_parity_tests.rs

This new test suite is critical infrastructure. It verifies that blocks built via build_payload pass validation via add_block_pipeline_bal.

Gap: The validate_corrupted_bal helper (lines 195-210) mutates the BAL and re-hashes the header, but doesn't test the case where the BAL hash itself is invalid (corrupted but matching hash). Consider adding a test for hash mismatches.

7. 2D Gas Inclusion Check

File: crates/blockchain/payload.rs (lines 656-670)

The EIP-8037 per-transaction 2D check is correctly placed before BAL recording:

// EIP-8037 ... per-tx 2D inclusion check against running block totals. Run BEFORE we touch...
if context.is_amsterdam
    && let Err(e) = check_2d_gas_allowance(...)
{
    return Err(e.into());
}

This prevents rejected transactions from polluting the BAL. ✅ Correct.

Testing & Tooling

8. Fixture Version Management

File: tooling/ef_tests/blockchain/Makefile

The separation of vectors (snobal-devnet-6) and vectors_zkevm (older bal@v5.6.1) prevents stale fixtures from clobbering new implementations. The test-stateless-zkevm target narrowing to eip8025_optional_proofs is a pragmatic workaround until zkevm regenerates fixtures.

Action Required: Add a TODO comment in the Makefile with a link to the zkevm fixture regeneration tracking issue so this narrowing can be removed later.

9. Known Issues Documentation

File: docs/known_issues.md

Excellent documentation of the 74 intentionally skipped Amsterdam tests due to the bal-devnet-6 vs devnet-7-prep semantic mismatch. The categorization by EIP (7702, 8037, 7928, etc.) is helpful for future triage.

Suggestion: Add the specific commit hash or date when these can be re-enabled (e.g., "Re-enable after fixture bump to snobal-devnet-7").

Minor Issues

10. Error Code Documentation

File: crates/networking/rpc/utils.rs (line 25)

The comment lists error codes -38001 to -38006, but the TooDeepReorg variant uses -38006. Verify this doesn't conflict with any existing Engine API specs (it appears correct per PR #786).

11. Prestate Tracer Storage Filtering

File: crates/vm/backends/levm/tracing.rs (lines 188-215)

The build_pre_state_map function filters storage to only include touched slots. The logic for merging initial_accounts_state with pre_snapshot correctly handles slots first accessed in the current transaction.

Optimization: The retain closure (lines 206-213) could be extracted to a helper for readability, but functionally correct.

Summary

Approvals:

• ✅ EIP-8037 reservoir and clamp-and-spill implementation matches spec complexity
• ✅ EIP-7928 BAL validation with shadow recorder is robust
• ✅ Fork choice reorg depth limits correctly implement safety checks
• ✅ Comprehensive test coverage for edge cases (SELFDESTRUCT refunds, nested halts, 0→N→0 SSTORE)

Recommendations:

1. High: Verify the credit_state_gas_refund split math against the latest EELS reference implementation commit (e87390b) before merging—this is the highest risk consensus code.
2. Medium: Add explicit test for TooDeepReorg error message clarity to distinguish spec vs implementation limits.
3. Low: Add TODO comment in Makefile for zkevm fixture regeneration tracking.

The PR is well-structured for a complex fork implementation. The builder/validator parity tests are particularly valuable for preventing consensus failures in production block building.

Automated review by Kimi (Moonshot AI) · kimi-k2.5 · custom prompt

🤖 Claude Code Review

Code Review: PR 6574 — bal-devnet-6 Support (EIP-7928 BAL + EIP-8037 2D Gas)

This is a large, well-motivated PR. The architecture and documentation are clear, and the overall approach is sound. There are a few correctness and robustness issues worth addressing before merge.

Overview

• EIP-7928 (BAL): u16 → u32 index widening, shadow recorder, builder/validator parity.
• EIP-8037 (2D Gas): Full clamp-and-spill reservoir model, cost_per_state_byte pinned to 1174 for devnet-6.
• Engine API: execution-apis PR 786 FCU semantics (TooDeepReorg, NewHeadAlreadyCanonical payload building).
• Tracing: prestateTracer for debug_traceTransaction / debug_traceBlockByNumber.

High Severity

1. Panic on None in NewHeadAlreadyCanonical branch (crates/networking/rpc/engine/fork_choice.rs, around the NewHeadAlreadyCanonical match arm)

get_block_header_by_hash returns Option<BlockHeader>. Using ? on it will propagate a Result error but will panic on None because Option::None cannot be propagated via ? without ok_or(). While the block was validated earlier, defensive handling is required:

// Current (panics on None):
let head_block = context.storage
    .get_block_header_by_hash(fork_choice_state.head_block_hash)?;

// Should be:
let head_block = context.storage
    .get_block_header_by_hash(fork_choice_state.head_block_hash)?
    .ok_or_else(|| RpcErr::Internal("head block header not found".into()))?;

2. prestateTracer misses read-only (warm) accounts (crates/vm/backends/levm/tracing.rs, find_touched_accounts)

The implementation iterates post_cache to find changed accounts. Accounts that are accessed (via EXTCODESIZE, BALANCE, SLOAD, etc.) but not modified never enter post_cache — they stay in initial_accounts_state. Geth's prestateTracer includes all accessed accounts, not just modified ones. This produces an incomplete prestate map for any tx that reads accounts without writing them. This is a semantic gap vs. geth.

3. apply_same_tx_selfdestruct_state_refund iterates accounts with a shrinking pool (crates/vm/levm/src/vm.rs)

When a tx selfdestructs multiple accounts, each iteration computes net_state_gas fresh but the loop also increments state_gas_refund_absorbed each iteration. This means the second account's refund is clamped against a smaller pool than the first, regardless of their combined entitlement. If EELS applies a single aggregate clamp to the total refund, this diverges from reference client behavior. Worth verifying against the EELS apply_same_tx_selfdestruct_state_refund implementation.

Medium Severity

4. Spurious -38006 TooDeepReorg on genesis-head FCU when finalized is nonzero (crates/networking/rpc/engine/fork_choice.rs, canonical_link_height computation)

When head.number == 0, saturating_sub(1) gives canonical_link_height = 0. The spec guard canonical_link_height < stored_finalized evaluates to 0 < stored_finalized, which is true if finalized is at any block above genesis. This triggers -38006 for a legitimate sync-from-scratch FCU(genesis) when the stored finalized is nonzero. A sync scenario where the CL sends FCU(genesis) after the node has been synced before could hit this.

5. L2 payload builder does not restore block_regular_gas_used / block_state_gas_used on apply_plain_transaction error (crates/l2/sequencer/block_producer/payload_builder.rs)

The L2 builder snapshots previous_block_regular_gas_used but only restores it in the invalid-message break branch, not in the apply_plain_transaction Err continue branch. If a tx passes the 2D check inside execute_tx but then fails with a state-gas overflow error (post-counter-update), the 2D counters are permanently inflated for subsequent txs in the same block. Contrast with the L1 builder (apply_tx_to_payload) where check_2d_gas_allowance runs before execution, so the Err path is safe. The L2 builder needs a matching pre-snapshot/restore pattern.

6. trace_block_prestate aborts on any tx execution error

If any tx in the block fails during replay, the function propagates the error and stops, leaving subsequent txs untraced. Geth continues tracing through failed txs (the prestate is still valid). The fix is to catch the error per-tx, record an empty or partial prestate, and continue iterating.

Low Severity / Maintainability

7. Three copies of EIP-7976/7981 floor gas logic — default_hook.rs, utils.rs (VM), and utils.rs::intrinsic_gas_floor (standalone). The parity tests catch most divergences now, but future edits require triple-updating. A single canonical function would eliminate the risk.

8. test_inner_create_deposit_oog_discard is #[ignore]'d with a stale calibration — the comment assumes cpsb(1_000_000) = 1, but with the current 1174 pin the test would also fail because state_gas_storage_set = 37_568 causes the outer CALL to OOG before reaching CREATE. The ignore comment should note that the test is broken under the 1174 pin, not just "disabled because of calibration."

9. Missing test: EIP-4844 blob tx through intrinsic_gas_dimensions — the parity suite covers plain transfer, CREATE, calldata+access list, and EIP-7702 auth-list, but not blob transactions. The Amsterdam is_amsterdam_activated fast-return in transaction_intrinsic_gas bypasses blob checks; if intrinsic_gas_dimensions handles them differently from the legacy path this would be silently wrong.

10. Deeply nested mixed-success/failure call trees for state_gas_refund_pending — the credit_state_gas_refund accounting chain is correct at 1-hop and 2-hop depth (as tested), but the interaction of pending refunds across 3+ hop call trees with mixed success/failure at intermediate frames is not tested. The debug_assert_eq! guards catch most invariant violations in debug builds, but production builds would silently compute incorrect gas.

Positive Notes

• The u16 → u32 widening is consistent across all encode/decode/validate paths and is well-guarded by RLP round-trip tests including the WIDE_IDX = u32::MAX / 2 boundary case.
• check_2d_gas_allowance formula correctly mirrors EELS fork.py:560-578, including the saturating_sub guard for intrinsic-underfunded txs.
• The parallel execution 2D check is correctly serialized in aggregation order, matching EELS serial semantics despite parallel execution.
• intrinsic_state_gas_charged double-set guard with debug_assert_eq! is a nice belt-and-suspenders pattern.
• BAL SYSTEM_ADDRESS filtering is correct and covered by parity tests in both directions.
• The fe959e3bd fix (build payload on NewHeadAlreadyCanonical with attributes + set_synced()) is correct per spec and the regression it clears (Hive Cancun Engine, Rpc Compat, Devp2p) validates the fix.

Automated review by Claude (Anthropic) · sonnet · custom prompt

🤖 Codex Code Review

1. High: crates/blockchain/blockchain.rs still rejects any Osaka+ tx with gas_limit > 16_777_216, but the VM explicitly drops that rule from Amsterdam onward in crates/vm/levm/src/hooks/default_hook.rs. The builder path already assumes Amsterdam txs can exceed the regular cap and only reserves min(tx.gas_limit, TX_MAX_GAS_LIMIT_AMSTERDAM) in crates/blockchain/payload.rs. Result: valid Amsterdam reservoir-style txs can never enter the mempool.

2. Medium: crates/blockchain/fork_choice.rs returns NewHeadAlreadyCanonical before the safe/finalized connectivity checks at crates/blockchain/fork_choice.rs, and crates/networking/rpc/engine/fork_choice.rs now treats that case as a full success and returns a head header for payload building. A disconnected safeHash / finalizedHash inside the finalized prefix can now get VALID (and potentially a payload ID) instead of InvalidForkChoiceState.

3. Medium: The new prestate tracer is built from pre/post cache diffs in crates/vm/backends/levm/tracing.rs, and the slot filtering at crates/vm/backends/levm/tracing.rs only keeps newly loaded or mutated slots. Pure reads of already-cached state disappear from the output, so SLOAD / BALANCE / EXTCODEHASH-style read sets can be missing from both tx-level and block-level prestate traces. This needs an explicit access tracker, not state diffing.

Open question: crates/vm/backends/levm/mod.rs says system calls “do not use intrinsic gas”, but is_system_call is only consumed by validate_gas_allowance; prepare_execution still unconditionally runs validate_min_gas_limit and add_intrinsic_gas. If that comment reflects the intended semantics, the flag is only half-wired.

Static review only; I didn’t run the suite.

Automated review by OpenAI Codex · gpt-5.4 · custom prompt

Greptile Summary

This PR implements EIP-7928 (Block Access List) and EIP-8037 (2D gas) for the bal-devnet-6 devnet, rolls up devnet-3→6 spec changes, adds prestateTracer support, and applies execution-apis PR 786 FCU semantics (finalized-prefix no-reorg restriction + -38006 TooDeepReorg). The implementation is complex and heavily commented, with extensive per-EIP test coverage.

• EIP-8037 (2D gas): Full implementation with clamp-and-spill reservoir model, sub-frame ExceptionalHalt reclassification, REVERT cascade propagation, CREATE top-halt formula, and SSTORE 0\u2192N\u21920 refunds via credit_state_gas_refund.
• EIP-7928 (BAL): Block-access-list index promoted from u16 to u32 throughout, shadow recorder for per-tx access validation in parallel execution, and SYSTEM_ADDRESS exclusion from BAL entries.
• Fork choice: NewHeadAlreadyCanonical now builds a payload and marks the node synced; TooDeepReorg error gated on the finalized prefix per execution-apis PR 786.

Confidence Score: 3/5

The 2D gas accounting changes are intricate; the interaction between set_delegation's auth refund and state_gas_reservoir_initial needs verification before merging.

When set_delegation credits the reservoir for existing EIP-7702 authorities, it reduces both state_gas_used and intrinsic_state_gas_charged by the refund but leaves state_gas_reservoir_initial at its pre-refund value. The refund_sender formula subtracts the stale reservoir_initial, causing the regular dimension to be overcounted by one refund-unit per existing authority, inflating block_regular_gas_used and tightening the 2D budget for subsequent transactions. The fork-choice and payload-builder changes are clean and safe.

crates/vm/levm/src/utils.rs (set_delegation refund + reservoir_initial interaction) and crates/vm/levm/src/hooks/default_hook.rs (refund_sender regular-gas formula)

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A[VM.execute] --> B[prepare_execution]
    B --> B1[add_intrinsic_gas
sets intrinsic_state_gas_charged
sets state_gas_reservoir_initial]
    B1 --> B2[validate_type_4_tx - set_delegation
adds refund to reservoir
subtracts refund from state_gas_used
NOT updating state_gas_reservoir_initial]
    B2 --> C[snapshot reservoir_at_top_message_entry]
    C --> D{is_create?}
    D -- yes --> E[handle_create_transaction]
    D -- no --> F[run_execution]
    E --> G[finalize_execution]
    F --> G
    G --> G1{Amsterdam and not success?}
    G1 -- REVERT --> G2[refill reservoir with execution_portion]
    G1 -- ExceptionalHalt --> G3[restore reservoir to entry
reclassify gross_spill to regular]
    G1 -- success --> G4[apply_same_tx_selfdestruct_refund]
    G2 --> H[refund_sender]
    G3 --> H
    G4 --> H
    H --> H1[regular_gas = raw_consumed
- intrinsic_state_charged
- reservoir_initial stale after auth-refund
- spill + reclassified]
    H1 --> I[report.gas_used = max_regular + state_gas]
    I --> J[block_regular_gas_used += tx_regular_gas
block_state_gas_used += tx_state_gas]

Fix the following 2 code review issues. Work through them one at a time, proposing concise fixes.

---

### Issue 1 of 2
crates/vm/levm/src/utils.rs:354-361
**`state_gas_reservoir_initial` not updated alongside reservoir on auth refund**

`set_delegation` adds `refund` to `state_gas_reservoir` but leaves `state_gas_reservoir_initial` (captured moments earlier in `add_intrinsic_gas`) unchanged. The `refund_sender` formula is `regular_gas = raw_consumed − intrinsic_state_charged − reservoir_initial − spill + reclassified`. Because `intrinsic_state_charged` is reduced by `refund` but `reservoir_initial` is not, the result overcounts the regular gas dimension by `refund` (= `STATE_BYTES_PER_NEW_ACCOUNT × cpsb ≈ 131,488`) per existing authority. This flows into `block_regular_gas_used` via `apply_plain_transaction`, so a block with N existing-authority type-4 txs reports `N × refund` extra regular-dimension gas, tightening the 2D budget for subsequent transactions.

### Issue 2 of 2
crates/vm/levm/src/utils.rs:339-367
**PR description contradicts final code for `set_delegation`**

The PR's "Notes for reviewers" states that commit `8045f5337` ("EIP-7702 set_delegation only credits state_gas_reservoir") reverts the `state_gas_used`/`intrinsic_state_gas_charged` subtractions and that "re-applying the subtraction is the bal-devnet-7 follow-up." However, commit `03e58d9ff` later in this same PR re-applies both subtractions. The final HEAD contains the subtractions, which directly contradicts the description reviewers are given. The description's "Notes for reviewers" section should be updated to reflect the actual final state.

_{Reviews (1): Last reviewed commit: "fix(test): drop redundant module-level i..." | Re-trigger Greptile}

Four findings inline — all minor. None blocking.

Body finding (not inline): the 74-fixture skip list in tools/ef_tests/blockchain/tests/all.rs + docs/known_issues.md is the right contract, but worth wiring a tracking issue so the skip list shrinks as snobal-devnet-7 fixtures land — easy to forget once devnet attention moves on.

The skip list is alredy gone in the bal-devnet-7 branch