storage: a short delta frame or an undecodable block header still panics a state rebuild

Open
#816 0 comments 0 reactions 0 assignees View on GitHub

Maintainers usually reply within 6 days

Nobody has claimed this yet.

Assessment

Difficulty
4/5
Estimated time
3-5 days
Newbie friendliness
48/100
Issue type
Bug
Clarity
Clearly specified
Activity status
Active
Tech stack
rust
Domain
backend, databases

Research direction

Start with beacon_state_delta.rs at unframe and state_writer.rs at reconstruct_beacon_state_bytes and fold_beacon_state_deltas; then trace block slot decoding through store.rs at try_block_slot_in and peek_beacon_block_slot. Check the additional BlockHeader::from_ssz_bytes(..).expect("valid header") reads in store.rs to decide the fix's scope. Done means truncated StateDiffs and BlockHeaders rows read through get_state return errors instead of panicking, with tests covering both cases.

Written by the indexing model from the issue text.

Description

lambdaclass/ethlambda#788 changed get_state's rebuild to return Error::CorruptState, rather than panic, when a stored state, delta or diff does not decode. A few decodes of disk-sourced bytes on the same path still panic. Line numbers are at #788's head (4fe2f243):

  • Delta frame header. beacon_state_delta::unframe (crates/storage/src/beacon_state_delta.rs:111) slices bytes[0..32], [32..40] and [40..48] without checking the length. A StateDiffs row shorter than its 48-byte header therefore panics instead of reading back as CorruptState. On the rebuild path it is called by reconstruct_beacon_state_bytes (crates/storage/src/state_writer.rs:548) and fold_beacon_state_deltas (state_writer.rs:601). The writer's nearest_snapshot_root (state_writer.rs:580) calls it too.
  • Block slot lookup. Every hop of a rebuild reads its slot through try_block_slot_in (crates/storage/src/store.rs:1121). Its beacon arm, peek_beacon_block_slot (store.rs:1073), indexes the row by its SSZ offset with no bounds check. Its lean arm .expect("valid header")s (store.rs:1132). reconstruct_state's final header read (state_writer.rs:679) does the same.

This was out of #788's scope because these panics already exist on main and that PR's docs name them. The header decode is also not specific to the rebuild:

  • block_slot, beacon_block_slot and nearest_snapshot_root share it.
  • Four more BlockHeader::from_ssz_bytes(..).expect("valid header") reads sit elsewhere in store.rs (2702, 3026, 3074, 3138).

Making only the rebuild tolerate a corrupt header row would leave the next header read panicking. #306 is related but covers backend Results swallowed by .expect(), not bytes that read back fine and then fail to decode.

A fix:

  • Add a try_unframe (split_at_checked(48)) whose error the rebuild maps to CorruptState, and keep unframe as the panicking wrapper for frames the writer just built.
  • Add a fallible beacon slot peek and lean header decode, so that try_block_slot_in returns the crate's Error with a decode failure as CorruptState (or a new CorruptBlockHeader { root }).
  • Decide whether the other header reads follow.
  • Tests: a truncated StateDiffs row and a truncated BlockHeaders row each read back from get_state as an error, not a panic.
Dominant language
Rust
Stars
86
Forks
34
Avg merge
9d 20h
Merged PRs (30d)
77

Getting set up

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

More from lambdaclass/ethlambda

All issues in lambdaclass/ethlambda

Similar issues

More Rust issues

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.