Headline
GHSA-xqqc-c5gw-c5r5: Tendermint light client verification not taking into account chain ID
Impact
Anyone using the tendermint-light-client and related packages to perform light client verification (e.g. IBC-rs, Hermes).
At present, the light client does not check that the chain IDs of the trusted and untrusted headers match, resulting in a possible attack vector where someone who finds a header from an untrusted chain that satisfies all other verification conditions (e.g. enough overlapping validator signatures) could fool a light client.
The attack vector is currently theoretical, and no proof-of-concept exists yet to exploit it on live networks.
Patches
Users of the light client-related crates can currently upgrade to v0.28.0.
Workarounds
None
References
Impact
Anyone using the tendermint-light-client and related packages to perform light client verification (e.g. IBC-rs, Hermes).
At present, the light client does not check that the chain IDs of the trusted and untrusted headers match, resulting in a possible attack vector where someone who finds a header from an untrusted chain that satisfies all other verification conditions (e.g. enough overlapping validator signatures) could fool a light client.
The attack vector is currently theoretical, and no proof-of-concept exists yet to exploit it on live networks.
Patches
Users of the light client-related crates can currently upgrade to v0.28.0.
Workarounds
None
References
- Light Client specification
References
- GHSA-xqqc-c5gw-c5r5
- informalsystems/tendermint-rs@5c32f31
Related news
Tendermint is a high-performance blockchain consensus engine for Byzantine fault tolerant applications. Versions prior to 0.28.0 contain a potential attack via Improper Verification of Cryptographic Signature, affecting anyone using the tendermint-light-client and related packages to perform light client verification (e.g. IBC-rs, Hermes). The light client does not check that the chain IDs of the trusted and untrusted headers match, resulting in a possible attack vector where someone who finds a header from an untrusted chain that satisfies all other verification conditions (e.g. enough overlapping validator signatures) could fool a light client. The attack vector is currently theoretical, and no proof-of-concept exists yet to exploit it on live networks. This issue is patched in version 0.28.0. There are no workarounds.