GHSA-g5xx-c4hv-9ccc json
CometBFT's state syncing validator from malicious node may lead to a chain split
published 2y ago modified last yr. also GO-2024-3112
- cvss
- 4.7
- version 3.1, GitHub
- cisa kev
- Not listed
- not in CISA's catalog
- epss
- —
- no score, EPSS covers CVEs only
- public exploits
- 0
- none known
Summary
**Name**: ASA-2024-009: State syncing validator from malicious node may lead to a chain split **Component**: CometBFT **Criticality**: Medium ([ACMv1.2](https://github.com/interchainio/security/blob/main/resources/CLASSIFICATION_MATRIX.md): I:Moderate; L: Possible) **Affected versions**: >= 0.34.0, <= 0.34.33, >=0.37.0, <= 0.37.10, >= 0.38.0, <= 0.38.11
### Summary
The state sync protocol retrieves a snapshot of the application and installs it in a fresh node. In order for this node to be ready to run consensus and block sync from the installed snapshot height, we also need to install a valid `State` in the node, which is the starting state from which it is able to validate new blocks and append them to the blockchain.
The `State` object used by state sync is computed using the light client protocol, which retrieves information about committed blocks from at least two RPC endpoints. The light client protocol performs several state validations and, in particular, compares the state provided by different RPC endpoints, looking for inconsistencies.
The `State` object contains, among other fields, a `Validators` field which stores the current validator set. A validator set is a list of validator addresses, public keys and associated voting powers, one per validator. It also stores, for historical reasons, the state of the proposer selection algorithm, in the form of the `ProposerPriority` field associated with each `Validator`.
While the light client is able to validate the `ValidatorSet` retrieved from RPC endpoints, this validation does not include the `ProposerPriority` field associated with each `Validator`. As a result, when state sync adopts RPC endpoints that, for unknown reasons, provide an invalid state of the proposer selection algorithm, the node will not be able to properly run the consensus protocol, as their local view of which validator is the proposer of a given round and height will disagree with the views of the correct validators. If an increasing number of validators state sync using RPC endpoints with invalid states, the network eventually halts.
### Patches
Release versions 0.34.34, 0.37.11, and 0.38.12 include a patch to address this issue.
In the patched versions, the light client protocol compares the `ProposerPriority` fields of the `ValidatorSet` instances retrieved from the RPC endpoints configured for state sync. If they differ, the computed `State` object is considered invalid and state sync will fail with an error.
### Workarounds
The issue is observed when validators run state sync using RPC nodes that are malicious or report invalid states for the proposer selection algorithm.
It is worth noting that non-malicious nodes running upstream software should never report an invalid state for the proposer selection algorithm. This situation may result from the adoption of nodes with customized code or which had their state, stored in local databases, manually updated.
When the network public's RPC endpoints have an invalid state for the proposer election algorithm, there, new validators should refrain from using state sync for bootstrapping or be sure that they configure for state sync RPC endpoints with a valid state of the proposer election algorithm.
A validator with an invalid state for the proposer selection algorithm will reject most of the proposed blocks and will have the network rejecting blocks it has proposed. It is also possible to manually compare the state of the proposer election algorithm of nodes by comparing the outputs of the `/validators?height=_` RPC endpoints. The outputs must fully match, including the ProposerPriority field associated with each validator.
### References
* [State Sync documentation](https://docs.cometbft.com/v0.38/core/state-sync)
This issue was reported to the Cosmos Bug Bounty Program on HackerOne on 12/08/24. If you believe you have found a bug in the Interchain Stack or would like to contribute to the program by reporting a bug, please see [https://hackerone.com/cosmos](https://hackerone.com/cosmos).
If you have questions about Interchain security efforts, please reach out to our official communication channel at [security@interchain.io](mailto:security@interchain.io).
For more information about CometBFT, please see [https://docs.cometbft.com/](https://docs.cometbft.com/).
For more information about the Interchain Foundation’s engagement with Amulet, please see [https://github.com/interchainio/security](https://github.com/interchainio/security).
Patch / fix
- go github.com/cometbft/cometbft/light 0.34.34, 0.37.11, 0.38.12
- go github.com/cometbft/cometbft 0.37.11, 0.38.12
Is GHSA-g5xx-c4hv-9ccc in your dependencies? Fencer shows which of your repositories are affected as soon as an advisory lands, and opens the fix as a pull request.
Check with FencerCVSS scores
| assigned by | version | score | vector |
|---|---|---|---|
| GitHub | 3.1 | 4.7 | CVSS:3.1/AV:N/AC:H/PR:N/UI:R/S:C/C:N/I:L/A:L |
Affected packages
| ecosystem | package | affected | fixed in |
|---|---|---|---|
| go | github.com/cometbft/cometbft/light | >=0.34.0, <0.34.34 >=0.37.0, <0.37.11 >=0.38.0, <0.38.12 |
0.34.34 0.37.11 0.38.12 |
| go | github.com/cometbft/cometbft | >=0.37.0, <0.37.11 >=0.38.0, <0.38.12 |
0.37.11 0.38.12 |
References
-
github.com
/cometbft/cometbft/commit/3937e00a339ee6b861d75997b4f6c87d867b74f2
web, patch
-
github.com
/cometbft/cometbft/commit/52c00a537f8f56ed94b4a5c8af6e3fecff468b55
web, patch
-
github.com
/cometbft/cometbft/security/advisories/GHSA-g5xx-c4hv-9ccc
web, advisory
-
github.com
/cometbft/cometbft
package
-
pkg.go.dev
/vuln/GO-2024-3112
web
Sources
- GitHub Advisory GHSA-g5xx-c4hv-9ccc
- OSV GO-2024-3112
Fencer Intel keeps one entry per vulnerability: the CVE id wins, then GHSA, and records that alias each other are merged, except that two CVEs always keep separate entries. The same data is available as JSON.