Impact

IRRd did not always filter password hashes in query responses relating to mntner objects and database exports. This may have allowed adversaries to retrieve some of these hashes, perform a brute-force search for the clear-text passphrase, and use these to make unauthorised changes to affected IRR objects.

This issue only affected instances that process password hashes, which means it is limited to IRRd instances that serve authoritative databases. IRRd instances operating solely as mirrors of other IRR databases are not affected.

The issue occurred:

• For mntner objects where all password hash names (MD5-PW and CRYPT-PW) were in lower or mixed case in the auth attribute. For these objects, hashes remained in the output of all queries of any method and all database exports made with the export_destination setting. Fortunately, objects in the common public IRR database virtually all use uppercase hash names which means very few of those objects were affected.
• For any GraphQL queries that queried the auth field on mntner objects.
• For any GraphQL queries that queried the objectText field on the journal field on mntner objects, if the nrtm_access_list setting permitted journal access.

The two GraphQL cases are visible in logs, allowing users to determine whether any existing objects had their hashes exposed.

Patches

This has been fixed in IRRd 4.2.3 and the main branch.
Versions in the 4.1.x series never were affected.

Workarounds

Users of the 4.2.x series are strongly recommended to upgrade. All users running a more recent version from the main branch should update to the latest version. Alternatively, but not recommended, apply the patch manually for 4.2.x (tested on 4.2.2) or main.

References

For further detail, see the release notes for 4.2.3.

For more information

If you have any questions or comments about this advisory:

• Open an issue or discussion in the GitHub repository
• Email us - especially if you think you may have found a further security issue