Impact
An attacker cooperating with a malicious homeserver can construct messages that legitimately appear to have come from another person, without any indication such as a grey shield.
Additionally, a sophisticated attacker cooperating with a malicious homeserver could employ this vulnerability to perform a targeted attack in order to send fake to-device messages appearing to originate from another user. This can allow, for example, to inject the key backup secret during a self-verification, to make a targeted device start using a malicious key backup spoofed by the homeserver.
These attacks are possible due to a protocol confusion vulnerability that accepts to-device messages encrypted with Megolm instead of Olm.
Patches
matrix-ios-sdk has been modified to only accept Olm-encrypted to-device messages.
Out of caution, several other checks have been audited or added:
- Cleartext
m.room_key, m.forwarded_room_key and m.secret.send to_device messages are discarded.
- Secrets received from untrusted devices are discarded.
- Key backups are only usable if they have a valid signature from a trusted device (no more local trust, or trust-on-decrypt).
- The origin of a to-device message should only be determined by observing the Olm session which managed to decrypt the message, and not by using claimed sender_key, user_id, or any other fields controllable by the homeserver.
Workarounds
As this attack requires coordination between a malicious home server and an attacker, if you trust your home server no particular workaround is needed. Notice that the backup spoofing attack is a particularly sophisticated targeted attack.
We are not aware of this attack being used in the wild, though specifying a false positive-free way of checking whether your key backup is malicious is challenging.
As an abundance of caution, to avoid malicious backup attacks, you should not verify your new logins using emoji/QR verifications methods until patched. Prefer using verify with passphrase.
References
Blog post: https://matrix.org/blog/2022/09/28/upgrade-now-to-address-encryption-vulns-in-matrix-sdks-and-clients
For more information
If you have any questions or comments about this advisory, e-mail us at security@matrix.org.
Impact
An attacker cooperating with a malicious homeserver can construct messages that legitimately appear to have come from another person, without any indication such as a grey shield.
Additionally, a sophisticated attacker cooperating with a malicious homeserver could employ this vulnerability to perform a targeted attack in order to send fake to-device messages appearing to originate from another user. This can allow, for example, to inject the key backup secret during a self-verification, to make a targeted device start using a malicious key backup spoofed by the homeserver.
These attacks are possible due to a protocol confusion vulnerability that accepts to-device messages encrypted with Megolm instead of Olm.
Patches
matrix-ios-sdk has been modified to only accept Olm-encrypted to-device messages.
Out of caution, several other checks have been audited or added:
m.room_key,m.forwarded_room_keyandm.secret.sendto_device messages are discarded.Workarounds
As this attack requires coordination between a malicious home server and an attacker, if you trust your home server no particular workaround is needed. Notice that the backup spoofing attack is a particularly sophisticated targeted attack.
We are not aware of this attack being used in the wild, though specifying a false positive-free way of checking whether your key backup is malicious is challenging.
As an abundance of caution, to avoid malicious backup attacks, you should not verify your new logins using emoji/QR verifications methods until patched. Prefer using verify with passphrase.
References
Blog post: https://matrix.org/blog/2022/09/28/upgrade-now-to-address-encryption-vulns-in-matrix-sdks-and-clients
For more information
If you have any questions or comments about this advisory, e-mail us at security@matrix.org.