Bug 1209572 (CVE-2015-1853)

Miroslav Lichvar of Red Hat reports:

An attacker knowing that NTP hosts A and B are peering with each other (symmetric association) can send a packet to host A with source address of B which will set the NTP state variables on A to the values sent by the attacker. Host A will then send on its next poll to B a packet with originate timestamp that doesn't match the transmit timestamp of B and the packet will be dropped. If the attacker does this periodically for both hosts, they won't be able to synchronize to each other. This is a known denial-of-service attack, described at [1].

According to the document the NTP authentication is supposed to protect symmetric associations against this attack, but that doesn't seem to be the case. The state variables are updated even when authentication fails and the peers are sending packets with originate timestamps that don't match the transmit timestamps on the receiving side.

This seems to be a very old problem, it's in the oldest ntp version I could find (xntp3.3wy). It's also in the NTPv3 (RFC 1305) and NTPv4 (RFC 5905) specifications, so other NTP implementations with support for symmetric associations and authentication may be vulnerable too.

For authentication using symmetric keys it should be sufficient to move the code updating the state variables after the MAC is validated (TEST5). For autokey that's not enough, it seems the attacker can still reset the protocol with crypto-NAK or CRYPTO_ASSOC messages and possibly other ways. I'm not sure if this can be fixed for autokey in general.

[1] https://www.eecis.udel.edu/~mills/onwire.html