TP240PhoneHome amplification works by starting a stream rather than by returning one oversized reply. It is a reflection DDoS attack that abuses a test facility left reachable in a business phone system, tracked as CVE-2022-26143 and exploited from February 2022, and a DDoS test meets it from two directions: whether your edge holds against a long, steady stream from very few sources, and whether any system of yours could be the one sending it.
How TP240PhoneHome amplification works
The reflection step is the familiar one: a request with a forged source address makes a third party send its answer to the victim. The amplification attack page covers that shared mechanism, and nothing about it changes here.
What changes is the shape of the answer. The abused feature is a diagnostic routine in the phone system's driver, meant for checking a connection by sending test traffic to the requester. It accepted the command from the internet without authentication, and once started, it kept sending.
So the attacker's cost is a single request, while the reflector's output is a sustained stream. The reported ratio, about 4.3 billion to one, was the highest publicly reported at the time. It compares the whole stream a single request set off with that request, not one reply with one query, which is why it sits orders of magnitude above the per-packet figures quoted for DNS or memcached on the amplification factor page.
The flaw carries a CVSS score of 9.8.
Why it matters under attack
The flood keeps going after the attacker stops. With most reflection vectors, the attacker must keep sending requests for the replies to keep arriving. Here, one request buys a long run of traffic. Traceback toward the attacker finds almost nothing to follow, and an attack can outlast the attacker's own interest in it.
Few reflectors, all of them identifiable. A handful of exposed systems can do what a large pool of resolvers does for DNS reflection. The source list reaching the victim is short and stable, which makes it a strong case for per-source rules rather than broad rate limits.
The reflector is a business system someone depends on. A phone system that is busy streaming test traffic is spending its uplink on someone else's attack. Its owner has their own availability problem, and may not notice until calls degrade or the transit bill arrives. A reflector of this kind is not an abandoned device; it is usually in production.
Hardening, and what a DDoS test measures
Two owners, two sets of controls:
- Apply the vendor patch for CVE-2022-26143 on any affected phone system.
- Block the test facility at the perimeter with an access list, so it is reachable only from management networks, if at all.
- On the receiving side, push a drop rule upstream. Because the sources are few and the stream is steady, a BGP Flowspec rule matching it can be precise without catching real users.
For the defender, the test reproduces the shape that matters: a long, even UDP stream from a handful of the tester's own real addresses, generated on the tester's own infrastructure and aimed at the in-scope edge, with no third-party appliance involved. The useful readings are how long the stream reaches the origin before mitigation engages, whether the rule that stops it is per-source or blunt, and whether that rule is lifted cleanly once the stream ends rather than left in place for weeks.
Then look at your own estate, with an authorised scan of your own address ranges for any voice or collaboration appliance whose management or diagnostic services answer from the internet. How a long-running reflected stream fits next to the bursty vectors a test usually combines is laid out in understanding DDoS attack vectors.