Hi, all. Here's something I wrote about removing a fingerprinting opportunity from the current TLS protocol. I was going to just send it to the tls list, but Ben Laurie suggested that I send it to the perpass list as well.
I have little standards-body experience; if this ought to turn into an RFC or something, I'd be happy to write a draft if need be. (Tor currently plans to implement this by patching OpenSSL; any feedback or advice would be appreciated. The first three approaches below are implemented in OpenSSL branches named "no_client_timestamp", "no_client_timestamp_v2" , and "no_client_timestamp_v3" in the git repository at https://github.com/nmathewson/openssl . We have long shipped our privacy-enhanced browser with a feature like this enabled through an NSS patch.) ==== SYNOPSIS: In TLS as currently specified, clients and servers each send their current view of "the time" in the clear as part of their handshake. This provides a fingerprint that can track users in spite of other attempts to make them untrackable. I propose several ways to stop doing sending this cleartext timestamp; some trivial and some more complex. ACKNOWLEDGEMENTS: Thanks to Adam Langley, Ben Laurie, and everybody else who gave me feedback here. PROBLEM: The gmt_unix_time field in the Random field in the TLS handshake provides a way for an observer to fingerprint clients. Despite the late date, much of the world is still not synchronized to the second via an ntp-like service. This means that different clients have different views of the current time, which provides a fingerprint that helps to track and distinguish them. This fingerprint is useful for tracking clients as they move around. It can also distinguish clients using a single VPN, NAT, or privacy network. (Tor's modified firefox avoids this by not sending the time.) Worse, some implementations don't send the current time, but the process time, or the computer's uptime, both of which are far more distinguishing than the current time() value. The information fingerprint here is strong enough to uniquely identify some TLS users (the ones whose clocks are hours off). Even for the ones whose clocks are mostly right (within a second or two), the field leaks a bit of information, and it only takes so many bits to make a user unique. Of course, I realize that the gmt_unix_time fingerprint is not the only way to track a typical client as it moves from network to network, and not the only way to distinguish different clients behind a NAT. BACKGROUND: WHY gmt_unix_time IN THE FIRST PLACE? According to third-hand reports (and correct me if I'm wrong!) gmt_unix_time was introduced in SSL 3.0 to prevent complete failure in cases where the PRNG was completely broken, by making a part of the Random field that would definitely vary between TLS handshakes. I doubt that this goal is really achieved, for two reasons: * On modern desktop environments, it's not really so strange to start two or more TLS connections within the same second. When this occurs, the gmt_unix_time fails in its intended purpose of preventing repeated Random values. * Nonwithstanding the gmt_unix_time workaround, TLS in general does not withstand broken PRNGs. For one example, if an implementation's PRNG is prone to repeating ClientRandom values, it is probably prone to repeating those some values as ephemeral Diffie-Hellman private values. Nevertheless, if the goal of providing unique-looking output WHY ELSE IS gmt_unix_time USED? The consensus among implementors I've asked seems to be that it's unwise to depend on any particular value or interpretation for the field. The TLS 1.2 standard, RFC 5246, says that "Clocks are not required to be set correctly by the basic TLS protocol; higher-level or application protocols may define additional requirements." Some implementations set the entire gmt_unix_time field randomly; this appears not to have broken TLS on the internet. At least one tool (tlsdate) uses the server-side value of the field as an authenticated view of the current time. PROPOSAL 1: Declare that implementations MAY replace gmt_unix_time either with four more random bytes, or four bytes of zeroes. Rationale: * Some implementations (like TorBrowser) are already doing this in practice. * It's sensible and simple. Implementors are quite unlikely to mess it up. PROPOSAL 2: Suppose that we do not accept the argument about the pointlessness of gmt_unit_time that I make above, and we want to preserve the security it allegedly provides. In that case, we can allow the following approach instead: Set the Random field, not to 32 bytes from your PRNG, but to the HMAC-SHA256 of any high resolution timer that you have, using 32 bytes from your PRNG as a key. In other words, replace this: Random.gmt_unix_time = time(); Random.random_bytes = get_random_bytes(28) with this: now = hires_time(); // clock_gettime(), or concatenate time() // with a CPU timer, or process // uptime, or whatever. key = get_random_bytes(32); Random = hmac_sha256(key, now); This approach is better than the status quo on the following counts: * It doesn't leak your view of the current time, assuming that your PRNG isn't busted. * It actually fixes the problem that gmt_unix_time purported to fix, by using a high-resolution time that's much less likely to be used twice. Even if the PRNG is broken, the value is still nonrepeating. I do not think this is not worse than the status quo: * It is unpredictable from an attacker's POV, assuming that the PRNG works. (Because an HMAC, even of known data, with an unknown random key is supposed to look random). PROPOSAL 3 As a hybrid alternative, one could set part of the ClientRandom field (perhaps 12 bytes or so) based on the HMAC of the time, and the rest of the ClientRandom field based on the PRNG: N = 12 now = hires_time(); key = get_random_bytes(32); mac = hmac_sha256(key, now)[0:N]; // First N bytes. rand = get_random_bytes(32 - N) Random = mac ++ rand // concatenate. But this seems to be getting pointlessly baroque. CONSIDERATIONS: I'd personally suggest proposal 1 (just set the field at random) for most users. Yes, it makes things a little worse if your PRNG can generate repeat values... but nearly everything in cryptography fails if your PRNG is broken. We might want to apply this fix on clients only. With a few exceptions (like hidden services) the server's view of the current time is not sensitive. Implementors might want to make this feature optional and on-by-default, just in case some higher-level application protocol really does depend on it. _______________________________________________ perpass mailing list [email protected] https://www.ietf.org/mailman/listinfo/perpass
