[Author Prev][Author Next][Thread Prev][Thread Next][Author Index][Thread Index]
[tor-bugs] #8751 [TorBrowserButton]: do something about TLS HELLO gmt_unix_time
#8751: do something about TLS HELLO gmt_unix_time
------------------------------+---------------------------------------------
Reporter: proper | Owner: mikeperry
Type: defect | Status: new
Priority: major | Milestone:
Component: TorBrowserButton | Version:
Keywords: | Parent: #3059
Points: | Actualpoints:
------------------------------+---------------------------------------------
= Assumptions =
When not using Tor...
- For example, when using wget or Firefox, the user's ISP and destination
server can watch TLS hello and thus learn about the client's clock.
- Many updaters in background are also using TLS.
When using Tor...
- For example, when using TBB, Tor exit nodes, the ISP's of Tor exit nodes
and destination servers can see client's clock.
These are the assumptions. [3] Please tell me if they are wrong.
= Problem =
NTP server admins can willingly or if their server gets compromised and
any man-in-the-middle can tamper with NTP replies and therefore introduce
a unique clock skew.
Almost no one is using authenticated NTP, because there are no
instructions in a forum or blog how to enable NTP authentication.
Therefore almost everyone uses standard configuration and is at risk.
Also due to a clock defect, low battery, clock can skew without tampering
with NTP.
Since the browser [1] transmits it in TLS HELLO (gmt_unix_time), it can be
used to track individual users. For example, a clock skew of +/-30 minutes
may not worry the user ("That damn clock is wrong again. I use my watch
instead.") but could identify the user even when using Tor.
Also adversaries who didn't introduce the clock skew could use it to
identify users. If the user visits a website under adversary control 2
without Tor for some non-anonymous activity, it knows the clock skew.
Later, if the user visits another website under adversary control, it can
see the same clock skew, which is at least a strong anonymity set
reduction.
= Solution =
RFC 5245 says.
> Clocks are not required to be set correctly by the basic TLS protocol;
So perhaps get ride of it entirely (replace it with some fixed time)?
> higher-level or application protocols may define additional
requirements.
Whatever that means.
= Implementation =
I have no idea.
= Related =
#3059
= Footnotes =
,,
[1] Also #1517 "Provide JS with reduced time precision" wouldn't help
much, since it wouldn't do something about bigger clock skews.
[2] Nowadays with services like google analytics and facebook like button,
there are servers which are present on a high percentage of all websites.
[3] Haven't used wireshark, but read http://www.moserware.com/2009/06
/first-few-milliseconds-of-https.html and http://wiki.wireshark.org/SSL.
--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/8751>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online
_______________________________________________
tor-bugs mailing list
tor-bugs@xxxxxxxxxxxxxxxxxxxx
https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-bugs