Open diracdeltas opened 6 years ago
discussed in meeting;
WRT tor, if this is not provided by site isolation then we should probably disable TLS ticket caching entirely
Here are the notes I took after reading the suggested mitigations in that paper:
thanks @fmarier
my current thinking is this would be ideal:
actually according to https://bugs.chromium.org/p/chromium/issues/detail?id=30877, chromium OTR sessions (aka incognito windows) should have session resumption disabled already. if that's the case, Tor mode is protected already
Is Tor actually an OTR session though — do guest windows have the same treatment?
The question about guest sessions will become moot after #2307.
It is unclear to me whether TLS session resumption is actually disabled altogether in off-the-record profiles, or whether it is merely partitioned between normal and off-the-record profiles; this part needs investigation.
That said, upstream is preparing to isolate TLS session caches by network isolation key, which will be the first party, rendering this entirely moot: http://crbug.com/974910
@diracdeltas @darkdh would this be fixed now that Tor uses OTR profile? https://github.com/brave/brave-core/pull/7069
since https://arxiv.org/abs/1810.07304 came out, many people are concerned about the possibility of advertisers tracking people using TLS session resumption data. brave should consider invalidating TLS session resumption tickets periodically to mitigate this.