Closed
Bug 1101528
Opened 11 years ago
Closed 10 years ago
Firefox uses the same TLS session ticket and/or ID between normal and private browsing
Categories
(Core :: Security: PSM, defect)
Tracking
()
RESOLVED
WORKSFORME
People
(Reporter: freddy, Unassigned)
Details
(Keywords: privacy)
Attachments
(1 obsolete file)
In private browsing mode, Firefox keeps using TLS session tickets established previously in non-private browsing mode.
| Reporter | ||
Comment 1•11 years ago
|
||
The Tor browser bundle deals with this by disabling caching session IDs altogether, setting noCache in the sslOptions to TRUE: https://lists.torproject.org/pipermail/tor-commits/2013-April/054489.html
Comment 2•11 years ago
|
||
Summary: Firefox uses the same TLS session ticket between normal and private browsing → Firefox uses the same TLS session ticket and/or ID between normal and private browsing
Comment 3•11 years ago
|
||
NSS has no concept of Private Browsing and it's going to be hard to teach it.
Status: UNCONFIRMED → NEW
Ever confirmed: true
Comment 4•11 years ago
|
||
Just found out that bug 967977 introduced a pref to disable the use of session identifiers.
Updated•11 years ago
|
Component: Private Browsing → Security: PSM
Product: Firefox → Core
Updated•11 years ago
|
Attachment #8535782 -
Flags: review?(nical.bugzilla)
Updated•11 years ago
|
Attachment #8535782 -
Attachment is obsolete: true
Attachment #8535782 -
Flags: review?(nical.bugzilla)
Comment 7•11 years ago
|
||
(In reply to Daniel Veditz [:dveditz] from comment #3)
> NSS has no concept of Private Browsing and it's going to be hard to teach it.
NSS does have a way of partitioning the session cache and as far as I remember Firefox uses that. So, this may be just a straight-up bug in NSS (looking in the default session cache for session tickets) or in Firefox (not using the cache partitioning mechanism correctly.)
How did you test/verify this?
Flags: needinfo?(fbraun)
| Reporter | ||
Comment 8•11 years ago
|
||
I did not test this and took the information provided on the tor mailing list for granted.
It would make sense to come up with an automated test, maybe Mike Perry (who authored the reference from comment 1) has one?
(Link in comment 3 has gone stale. See https://gitweb.torproject.org/torbrowser.git/tree/src/current-patches/firefox/0007-Disable-SSL-Session-ID-tracking.patch instead)
Flags: needinfo?(fbraun) → needinfo?(mikeperry)
Comment 9•11 years ago
|
||
In Tor browser, we completely disable TLS session tickets and SSL session identifiers via security.enable_tls_session_tickets and security.ssl.disable_session_identifiers (see also Bug 967977).
We did this primarily to avoid third party tracking/linkability via TLS, as we were not aware of the NSS partitioning features Brian mentioned in comment 7. We may revisit this if these features are easy to make use of.
I don't recall if we explicitly tested private browsing versus non-private browsing cross-over of TLS tickets or SSL session ids. We definitely don't have a test for this case, anyway.
Comment 10•11 years ago
|
||
This is the PSM code that I believe partitions the session cache with the PB flag
https://mxr.mozilla.org/mozilla-central/source/security/manager/ssl/src/nsNSSIOLayer.cpp#2587
so someone should actually verify that this happens in firefox
Status: NEW → UNCONFIRMED
Ever confirmed: false
| Reporter | ||
Updated•11 years ago
|
Flags: needinfo?(mikeperry)
Comment 11•10 years ago
|
||
Firefox/NSS is doing the right thing here.
Status: UNCONFIRMED → RESOLVED
Closed: 10 years ago
Resolution: --- → WORKSFORME
You need to log in
before you can comment on or make changes to this bug.
Description
•