projects/torbrowser/design/index.html.en (2b143bac2) - tor-webwml.git

index.html.en

<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>The Design and Implementation of the Tor Browser [DRAFT]</title><meta name="generator" content="DocBook XSL Stylesheets V1.78.1" /></head><body><div class="article"><div class="titlepage"><div><div><h2 class="title"><a id="design"></a>The Design and Implementation of the Tor Browser [DRAFT]</h2></div><div><div class="author"><h3 class="author"><span class="firstname">Mike</span> <span class="surname">Perry</span></h3><div class="affiliation"><div class="address"><p><code class="email">&lt;<a class="email" href="mailto:mikeperry#torproject org">mikeperry#torproject org</a>&gt;</code></p></div></div></div></div><div><div class="author"><h3 class="author"><span class="firstname">Erinn</span> <span class="surname">Clark</span></h3><div class="affiliation"><div class="address"><p><code class="email">&lt;<a class="email" href="mailto:erinn#torproject org">erinn#torproject org</a>&gt;</code></p></div></div></div></div><div><div class="author"><h3 class="author"><span class="firstname">Steven</span> <span class="surname">Murdoch</span></h3><div class="affiliation"><div class="address"><p><code class="email">&lt;<a class="email" href="mailto:sjmurdoch#torproject org">sjmurdoch#torproject org</a>&gt;</code></p></div></div></div></div><div><p class="pubdate">May 6th, 2015</p></div></div><hr /></div><div class="toc"><p><strong>Table of Contents</strong></p><dl class="toc"><dt><span class="sect1"><a href="#idp54432272">1. Introduction</a></span></dt><dd><dl><dt><span class="sect2"><a href="#components">1.1. Browser Component Overview</a></span></dt></dl></dd><dt><span class="sect1"><a href="#DesignRequirements">2. Design Requirements and Philosophy</a></span></dt><dd><dl><dt><span class="sect2"><a href="#security">2.1. Security Requirements</a></span></dt><dt><span class="sect2"><a href="#privacy">2.2. Privacy Requirements</a></span></dt><dt><span class="sect2"><a href="#philosophy">2.3. Philosophy</a></span></dt></dl></dd><dt><span class="sect1"><a href="#adversary">3. Adversary Model</a></span></dt><dd><dl><dt><span class="sect2"><a href="#adversary-goals">3.1. Adversary Goals</a></span></dt><dt><span class="sect2"><a href="#adversary-positioning">3.2. Adversary Capabilities - Positioning</a></span></dt><dt><span class="sect2"><a href="#attacks">3.3. Adversary Capabilities - Attacks</a></span></dt></dl></dd><dt><span class="sect1"><a href="#Implementation">4. Implementation</a></span></dt><dd><dl><dt><span class="sect2"><a href="#proxy-obedience">4.1. Proxy Obedience</a></span></dt><dt><span class="sect2"><a href="#state-separation">4.2. State Separation</a></span></dt><dt><span class="sect2"><a href="#disk-avoidance">4.3. Disk Avoidance</a></span></dt><dt><span class="sect2"><a href="#app-data-isolation">4.4. Application Data Isolation</a></span></dt><dt><span class="sect2"><a href="#identifier-linkability">4.5. Cross-Origin Identifier Unlinkability</a></span></dt><dt><span class="sect2"><a href="#fingerprinting-linkability">4.6. Cross-Origin Fingerprinting Unlinkability</a></span></dt><dt><span class="sect2"><a href="#new-identity">4.7. Long-Term Unlinkability via "New Identity" button</a></span></dt><dt><span class="sect2"><a href="#other-security">4.8. Other Security Measures</a></span></dt></dl></dd><dt><span class="sect1"><a href="#BuildSecurity">5. Build Security and Package Integrity</a></span></dt><dd><dl><dt><span class="sect2"><a href="#idp56215504">5.1. Achieving Binary Reproducibility</a></span></dt><dt><span class="sect2"><a href="#idp56237264">5.2. Package Signatures and Verification</a></span></dt><dt><span class="sect2"><a href="#idp56241792">5.3. Anonymous Verification</a></span></dt><dt><span class="sect2"><a href="#update-safety">5.4. Update Safety</a></span></dt></dl></dd><dt><span class="appendix"><a href="#Transparency">A. Towards Transparency in Navigation Tracking</a></span></dt><dd><dl><dt><span class="sect1"><a href="#deprecate">A.1. Deprecation Wishlist</a></span></dt><dt><span class="sect1"><a href="#idp56278768">A.2. Promising Standards</a></span></dt></dl></dd></dl></div><div class="sect1"><div class="titlepage"><div><div><h2 class="title" style="clear: both"><a id="idp54432272"></a>1. Introduction</h2></div></div></div><p>

This document describes the <a class="link" href="#adversary" title="3. Adversary Model">adversary model</a>,
<a class="link" href="#DesignRequirements" title="2. Design Requirements and Philosophy">design requirements</a>, and <a class="link" href="#Implementation" title="4. Implementation">implementation</a>  of the Tor Browser. It is current as of Tor Browser
4.5.

</p><p>

This document is also meant to serve as a set of design requirements and to
describe a reference implementation of a Private Browsing Mode that defends
against active network adversaries, in addition to the passive forensic local
adversary currently addressed by the major browsers.

</p><p>

For more practical information regarding Tor Browser development, please
consult the <a class="ulink" href="https://trac.torproject.org/projects/tor/wiki/doc/TorBrowser/Hacking" target="_top">Tor
Browser Hacking Guide</a>.

</p><div class="sect2"><div class="titlepage"><div><div><h3 class="title"><a id="components"></a>1.1. Browser Component Overview</h3></div></div></div><p>

The Tor Browser is based on <a class="ulink" href="https://www.mozilla.org/en-US/firefox/organizations/" target="_top">Mozilla's Extended
Support Release (ESR) Firefox branch</a>. We have a <a class="ulink" href="https://gitweb.torproject.org/tor-browser.git" target="_top">series of patches</a>
against this browser to enhance privacy and security. Browser behavior is
additionally augmented through the <a class="ulink" href="https://gitweb.torproject.org/torbutton.git/tree/" target="_top">Torbutton
extension</a>, though we are in the process of moving this functionality
into direct Firefox patches. We also <a class="ulink" href="https://gitweb.torproject.org/tor-browser.git/tree/browser/app/profile/000-tor-browser.js?h=tor-browser-31.6.0esr-4.5-1" target="_top">change
a number of Firefox preferences</a> from their defaults.

</p><p>
Tor process management and configuration is accomplished through the <a class="ulink" href="https://gitweb.torproject.org/tor-launcher.git" target="_top">Tor Launcher</a>
addon, which provides the initial Tor configuration splash screen and
bootstrap progress bar. Tor Launcher is also compatible with Thunderbird,
Instantbird, and XULRunner.

</p><p>

To help protect against potential Tor Exit Node eavesdroppers, we include
<a class="ulink" href="https://www.eff.org/https-everywhere" target="_top">HTTPS-Everywhere</a>. To
provide users with optional defense-in-depth against Javascript and other
potential exploit vectors, we also include <a class="ulink" href="http://noscript.net/" target="_top">NoScript</a>. We also modify <a class="ulink" href="https://gitweb.torproject.org/builders/tor-browser-bundle.git/tree/Bundle-Data/linux/Data/Browser/profile.default/preferences/extension-overrides.js" target="_top">several
extension preferences</a> from their defaults.

</p><p>

To provide censorship circumvention in areas where the public Tor network is
blocked either by IP, or by protocol fingerprint, we include several <a class="ulink" href="https://trac.torproject.org/projects/tor/wiki/doc/AChildsGardenOfPluggableTransports" target="_top">Pluggable
Transports</a> in the distribution. As of this writing, we include <a class="ulink" href="https://gitweb.torproject.org/pluggable-transports/obfs4.git" target="_top">Obfs4proxy</a>,
<a class="ulink" href="https://trac.torproject.org/projects/tor/wiki/doc/meek" target="_top">meek</a>,
<a class="ulink" href="https://fteproxy.org/" target="_top">FTE</a>, and <a class="ulink" href="https://crypto.stanford.edu/flashproxy/" target="_top">FlashProxy</a>.

</p></div></div><div class="sect1"><div class="titlepage"><div><div><h2 class="title" style="clear: both"><a id="DesignRequirements"></a>2. Design Requirements and Philosophy</h2></div></div></div><p>

The Tor Browser Design Requirements are meant to describe the properties of a
Private Browsing Mode that defends against both network and local forensic
adversaries.

</p><p>

There are two main categories of requirements: <a class="link" href="#security" title="2.1. Security Requirements">Security Requirements</a>, and <a class="link" href="#privacy" title="2.2. Privacy Requirements">Privacy Requirements</a>. Security Requirements are the
minimum properties in order for a browser to be able to support Tor and
similar privacy proxies safely. Privacy requirements are the set of properties
that cause us to prefer one browser over another.

</p><p>

While we will endorse the use of browsers that meet the security requirements,
it is primarily the privacy requirements that cause us to maintain our own
browser distribution.

</p><p>

The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL
      NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED",  "MAY", and
      "OPTIONAL" in this document are to be interpreted as described in
      <a class="ulink" href="https://www.ietf.org/rfc/rfc2119.txt" target="_top">RFC 2119</a>.

</p><div class="sect2"><div class="titlepage"><div><div><h3 class="title"><a id="security"></a>2.1. Security Requirements</h3></div></div></div><p>

The security requirements are primarily concerned with ensuring the safe use
of Tor. Violations in these properties typically result in serious risk for
the user in terms of immediate deanonymization and/or observability. With
respect to browser support, security requirements are the minimum properties
in order for Tor to support the use of a particular browser.

</p><div class="orderedlist"><ol class="orderedlist" type="1"><li class="listitem"><a class="link" href="#proxy-obedience" title="4.1. Proxy Obedience"><span class="command"><strong>Proxy
Obedience</strong></span></a><p>The browser
MUST NOT bypass Tor proxy settings for any content.</p></li><li class="listitem"><a class="link" href="#state-separation" title="4.2. State Separation"><span class="command"><strong>State
Separation</strong></span></a><p>

The browser MUST NOT provide the content window with any state from any other
browsers or any non-Tor browsing modes. This includes shared state from
independent plugins, and shared state from operating system implementations of
TLS and other support libraries.

</p></li><li class="listitem"><a class="link" href="#disk-avoidance" title="4.3. Disk Avoidance"><span class="command"><strong>Disk
Avoidance</strong></span></a><p>

The browser MUST NOT write any information that is derived from or that
reveals browsing activity to the disk, or store it in memory beyond the
duration of one browsing session, unless the user has explicitly opted to
store their browsing history information to disk.

</p></li><li class="listitem"><a class="link" href="#app-data-isolation" title="4.4. Application Data Isolation"><span class="command"><strong>Application Data
Isolation</strong></span></a><p>

The components involved in providing private browsing MUST be self-contained,
or MUST provide a mechanism for rapid, complete removal of all evidence of the
use of the mode. In other words, the browser MUST NOT write or cause the
operating system to write <span class="emphasis"><em>any information</em></span> about the use
of private browsing to disk outside of the application's control. The user
must be able to ensure that secure deletion of the software is sufficient to
remove evidence of the use of the software. All exceptions and shortcomings
due to operating system behavior MUST be wiped by an uninstaller. However, due
to permissions issues with access to swap, implementations MAY choose to leave
it out of scope, and/or leave it to the operating system/platform to implement
ephemeral-keyed encrypted swap.

</p></li></ol></div></div><div class="sect2"><div class="titlepage"><div><div><h3 class="title"><a id="privacy"></a>2.2. Privacy Requirements</h3></div></div></div><p>

The privacy requirements are primarily concerned with reducing linkability:
the ability for a user's activity on one site to be linked with their activity
on another site without their knowledge or explicit consent. With respect to
browser support, privacy requirements are the set of properties that cause us
to prefer one browser over another.

</p><p>

For the purposes of the unlinkability requirements of this section as well as
the descriptions in the <a class="link" href="#Implementation" title="4. Implementation">implementation
section</a>, a <span class="command"><strong>URL bar origin</strong></span> means at least the
second-level DNS name.  For example, for mail.google.com, the origin would be
google.com. Implementations MAY, at their option, restrict the URL bar origin
to be the entire fully qualified domain name.

</p><div class="orderedlist"><ol class="orderedlist" type="1"><li class="listitem"><a class="link" href="#identifier-linkability" title="4.5. Cross-Origin Identifier Unlinkability"><span class="command"><strong>Cross-Origin
Identifier Unlinkability</strong></span></a><p>

User activity on one URL bar origin MUST NOT be linkable to their activity in
any other URL bar origin by any third party automatically or without user
interaction or approval. This requirement specifically applies to linkability
from stored browser identifiers, authentication tokens, and shared state. The
requirement does not apply to linkable information the user manually submits
to sites, or due to information submitted during manual link traversal. This
functionality SHOULD NOT interfere with interactive, click-driven federated
login in a substantial way.

</p></li><li class="listitem"><a class="link" href="#fingerprinting-linkability" title="4.6. Cross-Origin Fingerprinting Unlinkability"><span class="command"><strong>Cross-Origin
Fingerprinting Unlinkability</strong></span></a><p>

User activity on one URL bar origin MUST NOT be linkable to their activity in
any other URL bar origin by any third party. This property specifically applies to
linkability from fingerprinting browser behavior.

</p></li><li class="listitem"><a class="link" href="#new-identity" title="4.7. Long-Term Unlinkability via &quot;New Identity&quot; button"><span class="command"><strong>Long-Term
Unlinkability</strong></span></a><p>

The browser MUST provide an obvious, easy way for the user to remove all of
its authentication tokens and browser state and obtain a fresh identity.
Additionally, the browser SHOULD clear linkable state by default automatically
upon browser restart, except at user option.

</p></li></ol></div></div><div class="sect2"><div class="titlepage"><div><div><h3 class="title"><a id="philosophy"></a>2.3. Philosophy</h3></div></div></div><p>

In addition to the above design requirements, the technology decisions about
Tor Browser are also guided by some philosophical positions about technology.

</p><div class="orderedlist"><ol class="orderedlist" type="1"><li class="listitem"><span class="command"><strong>Preserve existing user model</strong></span><p>

The existing way that the user expects to use a browser must be preserved. If
the user has to maintain a different mental model of how the sites they are
using behave depending on tab, browser state, or anything else that would not
normally be what they experience in their default browser, the user will
inevitably be confused. They will make mistakes and reduce their privacy as a
result. Worse, they may just stop using the browser, assuming it is broken.

</p><p>

User model breakage was one of the <a class="ulink" href="https://blog.torproject.org/blog/toggle-or-not-toggle-end-torbutton" target="_top">failures
of Torbutton</a>: Even if users managed to install everything properly,
the toggle model was too hard for the average user to understand, especially
in the face of accumulating tabs from multiple states crossed with the current
Tor-state of the browser.

</p></li><li class="listitem"><span class="command"><strong>Favor the implementation mechanism least likely to
break sites</strong></span><p>

In general, we try to find solutions to privacy issues that will not induce
site breakage, though this is not always possible.

</p></li><li class="listitem"><span class="command"><strong>Plugins must be restricted</strong></span><p>

Even if plugins always properly used the browser proxy settings (which none of
them do) and could not be induced to bypass them (which all of them can), the
activities of closed-source plugins are very difficult to audit and control.
They can obtain and transmit all manner of system information to websites,
often have their own identifier storage for tracking users, and also
contribute to fingerprinting.

</p><p>

Therefore, if plugins are to be enabled in private browsing modes, they must
be restricted from running automatically on every page (via click-to-play
placeholders), and/or be sandboxed to restrict the types of system calls they
can execute. If the user agent allows the user to craft an exemption to allow
a plugin to be used automatically, it must only apply to the top level URL bar
domain, and not to all sites, to reduce cross-origin fingerprinting
linkability.

</p></li><li class="listitem"><span class="command"><strong>Minimize Global Privacy Options</strong></span><p>

<a class="ulink" href="https://trac.torproject.org/projects/tor/ticket/3100" target="_top">Another
failure of Torbutton</a> was the options panel. Each option
that detectably alters browser behavior can be used as a fingerprinting tool.
Similarly, all extensions <a class="ulink" href="http://blog.chromium.org/2010/06/extensions-in-incognito.html" target="_top">should be
disabled in the mode</a> except as an opt-in basis. We should not load
system-wide and/or operating system provided addons or plugins.

</p><p>
Instead of global browser privacy options, privacy decisions should be made
<a class="ulink" href="https://wiki.mozilla.org/Privacy/Features/Site-based_data_management_UI" target="_top">per
URL bar origin</a> to eliminate the possibility of linkability
between domains. For example, when a plugin object (or a Javascript access of
window.plugins) is present in a page, the user should be given the choice of
allowing that plugin object for that URL bar origin only. The same
goes for exemptions to third party cookie policy, geolocation, and any other
privacy permissions.
     </p><p>
If the user has indicated they wish to record local history storage, these
permissions can be written to disk. Otherwise, they should remain memory-only. 
     </p></li><li class="listitem"><span class="command"><strong>No filters</strong></span><p>

Site-specific or filter-based addons such as <a class="ulink" href="https://addons.mozilla.org/en-US/firefox/addon/adblock-plus/" target="_top">AdBlock
Plus</a>, <a class="ulink" href="http://requestpolicy.com/" target="_top">Request Policy</a>,
<a class="ulink" href="http://www.ghostery.com/about" target="_top">Ghostery</a>, <a class="ulink" href="http://priv3.icsi.berkeley.edu/" target="_top">Priv3</a>, and <a class="ulink" href="http://sharemenot.cs.washington.edu/" target="_top">Sharemenot</a> are to be
avoided. We believe that these addons do not add any real privacy to a proper
<a class="link" href="#Implementation" title="4. Implementation">implementation</a> of the above <a class="link" href="#privacy" title="2.2. Privacy Requirements">privacy requirements</a>, and that development efforts
should be focused on general solutions that prevent tracking by all
third parties, rather than a list of specific URLs or hosts.
     </p><p>
Filter-based addons can also introduce strange breakage and cause usability
nightmares, and will also fail to do their job if an adversary simply
registers a new domain or creates a new URL path. Worse still, the unique
filter sets that each user creates or installs will provide a wealth of
fingerprinting targets.
      </p><p>

As a general matter, we are also generally opposed to shipping an always-on Ad
blocker with Tor Browser. We feel that this would damage our credibility in
terms of demonstrating that we are providing privacy through a sound design
alone, as well as damage the acceptance of Tor users by sites that support
themselves through advertising revenue.

</p><p>
Users are free to install these addons if they wish, but doing
so is not recommended, as it will alter the browser request fingerprint.
      </p></li><li class="listitem"><span class="command"><strong>Stay Current</strong></span><p>
We believe that if we do not stay current with the support of new web
technologies, we cannot hope to substantially influence or be involved in
their proper deployment or privacy realization. However, we will likely disable
high-risk features pending analysis, audit, and mitigation.
      </p></li></ol></div></div></div><div class="sect1"><div class="titlepage"><div><div><h2 class="title" style="clear: both"><a id="adversary"></a>3. Adversary Model</h2></div></div></div><p>

A Tor web browser adversary has a number of goals, capabilities, and attack
types that can be used to illustrate the design requirements for the
Tor Browser. Let's start with the goals.

</p><div class="sect2"><div class="titlepage"><div><div><h3 class="title"><a id="adversary-goals"></a>3.1. Adversary Goals</h3></div></div></div><div class="orderedlist"><ol class="orderedlist" type="1"><li class="listitem"><span class="command"><strong>Bypassing proxy settings</strong></span><p>The adversary's primary goal is direct compromise and bypass of 
Tor, causing the user to directly connect to an IP of the adversary's
choosing.</p></li><li class="listitem"><span class="command"><strong>Correlation of Tor vs Non-Tor Activity</strong></span><p>If direct proxy bypass is not possible, the adversary will likely
happily settle for the ability to correlate something a user did via Tor with
their non-Tor activity. This can be done with cookies, cache identifiers,
Javascript events, and even CSS. Sometimes the fact that a user uses Tor may
be enough for some authorities.</p></li><li class="listitem"><span class="command"><strong>History disclosure</strong></span><p>
The adversary may also be interested in history disclosure: the ability to
query a user's history to see if they have issued certain censored search
queries, or visited censored sites.
     </p></li><li class="listitem"><span class="command"><strong>Correlate activity across multiple sites</strong></span><p>

The primary goal of the advertising networks is to know that the user who
visited siteX.com is the same user that visited siteY.com to serve them
targeted ads. The advertising networks become our adversary insofar as they
attempt to perform this correlation without the user's explicit consent.

</p></li><li class="listitem"><span class="command"><strong>Fingerprinting/anonymity set reduction</strong></span><p>

Fingerprinting (more generally: "anonymity set reduction") is used to attempt
to gather identifying information on a particular individual without the use
of tracking identifiers. If the dissident or whistleblower's timezone is
available, and they are using a rare build of Firefox for an obscure operating
system, and they have a specific display resolution only used on one type of
laptop, this can be very useful information for tracking them down, or at
least <a class="link" href="#fingerprinting">tracking their activities</a>.

</p></li><li class="listitem"><span class="command"><strong>History records and other on-disk
information</strong></span><p>

In some cases, the adversary may opt for a heavy-handed approach, such as
seizing the computers of all Tor users in an area (especially after narrowing
the field by the above two pieces of information). History records and cache
data are the primary goals here. Secondary goals may include confirming
on-disk identifiers (such as hostname and disk-logged spoofed MAC address
history) obtained by other means.

</p></li></ol></div></div><div class="sect2"><div class="titlepage"><div><div><h3 class="title"><a id="adversary-positioning"></a>3.2. Adversary Capabilities - Positioning</h3></div></div></div><p>
The adversary can position themselves at a number of different locations in
order to execute their attacks.
    </p><div class="orderedlist"><ol class="orderedlist" type="1"><li class="listitem"><span class="command"><strong>Exit Node or Upstream Router</strong></span><p>
The adversary can run exit nodes, or alternatively, they may control routers
upstream of exit nodes. Both of these scenarios have been observed in the
wild.
     </p></li><li class="listitem"><span class="command"><strong>Ad servers and/or Malicious Websites</strong></span><p>
The adversary can also run websites, or more likely, they can contract out
ad space from a number of different ad servers and inject content that way. For
some users, the adversary may be the ad servers themselves. It is not
inconceivable that ad servers may try to subvert or reduce a user's anonymity 
through Tor for marketing purposes.
     </p></li><li class="listitem"><span class="command"><strong>Local Network/ISP/Upstream Router</strong></span><p>
The adversary can also inject malicious content at the user's upstream router
when they have Tor disabled, in an attempt to correlate their Tor and Non-Tor
activity.
     </p><p>

Additionally, at this position the adversary can block Tor, or attempt to
recognize the traffic patterns of specific web pages at the entrance to the Tor
network.

</p></li><li class="listitem"><span class="command"><strong>Physical Access</strong></span><p>
Some users face adversaries with intermittent or constant physical access.
Users in Internet cafes, for example, face such a threat. In addition, in
countries where simply using tools like Tor is illegal, users may face
confiscation of their computer equipment for excessive Tor usage or just
general suspicion.
     </p></li></ol></div></div><div class="sect2"><div class="titlepage"><div><div><h3 class="title"><a id="attacks"></a>3.3. Adversary Capabilities - Attacks</h3></div></div></div><p>

The adversary can perform the following attacks from a number of different 
positions to accomplish various aspects of their goals. It should be noted
that many of these attacks (especially those involving IP address leakage) are
often performed by accident by websites that simply have Javascript, dynamic 
CSS elements, and plugins. Others are performed by ad servers seeking to
correlate users' activity across different IP addresses, and still others are
performed by malicious agents on the Tor network and at national firewalls.

</p><div class="orderedlist"><ol class="orderedlist" type="1"><li class="listitem"><span class="command"><strong>Read and insert identifiers</strong></span><p>

The browser contains multiple facilities for storing identifiers that the
adversary creates for the purposes of tracking users. These identifiers are
most obviously cookies, but also include HTTP auth, DOM storage, cached
scripts and other elements with embedded identifiers, client certificates, and
even TLS Session IDs.

</p><p>

An adversary in a position to perform MITM content alteration can inject
document content elements to both read and inject cookies for arbitrary
domains. In fact, even many "SSL secured" websites are vulnerable to this sort of
<a class="ulink" href="http://seclists.org/bugtraq/2007/Aug/0070.html" target="_top">active
sidejacking</a>. In addition, the ad networks of course perform tracking
with cookies as well.

</p><p>

These types of attacks are attempts at subverting our <a class="link" href="#identifier-linkability" title="4.5. Cross-Origin Identifier Unlinkability">Cross-Origin Identifier Unlinkability</a> and <a class="link" href="#new-identity" title="4.7. Long-Term Unlinkability via &quot;New Identity&quot; button">Long-Term Unlinkability</a> design requirements.

</p></li><li class="listitem"><a id="fingerprinting"></a><span class="command"><strong>Fingerprint users based on browser
attributes</strong></span><p>

There is an absurd amount of information available to websites via attributes
of the browser. This information can be used to reduce anonymity set, or even
uniquely fingerprint individual users. Attacks of this nature are typically
aimed at tracking users across sites without their consent, in an attempt to
subvert our <a class="link" href="#fingerprinting-linkability" title="4.6. Cross-Origin Fingerprinting Unlinkability">Cross-Origin
Fingerprinting Unlinkability</a> and <a class="link" href="#new-identity" title="4.7. Long-Term Unlinkability via &quot;New Identity&quot; button">Long-Term Unlinkability</a> design requirements.

</p><p>

Fingerprinting is an intimidating
problem to attempt to tackle, especially without a metric to determine or at
least intuitively understand and estimate which features will most contribute
to linkability between visits.

</p><p>

The <a class="ulink" href="https://panopticlick.eff.org/about.php" target="_top">Panopticlick study
done</a> by the EFF uses the <a class="ulink" href="https://en.wikipedia.org/wiki/Entropy_%28information_theory%29" target="_top">Shannon
entropy</a> - the number of identifying bits of information encoded in
browser properties - as this metric. Their <a class="ulink" href="https://wiki.mozilla.org/Fingerprinting#Data" target="_top">result data</a> is
definitely useful, and the metric is probably the appropriate one for
determining how identifying a particular browser property is. However, some
quirks of their study means that they do not extract as much information as
they could from display information: they only use desktop resolution and do
not attempt to infer the size of toolbars. In the other direction, they may be
over-counting in some areas, as they did not compute joint entropy over
multiple attributes that may exhibit a high degree of correlation. Also, new
browser features are added regularly, so the data should not be taken as
final.

</p><p>

Despite the uncertainty, all fingerprinting attacks leverage the following
attack vectors:

</p><div class="orderedlist"><ol class="orderedlist" type="a"><li class="listitem"><span class="command"><strong>Observing Request Behavior</strong></span><p>

Properties of the user's request behavior comprise the bulk of low-hanging
fingerprinting targets. These include: User agent, Accept-* headers, pipeline
usage, and request ordering. Additionally, the use of custom filters such as
AdBlock and other privacy filters can be used to fingerprint request patterns
(as an extreme example).

</p></li><li class="listitem"><span class="command"><strong>Inserting Javascript</strong></span><p>

Javascript can reveal a lot of fingerprinting information. It provides DOM
objects such as window.screen and window.navigator to extract information
about the user agent.

Also, Javascript can be used to query the user's timezone via the
<code class="function">Date()</code> object, <a class="ulink" href="https://www.khronos.org/registry/webgl/specs/1.0/#5.13" target="_top">WebGL</a> can
reveal information about the video card in use, and high precision timing
information can be used to <a class="ulink" href="http://w2spconf.com/2011/papers/jspriv.pdf" target="_top">fingerprint the CPU and
interpreter speed</a>. In the future, new JavaScript features such as
<a class="ulink" href="http://w3c-test.org/webperf/specs/ResourceTiming/" target="_top">Resource
Timing</a> may leak an unknown amount of network timing related
information.

</p></li><li class="listitem"><span class="command"><strong>Inserting Plugins</strong></span><p>

The Panopticlick project found that the mere list of installed plugins (in
navigator.plugins) was sufficient to provide a large degree of
fingerprintability. Additionally, plugins are capable of extracting font lists,
interface addresses, and other machine information that is beyond what the
browser would normally provide to content. In addition, plugins can be used to
store unique identifiers that are more difficult to clear than standard
cookies.  <a class="ulink" href="http://epic.org/privacy/cookies/flash.html" target="_top">Flash-based
cookies</a> fall into this category, but there are likely numerous other
examples. Beyond fingerprinting, plugins are also abysmal at obeying the proxy
settings of the browser.

</p></li><li class="listitem"><span class="command"><strong>Inserting CSS</strong></span><p>

<a class="ulink" href="https://developer.mozilla.org/En/CSS/Media_queries" target="_top">CSS media
queries</a> can be inserted to gather information about the desktop size,
widget size, display type, DPI, user agent type, and other information that
was formerly available only to Javascript.

</p></li></ol></div></li><li class="listitem"><a id="website-traffic-fingerprinting"></a><span class="command"><strong>Website traffic fingerprinting</strong></span><p>

Website traffic fingerprinting is an attempt by the adversary to recognize the
encrypted traffic patterns of specific websites. In the case of Tor, this
attack would take place between the user and the Guard node, or at the Guard
node itself.
     </p><p> The most comprehensive study of the statistical properties of this
attack against Tor was done by <a class="ulink" href="http://lorre.uni.lu/~andriy/papers/acmccs-wpes11-fingerprinting.pdf" target="_top">Panchenko
et al</a>. Unfortunately, the publication bias in academia has encouraged
the production of a number of follow-on attack papers claiming "improved"
success rates, in some cases even claiming to completely invalidate any
attempt at defense. These "improvements" are actually enabled primarily by
taking a number of shortcuts (such as classifying only very small numbers of
web pages, neglecting to publish ROC curves or at least false positive rates,
and/or omitting the effects of dataset size on their results). Despite these
subsequent "improvements", we are skeptical of the efficacy of this attack in
a real world scenario, <span class="emphasis"><em>especially</em></span> in the face of any
defenses.

</p><p>

In general, with machine learning, as you increase the <a class="ulink" href="https://en.wikipedia.org/wiki/VC_dimension" target="_top">number and/or complexity of
categories to classify</a> while maintaining a limit on reliable feature
information you can extract, you eventually run out of descriptive feature
information, and either true positive accuracy goes down or the false positive
rate goes up. This error is called the <a class="ulink" href="http://www.cs.washington.edu/education/courses/csep573/98sp/lectures/lecture8/sld050.htm" target="_top">bias
in your hypothesis space</a>. In fact, even for unbiased hypothesis
spaces, the number of training examples required to achieve a reasonable error
bound is <a class="ulink" href="https://en.wikipedia.org/wiki/Probably_approximately_correct_learning#Equivalence" target="_top">a
function of the complexity of the categories</a> you need to classify.

</p><p>

In the case of this attack, the key factors that increase the classification
complexity (and thus hinder a real world adversary who attempts this attack)
are large numbers of dynamically generated pages, partially cached content,
and also the non-web activity of entire Tor network. This yields an effective
number of "web pages" many orders of magnitude larger than even <a class="ulink" href="http://lorre.uni.lu/~andriy/papers/acmccs-wpes11-fingerprinting.pdf" target="_top">Panchenko's
"Open World" scenario</a>, which suffered continuous near-constant decline
in the true positive rate as the "Open World" size grew (see figure 4). This
large level of classification complexity is further confounded by a noisy and
low resolution featureset - one which is also relatively easy for the defender
to manipulate at low cost.

</p><p>

To make matters worse for a real-world adversary, the ocean of Tor Internet
activity (at least, when compared to a lab setting) makes it a certainty that
an adversary attempting examine large amounts of Tor traffic will ultimately
be overwhelmed by false positives (even after making heavy tradeoffs on the
ROC curve to minimize false positives to below 0.01%). This problem is known
in the IDS literature as the <a class="ulink" href="http://www.raid-symposium.org/raid99/PAPERS/Axelsson.pdf" target="_top">Base Rate
Fallacy</a>, and it is the primary reason that anomaly and activity
classification-based IDS and antivirus systems have failed to materialize in
the marketplace (despite early success in academic literature).

</p><p>

Still, we do not believe that these issues are enough to dismiss the attack
outright. But we do believe these factors make it both worthwhile and
effective to <a class="link" href="#traffic-fingerprinting-defenses">deploy
light-weight defenses</a> that reduce the accuracy of this attack by
further contributing noise to hinder successful feature extraction.

</p></li><li class="listitem"><span class="command"><strong>Remotely or locally exploit browser and/or
OS</strong></span><p>

Last, but definitely not least, the adversary can exploit either general
browser vulnerabilities, plugin vulnerabilities, or OS vulnerabilities to
install malware and surveillance software. An adversary with physical access
can perform similar actions.

</p><p>

For the purposes of the browser itself, we limit the scope of this adversary
to one that has passive forensic access to the disk after browsing activity
has taken place. This adversary motivates our 
<a class="link" href="#disk-avoidance" title="4.3. Disk Avoidance">Disk Avoidance</a> defenses.

</p><p>

An adversary with arbitrary code execution typically has more power, though.
It can be quite hard to really significantly limit the capabilities of such an
adversary. <a class="ulink" href="http://tails.boum.org/contribute/design/" target="_top">The Tails system</a> can
provide some defense against this adversary through the use of readonly media
and frequent reboots, but even this can be circumvented on machines without
Secure Boot through the use of BIOS rootkits.

</p></li></ol></div></div></div><div class="sect1"><div class="titlepage"><div><div><h2 class="title" style="clear: both"><a id="Implementation"></a>4. Implementation</h2></div></div></div><p>

The Implementation section is divided into subsections, each of which
corresponds to a <a class="link" href="#DesignRequirements" title="2. Design Requirements and Philosophy">Design Requirement</a>.
Each subsection is divided into specific web technologies or properties. The
implementation is then described for that property.

</p><p>

In some cases, the implementation meets the design requirements in a non-ideal
way (for example, by disabling features). In rare cases, there may be no
implementation at all. Both of these cases are denoted by differentiating
between the <span class="command"><strong>Design Goal</strong></span> and the <span class="command"><strong>Implementation
Status</strong></span> for each property. Corresponding bugs in the <a class="ulink" href="https://trac.torproject.org/projects/tor/report" target="_top">Tor bug tracker</a>
are typically linked for these cases.

</p><div class="sect2"><div class="titlepage"><div><div><h3 class="title"><a id="proxy-obedience"></a>4.1. Proxy Obedience</h3></div></div></div><p>

Proxy obedience is assured through the following:
   </p><div class="orderedlist"><ol class="orderedlist" type="1"><li class="listitem"><span class="command"><strong>Firefox proxy settings, patches, and build flags</strong></span><p>

Our <a class="ulink" href="https://gitweb.torproject.org/tor-browser.git/tree/browser/app/profile/000-tor-browser.js?h=tor-browser-31.6.0esr-4.5-1" target="_top">Firefox
preferences file</a> sets the Firefox proxy settings to use Tor directly
as a SOCKS proxy. It sets <span class="command"><strong>network.proxy.socks_remote_dns</strong></span>,
<span class="command"><strong>network.proxy.socks_version</strong></span>,
<span class="command"><strong>network.proxy.socks_port</strong></span>, and
<span class="command"><strong>network.dns.disablePrefetch</strong></span>.

</p><p>

To prevent proxy bypass by WebRTC calls, we disable WebRTC at compile time
with the <span class="command"><strong>--disable-webrtc</strong></span> configure switch, as well
as set the pref <span class="command"><strong>media.peerconnection.enabled</strong></span> to false.

</p><p>

We also patch Firefox in order to provide several defense-in-depth mechanisms
for proxy safety. Notably, we <a class="ulink" href="https://gitweb.torproject.org/tor-browser.git/commit/?h=tor-browser-31.6.0esr-4.5-1&amp;id=8c6604d2b776f0d8e33ed9130c5f5b8cf744bac8" target="_top">patch
the DNS service</a> to prevent any browser or addon DNS resolution, and we
also <a class="ulink" href="https://gitweb.torproject.org/tor-browser.git/commit/?h=tor-browser-31.6.0esr-4.5-1&amp;id=c96c854c0eca21fed1362d1ddd164b657d351795" target="_top">patch
OCSP and PKIX code</a> to prevent any use of the non-proxied command-line
tool utility functions from being functional while linked in to the browser.
In both cases, we could find no direct paths to these routines in the browser,
but it seemed better safe than sorry.

</p><p>

During every Extended Support Release transition, we perform <a class="ulink" href="https://gitweb.torproject.org/tor-browser-spec.git/tree/audits" target="_top">in-depth
code audits</a> to verify that there were no system calls or XPCOM
activity in the source tree that did not use the browser proxy settings.
 </p><p>

We have verified that these settings and patches properly proxy HTTPS, OCSP,
HTTP, FTP, gopher (now defunct), DNS, SafeBrowsing Queries, all JavaScript
activity, including HTML5 audio and video objects, addon updates, WiFi
geolocation queries, searchbox queries, XPCOM addon HTTPS/HTTP activity,
WebSockets, and live bookmark updates. We have also verified that IPv6
connections are not attempted, through the proxy or otherwise (Tor does not
yet support IPv6). We have also verified that external protocol helpers, such
as SMB URLs and other custom protocol handlers are all blocked.

</p></li><li class="listitem"><span class="command"><strong>Disabling plugins</strong></span><p>Plugins have the ability to make arbitrary OS system calls and  <a class="ulink" href="http://decloak.net/" target="_top">bypass proxy settings</a>. This includes
the ability to make UDP sockets and send arbitrary data independent of the
browser proxy settings.
 </p><p>
Torbutton disables plugins by using the
<span class="command"><strong>@mozilla.org/plugin/host;1</strong></span> service to mark the plugin tags
as disabled. This block can be undone through both the Torbutton Security UI,
and the Firefox Plugin Preferences.
 </p><p>
If the user does enable plugins in this way, plugin-handled objects are still
restricted from automatic load through Firefox's click-to-play preference
<span class="command"><strong>plugins.click_to_play</strong></span>.
 </p><p>

In addition, to reduce any unproxied activity by arbitrary plugins at load
time, and to reduce the fingerprintability of the installed plugin list, we
also patch the Firefox source code to <a class="ulink" href="https://gitweb.torproject.org/tor-browser.git/commit/?h=tor-browser-31.6.0esr-4.5-1&amp;id=465cb8295db58a6450dc14a593d29372cbebc71d" target="_top">
prevent the load of any plugins except for Flash and Gnash</a>. Even for
Flash and Gnash, we also patch Firefox to <a class="ulink" href="https://gitweb.torproject.org/tor-browser.git/commit/?h=tor-browser-31.6.0esr-4.5-1&amp;id=e5531b1baa3c96dee7d8d4274791ff393bafd241" target="_top">prevent loading them into the
address space</a> until they are explicitly enabled.

</p></li><li class="listitem"><span class="command"><strong>External App Blocking and Drag Event Filtering</strong></span><p>

External apps can be induced to load files that perform network activity.
Unfortunately, there are cases where such apps can be launched automatically
with little to no user input. In order to prevent this, Torbutton installs a
component to <a class="ulink" href="https://gitweb.torproject.org/torbutton.git/tree/src/components/external-app-blocker.js" target="_top">
provide the user with a popup</a> whenever the browser attempts to launch
a helper app.

</p><p>

Additionally, modern desktops now pre-emptively fetch any URLs in Drag and
Drop events as soon as the drag is initiated. This download happens
independent of the browser's Tor settings, and can be triggered by something
as simple as holding the mouse button down for slightly too long while
clicking on an image link. We filter drag and drop events events <a class="ulink" href="https://gitweb.torproject.org/torbutton.git/tree/src/components/external-app-blocker.js" target="_top">from
Torbutton</a> before the OS downloads the URLs the events contained.

</p></li><li class="listitem"><span class="command"><strong>Disabling system extensions and clearing the addon whitelist</strong></span><p>

Firefox addons can perform arbitrary activity on your computer, including
bypassing Tor. It is for this reason we disable the addon whitelist
(<span class="command"><strong>xpinstall.whitelist.add</strong></span>), so that users are prompted
before installing addons regardless of the source. We also exclude
system-level addons from the browser through the use of
<span class="command"><strong>extensions.enabledScopes</strong></span> and
<span class="command"><strong>extensions.autoDisableScopes</strong></span>.

</p></li></ol></div></div><div class="sect2"><div class="titlepage"><div><div><h3 class="title"><a id="state-separation"></a>4.2. State Separation</h3></div></div></div><p>

Tor Browser State is separated from existing browser state through use of a
custom Firefox profile, and by setting the $HOME environment variable to the
root of the bundle's directory.  The browser also does not load any
system-wide extensions (through the use of
<span class="command"><strong>extensions.enabledScopes</strong></span> and
<span class="command"><strong>extensions.autoDisableScopes</strong></span>). Furthermore, plugins are
disabled, which prevents Flash cookies from leaking from a pre-existing Flash
directory.

</p></div><div class="sect2"><div class="titlepage"><div><div><h3 class="title"><a id="disk-avoidance"></a>4.3. Disk Avoidance</h3></div></div></div><div class="sect3"><div class="titlepage"><div><div><h4 class="title"><a id="idp55920416"></a>Design Goal:</h4></div></div></div><div class="blockquote"><blockquote class="blockquote">

The User Agent MUST (at user option) prevent all disk records of browser activity.
The user should be able to optionally enable URL history and other history
features if they so desire.

</blockquote></div></div><div class="sect3"><div class="titlepage"><div><div><h4 class="title"><a id="idp55921776"></a>Implementation Status:</h4></div></div></div><div class="blockquote"><blockquote class="blockquote">

We achieve this goal through several mechanisms. First, we set the Firefox
Private Browsing preference
<span class="command"><strong>browser.privatebrowsing.autostart</strong></span>. In addition, four Firefox patches are needed to prevent disk writes, even if
Private Browsing Mode is enabled. We need to

<a class="ulink" href="https://gitweb.torproject.org/tor-browser.git/commit/?h=tor-browser-31.6.0esr-4.5-1&amp;id=44b8ae43a83191bbf5161cbdbf399e10c1b943d0" target="_top">prevent
the permissions manager from recording HTTPS STS state</a>, <a class="ulink" href="https://gitweb.torproject.org/tor-browser.git/commit/?h=tor-browser-31.6.0esr-4.5-1&amp;id=e5abcb28f131aa96e8762212573488d303b3614d" target="_top">prevent
intermediate SSL certificates from being recorded</a>, <a class="ulink" href="https://gitweb.torproject.org/tor-browser.git/commit/?h=tor-browser-31.6.0esr-4.5-1&amp;id=ee34e122ac2929a7668314483e36e58a88c98c08" target="_top">prevent
the clipboard cache from being written to disk for large pastes</a>, and
<a class="ulink" href="https://gitweb.torproject.org/tor-browser.git/commit/?h=tor-browser-31.6.0esr-4.5-1&amp;id=c8e357740dd7bafa2a129007f27d2b243e36f4a2" target="_top">prevent
the content preferences service from recording site zoom</a>. We also had
to disable the media cache with the pref <span class="command"><strong>media.cache_size</strong></span>,
to prevent HTML5 videos from being written to the OS temporary directory,
which happened regardless of the private browsing mode setting.

</blockquote></div><div class="blockquote"><blockquote class="blockquote">

As an additional defense-in-depth measure, we set the following preferences:
<span class="command"><strong></strong></span>,
<span class="command"><strong>browser.cache.disk.enable</strong></span>,
<span class="command"><strong>browser.cache.offline.enable</strong></span>,
<span class="command"><strong>dom.indexedDB.enabled</strong></span>,
<span class="command"><strong>network.cookie.lifetimePolicy</strong></span>,
<span class="command"><strong>signon.rememberSignons</strong></span>,
<span class="command"><strong>browser.formfill.enable</strong></span>,
<span class="command"><strong>browser.download.manager.retention</strong></span>,
<span class="command"><strong>browser.sessionstore.privacy_level</strong></span>,
and <span class="command"><strong>network.cookie.lifetimePolicy</strong></span>. Many of these
preferences are likely redundant with
<span class="command"><strong>browser.privatebrowsing.autostart</strong></span>, but we have not done the
auditing work to ensure that yet.

</blockquote></div><div class="blockquote"><blockquote class="blockquote">

For more details on disk leak bugs and enhancements, see the <a class="ulink" href="https://trac.torproject.org/projects/tor/query?keywords=~tbb-disk-leak&amp;status=!closed" target="_top">tbb-disk-leak tag in our bugtracker</a></blockquote></div></div></div><div class="sect2"><div class="titlepage"><div><div><h3 class="title"><a id="app-data-isolation"></a>4.4. Application Data Isolation</h3></div></div></div><p>

Tor Browser Bundle MUST NOT cause any information to be written outside of the
bundle directory. This is to ensure that the user is able to completely and
safely remove the bundle without leaving other traces of Tor usage on their
computer.

</p><p>

To ensure TBB directory isolation, we set
<span class="command"><strong>browser.download.useDownloadDir</strong></span>,
<span class="command"><strong>browser.shell.checkDefaultBrowser</strong></span>, and
<span class="command"><strong>browser.download.manager.addToRecentDocs</strong></span>. We also set the
$HOME environment variable to be the TBB extraction directory.
   </p></div><div class="sect2"><div class="titlepage"><div><div><h3 class="title"><a id="identifier-linkability"></a>4.5. Cross-Origin Identifier Unlinkability</h3></div></div></div><p>

The Cross-Origin Identifier Unlinkability design requirement is satisfied
through first party isolation of all browser identifier sources. First party
isolation means that all identifier sources and browser state are scoped
(isolated) using the URL bar domain. This scoping is performed in
combination with any additional third party scope. When first party isolation
is used with explicit identifier storage that already has a constrained third
party scope (such as cookies and DOM storage), this approach is
referred to as "double-keying".

</p><p>

The benefit of this approach comes not only in the form of reduced
linkability, but also in terms of simplified privacy UI. If all stored browser
state and permissions become associated with the URL bar origin, the six or
seven different pieces of privacy UI governing these identifiers and
permissions can become just one piece of UI. For instance, a window that lists
the URL bar origin for which browser state exists, possibly with a
context-menu option to drill down into specific types of state or permissions.
An example of this simplification can be seen in Figure 1.

</p><div class="figure"><a id="idp55943472"></a><p class="title"><strong>Figure 1. Improving the Privacy UI</strong></p><div class="figure-contents"><div class="mediaobject" align="center"><img src="NewCookieManager.png" align="middle" alt="Improving the Privacy UI" /></div><div class="caption"><p></p>

This example UI is a mock-up of how isolating identifiers to the URL bar
origin can simplify the privacy UI for all data - not just cookies. Once
browser identifiers and site permissions operate on a URL bar basis, the same
privacy window can represent browsing history, DOM Storage, HTTP Auth, search
form history, login values, and so on within a context menu for each site.

</div></div></div><br class="figure-break" /><div class="sect3"><div class="titlepage"><div><div><h4 class="title"><a id="idp55946896"></a>Identifier Unlinkability Defenses in the Tor Browser</h4></div></div></div><p>

Unfortunately, many aspects of browser state can serve as identifier storage,
and no other browser vendor or standards body has invested the effort to
enumerate or otherwise deal with these vectors for third party tracking. As
such, we have had to enumerate and isolate these identifier sources on a
piecemeal basis. Here is the list that we have discovered and dealt with to
date:

</p><div class="orderedlist"><ol class="orderedlist" type="1"><li class="listitem"><span class="command"><strong>Cookies</strong></span><p><span class="command"><strong>Design Goal:</strong></span>

All cookies MUST be double-keyed to the URL bar origin and third-party
origin. There exists a <a class="ulink" href="https://bugzilla.mozilla.org/show_bug.cgi?id=565965" target="_top">Mozilla bug</a>
that contains a prototype patch, but it lacks UI, and does not apply to modern
Firefox versions.

</p><p><span class="command"><strong>Implementation Status:</strong></span>

As a stopgap to satisfy our design requirement of unlinkability, we currently
entirely disable 3rd party cookies by setting
<span class="command"><strong>network.cookie.cookieBehavior</strong></span> to 1. We would prefer that
third party content continue to function, but we believe the requirement for 
unlinkability trumps that desire.

</p></li><li class="listitem"><span class="command"><strong>Cache</strong></span><p>

In Firefox, there are actually two distinct caching mechanisms: One for
general content (HTML, Javascript, CSS), and one specifically for images. The
content cache is isolated to the URL bar domain by <a class="ulink" href="https://gitweb.torproject.org/tor-browser.git/commit/?h=tor-browser-31.6.0esr-4.5-1&amp;id=7c58be929777d386a03e1faaee648909151fd951" target="_top">altering
each cache key</a> to include an additional ID that includes the URL bar
domain. This functionality can be observed by navigating to <a class="ulink" href="about:cache" target="_top">about:cache</a> and viewing the key used for each cache
entry. Each third party element should have an additional "id=string"
property prepended, which will list the FQDN that was used to source it.

</p><p>

Additionally, because the image cache is a separate entity from the content
cache, we had to patch Firefox to also <a class="ulink" href="https://gitweb.torproject.org/tor-browser.git/commit/?h=tor-browser-31.6.0esr-4.5-1&amp;id=d8b98a75fb200268c40886d876adc19e00b933bf" target="_top">isolate
this cache per URL bar domain</a>.

</p></li><li class="listitem"><span class="command"><strong>HTTP Authentication</strong></span><p>

HTTP Authorization headers can be used to encode <a class="ulink" href="http://jeremiahgrossman.blogspot.com/2007/04/tracking-users-without-cookies.html" target="_top">silent
third party tracking identifiers</a>. To prevent this, we remove HTTP
authentication tokens for third party elements through a <a class="ulink" href="https://gitweb.torproject.org/tor-browser.git/commit/?h=tor-browser-31.6.0esr-4.5-1&amp;id=b8ce4a0760759431f146c71184c89fbd5e1a27e4" target="_top">patch
to nsHTTPChannel</a>.

</p></li><li class="listitem"><span class="command"><strong>DOM Storage</strong></span><p>

DOM storage for third party domains MUST be isolated to the URL bar origin,
to prevent linkability between sites. This functionality is provided through a
<a class="ulink" href="https://gitweb.torproject.org/tor-browser.git/commit/?h=tor-browser-31.6.0esr-4.5-1&amp;id=97490c4a90ca1c43374486d9ec0c5593d5fe5720" target="_top">patch
to Firefox</a>.

</p></li><li class="listitem"><span class="command"><strong>Flash cookies</strong></span><p><span class="command"><strong>Design Goal:</strong></span>

Users should be able to click-to-play flash objects from trusted sites. To
make this behavior unlinkable, we wish to include a settings file for all platforms that disables flash
cookies using the <a class="ulink" href="http://www.macromedia.com/support/documentation/en/flashplayer/help/settings_manager03.html" target="_top">Flash
settings manager</a>.

</p><p><span class="command"><strong>Implementation Status:</strong></span>

We are currently <a class="ulink" href="https://trac.torproject.org/projects/tor/ticket/3974" target="_top">having
difficulties</a> causing Flash player to use this settings
file on Windows, so Flash remains difficult to enable.

</p></li><li class="listitem"><span class="command"><strong>SSL+TLS session resumption</strong></span><p><span class="command"><strong>Design Goal:</strong></span>

TLS session resumption tickets and SSL Session IDs MUST be limited to the URL
bar origin.

</p><p><span class="command"><strong>Implementation Status:</strong></span>

We currently clear SSL Session IDs upon <a class="link" href="#new-identity" title="4.7. Long-Term Unlinkability via &quot;New Identity&quot; button">New
Identity</a>, we disable TLS Session Tickets via the Firefox Pref
<span class="command"><strong>security.enable_tls_session_tickets</strong></span>. We disable SSL Session
IDs via a <a class="ulink" href="https://gitweb.torproject.org/tor-browser.git/commit/?h=tor-browser-31.6.0esr-4.5-1&amp;id=a01fb747d4b8b24687de538cb6a1304fe27d9d88" target="_top">patch
to Firefox</a>. To compensate for the increased round trip latency from disabling
these performance optimizations, we also enable
<a class="ulink" href="https://tools.ietf.org/html/draft-bmoeller-tls-falsestart-00" target="_top">TLS
False Start</a> via the Firefox Pref
<span class="command"><strong>security.ssl.enable_false_start</strong></span>.
    </p></li><li class="listitem"><span class="command"><strong>Tor circuit and HTTP connection linkability</strong></span><p>

Tor circuits and HTTP connections from a third party in one URL bar origin
MUST NOT be reused for that same third party in another URL bar origin.

</p><p>

This isolation functionality is provided by the combination of a <a class="ulink" href="https://gitweb.torproject.org/tor-browser.git/commit/?h=tor-browser-31.6.0esr-4.5-1&amp;id=b3ea705cc35b79a9ba27323cb3e32d5d004ea113" target="_top">Firefox
patch to allow SOCKS usernames and passwords</a>, as well as a Torbutton
component that <a class="ulink" href="" target="_top">sets
the SOCKS username and password for each request</a>. The Tor client has
logic to prevent connections with different SOCKS usernames and passwords from
using the same Tor circuit. Firefox has existing logic to ensure that connections with
SOCKS proxies do not re-use existing HTTP Keep-Alive connections unless the
proxy settings match.  We extended this logic to cover SOCKS username and
password authentication, providing us with HTTP Keep-Alive unlinkability.

</p></li><li class="listitem"><span class="command"><strong>SharedWorkers</strong></span><p>

<a class="ulink" href="https://developer.mozilla.org/en-US/docs/Web/API/SharedWorker" target="_top">SharedWorkers</a>
are a special form of Javascript Worker Threads that have a shared scope
between all threads from the same Javascript origin.
     </p><p><span class="command"><strong>Design Goal:</strong></span>

SharedWorker scope MUST be isolated to the URL bar domain. A SharedWorker
launched from a third party from one URL bar domain MUST NOT have access to
the objects created by that same third party loaded under another URL bar domain.

</p><p><span class="command"><strong>Implementation Status:</strong></span>

For now, we disable SharedWorkers via the pref
<span class="command"><strong>dom.workers.sharedWorkers.enabled</strong></span>.

</p></li><li class="listitem"><span class="command"><strong>blob: URIs (URL.createObjectURL)</strong></span><p>

The <a class="ulink" href="https://developer.mozilla.org/en-US/docs/Web/API/URL/createObjectURL" target="_top">URL.createObjectURL</a>
API allows a site to load arbitrary content into a random UUID that is stored
in the user's browser, and this content can be accessed via a URL of the form
<span class="command"><strong>blob:UUID</strong></span> from any other content element anywhere on the
web. While this UUID value is neither under control of the site nor
predictable, it can still be used to tag a set of users that are of high
interest to an adversary.

</p><p>

URIs created with URL.createObjectURL MUST be limited in scope to the first
party URL bar domain that created them. We provide this isolation in Tor
Browser via a <a class="ulink" href="https://gitweb.torproject.org/tor-browser.git/commit/?h=tor-browser-31.6.0esr-4.5-1&amp;id=0d67ab406bdd3cf095802cb25c081641aa1f0bcc" target="_top">direct
patch to Firefox</a> and disable URL.createObjectURL in the WebWorker
context as a stopgap, due to an edge case with enforcing this isolation in
WebWorkers.

</p></li><li class="listitem"><span class="command"><strong>SPDY</strong></span><p>

Because SPDY can store identifiers, it is disabled through the
Firefox preference <span class="command"><strong>network.http.spdy.enabled</strong></span>.

</p></li><li class="listitem"><span class="command"><strong>Automated cross-origin redirects</strong></span><p><span class="command"><strong>Design Goal:</strong></span>

To prevent attacks aimed at subverting the Cross-Origin Identifier
Unlinkability <a class="link" href="#privacy" title="2.2. Privacy Requirements">privacy requirement</a>, the browser
MUST NOT store any identifiers (cookies, cache, DOM storage, HTTP auth, etc)
for cross-origin redirect intermediaries that do not prompt for user input.
For example, if a user clicks on a bit.ly URL that redirects to a
doubleclick.net URL that finally redirects to a cnn.com URL, only cookies from
cnn.com should be retained after the redirect chain completes.

</p><p>

Non-automated redirect chains that require user input at some step (such as
federated login systems) SHOULD still allow identifiers to persist.

</p><p><span class="command"><strong>Implementation status:</strong></span>

There are numerous ways for the user to be redirected, and the Firefox API
support to detect each of them is poor. We have a <a class="ulink" href="https://trac.torproject.org/projects/tor/ticket/3600" target="_top">trac bug
open</a> to implement what we can.

</p></li><li class="listitem"><span class="command"><strong>window.name</strong></span><p>

<a class="ulink" href="https://developer.mozilla.org/En/DOM/Window.name" target="_top">window.name</a> is
a magical DOM property that for some reason is allowed to retain a persistent value
for the lifespan of a browser tab. It is possible to utilize this property for
<a class="ulink" href="http://www.thomasfrank.se/sessionvars.html" target="_top">identifier
storage</a>.

</p><p>

In order to eliminate non-consensual linkability but still allow for sites
that utilize this property to function, we reset the window.name property of
tabs in Torbutton every time we encounter a blank Referer. This behavior
allows window.name to persist for the duration of a click-driven navigation
session, but as soon as the user enters a new URL or navigates between
HTTPS/HTTP schemes, the property is cleared.

</p></li><li class="listitem"><span class="command"><strong>Auto form-fill</strong></span><p>

We disable the password saving functionality in the browser as part of our
<a class="link" href="#disk-avoidance" title="4.3. Disk Avoidance">Disk Avoidance</a> requirement. However,
since users may decide to re-enable disk history records and password saving,
we also set the <a class="ulink" href="http://kb.mozillazine.org/Signon.autofillForms" target="_top">signon.autofillForms</a>
preference to false to prevent saved values from immediately populating
fields upon page load. Since Javascript can read these values as soon as they
appear, setting this preference prevents automatic linkability from stored passwords.

</p></li><li class="listitem"><span class="command"><strong>HSTS supercookies</strong></span><p>

An extreme (but not impossible) attack to mount is the creation of <a class="ulink" href="http://www.leviathansecurity.com/blog/archives/12-The-Double-Edged-Sword-of-HSTS-Persistence-and-Privacy.html" target="_top">HSTS
supercookies</a>. Since HSTS effectively stores one bit of information per domain
name, an adversary in possession of numerous domains can use them to construct
cookies based on stored HSTS state.

</p><p><span class="command"><strong>Design Goal:</strong></span>

There appears to be three options for us: 1. Disable HSTS entirely, and rely
instead on HTTPS-Everywhere to crawl and ship rules for HSTS sites. 2.
Restrict the number of HSTS-enabled third parties allowed per URL bar origin.
3. Prevent third parties from storing HSTS rules. We have not yet decided upon
the best approach.

</p><p><span class="command"><strong>Implementation Status:</strong></span> Currently, HSTS state is
cleared by <a class="link" href="#new-identity" title="4.7. Long-Term Unlinkability via &quot;New Identity&quot; button">New Identity</a>, but we don't
defend against the creation of these cookies between <span class="command"><strong>New
Identity</strong></span> invocations.
      </p></li></ol></div><p>
For more details on identifier linkability bugs and enhancements, see the <a class="ulink" href="https://trac.torproject.org/projects/tor/query?keywords=~tbb-linkability&amp;status=!closed" target="_top">tbb-linkability tag in our bugtracker</a>
  </p></div></div><div class="sect2"><div class="titlepage"><div><div><h3 class="title"><a id="fingerprinting-linkability"></a>4.6. Cross-Origin Fingerprinting Unlinkability</h3></div></div></div><p>

Browser fingerprinting is the act of inspecting browser behaviors and features in
an attempt to differentiate and track individual users. Fingerprinting attacks
are typically broken up into passive and active vectors. Passive
fingerprinting makes use of any information the browser provides automatically
to a website without any specific action on the part of the website. Active
fingerprinting makes use of any information that can be extracted from the
browser by some specific website action, usually involving Javascript.
Some definitions of browser fingerprinting also include supercookies and
cookie-like identifier storage, but we deal with those issues separately in
the <a class="link" href="#identifier-linkability" title="4.5. Cross-Origin Identifier Unlinkability">preceding section on identifier
linkability</a>.

</p><p>

For the most part, however, we do not differentiate between passive or active
fingerprinting sources, since many active fingerprinting mechanisms are very
rapid, and can be obfuscated or disguised as legitimate functionality.
Instead, we believe fingerprinting can only be rationally addressed if we
understand where the problem comes from, what sources of issues are the most
severe, and how to study the efficacy of defenses properly.

</p><div class="sect3"><div class="titlepage"><div><div><h4 class="title"><a id="fingerprinting-scope"></a>Sources of Fingerprinting Issues</h4></div></div></div><p>

All browser fingerprinting issues arise from one of four primary sources:
end-user configuration details, device and hardware characteristics, operating
system vendor and version differences, and browser vendor and version
differences. Additionally, user behavior itself provides one more source of
potential fingerprinting.

</p><p>

In order to help prioritize and inform defenses, we now list these sources in
order from most severe to least severe in terms of the amount of information
they reveal, and describe them in more detail.

</p><div class="orderedlist"><ol class="orderedlist" type="1"><li class="listitem"><span class="command"><strong>End-user Configuration Details</strong></span><p>

End-user configuration details are by far the most severe threat to
fingerprinting, as they will quickly provide enough information to uniquely
identify a user. We believe it is essential to avoid exposing platform
configuration details to website content at all costs. We also discourage
excessive fine-grained customization of Tor Browser by minimizing and
aggregating user-facing privacy and security options, as well as by
discouraging the use of additional addons. When it is necessary to expose
configuration details in the course of providing functionality, we strive to
do so only on a per-site basis via site permissions, to avoid linkability.

</p></li><li class="listitem"><span class="command"><strong>Device and Hardware Characteristics</strong></span><p>

Device and hardware characteristics can be determined in three ways: they can
be reported explicitly by the browser, they can be inferred through browser
functionality, or they can be extracted through statistical measurements of
system performance. We are most concerned with the cases where this
information is either directly reported or can be determined via a single use
of an API or feature, and prefer to place such APIs either behind site
permissions, alter their functionality to prevent exposing the most variable
aspects of these characteristics, or disable them entirely.

</p><p>

On the other hand, because statistical inference of system performance
requires many iterations to achieve accuracy in the face of noise and
concurrent activity, we are less concerned with this mechanism of extracting
this information. We also expect that reducing the resolution of Javascript's
time sources will significantly increase the duration of execution required to
extract accurate results, and thus make statistical approaches both
unattractive and highly noticeable due to excessive resource consumption.

</p></li><li class="listitem"><span class="command"><strong>Operating System Vendor and Version Differences</strong></span><p>

Operating system vendor and version differences permeate many different
aspects of the browser. While it is possible to address these issues with some
effort, the relative lack of diversity in operating systems causes us to
primarily focus our efforts on passive operating system fingerprinting
mechanisms at this point in time. For the purposes of protecting user
anonymity, it is not strictly essential that the operating system be
completely concealed, though we recognize that it is useful to reduce this
differentiation ability where possible, especially for cases where the
specific version of a system can be inferred.

</p></li><li class="listitem"><span class="command"><strong>User Behavior</strong></span><p>

While somewhat outside the scope of browser fingerprinting, for completeness
it is important to mention that users themselves theoretically might be
fingerprinted through their behavior while interacting with a website. This
behavior includes e.g. keystrokes, mouse movements, click speed, and writing
style. Basic vectors such as keystroke and mouse usage fingerprinting can be
mitigated by altering Javascript's notion of time. More advanced issues like
writing style fingerprinting are the domain of <a class="ulink" href="https://github.com/psal/anonymouth" target="_top">other tools</a>.

</p></li><li class="listitem"><span class="command"><strong>Browser Vendor and Version Differences</strong></span><p>

Due to vast differences in feature set and implementation behavior even
between different versions of the same browser, browser vendor and version
differences are simply not possible to conceal in any realistic way. It
is only possible to minimize the differences among different installations of
the same browser vendor and version. We make no effort to mimic any other
major browser vendor, and in fact most of our fingerprinting defenses serve to
differentiate Tor Browser users from normal Firefox users. Because of this,
any study that lumps browser vendor and version differences into its analysis
of the fingerprintability of a population is largely useless for evaluating
either attacks or defenses. Unfortunately, this includes popular large-scale
studies such as <a class="ulink" href="https://panopticlick.eff.org/" target="_top">Panopticlick</a> and <a class="ulink" href="https://amiunique.org/" target="_top">Am I Unique</a>.

</p></li></ol></div></div><div class="sect3"><div class="titlepage"><div><div><h4 class="title"><a id="fingerprinting-defenses-general"></a>General Fingerprinting Defenses</h4></div></div></div><p>

To date, the Tor Browser team has concerned itself only with developing
defenses for APIs that have already been standardized and deployed. Once an
API or feature has been standardized and widely deployed, defenses to the
associated fingerprinting issues tend to have only a few options available to
compensate for the lack of up-front privacy design. In our experience, so far
these options have been limited to value spoofing, subsystem reimplementation,
virtualization, site permissions, and feature removal. We will now describe
these options and the fingerprinting sources they tend to work best with.

</p><div class="orderedlist"><ol class="orderedlist" type="1"><li class="listitem"><span class="command"><strong>Value Spoofing</strong></span><p>

Value spoofing can be used for simple cases where the browser directly
provides some aspect of the user's configuration details, devices, hardware,
or operating system directly to a website. It becomes less useful when the
fingerprinting method relies on behavior to infer aspects of the hardware or
operating system, rather than obtain them directly.

</p></li><li class="listitem"><span class="command"><strong>Subsystem Reimplementation</strong></span><p>

In cases where simple spoofing is not enough to properly conceal underlying
device characteristics or operating system details, the underlying
subsystem that provides the functionality for a feature or API may need
to be completely reimplemented. This is most common in cases where
customizable or version-specific aspects of the user's operating system are
visible through the browser's featureset or APIs, usually because the browser
directly exposes OS-provided implementations of underlying features. In these
cases, such OS-provided implementations must be replaced by a generic
implementation, or at least an implementation wrapper that makes effort to
conceal any user-customized aspects of the system.

</p></li><li class="listitem"><span class="command"><strong>Virtualization</strong></span><p>

Virtualization is needed when simply reimplementing a feature in a different
way is insufficient to fully conceal the underlying behavior. This is most
common in instances of device and hardware fingerprinting, but since the
notion of time can also be virtualized, virtualization also can apply to any
instance where an accurate measurement of wall clock time is required for a
fingerprinting vector to attain high accuracy.

</p></li><li class="listitem"><span class="command"><strong>Site Permissions</strong></span><p>

In the event that reimplementation or virtualization is too expensive in terms
of performance or engineering effort, and the relative expected usage of a
feature is rare, site permissions can be used to prevent the usage of a
feature for cross-site tracking. Unfortunately, site permissions become less
effective once a feature is already widely overused and abused by many
websites, since warning fatigue typically sets in for most users after just a
few permission requests.

</p></li><li class="listitem"><span class="command"><strong>Feature or Functionality Removal</strong></span><p>

Due to the current bias in favor of invasive APIs that expose the maximum
amount of platform information, some features and APIs are simply not
salvageable in their current form. When such invasive features serve only a
narrow domain or use case, or when there are alternate ways of accomplishing
the same task, these features and/or certain aspects of their functionality
may be simply removed.

</p></li></ol></div></div><div class="sect3"><div class="titlepage"><div><div><h4 class="title"><a id="idp56040528"></a>Randomization or Uniformity?</h4></div></div></div><p>

When applying a form of defense to a specific fingerprinting vector or source,
there are two general strategies available. Either the implementation for all
users of a single browser version can be made to behave as uniformly as
possible, or the user agent can attempt to randomize its behavior, so that
each interaction between a user and a site provides a different fingerprint.

</p><p>

Although <a class="ulink" href="http://research.microsoft.com/pubs/209989/tr1.pdf" target="_top">some
research suggests</a> that randomization can be effective, so far striving
for uniformity has generally proved to be a better strategy for Tor Browser
for the following reasons:

</p><div class="orderedlist"><ol class="orderedlist" type="1"><li class="listitem"><span class="command"><strong>Randomization is not a shortcut</strong></span><p>

While many end-user configuration details that the browser currently exposes
may be safely replaced by false information, randomization of these details
must be just as exhaustive as an approach that seeks to make these behaviors
uniform. When confronting either strategy, the adversary can still make use of
any details which have not been altered to be either sufficiently uniform or
sufficiently random.

</p><p>

Furthermore, the randomization approach seems to break down when it is applied
to deeper issues where underlying system functionality is directly exposed. In
particular, it is not clear how to randomize the capabilities of hardware
attached to a computer in such a way that it either convincingly behaves like
other hardware, or such that the exact properties of the hardware that vary
from user to user are sufficiently randomized. Similarly, truly concealing
operating system version differences through randomization may require
multiple reimplementations of the underlying operating system functionality to
ensure that every operating system version is covered by the range of possible
behaviors.

</p></li><li class="listitem"><span class="command"><strong>Evaluation and measurement difficulties</strong></span><p>

The fact that randomization causes behaviors to differ slightly with every
site visit makes it appealing at first glance, but this same property makes it
very difficult to objectively measure its effectiveness. By contrast, an
implementation that strives for uniformity is very simple to measure. Despite
their current flaws, a properly designed version of <a class="ulink" href="https://panopticlick.eff.org/" target="_top">Panopticlick</a> or <a class="ulink" href="https://amiunique.org/" target="_top">Am I Unique</a> could report the entropy and
uniqueness rates for all users of a single user agent version, without the
need for complicated statistics about the variance of the measured behaviors.

</p><p>

Randomization (especially incomplete randomization) may also provide a false
sense of security. When a fingerprinting attempt makes naive use of randomized
information, a fingerprint will appear unstable, but may not actually be
sufficiently randomized to prevent a dedicated adversary.  Sophisticated
fingerprinting mechanisms may either ignore randomized information, or
incorporate knowledge of the distribution and range of randomized values into
the creation of a more stable fingerprint (by either removing the randomness,
modeling it, or averaging it out).

</p></li><li class="listitem"><span class="command"><strong>Usability issues</strong></span><p>

When randomization is introduced to features that affect site behavior, it can
be very distracting for this behavior to change between visits of a given
site. For the simplest cases, this will lead to minor visual nuisances.
However, when this information affects reported functionality or hardware
characteristics, sometimes a site will function one way on one visit, and
another way on a subsequent visit.

</p></li><li class="listitem"><span class="command"><strong>Performance costs</strong></span><p>

Randomizing involves performance costs. This is especially true if the
fingerprinting surface is large (like in a modern browser) and one needs more
elaborate randomizing strategies (including randomized virtualization) to
ensure that the randomization fully conceals the true behavior. Many calls to
a cryptographically secure random number generator during the course of a page
load will both serve to exhaust available entropy pools, as well as lead to
increased computation while loading a page.

</p></li><li class="listitem"><span class="command"><strong>Increased vulnerability surface</strong></span><p>

Improper randomization might introduce a new fingerprinting vector, as the
process of generating the values for the fingerprintable attributes could be
itself susceptible to side-channel attacks, analysis, or exploitation.

</p></li></ol></div></div><div class="sect3"><div class="titlepage"><div><div><h4 class="title"><a id="fingerprinting-defenses"></a>Specific Fingerprinting Defenses in the Tor Browser</h4></div></div></div><p>

The following defenses are listed roughly in order of most severe
fingerprinting threat first. This ordering is based on the above intuition
that user configurable aspects of the computer are the most severe source of
fingerprintability, followed by device characteristics and hardware, and then
finally operating system vendor and version information.

</p><p>

Where our actual implementation differs from an ideal solution, we separately
describe our <span class="command"><strong>Design Goal</strong></span> and our <span class="command"><strong>Implementation
Status</strong></span>.

</p><div class="orderedlist"><ol class="orderedlist" type="1"><li class="listitem"><span class="command"><strong>Plugins</strong></span><p>

Plugins add to fingerprinting risk via two main vectors: their mere presence
in window.navigator.plugins (because they are optional, end-user installed
third party software), as well as their internal functionality.

</p><p><span class="command"><strong>Design Goal:</strong></span>

All plugins that have not been specifically audited or sandboxed MUST be
disabled. To reduce linkability potential, even sandboxed plugins should not
be allowed to load objects until the user has clicked through a click-to-play
barrier.  Additionally, version information should be reduced or obfuscated
until the plugin object is loaded. For flash, we wish to <a class="ulink" href="https://trac.torproject.org/projects/tor/ticket/3974" target="_top">provide a
settings.sol file</a> to disable Flash cookies, and to restrict P2P
features that are likely to bypass proxy settings. We'd also like to restrict
access to fonts and other system information (such as IP address and MAC
address) in such a sandbox.

</p><p><span class="command"><strong>Implementation Status:</strong></span>

Currently, we entirely disable all plugins in Tor Browser. However, as a
compromise due to the popularity of Flash, we allow users to re-enable Flash,
and flash objects are blocked behind a click-to-play barrier that is available
only after the user has specifically enabled plugins. Flash is the only plugin
available, the rest are entirely
blocked from loading by the Firefox patches mentioned in the <a class="link" href="#proxy-obedience" title="4.1. Proxy Obedience">Proxy Obedience
section</a>. We also set the Firefox
preference <span class="command"><strong>plugin.expose_full_path</strong></span> to false, to avoid
leaking plugin installation information.

</p></li><li class="listitem"><span class="command"><strong>HTML5 Canvas Image Extraction</strong></span><p>

After plugins and plugin-provided information, we believe that the <a class="ulink" href="https://developer.mozilla.org/en-US/docs/HTML/Canvas" target="_top">HTML5
Canvas</a> is the single largest fingerprinting threat browsers face
today. <a class="ulink" href="http://www.w2spconf.com/2012/papers/w2sp12-final4.pdf" target="_top">Initial
studies</a> show that the Canvas can provide an easy-access fingerprinting
target: The adversary simply renders WebGL, font, and named color data to a
Canvas element, extracts the image buffer, and computes a hash of that image
data. Subtle differences in the video card, font packs, and even font and
graphics library versions allow the adversary to produce a stable, simple,
high-entropy fingerprint of a computer. In fact, the hash of the rendered
image can be used almost identically to a tracking cookie by the web server.

</p><p>

In some sense, the canvas can be seen as the union of many other
fingerprinting vectors. If WebGL is normalized through software rendering,
system colors were standardized, and the browser shipped a fixed collection of
fonts (see later points in this list), it might not be necessary to create a
canvas permission. However, until then, to reduce the threat from this vector,
we have patched Firefox to <a class="ulink" href="https://gitweb.torproject.org/tor-browser.git/commit/?h=tor-browser-31.6.0esr-4.5-1&amp;id=6a169ef0166b268b1a27546a17b3d7470330917d" target="_top">prompt before returning valid image data</a> to the Canvas APIs,
and for <a class="ulink" href="https://gitweb.torproject.org/tor-browser.git/commit/?h=tor-browser-31.6.0esr-4.5-1&amp;id=7d51acca6383732480b49ccdb5506ad6fb92e651" target="_top">access to isPointInPath and related functions</a>.
If the user hasn't previously allowed the site in the URL bar to access Canvas
image data, pure white image data is returned to the Javascript APIs.

</p><p>
     </p></li><li class="listitem"><span class="command"><strong>Open TCP Port and Local Network Fingerprinting</strong></span><p>

In Firefox, by using either WebSockets or XHR, it is possible for remote
content to <a class="ulink" href="http://www.andlabs.org/tools/jsrecon.html" target="_top">enumerate
the list of TCP ports open on 127.0.0.1</a>, as well as on any other
machines on the local network. In other browsers, this can be accomplished by
DOM events on image or script tags. This open vs filtered vs closed port list
can provide a very unique fingerprint of a machine, because it essentially
enables the detection of many different popular third party applications and
optional system services (Skype, Bitcoin, Bittorrent and other P2P software,
SSH ports, SMB and related LAN services, CUPS and printer daemon config ports,
mail servers, and so on). It is also possible to determine when ports are
closed versus filtered/blocked (and thus probe custom firewall configuration).

</p><p>

In Tor Browser, we prevent access to 127.0.0.1/localhost by ensuring that even
these requests are still sent by Firefox to our SOCKS proxy (ie we set
<span class="command"><strong>network.proxy.no_proxies_on</strong></span> to the empty string). The local
Tor client then rejects them, since it is configured to proxy for internal IP
addresses by default. Access to the local network is forbidden via the same
mechanism.

</p></li><li class="listitem"><span class="command"><strong>Invasive Authentication Mechanisms (NTLM and SPNEGO)</strong></span><p>

Both NTLM and SPNEGO authentication mechanisms can leak the hostname, and in
some cases the current username. The only reason why these aren't a more
serious problem is that they typically involve user interaction, and likely
aren't an attractive vector for this reason. However, because it is not clear
if certain carefully-crafted error conditions in these protocols could cause
them to reveal machine information and still fail silently prior to the
password prompt, these authentication mechanisms should either be disabled, or
placed behind a site permission before their use. We simply disable them.

</p></li><li class="listitem"><span class="command"><strong>USB Device ID Enumeration</strong></span><p>

The <a class="ulink" href="https://developer.mozilla.org/en-US/docs/Web/Guide/API/Gamepad" target="_top">GamePad
API</a> provides web pages with the <a class="ulink" href="https://dvcs.w3.org/hg/gamepad/raw-file/default/gamepad.html#widl-Gamepad-id" target="_top">USB
device id, product id, and driver name</a> of all connected game
controllers, as well as detailed information about their capabilities. This API
should be behind a site permission in Private Browsing Modes, or should present a generic 
controller type (perhaps a two button controller that can be mapped to the keyboard) in all cases.
We simply disable it via the pref <span class="command"><strong>dom.gamepad.enabled</strong></span>.

</p></li><li class="listitem"><span class="command"><strong>Fonts</strong></span><p>

According to the Panopticlick study, fonts provide the most linkability when
they are provided as an enumerable list in file system order, via either the
Flash or Java plugins. However, it is still possible to use CSS and/or
Javascript to query for the existence of specific fonts. With a large enough
pre-built list to query, a large amount of fingerprintable information may
still be available, especially given that additional fonts often end up
installed by third party software and for multilingual support.

</p><p><span class="command"><strong>Design Goal:</strong></span> The sure-fire way to address font
linkability is to ship the browser with a font for every language, typeface,
and style, and to only use those fonts at the exclusion of system fonts. We are
<a class="ulink" href="https://trac.torproject.org/projects/tor/ticket/13313" target="_top">currently
investigating</a> this approach, and our current favorite font sets for
this purpose are the <a class="ulink" href="http://www.droidfonts.com/droidfonts/" target="_top">Droid
fonts</a>, the <a class="ulink" href="http://hangeul.naver.com/" target="_top">Nanum fonts</a>,
and <a class="ulink" href="https://fedorahosted.org/lohit/" target="_top">Lohit fonts</a>. The Droid
font set is fairly complete by itself, but Nanum and Lohit have smaller
versions of many South Asian languages. When combined in a way that chooses the
smallest font implementations for each locale, these three font sets provide
coverage for the all languages used on Wikipedia with more than
10,000 articles, and several others as well, in approximately 3MB of compressed
overhead. The <a class="ulink" href="https://www.google.com/get/noto/" target="_top">Noto font
set</a> is another font set that aims for complete coverage, but is
considerably larger than the combination of the Droid, Nanum, and Lohit fonts.

</p><p><span class="command"><strong>Implementation Status:</strong></span>

In the meantime while we investigate shipping our own fonts, we disable
plugins, which prevents font name enumeration. Additionally, we limit both the
number of font queries from CSS, as well as the total number of fonts that can
be used in a document <a class="ulink" href="https://gitweb.torproject.org/tor-browser.git/commit/?h=tor-browser-31.6.0esr-4.5-1&amp;id=e78bc05159a79c1358fa9c64e565af9d98c141ee" target="_top">with
a Firefox patch</a>. We create two prefs,
<span class="command"><strong>browser.display.max_font_attempts</strong></span> and
<span class="command"><strong>browser.display.max_font_count</strong></span> for this purpose. Once these
limits are reached, the browser behaves as if
<span class="command"><strong>browser.display.use_document_fonts</strong></span> was set.

</p><p>

To improve rendering, we exempt remote <a class="ulink" href="https://developer.mozilla.org/en-US/docs/CSS/@font-face" target="_top">@font-face
fonts</a> from these counts, and if a font-family CSS rule lists a remote
font (in any order), we use that font instead of any of the named local fonts.

</p></li><li class="listitem"><span class="command"><strong>Monitor, Widget, and OS Desktop Resolution</strong></span><p>

Both CSS and Javascript have access to a lot of information about the screen
resolution, usable desktop size, OS widget size, toolbar size, title bar size,
and OS desktop widget sizing information that are not at all relevant to
rendering and serve only to provide information for fingerprinting. Since many
aspects of desktop widget positioning and size are user configurable, these
properties yield customized information about the computer, even beyond the
monitor size.

</p><p><span class="command"><strong>Design Goal:</strong></span>

Our design goal here is to reduce the resolution information down to the bare
minimum required for properly rendering inside a content window. We intend to
report all rendering information correctly with respect to the size and
properties of the content window, but report an effective size of 0 for all
border material, and also report that the desktop is only as big as the inner
content window. Additionally, new browser windows are sized such that their
content windows are one of a few fixed sizes based on the user's desktop
resolution. In addition, to further reduce resolution-based fingerprinting, we
are <a class="ulink" href="https://trac.torproject.org/projects/tor/ticket/7256" target="_top">investigating
zoom/viewport-based mechanisms</a> that might allow us to always report the
same desktop resolution regardless of the actual size of the content window,
and simply scale to make up the difference.  Until then, the user should also
be informed that maximizing their windows can lead to fingerprintability under
this scheme.

</p><p><span class="command"><strong>Implementation Status:</strong></span>

We automatically resize new browser windows to a 200x100 pixel multiple using
a window observer <a class="ulink" href="https://gitweb.torproject.org/torbutton.git/tree/src/chrome/content/torbutton.js#n3361" target="_top">based
on desktop resolution</a>. To minimize the effect of the long tail of large
monitor sizes, we also cap the window size at 1000 pixels in each direction.
Additionally, we patch Firefox to use the client content window size <a class="ulink" href="https://gitweb.torproject.org/tor-browser.git/commit/?h=tor-browser-31.6.0esr-4.5-1&amp;id=bd3b1ed32a9c21fdc92fc35f2ec0a41badc378d5" target="_top">for
window.screen</a>, and to <a class="ulink" href="https://gitweb.torproject.org/tor-browser.git/commit/?h=tor-browser-31.6.0esr-4.5-1&amp;id=a5648c8d80f396caf294d761cc4a9a76c0b33a9d" target="_top">report
a window.devicePixelRatio of 1.0</a>. Similarly, we <a class="ulink" href="https://gitweb.torproject.org/tor-browser.git/commit/?h=tor-browser-31.6.0esr-4.5-1&amp;id=3c02858027634ffcfbd97047dfdf170c19ca29ec" target="_top">patch
DOM events to return content window relative points</a>.

We also force popups to open in new tabs (via
<span class="command"><strong>browser.link.open_newwindow.restriction</strong></span>), to avoid
full-screen popups inferring information about the browser resolution. In
addition, we prevent auto-maximizing on browser start, and inform users that
maximized windows are detrimental to privacy in this mode.

</p></li><li class="listitem"><span class="command"><strong>Display Media information</strong></span><p>

Beyond simple resolution information, a large amount of so-called "Media"
information is also exported to content. Even without Javascript, CSS has
access to a lot of information about the device orientation, system theme
colors, and other desktop and display features that are not at all relevant to
rendering and also user configurable. Most of this
information comes from <a class="ulink" href="https://developer.mozilla.org/en-US/docs/Web/Guide/CSS/Media_queries" target="_top">CSS
Media Queries</a>, but Mozilla has exposed <a class="ulink" href="https://developer.mozilla.org/en-US/docs/Web/CSS/color_value#System_Colors" target="_top">several
user and OS theme defined color values</a> to CSS as well.

</p><p><span class="command"><strong>Design Goal:</strong></span>

CSS should not be able infer anything that the user has configured about their
computer. Additionally, it should not be able to infer machine-specific
details such as screen orientation or type.

</p><p><span class="command"><strong>Implementation Status:</strong></span>

We patch
Firefox to <a class="ulink" href="https://gitweb.torproject.org/tor-browser.git/commit/?h=tor-browser-31.6.0esr-4.5-1&amp;id=cf8956b4460107c5b0053c8fc574e34b0a30ec1e" target="_top">report
a fixed set of system colors to content window CSS</a>, and <a class="ulink" href="https://gitweb.torproject.org/tor-browser.git/commit/?h=tor-browser-31.6.0esr-4.5-1&amp;id=bbc138486e0489b0d559343fa0522df4ee3b3533" target="_top">prevent
detection of font smoothing on OSX</a>. We also always
<a class="ulink" href="https://gitweb.torproject.org/tor-browser.git/commit/?h=tor-browser-31.6.0esr-4.5-1&amp;id=e17d60442ab0db92664ff68d90fe7bf737374912" target="_top">report
landscape-primary</a> for the screen orientation.

</p></li><li class="listitem"><span class="command"><strong>WebGL</strong></span><p>

WebGL is fingerprintable both through information that is exposed about the
underlying driver and optimizations, as well as through performance
fingerprinting.

</p><p>

Because of the large amount of potential fingerprinting vectors and the <a class="ulink" href="http://www.contextis.com/resources/blog/webgl/" target="_top">previously unexposed
vulnerability surface</a>, we deploy a similar strategy against WebGL as
for plugins. First, WebGL Canvases have click-to-play placeholders (provided
by NoScript), and do not run until authorized by the user. Second, we
obfuscate driver information by setting the Firefox preferences
<span class="command"><strong>webgl.disable-extensions</strong></span> and
<span class="command"><strong>webgl.min_capability_mode</strong></span>, which reduce the information
provided by the following WebGL API calls: <span class="command"><strong>getParameter()</strong></span>,
<span class="command"><strong>getSupportedExtensions()</strong></span>, and
<span class="command"><strong>getExtension()</strong></span>.

</p><p>

Another option for WebGL might be to use software-only rendering, using a
library such as <a class="ulink" href="http://www.mesa3d.org/" target="_top">Mesa</a>. The use of
such a library would avoid hardware-specific rendering differences.

</p></li><li class="listitem"><span class="command"><strong>User Agent and HTTP Headers</strong></span><p><span class="command"><strong>Design Goal:</strong></span>

All Tor Browser users MUST provide websites with an identical user agent and
HTTP header set for a given request type. We omit the Firefox minor revision,
and report a popular Windows platform. If the software is kept up to date,
these headers should remain identical across the population even when updated.

</p><p><span class="command"><strong>Implementation Status:</strong></span>

Firefox provides several options for controlling the browser user agent string
which we leverage. We also set similar prefs for controlling the
Accept-Language and Accept-Charset headers, which we spoof to English by default. Additionally, we
<a class="ulink" href="https://gitweb.torproject.org/tor-browser.git/commit/?h=tor-browser-31.6.0esr-4.5-1&amp;id=e9841ee41e7f3f1535be2d605084c41ee9faf6c2" target="_top">remove
content script access</a> to Components.interfaces, which <a class="ulink" href="http://pseudo-flaw.net/tor/torbutton/fingerprint-firefox.html" target="_top">can be
used</a> to fingerprint OS, platform, and Firefox minor version.  </p></li><li class="listitem"><span class="command"><strong>Locale Fingerprinting</strong></span><p>

In Tor Browser, we provide non-English users the option of concealing their OS
and browser locale from websites. It is debatable if this should be as high of
a priority as information specific to the user's computer, but for
completeness, we attempt to maintain this property.

</p><p><span class="command"><strong>Implementation Status:</strong></span>

We set the fallback character set to set to windows-1252 for all locales, via
<span class="command"><strong>intl.charset.default</strong></span>.  We also patch Firefox to allow us to
<a class="ulink" href="https://gitweb.torproject.org/tor-browser.git/commit/?h=tor-browser-31.6.0esr-4.5-1&amp;id=4545ecd6dc2ca7d10aefe36b81658547ea97b800" target="_top">instruct
the JS engine</a> to use en-US as its internal C locale for all Date, Math,
and exception handling.

</p></li><li class="listitem"><span class="command"><strong>Timezone and Clock Offset</strong></span><p>

While the latency in Tor connections varies anywhere from milliseconds to
a few seconds, it is still possible for the remote site to detect large
differences between the user's clock and an official reference time source.

</p><p><span class="command"><strong>Design Goal:</strong></span>

All Tor Browser users MUST report the same timezone to websites. Currently, we
choose UTC for this purpose, although an equally valid argument could be made
for EDT/EST due to the large English-speaking population density (coupled with
the fact that we spoof a US English user agent).  Additionally, the Tor
software should detect if the users clock is significantly divergent from the
clocks of the relays that it connects to, and use this to reset the clock
values used in Tor Browser to something reasonably accurate. Alternatively,
the browser can obtain this clock skew via a mechanism similar to that used in
<a class="ulink" href="https://github.com/ioerror/tlsdate" target="_top">tlsdate</a>.

</p><p><span class="command"><strong>Implementation Status:</strong></span>

We set the timezone using the TZ environment variable, which is supported on
all platforms.

</p></li><li class="listitem"><span class="command"><strong>Javascript Performance Fingerprinting</strong></span><p>

<a class="ulink" href="http://w2spconf.com/2011/papers/jspriv.pdf" target="_top">Javascript performance
fingerprinting</a> is the act of profiling the performance
of various Javascript functions for the purpose of fingerprinting the
Javascript engine and the CPU.

</p><p><span class="command"><strong>Design Goal:</strong></span>

We have <a class="ulink" href="https://trac.torproject.org/projects/tor/ticket/3059" target="_top">several potential
mitigation approaches</a> to reduce the accuracy of performance
fingerprinting without risking too much damage to functionality. Our current
favorite is to reduce the resolution of the Event.timeStamp and the Javascript
Date() object, while also introducing jitter. We believe that Javascript time
resolution may be reduced all the way up to the second before it seriously
impacts site operation. Our goal with this quantization is to increase the
amount of time it takes to mount a successful attack. <a class="ulink" href="http://w2spconf.com/2011/papers/jspriv.pdf" target="_top">Mowery et al</a> found
that even with the default precision in most browsers, they required up to 120
seconds of amortization and repeated trials to get stable results from their
feature set. We intend to work with the research community to establish the
optimum trade-off between quantization+jitter and amortization time, as well
as identify highly variable Javascript operations. As long as these attacks
take several seconds or more to execute, they are unlikely to be appealing to
advertisers, and are also very likely to be noticed if deployed against a
large number of people.

</p><p><span class="command"><strong>Implementation Status:</strong></span>

Currently, our mitigation against performance fingerprinting is to
disable <a class="ulink" href="http://www.w3.org/TR/navigation-timing/" target="_top">Navigation
Timing</a> through the Firefox preference
<span class="command"><strong>dom.enable_performance</strong></span>, and to disable the <a class="ulink" href="https://developer.mozilla.org/en-US/docs/Web/API/HTMLVideoElement#Gecko-specific_properties" target="_top">Mozilla
Video Statistics</a> API extensions via the preference
<span class="command"><strong>media.video_stats.enabled</strong></span>.

</p></li><li class="listitem"><span class="command"><strong>Keystroke Fingerprinting</strong></span><p>

Keystroke fingerprinting is the act of measuring key strike time and key
flight time. It is seeing increasing use as a biometric.

</p><p><span class="command"><strong>Design Goal:</strong></span>

We intend to rely on the same mechanisms for defeating Javascript performance
fingerprinting: timestamp quantization and jitter.

</p><p><span class="command"><strong>Implementation Status:</strong></span>
We have no implementation as of yet.
     </p></li><li class="listitem"><span class="command"><strong>Operating System Type Fingerprinting</strong></span><p>

As we mentioned in the introduction of this section, OS type fingerprinting is
currently considered a lower priority, due simply to the numerous ways that
characteristics of the operating system type may leak into content, and the
comparatively low contribution of OS to overall entropy. In particular, there
are likely to be many ways to measure the differences in widget size,
scrollbar size, and other rendered details on a page. Also, directly exported
OS routines (such as those from the standard C math library) expose
differences in their implementations through their return values.

</p><p><span class="command"><strong>Design Goal:</strong></span>

We intend to reduce or eliminate OS type fingerprinting to the best extent
possible, but recognize that the effort for reward on this item is not as high
as other areas. The entropy on the current OS distribution is somewhere around
2 bits, which is much lower than other vectors which can also be used to
fingerprint configuration and user-specific information.  You can see the
major areas of OS fingerprinting we're aware of using the <a class="ulink" href="https://trac.torproject.org/projects/tor/query?keywords=~tbb-fingerprinting-os" target="_top">tbb-fingerprinting-os
tag on our bug tracker</a>.

</p><p><span class="command"><strong>Implementation Status:</strong></span>

At least three HTML5 features have different implementation status across the
major OS vendors and/or the underlying hardware: the <a class="ulink" href="https://developer.mozilla.org/en-US/docs/DOM/window.navigator.battery" target="_top">Battery
API</a>, the <a class="ulink" href="https://developer.mozilla.org/en-US/docs/DOM/window.navigator.connection" target="_top">Network
Connection API</a>, and the <a class="ulink" href="https://wiki.mozilla.org/Sensor_API" target="_top">Sensor API</a>. We disable these APIs through the Firefox preferences
<span class="command"><strong>dom.battery.enabled</strong></span>,
<span class="command"><strong>dom.network.enabled</strong></span>, and
<span class="command"><strong>device.sensors.enabled</strong></span>.

</p></li></ol></div><p>
For more details on fingerprinting bugs and enhancements, see the <a class="ulink" href="https://trac.torproject.org/projects/tor/query?keywords=~tbb-fingerprinting&amp;status=!closed" target="_top">tbb-fingerprinting tag in our bug tracker</a>
   </p></div></div><div class="sect2"><div class="titlepage"><div><div><h3 class="title"><a id="new-identity"></a>4.7. Long-Term Unlinkability via "New Identity" button</h3></div></div></div><p>

In order to avoid long-term linkability, we provide a "New Identity" context
menu option in Torbutton. This context menu option is active if Torbutton can
read the environment variables $TOR_CONTROL_PASSWD and $TOR_CONTROL_PORT.

</p><div class="sect3"><div class="titlepage"><div><div><h4 class="title"><a id="idp56156768"></a>Design Goal:</h4></div></div></div><div class="blockquote"><blockquote class="blockquote">

All linkable identifiers and browser state MUST be cleared by this feature.

</blockquote></div></div><div class="sect3"><div class="titlepage"><div><div><h4 class="title"><a id="idp56158016"></a>Implementation Status:</h4></div></div></div><div class="blockquote"><blockquote class="blockquote"><p>

First, Torbutton disables Javascript in all open tabs and windows by using
both the <a class="ulink" href="https://developer.mozilla.org/en-US/docs/XPCOM_Interface_Reference/nsIDocShell#Attributes" target="_top">browser.docShell.allowJavascript</a>
attribute as well as <a class="ulink" href="https://developer.mozilla.org/en-US/docs/XPCOM_Interface_Reference/nsIDOMWindowUtils#suppressEventHandling%28%29" target="_top">nsIDOMWindowUtil.suppressEventHandling()</a>.
We then stop all page activity for each tab using <a class="ulink" href="https://developer.mozilla.org/en-US/docs/XPCOM_Interface_Reference/nsIWebNavigation#stop%28%29" target="_top">browser.webNavigation.stop(nsIWebNavigation.STOP_ALL)</a>.
We then clear the site-specific Zoom by temporarily disabling the preference
<span class="command"><strong>browser.zoom.siteSpecific</strong></span>, and clear the GeoIP wifi token URL
<span class="command"><strong>geo.wifi.access_token</strong></span> and the last opened URL prefs (if
they exist). Each tab is then closed.

</p><p>

After closing all tabs, we then emit "<a class="ulink" href="https://developer.mozilla.org/en-US/docs/Supporting_private_browsing_mode#Private_browsing_notifications" target="_top">browser:purge-session-history</a>"
(which instructs addons and various Firefox components to clear their session
state), and then manually clear the following state: searchbox and findbox
text, HTTP auth, SSL state, OCSP state, site-specific content preferences
(including HSTS state), content and image cache, offline cache, offline
storage, Cookies, crypto tokens, DOM storage, the safe browsing key, and the
Google wifi geolocation token (if it exists). We also clear NoScript's site
and temporary permissions, and all other browser site permissions.

</p><p>

After the state is cleared, we then close all remaining HTTP keep-alive
connections and then send the NEWNYM signal to the Tor control port to cause a
new circuit to be created.
     </p><p>

Finally, a fresh browser window is opened, and the current browser window is
closed (this does not spawn a new Firefox process, only a new window). Upon
the close of the final window, an unload handler is fired to invoke the <a class="ulink" href="https://developer.mozilla.org/en-US/docs/Mozilla/Tech/XPCOM/Reference/Interface/nsIDOMWindowUtils#garbageCollect%28%29" target="_top">garbage
collector</a>, which has the effect of immediately purging any blob:UUID
URLs that were created by website content via <a class="ulink" href="https://developer.mozilla.org/en-US/docs/Web/API/URL/createObjectURL" target="_top">URL.createObjectURL</a>.

</p></blockquote></div><div class="blockquote"><blockquote class="blockquote">
If the user chose to "protect" any cookies by using the Torbutton Cookie
Protections UI, those cookies are not cleared as part of the above.
    </blockquote></div></div></div><div class="sect2"><div class="titlepage"><div><div><h3 class="title"><a id="other-security"></a>4.8. Other Security Measures</h3></div></div></div><p>

In addition to the above mechanisms that are devoted to preserving privacy
while browsing, we also have a number of technical mechanisms to address other
privacy and security issues.

</p><div class="orderedlist"><ol class="orderedlist" type="1"><li class="listitem"><a id="security-slider"></a><span class="command"><strong>Security Slider</strong></span><p>

In order to provide vulnerability surface reduction for users that need high
security, we have implemented a "Security Slider" to allow users to make a
tradeoff between usability and security while minimizing the total number of
choices (to reduce fingerprinting). Using metrics collected from
Mozilla's bug tracker, we analyzed the vulnerability counts of core
components, and used <a class="ulink" href="https://github.com/iSECPartners/publications/tree/master/reports/Tor%20Browser%20Bundle" target="_top">information
gathered from a study performed by iSec Partners</a> to inform which
features should be disabled at which security levels.

</p><p>

The Security Slider consists of four positions:

</p><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><span class="command"><strong>Low</strong></span><p>

At this security level, the preferences are the Tor Browser defaults.

</p></li><li class="listitem"><span class="command"><strong>Medium-Low</strong></span><p>

At this security level, we disable the ION JIT
(<span class="command"><strong>javascript.options.ion.content</strong></span>), TypeInference JIT
(<span class="command"><strong>javascript.options.typeinference</strong></span>), ASM.JS
(<span class="command"><strong>javascript.options.asmjs</strong></span>), WebAudio
(<span class="command"><strong>media.webaudio.enabled</strong></span>), MathML
(<span class="command"><strong>mathml.disabled</strong></span>), block remote JAR files
(<span class="command"><strong>network.jar.block-remote-files</strong></span>), and make HTML5 audio and
video click-to-play via NoScript (<span class="command"><strong>noscript.forbidMedia</strong></span>).

</p></li><li class="listitem"><span class="command"><strong>Medium-High</strong></span><p>

This security level inherits the preferences from the Medium-Low level, and
additionally disables the baseline JIT
(<span class="command"><strong>javascript.options.baselinejit.content</strong></span>), disables Graphite
(<span class="command"><strong>gfx.font_rendering.graphite.enabled</strong></span>) and SVG OpenType font
rendering (<span class="command"><strong>gfx.font_rendering.opentype_svg.enabled</strong></span>) and only
allows Javascript to run if it is loaded over HTTPS and the URL bar is HTTPS
(by setting <span class="command"><strong>noscript.global</strong></span> to false and
<span class="command"><strong>noscript.globalHttpsWhitelist</strong></span> to true).

</p></li><li class="listitem"><span class="command"><strong>High</strong></span><p>

This security level inherits the preferences from the Medium-Low and
Medium-High levels, and additionally disables remote fonts
(<span class="command"><strong>noscript.forbidFonts</strong></span>), completely disables Javascript (by
unsetting <span class="command"><strong>noscript.globalHttpsWhitelist</strong></span>), and disables SVG
images (<span class="command"><strong>svg.in-content.enabled</strong></span>).

</p></li></ul></div></li><li class="listitem"><a id="traffic-fingerprinting-defenses"></a><span class="command"><strong>Website Traffic Fingerprinting Defenses</strong></span><p>

<a class="link" href="#website-traffic-fingerprinting">Website Traffic
Fingerprinting</a> is a statistical attack to attempt to recognize specific
encrypted website activity.

</p><div class="sect3"><div class="titlepage"><div><div><h4 class="title"><a id="idp56192352"></a>Design Goal:</h4></div></div></div><div class="blockquote"><blockquote class="blockquote"><p>

We want to deploy a mechanism that reduces the accuracy of <a class="ulink" href="https://en.wikipedia.org/wiki/Feature_selection" target="_top">useful features</a> available
for classification. This mechanism would either impact the true and false
positive accuracy rates, <span class="emphasis"><em>or</em></span> reduce the number of web pages
that could be classified at a given accuracy rate.

</p><p>

Ideally, this mechanism would be as light-weight as possible, and would be
tunable in terms of overhead. We suspect that it may even be possible to
deploy a mechanism that reduces feature extraction resolution without any
network overhead. In the no-overhead category, we have <a class="ulink" href="http://freehaven.net/anonbib/cache/LZCLCP_NDSS11.pdf" target="_top">HTTPOS</a> and
<a class="ulink" href="https://blog.torproject.org/blog/experimental-defense-website-traffic-fingerprinting" target="_top">better
use of HTTP pipelining and/or SPDY</a>. 
In the tunable/low-overhead
category, we have <a class="ulink" href="http://freehaven.net/anonbib/cache/ShWa-Timing06.pdf" target="_top">Adaptive
Padding</a> and <a class="ulink" href="http://www.cs.sunysb.edu/~xcai/fp.pdf" target="_top">
Congestion-Sensitive BUFLO</a>. It may be also possible to <a class="ulink" href="https://trac.torproject.org/projects/tor/ticket/7028" target="_top">tune such
defenses</a> such that they only use existing spare Guard bandwidth capacity in the Tor
network, making them also effectively no-overhead.

</p></blockquote></div></div><div class="sect3"><div class="titlepage"><div><div><h4 class="title"><a id="idp56199248"></a>Implementation Status:</h4></div></div></div><div class="blockquote"><blockquote class="blockquote"><p>
Currently, we patch Firefox to <a class="ulink" href="https://gitweb.torproject.org/tor-browser.git/commit/?h=tor-browser-31.6.0esr-4.5-1&amp;id=20a59cec9886cf2575b1fd8e92b43e31ba053fbd" target="_top">randomize
pipeline order and depth</a>. Unfortunately, pipelining is very fragile.
Many sites do not support it, and even sites that advertise support for
pipelining may simply return error codes for successive requests, effectively
forcing the browser into non-pipelined behavior. Firefox also has code to back
off and reduce or eliminate the pipeline if this happens. These
shortcomings and fallback behaviors are the primary reason that Google
developed SPDY as opposed to simply extending HTTP to improve pipelining. It
turns out that we could actually deploy exit-side proxies that allow us to
<a class="ulink" href="https://gitweb.torproject.org/torspec.git/tree/proposals/ideas/xxx-using-spdy.txt" target="_top">use
SPDY from the client to the exit node</a>. This would make our defense not
only free, but one that actually <span class="emphasis"><em>improves</em></span> performance.

</p><p>

Knowing this, we created this defense as an <a class="ulink" href="https://blog.torproject.org/blog/experimental-defense-website-traffic-fingerprinting" target="_top">experimental
research prototype</a> to help evaluate what could be done in the best
case with full server support. Unfortunately, the bias in favor of compelling
attack papers has caused academia to ignore this request thus far, instead
publishing only cursory (yet "devastating") evaluations that fail to provide
even simple statistics such as the rates of actual pipeline utilization during
their evaluations, in addition to the other shortcomings and shortcuts <a class="link" href="#website-traffic-fingerprinting">mentioned earlier</a>. We can
accept that our defense might fail to work as well as others (in fact we
expect it), but unfortunately the very same shortcuts that provide excellent
attack results also allow the conclusion that all defenses are broken forever.
So sadly, we are still left in the dark on this point.

</p></blockquote></div></div></li><li class="listitem"><span class="command"><strong>Privacy-preserving update notification</strong></span><p>

In order to inform the user when their Tor Browser is out of date, we perform a
privacy-preserving update check asynchronously in the background. The
check uses Tor to download the file <a class="ulink" href="https://check.torproject.org/RecommendedTBBVersions" target="_top">https://check.torproject.org/RecommendedTBBVersions</a>
and searches that version list for the current value for the local preference
<span class="command"><strong>torbrowser.version</strong></span>. If the value from our preference is
present in the recommended version list, the check is considered to have
succeeded and the user is up to date. If not, it is considered to have failed
and an update is needed. The check is triggered upon browser launch, new
window, and new tab, but is rate limited so as to happen no more frequently
than once every 1.5 hours.

</p><p>

If the check fails, we cache this fact, and update the Torbutton graphic to
display a flashing warning icon and insert a menu option that provides a link
to our download page. Additionally, we reset the value for the browser
homepage to point to a <a class="ulink" href="https://check.torproject.org/?lang=en-US&amp;small=1&amp;uptodate=0" target="_top">page that
informs the user</a> that their browser is out of
date.

</p><p>

We also make use of the in-browser Mozilla updater, and have <a class="ulink" href="https://gitweb.torproject.org/tor-browser.git/commit/?h=tor-browser-31.6.0esr-4.5-1&amp;id=bcf51aae541fc28de251924ce9394224bd2b814c" target="_top">patched
the updater</a> to avoid sending OS and Kernel version information as part
of its update pings.

</p></li></ol></div></div></div><div class="sect1"><div class="titlepage"><div><div><h2 class="title" style="clear: both"><a id="BuildSecurity"></a>5. Build Security and Package Integrity</h2></div></div></div><p>

In the age of state-sponsored malware, <a class="ulink" href="https://blog.torproject.org/blog/deterministic-builds-part-one-cyberwar-and-global-compromise" target="_top">we
believe</a> it is impossible to expect to keep a single build machine or
software signing key secure, given the class of adversaries that Tor has to
contend with. For this reason, we have deployed a build system
that allows anyone to use our source code to reproduce byte-for-byte identical
binary packages to the ones that we distribute.

</p><div class="sect2"><div class="titlepage"><div><div><h3 class="title"><a id="idp56215504"></a>5.1. Achieving Binary Reproducibility</h3></div></div></div><p>

The GNU toolchain has been working on providing reproducible builds for some
time, however a large software project such as Firefox typically ends up
embedding a large number of details about the machine it was built on, both
intentionally and inadvertently. Additionally, manual changes to the build
machine configuration can accumulate over time and are difficult for others to
replicate externally, which leads to difficulties with binary reproducibility.

</p><p>
For this reason, we decided to leverage the work done by the <a class="ulink" href="http://gitian.org/" target="_top">Gitian Project</a> from the Bitcoin community.
Gitian is a wrapper around Ubuntu's virtualization tools that allows you to
specify an Ubuntu version, architecture, a set of additional packages, a set
of input files, and a bash build scriptlet in an YAML document called a
"Gitian Descriptor". This document is used to install a qemu-kvm image, and
execute your build scriptlet inside it.
   </p><p>

We have created a <a class="ulink" href="https://gitweb.torproject.org/builders/tor-browser-bundle.git/tree/refs/heads/master" target="_top">set
of wrapper scripts</a> around Gitian to automate dependency download and
authentication, as well as transfer intermediate build outputs between the
stages of the build process. Because Gitian creates an Ubuntu build
environment, we must use cross-compilation to create packages for Windows and
Mac OS. For Windows, we use mingw-w64 as our cross compiler. For Mac OS, we
use crosstools-ng in combination with a binary redistribution of the Mac OS 10.6
SDK.

</p><p>

The use of the Gitian system eliminates build non-determinism by normalizing
the build environment's hostname, username, build path, uname output,
toolchain versions, and time. On top of what Gitian provides, we also had to
address the following additional sources of non-determinism:

</p><div class="orderedlist"><ol class="orderedlist" type="1"><li class="listitem"><span class="command"><strong>Filesystem and archive reordering</strong></span><p>

The most prevalent source of non-determinism in the components of Tor Browser
by far was various ways that archives (such as zip, tar, jar/ja, DMG, and
Firefox manifest lists) could be reordered. Many file archivers walk the
file system in inode structure order by default, which will result in ordering
differences between two different archive invocations, especially on machines
of different disk and hardware configurations.

</p><p>

The fix for this is to perform an additional sorting step on the input list
for archives, but care must be taken to instruct libc and other sorting routines
to use a fixed locale to determine lexicographic ordering, or machines with
different locale settings will produce different sort results. We chose the
'C' locale for this purpose. We created wrapper scripts for <a class="ulink" href="https://gitweb.torproject.org/builders/tor-browser-bundle.git/tree/gitian/build-helpers/dtar.sh" target="_top">tar</a>,
<a class="ulink" href="https://gitweb.torproject.org/builders/tor-browser-bundle.git/tree/gitian/build-helpers/dzip.sh" target="_top">zip</a>,
and <a class="ulink" href="https://gitweb.torproject.org/builders/tor-browser-bundle.git/tree/gitian/build-helpers/ddmg.sh" target="_top">DMG</a>
to aid in reproducible archive creation.

</p></li><li class="listitem"><span class="command"><strong>Uninitialized memory in toolchain/archivers</strong></span><p>

We ran into difficulties with both binutils and the DMG archive script using
uninitialized memory in certain data structures that ended up written to disk.
Our binutils fixes were merged upstream, but the DMG archive fix remains an
<a class="ulink" href="https://gitweb.torproject.org/builders/tor-browser-bundle.git/tree/gitian/patches/libdmg.patch" target="_top">independent
patch</a>.

</p></li><li class="listitem"><span class="command"><strong>Fine-grained timestamps and timezone leaks</strong></span><p>

The standard way of controlling timestamps in Gitian is to use libfaketime,
which hooks time-related library calls to provide a fixed timestamp. However,
due to our use of wine to run py2exe for python-based pluggable transports,
pyc timestamps had to be addressed with an additional <a class="ulink" href="https://gitweb.torproject.org/builders/tor-browser-bundle.git/tree/gitian/build-helpers/pyc-timestamp.sh" target="_top">helper
script</a>. The timezone leaks were addressed by setting the
<span class="command"><strong>TZ</strong></span> environment variable to UTC in our descriptors.

</p></li><li class="listitem"><span class="command"><strong>Deliberately generated entropy</strong></span><p>

In two circumstances, deliberately generated entropy was introduced in various
components of the build process. First, the BuildID Debuginfo identifier
(which associates detached debug files with their corresponding stripped
executables) was introducing entropy from some unknown source. We removed this
header using objcopy invocations in our build scriptlets, and opted to use GNU
DebugLink instead of BuildID for this association.

</p><p>

Second, on Linux, Firefox builds detached signatures of its cryptographic
libraries using a temporary key for FIPS-140 certification. A rather insane
subsection of the FIPS-140 certification standard requires that you distribute
signatures for all of your cryptographic libraries. The Firefox build process
meets this requirement by generating a temporary key, using it to sign the
libraries, and discarding the private portion of that key. Because there are
many other ways to intercept the crypto outside of modifying the actual DLL
images, we opted to simply remove these signature files from distribution.
There simply is no way to verify code integrity on a running system without
both OS and co-processor assistance. Download package signatures make sense of
course, but we handle those another way (as mentioned above).

</p></li><li class="listitem"><span class="command"><strong>LXC-specific leaks</strong></span><p>

Gitian provides an option to use LXC containers instead of full qemu-kvm
virtualization. Unfortunately, these containers can allow additional details
about the host OS to leak. In particular, umask settings as well as the
hostname and Linux kernel version can leak from the host OS into the LXC
container. We addressed umask by setting it explicitly in our Gitian
descriptor scriptlet, and addressed the hostname and kernel version leaks by
directly patching the aspects of the Firefox build process that included this
information into the build. It also turns out that some libraries (in
particular: libgmp) attempt to detect the current CPU to determine which
optimizations to compile in. This CPU type is uniform on our KVM instances,
but differs under LXC. We are also investigating currently
<a class="ulink" href="https://trac.torproject.org/projects/tor/ticket/12240" target="_top">oddities related to
time-based dependency tracking</a> that only appear in LXC containers.

</p></li></ol></div></div><div class="sect2"><div class="titlepage"><div><div><h3 class="title"><a id="idp56237264"></a>5.2. Package Signatures and Verification</h3></div></div></div><p>

The build process generates a single sha256sums.txt file that contains a sorted
list of the SHA-256 hashes of every package produced for that build version. Each
official builder uploads this file and a GPG signature of it to a directory
on a Tor Project's web server. The build scripts have an optional matching
step that downloads these signatures, verifies them, and ensures that the
local builds match this file.

</p><p>

When builds are published officially, the single sha256sums.txt file is
accompanied by a detached GPG signature from each official builder that
produced a matching build. The packages are additionally signed with detached
GPG signatures from an official signing key.

</p><p>

The fact that the entire set of packages for a given version can be
authenticated by a single hash of the sha256sums.txt file will also allow us
to create a number of auxiliary authentication mechanisms for our packages,
beyond just trusting a single offline build machine and a single cryptographic
key's integrity. Interesting examples include providing multiple independent
cryptographic signatures for packages, listing the package hashes in the Tor
consensus, and encoding the package hashes in the Bitcoin blockchain.

</p><p>

The Windows releases are also signed by a hardware token provided by Digicert.
In order to verify package integrity, the signature must be stripped off using
the osslsigncode tool, as described on the <a class="ulink" href="https://www.torproject.org/docs/verifying-signatures.html.en#BuildVerification" target="_top">Signature
Verification</a> page.

</p></div><div class="sect2"><div class="titlepage"><div><div><h3 class="title"><a id="idp56241792"></a>5.3. Anonymous Verification</h3></div></div></div><p>

Due to the fact that bit-identical packages can be produced by anyone, the
security of this build system extends beyond the security of the official
build machines. In fact, it is still possible for build integrity to be
achieved even if all official build machines are compromised.

</p><p>

By default, all tor-specific dependencies and inputs to the build process are
downloaded over Tor, which allows build verifiers to remain anonymous and
hidden. Because of this, any individual can use our anonymity network to
privately download our source code, verify it against public signed, audited,
and mirrored git repositories, and reproduce our builds exactly, without being
subject to targeted attacks. If they notice any differences, they can alert
the public builders/signers, hopefully using a pseudonym or our anonymous
bug tracker account, to avoid revealing the fact that they are a build
verifier.

</p></div><div class="sect2"><div class="titlepage"><div><div><h3 class="title"><a id="update-safety"></a>5.4. Update Safety</h3></div></div></div><p>

We make use of the Firefox updater in order to provide automatic updates to
users. We make use of certificate pinning to ensure that update checks cannot
be tampered with, and we sign the individual MAR update files with an offline
signing key.

</p><p>

The Firefox updater also has code to ensure that it can reliably access the
update server to prevent availability attacks, and complains to the user after 48
hours go by without a successful response from the server. Additionally, we
use Tor's SOCKS username and password isolation to ensure that every new
request to the updater (provided the former got issued more than 10 minutes ago)
traverses a separate circuit, to avoid holdback attacks by exit nodes.

</p></div></div><div class="appendix"><h2 class="title" style="clear: both"><a id="Transparency"></a>A. Towards Transparency in Navigation Tracking</h2><p>

The <a class="link" href="#privacy" title="2.2. Privacy Requirements">privacy properties</a> of Tor Browser are based
upon the assumption that link-click navigation indicates user consent to
tracking between the linking site and the destination site.  While this
definition is sufficient to allow us to eliminate cross-site third party
tracking with only minimal site breakage, it is our long-term goal to further
reduce cross-origin click navigation tracking to mechanisms that are
detectable by attentive users, so they can alert the general public if
cross-origin click navigation tracking is happening where it should not be.

</p><p>

In an ideal world, the mechanisms of tracking that can be employed during a
link click would be limited to the contents of URL parameters and other
properties that are fully visible to the user before they click. However, the
entrenched nature of certain archaic web features make it impossible for us to
achieve this transparency goal by ourselves without substantial site breakage.
So, instead we maintain a <a class="link" href="#deprecate" title="A.1. Deprecation Wishlist">Deprecation
Wishlist</a> of archaic web technologies that are currently being (ab)used
to facilitate federated login and other legitimate click-driven cross-domain
activity but that can one day be replaced with more privacy friendly,
auditable alternatives.

</p><p>

Because the total elimination of side channels during cross-origin navigation
will undoubtedly break federated login as well as destroy ad revenue, we
also describe auditable alternatives and promising web draft standards that would
preserve this functionality while still providing transparency when tracking is
occurring.

</p><div class="sect1"><div class="titlepage"><div><div><h2 class="title" style="clear: both"><a id="deprecate"></a>A.1. Deprecation Wishlist</h2></div></div></div><div class="orderedlist"><ol class="orderedlist" type="1"><li class="listitem"><span class="command"><strong>The Referer Header</strong></span><p>

We haven't disabled or restricted the Referer ourselves because of the
non-trivial number of sites that rely on the Referer header to "authenticate"
image requests and deep-link navigation on their sites. Furthermore, there
seems to be no real privacy benefit to taking this action by itself in a
vacuum, because many sites have begun encoding Referer URL information into
GET parameters when they need it to cross HTTP to HTTPS scheme transitions.
Google's +1 buttons are the best example of this activity.

</p><p>

Because of the availability of these other explicit vectors, we believe the
main risk of the Referer header is through inadvertent and/or covert data
leakage.  In fact, <a class="ulink" href="http://www2.research.att.com/~bala/papers/wosn09.pdf" target="_top">a great deal of
personal data</a> is inadvertently leaked to third parties through the
source URL parameters.

</p><p>

We believe the Referer header should be made explicit, and believe that CSP
2.0 provides a <a class="ulink" href="http://www.w3.org/TR/CSP11/#directive-referrer" target="_top">decent step in this
direction</a>. If a site wishes to transmit its URL to third party content
elements during load or during link-click, it should have to specify this as a
property of the associated HTML tag or CSP policy. With an explicit property
or policy, it would then be possible for the user agent to inform the user if
they are about to click on a link that will transmit Referer information
(perhaps through something as subtle as a different color in the lower toolbar
for the destination URL). This same UI notification can also be used for links
with the <a class="ulink" href="https://developer.mozilla.org/en-US/docs/HTML/Element/a#Attributes" target="_top">"ping"</a>
attribute.

</p></li><li class="listitem"><span class="command"><strong>window.name</strong></span><p>
<a class="ulink" href="https://developer.mozilla.org/En/DOM/Window.name" target="_top">window.name</a> is
a DOM property that for some reason is allowed to retain a persistent value
for the lifespan of a browser tab. It is possible to utilize this property for
<a class="ulink" href="http://www.thomasfrank.se/sessionvars.html" target="_top">identifier
storage</a> during click navigation. This is sometimes used for additional
XSRF protection and federated login.
   </p><p>

It's our opinion that the contents of window.name should not be preserved for
cross-origin navigation, but doing so may break federated login for some sites.

</p></li><li class="listitem"><span class="command"><strong>Javascript link rewriting</strong></span><p>

In general, it should not be possible for onclick handlers to alter the
navigation destination of 'a' tags, silently transform them into POST
requests, or otherwise create situations where a user believes they are
clicking on a link leading to one URL that ends up on another. This
functionality is deceptive and is frequently a vector for malware and phishing
attacks. Unfortunately, many legitimate sites also employ such transparent
link rewriting, and blanket disabling this functionality ourselves will simply
cause Tor Browser to fail to navigate properly on these sites.

</p><p>

Automated cross-origin redirects are one form of this behavior that is
possible for us to <a class="ulink" href="https://trac.torproject.org/projects/tor/ticket/3600" target="_top">address
ourselves</a>, as they are comparatively rare and can be handled with site
permissions.

</p></li></ol></div></div><div class="sect1"><div class="titlepage"><div><div><h2 class="title" style="clear: both"><a id="idp56278768"></a>A.2. Promising Standards</h2></div></div></div><div class="orderedlist"><ol class="orderedlist" type="1"><li class="listitem"><a class="ulink" href="http://web-send.org" target="_top">Web-Send Introducer</a><p>

Web-Send is a browser-based link sharing and federated login widget that is
designed to operate without relying on third-party tracking or abusing other
cross-origin link-click side channels. It has a compelling list of <a class="ulink" href="http://web-send.org/features.html" target="_top">privacy and security features</a>,
especially if used as a "Like button" replacement.

</p></li><li class="listitem"><a class="ulink" href="https://developer.mozilla.org/en-US/docs/Persona" target="_top">Mozilla Persona</a><p>

Mozilla's Persona is designed to provide decentralized, cryptographically
authenticated federated login in a way that does not expose the user to third
party tracking or require browser redirects or side channels. While it does
not directly provide the link sharing capabilities that Web-Send does, it is a
better solution to the privacy issues associated with federated login than
Web-Send is.

</p></li></ol></div></div></div></div></body></html>

tor-webwml.git

Update Tor Browser design doc.