December 2010 Progress Report

by phobos | January 11, 2011

New Releases

  • On December 14, we released updated Tor Browser Bundles for Windows, OSX, and
    Linux, https://vbdvexcmqi.oedi.net/blog/new-tor-browser-bundle-packages-0
  • On December 17th, we released an updated -stable version of Tor, 0.2.1.28.
    Tor 0.2.1.28 does some code cleanup to reduce the risk of remotely exploitable
    bugs. Thanks to Willem Pinckaers for notifying us of the issue. The Common
    Vulnerabilities and Exposures project has assigned CVE-2010-1676 to this issue.
    We also took this opportunity to change the IP address for one of our directory
    authorities, and to update the geoip database we ship.
    https://vbdvexcmqi.oedi.net/blog/tor-02128-released-security-patches
  • On December 17th, we released an updated -alpha version of Tor,
    0.2.2.20-alpha. Tor 0.2.2.20-alpha does some code cleanup to reduce the risk of remotely exploitable bugs. We also fix a variety of other significant bugs, change the IP address for one of our directory authorities, and update the minimum version
    that Tor relays must run to join the network. https://vbdvexcmqi.oedi.net/blog/tor-02220-alpha-out-security-patches

Enhancements that make Tor a better tool for users in censored
countries.

  • Mike spent the past month and a half primarily working on preparing
    Torbutton for Firefox 4. This was a rather difficult task, as a lot has changed
    in this release, and the Javascript debugger doesn't yet support Firefox 4. The
    new mechanism works just fine for replacing XPCOM components. He also took the
    opportunity to clean up the code a bit. Firefox 3.5 and Firefox 4 both added
    some new APIs that make our job easier. We no longer rely so heavily on
    reimplementing pieces of Firefox using XPCOM re-registration. In fact, the only
    component we still need to hook is the external app launcher, to provide our
    warning message.
  • Mike reviewed a proposed Chrome API at: https://groups.google.com/a/chromium.org/group/chromium-dev/browse_thre…. The main goal there was to try to steer them away from declarative models and more towards blocking callback APIs that will work better for us. We'll see if it works.

Blocking Resistance

Steven worked on plans of how to make Tor look more like other protocols, for
when Tor's HTTPS-like cloaking is broken or someone blocks Tor along with HTTPS.
The proposal is to allow Tor bridges to be configured to use one or more plugins
which offer translation between Tor and obfuscated-Tor. There is now a proposal
draft here:

https://gitweb.torproject.org/tor.git/blob/HEAD:/doc/spec/proposals/ide…. Steven wrote a first proof-of-concept (albeit not compliant with the proposal), for a HTTP-like protocol, and put the code here: https://gitweb.torproject.org/sjm217/pluggable-transport.git. A screenshot of the a Wireshark dump of Steven successfully accessing check.torproject.org is at: http://www.cl.cam.ac.uk/~sjm217/volatile/pluggable-transport.png.

Outreach and Advocacy

Bridge relay and bridge authority work
Jacob and Runa continue work on making it easy for people to become Tor bridge
relays by default. The Torouter project is very much an alpha code quality project, but making progress on OpenWRT-based wireless access points and the Excito B3 hardware. The project page is kept up-to-date at https://trac.torproject.org/projects/tor/wiki/TheOnionRouter/Torouter.

Scalability

  • Karsten wrote a technical report with an "Overview of Statistical Data in the Tor Network". The idea is to give this report to interested researchers who want to know what statistical data we have. The report is available at https://metrics.torproject.org/papers/data-2010-12-29.pdf.
  • Erinn and Sebastian worked on some Thandy & Hudson hacking accomplishing:
    1) Hudson and Windows finally cooperate, so now we can have a Windows autobuilder; and 2) setup a basic instance of Thandy-the secure software update platform.
  • Mike did a quick modification to the TorFlow statsplitter.py script to have it output the results from the extra-info descriptors to give us breakdowns on port stats in more readable percentages, to compare the default exit policy of blutmagie to Amunet. It looks like the default exit policy causes you to write a heck of a lot more data, and over half of the read data is misc ports. Using these stats to produce new consensus weights to account for this seems like a good task to do.

Translations
We fully migrated from our own Pootle-based translation system to Transifex, https://www.transifex.net/projects/p/torproject/. A number of translations for various products have already come through for German, Arabic, Burmese, Simplified Chinese, Dutch, Finnish, French, Indonesian, Italian, Norwegian, Persian, Polish, and Romanian.

Comments

Comments are closed.