tor-webwml.git

en/volunteer.wml   1) ## translation metadata

en/volunteer.wml   2) # Revision: $Revision$

volunteer.html     3) 

en/volunteer.wml   4) #include "head.wmi" TITLE="Volunteer"

volunteer.html     5) 
volunteer.html     6) <div class="main-column">
volunteer.html     7) 
volunteer.html     8) <!-- PUT CONTENT AFTER THIS TAG -->

en/volunteer.wml   9) <h2>Three things everyone can do now:</h2>

volunteer.html    10) <ol>

en/volunteer.wml  11) <li>Please consider <a href="<page docs/tor-doc-server>">running

volunteer.html    12) a server</a> to help the Tor network grow.</li>

en/volunteer.wml  13) <li>Tell your friends! Get them to run servers. Get them to run hidden

volunteer.html    14) services. Get them to tell their friends.</li>

en/volunteer.wml  15) <li>We are looking for funding and sponsors. If you like Tor's goals, please

en/volunteer.wml  16)   <a href="<page donate>">take a moment to donate to support further
en/volunteer.wml  17)   Tor development</a>. Also, if you know any

en/volunteer.wml  18)   companies, NGOs, agencies, or other organizations that want communications

en/volunteer.wml  19)   security, let them know about us.</li>

volunteer.html    20) </ol>
volunteer.html    21) 

en/volunteer.wml  22) <a id="Usability"></a>

en/volunteer.wml  23) <h2><a class="anchor" href="#Usability">Supporting Applications</a></h2>

volunteer.html    24) <ol>

en/volunteer.wml  25) <li>We need good ways to intercept DNS requests so they don't "leak" their
en/volunteer.wml  26) request to a local observer while we're trying to be anonymous. (This
en/volunteer.wml  27) happens because the application does the DNS resolve before going to
en/volunteer.wml  28) the SOCKS proxy.)</li>
en/volunteer.wml  29) <ul>
en/volunteer.wml  30) <li>We need to <a
en/volunteer.wml  31) href="http://wiki.noreply.org/noreply/TheOnionRouter/TSocksPatches">apply
en/volunteer.wml  32) all our tsocks patches</a> and maintain a new fork. We'll host it if
en/volunteer.wml  33) you want.</li>
en/volunteer.wml  34) <li>We should patch Dug Song's "dsocks" program to use Tor's
en/volunteer.wml  35) <i>mapaddress</i> commands from the controller interface, so we
en/volunteer.wml  36) don't waste a whole round-trip inside Tor doing the resolve before
en/volunteer.wml  37) connecting.</li>
en/volunteer.wml  38) <li>We need to make our <i>torify</i> script detect which of tsocks or
en/volunteer.wml  39) dsocks is installed, and call them appropriately. This probably means
en/volunteer.wml  40) unifying their interfaces, and might involve sharing code between them
en/volunteer.wml  41) or discarding one entirely.</li>
en/volunteer.wml  42) </ul>

volunteer.html    43) <li>People running servers tell us they want to have one BandwidthRate
volunteer.html    44) during some part of the day, and a different BandwidthRate at other parts
volunteer.html    45) of the day. Rather than coding this inside Tor, we should have a little

en/volunteer.wml  46) script that speaks via the <a href="<page gui/index>">Tor Controller Interface</a>,

volunteer.html    47) and does a setconf to change the bandwidth rate. Perhaps it would run out
volunteer.html    48) of cron, or perhaps it would sleep until appropriate times and then do
volunteer.html    49) its tweak (that's probably more portable). Can somebody write one for us

en/volunteer.wml  50) and we'll put it into <a href="<svnsandbox>contrib/">contrib/</a>?

en/volunteer.wml  51) This is a good entry for the <a href="<page gui/index>">Tor GUI

en/volunteer.wml  52) competition</a>.
en/volunteer.wml  53)   <!-- We have a good script to adjust stuff now, right? -NM -->
en/volunteer.wml  54) </li>

en/volunteer.wml  55) <li>Tor can <a

volunteer.html    56) href="http://wiki.noreply.org/noreply/TheOnionRouter/TorFAQ#ChooseEntryExit">exit

en/volunteer.wml  57) the Tor network from a particular exit node</a>, but we should be able
en/volunteer.wml  58) to specify just a country and have something automatically pick. The
en/volunteer.wml  59) best bet is to fetch Blossom's directory also, and run a local Blossom
en/volunteer.wml  60) client that fetches this directory securely (via Tor and checking its
en/volunteer.wml  61) signature), intercepts <tt>.country.blossom</tt> hostnames, and does
en/volunteer.wml  62) the right thing.</li>

volunteer.html    63) <li>Speaking of geolocation data, somebody should draw a map of the Earth
volunteer.html    64) with a pin-point for each Tor server. Bonus points if it updates as the

en/volunteer.wml  65) network grows and changes. Unfortunately, the easy ways to do this involve
en/volunteer.wml  66) sending all the data to Google and having them draw the map for you. How
en/volunteer.wml  67) much does this impact privacy, and do we have any other good options?</li>

volunteer.html    68) </ol>
volunteer.html    69) 

en/volunteer.wml  70) <a id="Documentation"></a>
en/volunteer.wml  71) <h2><a class="anchor" href="#Documentation">Documentation</a></h2>

volunteer.html    72) <ol>

en/volunteer.wml  73) <li>We hear that Tor users can fall victim to anonymity-breaking attacks
en/volunteer.wml  74) from javascript, java, activex, flash, etc, if they don't disable
en/volunteer.wml  75) them. Are there plugins out there (like NoScript for Firefox) that make
en/volunteer.wml  76) it easier for users to manage this risk? What is the risk exactly?</li>
en/volunteer.wml  77) <li>Is there a full suite of plugins that will replace all of Privoxy's
en/volunteer.wml  78) functionality for Firefox 1.5+? We hear Tor is much faster when you take
en/volunteer.wml  79) Privoxy out of the loop.</li>
en/volunteer.wml  80) <li>Please help Matt Edman with the documentation and how-tos for his

en/volunteer.wml  81) Tor controller,
en/volunteer.wml  82) <a href="http://vidalia-project.net/">Vidalia</a>.</li>

en/volunteer.wml  83) <li>Evaluate and document
en/volunteer.wml  84) <a href="http://wiki.noreply.org/wiki/TheOnionRouter/TorifyHOWTO">our
en/volunteer.wml  85) list of programs</a> that can be configured to use Tor.</li>

volunteer.html    86) <li>We need better documentation for dynamically intercepting

en/volunteer.wml  87) connections and sending them through Tor. tsocks (Linux), dsocks (BSD),

en/volunteer.wml  88) and freecap (Windows) seem to be good candidates, as would better
en/volunteer.wml  89) use of our new TransPort feature.</li>

en/volunteer.wml  90) <li>We have a huge list of <a href="http://wiki.noreply.org/noreply/TheOnionRouter/SupportPrograms">potentially useful

volunteer.html    91) programs that interface to Tor</a>. Which ones are useful in which
volunteer.html    92) situations? Please help us test them out and document your results.</li>

en/volunteer.wml  93) <li>Help translate the web page and documentation into other
en/volunteer.wml  94) languages. See the <a href="<page translation>">translation
en/volunteer.wml  95) guidelines</a> if you want to help out. We also need people to help
en/volunteer.wml  96) maintain the existing Italian, French, and Swedish translations -
en/volunteer.wml  97) see the <a href="<page translation-status>">translation status
en/volunteer.wml  98) overview</a>.</li>

en/volunteer.wml  99) <li>Can somebody walk us through whether we want to go the
en/volunteer.wml 100) <a href="http://www.cacert.org/">cacert</a> route for our
en/volunteer.wml 101) SSL <a href="<page documentation>#Developers">SVN repository?</a></li>

volunteer.html   102) </ol>
volunteer.html   103) 

en/volunteer.wml 104) <a id="Coding"></a>
en/volunteer.wml 105) <h2><a class="anchor" href="#Coding">Coding and Design</a></h2>

volunteer.html   106) <ol>

en/volunteer.wml 107) <li>Tor servers don't work well on Windows XP. On
en/volunteer.wml 108) Windows, Tor uses the standard <tt>select</tt> system
en/volunteer.wml 109) call, which uses space in the non-page pool. This means
en/volunteer.wml 110) that a medium sized Tor server will empty the non-page pool, <a
en/volunteer.wml 111) href="http://wiki.noreply.org/noreply/TheOnionRouter/WindowsBufferProblems">causing

en/volunteer.wml 112) havoc and crashes</a>. We should probably be using overlapped IO

en/volunteer.wml 113) instead. One solution would be to teach libevent how to use overlapped IO
en/volunteer.wml 114) rather than select() on Windows, and then adapt Tor to the new libevent
en/volunteer.wml 115) interface.</li>
en/volunteer.wml 116) <li>Because Tor servers need to store-and-forward each cell they handle,
en/volunteer.wml 117) high-bandwidth Tor servers end up using dozens of megabytes of memory
en/volunteer.wml 118) just for buffers. We need better heuristics for when to shrink/expand
en/volunteer.wml 119) buffers. Maybe this should be modelled after the Linux kernel buffer

en/volunteer.wml 120) design, where we have many smaller buffers that link to each other,

en/volunteer.wml 121) rather than monolithic buffers?</li>
en/volunteer.wml 122) <li>We need an official central site to answer "Is this IP address a Tor
en/volunteer.wml 123) server?" questions. This should provide several interfaces, including
en/volunteer.wml 124) a web interface and a DNSBL-style interface. It can provide the most
en/volunteer.wml 125) up-to-date answers by keeping a local mirror of the Tor directory
en/volunteer.wml 126) information. Bonus points if it does active testing through each exit
en/volunteer.wml 127) node to find out what IP address it's really exiting from.</li>

en/volunteer.wml 128) <li>We need a measurement study of <a
en/volunteer.wml 129) href="http://www.pps.jussieu.fr/~jch/software/polipo/">Polipo</a>
en/volunteer.wml 130) vs <a href="http://www.privoxy.org/">Privoxy</a>. Is Polipo in fact
en/volunteer.wml 131) significantly faster, once you factor in the slow-down from Tor? Are the
en/volunteer.wml 132) results the same on both Linux and Windows? Related, does Polipo handle
en/volunteer.wml 133) more web sites correctly than Privoxy, or vice versa? Are there stability
en/volunteer.wml 134) issues on any common platforms, e.g. Windows?</li>
en/volunteer.wml 135) <li>It would be great to have a LiveCD that includes the latest
en/volunteer.wml 136) versions of Tor, Polipo or Privoxy, Firefox, Gaim+OTR, etc. There are
en/volunteer.wml 137) two challenges here: first is documenting the system and choices well
en/volunteer.wml 138) enough that security people can form an opinion on whether it should be
en/volunteer.wml 139) secure, and the second is figuring out how to make it easily maintainable,
en/volunteer.wml 140) so it doesn't become quickly obsolete like AnonymOS. Bonus points if
en/volunteer.wml 141) the CD image fits on one of those small-form-factor CDs.</li>
en/volunteer.wml 142) <li>We need a distributed testing framework. We have unit tests,

en/volunteer.wml 143) but it would be great to have a script that starts up a Tor network, uses
en/volunteer.wml 144) it for a while, and verifies that at least parts of it are working.</li>

en/volunteer.wml 145) <li>Right now the hidden service descriptors are being stored on just a
en/volunteer.wml 146) few directory servers. This is bad for privacy and bad for robustness. To
en/volunteer.wml 147) get more robustness, we're going to need to make hidden service
en/volunteer.wml 148) descriptors even less private because we're going to have to mirror them
en/volunteer.wml 149) onto many places. Ideally we'd like to separate the storage/lookup system
en/volunteer.wml 150) from the Tor directory servers entirely. The first problem is that we need
en/volunteer.wml 151) to design a new hidden service descriptor format to a) be ascii rather
en/volunteer.wml 152) than binary for convenience; b) keep the list of introduction points
en/volunteer.wml 153) encrypted unless you know the <tt>.onion</tt> address, so the directory
en/volunteer.wml 154) can't learn them; and c) allow the directories to verify the timestamp
en/volunteer.wml 155) and signature on a hidden service descriptor so they can't be tricked
en/volunteer.wml 156) into giving out fake ones. Second, any reliable distributed storage
en/volunteer.wml 157) system will do, as long as it allows authenticated updates, but as far
en/volunteer.wml 158) as we know no implemented DHT code supports authenticated updates.</li>

en/volunteer.wml 159) <li>Tor 0.1.1.x and later include support for hardware crypto accelerators
en/volunteer.wml 160) via

volunteer.html   161) OpenSSL. Nobody has ever tested it, though. Does somebody want to get
volunteer.html   162) a card and let us know how it goes?</li>

volunteer.html   163) <li>Perform a security analysis of Tor with <a
volunteer.html   164) href="http://en.wikipedia.org/wiki/Fuzz_testing">"fuzz"</a>. Determine

en/volunteer.wml 165) if there are good fuzzing libraries out there for what we want. Win fame by

volunteer.html   166) getting credit when we put out a new release because of you!</li>

volunteer.html   167) <li>Tor uses TCP for transport and TLS for link
volunteer.html   168) encryption. This is nice and simple, but it means all cells
volunteer.html   169) on a link are delayed when a single packet gets dropped, and
volunteer.html   170) it means we can only reasonably support TCP streams. We have a <a
volunteer.html   171) href="http://wiki.noreply.org/noreply/TheOnionRouter/TorFAQ#TransportIPnotTCP">list

en/volunteer.wml 172) of reasons why we haven't shifted to UDP transport</a>, but it would
en/volunteer.wml 173) be great to see that list get shorter. We also have a proposed <a

en/volunteer.wml 174) href="<svnsandbox>doc/tor-spec-udp.txt">specification for Tor and

en/volunteer.wml 175) UDP</a> &mdash; please let us know what's wrong with it.</li>

volunteer.html   176) <li>We're not that far from having IPv6 support for destination addresses
volunteer.html   177) (at exit nodes). If you care strongly about IPv6, that's probably the
volunteer.html   178) first place to start.</li>

volunteer.html   179) </ol>
volunteer.html   180) 

en/volunteer.wml 181) <a id="Research"></a>
en/volunteer.wml 182) <h2><a class="anchor" href="#Research">Research</a></h2>

volunteer.html   183) <ol>

volunteer.html   184) <li>The "website fingerprinting attack": make a list of a few
volunteer.html   185) hundred popular websites, download their pages, and make a set of
volunteer.html   186) "signatures" for each site. Then observe a Tor client's traffic. As
volunteer.html   187) you watch him receive data, you quickly approach a guess about which
volunteer.html   188) (if any) of those sites he is visiting. First, how effective is
volunteer.html   189) this attack on the deployed Tor codebase? Then start exploring
volunteer.html   190) defenses: for example, we could change Tor's cell size from 512
volunteer.html   191) bytes to 1024 bytes, we could employ padding techniques like <a
volunteer.html   192) href="http://freehaven.net/anonbib/#timing-fc2004">defensive dropping</a>,
volunteer.html   193) or we could add traffic delays. How much of an impact do these have,
volunteer.html   194) and how much usability impact (using some suitable metric) is there from
volunteer.html   195) a successful defense in each case?</li>

volunteer.html   196) <li>The "end-to-end traffic confirmation attack":
volunteer.html   197) by watching traffic at Alice and at Bob, we can <a
volunteer.html   198) href="http://freehaven.net/anonbib/#danezis:pet2004">compare
volunteer.html   199) traffic signatures and become convinced that we're watching the same
volunteer.html   200) stream</a>. So far Tor accepts this as a fact of life and assumes this
volunteer.html   201) attack is trivial in all cases. First of all, is that actually true? How
volunteer.html   202) much traffic of what sort of distribution is needed before the adversary
volunteer.html   203) is confident he has won? Are there scenarios (e.g. not transmitting much)
volunteer.html   204) that slow down the attack? Do some traffic padding or traffic shaping
volunteer.html   205) schemes work better than others?</li>

volunteer.html   206) <li>The "routing zones attack": most of the literature thinks of
volunteer.html   207) the network path between Alice and her entry node (and between the
volunteer.html   208) exit node and Bob) as a single link on some graph. In practice,
volunteer.html   209) though, the path traverses many autonomous systems (ASes), and <a
volunteer.html   210) href="http://freehaven.net/anonbib/#feamster:wpes2004">it's not uncommon
volunteer.html   211) that the same AS appears on both the entry path and the exit path</a>.
volunteer.html   212) Unfortunately, to accurately predict whether a given Alice, entry,
volunteer.html   213) exit, Bob quad will be dangerous, we need to download an entire Internet
volunteer.html   214) routing zone and perform expensive operations on it. Are there practical
volunteer.html   215) approximations, such as avoiding IP addresses in the same /8 network?</li>

volunteer.html   216) <li>Tor doesn't work very well when servers have asymmetric bandwidth
volunteer.html   217) (e.g. cable or DSL). Because Tor has separate TCP connections between
volunteer.html   218) each hop, if the incoming bytes are arriving just fine and the outgoing
volunteer.html   219) bytes are all getting dropped on the floor, the TCP push-back mechanisms
volunteer.html   220) don't really transmit this information back to the incoming streams.
volunteer.html   221) Perhaps Tor should detect when it's dropping a lot of outgoing packets,
volunteer.html   222) and rate-limit incoming streams to regulate this itself? I can imagine
volunteer.html   223) a build-up and drop-off scheme where we pick a conservative rate-limit,
volunteer.html   224) slowly increase it until we get lost packets, back off, repeat. We
volunteer.html   225) need somebody who's good with networks to simulate this and help design
volunteer.html   226) solutions; and/or we need to understand the extent of the performance
volunteer.html   227) degradation, and use this as motivation to reconsider UDP transport.</li>
volunteer.html   228) <li>A related topic is congestion control. Is our
volunteer.html   229) current design sufficient once we have heavy use? Maybe
volunteer.html   230) we should experiment with variable-sized windows rather
volunteer.html   231) than fixed-size windows? That seemed to go well in an <a
volunteer.html   232) href="http://www.psc.edu/networking/projects/hpn-ssh/theory.php">ssh
volunteer.html   233) throughput experiment</a>. We'll need to measure and tweak, and maybe
volunteer.html   234) overhaul if the results are good.</li>

volunteer.html   235) <li>To let dissidents in remote countries use Tor without being blocked
volunteer.html   236) at their country's firewall, we need a way to get tens of thousands of
volunteer.html   237) relays, not just a few hundred. We can imagine a Tor client GUI that
volunteer.html   238) has a "help China" button at the top that opens a port and relays a
volunteer.html   239) few KB/s of traffic into the Tor network. (A few KB/s shouldn't be too
volunteer.html   240) much hassle, and there are few abuse issues since they're not being exit
volunteer.html   241) nodes.) But how do we distribute a list of these volunteer clients to the
volunteer.html   242) good dissidents in an automated way that doesn't let the country-level
volunteer.html   243) firewalls intercept and enumerate them? Probably needs to work on a

en/volunteer.wml 244) human-trust level. See our <a
en/volunteer.wml 245) href="http://wiki.noreply.org/noreply/TheOnionRouter/TorFAQ#China">FAQ
en/volunteer.wml 246) entry</a> on this, and then read the <a
en/volunteer.wml 247) href="http://freehaven.net/anonbib/topic.html#Communications_20Censorship">censorship
en/volunteer.wml 248) resistance section of anonbib</a>.</li>

volunteer.html   249) <li>Tor circuits are built one hop at a time, so in theory we have the
volunteer.html   250) ability to make some streams exit from the second hop, some from the
volunteer.html   251) third, and so on. This seems nice because it breaks up the set of exiting
volunteer.html   252) streams that a given server can see. But if we want each stream to be safe,
volunteer.html   253) the "shortest" path should be at least 3 hops long by our current logic, so
volunteer.html   254) the rest will be even longer. We need to examine this performance / security
volunteer.html   255) tradeoff.</li>
volunteer.html   256) <li>It's not that hard to DoS Tor servers or dirservers. Are client
volunteer.html   257) puzzles the right answer? What other practical approaches are there? Bonus
volunteer.html   258) if they're backward-compatible with the current Tor protocol.</li>

volunteer.html   259) </ol>
volunteer.html   260) 

en/volunteer.wml 261) <a href="<page contact>">Let us know</a> if you've made progress on any
en/volunteer.wml 262) of these!

volunteer.html   263) 

en/volunteer.wml 264)   </div><!-- #main -->

volunteer.html   265) 

en/volunteer.wml 266) #include <foot.wmi>