- tor-dev - lists.torproject.org

Re: [tor-dev] Using ScrambleSuit with something other than Tor
by Philipp Winter 11 Jul '14

11 Jul '14

Hi Yuhao, On Fri, Jul 11, 2014 at 08:20:25PM +0800, Yuhao Dong wrote: > I'm Yuhao Dong (undergrad student at University of Waterloo) and I'm > currently doing research on, surprise, my own traffic obfuscation > system. Cool! I'm in Toronto until the middle of August -- in case you are back until then and want to meet up in person. > I'd like to compare it to ScrambleSuit; however I am > currently on break in China. It is going to be really hard to set up > Tor in a way that works; besides, I am using a test suite that > transmits specially-shaped traffic over HTTP. I would like to be > able to see what happens if I apply ScrambleSuit to that traffic > instead of Tor. That sounds good. Given Tor's high variance in latency and throughput, it's more effective to compare the performance of pluggable transports without Tor, in a LAN setting. > Is there a way to build a generic ScrambleSuit SOCKS5 proxy? In > other words, it would be awesome if I could set up two servers on my > LAN: > > web browsing client <-SOCKS-> first LAN server <-ScrambleSuit-> > second LAN server <-> Internet > > so that arbitrary traffic can be tunneled through ScrambleSuit, and > I can do packet analysis on the ScrambleSuit link. > > I understand that Tor's "pluggable transport" system uses SOCKS5 > underneath; however the spec seems very vague on how to precisely > get that local SOCKS5 address. How can I use ScrambleSuit in the way > I described, to encapsulate something other than Tor? It sounds like you basically want to have obfsproxy with ScrambleSuit on both your LAN servers. The first LAN server could be started like this: python /path/to/obfsproxy \ --log-min-severity=debug \ --data-dir=/tmp/scramblesuit-client \ scramblesuit \ --password=BANANASAPPLESCOCONUTSPEACHESEGGS \ --dest SECOND_LAN_SERVER:50000 \ client 0.0.0.0:1080 The second LAN server could be started like this: python /path/to/obfsproxy \ --log-min-severity=debug \ --data-dir=/tmp/scramblesuit-server \ scramblesuit \ --password=BANANASAPPLESCOCONUTSPEACHESEGGS \ --dest SOCKS_PROXY:1234 \ server 0.0.0.0:50000 SOCKS_PROXY can be Tor or an arbitrary SOCKS proxy which allows direct connections to your Internet link. You can then configure arbitrary applications to use the SOCKS5 proxy running on your first LAN server. Hope that helps, Philipp

1 0

SENDME cells
by Gareth Owen 09 Jul '14

09 Jul '14

(sorry re-post - forgot subject) Dear all I'm working on a tor research project and am having difficulty understanding the SENDME cells. The tor spec acknowledges that it isn't particularly clear so I would welcome some clarification please. The spec says that *all* nodes in a circuit maintain a send and receive window, and that this window is decremented on each RELAY_DATA and incremented on each SENDME. Cells that are neither of these do not affect the window size. The problem I have understanding is, that only edge nodes will know whether a cell is a RELAY_DATA, the intermediate nodes only know that its a RELAY but not what type. So, if only RELAY_DATA decrements the window size, and intermediate nodes cannot spot these, what point is there in intermediate nodes having a window? Any help greatly appreciated. Gareth --- Dr Gareth Owen Senior Lecturer School of Computing, University of Portsmouth W: ghowen.me

3 4

(Extra) roadmapping PT meeting this Friday (16:00UTC 11th of July 2014)
by George Kadianakis 08 Jul '14

08 Jul '14

Hello friends, the PT meeting last week was a bit quiet because it was conflicting with the Tor dev meeting. To compensate for that, we are thinking of throwing an extra PT meeting this Friday. The agenda can be modified according to the participants, but we are planning to spend some time writing down the fictional roadmap designed during the dev meeting to a wiki page, so that we can better plan PT development in the months ahead If you are in a roadmapping mood, feel free to pass by and help. As always, the meeting is going to occur at 16:00 UTC. Place is the #tor-dev IRC channel in the OFTC network. Cheers!

1 0

Re: [tor-dev] [HTTPS-Everywhere] [GSoC] HTTPS Everywhere secure ruleset update mechanism update
by Yan Zhu 08 Jul '14

08 Jul '14

(resending to tor-dev with tp.o email address) On 07/08/2014 03:42 AM, Yan Zhu wrote: > On 07/08/2014 12:07 AM, Jeroen Massar wrote: >> On 2014-07-07 20:40, Red wrote: >> [.. lots of cool work being worked on ..] >> >> Hi Zack, >> >> Seems you are doing lots of cool stuff ;) >> >> But I am one of those strange people who really hate it that every >> separate tool has their own updater (which can be used for tracking a >> user, as the set of updater tools polling servers makes a fingerprint in >> the same way other flows make a fingerprint). > > Hi Jeroen, > > This makes a lot of sense. I'm aware of the fingerprintability concern, > and EFF tech projects generally try to mitigate it by polling the update > servers at randomized intervals over fresh Tor circuits if possible. For > this project, we initially proposed polling for an update when the > browser starts and every 3 hours plus some random, evenly-distributed > number of milliseconds between 0 and 300000. I'm curious if others have > more refined suggestions! > >> >> And thus I run Little Snitch and block those updates. Till I deem it a >> good time for the update to be done and trigger it manually. >> >> As such, when you get to the stage of adding features, it would be good >> if there was: >> - an option to disable the auto fetching > > Yes, this would be fairly easy to add. > >> - an option to trigger the fetching > > Probably also easy. > >> - to feed the update mechanism with a pre-fetched file >> (eg provided through a different update mechanism) > > Since the update mechanism is just an XHR that downloads a new ruleset > library from a hardcoded static URL and replaces the existing one in the > Firefox profile directory, you could fetch-and-replace this manually via > any number of mechanisms. :) > > Also, the ruleset libraries will still ship with extension updates, so > you could disable ruleset updates and just wait for the next HTTPS > Everywhere release. > > -Yan > >> >> Greets, >> Jeroen >> >> _______________________________________________ >> tor-dev mailing list >> tor-dev(a)lists.torproject.org >> https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-dev >> > > > > > _______________________________________________ > HTTPS-Everywhere mailing list > HTTPS-Everywhere(a)lists.eff.org > https://lists.eff.org/mailman/listinfo/https-everywhere > -- Yan Zhu <yan(a)eff.org>, <yan(a)torproject.org> Staff Technologist Electronic Frontier Foundation https://www.eff.org 815 Eddy Street, San Francisco, CA 94109 +1 415 436 9333 x134

2 1

Re: [tor-dev] [HTTPS-Everywhere] [GSoC] HTTPS Everywhere secure ruleset update mechanism update
by Yan Zhu 08 Jul '14

08 Jul '14

(resending to tor-dev with tp.o email address) On 07/08/2014 03:30 AM, Yan Zhu wrote: > On 07/08/2014 02:55 AM, Ben Laurie wrote: >> On 7 July 2014 19:40, Red <redwire(a)riseup.net> wrote: >>> Despite the fact that the process for producing the signature in >>> question[2] seemed to work fine- Openssl was able to generate and verify >>> the signature, the testing code calling the verifyData[3] function used >>> for verification was returning an undocumented NS_ERROR_FAILURE >>> exception. I had spent a great deal of time asking for support in >>> relevant Firefox extension development IRC channels, reading source code >>> from unit tests for the nsIDataSignatureVerifier component, and >>> experimenting with alternative openssl commands in order to try to >>> figure out why this error was occurring. >> >> Looking at the pk1sign source, it looks like the signature needs to be >> in base64. Was that what you were using? >> >> Do you have a test case that fails using command line tools? > > I think Zack's original failing test case was generated via something like: > $ openssl rsautl -sign -in update.digest -out signtmp.sig -inkey privkey.pem > $ openssl base64 -in signtmp.sig -out update.json.sig > > as described in the original spec that we wrote: > https://github.com/redwire/https-everywhere/blob/makeJSONManifest/doc/updat… > > Here is the diff between the failing test and the passing test: > https://github.com/redwire/https-everywhere/commit/8b3c85d9d90d679e8b699701…. > I generated the data for the passing test with pk1sign. > > The documentation for nsIDataSignatureVerifier does not really describe > the expected data format for the signature [1], so it took a while to > figure out that it expects a very specialized form [2]. > > [1] > https://developer.mozilla.org/en-US/docs/Mozilla/Tech/XPCOM/Reference/Inter… > [2] https://bugzilla.mozilla.org/show_bug.cgi?id=685852#c0 > > >> _______________________________________________ >> tor-dev mailing list >> tor-dev(a)lists.torproject.org >> https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-dev >> > > > > > _______________________________________________ > HTTPS-Everywhere mailing list > HTTPS-Everywhere(a)lists.eff.org > https://lists.eff.org/mailman/listinfo/https-everywhere > -- Yan Zhu <yan(a)eff.org>, <yan(a)torproject.org> Staff Technologist Electronic Frontier Foundation https://www.eff.org 815 Eddy Street, San Francisco, CA 94109 +1 415 436 9333 x134

1 0

[GSoC] HTTPS Everywhere secure ruleset update mechanism update
by Red 08 Jul '14

08 Jul '14

Hello everyone, I would like to quickly summarize what has been going on for the last couple of weeks of my work. Due to my mentor, Yan, being away on travel, we haven't had time to have a formal weekly meeting on IRC as per usual. We have, however, been emailing on the HTTPS Everywhere mailing list frequently to discuss some of the problems I've been working on last week. We have also compiled a list of things I am to work on for the next week, which I will summarize here. For the past week, I have been grappling with a problem pertaining to digital signature verification using the nsIDataSignatureVerifier XPCOM component designed to handle this task. I had written a few tests[1] that would use the testing mechanism that Yan built to make sure that the hashing and signature verification my project would rely on for security was working. The testing mechanism was built after our last meeting the Friday before last, during which time we realized we could write a separate Firefox extension using the addon SDK (and thus, Jetpack's testing suite) and import the HTTPS Everywhere component into that extension for testing purposes. Despite the fact that the process for producing the signature in question[2] seemed to work fine- Openssl was able to generate and verify the signature, the testing code calling the verifyData[3] function used for verification was returning an undocumented NS_ERROR_FAILURE exception. I had spent a great deal of time asking for support in relevant Firefox extension development IRC channels, reading source code from unit tests for the nsIDataSignatureVerifier component, and experimenting with alternative openssl commands in order to try to figure out why this error was occurring. Yan was able to get the test to pass by generating a key and signature using the NSS tools. However, she has said that this process is more involved than we would like, and is not likely feasible to accomplish on EFF's airgapped machine hosting its offline private signing key due to the lack of availability of NSS tools. To overcome this limitation, I will be porting one of either the Uhura tool that has been used in the past, or the pk1sign program[4] Yan had found referenced in a Bugzilla report. For now, that's what I'll be doing. I am feeling optimistic that, once this issue of generating an appropriate (i.e. verifiable by nsIDataSignatureVerifier) signature is resolved, writing tests for and refactoring the secure update mechanism I am building will be complete before long. I have compiled some of the discussion Yan and I have had via email into my weekly meeting notes[5], despite there having not been an actual meeting. As usual, I welcome any advice and input! Cheers, Zack [1]: https://github.com/redwire/https-everywhere/blob/feature/tests/https-everyw… [2]: https://github.com/redwire/https-everywhere/blob/rulesetUpdating/doc/update… [3]: https://developer.mozilla.org/en-US/docs/Mozilla/Tech/XPCOM/Reference/Inter… [4]: http://dxr.mozilla.org/mozilla-central/source/security/nss/cmd/pk1sign/pk1s… [5]: https://gist.github.com/redwire/b62f03905a826e79947a#week-8

3 2

[PATCH] entrynodes.c: Improve readability (issue 9971)
by Martin Kepplinger 08 Jul '14

08 Jul '14

Maybe we can do away with issue 9971 and improve readability: https://trac.torproject.org/projects/tor/ticket/9971 As response to this mail, I'll append seperate patches for the 3 sub-issues mentioned. They apply (and are build-tested) seperately to the current tree, so a maintainer can quickly pick one of them up, if happy with it. There shouldn't be a conflict in *1* and *2* but I could append a patchset anyway. sub-issue *3* is *only* a call for advice and will be written on top of the other ones, after discussion. 1* rename entry_guard_t's made_contact to used_so_save_if_down I think that's readable. Is that about what arma had in mind? 2* rename for_discovery argument of add_an_entry_guard() to I like probationary more than provisional. Those 2 are suggested in issue 9971. I chose forced_probationary for now, because isn't it strictly a suboptimal situation in terms of desired 'grade of anonymity'. What do you think? 3* NEEDS REVIEW FIRST: regarding the int arguments of add_an_entry_guard(). I look at: node_t *chosen is a node to add. prepend is set if the guard should become first in the list? there are 2 users of add_an_entry_guard() that pass it a chosen node. One is a bridge (prepend) and the other one is a user-defined node (!prepend), so: Given the fact that the list is not supposed to be long and the two 'users' are somewhat similar, why not prepend the node if explicitly given?

1 4

[PATCH] remove test for remote SOCKS Port at config validation
by Martin Kepplinger 08 Jul '14

08 Jul '14

(why doesn't git send-email work for this list?) this is UNTESTED and for now, more of a question if this may suffice to close ticket 4019: -- Martin Kepplinger http://martinkepplinger.com

1 1

[ot] tor2web static maps
by grarpamp 07 Jul '14

07 Jul '14

I think I saw these posted here earlier from a tor2web person. The url's seem unreliable at times. Are there new, updated, or canonical links to this data? Or to a document outlining them? Thanks. https://antani.tor2web.org/antanistaticmap/stats/yesterday https://antani.onion.to/antanistaticmap/stats/yesterday

2 2

[PATCH] Add some words about meek in proposal 203
by Lunar 07 Jul '14

07 Jul '14

This is mostly David Fifield's words from an email exchange. --- I re-read proposal 203 the other day and wondered how it was related to the meek pluggable transport. As I might not be the only one, I thought it could be worthwhile to share David's answer. Feel free to improve! proposals/203-https-frontend.txt | 28 ++++++++++++++++++++++++++++ 1 file changed, 28 insertions(+) diff --git a/proposals/203-https-frontend.txt b/proposals/203-https-frontend.txt index 26101b3..df30cd5 100644 --- a/proposals/203-https-frontend.txt +++ b/proposals/203-https-frontend.txt @@ -245,3 +245,31 @@ Side note: What to put on the webserver? "Something to add to your HTTPS website" rather than as a standalone installation. +Related work: + + meek [1] is a pluggable transport that uses HTTP for carrying bytes + and TLS for obfuscation. Traffic is relayed through a third-party + server (Google App Engine). It uses a trick to talk to the third + party so that it looks like it is talking to an unblocked server. + + meek itself is not really about HTTP at all. It uses HTTP only + because it's convenient and the big Internet services we use as cover + also use HTTP. meek uses HTTP as a transport, and TLS for + obfuscation, but the key idea is really "domain fronting," where it + appears to the censor you are talking to one domain (www.google.com) + but behind the scenes you are talking to another + (meek-reflect.appspot.com) The meek-server program is an ordinary + HTTP (not necessarily even HTTPS!) server, whose communication is + easily fingerprintable; but that doesn't matter because the censor + never sees that part of the communication, only the communication + between the client and CDN. + + One way to think about the difference: if a censor (somehow) learns + the IP address of a bridge as described in this proposal, it's easy + and low-cost for the censor to block that bridge by IP address. meek + aims to make it much more expensive: even if you know a domain is + being used (in part) for circumvention, in order to block it have to + block something important like the Google frontend or CloudFlare + (high collateral damage). + +1. https://trac.torproject.org/projects/tor/wiki/doc/meek -- 1.7.10.4

3 3