19:00:31 <GeKo> #startmeeting tor-browser 19:00:31 <MeetBot> Meeting started Mon Dec 19 19:00:31 2016 UTC. The chair is GeKo. Information about MeetBot at http://wiki.debian.org/MeetBot. 19:00:31 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic. 19:00:34 <GeKo> hi all! 19:00:38 <boklm> hi 19:00:43 <isabela> o/ 19:00:46 <GeKo> welcome to the last meeting for this year! 19:01:01 <GeKo> i guess we start as usual with the updates. 19:01:07 <GeKo> who wants to go first today? 19:01:53 * boklm can go first 19:01:57 <GeKo> go! 19:02:14 <boklm> This past week I was at the reproducible build summit, and I worked on #17380. 19:02:21 <boklm> This week I'm planning to write a short blog post about the reproducible build summit (similar to the one from last year), and continue working on #17380. 19:02:32 <boklm> I'm also planning to take some days off 19:02:41 <boklm> That's it for me 19:03:00 <Yawning> hi browser people 19:03:15 <isabela> boklm: ximin will also work on one as part of the tor at the heart series 19:03:35 <isabela> boklm: do you want me to add you to this email thread? 19:03:46 <boklm> isabela: ah, yes 19:03:50 <isabela> :) will do 19:04:39 <GeKo> i can go next i guess 19:04:59 <GeKo> last week i got the releases out and it seems all went well 19:05:22 <GeKo> i cut down my backlog and worked mainly on design documenation update 19:05:42 <GeKo> this week i mainly plan on traveling and continuing to work on the design doc update 19:05:53 <GeKo> that's it for me 19:06:06 <Yawning> the sandbox was released. it had bugs 19:06:08 <Yawning> I fixed 19:06:17 <Yawning> gonna do the updater background/robustness stuff 19:06:24 <Yawning> while I can 19:06:53 <Yawning> (I assume a trivial patch to torbutton to integrate that into the UI is possible for the next release) 19:07:13 <GeKo> i think so 19:07:16 <Yawning> (it's literally, "when the check for update thing is cliecked, send a control port command if an env var is set") 19:07:32 <Yawning> I think it will be low risk, because it never will get triggered on unsandboxed builds 19:07:41 <Yawning> and it can't get more broken than "the thing does nothing" 19:08:52 <Yawning> I'm not sure what will happen past the updater changes, we'll see 19:08:56 <Yawning> that's it for me 19:09:02 <mcs> Do you mean the “Check for Tor Browser Update” menu item or also the about box, etc.? 19:09:48 <Yawning> just the menu item 19:09:53 <Yawning> the about box thing doesn't appear 19:10:14 <Yawning> it says "updates disabled by system administrator" 19:10:15 <mcs> That sound fairly easy and fairly safe then. 19:10:16 <mcs> sounds 19:10:20 <mcs> ah, right 19:13:08 * mcs will go next 19:13:16 <mcs> Last week, Kathy and I responded to OSX sandbox issues that have been reported (see #20989 and #21009). 19:13:22 <mcs> We also took another look at the Tor Browser roadmap and will send GeKo some feedback later today. 19:13:29 <mcs> In addition to that, we will continue to investigate the OSX sandbox problems this week. 19:13:33 <mcs> Coming up, we will be taking some time off to enjoy the Christmas and New Year holidays and to take care of end-of-year things for our company. 19:13:37 <mcs> That’s all for now. 19:14:33 * arthuredelstein can go 19:14:57 <arthuredelstein> Hi Everyone. :) This week I worked on #20680. 19:15:04 <arthuredelstein> I have about 18 patches left to go. I have some questions about a few of them and I will be sending out emails about these soon. 19:15:15 <arthuredelstein> So this week I will continue to work on rebasing and also I will look into #20981. 19:15:30 <GeKo> neat 19:15:33 <arthuredelstein> And I'll be taking time off next week as well. :) 19:15:40 <Yawning> (ESR52 in this context is just geck-dev?) 19:15:43 <Yawning> *gecko 19:15:45 <arthuredelstein> That's it for me. 19:15:55 <arthuredelstein> ESR52 is the aurora branch of gecko-dev 19:15:55 <mcs> “18 patches left to go” means we still have to many patches ;) 19:15:58 <Yawning> FF52 doesn't exist yet right? 19:16:00 <Yawning> ah k 19:16:14 <arthuredelstein> It's not ESR yet ;) 19:16:17 <mcs> too many also 19:16:19 <GeKo> no, still vaportware :) 19:16:28 <GeKo> vaporware even 19:16:36 <mcs> It is great to get an early start of the rebasing! 19:16:49 <mcs> umm… on the rebasing 19:16:56 <arthuredelstein> mcs: We do still have too many, but not having to rebase the isolation patches is definitely a big help! 19:17:04 <GeKo> + getting the compiling going 19:17:31 <Yawning> "figuring out what is broken in my sandbox" 19:17:48 <GeKo> fwiw: i spent another deal of my spare time to get all issues for mingw-w64 on our radar 19:18:07 <GeKo> i got one patch uplifted but there are still things that are broken :/ 19:18:26 <GeKo> and it seems this time jacek does not have the time to fix all the tricky issues for us. 19:18:35 <GeKo> so... exciting! 19:19:33 <GeKo> anyway, who else is here for a status update? 19:19:37 <Synzvato> I could go 19:19:57 <GeKo> hi! 19:20:02 <Synzvato> Hi :) 19:20:04 <Synzvato> This week I have done some more work on 20815 - preventing multiple instances of Orfox settings tabs, adding display logic to tor-browser-settings page elements, and implementing add-on setting persistance. 19:20:12 <Synzvato> #20815 19:22:09 <Synzvato> Will be available at #tor-mobile right after this meeting to discuss the upcoming alpha release 19:22:19 <GeKo> so where are we right now? 19:22:22 <GeKo> ah, okay 19:23:00 <GeKo> do you need anything from the other tor-browser folks? 19:23:08 <GeKo> input or something? 19:23:29 <Synzvato> The foundation is done, and the UX is functional 19:24:12 <Synzvato> Perhaps feedback on the current approach, and the goal of turning this into an embeddable WebExtension "module" 19:24:30 <Synzvato> Any feedback is welcome, of course :) 19:24:57 <Synzvato> https://git.synz.io/synzvato/tor-browser-settings/tree/master 19:26:55 <GeKo> okay. 19:27:17 <GeKo> ok, i guess we move on to the discussion part 19:27:25 <GeKo> first: next meeting time 19:28:01 <GeKo> i'll be traveling jan 2nd so i'd propose having our next regular meeting on january 9th 19:28:08 <GeKo> does that work for everyone? 19:28:16 <Yawning> yeah 19:28:18 <mcs> yes 19:28:19 <Synzvato> Sure 19:28:25 <boklm> yes 19:28:55 <GeKo> seems so, good. i'll write a respective mail to tbb-dev these days then 19:29:23 <GeKo> do we have anything else up for discussion in today's meeting? 19:29:33 <Yawning> we should always kill the FF JS JIT 19:29:37 <Yawning> regardless of security slider setting 19:29:41 <Yawning> because it is dangerous and evil 19:30:05 <GeKo> oha :) 19:30:30 <Yawning> we should also consider our historical stance of "people will get confused" as a reason for defaulting the slider to low 19:30:46 <Yawning> and default it to high, so that people who want low security have to opt into it 19:31:11 <Yawning> the 2nd thing is probably more controversial than the first 19:31:12 <Yawning> :P 19:31:54 <GeKo> while i agree on JIT being scary i think the solution is a good sandbox :) 19:32:21 <Yawning> if the jit is scary that it takes containment that doesn't exist on all platforms yet 19:32:27 <Yawning> it shouldnt' be enabled in the name of performance 19:32:35 <Yawning> that's my opinion anyway >.> 19:32:53 <mcs> Is the idea of a JIT scary in general or is the Mozilla implementation scary? 19:32:59 <arthuredelstein> Also even with a sandbox a JIT doesn't necessarily stop deanonymization. 19:33:02 <Yawning> it's scary in general 19:33:06 <Yawning> arthuredelstein: correct 19:33:26 <Yawning> #21011 19:33:27 <arthuredelstein> If we had perfect memory hardening then maybe it would be OK. But the world is a long way from that still. 19:33:45 <Yawning> the jit still would let bitfipping attacks and stuff work 19:33:53 <Yawning> better even with memory hardening 19:34:04 <Yawning> because most people don't have DDR4 or ECC memory 19:35:09 <Yawning> nb: I'm not sure if newer DDR is sufficient to solve the read disturb problem 19:35:49 <Yawning> (i changed the sandbox security slider default to high, so people need to opt into the JIT, but still) 19:37:03 <GeKo> looking at the fun in #19400 19:37:27 <GeKo> we got an idea who is using these things: 19:37:33 <GeKo> facebook and the like 19:37:58 <Yawning> ooof 19:38:09 <Yawning> it's just performance right? 19:38:23 <Yawning> like, with the JIT disabled stuff will just be slower rather than non-functional? 19:38:35 <arthuredelstein> yes, afaik 19:38:39 <GeKo> slower in a way that it is basically non-functional 19:39:06 <Yawning> the fuck are they doing 19:39:13 <GeKo> i myself thought once debugging mega.nz issues that this breaks functionality 19:39:18 <Yawning> this isn't like "I want to play quak2 in my browser" or something right? 19:39:26 <Yawning> *quake 19:39:31 <GeKo> but no it took just minutes to load (literally) instead of seconds 19:39:37 <GeKo> no 19:39:51 <Yawning> my burning hatred for javascript rises 19:40:23 <GeKo> if it were just some gaming stuff that would be pretty easy to handle 19:40:39 <Yawning> Do we have any idea on how many people use tor browser with the security slider set to "YOLO" 19:40:53 <Yawning> vs increasing it? 19:41:30 <GeKo> i have no hard data 19:41:53 <Yawning> I have no sensible solution to this 19:42:06 <Yawning> that doesn't boil down to "change the default and educate people better about the risks" 19:42:14 <Yawning> "and set everyone that uses asm.js on fire" 19:42:22 <GeKo> yeah, i know 19:42:58 <arthuredelstein> Do we care about mega.nz? Maybe turning of JIT is worth the few sites that break. 19:43:17 <Yawning> it's what, a file host? 19:43:19 <arthuredelstein> The scripts in mega.nz are super sketchy for multiple reasons. 19:43:22 <Yawning> people will probably complain 19:43:24 <arthuredelstein> yes 19:43:36 <GeKo> sure. but facebook? 19:43:45 <Yawning> we should rename "Low" to "Unsafe" 19:43:51 <arthuredelstein> Does facebook not work with JIT disabled? 19:44:32 <GeKo> well "work", i don't know i did not spend my time doing performance measurements there 19:44:43 <Yawning> is it unusably slow? 19:44:52 <Yawning> I've never used spacebook 19:45:21 <GeKo> and it it is not facebook then it is an other major site that deploys that shit and makes our users screaming "Tor Browser is broken!1!" 19:45:26 <GeKo> *and if 19:45:29 <Yawning> yeah 19:45:41 <arthuredelstein> One option could be to disable JIT for non-https. 19:45:51 <Yawning> I think renaming the seucirty slider to Unsafe and making uses having to opt into it is a good idea 19:45:53 <arthuredelstein> Though I don't know how hard that is. 19:45:54 <Yawning> arthuredelstein: that solves nothing 19:46:02 <arthuredelstein> Yawning: why? 19:46:06 <Yawning> "now the badguys need a Lets encrypt cert to heap spray you" 19:46:22 <Yawning> I do not see that as a large improvement 19:46:25 <Yawning> >.> 19:46:39 <arthuredelstein> I think it would be a significant improvement over the status quo. 19:46:45 <arthuredelstein> Where any exit node can do it. 19:46:46 <Yawning> yeah 19:46:51 <Yawning> true 19:47:10 <arthuredelstein> I agree better still is to disable the JIT completely. 19:47:14 <Yawning> the more I think about it, the more renaming low to something that conveys that it is really dangerous 19:47:17 <Yawning> is a good idea 19:47:27 <Yawning> and new installs should be set to something higher 19:47:39 <Yawning> so when sites break on everything but "Unsafe" 19:47:42 <Yawning> it's obvious who to blame 19:48:04 <GeKo> and your users are already gone 19:48:10 <Yawning> mm 19:49:21 <Yawning> I dunno, have we had a browser vuln that didn't require JS apart from RELAY_EARLY 19:49:36 <Yawning> at some point I think we should say "enough is enough, that shit is off by default" 19:49:46 <Yawning> but I'm starting to rant, so I'll stop now 19:49:51 <GeKo> well JS is not necessarily JIT 19:49:55 <Yawning> because this is a complicated problem 19:50:15 <GeKo> or did you want to get rid of JS as well while you are at it? 19:50:24 <GeKo> :) 19:50:28 <Yawning> well 19:50:38 <Yawning> I think people should have to explicitly increase their attack surface 19:50:40 <Yawning> yes 19:50:48 <Yawning> rather than explicitly increase their security 19:51:09 <arthuredelstein> One problem is the security slider is browser-wide instead of site-specific. 19:51:16 <arthuredelstein> Not that that's easy to solve. 19:51:22 <Yawning> yeah 19:51:34 <Yawning> it'd be nice if it was like noscript 19:51:42 <Yawning> "allow unsafe browser things for this site" 19:51:44 <Yawning> or whatever 19:51:49 <Yawning> but that's probably a huge rat's netst 19:51:57 <Yawning> and months of dev time if not ore 19:52:58 <Yawning> sorry to rant 19:53:00 <Yawning> :/ 19:53:14 <GeKo> alright, i guess this is one of the discussions that will keep on our plate for the near future :) 19:53:19 <GeKo> no worries 19:53:39 <arthuredelstein> I think Yawning is making an important point. We need more data to decide what we can afford to turn off. 19:53:51 <arthuredelstein> Not sure how to get that data though. 19:54:25 <mcs> We can also consider adding more text to help people choose wisely (and maybe prompting them). 19:54:41 <Yawning> maybe make setting it part of the setup wizard 19:54:56 <Yawning> like, after you configure tor, add a step where it's like 19:55:14 <Yawning> "pick your poison using mega/facebook and geting owned by the FBI, or being secure" 19:55:15 <Yawning> >.> 19:55:23 <GeKo> hah 19:55:30 <GeKo> i think we had this idea back then 19:55:48 <GeKo> but we seemed to think that this would be pretty overwhelming for the normal user 19:56:04 <GeKo> who is usually just opening the browser and starts surfing 19:56:36 <arthuredelstein> What about a periodic reminder to users who are in the low setting? 19:56:52 <arthuredelstein> Like "We notice you are still on the low security setting? Would you like to increase it by any chance?" 19:56:54 <Yawning> like, I know we display the thing on first launch directing people to it 19:57:11 <Yawning> maybe we should bother the UI/UX team 19:57:19 <GeKo> we should 19:58:29 <GeKo> alright. i think we are done for today. thanks for the meeting and i hope all of you get some rest over the holidays 19:58:39 <Yawning> safe travels to those that are 19:58:51 <mcs> yes, have a great Christmas everyone! 19:58:58 <GeKo> + hopefully 2017 will be less crazy 19:59:14 <Synzvato> Thanks, same, happy holidays everyone! 19:59:15 <arthuredelstein> Happy holidays everybody! 19:59:18 <GeKo> (i know, i know but one can still dream) 19:59:23 <GeKo> *BAF* 19:59:26 <GeKo> #endmeeting