=== oCrazyLemn is now known as CrazyLemon [07:26] Hello, world!\n [08:02] good morning desktopers [08:02] hai [08:04] yo [08:06] hey seb128 [08:06] mvo, hey ;-) wie gehts? [08:06] and good morning mlankhorst and Laney [08:06] same! [08:06] seb128: good, thanks. looking at upgrade bugs [08:07] that is medium fun but important :) [08:08] * mvo makes up for it by drinking huge amounts of tea [08:08] (to add fun ot it I mean) [08:08] never upload chromium [08:09] to be more precise, don't subscribe to its bugs [08:09] Laney, living the life of the maintainer of a popular (and buggy) application? ;-) [08:10] a maintainer who has an overused "mark all as read" button :P [08:10] lol [08:11] haha - or a worn out "d" key [08:13] I've got a script which subscribes me to bug mail for packages I upload [08:13] which is fine most of the time, but not when I upload chromium just to change a recommends to a suggests [08:13] *cough* [08:13] bonjour seb128 [08:13] hey mvo [08:14] morning Laney [08:14] hey mlankhorst! [08:14] hey pitti [08:14] guten morgen pitti [08:14] I should check this channel more often :) [08:14] its good fun here [08:15] was ist los? [08:15] pitti, guten tag! wie gehts? [08:15] hai [08:15] seb128: mediocre, caught a cold :( [08:15] pitti, :-( [08:15] pitti, get better! I hope it's over for the easter w.e [08:15] but yesterday was fun, we went to the Munich theater to see the West Side Story (my wife's bday present) [08:15] pitti: oh, get well! [08:15] pitti, oh, nice [08:16] I hadn't seen that before [08:16] with an original NY theater cast === tkamppeter_ is now known as tkamppeter [08:26] didrocks, see, pitti seems to like croissants au chocolat, he +1ed my post! [08:26] seb128: oui, je les aime ! [08:26] yeah, I had to reflect reality and just answered! [08:26] pitti, moi aussi, mais le pauvre Didier ne connaît pas les croissants au chocolat [08:26] it's clearly a German thing [08:27] he says that they don't have those in Lyon [08:27] comment est-ce c'est possible ? [08:27] "que" [08:27] seb128: we have french cuisine in Lyon! [08:27] only things on THE list :) [08:27] pitti, je ne sais pas non plus, dommage pour eux :/ [08:28] seb128: mais du gâteau de rhubarbe est encore mieux :) [08:28] rhubarbe… pas très fan ici [08:28] gâteau ou tarte ? [08:28] tarte à la rhubarbe ! [08:29] seb128: ma femme a fait le tarte pour mon anniversaire, yummy [08:30] pitti, une tarte à quoi ? à la rhubarbe ? [08:30] c'est pas encore la saison, si ? [08:30] oui [08:30] seb128: non, nous pouvons acheter de la rhubarbe dans nos "market" [08:32] seb128: de avril à juin [08:32] seb128: c'était un printemps tôt :) [08:33] pitti, tu as raison ;-) [09:39] dbarth: I think alex_abreu, Laney and you talked about the migration script which readded the amazon entry in launcher if people have removed it. Is that fixed now? [09:40] don't think so, but alex_abreu said he thought he knew what to do [09:41] what's the bug num. for that one? [09:41] Don't know, I asked him on irc yesterday [09:42] hmm ok [09:42] Laney: dbarth: this needs to be fixed before we release trusty though [09:42] (and should go on -updates) [09:42] and the issue is that the amazon launcher comes back [09:42] yup [09:42] * didrocks can already imagine the bad PR with "Canonical forces amazon in the launcher where I removed it" [09:42] the solution is really easy [09:42] oh no, please ;) [09:42] just put back the script to the older name [09:42] right [09:42] so, people transitionned are already transitionned [09:43] for the new part of the script [09:43] brings a second script with a new name [09:43] but don't mix both in the same script [09:43] hmm, taking a look [09:43] what changed? [09:43] so people on past LTS won't get it added [09:43] Laney: the name of the script [09:43] so never seen -> executed [09:43] I mean in the script [09:43] Laney: I didn't check yet [09:44] okay [09:44] was hoping to bring the topic and having people with more knowledge knowing :) [09:44] https://launchpad.net/ubuntu/+source/webapps-applications/2.4.17+14.04.20140411-0ubuntu1 [09:44] well, then we could take the opportunity to do my fix from yesterday [09:44] * Update session migration script to remove all the cruft [09:44] ~/.local/share/applications by supported webapps [09:44] and avoid having that grimy dependency on libunity-core-whateveritis [09:44] yup [09:44] Laney: this one landed i think [09:44] I know [09:45] I suggested an IMO nicer fix to check that at runtime instead [09:45] https://bugs.launchpad.net/ubuntu/+source/webapps-applications/+bug/1308076 [09:45] Launchpad bug 1308076 in webapps-applications (Ubuntu) "/usr/share/session-migration/scripts/install-default-webapps-in-launcher-and-cleanup-local.py:5:g_settings_set_property:object_set_property:g_object_new_internal:g_object_new_valist:g_object_new" [High,Fix released] [09:45] ah i see [09:45] Laney, can you handle that/make sure the fix land? [09:45] can do [09:45] well, so you want to change robru's fix with another one? [09:45] dunno [09:45] i don't mind, that's packaging black magic to me [09:46] as long as it covers more bases [09:46] I like it more, but who knows [09:46] he said he thought it was still technical debt, not sure I agree [09:46] anyways we can do the first fix I guess [09:47] and so the first fix landed afaict [09:47] unrenaming the script [09:47] and doing the new part in a second one [09:47] yeah, marked as landed in citrain [09:47] ok [09:47] now on to the amazon one [10:06] I wonder if I should rename the "new" script [10:06] install-... isn't what it does now [10:07] Will do, it shouldn't be bad to run it an extra time [10:07] dbarth: btw, not sure it was discussed, but this is now broken with the new webapps: http://developer.ubuntu.com/web/tutorial/ [10:08] dbarth: if a website is setting a userscript on their site to show a webapps, from what I heard (didn't try myself), it won't be shown [10:17] didrocks: a bad exit code means a script will be re-run the next time, yes ? [10:17] but no error or anything like that [10:17] Laney: right [10:18] ok [10:18] g_printerr("Exited with an error\nstdout: %s\nstderr: %s\n", stdout, stderr); [10:18] in logs [10:18] It's "the schema isn't installed, don't do anything but try again next time" [10:18] == sys.exit(1) [10:19] sounds good :) [10:20] cheers [10:27] seb128: dbarth: https://code.launchpad.net/~laney/webapps-applications/dont-readd-launcher-icon/+merge/216057 please to review [10:28] I just copied the old script back [10:28] and removed that part of the new one [10:28] uh, faster [10:28] hmm [10:28] ? [10:29] (looks like from what you tell is the right strategy to me) [10:32] Laney, didrocks, dbarth: looks fine to me as well [10:33] I saw that you suggested this runtime check too [10:33] on the original bug report [10:33] ;-) [10:33] yup +1 [10:34] adding a silo request [10:35] ok well if you want to get it trained into the queue then there's a minor chance of it being accepted [10:35] otherwise 0-day sru which is fine [10:35] *grumble* *grumble* fscking logscreen forces me to kill my debugging session and loose a trillion carefully planted breakpoints. A logscreen that lets me Alt-Tab though apps (which are not shown of course), but not unlock the screen is ... not helpful. [10:35] Sweetshark: what unity version is running? can you reproduce it? [10:36] thanks Laney [10:36] np [10:38] unity 7.2.0+14.04.20140404-0ubunt(here the terminal ran out of chars) [10:39] Try upgrading ;-) [10:39] Laney: I have no welldefined reproduction scenario, but its not the first time this happened. [10:39] Lots of issues like that were fixed in >> 0404 [10:43] man that chromium notifications indicator is ugly [10:43] Laney, getting used to be the chromium maintainer? ;-) [10:45] Laney: :( [10:45] * Laney hugs qengho [10:47] qengho: Is the indicator an upstream thing or something you've done? [10:47] Sorry for sounding mean :( [10:51] Laney: It's not me. Don't feel bad for that. [10:52] In that case I don't know what a Chromium notification is or why I need an indicator for it the whole time :-) [10:53] Laney: Can you show me a screenshot? [10:56] qengho: http://ubuntuone.com/0Ejd0uop0SwgC8HLiR7hUL [10:56] (I'll miss being able to do that) === MacSlow is now known as MacSlow|lunch [11:14] Laney, how did you activate that? [11:15] I have no idea [11:15] not on purpose [11:16] Maybe it was when I got a notification on Twitter (I'm using tweetdeck inside chromium) [11:17] Laney: I have honestly never seen that. I have seen little pop-up windows in bottom-right corner, and I was going to agree they're ugly. [11:17] Someone send me a tweet :-) [11:17] * Laney posts something controversial [11:17] lol [11:18] I guess seb128 already did on g+ [11:18] roh [11:19] * Laney doesn't have seb128 there :-o [11:20] Laney: you are just missing some fake photos :p [11:20] it's like the moon landing right [11:20] didrocks, you are loosing on that post, we are 3 to say "yummy" and you are alone to say "fake" [11:20] he just faked it up in a studio [11:20] seb128: hum, that's not what Julie is saying [11:20] Laney: exactly! [11:20] didrocks, you both are counting for 1 :p [11:21] roh [11:22] didrocks, I don't want to say anything but Julie just admitted on the same post that you eat cat food [11:22] seb128: I saw that, but it seems you don't understand irony :p [11:22] bad for you ;) [11:23] tssss [11:24] didrocks / seb128: want to check landing-004 doesn't re-add icons for you? [11:25] didrocks, btw I saw Julie (the real one on TV, with a "e" in the name :p) cooking some of those pink stuff on TV yesterday [11:25] Laney, on it [11:28] seb128: ahah ;) [11:28] seb128, what's the word on the street regarding trusty? everything looking good? [11:29] rickspencer3, there were some rounds of bugfixes/respin since yesterday (mostly langpack install issues), but things look under control/good [11:30] Still another go around for that problem [11:30] OEM installs getting the wrong language settings [11:30] * seb128 test Laney's migration script fix, session restart, brb [11:30] Seems to be under control though [11:30] well, tomorrow is release day ... [11:31] Yup, fix is uploaded [11:31] Laney, seb128 so sounds like one last respin for desktop today? [11:31] AIUI, but I'm not in the room [11:31] rickspencer3, hopefully one and we are good yes [11:31] nice [11:32] other than that it's pretty good I think [11:34] Laney, no icon added after a session restart [11:34] cool [11:34] thanks for checking [11:35] new kettle! [11:35] yw [11:37] * Laney MPs a gsettings-qt symbols file [11:54] dbarth: do you want to check 004? [11:54] I'd like to publish it quickly so it has a chance of going in [11:58] doing this now ;-) [11:58] Laney: sure [11:59] just doing now (removed amazon first) [12:01] I think you might get it readded if you installed in the 5 days since the script was renamed [12:01] but that's probably acceptable [12:01] just re-starting my session to make sure [12:02] Laney: yeah, it's the "risk" of running a development release :) [12:03] (seems acceptable if we only had those kind of issues :p) [12:04] Laney: ok, that's a land then [12:04] landing [12:04] well, anyway [12:04] dbarth: thanks, already pressed the button ;-) [12:05] cheers for checking [12:05] see that [12:05] on that note, going to lunch [12:05] Laney: the icing on the cake is that i don't even have to ping the release team about the unapproved queue ;) [12:05] will walk to the cafe in the sun, nice day for it :-) [12:06] Laney: yup, enjoy [12:06] Laney: don't take any croissant please! === MacSlow|lunch is now known as MacSlow === Sweetsha1k is now known as Sweetshark [13:16] Laney: FWIW libreoffice autopkgtest magically self healed: https://jenkins.qa.ubuntu.com/view/Trusty/view/AutoPkgTest/job/trusty-adt-libreoffice/ [13:19] Sweetshark: witchcraft? [13:20] time for some exercice, bbiab [13:22] Laney: "any sufficiently advanced technology is indistingushable from magic" [13:22] Laney: teaches you something about how advanced LibreOffice is ;) [13:31] Laney, didrocks https://code.launchpad.net/~abreu-alexandre/webapps-applications/dont-forcefully-add-amazon-webapp/+merge/216100 [13:36] alex_abreu: thanks - we already uploaded something similar to that earlier to try and get it into the release [13:36] check the trusty-proposed branch [13:37] but in that one we restored the old script to add the launcher entry too [13:37] Laney, yeah sorry for the delay, was busy w/ other things [13:37] you think that's not needed? [13:40] Laney, mmmh yeah yours is better actually [13:40] ok [13:40] Laney, why do you remove libunity-core ? [13:40] added a runtime check for it instead [13:41] Laney, the SettingsSchemaSource ? [13:41] y [13:41] a [13:41] ok +1 [13:41] avoids having that fugly dependency [13:41] :) [13:41] agree [13:46] wowzers, my PC's locked up hard [13:46] all I did was try the G+ webapp [13:57] twice now [13:57] someone else want to try this? [13:58] using nouveau if that helps [13:58] mlankhorst: maybe you [14:00] http://paste.ubuntu.com/7261615 http://paste.ubuntu.com/7261622 [14:04] card's dead [14:04] oh looks like fun corruption [14:05] and a newer card than I have :P [14:05] card specific? [14:05] no idea [14:05] launch the G+ webapp and use it a bit [14:05] eventually (< 1 min) it locks my system up [14:05] looks like the real fun is in 'Unknown handle 0x0000001f' which means that the kernel disagrees with userspace about whether some buffers are valid or not === JanC_ is now known as JanC [14:29] Laney: can I run it from the command line? [14:30] Exec=unity-webapps-runner -n 'R29vZ2xlUGx1cw==' -d 'plus.google.com' --store-session-cookies %u [14:30] that's what the desktop file runs [14:59] no crash yet [14:59] I never managed to get signed in [14:59] time to expense a new card! [15:03] just as i finished typing it.. [15:03] :/ [15:03] you got it? [15:03] what's the facebookmessenger app that keep coming back into my launchers [15:04] Laney: yeah after i hit enter on 'no crash yet' [15:04] well, that's promising [15:04] as in, if anyone in the team can fix it :P [15:04] no! It's downright terrible.. [15:04] xclaesse: Hrm. $ grep face ~/.config/autostart/* [15:04] qengho, nothing [15:05] xclaesse: How often does it "keep coming back"? [15:06] that app seems completely useless, it just dispaly the same facebook than I have in firefox already [15:06] qengho, literraly every time I'm not looking [15:06] Oh, a web-app maybe? [15:07] ah ok, each time I reload facebook tab in firefox [15:07] that facebook icon appear in the launchers [15:10] uh, yeah, that happens here too [15:11] also for other webapps [15:11] alex_abreu: do you know about this? [15:11] Laney: bleh I suspect something that valgrind will turn up anyway [15:12] * qengho thinks hard about Valgrind on GPUs. [15:12] Laney, mmmh no [15:13] qengho: no it's a handle issue, some handle is annotated by userspace but the handle is invalid :P [15:17] alex_abreu: ok, I'll file it, which package is the firefox webapps integration? [15:17] Laney, unity-firefox-extension [15:17] ty === m_conley_away is now known as m_conley [15:18] Laney, xclaesse, alex_abreu: I guess that's how I got the amazon icon back on my launcher the other day btw [15:18] Laney, since I didn't have webapps-applications-common installed [15:19] maybe [15:19] I can't trigger it with amazon but could be [15:20] Laney, I had it after visiting your kettle stuff I think [15:20] try again? [15:20] yeah, I couldn't reproduce [15:20] I guess you can do with facebook or g+ though [15:20] but I can't reproduce on fb either [15:20] huh [15:20] no [15:20] it seems to depends of the url [15:20] or how you access the site [15:20] not sure [15:20] I just go to facebook.com in firefox [15:21] not logged in if that makes a difference [15:21] yeah, can't reproduce... [15:21] oh well [15:22] that made amazon listed first in my dash as most recently used though [15:22] (trying on amazon) [15:24] Laney: yeah freed memory use in egl_gallium, probably the cause of this bug :P [15:24] that's usually inadvisable [15:24] can't look further, no debug symbols yet [15:27] ok tomorrow :P [15:29] bug #1308625 [15:29] Launchpad bug 1308625 in unity-firefox-extension (Ubuntu) "Visiting a website with a webapp re-adds the launcher icon every time" [Undecided,New] https://launchpad.net/bugs/1308625 [15:30] I don't remember what, if anything, I picked when it asks you if you want to 'install' the site [15:30] or how to get that choice back [15:31] Laney, I can confirm your issue btw [15:31] which one? [15:31] Laney, I had the webapp integration disable, I did that ages ago, I didn't remember it [15:31] Laney, fb adding itself to the launcher [15:31] ah [15:31] you mean you had the extension off? [15:31] yes [15:32] nod [15:32] * Laney adds that to the bug [15:33] everyone must disable it or we'd certainly have heard about this by now, you'd think [15:33] ... [15:33] or keep the icon there I guess [15:33] or don't use fb :p [15:34] yeah it is pretty niche [15:34] our devel release userbase is for sure not the same set than the ones who are going to run the stable version [15:34] I don't use it on that machine or I'd have noticed sooner [15:34] hum, visiting g+ adds googledocs to the launcher [15:35] dbarth, alex_abreu, are you looking at those issue? [15:35] that seems like a potential release blocker to me [15:36] seb128: same as amazon, or is that a new one? [15:36] seb128: hmm no, ok ,taking a look [15:36] dbarth, https://launchpad.net/bugs/1308625 [15:36] dbarth, it's happening with facebook, g+ at least [15:36] seb128: Error: Could not gather data from Launchpad for bug #1308625 (https://launchpad.net/bugs/1308625). The error has been logged [15:36] good old LP going down [15:37] yeah,seems so [15:37] release time, not a good time for being launchpad [15:37] dbarth, well, basically is "enable webapp integration in firefox, visit facebook -> get an icon for fb in the launcher" [15:37] right [15:37] dbarth, if you unpin it, the icon comes back every time you visit the site [15:37] remove facebook first if it's already there [15:38] ok, checking in a guest session; any particular rev. of the firefox ext? [15:38] same on g+ (it doesn't even add the right site there, you get a gdrive launcher) [15:38] dbarth, trusty [15:38] current trusty versions [15:38] I get the G+ one [15:39] could be random on google sites? [15:39] shrug [15:39] oh, I don't have unity-webapps-googleplus in fact [15:40] I'll let these guys debug it now :) [15:40] yeah [15:40] xclaesse: thanks for reporting [15:40] dbarth, alex_abreu: let us know what you figure out [15:40] xclaesse, thanks! === gatox is now known as gatox_lunch [15:42] shrug, does the same on launchpad [15:42] do we enable that integration by default? [15:42] * seb128 boots test box [15:43] I don't see it there either [15:43] wonder why [15:43] I probably said 'never' ages ago [15:45] Laney, we don't install any of the unity-webapps-* by default in trusty [15:45] yeah I checked I have these ones installed though [15:45] that's likely why nobody ran into the issue [15:45] same here [15:45] so either we tried that [15:45] or they were installed in older releases [15:46] I had it installed as far back as my apt/dpkg logs go [15:47] 2013-05-16 [15:47] could have done it manually though [15:48] or that's what the install prompt does/did [15:53] Laney, seb128 ok a fix is on the way [15:53] this guy is fast [15:53] Laney, seb128 https://code.launchpad.net/~abreu-alexandre/unity-firefox-extension/fix-launcher-pin-on-reload/+merge/216148 [15:54] ty, building that [15:59] alex_abreu: yep, no new icon, thanks [16:00] dbarth, can you organize a landing for that fix? [16:00] Laney, ok great, ... [16:00] should get it reviewed [16:00] it's not obvious why it's right to me [16:01] yeah, to me neither [16:01] we should also check that they still get added on first use/when they should [16:02] rsalveti, hum, committing to trunks/uploading of projects without asking for review or even pinging on IRC now? :-( [16:02] seb128: packaging changes, and just adding | [16:02] for the x86 emulator [16:02] not changing anything else [16:03] yeah, packaging changes can be buggy [16:03] that was the direction I took with the landing team for that sort of changes [16:03] seb128: right, I'm a core-dev [16:03] I could just upload it by hand as well [16:03] and asking is also explaining to others what you commit to their trunk and why :p [16:03] rsalveti, the fact that you can do it doesn't make it right though [16:03] I just did the sync so the trunk is also in sync with the archive [16:03] rsalveti, yeah, my problem is not with the way the landing is done [16:03] right, but it doesn't mean I need to ping someone before doing it [16:04] is about committing changes to project you don't usually contribute to without even bothering give a notice/ping on IRC [16:04] seb128, Laney I did test that & other use cases [16:04] well, I didn't change the code, and check the review, it's minimal [16:05] landing for this is a huge pita [16:05] for a minor packaging change [16:05] rsalveti, let's agree to disagree, it feels wrong in principle to me but maybe that's just me being old school [16:05] well, the old school way would be me uploading it directly [16:05] and not pushing to trunk [16:05] rsalveti, yeah, as said my issue is not with the way it landed, it's about doing changes on packages actively maintained by somebody else without asking/let a fyi first [16:05] but then someone from the landing team would need to do that later on [16:06] alex_abreu: I mean I can't tell what that realInit call is for and if it was ever useful [16:06] but maybe nobody else exists who knows this code anyway [16:06] seb128: asking for changes when minimal? [16:06] this is not debian [16:06] rsalveti, I soviet Debian you would get list flamed for uploading a package without talking to the maintainer :p [16:06] I->in* [16:06] right, but that's why I'm not doing that there [16:06] :-) [16:06] haha, fair enough [16:07] seb128: but I'll ping you next time if you prefer that, np with that [16:07] rsalveti, well as said, no big deal, it's maybe me but I try to ping/let a quick fyi before uploading a package if it's usually actively maintained by somebody/a team in Ubuntu [16:07] Laney, yeah, I asked for a review, ... the thing is that it is a left over from the past. The extension was much simplified & now only takes care of very little [16:07] rsalveti, thanks [16:08] rsalveti, like if you had asked "ok if I upload a small control fix for the emulator" I would have said "yes, no problem, go for it" and we would have spared that discussion ;-) [16:08] anyway moving on [16:08] rsalveti, thanks for the fix btw ;-) [16:08] seb128: np, I thought it was fine because I had that discussion with the landing team before doing such changes [16:08] so I had their 'ack' [16:09] but anyway, pinging you next time :-) [16:09] rsalveti, it's not their project :p [16:09] fair enough [16:09] :-) [16:09] but yeah, enough discussion about a minor detail [16:09] thanks ;-) [16:10] rsalveti, btw another approach is to email ubuntu-devel@ saying "I"m going to go through the package that have that depends to add an alternative, let me know if you are in this case and have an issue with an upload" [16:10] sure [16:10] rsalveti, that way you don't have to ping individual projects and you let everyone know waht's going on [16:10] [16:10] ;-) [16:17] bregma, Trevinho, ChrisTownsend: is anyone looking at why compiz is making update-manager not resizable? that's a regression from this cycle and quite annoying, if you try to see the details of an upgrade you have like 1 line of content and can't change the height to see more [16:18] seb128: I will [16:18] seb128: it's resizable btw... but the hints gets updated with some delay [16:18] Trevinho, you should share some of the bug with your team ;-) [16:18] seb128: playing with the window size gets the power back [16:19] seb128: hehe, of course, just that I know where to put the hands [16:19] Trevinho, not it's not resizable for me, I don't have the cursor changing on borders [16:19] no* === gatox_lunch is now known as gatox [16:36] Trevinho, same thing happens with the apport dialog, btw, I used to be able to resize it now it's fixed size and I can't see most stack dumps... maybe those windows have something in common>? [16:37] bregma: might be... as I ignored one event that I thought it was duplicated, while it might not be the case === maclin__ is now known as maclin === dpm is now known as dpm-afk === psivaa_ is now known as psivaa === mjohnson151 is now known as mjohnson15 === alex_abreu is now known as alex-abreu [17:06] * Laney is off [17:06] seb128: do you think we need to insert some stuff in the release notes? [17:06] https://wiki.ubuntu.com/TrustyTahr/ReleaseNotes#Desktop looks bare [17:06] Laney, we should [17:07] bregma, can you get an unity7 section in ^ [17:07] that's for known issues [17:07] * bregma looks [17:07] there's an 'updated packages' section for good stuff [17:07] known issues could be e.g. the grab lockscreen bug [17:08] probably others too, need to think [17:08] will be back later / tomorrow morning to work on that a bit [17:08] but now I have to go: https://twitter.com/DepotNottingham/status/456410825117671424/photo/1 [17:08] :-) [17:08] Laney, I guess that be filed tomorrow morning still? [17:08] Laney, happy climbing ;-) [17:08] sorry? [17:09] that can be* [17:09] Laney, the release notes [17:09] I think we'll have the morning to work on it [17:09] ok [17:09] so good [17:09] let's do that then [17:09] but you could verify with ubuntu-release [17:09] i'm not doing that stuff [17:09] right, but I think I'm going to call it a day soon as well [17:09] nod [17:09] let's see, I might be back later in front of the TV to write some stuff [17:09] otherwise it's going to be for tomorrow [17:09] we have our .nz overnight guy who might have time ;-) [17:10] oh, that's right! [17:10] I'm going to drop an email to robert_ancell [17:11] our man in the land of the upside down [17:11] ok, back later [17:11] bye! [17:13] have fun, ttyl! === mjohnson151 is now known as mjohnson15 [17:55] robru, did you have any trouble with smuxi and the notifications menu? [17:56] qengho, notifications *menu* seems fine for me, the problem I'm having is that I can't get it to make an audible beep on notifications. And yet it makes the beep in gnome-shell, so it seems more like a unity issue... === psivaa is now known as psivaa-afk [17:58] robru: say my name. I'll get a screenshot. [17:58] qengho, ok [17:58] robru: Hrm, again? [17:59] qengho, qengho, qengho ! [18:02] robru: thanks. http://i.imgur.com/RM2nQ45.png [18:03] qengho, nope, never saw that [18:05] robru: do you use many workspaces? [18:05] qengho, nope, just the one (but two screens though) [18:05] robru: I just discovered I get no notification menu update if smuxi is focused, even if it's in a different channel than focused one. [18:06] I don't know if focus/visibility is different. [18:06] qengho, right, if the window is focused then smuxi assumes you're paying attention and doesn't do any notification [18:08] qengho, actually what's annoying me about smuxi now is the when I wake up each morning and it tells me I have x pings in a given channel, but then I have to scrollllll so far through the scrollback to find them. I might give quassel another try because it has that special notification area that just shows all your pings at once... [18:09] What's the best way to bisect this menu problem? Before we added DBus security, I'd have just snooped it. Is that still easiest? [18:10] qengho, I don't know much about dbus personally ;-) [18:11] robru: lucky. === dpm-afk is now known as dpm [18:34] qengho, what's the issue? === dpm is now known as dpm-afk [18:52] seb128, I want to find the problem with my IRC notifications. http://i.imgur.com/RM2nQ45.png [18:53] seb128: I suspect the client is doing something wrong. [18:53] qengho, that seems like an indicator-messages issue, larsu can probably help you if you ping him during european work hours [18:53] or client issue, but larsu knows the api/can help you debugging [18:53] seb128: thx. [20:13] robert_ancell, jasoncwarner, hey is https://bugs.launchpad.net/ubuntu/+source/gnome-screensaver/+bug/1308572 on your radar at all? [20:13] Launchpad bug 1308572 in gnome-screensaver (Ubuntu) "Ubuntu 14.04: security problem in the lock screen" [Undecided,New] [20:13] rickspencer3, I hadn't seen it but bregma should be looking at that (it's all in unity now) [20:14] robert_ancell, ok, thanks [20:17] Trevinho: can you take a look at the bug rickspencer3 mentioned above? [20:19] jasoncwarner, rickspencer3: ok... I'm moving the bug to unity btw [20:19] beat you :) [20:19] Laney: oh :) [20:20] Trevinho, just bumped the prio... [20:20] Thanks Trevinho IMO this should be top of our list right now. Thanks! [20:20] hopefully didn't step on anyones toes [20:20] Trevinho, i can take a look as well [20:20] jasoncwarner: I can't reproduce btw... [20:20] Trevinho, i can [20:21] bschaefer: ok, cool [20:21] Trevinho, just type an incorrect password [20:21] Trevinho, and just hold enter down [20:21] bschaefer: yeah, but it goes well here [20:21] * bschaefer assumes it the TextEntry being popped off the queue [20:21] just hold down enter [20:21] you don't even need to put a password in [20:21] yeah just did that as well [20:21] rickspencer3: yes, I'm trying but I can't :) [20:21] lucky me :P [20:21] Trevinho, did you accidentally fix it :)? [20:22] Trevinho, what unity branch are you running? Im a bit behind trunk [20:22] i think [20:22] bschaefer: mh, I think no... I'm on trunk + hacks now... [20:22] to repro I just cntrl-l and hold down the shift key until it crashes [20:22] then I am into Unity [20:23] Trevinho, yeah i've a crash dealing with a lockscreen tooltip...hopefully thats not where im crashing [20:23] I guess we need to fix either the crasher or make it so that unity doesn't comeback when it crashes [20:23] rickspencer3, well you can always restart unity in a tty... [20:23] to get around the lock screen (idk if thats been fixed) [20:23] rickspencer3: well, it should come up locked if it crashes in such state [20:23] bschaefer, do you not need a password for the 'puter to log in to do taht? [20:23] Trevinho, right, but the first step is to get it into a shippable state [20:24] rickspencer3, yeah, so its not as critical [20:25] Trevinho, was that bit fixed? When you restart unity, if in lockscreen, to show the lockscreen still? [20:27] bschaefer: sorry? We don't have that yet [20:27] Trevinho, shoots [20:27] Trevinho, thats half the fix for this bug [20:27] * bschaefer will focus on the crash it self [20:28] bschaefer: it would be trivial to achieve though... Just touching/removing a cache file when entering/exiting the lockscreen so that when loading unity we check if locking or not [20:28] Trevinho, we can do that for now [20:28] or possibly the way to do it :) [20:29] so anytime we setup unity and the lockscreen controller, check if that file exists, if so, then lock the screen [20:29] exactly [20:29] Trevinho, that should be a really quick way around the current security issue (though crashes are bad bad BAD!!!) [20:29] bschaefer: I can do it, if you move on the crash [20:29] Trevinho, sounds good [20:29] bschaefer, right, the crash should be fixed, but it seems like that method is necessary anyway [20:31] rickspencer3, yeah, ... hopefully we can get that landed asap to fix the release... [20:31] groovy [20:31] bschaefer, Trevinho will it help if I ask for status every 5 minutes [20:31] ? [20:32] j/k [20:32] haha [20:32] it motivates Trevinho [20:32] you need to juju deploy a status pinger to all the manager nodes [20:32] Trevinho, ill set up a new bug for you :) [20:32] crap, just tried to reproduce and now unity is gone [20:32] but at least it returned locked [20:32] rickspencer3: eheh :) [20:36] Trevinho, actually ill turn that current bug into the main problem you're looking at and ill make a new bug for the crash [20:37] bschaefer: ok, if you've the bt to attach is better to go with a new [20:37] Trevinho, i dont, gdb doesn't like my laptop due to schemas :( [20:37] ill give it a go though [20:41] Trevinho, if you cant reproduce the crash, you can always restart unity from a tty to test that fix :) [20:42] bschaefer: indeed [20:42] Trevinho, also try unpluging your extra monitor :) [20:42] bschaefer: or, also from main view [20:42] Trevinho, couldn't get the crash with a second monitor, with 1 i can [20:42] Trevinho, or that ;) [20:43] bschaefer: ah so maybe it's why I can't reproduce it, I'm two monitors [20:44] Trevinho, yeah [20:44] Trevinho, you always do :) [20:44] Trevinho, but cool, got the stack trace [20:45] Trevinho, interesting: http://paste.ubuntu.com/7263684/ [20:45] looks like an issue with future (some message hanging aroud) and calling a glib signal === Ursinha is now known as Ursinha-afk [20:46] Hey robert_ancell could you help test the fix that Trevinho is working on when he's done? [20:47] bschaefer: I'd avoid the problem from the root btw, by avoiding to emit too many signals when enter is pressed, and if we didn't get anything back yet [20:47] bschaefer: basically unless we don't have a reply, we should avoid to send anything [20:47] Trevinho, cool, andyrock said he'll take a look [20:47] Trevinho, yeah [20:48] Trevinho, andyrock said he knew whats causing it, so with that fix + your fix (should be a yay!) [20:48] jasoncwarner, sure === Ursinha-afk is now known as Ursinha === Ursinha is now known as Ursinha-afk [21:00] Thanks robert_ancell [21:00] Trevinho, let me know when you have a branch to test [21:00] robert_ancell: indeed [21:01] Hey, i have a solution to respawning compiz locked again, using upstart. [21:01] so when compiz crash, instead of just respawning compiz we add a second job to respawn compiz and lock the screen. [21:01] this should protect from any crashes caused to lock screen WM. [21:01] (in the x11 world) [21:02] except for the ones that introduce an infinite loop you can't break out of because compiz crashing results in the screen being locked, when you try to unlock it :) [21:03] Trevinho, robert_ancell, jasoncwarner I should be able to test as well [21:03] just let me know if you want any help === psivaa-afk is now known as psivaa === Ursinha-afk is now known as Ursinha [21:36] rickspencer3, https://code.launchpad.net/~andyrock/unity/fix-1308572/+merge/216216 [21:36] and, nice [21:37] andy, I guess that fixes the crash, right? [21:37] sorry about that that... but sometimes nux is just stupid [21:37] rickspencer3, yep [21:38] rickspencer3, testing it right now :) [21:39] nice [21:39] nice andyrock fingers crossed ;) [21:40] can we get it in for the final image? [21:41] jasoncwarner: thanks for the quick approval ;) [21:42] np Sweetshark [21:42] andyrock I think we are going to hold up image for this fix [21:42] infinity fyi, fix is coming...testing to commence. [21:42] jasoncwarner, we still need Trevinho fix :) [21:42] bschaefer: yeah [21:43] excellent work guys [21:43] desrt: wasnt it you saying: I wont support the lib on a bazillion platforms. Sit down and take a look at this in amazement: http://www.openbsd.org/cgi-bin/cvsweb/src/lib/libssl/src/apps/asn1pars.c?rev=1.15 [21:43] rickspencer3, jasoncwarner, robert_ancell: this is the other: https://code.launchpad.net/~3v1n0/unity/relocks-on-crashes/+merge/216221 [21:43] * bschaefer needs a faster computer... [21:43] bschaefer: ^ [21:43] Trevinho, awesome [21:43] jasoncwarner: Shiny. Whatever fix occurs for the rexec issue, do let's make sure it doesn't do anything silly like lock after login or other such failure modes one might think about. :) [21:43] andyrock, we should work hard to get this fix and the fix that protects against other crashes as well ready today/tonight so that we can respin and get it into the final image [21:43] * bschaefer gets that compiling [21:44] Sweetshark: beautiful [21:44] Trevinho, bschaefer, andyrock great work guys! [21:44] Sweetshark: and not too dissimilar to a mess that i had to recently unravel [21:45] Trevinho: OnUnlockedSCreen with a capital C? :) [21:45] infinity: yeah I've fixwed that locally, pusjhing in a sec [21:45] Trevinho: Is there a cleanup here, or if I crash my laptop while locked, will a reboot lead to login->lock->login? [21:45] * Trevinho has evidently typo issues :P [21:46] infinity: I was thinking that just now... [21:46] desrt: I assume so. Its been a while since I last dared to look at our main(). [21:46] infinity: mh, no.. so maybe it's better to save in temp... at that point [21:47] also if temp can be configured not to be cleaned up.. [21:47] desrt: slightly unnerving that there is still a comment saying "/* hmm... this is a little evil but it works */" in openssls main ... [21:48] Trevinho: /tmp is usually not the right answer, security-wise, cause then you need to get all mktemp-happy and audit that whole mess. User's ~/.cache or whatever is less error-prone. [21:48] Trevinho: Especially if you're relying on it being a well-known-name on restart. [21:48] andyrock, woot no crash! [21:50] rickspencer3, andyrock branch works for me, and is approved (off to test Trevinho) [21:50] nice! [21:51] infinity: yeah, I would prefer to avoid that... Mh, so... The only thing we can do is probably adding it to the startup script [21:53] oh, I was just about to ask the same question as infinity (about cleaning up on crash) [21:53] Trevinho: Yeah, cleaning the semaphore in /usr/share/upstart/sessions/unity7.conf would likely do the right thing. [21:53] wouldn't /run/user be a better place? [21:53] (rather than /tmp or ~/.cache) [21:54] chrisccoulson: mh, oh, right... that might be probably better [21:54] chrisccoulson: Oh, indeed, I keep forgetting we have that. [21:54] That would solve needing to clean the semaphore on system restart. [21:54] right [21:54] * Trevinho moves there [21:54] nice :) [21:54] bschaefer andyrock nice. looks like #1 is all set ? [21:55] jasoncwarner, yup, and Trevinho is hammering out the finishing details on #2 [21:55] awesome, great work everyone! thanks for jumping on this. [21:59] Trevinho, nice :) [21:59] Trevinho, working here for the #2 case [22:00] bschaefer Trevinho already? [22:00] what kind of black magic is this? [22:00] o/ [22:00] jasoncwarner: now should be done [22:00] sometimes I miss you guys [22:00] thumper, :) [22:01] * bschaefer has done nothing [22:01] but tested [22:01] haha [22:01] thumper: you should always miss those guys ;) [22:01] * thumper is feeling nostalgic [22:01] thumper: <3 [22:01] :D [22:01] jasoncwarner: how goes the home gym purchase? [22:02] thumper: slow...have to slowly convince the wife it's a good idea (aka, her idea) [22:02] haha [22:02] robert_ancell: , seems both should be fixed. mind looping in and helping to verify? [22:02] jasoncwarner, still compiling the first one... [22:02] https://code.launchpad.net/~3v1n0/unity/relocks-on-crashes/+merge/216221 [22:02] infinity: FYI, looks like #1 and #2 are both fixed. bschaefer verified, looping in more people to test [22:02] robert_ancell, i was lucky to have half compiled unity branches :) [22:03] jasoncwarner, always a good idea [22:03] bschaefer, I was wondering what you had! [22:03] robert_ancell, haha [22:04] * bschaefer reboots [22:05] chrisccoulson: the runtime dir is cleaned up also on login/logout, right? [22:06] Trevinho, that's a good question. does upstart manage this? [22:06] I wouldn't expect it to be, but does it matter? [22:06] The main case we care about is system crash, when /run will obviously be wiped. [22:07] The screen won't be locked during logout. [22:08] chrisccoulson, wgrant: well probably it won't happne if you're also logged in a nother tty or ssh... [22:08] Right, that's why I doubt it happens at all. [22:08] wgrant: I care about it because if in that case an user goes to tty1, then kills gnome-session at next login it will get a double lock screen [22:08] wgrant: both lightdm and unity lockscreen [22:09] wgrant: cleaning it up on unity destruction should help, but not if there's another crash there (hard killing the session?) [22:13] Trevinho: Well, if they kill gnome-session then they deserve what they get? :) [22:13] Yeah, that's a weird case to cater to. [22:13] wgrant: eheh ok :) [22:13] Real users have machines crash. [22:13] just deleteing the marker on unlock should be enough no? worst case they get relocked, and when they unlock it gets removed and they are good [22:13] It seems most important to fail as safely as we can. [22:13] They don't kill random processes. [22:13] * Trevinho tries to things to corner cases as well [22:13] think* [22:14] if they randomly kill compiz and survive that, they get locked, and can unlock [22:14] well it'll only happen if they were in the lockscreen [22:15] that file is only generated when in the lockscreen [22:15] Right. [22:15] So, /run/user is absolutely the right thing here, since it also can't go wrong if your /home goes readonly, etc. [22:15] apw: it depends how they kill it, since on sigterm it will gets cleared [22:16] the only edge cause i can think of is restarting while in the lockscreen [22:16] The only weird failure mode is if you actually start randomly killing processes for kicks, and then you get two login screens (maybe), not world-ending. [22:16] which putting it in /run/user will fix that [22:16] hey all, robotfuel is going to test the fixes as well. robert_ancell is probably still compiling unity. robert_ancell and robotfuel could you let me know when/what you find? [22:16] bschaefer: Reboot while locked is handled by /run clearing on reboot [22:16] infinity, well unity-greeter shouldn't start up? [22:16] will do. I'm at 60% at the moment [22:16] wgrant, cool, were there any other edge cases? [22:16] bschaefer: Depends on when you went r/o. ;) [22:16] infinity, right... lockscreens are interesting haha [22:17] Anyhow, it seems the advantages of /run wildly outweight the potential benefits of ~/.cache [22:17] outweigh, too. [22:18] bschaefer: if you want to do a new test, btw... I've pushed the cleaned up version [22:18] Trevinho, sweet, will do! [22:19] bregma, yay you've returned! [22:28] Trevinho, robert_ancell, everyone else thx for your help [22:28] olli_: np ;) [22:28] Trevinho, let me know if you want me to test something [22:29] olli_: the branches are lp:~3v1n0/unity/relocks-on-crashes and lp:~andyrock/unity/fix-1308572 [22:29] Trevinho, \o/ [22:29] Trevinho, works now on a reboot :) [22:29] Trevinho, we need to get a silo for test, not sure if bregma has resolved his internet issues [22:29] Trevinho, do we have .debs? [22:29] ah [22:30] happy to give the silo a spin [22:30] olli_: eh, not yet I believe... bregma will set it up in minutes I guess [22:30] olli_, i've tested both branches, and confirmed them working [22:30] robert_ancell, is also testing [22:30] bschaefer, alrighty [22:30] 72%... [22:30] robert_ancell, you should always have a nice (~5GB) unity build ready to compile a new unity branch! (joking) [22:31] bschaefer, heh :) I guess it's nice to have a complete build from scratch to be 100% sure everything is working [22:31] robert_ancell, this is very true! [22:32] Trevinho, bregma wanted either you or I to poke robru to get a silo [22:32] Trevinho, it makes it easier for others to test as well, from a ppa [22:34] Trevinho, code looks good as well, branch approved [22:35] bschaefer: cool [22:35] bschaefer, i'm back now, silos for all! [22:35] robru, \o/ [22:36] robru, would you be able to place these two branch in them, and start the build? [22:36] lp:~3v1n0/unity/relocks-on-crashes and lp:~andyrock/unity/fix-1308572 [22:36] bschaefer, got some branches for something? I hope you're not wanting to land something in desktop trusty... [22:36] bschaefer, yeah I can help you with that. you know bout the spreadsheet right? [22:36] robru: yes, they do and yes, they shoudl ;) [22:36] robru: we have [22:36] robru, we are un fortunately, critical security issues [22:36] robru, yeah, but i don't think i hvae the power! [22:36] robru: these are critical security things in the lockscreen. [22:36] Yeah XDG_RUNTIME_DIR is what you want [22:37] jasoncwarner, ok ok I'm on it [22:37] It lives as long as the user is logged in somewhere [22:37] robru, then you very much! [22:37] thank* [22:37] and will be cleared on a reboot or full logout [22:37] bschaefer, do you have merges for those branches? I need merges [22:37] robru, yeah let me get them [22:37] https://code.launchpad.net/~3v1n0/unity/relocks-on-crashes/+merge/216221 [22:38] https://code.launchpad.net/~andyrock/unity/fix-1308572/+merge/216216 [22:38] robru, ^ [22:39] bschaefer, thanks. [22:39] robru, thank you :) [22:39] bschaefer, jasoncwarner Trevinho : ok, got those building in silo 8: https://ci-train.ubuntu.com/job/landing-008-1-build/44/console [22:40] sweet [22:40] jasoncwarner, is the release team aware of this? [22:40] olli_: then, if you want to test they will be there in some minutes ^ [22:40] robru: infinity is looped in [22:41] jasoncwarner, ok great, so when I ask him to accept it he won't be surprised ;-) [22:41] We're all eagerly awaiting it. [22:41] robru: shouldn't be ;) robert_ancell and robotfuel are also testing when they can. robert_ancell is trying to build unity...probably at 78% right now ;) [22:42] 79 [22:42] I am compiling as well [22:42] robert_ancell: unrelated, but why we don't get the PropertyChanged signal from lightdm https://code.launchpad.net/~andyrock/unity/lp-1281058/+merge/215331/comments/513206 ? [22:43] robert_ancell: if you want to speed things up (and you're not building the debs), just build things inside build/plugins/unityshell and build/UnityCore [22:43] Trevinho, I guess it was never implemented - please file a bug [22:44] robert_ancell: ok, fine... Also, I've noticed that when the unity_Greeter is already there (i.e. you call .Lock on display manager, then you move with ctrl+alt+fX to the prev session, you unlock and back in, then again you recall .Lock), then unity don't get the Locked signal [22:45] robert_ancell: thus we don't lock... [22:45] robert_ancell: gdbus monitor said it's not sent to anyone... [22:45] Trevinho, this is Locked from logind? [22:45] robert_ancell: yes [22:46] Trevinho, interesting. Please file a bug and we can make a test for that and check it does the right thing [22:46] robert_ancell: ok, in lightdm or what? [22:46] Trevinho, yes, put it in lightdm [22:52] hi everyone. I have to step out for about 15 minutes, just fyi. [22:58] Trevinho, is it done? [22:59] olli_: still building at https://launchpad.net/~ci-train-ppa-service/+archive/landing-008/ [22:59] olli_: amd64 is doing the pkgs btw, so it's matter of few minutes [23:00] robert_ancell: here you are some details [23:00] https://bugs.launchpad.net/lightdm/+bug/1308789 [23:00] Launchpad bug 1308789 in Light Display Manager "logind Locked signal is not emitted when (unity_)greeter is already running" [Undecided,New] [23:00] Trevinho, thanks [23:02] robert_ancell: same happens with g-s (double checked now), so it's not unity fault (and it would be weird as we don't get the signals) [23:04] robert_ancell: the only thing that I see differnt with g-s is that I also get /org/freedesktop/login1/session/c2: org.freedesktop.DBus.Properties.PropertiesChanged ('org.freedesktop.login1.Session', @a{sv} {}, ['IdleHint', 'IdleSinceHint', 'IdleSinceHintMonotonic']), do you know what is that related to? [23:05] Trevinho, I think that's the same as the idle that comes from gnome-screensaver. Though I don't know if anything populates it [23:05] jasoncwarner, olli_: packages are finally at https://launchpad.net/~ci-train-ppa-service/+archive/landing-008/+packages for your tests [23:07] bschaefer, Trevinho, robert_ancell, robotfuel, do we have a testplan we are using to validate this one [23:08] Trevinho: there are actually two fixes in there [23:08] err self-ping [23:08] olli_: there are actually two fixes in there [23:09] olli_: the first one is: it should not crash when keeping enter pressed in the lockscreen [23:10] olli_: then, the other fix is about making sure that if unity crashed when in lockscreen, then it relocks when run [23:10] olli_: so for this one, just run your normal session, lock it [23:10] olli_: go to tty1 and do something like killall -9 compiz (the -9 is important, otherwise it will not be like a crash) [23:11] k [23:11] olli_: go back to tty7 and unity should be reloading, but locked [23:12] Trevinho, try sleep 5; killall -9 compiz [23:12] and go to 7 to watch [23:12] andyrock, bschaefer, also confirm lp:~andyrock/unity/fix-1308572 fixes the crash [23:12] didn't recover for me [23:12] olli_: further test would be doing the same, but after killing it, reboot... And not-experiencing the 2nd lock after lightdm (quite unlikely as it's ensured by the kernel, but... just to be sure) [23:12] olli_: oh, yeah indeed.. you can just run a sleep 5 && killall -9 compiz, then Super+L [23:13] olli_: weird I didn't mention that since it's the way i tested it :D [23:13] Trevinho, does https://launchpad.net/~ci-train-ppa-service/+archive/landing-008/+packages contain both lp:~andyrock/unity/fix-1308572 and lp:~3v1n0/unity/relocks-on-crashes? Do you want confirmation on the latter? [23:13] Trevinho, I kicked it off on tty1 [23:13] and it the lockscreen didn't come back [23:13] robert_ancell: yes it has both [23:14] olli_: you mean, once you're back on tty7 that unity is running normally? [23:14] no, it's not [23:14] Trevinho, go to tty1 [23:15] sleep 5; killall -9 compiz [23:15] in a clean / newly started session [23:15] that works as expected [23:15] however, if you do it a 2nd time without login in I don't get the lockscreen [23:16] iow, compiz can't crash multiple consecutive times [23:16] olli_: that's because it doesn't get reloaded twice maybe? :o that should be upstart doing it... mhmh [23:17] Trevinho, even when I log in between the 2 tries I am not getting the lock screen on the 2nd try [23:17] olli_: let me test, but I think it's not unity's fault, mostly it doesn't get reloaded more than a certain limit I think [23:18] jasoncwarner, ^ [23:19] uh, guys? I just tried silo 8 in a VM... after installing the updated unity, it doesn't restart after a crash, locked or not, killall -9 makes it go away permanently. [23:19] not quite out of the woods yet? [23:19] jasoncwarner, gonna test more [23:20] jasoncwarner, I think the security issue is fixed [23:20] robru: this is just weird, since we didn't change anything that should be related to that :o mhmh [23:20] i.e. can't crash Unity anymore by keep pushing Enter [23:21] olli_, yeah, I couldn't reproduce the Enter key crash in an unpatched VM either. [23:21] olli_: acked [23:21] jasoncwarner, what's left is that 2 consecutive crashes of Unity while being locked doesn't bring UNity back [23:21] olli_: ok, thanks for the clarification [23:21] so, you can argue whether the latter is a ship stopper still [23:22] I think we need to keep pushing atm [23:22] but got to be aware that the security aspect is solved [23:22] robru, interesting, did you keep pushing Enter for like 30sec? [23:22] olli_, ok, just confirmed, same. the first killall makes unity come back locked, second killall makes unity not come back. [23:22] I've seen thatbefore thischange [23:22] olli_: I think this problem isn't a new thing btw... I don't use unity run by upstart much, but I noticed it doesn't restart always... [23:22] olli_, i held down the enter key for one full minute. [23:22] I could only crash it maybe 2 or 3 times before it stopped autorestarting. [23:23] But the Enter crash is reliable within 30s or so [23:23] robru, that makes for sucky repro testing ;) [23:23] I'm testing in a VM, it's quite slow, I guess it's a race condition or something. [23:23] let me try the 2 consecutive crash thing on this machine real quick, this is unpatched [23:23] brb [23:23] robru: That's not a new behaviour. We saw the same pre-patch when reproducing the crash in the first place. [23:23] when you guys hold down enter, do you see the lock screen check and recheck rapidly? because it seemed to only do it quite slowly for me [23:24] robru: (ie: it's not a regression that you can't keep crashing it forever) [23:24] Trevinho, olli_, do we strictly need the "keep lock on unity crash" fix? Note that killing gnome-screensaver would break the lock in the existing case [23:24] infinity, ok [23:24] robert_ancell: That's the fix that matters. [23:24] Killing gnome-screensaver would break it, but gnome-screensaver is simpler, well-tested code. [23:24] There will be other crashers. [23:24] robert_ancell: The crash isn't the security bug, the fact that it's unlocked after a crash is. [23:25] We need to prevent the crashers from becoming security bugs in the first place. [23:25] infinity, and that seems to be fixed - sorry I dropped just when you started [23:25] sure, it just feels like this is a bit of an enhancement over the status quo rather than a fix. Though of course the likelihood of unity crashing is higher than gnome-screensaver. [23:25] killall compiz unlocks the screen. [23:26] robotfuel, under which circumstances? [23:26] robert_ancell: Enhancement is to be encouraged. :P [23:26] A non-crash exit should unlock the screen. [23:26] infinity, this close to release? :) [23:27] olli: wait for the screen to to dim and lock, switch terminals to tty1 login and type killall compiz [23:27] robert_ancell: For a security hole? Yes. [23:27] robotfuel, will try [23:27] robotfuel: The crash case is the interesting one; killall will send SIGTERM for a clean shutdown by default. [23:27] robotfuel: A sigkill isn't a security hole. [23:27] unity should be cleaning up on SIGTERM. [23:27] Trevinho, ^ what robotfuel says [23:27] Or term. [23:27] SIGKILL or SIGSEGV should both work fine for reproducing this. [23:28] olli: A TERM isn't a crash. [23:28] don't we send a SIGTERM to unlock, isn't that normal [23:28] But SIGSERGV isn't the interesting one. [23:28] infinity: I think what robert_ancell is asking, if I read it right, is would we stop the release for the same behavior in gnome-screensaver (which, I believe, has done this for quite some time as well). again, if I read that right [23:28] SIGSEGV *is* the interesting one. [23:28] infinity, ah, didn't get that robotfuel missed -9 [23:28] jasoncwarner: If there was any evidence that gnome-screensaver had crashed at some point in the last 7 years, sure. [23:28] jasoncwarner: Turns out that compiz and unity do a lot. [23:30] infinity: maybe I'm misunderstanding. perhaps someone could rephrase the current issue? As I understand it, the security hole is fixed....but.... [23:30] olli ^ [23:30] jasoncwarner, I think infinity is arguing that killall -9 is not valid [23:30] SIGKILL vs SIGSEGV [23:30] jasoncwarner, infinity ok guys, well, the package as built in silo 8 contains the behavior that unity restarts locked after a crash, I guess that is what we desire? let me know if you want me to hit 'publish' [23:30] wgrant, robert_ancell: gnome-screensaver? it's not anymroe in trusty (unless you don't have some a11y settings on)... [23:31] wgrant: unity *does* cleanup in SIGTERM [23:31] I know, and I know. [23:31] Trevinho: Right. Which it should. [23:31] gnome-screensaver's reliability is important, because it's what we have to measure against. [23:31] Trevinho: The point is that people were (incorrectly) using TERM to fake a "crash". :) [23:31] The point is that SIGKILL, SIGABRT, SIGSEGV are all good for testing [23:31] SIGTERM is not. [23:32] infinity: ah, indeed... but SIGKILL is fine, sin't it? [23:32] Yes. [23:32] thanks olli [23:32] wgrant: I underlined that on the test-case I wrote above... :) [23:32] robru: That is what we want, yes. Both the lock-on-crash and, optionally (but highly desired) the fix for the crash itself. [23:32] infinity, I tested with killall -9 [23:32] i was also testing with -9 [23:32] and couldn't get into unity [23:33] now doing a -11 also doesn't get me into unity [23:33] even though it takes way longer for compiz to die [23:33] back to the lockscreen though [23:33] That's correct. [23:33] -11 will invoke aport, so it will coredump. [23:33] == slow [23:35] alrighty, so do we have agreement that the security issue is sufficiently addressed? [23:36] infinity, ^ [23:36] Have we etseted that edge cases such as system crashes while locked don't require double login? [23:36] But it sounds like it works, and the code looks fine. [23:36] wgrant: it needs if you use your tty powers... :/ [23:36] wgrant: not if you reboot [23:37] I think once we agree that the sec issue is addressed I'd do some more general testing [23:37] We just need to make sure it's not regressing anything, given how late this is. [23:37] just to make sure we didn't catch a nasty regression [23:37] ack [23:37] Right. [23:37] ok guys here's the thing [23:37] I did some guest session test, but multiple user tests might be interesting. [23:37] Trevinho: Right, but we can't really avoid that, and it's not a normal user case. [23:37] Power failures are a normal user case :) [23:37] so doing killall -9 makes it relaunch locked the first time, but then the second time it doesn't relaunch at all. that's supposedly expected [23:37] Right [23:37] but eg I was testing this with a terminal open [23:37] wgrant: yeah, indeed :) [23:38] and when "unity doesn't relaunch", it leaves the terminal open [23:38] Okay, has anyone also done a fresh reboot with this unity to make sure it doesn't do anything silly on first login? [23:38] The failure to restart the second time is not a regression [23:38] It is a security vuln, but that's probably harder. [23:38] I see no way from the code that it would, but. [23:38] infinity, I did [23:38] olli: Thanks. [23:38] can't type in it, but I can copy & paste with the mouse in the terminal. so effectively I have access to the terminal to issue commands in a limited way. still seems like a security hole to me [23:38] will do again just to be sure [23:38] eg if i'm an attacker and i have the ability to take down compiz, all i have to do is do it twice... [23:38] It would be nice to track down the double restart issue [23:39] Trevinho, ^ [23:39] this thing about "unity doesn't relaunch" I think it's something related to wront config from gnome-session/upstart since it doens't seem to be managed by upstart at pstree level, but maybe it's just because upstart runs throught g-s? [23:39] But I don't think it's absolutely essential, as it requires a triggerable crash to exploit, and is possibly awkward to fix now. [23:39] But it should certainly be investigated. [23:39] ok [23:40] well I'm happy to publish this unity then, do we all agree it's a good idea? [23:40] * Trevinho abstains, but would say +1 :P [23:40] robru: Do it. [23:40] ok [23:40] Agrred. [23:40] robru: +1 [23:41] oh lol, still building on armhf and powerpc... [23:41] Trevinho, why, just because it's like... 2am [23:41] :) [23:41] powerpc? did someone say powerpc? don't even get me started..... [23:42] robru, can yuou pls update the bug with your latest finding re terminal [23:42] olli, sure [23:42] jasoncwarner, I'll update the incident report to document the decision [23:42] thanks olli [23:42] infinity, is there anything else left to do? [23:42] for today [23:42] I think we need to follow up on the double crash [23:42] asap [23:43] jasoncwarner: I'll hook you up with (much) faster ppc buildds after release. [23:43] infinity: :) [23:44] Trevinho andyrock robotfuel robert_ancell robru infinity thank you all for working this so late in the game (and day for some!). [23:44] awesome to see how people rallied around it....not that we like seeing it b/c of a bug, obviously ;) [23:44] jasoncwarner: np, it's mostly my normal work time :) [23:45] Trevinho: ah, to be young again ;) [23:45] or...perhaps....not old :) [23:46] jasoncwarner: not that young... unfortunately :/ but time is not my friend [23:46] Trevinho: :) indeed! [23:47] jasoncwarner, hah, not even 5PM here ;-) [23:47] jasoncwarner, but you know that! [23:49] olli, https://bugs.launchpad.net/ubuntu/+source/unity/+bug/1308572/comments/6 [23:49] Launchpad bug 1308572 in unity (Ubuntu) "Ubuntu 14.04: security problem in the lock screen" [Critical,In progress] [23:49] thx robru [23:49] you're welcome [23:49] robru, can you pls ping me once you have hit release [23:49] olli, sure [23:49] thx [23:54] robru: That ppc build should be done in ~5m or so. Sorry, if I'd known the build was in flight earlier, I would have aimed it at sagari. [23:54] infinity, no worries.