[06:01] hi [08:05] Buenos días! [08:10] moin! [08:11] 'sup, pitti + hikiko? [08:11] hi desrt [08:12] happy thursday :) [08:13] happy thursday :) [08:13] * desrt medicates herself [08:13] mmm. coffee. [08:14] lol [08:16] hey hey desrt! [08:16] hello hikiko [08:16] hi pitti [08:16] happy thursday [08:17] pitti: why are you in spain? [08:27] hey desrt hikiko pitti [08:27] good morning desktopers [08:28] hey seb :) [08:29] hi seb128 [09:01] Sweet5hark, hey, bdmurray pointed you https://errors.ubuntu.com/problem/3f5546617f0b197529d734bee9ae770fb485b92d as being reported only against the yakkety SRU version, could you check if you think that might be a regression? [09:01] hey willcooke [09:03] HO BOY [09:03] it's dark today [09:03] hey Laney [09:03] you got confused, it's 3am, go back to bed :p [09:04] what up seb128 OG [09:04] oh sounds good [09:04] :-) [09:04] bonjour seb128, 'sup seb128 [09:04] err, Laney [09:04] lol [09:04] it's not sunny here either today [09:04] but less cold, which was nice for tennis yesterday [09:04] not just not sunny [09:04] * pitti appreciates the sun outside and puts on a smug face [09:04] there must be a lot of cloud [09:04] glooooooooooomy [09:04] seb128: so the mesa/gtk+ thing, was there a way to trigger it? [09:04] but indeed it's like 12° [09:05] hey pitti! [09:05] are you having fun at the sprint? [09:05] Morning all [09:05] tjaalton, hey, xnox gets it building ubiquity apparently, he might be able to provide you a smaller testcase [09:05] Morning desktopers [09:05] hey davmor2 flexiondotorg [09:05] o/ seb128 davidcalle Laney [09:06] wrong d ;-) [09:06] Oops, davmor2 I mean. But hello to David too :-) [09:08] * Laney nods flexiondotorg [09:08] how's life? [09:08] tjaalton: https://objectstorage.prodstack4-5.canonical.com/v1/AUTH_77e2ada1e7a84929a74ba3b87153c0ac/autopkgtest-zesty/zesty/amd64/g/gtk+3.0/20161201_093323_fa84e@/log.gz has a segfault, is that it? [09:08] Laney: guess so [09:09] then running gtk's autopkgtest with the -proposed mesa seems good [09:10] tjaalton, rebuild ubiquity is good enough test, and is relatively quick. [09:10] @flexiondotorg o/ [09:10] davidcalle: Error: "flexiondotorg" is not a valid command. [09:10] * davidcalle blames telegram [09:10] ./debian/rules tests [09:10] tjaalton, ^ if you want to run it interactively. [09:11] xnox, Laney: ok, I'll do that [09:11] tjaalton, just a guess but maybe simply try to start something gtkish under xfvb [09:11] right [09:15] xvfb-run gnome-terminal fails [09:15] so that's easy enough [09:15] to repro [09:21] night TheMuso [09:21] and morning all [09:21] Morning willcooke [09:33] * amd64: python-sfepy, python-tables, python-tables-dbg, python-tables-lib, python3-tables, python3-tables-dbg, python3-tables-lib, spykeviewer, vitables [09:33] so close [09:33] hi willcooke [10:10] SUCCESS (848/0) [10:12] morning desktopers [10:43] Laney, congrats! [10:43] hey ricotz [10:53] seb128: that is ignoring tests, still need to look at those [10:55] seb128: whoops, missed that. [10:57] seb128: hmm, curious: from the stats it didnt happen before, so it might be a regression ... or the update changed the stacktrace enough to make it appear as a "different" issue. [10:59] even more curious it has RecoveryUI::impl_doEmergencySave in the stacktrace, that is: this is during the "libreoffice crashed, do you want to recover you files" dialog. [11:04] the crash dialog is on fire, do you want to put it out and try to crash again? === PaulW2U_ is now known as PaulW2U [12:20] laney, pitti: is 1637758 really fixed? I have same bug again... [12:31] Same bug or same symptoms? [12:32] loginctl shows a lightdm session staying open? [12:32] session stays open [12:32] * Laney shrugs [12:32] don't know why it would have stopped working [12:32] what did you update? [12:33] don't know, but it is fully updated zesty without ppas. also i dont use proposed. [12:33] restarted at least 3 times [12:33] check the logs? [12:34] what logs? [12:35] Also I probably did not restart at least for week so there probably was many updates. [12:35] systemd is 232-7, lightdm is 1.21.1-0ubuntu2 [12:36] Have you tried a virtual machine to see if it happens in there? [12:37] I think that vm works, but will check now. [12:38] don't see it on my fully updated systems [12:43] vm works fine [12:45] no idea atm, sorry [12:46] I reinstalled lightdm, will try to reboot [12:52] still same problem :( [12:56] hm, WFM too, no immediate idea [13:47] pitti: Hi! Is there a way to read session environment variables like XDG_SESSION_TYPE in a autopkgtest ? === hikiko is now known as hikiko|ln [14:00] om26er: I'm not sure what you mean? [14:01] pitti: autopilot relies on XDG_SESSION_TYPE to check which session its running under. So when I run an autopilot test through autopkgtest, it fails as apparently that env variable is not available. [14:02] om26er: right, it's set by lightdm, and nothing logs in via lightdm there [14:02] om26er: your test can of course make up some value [14:02] om26er: I suppose you run them through Xvfb, so "x11" sounds right [14:03] pitti: in this case its run directly on a touch device, so 'mir' [14:17] pitti, laney: lightdm.log show that login1_service_terminate_session was called only for my second seat - seat1. It is not called for seat0 in my case. [14:19] Activating ConsoleKit session c1; Activating login1 session c2; what? why it thinks that seat0 is consolekit session? [14:31] WARNING: Session pid=1529: Error reading from session: Interrupted system call [14:33] pitti,laney: any idea why lightdm fails to get login1_session_id? [14:59] muktupavels: sorry, not out of my head; I have zero experience/knowledge about multi-seat I'm afraid [15:01] pitti: it might not be related to multi-seat. it looks like first season c1 fails to get login1_session_id [15:02] it is null and login1_service_terminate_session is not called [15:27] pitti, laney: for first session login1_session_id is null, but console_kit_cookie is set to c1. all other sessions created has login1_session_id set to c2, c3 and c4. console_kit_cookie for these sessions are null. [15:31] muktupavels: if you're debugging, I suggest you move backwards until you find out why that happens [15:31] ? [15:37] "?"? [15:38] what do you think with moving backwards? [15:38] if you found out that a variable is null when it shouldn't be, go and find out where that comes from [15:38] my problem is that for some reason lightdm fails to get session id and because of that I am affected with that bug again. [15:38] and then find out why that bit doesn't work [15:38] etc, until you find the bug [15:39] n_read = read (session->priv->from_child_output, buf, count); [15:39] that fails no idea what to do next [15:41] will reboot again === hikiko|ln is now known as hikiko [16:16] lol [16:16] apparently I am not for lunch 5hrs [16:16] I forgot to change my status again :/ [16:16] :p [16:34] https://bugs.launchpad.net/ubuntu/+source/init-system-helpers/+bug/1642966 [16:34] Ubuntu bug 1642966 in systemd (Ubuntu) "package cups-daemon 2.1.3-4 failed to install/upgrade: subprocess new pre-removal script returned error exit status 1" [High,Confirmed] [16:34] * flexiondotorg just popping out for an hour of so. Back later... [16:34] pitti - can you advise on who a suitable systemd expert is who might be able to help tkamppeter get moving on that ^ [16:35] that = bug, not flexiondotorg popping out :) [16:49] pitti, laney: https://code.launchpad.net/~albertsmuktupavels/lightdm/retry-interrupted-read/+merge/312845 [17:01] muktupavels: lightdm turns off SA_RESTART? [17:01] interesting that it's so reproducible for you [17:01] will let robert_ancell review [17:02] I don't know, but with that patch now my session works [17:02] first reboot that worked today. [17:04] interesting [17:04] it's a difficult area [17:04] well done for finding it [17:04] http://bazaar.launchpad.net/~lightdm-team/lightdm/trunk/revision/2406 [17:04] any chance that this has something to do with my problem? [17:05] https://bazaar.launchpad.net/~lightdm-team/lightdm/trunk/view/head:/src/process.c#L460 [17:06] I would guess that it's this bit [17:06] tjaalton, so, from that mesa bug, are you gonna upload the patch that is proposed on that upstream bug tracker? [17:06] if that had SA_RESTART too it would restart the read() instead of giving EINTR [17:06] imho it's worth uploading that botch into zesty [17:06] that's my hunch --- just from reading the code / guessing [17:06] retrying on EINTR in this case is valid [17:07] assuming that putting the flag there is wrong [17:07] you've got something that can be reviewed now anyway [17:07] xnox: libepoxy uploaded already [17:08] tjaalton, cool! [17:09] 3h ago, tests are slow it seems [17:53] k, going to sing some christmas carols [17:53] o/ [17:55] see ya Laney === marga_ is now known as marga === JanC is now known as Guest38161 === JanC_ is now known as JanC [18:30] dinner time, night all [21:13] robert_ancell: https://code.launchpad.net/~albertsmuktupavels/lightdm/retry-interrupted-read/+merge/312845 [21:14] muktupavels, did you think pittis suggestion was better? [21:14] I don't know. [21:15] muktupavels, I don't know either. I was just going to trust pitti on this one :) [21:15] is there any reason why SA_RESTART was not used when that code was written? [21:15] muktupavels, because I don't know exactly what it does [21:16] muktupavels, did you see this code breaking or is this just precautionary? [21:17] For some reason read is interruped for me every time i start pc. because of that login1_session_id is NULL. [21:17] ah [21:18] muktupavels, is there a bug report to link it to? [21:18] and because that lightdm session is not terminated fully [21:18] I think no. [21:19] because of that 1637758 bug still affects me. [21:20] muktupavels, if you could try the fix in process.c you suggested and that works, I'm happy to land it [21:20] robert_ancell: so you preffer to use SA_RESTART? do you need new/updated merge proposal or you will just push that yourself. [21:21] ok, will test that first [21:21] muktupavels, If you can make the MP that would be great. Since I haven't reproduced it here it would be better if you can say it works. [21:22] * robert_ancell hates Unix signals... [21:23] RAOF, are you familiar with xkbcomp at all? [21:33] robert_ancell: seems to work, branch updated. [21:33] muktupavels, thanks [21:34] muktupavels, can you link to that bug so we can SRU it? [21:34] what bug? [21:35] or you want that I create new bug for this? [21:36] muktupavels, if there's not one please create one, yes [21:47] robert_ancell: https://bugs.launchpad.net/lightdm/+bug/1648637 [21:47] Ubuntu bug 1648637 in Light Display Manager "handle interruped read and write calls" [Undecided,New]