=== flux__ is now known as o === o is now known as Guest41307 === alan_g is now known as alan_g|lunch [13:30] tsdgeos, "patches in the bug tracker have an history of not being seen" :( [13:30] tsdgeos, thanks for tip. You say "add some reviewers" -- do I have to know names? [13:32] mterry: git log/blame [13:32] tsdgeos, heh fair [13:32] tsdgeos, where do I push my changes to to create a new review ticket? [13:33] git push ssh://YOUR_GERRIT_USER@codereview.qt-project.org:29418/qt/qtdeclarative HEAD:refs/for/BRANCH [13:33] BRANCH being 5.6 probably [13:33] i don't think they'll accept any fix for 5.5 atm [13:33] tsdgeos, makes sense. OK thanks! [13:33] tsdgeos, see! This is what I meant about being deep into Qt :) [13:38] tsdgeos, does anyone have push rights? [13:38] mterry: "yes" [13:38] that only creates a merge request [13:39] not push to the branch itself [13:39] tsdgeos, but I presumably don't have ssh rights, in order to grab the commit-msg hook? Something like: scp -p -P 29418 codereview.qt-project.org:hooks/commit-msg .git/hooks/ [13:39] I guess I can find a copy on my disk from some other gerrit instance [13:40] mterry: you setup https://codereview.qt-project.org/#/settings/ssh-keys ? [13:41] tsdgeos, yes apparently in the past I did [13:41] then the scp should work [13:41] if the correct key is getting used for that server [13:41] humph. I get permission denied [13:41] otherwise you may not get to push either i guess [13:41] the key in my account matches my id_rsa.pub [13:42] mterry: did you also agree to the agreement in https://codereview.qt-project.org/#/settings/agreements ? [13:42] tsdgeos, "Verified Corporate Corporate contributor agreement" [13:43] looks good [13:43] then no idea [13:44] i can give you the hook if you want [13:44] tsdgeos, but no good if I can't push either. Is there gerrit documentation so I can go over the steps I need to do? [13:45] mterry: https://wiki.qt.io/Setting_up_Gerrit === alan_g|lunch is now known as alan_g [13:55] mterry: wow it has two Corporates in there that is a very Corporate Corporation then :) [13:56] davmor2, spared no expense [14:07] tsdgeos, ok well figured out commit-msg I guess (by downloading separately). But still hit the push "permission denied" problem. According to wiki too, I seem to have everything in order [14:07] i'd say it's the key [14:07] try stracing to make sure the proper key is being tried [14:08] tsdgeos, aha! [14:08] or i guess there's some SSH envvar that will tell you that too [14:08] tsdgeos, I figured it out. it was the key. I was tricked [14:08] tsdgeos, https://codereview.qt-project.org/#/c/147556/ [14:09] tsdgeos, so it's not rude to just volunteer people to review this? [14:10] mterry: it's how it works [14:10] otherwise people won't really have a look at it [14:20] tsdgeos, another gerrit question: if I make a change, how do I get it to update my ongoing review? [14:21] mterry: git commit -a --amend [14:21] basically make it be the same thing [14:21] and then push again [14:21] the change-id thing will make it update [14:23] mterry: did i make sense? [14:24] i guess i did since you reuploaded the change :D === pete-woods1 is now known as pete-woods [15:28] tedg, your symbols file is out of date again [15:30] :-( [15:30] Oh, because I added the find and valid functions. [15:37] mterry: Fixed, also bregma says there are lttng building issues on armhf [15:37] tedg, I haven't noticed failures when building on arm [15:39] dednick, your turn [15:40] mterry: I assume that's because you live somewhere warmer than bregma and he's using build failures to warm his house. [15:40] :) [15:58] Saviq: fresh Vivid setup on LAPTOP, added overlay PPA, then installed unity8-desktop-session-mir [15:58] unity8 doesn't boot, turns out mir-client-platform-mesa3 is not installed [15:58] and neither is mir-platform-graphics-mesa-km7 [16:07] also missing mir-platform-input-evdev4 [16:13] faenil: I guess the ubuntu8-desktop-session-mir metapackage should depend on those [16:13] greyback_: I agree [16:13] greyback_: still can't get unity8 to boot on vivid+overlay, even after installing those [16:13] usc now runs [16:13] unity8-dash says it can't connect to mir server [16:17] faenil: does unity8 start? [16:17] greyback_: yeah "using nested cursor" is the last line [16:17] using kms7, evdev4, mesa-kms 0.18.1 [16:17] faenil: can you move the cursor? Does log-in work? [16:18] greyback_: yes I can move the cursor, just no UI [16:18] no unity8 login [16:18] faenil: can you pastebin the whole unity8.log please? [16:20] faenil: can you also gdb attach to unity8 and past the trace? [16:21] one thing at a time :D [16:21] I can't finish debugging something that I find other bugs, it's a recursive bitch XD [16:21] lol(lol) [16:24] XD [16:26] greyback_: tsdgeos unity8 is not running :| [16:26] crashed, though I can move cursor [16:26] faenil: USC drawing the cursor [16:26] let's see if apport-cli can help [16:27] greyback_: ah ok [16:33] tsdgeos: mmm crash has a not very useful bt, except last call is XGetXCBConnection [16:33] which could still be a nice hint [16:33] faenil: yeah wrong qpa [16:34] tsdgeos: missing config pkg? [16:34] faenil: no idea, why are you trying vivid+unity8-desktop? [16:34] is that a config we care about? [16:34] to debug text rendering issues [16:34] we have issues on xenial [16:35] then fix xenial :) [16:35] so I was setting up vivid to see if it's a regression [16:35] tsdgeos: you fix it? :) [16:35] tsdgeos: we're trying to understand what's wrong... [16:35] ok [16:37] faenil: just make sure unity8 is using the correct qpa [16:37] you can probably just force it somewhere [16:37] tsdgeos: sounds like a workaround :) [16:37] faenil: you're just tryuing to see the rendering, aren't you? [16:38] who sets the QPA usually? it's not in unity8.conf is it? [16:38] why would you care it's a workaround? [16:38] tsdgeos: yeah, well, if we fix it for everyone, that's even better isn't it :D [16:38] faenil: as said, i don't think we care for the vivid+unity8-desktop scenario [16:38] i can be wrong though [16:38] okay [16:42] faenil, if unity8 stuck, might be the gst clutter thing again [16:42] bug #1525285 [16:42] bug 1525285 in clutter-gst-3.0 (Ubuntu) "inspecting clutter plugin hangs outside X11" [High,Confirmed] https://launchpad.net/bugs/1525285 [16:42] Saviq: not stuck, crashes [16:43] faenil, ah, that's "easier" ;P [16:44] org.gnome.ScreenSaver and org.freedesktop.Notifications say they can't connect to Mir [16:46] Saviq: do you know who's supposed to set the QPA plugin [16:46] Saviq: he's on vivid, the clutter thing is xenial only? or both? [16:46] vivid+ overlay ppa [16:46] tsdgeos, never tried on vivid [16:48] tsdgeos: thanks for picking up the "crash on logout" bug [16:49] faenil: was just some ldd'ing [16:49] tsdgeos: I reported already who was using what [16:49] they just didn't believe me :D [16:50] After inspection with ldd, I found out that [16:50] - unity8 doesn't explicitly load libprotobuf [16:50] - /usr/bin/telephony-service* (-handler, -indicator, -approver) all link to libprotobuf.so.9 [16:50] maybe they'll believe you :D [16:51] yeah well you missed the important one [16:51] telephony-service without any suffix? alright... [16:52] libtelephonyservice-qml.so [16:52] i.e. not an app but a plugin we load [16:52] i probably should write that in the bug somewhere [16:52] but it's loaded through telephony-service isn't it? [16:52] n [16:52] o [16:52] it's loaded though import Ubuntu.Telephony 0.1 [16:52] telephony-service (though libphonenumber) uses libprotobuf [16:53] alright [16:53] then I misunderstood that [16:53] here's a quickie: what's the best way to force the Unity 8 scaling factor on my desktop? I've been forcing GRID_UNIT_PX in unity.desktop, but is there a better way? [16:53] then what ldd option did you use? or did you recursively ldd the so loaded by unity8? [16:53] bregma, /etc/environment, really [16:54] faenil: basically yeah [16:54] ok [16:55] tsdgeos: I added "initctl set-env --global QT_QPA_PLATFORM=ubuntumirclient" to unity8.conf, no joy [16:56] faenil: i may be wrong, but i think you may need mirserver there [16:56] tsdgeos: how does usc run without it? [16:57] faenil: no idea, you're really talking to the one that has less ideas about this whole stuff [16:57] :D [16:57] libmirserver36 already at the latest version [16:57] tsdgeos: hehe alright :D sorry for bothering then! :) [16:57] I'll resume pestering my dear greyback_ then :D [16:57] he doesn't mind [16:57] :D [16:58] faenil: let's start at the top, unity8 log please [16:59] greyback_: yessir, rebooting already ;) [16:59] although it only shows "loading module" blablabla [16:59] and then "using nested cursor" [17:00] greyback_: http://pastebin.ubuntu.com/14680527/ [17:01] Saviq: do you want me to create a bug about the missing dependencies? [17:01] faenil: Saviq travelling atm. I think you should [17:01] nope [17:01] faenil, yes please, we need to find out if unity8-desktop-session-mir is the right place, though [17:02] Saviq: okay, I'll report it as desktop-session-mir and then we'll go from there [17:02] faenil: ok, that implies mir at fault. [17:02] greyback_: always your fault, I knew it [17:02] faenil: I'm not mir ;) Can you attach gdb and see where the crash is? [17:03] greyback_: no time, unity8 dies before I can attach [17:03] faenil: I usually do "sudo gdb -p `pidof unity8`" [17:03] ok, then need manual launch. [17:03] DESKTOP_SESSION=unity8-mir \ [17:03] MIR_SOCKET=$XDG_RUNTIME_DIR/mir_socket \ [17:03] MIR_SERVER_PROMPT_FILE=1 \ [17:03] MIR_SERVER_HOST_SOCKET=/run/lightdm-mir-0 \ [17:03] MIR_SERVER_FILE=$XDG_RUNTIME_DIR/mir_socket \ [17:04] MIR_SERVER_NAME=session-0 \ [17:04] UNITY_INDICATOR_PROFILE=desktop \ [17:04] QT_QPA_PLATFORM=mirserver /usr/bin/unity8 [17:04] oops, sorry [17:04] faenil: first, do "stop unity8" to stop it relaunching continually [17:04] it doesn't seem to be relaunching [17:04] faenil: then, run that command, pasted here again: http://pastebin.ubuntu.com/14680565/ [17:04] upstart might have given up on it [17:05] ok, can I do that from unity7? or do I need tty? [17:05] faenil: should work from unity7 [17:05] tsdgeos, https://code.launchpad.net/~saviq/qtmir/use-qstandardpaths-cache/+merge/282394/comments/721971 [17:05] faenil: as long as usc running [17:11] Saviq: https://bugs.launchpad.net/ubuntu/+source/unity8/+bug/1538659 [17:11] Launchpad bug 1538659 in unity8 (Ubuntu) "unity8-desktop-session-mir is missing mir-platform-* and mir-client-* dependencies" [Undecided,New] [17:11] faenil, thx [17:11] Saviq: just noticed the version of desktop-sessio-mir is really old [17:12] greyback_: and do I need special args to run usc? [17:12] ok, I'll just do from tty [17:13] faenil: yeah. I think you need this: /usr/sbin/unity-system-compositor --disable-inactivity-policy=true --on-fatal-error-abort --file /run/lightdm-mir-0 --from-dm-fd 12 --to-dm-fd 21 --vt 8 --enable-hardware-cursor=true [17:14] faenil: but that can depend on the hardware you've got. You might be better off letting lightdm execute it for you [17:14] tedg, supported orientations parsing isn't quite right [17:14] yeah [17:14] I'll just copy [17:14] your cmd [17:14] faenil: typically in this situation, I let unity8 crash, then "stop unity8" and then run the unity8 command I gave you above [17:15] tedg, if the value is empty, we should return "all". Only if there's exactly one string and it's "primary" should we treat the orientation as primary (which I'm guessing is what none of the flags being set implies) [17:15] ok, yeah I should have just put it in a file [17:16] dednick: hey, have pushed changes to lp:~unity-team/qtubuntu/screen-info/ - I hope it'll make your automatic visibility hack easier (UbuntuWindow has its own formFactor property) [17:18] greyback_: cannot open display [17:18] (unity8:3689) Gtk-WARNING **: cannot open display: [17:18] faenil: is there more output that just that? [17:19] nope [17:19] greyback_: sorry, as in, no other error [17:19] faenil: QT_QPA_PLATFORM=mirserver definitely set? [17:19] I have Found driver mesa-x11, found mesa-kms, Mir server created [17:19] before that [17:20] I'm doing "source cmd" where cmd is a file with a copypaste of your pastebin [17:20] faenil: is "qtmir-desktop" installed? [17:20] so it should be correct [17:20] yes [17:20] faenil: please run previous with QT_DEBUG_PLUGINS=1 and look for it trying to load "mirserver" [17:21] if in a VT, you'll need to pipe into a file [17:21] oki [17:21] greyback_: ta. i'll take a look [17:22] dednick: FYI I don't anticipate any more major changes to that branch now [17:23] greyback_: loaded library libqpa-mirserver, then unload succeeded on the same file [17:23] (grep) [17:24] faenil: it usually says why it unloaded it [17:24] greyback_: there's cannot open display before it unloads [17:24] mterry: K, actually looking at those tests now. Do you think there should be an explicit primary? [17:25] mterry: Or I guess we could do a member function that was primary() that checked the null. [17:25] tedg, I don't think we *need* one -- primary should only be used when no other is set. So it's a reasonable fallback if none are set [17:25] faenil: hmmm. Could you please do: "unset QT_QPA_PLATFORMTHEME" [17:26] faenil: the appmenu-qt5 plugin could be to blame. It is responsible for exporting menus over dbus, but does assume X11 is there. Might explain your error [17:26] greyback_: went further this time...blabla loading module, found drivers--- ERROR: QMirServer ;or faoòed tp start [17:26] QT_QPA_PLATFORMTHEME=appmenu-qt5 <- definitely caused me trouble in the past. [17:26] oops [17:26] Mir failed to start [17:27] ERROR: QMirServer - Mir failed to start [17:27] any reason why printed above? [17:27] that's after unsetting that envvar [17:27] nope [17:27] ok, good. that env var a problem, please elave it unset [17:27] just "loaded module blalba, found driver blabla, Selected driver mesa-kms, then ERROR" [17:28] unusual [17:28] mir usually says why it failed to start [17:29] can give you the log, let me reboot [17:29] isn't there anything to create pastebins from tty? :D [17:29] faenil: can you remove the appmenu-qt5 package before you do? [17:30] greyback_: ok [17:30] faenil: probably, but working in a VT is a pain. I use second machine to SSH into problem machine [17:30] hehe [17:30] yeah I should do the same [17:31] try setting "DISPLAY= " [17:31] as an env var in command line [17:32] greyback_: pastebin 14680756 [17:32] dednick: will try [17:33] meh. it's not complaining about display anymore. so probably not [17:33] faenil: does it take about 10 seconds to give that error? [17:33] dednick: yep [17:34] faenil: when you run unity8, try switching to the VT USC is on [17:34] it's possibly mir timing out waiting for random bits. try moving your mouse around/keyboard mashing while starting. [17:35] when USC not on the active VT, I suspect it hangs and refuses to allow new clients (like u8 here) to connect [17:36] greyback_: tried [17:36] greyback_: segfault now [17:37] mterry, if I install your use-ual-info branch, Unity 8 won't start (it appears to terminate with "what(): AppID is empty") [17:37] mterry, josharenson ^ you had this issue? [17:37] I have 10 secs to gdb attach should be doable [17:37] :D [17:37] bregma, dednick: make sure you have the latest of both use-ual-info and tedg's app-object branch -- they're moving targets [17:37] faenil: ok. Do you really want to spend time investigating? [17:38] greyback_: I have 15mins left, I'll give it that much [17:38] ok [17:38] mterry, build both from a fresh checkout within the last 3 hours [17:38] mterry: huh? [17:39] dednick: tried DISPLAY= , no change [17:39] app-object is fine, as long as LXD is installed (?!?) [17:39] bregma, my statement stands :) tedg changed API recently and I updated to match this morning, I'm not sure when exactly you grabbed it [17:39] dednick, thought you were talking about the other statement above you [17:39] mterry, I grabbed it at 09:46 EST [17:39] bregma, grab again is what I'm saying [17:39] mterry: unity8 stopping because of "QMirServer - Mir failed to start" [17:40] mterry: you and josharenson were getting it at sprint. [17:40] something to do with random data? [17:40] dednick, yes we saw that at the sprint... josharenson may remember how to get out of it [17:40] dednick, oh we did hit that issue... yes we were seeing that error when that happened [17:40] mterry, dednick, that problem can be worked around by generating enough entropy before logging in [17:40] dednick, move your mouse around during boot [17:40] dednick: mterry: I don't remember getting that issue.. I remember display id being empty [17:40] was only on boot? [17:41] josharenson, this error came up due to internal timeouts (30s) [17:41] dednick: Yes, but display id and app id would be different things [17:41] faenil, apt-get install pastebinit [17:41] dednick, we only saw it on boot, but just because we were booting a lot. Could happen anytime [17:41] ah, I actually caught up reading... [17:42] faenil: you tried moving mouse around when starting? [17:42] dednick: yep [17:42] flux__: thanks! [17:42] greyback_: mm I attached to unity8, but when I "continue" I get get thread exits [17:43] no segfaults.. [17:43] could it be something else segfaulting? [17:43] erm. could be something to do with u8 not finding the usc socket. [17:43] faenil: hard to say without being able to try myself [17:43] dednick: mir usually reports an error in that case [17:44] well, otherwise gdb would trap the segfault [17:44] and it doesn't, so it's not unity8 [17:44] faenil: does /run/lightdm-mir-0 exist? [17:44] that USC's socket that unity8 is trying to connect to [17:45] yup, it's there [17:45] i got to run. [17:45] faenil: it may be that since mir fails to start in 10 seconds, qtmir shuts everything down cleanly [17:45] cleanly with a segfault :D [17:45] Ah yes, we only hit the entropy issue because we changed that timeout [17:45] greyback_: yeah, that is where the error comes from. [17:46] faenil: I've also noticed that gdb can delay mir startup so much 10 seconds is not enough [17:46] i saw the code a few days ago [17:47] Unity8's bt is showing MirServerThread::waitForMirStartup [17:47] (before I type "continue") [17:47] "t a a bt" please [17:47] it's not crashing. get the full thread trace [17:47] that's thread 1 [17:47] yeah, it's not crashing yet at this point [17:47] faenil: thread 1 is waiting for a mir thread [17:48] I want to know what all threads are doing [17:48] "gdb$ thread apply all bt" [17:48] * dednick out. [17:48] o/ [17:48] there's a mir_connect_sync [17:49] then start_thread, another start_thread [17:49] a poll [17:49] and that waitForMirStartup [17:49] the one with mir_connect_sync is on a cond_wait [17:50] the other 2 start_thread I don't know, missing debug info of liblttng-ust [17:50] 3 start_thread actually [17:50] one from libmircommon, but again missing dbg info [17:50] I'll have to install symbols first [17:50] mir_connect_sync would be the cluprit [17:51] the last lines there are cond_wait <-- condition_variable::wait <-- ?? in libmirclient <-- mir_connect_sync [17:51] faenil: I think you should try from a second machine, I suspect VT launching might be hurting you. [17:52] faenil: yeah, that command should be super-fast, not blocking [17:52] ok [17:52] that is unity8 trying to connect to USC [17:52] some reason that's blocking, as if USC not responding [17:55] greyback_: segfault in a couple of secs [17:55] (machine is on USC VT) [17:57] faenil: I presume because of qtmir timing out aiting for mir server [17:57] waiting [17:57] so need to see why USC not accepting unity8's connection [17:58] can you watch /var/log/lightdm/unity-system-compositor.log - it should print when a client tries to connect & disconnect [17:58] sure [17:59] Opening session session-0 [17:59] Closing session session-0 [17:59] that's what happens [18:01] greyback_: ps my laptop is on vivid + overlay ppa as well, and it can run unity8 [18:01] so it's not "impossible" :D [18:01] faenil: curious. Perhaps I've a wrong MIR_SERVER_NAME set === alan_g is now known as alan_g|EOD [18:02] faenil: that env var is how USC identifies clients who connect to it. A poor way of ensuring only unity8 connects to it, by USC only allowing a session with a particular name [18:02] "session-0" used to work [18:02] I see.. [18:03] greyback_: Opening session session-0 [18:03] right [18:03] I have the same on my working laptop [18:03] ok, so that seems ok [18:03] so it should be fine [18:04] * greyback_ scratches head [18:04] I'm not really sure what's wrong so [18:05] I installed it on my laptop using ubuntu-desktop-next [18:05] which is outdated [18:05] but maybe it has some key pkg whcih I'm missing? [18:09] greyback_: Jan 27 18:05:57 usertesting-Vostro-3550 kernel: MirServerThread[1946]: segfault at 0 ip 00007f0049fbb6e7 sp 00007f0040c91558 error 4 in libX11-xcb.so.1.0.0[7f0049fbb000+1 [18:10] faenil: we should never see X11. So check QT_QPA_PLATFORMTHEME is definitely unset (or appmenu-qt5 removed) and try to get a backtrace which might tell us where it comes from [18:11] but that's a different complaint to above [18:11] it's a segfault [18:12] sure, but you can get segfault in many different ways [18:12] can only fix one at a time [18:13] heh : [18:13] :) [18:14] I unset the var [18:14] but now it says /run/user/1000/mir-socket doesn't exist [18:15] ah nvm [18:15] bash error [18:16] greyback_: still same segfault as above [18:16] that env var is unset [18:17] faenil: backtrace will help show why [18:18] greyback_: right I haven't tried again since using ssh [18:20] greyback_: http://pastebin.ubuntu.com/14681213/ [18:21] good I get the segfault in gdb now \o/ [18:21] let me grab bt [18:21] faenil: yay! [18:21] nooooooo ssh dropped [18:21] :'( [18:22] ok back [18:22] phew [18:22] faenil: one tip, if you get a SIGILL while starting a mir server (like unity8, it's normal. To skip, do "handle SIGILL nostop" [18:22] http://pastebin.ubuntu.com/14681229/ [18:22] greyback_: ok, thanks [18:23] hmm, so it's via EGL [18:23] I used to "kill -SIGCONT unity8" :) [18:23] to make unity8 boot [18:23] ok, that means the version of MESA installed on your machine is missing mir specific patches [18:24] Installed: 10.5.9-2ubuntu1~vivid2 [18:24] faenil: possible the mir patches are not in vivid+o at all [18:25] you'd need to ask the mir guys [18:25] greyback_: how comes it runs on my laptop then... [18:25] dunno [18:26] greyback_: my laptop has the same version [18:26] faenil: can you set "EGL_PLATFORM=mir EGL_LOG_LEVEL=debug" and try unity8 once more [18:26] we're getting into territory where I'm not an expert, you'd need mir team to help here [18:27] greyback_: UI! \o/ [18:27] omg, phone wizard UI, alright...better than nothing :D [18:28]