[01:11] mibofra: Yo! === chihchun_afk is now known as chihchun === chihchun is now known as chihchun_afk === chihchun_afk is now known as chihchun [07:45] You are in a twisty maze of classes, all named Surface… [07:53] Look at implementation [07:53] nick anpok [07:53] oops === anpok_ is now known as anpok [08:02] Victory! [08:02] good that was a short one [08:03] In slightly less winning news it seems that my N10 doesn't have enough RAM to build Mir :/ [08:04] we could try gold linker [08:04] where we means you [08:05] Nope; cc1plus gets OOMed. [08:05] oh [08:05] I'll add a bunch of swap, but tomorrow. [08:07] RAOF: That maze is actually much smaller than it used to be :S [08:22] duflu: @"...MIRCOMMON_2. Actually, that's a mistake..." as we've not branched a release why not revert? [08:22] alan_g: Because it's harmless to keep. Both mirplatform and mirserver are getting correctly bumped in 0.7 anyway [08:24] /shrug OK. (It just makes mircommon look less stable than it is) [08:24] alan_g: Yeah I know but doesn't matter. Users of libmirclient for example (/usr/lib/x86_64-linux-gnu/libgdk-3.so.0) only have linkage to libmirclient.so.* and not its dependencies. So only servers are affected and they need rebuilding anyway [08:28] * duflu tries forcing double buffering on to Unity8 to see what happens [08:45] is mirping/mirscreencast segfaulting known, or should I file a bug? [08:45] this is on mako, image 195 [09:00] Chipaca: Doesn't sound familiar. If in doubt then log a bug [09:05] https://bugs.launchpad.net/mir/+bug/1358191 [09:05] Ubuntu bug 1358191 in Mir "mirping, mirscreencast segfault" [Undecided,New] === tvoss is now known as tvoss|afk [10:41] * alan_g wonders why he suddenly sees a load of signed/unsigned comparison errors in test_stream_transport.cpp when cross-compiling === tvoss|afk is now known as tvoss [11:08] alan_g: hm wrong/outdated gtest headers [11:08] in partial chroot [11:08] in chrott/usr/src/gtest or gmock sth.. [11:09] anpok: Hmm. Thanks for the suggestion (I thought I'd updated but...) [11:09] yes but the old one might not have been removed [11:09] at least that was my mileage [11:21] anpok: my mileage too. :) === alan_g is now known as alan_g|lunch === alan_g|lunch is now known as alan_g === pete-woods is now known as pete-woods|lunch [12:52] anyone seeing anything like [libprotobuf ERROR google/protobuf/descriptor_database.cc:57] File already exists in database: mir_protobuf_wire.proto [12:52] in the demo servers? [12:54] kdub: I've seen it in an MP - not reproduced yet [12:54] https://jenkins.qa.ubuntu.com/job/mir-team-mir-development-branch-utopic-amd64-ci/937/console [12:54] yeah, I'm seeing it when starting a demo server, let me see if I can reproduce that test failure [12:55] yeah, I see it there too [12:59] * alan_g wonders why he can't reproduce [13:00] I'm kinda suspecting a protobuf upgrade happened? still chasing that theory though [13:00] oh, nvm, an older revision of mir was okay [13:15] robotfuel, is ~chris.gagnon/+junk/mir-demo-runner the python script that runs the demo programs in CI? [13:16] kdub: no that is the old script [13:17] robotfuel, ah, where's the new one? [13:17] kdub: lp:~mir-team/+junk/mir-medium-test-runner-for-jenkins [13:17] robotfuel, ah, thanks [13:17] I was hunting before you asked :D [13:19] robotfuel, right, but that installs a mir-demo-tester [13:19] kdub: http://bazaar.launchpad.net/~mir-team/+junk/mir-medium-test-runner-for-jenkins/view/head:/mir_install_packages.sh has the distupgrade you don't want ? [13:19] robotfuel, no, I'm just trying to do a post-mortem on a regression bug, and want to know which of the demo servers the jenkins job runs [13:20] kdub: mir_integration_tests mir_acceptance_tests mir_demo_client_basic mir_demo_client_fingerpaint mir_demo_client_eglflash mir_demo_client_eglplasma mir_demo_client_multiwin mir_demo_client_egltriangle mir_performance_tests [13:20] kdub: that's part of the jenkins job [13:21] robotfuel, right, but the mir_demo_client_* surely run against one of the demo servers [13:21] http://bazaar.launchpad.net/~chris.gagnon/+junk/mir-medium-test-runner-for-jenkins/view/head:/mir_install_packages.sh [13:22] kdub: http://bazaar.launchpad.net/~mir-team/+junk/mir-medium-test-runner-for-jenkins/view/head:/mir-mediumtest-runner.sh line 194 [13:23] like, it runs phablet-test-run against a 'mir-demo-tester' http://bazaar.launchpad.net/~chris.gagnon/+junk/mir-medium-test-runner-for-jenkins/view/head:/mir-mediumtest-runner.sh#L178 [13:24] oh, so switching to the same repo :) [13:24] kdub: phablet-test-run -x just executes a command on the phone and gets the return code. [13:24] sure, and the command it runs is this mir-demo-tester [13:25] which isnt part of the mir scripts [13:28] kdub: ah http://bazaar.launchpad.net/~chris.gagnon/+junk/mir-demo-runner/files [13:28] http://bazaar.launchpad.net/~chris.gagnon/+junk/mir-demo-runner/view/head:/bin/mir-demo-tester [13:28] robotfuel, ah, yes [13:28] looks like what I was looking for, thanks === pete-woods|lunch is now known as pete-woods [14:03] kdub: the protobuf error - do you get that on mesa or android? (Or both) Default build options? [14:05] alan_g, i got it on android, but interestingly the process of bisecting fixed it even on tip [14:05] so maybe some sort of problem mixing-and-matching libraries [14:06] kdub: that was my growing suspicion - but CI ought to be a clean env [14:06] yeah, it should be... [14:06] quite... [14:07] morning === dandrader is now known as dandrader|lunch [15:06] hi kdub [15:54] So it's looking like there's an error in qtmir when a prompt session is put on top of another prompt session. [15:54] qtmir.surfaces: MirSurfaceManager::refreshPromptSessionSurfaces - No Application for prompt session [15:55] greyback_, Any ideas there? ^ [15:55] Or perhaps dednick ^ [15:55] tedg: dednick would be better [15:56] tedg: what are you doing in the trust helper? what does trust session on trust session mean? [15:56] dednick, Dash → Pay UI → Online Accounts [15:57] dednick, So someone purchases something, and before we can process they need to login to Ubuntu One. [15:57] dednick, So the OA trusted prompt session needs to overlay on Pay UI. [15:57] dednick, Which is, itself a trusted prompt over the dash. [15:58] tedg: is it multiple prompt providers or multiple prompt sessions? [15:58] dednick, both [15:58] tedg: we don't support multiple prompt sessions. [15:58] yet [15:59] dednick, When do you expect to? [16:00] tedg: there is no eta on it. there wasn't supposed to be a requirement for it yet. [16:00] tvoss: ^ [16:02] dednick, you mentioned multiple prompt sessions could easily be supported, modulo trust-store support for that === chihchun is now known as chihchun_afk [16:02] dednick, as ted's part does not use the trust-store, we could easily unblock him here and I'll take care of the trust-store side of things [16:02] greyback_, thoughts? [16:05] tvoss: yeah, we can remove the guard against it easily enough [16:05] tvoss: sounds do-able, just requires more careful book-keeping in qtmir [16:05] tedg, so: do you need this? [16:05] tedg: what pid are you providing the online ui? [16:05] OA ui i mean [16:06] tvoss, Yes, we kinda do. [16:06] dednick, I'd have to check with mardy, but the request he's getting is from Pay UI, so I imagine that's the PID he's giving. [16:06] dednick, greyback_ with that: go :) [16:06] dednick, greyback_ obviously, if we find an easier way: even better [16:07] could someone please log a bug against qtmir with the exact requirements? Just to track the task [16:08] K, I can do that. [16:09] tedg, mardy: a prompt session will only be overlayed on a application in unity8. [16:10] dednick, Okay, so we need to add a unity8 task as well then? [16:11] tedg: presumably there's no application related to the OA? it's just the prompt you want? [16:11] dednick, In this case, yes. [16:12] tedg: then the prompt session must be given the pid of the dash in this case (same as the payment prompt session) [16:13] dednick, That's going to be difficult… can you just go up the tree in Mir until you hit an application? [16:15] tedg: if you want something overlayed on an app, we need the pid of the app. [16:15] dednick, I want it overlayed on the prompt provider. [16:15] dednick, greyback_, bug 1358388 [16:15] bug 1358388 in QtMir "Trusted Prompt Sessions should be able to be on a Trusted Prompt Provider" [Undecided,New] https://launchpad.net/bugs/1358388 === dandrader|lunch is now known as dandrader === chihchun_afk is now known as chihchun === alan_g is now known as alan_g|EOD [18:11] kgunn, hi === dandrader is now known as dandrader|afk [18:25] mibofra: hi [18:43] kgunn, after Friday I've done some more debug, and I've found what's of the mir's setup goes in segfault [18:49] kgunn, it seems the spinner [18:49] damn the spinner :) [18:49] mibofra: i think there's a fix for that working its way through the wickets [18:50] uhm [18:50] if you take the devel-proposed image it should most likely be in that [18:50] AlbertA: ^ i'm not making that up am i? [18:53] kgunn, and it seems to be the same reason of the segfault of some of the egl tests [18:53] *the not working ones [18:54] I'm retuning (one second that I put an antenna on the 3G modem) [18:57] ok now I've a connection that can be called a connection [18:58] kgunn, anyway that's the log: http://paste.ubuntu.com/8082137/ [19:00] camako: ^ check that out [19:00] seems like a race [19:00] * camako looks [19:01] mibofra: so tell me again, what hw & sw configuration are you on ? [19:01] (lol, anyway ok) [19:04] kgunn, the hw is a samsung galaxy s3 phone (gt-i9300), the host system is android (cyanogenmod 11). Ubuntu touch is in chroot. The systems shares the hw on the phone. So simply I kill the services (such as surfaceflinger, or wpa_supplicant on the host system) and load the ones in ubuntu, so I route the functions to it. [19:04] In fact, network manager and pulse for example work. When I've to use android simply I do the opposite (kill the services on ubuntu and restart the on android) [19:05] anyway, because the devices are shared, for example, if I setup the connectivity on andorid or ubuntu, I can use the iface with the same ip and setup on both the systems [19:06] Example: android setups a connection to my ap with ip 192.168.1.4, if I do an ifconfig on the chroot system (ubuntu) I see the iface with that ip and I can use ping, wget and more [19:07] Ex. of the opposite: if I run xboxdrv for my wireless xbox controller, under the chroot system, I can use it on android too [19:08] kgunn, was I clear or do you need other informations :) ? [19:10] Simply I'm trying this solution to don't do a porting of ubuntu touch. If this works, more or less also my tablet and other arm7 devices should run ubuntu touch in the same way [19:14] kgunn: umm I [19:14] I'm not sure I don't recall a fix for setup [19:15] AlbertA, just if you're curios to see my problem: http://paste.ubuntu.com/8082137/ [19:16] what's the stack trace for the segfault? [19:19] AlbertA, that's the only output form logcat http://paste.ubuntu.com/8082278/ [19:19] mibofra: I mean run the client with gdb [19:19] ah ok [19:19] then do bt for the stack trace after segfault [19:24] AlbertA, I've to install the debug symbols, just a minute === dandrader|afk is now known as dandrader [19:38] AlbertA, I can't find a package for the debugging symbols of the spinner, I've to recompile all the compositor+spinner to get a copy with the debug symbols? [19:38] mibofra: https://wiki.ubuntu.com/DebuggingProgramCrash#Debug_Symbol_Packages [19:39] thanks [19:47] AlbertA2, I'm pass to the shell gdb /usr/bin/unity-system-compositor-spinner, run with run -m /run/mir_socket because I've the socket there, but I get: Cannot exec /usr/bin/unity-system-compositor-spinner -c exec /usr/bin/unity-system-compositor-spinner -m /run/mir_socket. [19:48] mibrofra: what about just the demo server and clients? [19:48] mibofra: apt-get install mir-demos [19:48] then stop lightdm [19:48] and do a mir_demo_server_basic on one terminal [19:48] and mir_demo_server_egltriangle in another [19:49] started with gdb [19:49] AlbertA2, I've the demos and the clients, lightdm is stopped already. The only client that works is the render_to_fb. The egltriangle and similar go as the spinner in segfault. [19:50] anyway I'll try to debug them [19:50] right, [19:50] so I'm trying to get at the stack trace [19:50] with egltriangle [19:50] just to see what's failing [19:50] there [19:50] ok [19:50] let's try [19:52] AlbertA, Cannot exec /usr/bin/mir_demo_client_egltriangle -c exec /usr/bin/mir_demo_client_egltriangle . [19:52] Error: No such file or directory [19:53] with a gdb /usr/bin/mir_demo_client_egltriangle and into gdb a run [19:53] AlbertA2 [19:54] try without the full path === mibofra_ is now known as mibofra [20:00] ok I saw netsplitted, fantastic [20:00] *was [20:04] uhm AlbertA2 maybe I can solve the gdb problem [20:04] ok === dandrader is now known as dandrader|afk [20:05] ok AlbertA2 I've solved it [20:05] so I've debugged the spinner [20:06] AlbertA2 the stack: http://paste.ubuntu.com/8082638/ [20:06] how about egltriangle? [20:08] AlbertA, that's all the out for the spinner, http://paste.ubuntu.com/8082645/. Ok now I'll try with the egltriangle [20:09] AlbertA2, http://paste.ubuntu.com/8082652/ the same [20:09] umm so it doesnt say who is calling strerror [20:10] I guess then can you try running the client with valgrind? [20:10] I think yes [20:17] AlbertA, the spinner http://paste.ubuntu.com/8082700/ the egltriangle http://paste.ubuntu.com/8082704/, in all this think that the only demo that I can see an output on the screen is the render_to_fb [20:17] mibofra: so you installed the dbg symbols for libmirclient8? [20:18] AlbertA2, I think so [20:18] let's see [20:19] AlbertA2, installed [20:20] well it offers no real clues...:) [20:21] other than maybe the android driver blob is failing [20:21] and it hitting a code path that has bugs in it [20:21] :) [20:22] I mean the sig fault is originated by libcutils which is an android library [20:22] AlbertA2, I've installed them after I've written installed XD [20:22] oh [20:23] and do you get a fuller stack trace? [20:23] the debug symbols for libmirclient8 [20:23] let's see [20:24] AlbertA2, I'm trying with the spinner, from gdb I've the same output as before. [20:26] AlbertA2, valgrind for the spinner but I think it's the same as before too: http://paste.ubuntu.com/8082787/ [20:26] AlbertA2, but I wonder why the render_to_fb demo works [20:26] :D [20:32] well its not going through the client for one [20:32] the client driver context may be having the failure [20:33] what does dmesg say? [20:34] anything related to the display driver or gpu driver? [20:36] AlbertA2, http://paste.ubuntu.com/8082851/ that's the dmesg [20:36] try grepping something [20:36] maybe it will seem more useful for you than me [20:39] AlbertA2, anyway I think, I've xorg working fine, if I use mir as sys compositor with x as described here: http://unity.ubuntu.com/mir/using_mir_on_pc.html can I use unity8 on it? [20:39] (only a curiosity) [20:41] mibofra [20:41] mibofra: if you get the examples working [20:41] then most likely unity8 will run fine [20:42] AlbertA, can they be executed with mir working as compositor of X? [20:42] mibofra: well unity8 acts as a client of mir [20:43] so ok [20:43] well technically a nested server [20:43] ah [20:43] a server that is actually a client of the main system compositor mir [20:47] mibofra: well nothing stands out in the dmesg output... [20:47] ok [20:48] mibofra: I guess the next thing you could do is step through the egltriangle client... to see if you can narrow down where it crashes [20:48] ok === dandrader|afk is now known as dandrader [22:25] AlbertA, anyway thank for the time you spend with me :) [22:25] (maybe you had something more interesting/important to do) === RAOF_ is now known as RAOF