[09:11] greyback: @MirBlob I get the impression that we're not quite in sync but I'm not yet sure where the differences are. Should we resolve in MP comments/IRC/hangout? [09:27] alan_g: no need, am satisfied [09:27] sorry, that thread keeps popping out of my mind [09:28] greyback: thanks. (I still have that niggling feeling - but that's for the future.) [09:28] ack [10:03] greyback: Actually I don't know how to connect clients to USC. It's the trust magic... and even when I get a connection the client exits immediately [10:04] clients have to be fullscreen on all outputs.. [10:06] Opening session egldemo [10:06] Mir chose pixel format 3. [10:06] Using pixel format 4. [10:06] Current active output is 1920x1200 +0+0 [10:06] Can't create a surface [10:06] Closing session egldemo [10:10] duflu: set MIR_SERVER_NAME=session-0 [10:10] greyback: K, ta [10:17] duflu: FWIW you can also use --debug-without-dm --debug-active-session-name for tests with USC [10:40] alan_g: what does debug-without-dm do? [10:41] Nope, still no luck [10:41] Stubs out the display manager related logic [10:41] And it's getting late [10:42] alan_g: You need to pass those to USC or the nested server? [10:42] duflu: USC [10:42] alan_g: Yeah, no luck [10:42] Same failures [10:43] But I did just implement Xmir rootless resizing on desktop and phone. So that's something [10:43] Fullscreen solitare [10:43] FTW [10:44] duflu: if you're running demo server the default debug-active-session-name of "nested-mir@:/run/user/1000/mir_socket" is what you need [10:44] alan_g: Can you email me a full working set of command lines? [10:44] I need to start finishing up [10:45] MIR_SERVER_NAME=session-0 MIR_SOCKET=/run/mir_socket MIR_HOST_SERVER_FILE=$XDG_RUNTIME_DIR/mir_socket mir_proving_server & [10:45] that tends to work for me [10:45] duflu: I'll have to do that later. (Don't have a working USC built at the moment) [10:45] alan_g: No problem. I'm using the default wily install [10:45] Me too [10:46] Guess I should just install USC [10:47] duflu: oh xmir is now happy with the color formats on phone? [10:47] anpok: Trunk git Xmir :) [10:47] I guess we can do a build next week-ish [10:48] duflu: cairo would need that happiness too [10:56] Hmm I used to be able to run USC with "sudo unity-system-compositor --debug-without-dm --arw-file" but now it tries (and fails) to access dbus. [11:01] duflu: I think that "feature" of USC is broken today [11:06] alf_: with display configuration, is "clone" mode something it can express? [11:07] greyback_: yes, the (x,y,w,h) of the two displays in the virtual coordinate space just need to overlap === marcusto_ is now known as marcustomlinson [11:10] alf_: aha yes [11:10] thanks [11:23] greyback_: clone mode is not always ideal though. More modes are possible that we don't have names for yet... https://bugs.launchpad.net/mir/+bug/1395416 [11:23] Ubuntu bug 1395416 in Mir "Clone mode forces all outputs to render at the rate of the slowest one" [Medium,Triaged] [11:23] * duflu -> weekend [11:23] duflu: ack [11:23] enjoy weekend! [11:23] o/ === alan_g is now known as alan_g|lunch === alan_g|lunch is now known as alan_g === ChrisTownsend1 is now known as ChrisTownsend === pixel is now known as Guest30724 === dandrader_ is now known as dandrader [14:55] guys, any idea what can cause http://paste.ubuntu.com/12449481/ on the desktop? [14:57] tsdgeos: not having root? [14:57] alan_g: i am root [14:58] error is [14:58] argument is not valid [14:58] or somethning like that when i translate back to enlgis [14:58] h [14:59] alf_: ^^? [15:00] greyback_ suggests i need to tear down lightdm? [15:01] back in aminute [15:01] tsdgeos: are you running from a VT? [15:03] tsdgeos: also, what exactly are you running? Is this a demo server or USC or unity8 or ...? [15:03] alf_: unity8 [15:09] ok, greyback_ brought me a bit forward [15:09] i get unity8 to segfault at least :D [15:09] segfaults in mir+X11 :S [15:10] http://pastebin.ubuntu.com/12449639/ [15:11] alf_: ↑ [15:13] tsdgeos: Can you install libmirserver33-dbsym? [15:13] on it [15:16] alf_: http://pastebin.ubuntu.com/12449714/ [15:17] want me to install the dbgsym for egl and xcb too? [15:18] tsdgeos: could you start unity8 with MESA_DEBUG=1 EGL_LOG_LEVEL=debug - in case they print anything useful [15:18] sure [15:19] greyback_: nothing interesting i'd say http://pastebin.ubuntu.com/12449736/ [15:20] libEGL debug: Native platform type: x11 (autodetected) <- that's wrong [15:20] tsdgeos: (guess) unset DISPLAY ? [15:21] it's not set [15:21] yeah, was bit of a gamble that [15:21] greyback_: what should it say? kms? [15:23] libEGL debug: Native platform type: drm (autodetected) <- what I get [15:23] tsdgeos: can you please pastebin the whole log [15:23] alf_: that's the whole log :D [15:23] last line == segmentation fault [15:23] EGL_PLATFORM=drm [15:24] tsdgeos: doesn't unity8/mir print anything else before crashing? [15:24] nope [15:24] tsdgeos: does that env var help? [15:24] greyback_: it now says drm, but still crashes :/ [15:25] ouch [15:25] tsdgeos: what graphics chipset has this machine? [15:25] intel+nvidia monster [15:26] tsdgeos: what's the output of ls /usr/lib/x86_64-linux-gnu/mir/server-platform ? [15:26] unity8@xps:~$ ls /usr/lib/x86_64-linux-gnu/mir/server-platform [15:26] graphics-mesa-kms.so.3 graphics-mesa-kms.so.4 server-mesa-x11.so.4 [15:27] tsdgeos: can you move server-mesa-x11.so.4 temporarily out of there please and try again [15:27] same thing [15:27] tsdgeos: move graphics-mesa-kms.so.3 ? [15:28] same thing [15:28] tsdgeos: does a demo server work? [15:29] usc is running [15:29] is that enough? [15:30] tsdgeos: yes, although I would like to try a pure demo server in nested configuration to see if we can get more info [15:30] alf_: what do i run? [15:30] tsdgeos: stop lightdm [15:31] tsdgeos: do you have ssh access to that machine? [15:31] tsdgeos: actually you don't need to stop lightdm [15:31] tsdgeos: just switch to a VT [15:32] i'm in via ssh already [15:32] lighttdm stopped [15:32] tsdgeos: ok, then switch to a VT [15:32] tsdgeos: sudo bin/mir_demo_server [15:33] tsdgeos: did you get a cursor ? [15:33] tsdgeos: (which you can move) [15:33] yes [15:34] tsdgeos: ok, from another ssh session: sudo mir_demo_server --host-socket /tmp/mir_socket -f /tmp/mir_nested [15:34] tsdgeos: and from a third ssh session: sudo mir_demo_client_egltriangle -m /tmp/mir_nested [15:34] is that second thing supposed to return? [15:35] tsdgeos: not immediately, did it output anything? [15:35] http://pastebin.ubuntu.com/12449924/ [15:36] it's crashing the same as unity8 [15:36] just doesn't say [15:36] but if run in gdb i get the same bt [15:36] tsdgeos: can you please paste the log from the first mir_demo_server (the host) [15:37] where do i get it? the VT has the mouse [15:37] is it logged somewhere? [15:37] or should i kill it and start it with &> ? [15:37] tsdgeos: no, use ctrl-alt-backspace to quit [15:38] tsdgeos: then also start it from an ssh session (just be sure a VT is active in the target computer) [15:38] alf_: http://pastebin.ubuntu.com/12449956/ this is the first command output [15:47] tsdgeos: ok, so could you get the backtrace of the second (nested) demo server, with egl debug symbols if possible [15:48] alf_: do i force drm with EGL_PLATFORM=drm ? [15:48] tsdgeos: no, let it run normally [15:49] alf_: http://pastebin.ubuntu.com/12450102/ [15:49] one with xcb symbols coming [15:50] tsdgeos: thanks [15:50] alf_: http://pastebin.ubuntu.com/12450115/ [15:51] forcing drm http://pastebin.ubuntu.com/12450115/ [15:52] forcing drm + valgrind http://pastebin.ubuntu.com/12450137/ [15:53] and now with more symbols [15:53] http://pastebin.ubuntu.com/12450161/ [15:54] any idea if your intel or your nvidia chip is the one being used? [15:55] in theory the intel one [15:58] tsdgeos: is this latest wily? [15:58] alf_: no vivid+overlay [15:59] fwiw the egltriangle client has the same problem [16:00] FWIW the simplest client that shows anything is multiwin [16:00] bregma: Does dual landing copy the packages from wily to vivid, or does it rebuild them from source for wily? [16:01] bregma: sorry, I meant rebuild them from source for vivid [16:01] alf_, it tweaks the wily changelog and rebuilds against vivid+overlay [16:01] alf_: nothing would work without rebuilding [16:01] so, it's GCC-5 safe [16:02] alan_g: yeah, I just wanted to make sure ... [16:02] better to be sure [16:03] tsdgeos: is there a reason you want vivid+overlay? [16:04] alf_: becuase is what we release products on [16:04] i have a wily chroot [16:04] but i'm getting [16:04] std::exception::what: Failed to find platform for current system [16:04] when running the demo_server [16:05] tsdgeos: @products, well, not for the desktop (i think) :) [16:05] i guess it missses some stuff [16:06] tsdgeos: tried updating mir-graphics-drivers-desktop? [16:07] yeah [16:08] has anyone ever run mir in a chroot? [16:08] camako: I can't seem to get qmlscene to connect to the mir socket from mir_demo_server, any idea what I'm doing wrong? [16:08] I think I saw that at the PD sprint [16:08] mhall@mhall-thinkpad:~/projects/uReadIt/work$ MIR_SOCKET=/tmp/mir_socket QT_QPA_PLATFORM=ubuntumirclient qmlscene Main.qml -- [16:08] Loading module: 'libubuntu_application_api_desktop_mirclient.so.3.0.0' [16:08] UbuntuClientIntegration: connection to Mir server failed. Check that a Mir server is [16:08] running, and the correct socket is being used and is accessible. The shell may have [16:08] rejected the incoming connection, so check its log file === dandrader is now known as dandrader|afk [16:09] do I need a compositor like Unity running in between mir_demo_server and qmlscene? [16:10] mhall119, default location for mir socket is the XDG_RUNTIME_DIR [16:10] mhall119: no. but the client needs access to the socket [16:11] ok, got it to run from the chroot [16:11] let¡s try the ohter thing [16:11] if you run the server as root the socket is /tmp/mir_socket (but you need --arw-file on the server to make it globally accessible) [16:12] camako: I specified --file /tmp/mir_socket when launching mir_demo_server [16:12] alf_: wooooohooo, it works [16:12] alan_g: I ran the server as my user [16:12] so mir is broken in vivid+overlay for my configuration [16:12] srwxrwxr-x 1 mhall mhall 0 Sep 18 12:06 /tmp/mir_socket [16:12] works in wily [16:13] tsdgeos: given that it is the same version, something mesa has changed [16:13] tsdgeos: in mesa .. [16:13] :/ [16:13] mhall119, how about mir demo apps, do they work ok? [16:14] camako: ah ha, they didn't at first, but after installing mir-client-platform-mesa3 they do [16:14] and now my qmlscene app doesn't segfault either, but still doesn't display [16:15] mhall119, is this under x? [16:16] mhall119, does the qmlscene work with the mesa-kms platform, and fail with mesa-x11? [16:16] camako: I'm running Unity 7, so yes under X, I don't know which mesa-* qmlscene is using [16:17] when I run the qmlscene app, it shows a white horizontal bar in the Mir on X window, like at hte top of the multiwin demo windows [16:18] mhall119: that's Mir's "title bar" [16:18] mhall119, it's the server using the platform not the app... Can you try your experiment by running the demo server in a vt? [16:19] mhall119, to rule out/confirm that it's mir-on-x [16:19] causing the problem [16:19] camako: how do I do that? [16:19] camako: could this be the intel/mesa extension that got fixed in Wily? [16:19] provide a '--vt 1' option to the demo server and run as sudo [16:20] alan_g, yeah that's what I'm thinking [16:20] camako: Unknown command line options: --vt 1 [16:21] ?? [16:21] mhall119: you need to specify --platform-graphic-lib /usr/lib/x86_64-linux-gnu/mir/server-platform/graphics-mesa-kms.so.4 [16:21] alf_: oh well, now i know i need wily, monday more [16:21] * tsdgeos waves [16:21] (Or equivalent) [16:27] alan_g: camako: it kind of works on vt 1 [16:27] lots of flickering, but I can see the app's content [16:28] mhall119, I think it is due to the gles mesa extension that got fixed recently [16:28] mhall119, does your system use intel GPU? [16:29] You want to update your system to latest wily [16:31] camako: yes, intel [16:31] camako: I'm on vivid + operlay PPA [16:32] mhall119, yeah I don't think vivid has that fix [16:32] camako: do you know what version of the package has the fix? [16:35] mhall119, I don't exactly know what package the fix was in... some intel gpu and/or display driver. [16:36] camako: ok [16:37] mhall119, lemme look up the extension and then we can google it [16:37] is this what's causing it not to display the window on X11? [16:37] yes [16:37] ok [16:54] mhall119, I am pretty sure the fix is in the libegl1-mesa package... My version is libegl1-mesa: [16:54] Installed: 10.6.3-1ubuntu1 [16:55] mhall119, but for vivid, it's 10.5.* where this fix doesn't exist === alan_g is now known as alan_g|EOW [17:18] camako: yup, I have 10.5.9, any chancce of 10.6.* going into the vivid overlay ppa? [17:19] mhall119, I don't know how these kinds of decisions are taken === dandrader|afk is now known as dandrader [18:02] :( Xmir is not working for me on wily [18:02] Xmir :2 --desktop_file_hint=geany.desktop [18:03] oh well :( [18:03] at least i get a black window :P [18:08] -damage [18:09] some improvements are on xmir git [18:30] Hi, kgunn are you online? I compiled mir on my phone, but it still doesn't work. I think that I need to compile the platform-api for my device? [18:30] bkchr: hey [18:30] bkchr: what exactly did you compile ? [18:31] like a different branch ? [18:32] or did you compile the src associate with the image you're using ? [18:32] ahh I see if taken the compile for pc tutorial :( [18:33] No, I switched the image to a newer one [18:33] To one from yesterday [18:33] I used this guide http://unity.ubuntu.com/mir/building_source_for_pc.html [18:33] bkchr: ah...inclding the bzr branch lp:mir bit ? [18:33] so that's actually our devel.... [18:34] instead, what you might want to do is either use lp:mir/0.15 (cause that's what's in the current image) [18:34] rc-proposed that is [18:34] Running mir_demo_server I got the following http://pastebin.com/amEReypq [18:34] (stable would be lp:mir/0.14) [18:35] otherwise you could grab the src from the appropriate ppa [18:35] rc-proposed uses [18:36] bkchr: runnig as root? [18:36] You think that switching the source would help me? :D Do [18:36] Yeah it's running as root [18:36] https://launchpad.net/~ci-train-ppa-service/+archive/ubuntu/stable-phone-overlay/+sourcepub/5379893/+listing-archive-extra [18:37] you'd need to download [18:37] mir_0.15.1+15.04.20150903-0ubuntu1.diff.gz (52.8 KiB) [18:37] mir_0.15.1+15.04.20150903-0ubuntu1.dsc (4.4 KiB) [18:37] mir_0.15.1+15.04.20150903.orig.tar.gz (1.6 MiB) [18:37] bkchr: which device? [18:37] and build it from that [18:37] or if you're on stable [18:37] you'd get it from stable-snapshot [18:38] kgunn: hm looks like hwc problem.. or broken implementation hwc set returning -1.. if that happens with lp:mir i would assume it might also happen with 0.15 [18:38] ah [18:39] at least that part hasent seen a lot of changes since 0.14 [18:39] anpok: oneplus one [18:39] anpok: but i think bkchr is sstill mixing and matching mir's there...and gonna leat to other probs with needing to rebuild qtmir/u-s-c etc [18:39] yes [18:40] bkchr: here it would be interesting what parameters are passed to hwc prepare and hwc set .. because it is odd that prepare went fine and set fails.. [18:40] i guess kdub and AlbertA would be the best candidates for remote debugging [18:44] and actually, kdub might know how to just do a trick to skip hwc [18:44] and rely on gles alone [18:45] nah, we always go through hwc these days [18:45] unless bkchr's phone is really really old [18:45] although, if you want to try a deprecated backup, move hwcomposer.*.so out of /system/lib/hw so mir can't find it, maybe the fb module will work [18:51] kgunn, so I should use either the 0.15 branch or the packages you proposed? [18:53] bkchr: are you on an image which is from rc-proposed ? or stable ? [18:54] bkchr: do system-image-cli -i [18:54] rc-proposed [18:54] ah nice [18:54] yeah lp:mir/0.15 will do it [18:54] same instructions for building otherwise [18:54] k [18:55] it's building [18:55] bkchr: sorry for confusion, i swear we're not trying to be difficult on purpose, that's all on accident ;-P [18:58] no problem^^ It already was a mess to get everything working so far. [19:18] i bet === dandrader is now known as dandrader|afk [19:57] camako: is it possible to use mirscreencast to get a screenshow of mir_demo_server? [19:58] mhall119, yes [19:59] I can't seem to get it to work [19:59] [1442605661.898312] MirBufferStreamAPI: Caught exception at client library boundary (in mir_buffer_stream_get_graphics_region): /build/mir-Xfl1I9/mir-0.15.1+15.04.20150903/src/platforms/mesa/client/client_buffer.cpp(63): Throw in function {anonymous}::ShmMemoryRegion::ShmMemoryRegion(const std::shared_ptr&, int, const mir::geometry::Size&, mir::geometry::Stride, MirPixelFormat) [19:59] Dynamic exception type: N5boost16exception_detail10clone_implINS0_19error_info_injectorISt13runtime_errorEEEE [19:59] std::exception::what: Failed to mmap buffer [19:59] 13, "Permission denied" [20:00] or sometimes just "Failed to create screencast" with no other output [20:00] mhall119, you need to make sure you are using software buffer [20:00] when you create the surface [20:00] how do I do that? [20:00] o thats screencast? [20:00] mhall119, are you writing a mir client atm? or just using screencast? [20:00] bschaefer: it's mirscreencast with -n1 [20:00] IIRC screencast was broken... [20:01] but *should* have been fixed [20:01] just trying to take a picture of what it's doing [20:01] but i dont remember what version fixed it :( [20:01] yeah [20:01] bschaefer: I'mstill on vivid, maybe it's fixed in wily [20:01] yeah i think duflu fixed in in 0.15 (i think?) [20:03] yet another reason to upgrade I guess === dandrader_ is now known as dandrader