=== DalekSec_ is now known as DalekSec === duflu_ is now known as duflu === chihchun_afk is now known as chihchun [04:38] Ah, of course. *That's* why you don't do a compile in an actual armhf chroot. [06:41] * RAOF apparently has an xorgRootess [06:43] RAOF: +l ? [06:44] No; typo :) [06:44] +Mir ? [06:44] But it seemed like a funny typo. [06:44] This *should* work, and be rootless... [07:24] hello, I'm installing unity-system-compositor & xmir on amd64 trusty, everything seems running fine under /var/log/lightdm, but the screen is blank.. any idea? [07:26] # tarball of log https://mega.co.nz/#!hMxD1ArC!e7AevxbOyidBBaZZAlXP7SoZdkMlT8b7Qzi3dDel2OI [07:30] penk: I can't see any reason in those logs. Could you please "ubuntu-bug unity-system-compositor" ? [07:30] duflu: sure I'll get to that [07:31] penk: Everything works without unity-system-compositor? [07:31] duflu: yes, with radeon driver [07:31] Hmm, I should revive my radeon machine and update it [07:31] duflu: I can hear the lightdm sound, but the screen is black [07:32] penk: Is the monitor LED showing it's asleep (like orange instead of green)? [07:32] duflu: I'm pretty sure the screen & system is awake [07:33] penk: So you can see the backlight is on, but pixels are black? [07:33] duflu: exactly [07:34] penk: When you can, please try "Ctrl+Alt+F1". Can you reach the VT that way? Also please try installing mir-demos and running "mir_demo_client_egltriangle". Does it appear on top of the blackness? [07:35] duflu: vt switching is working [07:35] duflu: sometimes the compositor gives me VT I/O error though [07:35] duflu: I'll try mir-demos later [07:35] That's probably not cool [07:52] duflu: I see mir_demo_client_egltriangle running on the screen [07:53] penk: Cool. Then the bug is just XMir. Please log it against that project [07:53] duflu: got it, thx [09:08] Now showing in a browser near you: http://timvideos.us/wool [09:24] duflu: what is shown next? [09:24] anpok: Too late I think ... http://linux.conf.au/programme/schedule [09:25] oh zero copy composition [09:26] kdub: I could not reproduce the error, but hey got a different one instead - mir_integration_tests fails to create a fb [09:27] That's OK. Keith being Keith only talked about X. And when asked about something other than X (Wayland) he said he wasn't interested. And had nothing to say [09:27] hehe [09:50] duflu: @bug 1239955 on CI - have we asked #ubuntu-ci-eng if they can poke the hardware? [09:50] bug 1239955 in Mir "integration-tests hang/fail in AndroidGPUDisplay.gpu_display_ok_with_gles when the display is asleep: what(): error posting with fb device" [Medium,Triaged] https://launchpad.net/bugs/1239955 [09:52] alan_g: Assuming it's the same cause. There might be other triggers... [09:54] on n10 that test always fails for me [09:54] but with a different exception [09:57] here gralloc is queried for a framebuffer_device_t, a pixel format is taken from that structure [09:57] the pixel format is RGBA_8888 [09:57] but the no EGLConfig that is querried matches that visual id [09:58] the most similar one is BGRA_8888 [09:59] the other thing I noticed was that whenver i run the tests in order the display is turned off around that test.. [09:59] seems like the tests before that one cause that in some way? [10:04] forget the last two statements - just some buffers are flipped where two of the three are black. display is not turned off.. [10:21] duflu: it isn't just the one device. The difference seems to be they moved to trusty-proposed 119 today. Gonna try that here. [10:21] Fair enough. We don't have much device-specific code === Prf_Jako1 is now known as Prf_Jakob === davmor2_ is now known as davmor2 === chihchun is now known as chihchun_afk === pete-woods is now known as pete-woods|fight === pete-woods|fight is now known as pete-woods|away [12:42] alf__: I'm not sure what to try next with bug 1239955 (which is blocking CI) - any suggestions before I MP disabling the test? [12:42] bug 1239955 in Mir "integration-tests hang/fail in AndroidGPUDisplay.gpu_display_ok_with_gles when the display is asleep: what(): error posting with fb device" [Critical,Triaged] https://launchpad.net/bugs/1239955 [12:44] alan_g: @"Initially I saw intermittent failures if unity8 wasn't stopped", is unity8 not being stopped while running our tests in CI? === chihchun_afk is now known as chihchun [12:44] alf__: it is stopped - I've checked [12:47] alan_g: does "sudo powerd-cli display on bright &" make any difference locally? === chihchun is now known as chihchun_afk [12:48] alf__: yes. Seems to work as well as the power button === Trevinho_ is now known as Trevinho [12:51] alf__: revise that - still failing intermittently [12:53] alan_g: no more ideas off the top of my head [12:57] alf__: OK I'll MP disabling the test as a workaround. === pete-woods|away is now known as pete-woods === alan_g is now known as alan_g|lunch === dandrader is now known as dandrader|afk [13:41] alf__: do you have a nexus10? [13:41] anpok: yes [13:41] anpok: ? [13:42] do the integration tests run for you on the device? [13:42] anpok: haven't tried recently... do you want me to? [13:43] it would be nices [13:43] -s [13:43] because I dont think that my issues are hardware related [13:43] anpok: ok [13:44] I believe that the egl config selection needs a different logic [13:50] anpok: ok, so should I just run the integration tests from development-branch or do you need some special branch/config? [13:50] just that [13:50] or even just limit to the AndroidGPUDisplay.* tests === dandrader|afk is now known as dandrader [14:01] anpok: I get "could not select EGL config for use with framebuffer" [14:01] anpok: for gpu_display_ok_with_gles [14:01] me too [14:03] it happens because we try to select an egl config with the same "visual id" as the one mentioned in the framebuffer_device_t [14:09] anpok: I am not familiar with what visual id means for Android, but on the desktop it's not directly related to a pixel format [14:10] hm it seems like on android it matches the pixel format integer found in system/graphics.h [14:12] HAL_PIXEL_FORMAT_RGBA_8888 = 1, [14:12] HAL_PIXEL_FORMAT_RGBX_8888 = 2, [14:13] anpok: it seems because our code checks for it or because you found some reference of that fact elsewhere? :) === alan_g|lunch is now known as alan_g [14:16] our code checks for it :) [14:16] .. and the values seem to make sense [14:16] so docs would be awesome :) [14:17] anpok: if a print out the id values I get: http://paste.ubuntu.com/6720922/ (which supports this) [14:17] yes [14:19] anpok: so we are looking for RGBA8888 but nexus 10 provides BGRA8888 [14:20] or RGBX for the frame buffer [14:21] I wonder if is necessary to have an exact match here [14:23] I guess that code path is only used in the unit tests.. [14:23] s/unit/integration [14:30] hm the android gl_context could also make use of the surfaceless context class [14:37] anpok: the framebuffer_device_t reports that it uses RGBA8888 but Nexus 10 doesn't seem able to render to it. As you said this is the fallback (non-HWC) path, which Nexus 10 probably won't ever need to use. === dandrader is now known as dandrader|lunch [15:03] ok if a different one is picked posting silently just fails === alan_g is now known as alan_g|tea [15:49] racarr: anyway...the more i thot about the autopilot input prob thomi & Saviq were talking about.... [15:49] i got confused.. [15:49] is it really the at the events go missing ? [15:50] it soundslike the device [15:50] or is it just that the restart causes mir to get confused b/c the "rules" aren't realy changed...even thos inotify says ok? [15:50] doesntget probed properly [15:50] ...or are there really 2 problems both there [15:50] racarr: ok...so more that latter [15:50] mm I think so [15:51] someone (i think thomi) was scaring me that we had a 2nd prob of events going missing (even after proper probing) [15:58] racarr: is something "eating" your spaces and apostrophes or it it just quick typing? [15:58] is it [16:00] i'm quessing an old tea spill on the space bar === greyback is now known as greyback|away [16:03] alf__: Space key eaten mostly [16:04] via tea yes lol === alan_g|tea is now known as alan_g [16:22] robotfuel, where can i find the script that sets up the phone on CI? [16:23] kdub: pulling it up now... [16:25] thanks [16:25] kdub: lp:~chris.gagnon/+junk/mir-medium-test-runner-for-jenkins [16:26] kdub: http://s-jenkins.ubuntu-ci:8080/view/Mir/job/mir-mediumtests-runner-mako/configure if you need more info [16:37] kdub: fyi...i was gonna try the fab-4 integration tests...since its now in archive, will likely ressurect the kill surfflinger topic as soon as i have confidence [16:38] not sure what that means :) [16:39] like, remove sf from the image? [16:41] kdub: well...at least stop having it available for unity8....only as a tool for bringup [16:41] meaning, we won't be putting effort in maintaining unity8 code [16:41] which enables surf flinger [16:41] to be reliably used [16:43] kdub, mainly drop the ability to run it [16:43] as i understood (as a plumber) [16:46] right, sounds okay to me === greyback|away is now known as greyback [17:56] is it possible to enable hardware cursor manually when using XMir? why is it still not supported, any reasons? [18:01] alf__: ^ if you're on [18:02] I really would love to use XMir, but a slow/laggy mouse cursor is a no-go === alan_g is now known as alan_g|EOD [18:03] brainwash_: right, thanks for trying xmir and using...just a matter of effort really, we've shifted our focus towards [18:04] building out features around tablet/unity8 shell [18:04] since xmir didn't make it in by default...the way [18:04] scheduling lines up wrt releases etc [18:04] so XMir gets abandoned slowly? [18:04] it makes more sense for us to pour it on for unity8-mir & "rootless" x on unity8-mir [18:05] rootless like xwayland, right? [18:05] brainwash_: not abandoned per se, but feature stagnant [18:05] brainwash_: rootless will leverage a lot of what was done in mir [18:06] oops/mir/xmir [18:06] alright, but I'm still interested if hardware cursor support is possible :) [18:06] brainwash_: was trying to hail someone smarter than me :) [18:07] brainwash_: depending on your timezone...if you pop in say in about 8 hours or so...you'd get a better answer [18:07] when the australs come onlien [18:07] online even [18:08] alight, I'll setup my server then and rejoin with my irc bouncer, so I'll don't miss the answer, thanks :) [18:08] brainwash_: sorry i wasn't better help... [18:08] again thanks for trying out xmir === dandrader|lunch is now known as dandrader [19:21] ricmm_, poke about libhybris === dandrader is now known as dandrader|afk === dandrader|afk is now known as dandrader [20:51] mterry_: wip [21:48] * kgunn hugs ricmm_ [21:50] rsalveti, are you okay with https://code.launchpad.net/~kdub/mir/less-3rd-party/+merge/199851 [21:50] basically making mir a dependency of android-headers [21:50] hm, lemmecheck [21:53] kdub: looks good [21:53] cool, thanks! [21:54] *making mir depend-on android-headers, rather [22:12] brainwash_: Yeah, HW cursor support is possible & needed. [22:12] brainwash_: It'll get done before we're done, but with phone/tablet being the current focus for Mir development it's understandably lower importance :) [22:13] brainwash_: On the other hand, it could get done sooner if you wanted to do it :) [22:14] RAOF: so it's not a trivial thing :/ [22:14] Oh, it's reasonably simple. [22:14] on top of that, I lack the knowledge to implement this [22:39] For those not watching #wayland, http://www.anandtech.com/show/7641/amd-demonstrates-freesync-free-gsync-alternative-at-ces-2014, is super-cool. [22:50] yeah.. RAOF that is pretty cool [22:50] and giving it away...hmmm [22:51] Apparently available on all recent AMD GPUs, too.