=== Amaranth_ is now known as Amaranth === Bernardo is now known as jbs === Bernardo|away is now known as Bernardo [08:19] finally made it to the UDS \o/ [10:20] Bernardo http://ubuntuforums.org/showpost.php?p=9291537&postcount=798 [10:30] lucazade: composite and render should be enabled by default, why did you put them in your xorg.conf? [10:30] the same applies to aiglx [10:30] to make test easier [10:30] yes i know [10:31] aah, to disable them when you need it [10:31] i cannot remember every option, so.. [10:32] i leave them [10:34] hi tseliot [10:35] hi bryceh [10:38] lucazade: did you put the 3d driver package in some ppa? [10:39] not yet... maybe slonua was uploading to the gma500 ppa on launchpad [10:39] not sure [10:40] ok [10:41] I think Sarvatt has the rest of the required packages in his ppa [10:43] do you a have link to his ppa? [10:43] or git repo? [10:44] oh : http://www.internetling.com/2010/05/12/howto-intel-gma-500-poulsbo-on-ubuntu-10-04-lucid-lynx/ [10:45] his ppa: https://edge.launchpad.net/~sarvatt/+archive/psb [10:47] RAOF, ^^^ [10:48] bryceh: Thanks [10:50] RAOF, tseliot, you guys might want to contact that Lucazade guy and see what changes he did to the packages to get them to work [10:50] looks to me like it might even be sru-worthy since it seems to have no debs for non-psb bits [10:50] or mm at least [10:50] bryceh: getting things to work is one thing, supporting them is another [10:51] we could put the packages in the x-updates PPA when they are ready [10:52] * tseliot knows what it means to try to support poulsbo... [10:55] tseliot, true [10:55] oh hey I'm blind, lucazade is here on the channel :-) sorry lucazade! [10:57] my touchscreen uses poulsbo (and I have a few other devices with the same chip) therefore I'll be able to test it and fix the packages if something is wrong [11:03] bryceh: svn repository for gma500... https://gma500.googlecode.com/svn/trunk/ [11:04] bryceh or lp:gma500 [11:05] RAOF, wayland progress! made it past the Makefile.in error [11:06] tseliot, great [11:08] lucazade, thanks [11:08] bryceh: yay! [11:09] lucazade, can you tell me what your interest is with gma500? any interest helping with the maintenance of it in ubuntu more generally? [11:14] bryceh.. my interest in gma500 is only for personal use, just a challange with my knowledge. I'd like to help and maintain but i'm not really a driver developer so I honestly don't know if I am able to. [11:14] really thanks for your offer [11:18] * ilmari should get around to forwarding bug #568138 upstream [11:18] Launchpad bug 568138 in xserver-xorg-video-intel "[arrandale] deadlock in i915_gem_madvise_ioctl" [Undecided,Confirmed] https://launchpad.net/bugs/568138 [11:23] lucazade, ok no prob, well stick around, you sound like you'd be quite helpful if other users have some questions :-) [12:30] hi [12:30] lucazade: saw that you got it working with gnome, great [12:31] I see bryceh has left... :( [12:32] lucazade: the problems you had with compiz probably can't be solved with this version of the binary blobs [12:33] lucazade: the server part only supports glx 1.2, and if compiz is calling a 1.3 function, then we can't get to work without much effort [12:34] yes, you're right... i can live without compiz on a netbook :) [12:36] metacity compositor is enough [12:44] Now.. back to try getting some useful information out of x crashes [12:44] :) [12:45] I've followed https://wiki.ubuntu.com/X/Backtracing but running X under gdb seems impossible on a 1101HA, always goes to 100%cpu [12:46] different subject - how do I tell launchpad which folders in bazaar it should build as packages? [12:47] i've never build ppa from bzr.. always use dput [12:47] Bernardo: what's the problem? Does X crash? [12:47] yes [12:47] with shadowfb false [12:48] do you have an X log with the backtrace? [12:48] http://code.google.com/p/gma500/issues/detail?id=21 [12:49] and [12:49] http://code.google.com/p/gma500/issues/detail?id=15 [12:49] 15 [12:49] you're fast... :) [12:49] eheh [12:50] tseliot: my machine has the xorg-core and xorg-core-dbg packages from bryceh's purple ppa, but that seems to be all info I can get from a backtrace [12:51] you won't get more than that without gdb [12:52] I thought as much, but gdb here (running on a ssh session) hangs with xorg taking 100% cpu when I try to reproduce the crash [12:52] I've left it running overnight, but after 8 hours it was still hung and wouldn't answer to ctrl-c [12:52] you could get X to dump core, and run gdb on that [12:53] (with the -core option; should get a core file in /etc/X11/ after the crash) [12:53] let me try now... [12:58] jcristau: thanks, now I have 75M core to play with... :) [13:00] Bernardo: I would like to see the full backtrace from that core [13:01] sure [13:01] it seems to end up inside pixman_blt_sse2 [13:02] did you install libpixman-1-0-dbg? [13:09] Should be: [13:09] libpixman-1-0-dbg: [13:09] Instalado: 0.16.4-1ubuntu2 [13:09] Candidato: 0.16.4-1ubuntu2 [13:09] tseliot: http://pastebin.com/gzvrhRcU [13:10] thanks [13:10] tseliot: anything else you want, please ask. I'd love to get this working again (before it breaks with kernel 2.6.34) [13:11] ok [13:11] src_bits=0x0, dst_bits=0x0 sounds wrong somehow. [13:13] yes, it makes very little sense [13:14] that seems to happen after #17, miDoCopy, the others below the values still seem to make sense [13:16] I forgot I had removed 'Option "NoTrapSignals" "true"' from xorg.conf, do you want another backtrace with that enabled? [13:17] shouldn't make a difference [13:19] ok, anything you want me to patch jusk ask. [13:20] sure [13:22] is it th same issue? http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=580119 [13:22] Debian bug 580119 in xserver-xorg-core "Hang in pixman_blt_sse2()" [Normal,Open] [13:27] i don't think so. not sure though. [13:28] that one is a hang, not a segv, and it's using intel, not psb (which means not exa). [13:36] leaving now, I'll be back in a couple of hours === Bernardo is now known as Bernardo|away [13:58] lucazade: can you reproduce the problem if you set Option "MigrationHeuristic" to "always" instead of "greedy"? [13:58] in your xorg.conf, that is [14:00] Bernardo|away: ^^ === Bernardo|away is now known as Bernardo [15:37] tseliot: ok, will try now [15:37] thanks [15:39] with always it will crash on kdm login screen [15:39] let me check the core [15:39] Bernardo: also, please try "smart" instead of "always" [15:40] http://pastebin.com/MNgfDDLg [15:40] that is for always [15:40] I'll try smart now [15:40] bryceh: so a smoke test on the RS600 rotation bug passed; KMS doesn't work on it. [15:42] http://pastebin.com/qN63Zgab [15:42] jg, the smoke test passed with KMS off, but failed with it on? [15:42] * tseliot has a look at the log [15:43] with smart I don't even get a graphic, only a flashing cursor on the top left [15:43] ah, ok [15:44] but X is started, according to the backtrace I posted [15:50] bryceh: doesn't work at all with kms on. rotate works with kms off. [15:51] bryceh: as X wont run at all on KMS on the chip, rotation is moot. [15:51] (nor does the console work with kms enabled). [16:07] bryceh: also, I have a brand-spanking, new HP Elitebook 2540p notebook; neither Ubuntu nor Fedora will install on it. ajax took a look; it appears to be the first eDP laptop in captivity he's heard of. What is happening from the conversation ajax and keithp had is that the X server is trying to set the frequency on the eDP internal panel, that it should not mess with, and loses. Ironically, if I plug an external panel into the external DP connector, it drives ju [16:07] st fine (but it's hard to install when you can't see what you are doing, short of a text based install). [16:07] bryceh: I haven't had a chance to catch up with ajax to see if he has a patch yet. [16:26] tseliot: anything else? [16:28] jg: http://lists.freedesktop.org/archives/intel-gfx/2010-May/006847.html -- doesn't look promising though https://bugs.freedesktop.org/show_bug.cgi?id=28070 [16:28] Freedesktop bug 28070 in Driver/intel "[Arrandale] No output (black) on eDP" [Major,New] [16:28] Bernardo: no, thanks. I think we'll have to settle with "greedy" then. I'm wondering what can be causing that null pointer... [17:19] tseliot: the bug seems similar to https://bugs.launchpad.net/ubuntu/+source/xorg-server/+bug/539772 but I am already using the packages from bryceh that should fix that [17:19] Launchpad bug 539772 in xorg-server "[MASTER] Lucid 2.6.32-16 crashed to login screen - miCopyRegion" [High,Fix committed] [17:27] Bernardo: while I don't think it's the same problem I think I can try a similar solution by checking that the pointer in the source drawable is not null [17:28] I need to look more into this issue [17:29] Ok, if you want me to test that, I'm here. [17:35] Bernardo: ah, did you say that you're already using the xserver from ppa:bryceharrington/purple ? [17:35] yes [17:35] When I first got a backtrace, I googled for something similar, and found that [17:36] aah, ok [17:36] tried to see if it fixed the problem, but it didn't [17:36] as I thought, it's a different problem [17:36] but at least it gave me debug symbols for xorg-core [17:36] but it's still useful as a reference [17:36] I thought as much [18:08] Bernardo: I've written a patch so as to check that point earlier. I'm trying to build X here before I upload to a PPA [18:08] s/point/pointer/ [18:22] Bernardo: please try 2:1.7.6-2ubuntu8~lp539772tseliot1 when it's built: https://launchpad.net/~albertomilone/+archive/ppa [18:23] tseliot: sure [18:23] thanks! [18:23] thanks for testing [18:28] ok, seems like all three (common, core, and core-dbg) are there, so I'll install them now [18:29] great [18:29] starting x without shadowfb... [18:29] Bernardo: wait it says that it's still building here [18:30] make sure that the version ends with tseliot1 [18:30] Ah... I got the ones without. Only looked at 539772. Doh... [18:30] https://launchpad.net/~albertomilone/+archive/ppa/+build/1738412 [18:31] Bernardo: the packages seem to be available here ^^ [18:31] Yes, they are just awaiting publication, I'll try downloading from there [18:31] ok [18:34] installing now [18:35] good [18:36] still crashes... :( [18:37] http://pastebin.com/jaTCqM1g [18:38] * tseliot has a look at the log === Bernardo is now known as Bernardo|away [18:57] Bernardo: anything to test? [19:01] * tseliot -> dinner === Bernardo|away is now known as Bernardo [19:20] * Bernardo just finished dinner... :) [19:20] lucazade: tseliot made a patch to try to fix the crash, but unfortunately it wasn't successful [19:44] Bernardo ok === Bernardo is now known as Bernardo|away [20:17] Bernardo|away, lucazade: I've just uploaded another patch (make sure you install 2:1.7.6-2ubuntu8~lp539772tseliot2) https://launchpad.net/~albertomilone/+archive/ppa [20:18] xorg-server ? [20:18] yep, it hasn't been built yet though [20:19] add the ppa and update? [20:20] yes, when the package are ready [20:20] packages [20:20] ok i'l wait [20:20] thanks [20:28] tseliot failed to build [20:32] let me check [20:34] I found the problem... let me fix it [20:34] ok [20:41] lucazade: ok, re-uploaded (~lp539772tseliot3) [20:41] tseliot perfect [20:42] i'm in queue as the package [20:44] ok === Bernardo|away is now known as Bernardo [20:56] tseliot, lucazade: I see we have another patch to test? [20:56] Bernardo: I'd be glad if you could test ~lp539772tseliot3 too [20:56] it's still building [20:56] ok, I'll wait... gnawing my finger nails... [20:57] :-) [20:59] jcristau: http://www.openbsd.org/cgi-bin/cvsweb/xenocara/driver/xf86-video-intel/ [20:59] (if you didn't find it yet) [21:00] Bernardo, lucazade: https://launchpad.net/~albertomilone/+archive/ppa/+build/1738531 [21:01] tseliot refreshed ppa but still display tseliot1.. maybe some mins [21:01] ok, going to download it [21:02] yes, they haven't been published yet [21:02] ok [21:03] installing... [21:04] Hey guys, thanks for working on the poulsbo stuff for Lucid [21:04] testing now... [21:05] tseliot: I think you got it! [21:05] yes! [21:05] didn't crash as usual, and 3d works [21:05] grande Alberto [21:06] Wow ! Does that mean that I'll get 3D on my mini 10 with that guys ? [21:06] glblur gives 16.6 fps [21:06] Bernardo, lucazade: really? [21:06] 19.9 with overclocking [21:07] still one problem wth bitmaps [21:07] still installing here [21:07] the eee-control tray icon is empty, the kde native ones show [21:07] LaMs: thank tselliot for it [21:08] I just ported stuff from mandriva, tseliot seems to have found the main problem with 3d [21:08] tselliot: I understand that you are the one who made 3d possible, so a big thank you ! [21:08] Sarvatt: oga gave me a link to his openbsd git repo, thanks [21:08] ricotz: huh did linux - 2.6.34-2.8 fail to build? I copied that over a few hours ago but your linux - 2.6.34-2.7~lucid~ricotz1 is in there? [21:09] LaMs: it seems that (much to my surprise) I have worked around the problem, yes [21:09] tseliot: can I make the announcement on the ubuntuforums thread to tell people to add the ppa, or do you want do hold or do it yourself? [21:09] I can test for you guys and I may help a little bit if needed. I'm an embedded developer here at work and have some experience with kernel driver [21:09] Bernardo, lucazade: I can make my patch smaller so that RAOF can adopt it [21:10] Just don't have any experience with Xorg at all [21:10] Bernardo: sure, go ahead [21:10] LaMs: add the gma500 ppa (or run lucazade's script), then add alberto milone's ppa ( https://launchpad.net/~albertomilone/+archive/ppa) and upgrade [21:10] Sarvatt, yeah 2.8 has a wrong ABI folder and failed, there is no real difference to 2.7 [21:10] lams kernel 2.6.34 break psb-kernel-source [21:10] LaMs [21:11] theres a fix for 2.6.34 on the intel-gfx lists [21:11] lucazade: Is 2.6.34 a kernel that Lucid will support ? [21:11] tseliot: thanks a lot. There is still the bitmaps issue, and now we can look at problems with 2.6.34 - but maybe those are solved now [21:11] lucazade: how is it going? [21:11] LaMs: nope [21:11] tseliot icons are not visible [21:11] Bernardo: bitmaps issue? Do you have a bug report I can look at? [21:12] but it booted ok [21:12] there will probably be a IEGD release here soon that supports 10.04 [21:12] the one I just was talking about - eee-control icon isn't visible, all the kde standard ones (which I think are svgs) can be seen [21:12] once meego comes out [21:12] Sarvatt: I thought meego will have xorg 1.8? [21:12] yup [21:13] So either we use the xorg-edgers ppa, or you guys are planning backporting 1.8 to lucid? [21:13] a IEGD release would save us a lot of work [21:14] sigh iegd. that shit still exists? [21:14] if it finally fixes suspend/resume [21:14] no but it'd be trivial for them to support xserver 1.7 if they do 1.8 [21:14] Sarvatt: if they feel like it [21:14] tseliot i think bitmap issue is the last one [21:14] jcristau: do nightmares exist? [21:15] tseliot: I thought the nightmare is the gma500/poulsbo itself [21:15] the driver is the problem [21:15] shows you a glimpse of cheap, fast, low power graphics, but with crappy binary blobs tying it down. [21:15] poulsbo is a dream even compared to arm GPU's :D [21:16] * Bernardo dabbles wit a zauros for quite some time [21:16] * Bernardo has a zaurus C1000 for the taking, if someone wants to play with a old arm system... [21:17] well I *think* poulsbo and some arm gpus share some intellectual property [21:18] Sarvatt: I won't trade my 1101HA for any laptop or netbook costing even 5 times as much. It is light, the right size, and when the drivers work I can watch full hd with it. And even when they don't work that well I can do my work with it. [21:18] As PowerVR is concerned yes [21:18] From my readings ... [21:18] yep [21:19] tseliot: I can see other problems, probably bitmap related, progress bars in kde now have a "grey rectangle within darker grey rectangle" background [21:20] but the desktop bitmap shows without problems [21:21] Bernardo: yes, I just worked around the problem by preventing the driver from trying to copy null drawables [21:21] here on gnome i can't see any icons, wallpapers is visible, glxgears seems to run ok [21:21] ok [21:21] lucazade: Any suspicious message in system logs ? [21:22] let me see if i see any [21:23] nothing in dmesg [21:24] yes, it's an X thing [21:24] not kernel related [21:24] in Xorg.0.log there is the usual "(EE) PSB(0): First SDVO output reported failure to sync or input is not trainded!!!" [21:24] nothing much more [21:25] the "loading" icon in kde is corrupted [21:25] Xsession-error nothing, trying to start gnome-session --debug [21:25] Do you see the corruption after trying to reproduce the previous problem or as soon as you start X? [21:26] I only saw corruption now, the usual result is a blank space where a application icon should be [21:27] ok but as soon as you started X? [21:27] as soon as I got the kde desktop [21:27] ok [21:27] another symptom, scrolling in firefox scrolls only a bit of the page, the rest remains [21:27] I maybe I'm killing sane drawables too [21:28] let me reduce the patch and rebuild the packages [21:28] ok [21:29] Ok I'm leaving [21:29] I'll get back from home with my mini 1010 in hand [21:29] I've not upgrade it yet [21:30] chow and great work guys [21:31] scrolling in konsole also has problems, only scrolls the last two lines [21:34] Bernardo, lucazade: ok, I've just uploaded ~lp539772tseliot4 to my ppa [21:35] tseliot ok [21:36] ok [21:37] I see the build will start in 10 minutes [21:40] * tseliot -> brb [21:43] at least vaapi works - fullHD! [21:43] :D [22:09] ok, it just build, so final test... [22:14] Still doesn't crash, but scrolling is still broken in firefox and console, and the eee-control icon is still invisible [22:14] have to go now, see you guys tomorrow. [22:14] good night! [22:14] see you Bernardo [22:14] night === Bernardo is now known as Bernardo|away [22:14] bye lucazade [22:14] same issue here [22:15] tseliot ping [22:16] i have to go.. [22:16] bye [22:17] lucazade - away [22:31] Bernardo|away, lucazade: ok, thanks for testing. I'll investigate the issue tomorrow