[00:53] ugh our older kernel-package doesn't work with the newer kernels anymore, this package is such a nightmare to follow and has changed so much since the 11.015 we're using [00:56] utsrelease.h moved to include/generated it looks like, think it might still work if you dont specific any -append-to-version.. hope so because I need to test a kernel with an extra debugging info patch for this intel execbuff while wedged bug [03:06] Hello - I am using the "radeon" module with Ubuntu 9.10 - and any time an application uses SDL to go full screen, the monitor turns off [03:07] what is a way to start troubleshooting this issue? [03:10] thanks for your continued work with incorporating the open drivers from AMD into Ubuntu 10.04 [08:27] are the libgl1-mesa-* 7.7-0ubuntu5 updates stable? i dont see any accompanying xserver updates [08:27] on Lucid^ [08:28] ATI [09:37] tseliot: hi, what's your reference for saying that alternatives don't need to have the same set of slaves? [09:37] if you're basing this on mandriva, then all bets are off, they don't use the dpkg implementation of update-alternatives... [09:38] slangasek: I'll find it. BTW I remember than they patched their update-alternatives [09:38] I'll ask them about those patches [09:39] they told me that they patched it so that it could deal better with links to files that don't exist [09:39] * slangasek nods [09:40] slangasek: if the patches are reasonable we might include them [09:42] yes, but that doesn't help users who are still running the dpkg from the old Ubuntu release at time of upgrade; I would prefer a solution on the libglx side that's robust wrt known package manager bugs... [09:43] anyway, this is still all speculative on my part, I haven't shown that this is related to the missing files problem yet :) [09:43] right [09:44] if I see what their patches do, maybe I can reproduce their behaviour in our preinst scripts [09:44] it would still be nice to find the cause of our problem though [10:57] slangasek: just another note: I think we can safely exclude nvidia from this as the problem took place on systems that use only intel and that didn't have any kind of nvidia driver installed [10:58] the kernel devs, pitti and other people used only -intel [10:58] * slangasek nods [11:31] tseliot, are we expecting an upload to fix this lost files thing? i only see u5 still [11:33] apw: u5 was supposed to fix this (which in part explains why some installations have the problem while other don't). We can't fix this weird upgrade issue for good if we don't know why our fix doesn't work reliably [11:34] apw: quoting the message that I sent to ubuntu-devel@: [11:34] This shouldn't happen in dist-upgrades from Karmic to Lucid or in new installations but only in some cases after updating a Lucid system. [11:35] tseliot, so is there anything more you need from us, or should i tell those affected '--reinstall' [11:36] slangasek: ^^ [11:36] * tseliot ran out of ideas [11:37] apw: if anyone has a system they haven't yet --reinstalled to fix, /var/lib/dpkg/alternatives/gl_conf would be nice [11:37] slangasek, will see who's left [11:39] tseliot: oh... just found it [11:39] what is it? [11:39] $ sudo update-alternatives --install /etc/ld.so.conf.d/GL.conf gl_conf /usr/lib/mesa/ld.so.conf 500 --slave /usr/lib/GL gl_libraries /usr/lib/mesa --slave /usr/lib/xorg/extra-modules xorg_extra_modules /usr/lib/xorg/x11-extra-modules [11:39] update-alternatives: warning: alternative /usr/lib/standard-x11/ld.so.conf (part of link group gl_conf) doesn't exist. Removing from list of alternatives. [11:39] /usr/lib/standard-x11/ld.so.conf vs. /usr/lib/standard-x11/standard.conf [11:40] are you trying to tell me that I'm removing the wrong alternative? [11:40] let me check [11:42] d'oh [11:43] slangasek: that's the cause [11:43] I should have removed /usr/lib/standard-x11/ld.so.conf [11:43] let me fix it [11:43] * tseliot blushes [11:43] we both missed it :) [11:43] go ahead and upload when you have it fixed [11:44] ok [11:47] slangasek: shall I bump the revision we're checking? dpkg --compare-versions "$2" lt-nl 2:1.7.3.902-1ubuntu5 [11:47] to dpkg --compare-versions "$2" lt-nl 2:1.7.3.902-1ubuntu6 [11:48] as some systems might still have that alternative installed [11:48] yes, should be bumped [11:49] if they don't have one then the following command shouldn't cause the preinst to fail: [11:49] update-alternatives --remove gl_conf /usr/lib/standard-x11/ld.so.conf || true [11:49] correct [11:49] ok [11:53] slangasek: here's the diff: http://pastebin.ubuntu.com/356525/ [11:53] tseliot: ack, looks right to me [11:53] perfet [11:53] perfect [11:57] slangasek, apw: uploaded [11:58] apw: it would be nice if you could try xserver-xorg-core 2:1.7.3.902-1ubuntu6 when it's available [16:01] apw: did the fix work? [17:54] lucid worked fine last few weeks for me, but after installing updates yesterday + reboot: now I get blackscreen instead of GDM ... dmesg/xorglog contains intel GPU hang errors --> http://pastebin.com/m19852642 [17:55] is this a known bug? if so, what is the LP bug number? [18:20] Sarvatt, any news about the execbuf error? [18:47] Duke`: what is the LP bug number for the execbuf error? [18:48] LP I don't know, but I have the freedesktop bug I think [18:48] https://bugs.freedesktop.org/show_bug.cgi?id=25475 [18:48] Freedesktop bug 25475 in Driver/intel "[i915] Xorg crash / Execbuf while wedged" [Critical,New] [18:50] okay that looks like what I have using lucid right now === ubottu is now known as ubott2 [22:14] Hi [22:15] I have serious problems with the latest versions of the xserver in ubuntu lucid [22:17] with kms enabled i get a black screen and the system hangs [22:17] without no screens are found and Xorg segfaults [22:18] but this does not match the latest bug reports (xorg segfaults) on bugs.ubuntu.com [22:22] xserver-xorg: 7.5+1ubuntu1, xserver-xorg-vido-intel: 2.9.1-1ubuntu1 [22:23] “sudo aptitude reinstall xserver-xorg-core” should fix it. [22:23] sounds easy enough to try *g* thx [22:26] it did indeed ... my i ask why? [22:31] thanks anyway ... good night utc+1