[00:51] did anyone else test the 2.6.24 kernel? [00:51] i dont get any other kernel module to work, no nvidia, no fglrx, nothing... === asac_ is now known as asac === TheMuso_ is now known as TheMuso [12:33] Problem is a kernel oops on boot. Stock ubuntu-server kernel oopses, and I can get it to come up. It hangs while setting up udev. I can get the system to come up by killing everything (either ctrl-alt-del or magic-sysrq + k). [12:33] any hints? [12:34] I compiled a new kernel that was well-suited to my hardware (which is: Core 2 Duo at 2.2 Ghz in a Gigabyte GA-G33M-DS2R, 3 SATA hdds on the Intel fakey-RAID (running straight AHCI, no goofy RAID), and DVD-ROM on the other IDE). [12:34] I havent tried the desktop kernel work though.. [12:36] I could try jumping runlevels. Trouble is udev starts in single-user, too. [12:36] Can you start it with init=/bin/bash ? [12:39] IntuitiveNipple: I can start it, eventually, with a full system. But it will oops while booting. It occurs to me that I may not have tried futzing with init - another good thought. [12:41] It starts from an initrd that has bbox... it'll have to be /bin/sh, not /bin/bash. But I'll see if that resolves the oops. [12:41] I'm thinking if this is an udev problem by disabling udev and booting, then dmesg | grep -i "error" [12:42] Or any more possible thoughts on this? [12:43] I was thinking along similar lines. Does it oops whilst udev is initialising, or before udev starts. Your previous comment on that was unclear. [12:43] I will see if I can find anything more. Aside from the oops, I have not found much in dmesg. [12:44] If it is whilst udev is initialising stuff, then I'd drop some /etc/udev/rules.d/ rules that I suspected, try to narrow down the precise cause. [12:45] Too many unknowns to narrow it down reliably... [12:45] Have you got the log showing the oops? [12:45] IntuitiveNipple: It appears to be while udev is initializing. I magic-sysrq-t'd while it was hung, and the only userland processes were related to udev startup. [12:46] IntuitiveNipple: Should have, yes. [12:46] let me get to this box and pastebin it. [12:46] Best to post a bug report to Launchpad, attach it plus an output of lspci -nn [12:47] You think its a bug? [12:47] If it oops, yes! [12:47] I want to be sure not to waste your(Developers) time, [12:47] Which kernel is this? Gutsy? [12:48] IntuitiveNipple: There is more funkiness to this setup. Root is on one of the SATA drives. /home and /var are on an MD RAID5 across the three drives. I do not know if that is giving it fits. Doesn't appear to. [12:48] Yes, gutsy. [12:48] If it turns out not to be a bug, it'll only get marked invalid, but once it is in launchpad people will look at it and if they have input, you'll hear about it [12:48] ooo ooo ooo [12:48] My kernel is 2.6.23.8, and has the same problem. [12:49] I'm not sure about this, *but* I'm sure I saw a conversation along the lines of, Ubuntu can't cope with separate / and /var, and I'm sure that is in launchpad [12:50] Are you a dev? [12:50] I'm on the kernel ACPI team, and I do a lot of debug work [12:50] Yeah its gotta be there on launchpad [12:51] COuld you direct me to that bit in launchpad or where do i go here on? [12:52] Are you using LVM? [12:52] https://bugs.launchpad.net/ubuntu/ [12:53] IntuitiveNipple: Just md [12:54] I wish I could find mention of that /var issue... I may have just seen people talk about on IRC [12:54] or, it may be in the mailing list! Grrr [12:55] IntuitiveNipple: I have logs showing the oops. I apparently no longer have the log showing the process dump, which is disappointing. I'll regenerate that one. [12:56] then post the launchpad bug # here [12:57] In the errors, he most important line appears to be [ 40.923128] BUG: unable to handle kernel paging request at virtual address f8b7fffc [12:57] s/he/the/ [12:57] Then it dumps an oops. [12:58] IntuitiveNipple: Here as in ? [12:58] oh you mean I paste it here? [12:58] Ok. [12:59] in this room, so any dev that looks at the back-history can click through to the bug [12:59] If you type bug #163011 [12:59] Launchpad bug 163011 in linux-source-2.6.22 "[FTBFS] unknown field ‘aio_reserved3’ building User Mode Linux" [High,Triaged] https://launchpad.net/bugs/163011 [12:59] See? uboto provides a link for you [12:59] How do I add it? [12:59] Ok [13:00] So I post it in launchpad and just give you the number later I guess? [13:00] So they can look through it [13:00] Once you've report the bug, simply type in here bug # [13:00] Gotcha [13:00] Yes [13:01] Thanks [13:02] i wanted to disable acpi.. I really wanted to avoid it. That seems to be the solution to a lot of these issues. [13:05] I've seen a lot of issues with acpid loading modules that crash a system, and stuff like cpuspeed (throttles the CPU when there is not much load) causing strange errors [13:06] Anyhow Im filing a bug report [13:12] Unless the motherboard has a broken DSDT, ACPI should be fine. With Ubuntu you can correct the DSDT and load the fixed one at boot-time [13:15] Good so that can be overwritten [13:18] Generally, the Linux approach is, the ACPI interpreter works to specification (hah!) so if it breaks, it is an issue with the BIOS (and it usually is!). Windows will include workarounds to play nicely. There's another approach that says if it works with Windows it should work with Linux. [13:19] IntuitiveNipple: 163551 [13:20] bug #163551 [13:20] Launchpad bug 163551 in ubuntu "Kernel oops during boot on gutsy server" [Undecided,New] https://launchpad.net/bugs/163551 [13:23] That'll do? [13:23] It looks to be the AGP bridge [13:25] That was my first thought, so I disabled the AGP bridge in the custom kernel (I don't need it). [13:26] It just blew out. It now occurs to me that that may have been because of a bad initrd, however - it didn't oops, it just collapsed trying to mount root. [13:26] the fault appears to happen at intel_i915_configure() [13:27] Would it be helpful for me to compile this kernel without agp and a decent initrd? I want to recompile anyway, because I want a monolithic kernel. [13:30] you could try blacklisting intel_agp [13:32] That would certainly be quicker. Let me see if I made that a module. [13:35] IntuitiveNipple: .. stock kernel. Sorry, I'm a bit slow today. [13:36] lol :) [13:36] I know how it is [13:36] IntuitiveNipple: no oops. Boots perfectly. [13:37] ok... can you do without intel_agp? [13:37] You'd best add a comment to the bug describing that as a workaround for the issue [13:38] So that does it for me. I'll update the bug. [13:38] Thanks [13:38] IntuitiveNipple: Absolutely. This will be headless for most of its life. [13:38] I wish I was! [13:39] Thank you for the help. [13:39] You're welcome [13:41] So, next question... it _appears_ to be running fine with /var off on its own. Is this going to create a problem for me down the road? I like having /var contained, if at all possible. [13:42] I wish I could remember where I saw the discussion on that, but I can't. Just test well and then ignore me :p [13:43] heh. Wilco. Thanks for the heads up in any event. [13:47] It *may* have been to do with ubiquity and the desktop CD installer, so that wouldn't affect you [13:48] It would be very annoying if some system cared much about an adminisrative decision. [13:49] I've seen similar problems elsewhere that made a bit of sense. For instance, sometimes if you have to start something funky in order to mount /var, there will be a problem with delayed writes. This happens when people crypto /var, for instance, and the networking subsystem is started first and tries to write to it. [13:49] Thinking about, the installer would make sense. Because a separate /car implies it'd have to 'know' to mount / then mkdir /var, then mount /var [13:49] /car??? lol [13:50] lol [13:50] Okay. For what it's worth, I didn't install this way - I installed onto a monolithic fs, then added the raid arrays and mounted them. So there is _a_ /var right away, just not the one that will be there when daemons start. [14:02] ok IntuitiveNipple Thanks a lot [14:02] I'll be back later:) [14:02] don't hurry.. I can do without any more bugs today... I'm all full up :) [14:03] Haha I'll still see you in a bit:) Nice talkin to you [14:15] moin [16:57] what os-virtualization (like user-mode-linux, openvz,...) is supported by ubuntu? [17:36] socketErr: uml, qemu, and kvm are all in universe. [17:38] as well as virtualbox-ose [17:52] something must be in 2.6.23 that fixes this [17:52] http://ftp.tu-chemnitz.de/pub/linux/ubuntu-releases/kubuntu/gutsy/MD5SUMS [17:52] err [17:52] https://bugs.launchpad.net/linux/+bug/54294 [17:52] Launchpad bug 54294 in linux-source-2.6.20 "Failed to allocate mem resource #6 ... (dup-of: 159241)" [Low,Confirmed] [17:52] Launchpad bug 159241 in linux "[Tracking Bug] PCI resource allocation warnings" [Unknown,Fix released] [17:52] that [17:52] some laptops can not boot at all [17:53] also 2.6.23 fixed another issue with intel onboard gfx which could not detect the right screenres from vga monitor