=== ogra_ is now known as ogra [10:39] Hello good people, is there a way that I can get atl1c on 9.04 with 2.6.30? [10:47] 2.6.30? [10:49] Yeah, I have an acer timeline 3810t, and 2.6.30 fixes a lot of problems with it. I installed it from the deb on kernel.ubuntu.com [10:50] norkakn, is that supported in the karmic kernel? might be worth trying the karmic kernel on there [10:50] my expectation is that the karmic kernel would work ok on jaunty, no guarentees of course [10:50] I think it is, how do I procure the karmic kernel? [10:51] My best experience was somehow apw's ppa [10:53] https://launchpad.net/~apw/+archive/daily [10:54] apw, reminds me, if you got time, you could upload the latest and greatest up to now. ;-) [10:54] smb, thats a good idea, its a bumper so it needs testing === woozy__ is now known as woozy_ [11:05] awe, well, it does the same thing that 2.6.31 from the mainline repo does, which is give me a big blank, black screen of nothingness [11:16] norkakn, oh yeah you need to tell it not to use KMS [11:16] i915.modeset=0 on the boot command line [11:18] apw: thanks, trying now [11:24] YES! [11:24] norkakn, yay [11:24] I'd tried the 31s before, but I didn't know about the KMS magic, thank you very much [11:25] norkakn, np, its not at all obvious [11:27] apw, I love this little machine, but nothing about it ever seems obvious. If my computers begin to be able to sleep, my life will be complete [11:27] suspedn not working then? [11:27] worth trying with karmic, works pretty damn well today for me [11:28] No, it is a known bug with the 3810T, no one seems sure if it is a BIOS bug or a Linux bug. [11:28] I'll try it out on my desktop though === cjwatson_ is now known as cjwatson [11:39] forgive me guys but i have this problem and i need to ask you: i have 2.6.28-15-generic on jaunty, now i need to recompile to change one small option, and i get 2.6.28.10-generic.. this way i can't use the backports and restricted modules so how can i recompile those packege too OR change the kernel version to exactly match the official one? [11:41] there are a lot of tutorials, but i can't found a sigle one that solve this problem... [11:45] Notch-1, if you are getting 2.6.28-10-generic you are using the wrong version of the source for the kernel [11:49] apw: 2.6.28*.*10 [11:49] i did apt-get install linux-source-2.6.28 ... [11:49] apw: It's just a wrong EXTRAERSION [11:49] mjg59: yes, but i can't change it, look at man make-kpkg [11:50] i can only change revision and --append-to-version ... [11:50] mjg59, oh yeah a . ... hrm [11:50] and the local version inside menuconfig, but i don't know this one [11:50] Notch-1, that implies you used the wrong tools to build it [11:50] apw: .10 will be the upstream stable point release EXTRAVERSION [11:50] if you use debuild it ought to get it right for you [11:51] somebody told me that -15-generic it's ubuntu notation while .10 is standard notation... [11:51] mjg59, yeah just didn't see it, saw what i thought it said not what it actually says [11:51] Notch-1, yep thats true [11:51] * apw idly wonder what option you need to change and why [11:52] apw: what's debuild? [11:52] its the tool for turning a source tree into a real binary .deb for installation [11:52] and basically is how packages get made in ubuntu [11:52] ah fine [11:53] i can use it to change this extraversion? [11:54] anyway it seems that the package linux.source and the kernel i'm running are very different, how can i compile a kernel like the original one? what options/files are missing? [11:54] so i can install backports and restricted.... [11:55] the problem is if you change an option and that changes the ABI then you can't install backports and restricted safely even if it lets you do it [11:55] hense my question as to the option you are changing [11:55] ah nice :D [11:55] it's the loop module [11:55] those other packages are built agianst that specific kernel [11:55] what you doing to it [11:56] recently it became included in the kernel, but for some reason me and other people need it at least as module... [11:56] that would probabally just about be safe ABI wise [11:57] the are a lot of posts on the net, everybody bothered recompiling the kernel to get the system back to work... but i don't think it's a complete solution, without the ability lo use backports and restricted modules.... [11:57] i would get the source either from 'apt-get source linux' or from our git tree, checking out the tag for the version you need [11:57] and then i'd change the config, and personally upload it to a PPA for building [11:59] * apw wonders if this is this aes loop thing that the maintainer won't simply change the name of to aesloop so that it can coexist with the normal loop module [12:00] nor will they work with upstream to get it into the kernel [12:00] apw: i don't think it's just a name problem... [12:01] as i recall its a philosphy issue, making life hard for everyone else problem [12:01] yes, but seems that noone need it upstrem so... [12:01] anyway, what if i just do make and replace the executable by hand? [12:02] no noone notice the new kernel... and i can install whatever i need... what do you think about that solution? [12:03] as in the module? [12:03] you should be able to build the module externally and just replace the loop one in the lib/modules [12:03] nono [12:03] bah ignroe that [12:03] i mean the kernel [12:03] i am losing my mind [12:03] :D [12:04] you could do that, but it wouldn't be very pretty [12:04] i would build your own version of the whole thing, with that module replaced [12:04] sure? :D [12:04] but i will still have the backport issue... [12:05] now i'm doing apt-get source linux-image-2.6.28-15-generic to see if this source are more similar to the original kernel... but i'm afraid :D [12:07] as we know the abi change (and there will be one) is benign mostly, you should be able to make a like 2.6.28-15.NNaesloop1 kernel in a PPA which can then be installed and the backports etc still work with it [12:08] excuse me, what's a ppa? :P [12:13] ppa is a personal package archive, which is like having your own ubuntu archive on the net [12:15] ah right, launchpad :D [12:15] have this features, now i remember thanks [12:17] yeah launchpaddy thing === ogra_ is now known as ogra === ogra_ is now known as ogra [15:12] Keybuk, are we expecting usplash to be absent with karmic tip? [15:15] yes [15:16] apw, did 'quiet' disappear from the grub line? [15:16] rtg nope they are still there [15:17] Keybuk, i assume there is a plan for the mess i the gap? [15:17] right [15:17] in fact, not having usplash is partially so we can see the mess we need to clean up === bjf-afk is now known as bjf [16:36] Keybuk, I've a server that won't mount a /dev/md0p1 partition at boot time. in fact, it borks the boot altogether if /home is mounted on this partition. what title should I use for the bug report that will catch your eye? [16:36] you're not the first person to say that [16:37] it breaks inside the initramfs, or breaks during normal boot? [16:37] Keybuk, can't tell. its got a bunch of udev messages [16:38] I think it must be during normal boot [16:39] ok great [16:39] you have it in front of you now? [16:40] Keybuk, that noisy hoover? of course not. its inconveniently off in another room [16:40] if you could boot it and try a few things, I'd appreciate it [16:40] boot with init=/bin/bash [16:40] and get the output of "blkid" [16:40] (just on its own) [16:40] Keybuk, can do. what would you like [16:40] ? [16:41] Keybuk, ok, back in a sec [16:44] Keybuk, ok, blkid shows /dev/md0p1 with a UUID and file system type of ext4 (but its not mounted even though its in /etc/fstab) [16:44] right [16:44] but, does it also show the same UUID for the underlying filesystem? [16:46] Keybuk, I'm not sure what you mean. it only shows 1 UUID [16:46] blkid only outputs one thing? [16:46] for that partition, yes [16:47] I don't understand [16:47] blkid should output lots of lines [16:47] and it does [16:47] ok, can you take a picture of them or something? [16:48] Keybuk, yep, gimme a bit [16:48] also mdadm -D /dev/md0 would be helpful [16:53] Keybuk, http://kernel.ubuntu.com/~rtg/md0 [17:01] rtg: great, and what's in /etc/fstab and /proc/self/mountinfo ? [17:02] on sec [17:05] Keybuk, http://kernel.ubuntu.com/~rtg/md0/imgp0763.jpg [17:06] thanks [17:06] that's very useful [17:06] so, I know what's happening but not how to fix it yet [17:07] for some reason, it's showing the UUID and filesystem type, etc. for sdb1 as it is for md0p1 [17:07] is it because mdadm isn't getting run? [17:07] and sdb1 is winning [17:07] so it tries to mount that, but fails [17:07] mdadm must be being run, your array is visible, up and enabled [17:08] I guess for now I can just mount it later in the process. [17:08] yeah [17:08] I need to debug this a bit more really [17:08] I've not seen these md0p1 things before [17:09] I didn't even know you *could* partition raid arrays [17:09] usually I see /dev/md0 ;) [17:09] I think this is a udev bug [17:09] but it may also be a blkid bug, since blkid shouldn't be reporting raid members as filesystems [17:10] Keybuk, its been awhile since i set this up, but I beleive partitioning was part of the setup documentation [17:10] right [17:10] if you fancy an experiment [17:10] try downgrading udev back to 146 [17:11] can do. [17:11] but don't just replace the package [17:11] because that won't work so well [17:11] instead grab the deb, and get the udevd and udevadm binaries out of it [17:11] and put those in place [17:11] (back up the 147 ones) [17:11] and see if that makes a difference [17:11] k, first I have go unwire this thing to get it to boot normally. how does one stop grub2 these days? [17:12] hold down shift [17:12] wasn't it the shift key? [17:12] hmm, no joy there [17:13] you couldn't get grub2? [17:13] or udev didn't help? [17:19] Keybuk, I can't get grub2 to stop during the boot process so that I could modify the kernel command line, so I edited /boot/grub/grub.cfg and rebooted. now I'm kind of stuck 'cause / is mounted readonly. [17:19] mount -o rw,remount / [17:20] Keybuk, yeah, I just came back in here to read the man page :) [17:20] (just be sure to mount -o ro,remount / or SysRq-U before rebooting :p) [17:24] Keybuk, udev 146 has already been obsoleted from the archive. do you have an amd64 copy lying around? [17:41] # The default value of the child_runs_first scheduler sysctl knob has been changed to "false." This causes the parent process to continue running after a fork() rather than yielding immediately to the child process. See this article for more information on 2.6.32 scheduler changes. [17:41] oh [17:41] sigh [17:41] happy days [17:41] now we see all the race conditions that people think they fixed, suddenly break again [17:41] admittedly, their fault for not actually fixing it, but hey :p [17:42] Keybuk, is that the root of the udev issue? [17:43] huh, no [17:43] sorry, was just musing on the 2.6.32 merge window [17:43] rtg: you can get udev 146 from LP still [17:44] https://launchpad.net/ubuntu/karmic/+source/udev/146-1 [17:44] there should be a "karmic amd64" link there [17:46] Keybuk, http://launchpadlibrarian.net/30822986/udev_146-1_amd64.deb [17:46] there you go :) [17:47] Keybuk, replace both udevadm and udevd ? [17:47] yes [17:49] yay [17:49] https://edge.launchpad.net/ubuntu/karmic/+source/udev/146-1 [17:49] err [17:49] proc-connector-add-event-for-process-becoming-session-leader.patch [17:49] is in akpm's big list of patches [17:53] Keybuk, no difference in nehavior. /home.md0p1 is still not auto-mounted from the fstab. uit works if I put 'mount -a' in /etc/rc.local [17:53] with -146 (that is) [17:53] ok [17:53] that's good to know [18:03] rtg: can you check something for me [18:03] ls /sys/block/sdb [18:04] do you have sdb1 in there? [18:04] if so [18:04] fdisk -l /dev/sdb [18:04] root@tyler-b:~# ls /sys/block/sdb [18:04] alignment_offset capability device holders queue removable sdb1 slaves subsystem uevent [18:04] bdi dev ext_range power range ro size stat trace [18:05] root@tyler-b:~# fdisk -l /dev/sdb [18:05] Disk /dev/sdb: 500.1 GB, 500107862016 bytes [18:05] 2 heads, 4 sectors/track, 122096646 cylinders [18:05] Units = cylinders of 8 * 512 = 4096 bytes [18:05] Disk identifier: 0xe3da5fd4 [18:05] Device Boot Start End Blocks Id System [18:05] /dev/sdb1 1 488386496 1953545982 83 Linux [18:05] aha! [18:05] *but* your mdadm says that "/dev/sdb" is the RAID member [18:05] not /dev/sdb1 [18:05] right [18:05] what does blkid -p /dev/sdb say? [18:05] root@tyler-b:~# blkid -p /dev/sdb [18:05] /dev/sdb: VERSION="0.90.0" UUID="8d408524-0506-f40d-4659-8b40d611b84e" TYPE="linux_raid_member" USAGE="raid" [18:06] AHA! [18:06] doesn't mdadm put some crap in fron of the partition table? [18:06] apparently not [18:08] maybe mdadm is putting its metadata in the MBR? [18:08] (is bcmwl a bit busted in α6 btw?) [18:09] dunno about bcmwl. it should be under jockey control [18:09] is there room in the MBR? it doesn't sound like a reasonable place for it [18:11] no, no idea [18:11] udevinfo -q all -n sdb [18:11] udevinfo -q all -n sdb1 [18:11] udevinfo -q all -n md0 [18:11] udevinfo -q all -n md0p1 [18:11] would be handy [18:11] Keybuk, is that from a dev package? [18:11] sorry [18:11] udevadm info [18:11] fingers still haven't learned ;) [18:12] hmm, spewage. I'll collect this in a file [18:14] this is a RAID 0 isn't it [18:15] Keybuk, yes, 4 spindles [18:15] I have a hypothesis [18:15] Keybuk, http://kernel.ubuntu.com/~rtg/md0/udev-info.txt [18:16] thanks [18:17] ok [18:17] well there's a couple of interesting bits there [18:17] where did you read this documentation about partitioning your RAID like that? [18:17] Keybuk, hmm, probably wikipedia. lemme look [18:18] ok [18:18] that's all I need for now [18:18] bbl [18:18] Keybuk, k, I'm travelling to PDX starting in a couple hours === bjf is now known as bjf-afk === bjf-afk is now known as bjf