=== thegodfather is now known as fabbione [09:58] hello [09:58] i need some help with the ubuntu kernel build process [09:58] may i ask here? [09:59] i'm using https://help.ubuntu.com/community/Kernel/Compile [09:59] ..and a checkout of commit 273f7b551a420580307fa414fe616f0e276a4035 [10:00] it should match my current kernel 2.6.24-16-generic [10:01] i did some minor changes to the i386/config.generic because i need a SLAB allocator [10:02] debian/scripts/misc/oldconfig i386 # everything fine so far [10:03] then i do a ... AUTOBUILD=1 NOEXTRAS=1 fakeroot debian/rules binary-generic [10:03] it tells me /usr/src/linux-2.6.24-16 is not clean, please run 'make mrproper' [10:04] however "git-ls-files --others" says my HEAD was clean [10:05] do you know what i'm doing wrong? [10:09] fs_: Use the KernelMaintenance and KMStarter documents found at https://wiki.ubuntu.com/KernelTeam/KnowledgeBase [10:09] fs_: I personally would start from scratch with a clean kernel source tree and see if you can build a kernel without patching it first by following wiki instructions [10:10] fs_: the instructions you are following are for a very old kernel [10:12] i'll try that [10:48] cking: if i don't change anything it build's fine, sorry it took so long. [10:48] i'll now try to follow https://wiki.ubuntu.com/KernelMaintenance [10:48] good plan.. :-) [10:50] fs_: just take care that you don't run the in-kernel build system (make menuconfig, oldconfig, etc.). That confuses the Ubuntu build system. [10:50] ok [10:50] i'll try and see -- hope so :) [10:55] can i start with "debian/rules startnewrelease" or do i have to "debian/scripts/misc/getabis 2.6.24 16.30" first? [11:17] cjwatson: I now have further info on http://pastebin.ca/1002222 [11:17] http://www.peters-cnc-ecke.de/forumupload/uploadFiles2008/11479_115669130125_IMGP1627.JPG [11:18] this isn't really my area ... [11:25] ok, I'll ask smb if he's around later [11:26] would he be more appropriate? === lamont` is now known as lamont [11:48] alex_joni: well, certainly somebody on the kernel team, I don't know exactly [11:49] cjwatson: thanks [12:04] i followed the instructions on https://wiki.ubuntu.com/KernelMaintenance very carefully, but i still get the same error [12:05] still ... "/usr/src/linux-2.6.24-16 is not clean, please run 'make mrproper'" [12:06] after " scripts/kconfig/conf -s arch/x86/Kconfig" [12:08] i started at "Development cycle" with a fresh checkout [12:08] < fs_> i did some minor changes to the i386/config.generic <- where exactly is that file? [12:08] at debian/config/i386/config.generic [12:09] i took the /boot/config-2.6.24-16-generic ran make menuconfig on that changed SLUB to SLAB [12:09] cp'ed it to debian/config/i386/config.generic [12:10] when you run make menuconfig you taint your sources [12:11] put the config.generic somewhere safe, then revert all changes to your git checkout [12:11] tried both "debian/rules updateconfigs" and the long version with the for loop under "Updating configs" [12:11] git reset --hard (or soemthing like that) [12:11] aha [12:13] i think i'll start from scratch [12:14] do you mean i should make the change to debian/config/i386/config.generic with an editor [12:15] you say the menuconfig on the sources was bad -- i won't do this now. [12:15] right [12:16] oh lord [12:16] if you do any changes on the sources directly (even running make menuconfig or such) it will confuse the ubuntu scripts [12:16] all right! [12:16] fs_: at least that's how I made it work :) [12:17] but you should simply get a clean checkout, put your new config in place, use dch to bump the version number, and build the binary package [12:17] ok [12:17] (remember that newer versions released by Ubuntu will one day probably update your locally installed package) [12:18] so I would uninstall the meta packages, and use a very high number for the kernel package [12:18] oh thats ok, i can put them on hold [12:18] if you're carefull.. then it's ok [12:19] i have to get debs first [12:19] dch with any args? [12:19] dch -i to bump version [12:26] is that "git-update-index debian/config/*/*"... mandatory? [12:32] alex_joni: hit me! :) [12:32] it's building [12:32] how great! :D [12:32] fs_: you should build/configure outside the kernel source tree to avoid tainting the sources. e.g. make O=`pwd`/../build menuconfig, where 'build' is a directory above the Ubuntu kernel source [12:33] amit: yes! and i'll never forget [12:33] ...until building changes ;) [12:34] if that works .. i've to go for the lum and lrm's [12:35] fs_: I've put something in for UDS to create a debian/rules target to automate this [12:36] amitk: are you acustomed with ata drivers? [12:36] amit: what's an UDS? [12:36] fs_: Ubuntu Developers Summit [12:37] alex_joni: not an expert [12:37] amit: that's really, nice [12:37] amitk: mmkay.. had this issue: http://www.peters-cnc-ecke.de/forumupload/uploadFiles2008/11479_115669130125_IMGP1627.JPG [12:38] alex_joni: intel chipset? [12:39] amitk: I think so [12:39] amitk: regular kernel (-generic) seems to boot ok, this is using a custom flavour [12:40] alex_joni: what flavour? [12:42] -rtai [12:43] alex_joni: your own flavour? [12:44] amitk: yes [12:46] amitk: wonder if it's related to bug #206635 [12:47] https://bugs.launchpad.net/ubuntu/+source/linux/+bug/206635 [12:47] alex_joni: You have to give me context before I can help you. Did you change the config from -generic? When does this appear? Did you rebuild initramfs? etc. [12:49] alex_joni: or if you've already talked to somebody just wait. They should be coming online soon :) [12:49] amitk: the issue is that it's not my machine, will try to bring the guy with the machine in here (I mostly built the kernel debs) [12:50] amitk: I understand more info is needed, so I mostly would appreciate pointers in what direction I should start looking [12:50] amitk: it's a changed config (basicly all power management is switched off: ACPI, APM, etc) [12:53] alex_joni: dmesg (if you even boot), lspci -vvnn, details of config changes are needed. BTW, you said that -generic works, right? [12:53] amitk: yes [12:53] alex_joni: in that case, lsmod output is also helpful with generic. [12:53] amitk: beeing that early in the stage it's probably something in initramfs? (I mean it doesn't get to load modules from lum..) [12:54] alex_joni: have you recreated your initramfs? [12:54] alex_joni: https://wiki.ubuntu.com/KernelTeamBugPolicies for more info [12:55] amitk: yes [12:55] amitk: I was a bit worried about submitting a bug at launchpad.. as it might not be something wrong in the ubuntu source/packages [12:55] but rather something I didn't do right.. [12:56] alex_joni: it isn't a bug if the default Ubuntu kernel works :) But it will help you have all this info on a single page in order to get help [12:57] amitk: ok, thanks.. will gather more data [12:58] amitk: different topic: machine freeze and blinking keyboard lights means kernel panic? [13:03] alex_joni: almost always :) [13:10] amitk: cool, thanks [13:42] I am using Ubuntu 8.04 64-bit version, if i install any 32-bit applications like browsers, datbase clients ...etc. The 32-bit based applications not able to reach /etc/resolv.conf file to communicate to the network. Any one tell me how to resolve this issue ? [13:43] shashi: ask on #ubuntu for support === maks__ is now known as maks_ [13:52] where can i get ubuntu source code [13:53] see https://wiki.ubuntu.com/KernelGitGuide [13:58] what language is it written [13:59] i mean the ubuntu kernel [14:01] in c or in english afaik? [14:01] i want to learn and understand about ubuntu kernel.. [14:02] nice :) [14:02] the ubuntu-kernel is a linux-kernel [14:05] anything more to tel me about kernel.... and can i download the source code... and get started [14:06] depends on what you want to do [14:06] linux is http://kernel.org/ [14:08] i actually want to learn how a kernel is written especially about linux(my sweetheart:-[) .. thats what i am interested in [14:08] there are serval books about that issue [14:09] what are the pre-requisites.. i mean the language and skill i need to know ?? [14:10] it's mainly the c programming language [14:11] then... any thing else to learn before i jump into linux kernel [14:11] ??? [14:12] there is so much to learn [14:12] If you don't already know C pretty well, the linux kernel is not a good place to start. [14:14] C means... what subconcept you actually mean... do i need to know pointers well ?? then about various datastructures ??? [14:16] can you pls confirm that suspend-to-disk does not work with swap on file? [14:18] xivulon: Correct [14:19] devindia: there are serval books about the linux kernel one is "linux device drivers". you can find it at http://lwn.net/Kernel/LDD3/ [14:19] mjg59: would it be possible to disable hibernation within pm-util if any device in swap -s is on file? [14:19] I had a couple of users ending up with a frozen system [14:20] I can file a bug if you wish [14:20] That would be good [14:21] is this a place for kernel developers to chat?? hope i am not disturbing anyone?? [14:22] mjg59 bug #224697 [14:23] devindia: this is channel for Ubuntu kernel development. For general-purpose kernel change you should go to a channel from kernelnewbies.org [14:23] s/change/chat/ [14:25] ok... i will do that [14:26] but i want to know about ubuntu kernel..... the site is telling me about something else [14:27] does ubuntu use launchpad for its development ?/ [14:29] devindia: ubuntu kernel is basically the linux kernel + some extra patches. Everything you learn at kernelnewbies is applicable to the Ubuntu kernel. [14:29] devindia: yes, LP is used for development [14:30] development in the sense i asked for "code" if yes... can i get the sourcecode from launchpad [14:33] devindia: most likely you get the linux kernel from www.kernel.org [14:34] devindia: the ubuntu kernel code is at http://kernel.ubuntu.com/git [14:37] where is git coming in... if ubuntu uses launchpad.. it must use bazaar... am i right? [14:38] devindia: no, the kernel is an exception [14:39] In wubi I have a "Common Workarounds" boot mode for the installer [14:40] so only patches are made using bazaar [14:40] At the moment it is mostly acpi=off, but what would be a reasonable combination? [14:40] ??? [14:40] I was thinking: "Common Workarounds Boot" => xforcevesa acpi=off noapic nolapic all_generic_ide floppy=off irqpoll [14:41] is this okish? [14:41] devindia: EVERYTHING is in git [15:09] rtg: you've tagged the cgroups bug verification-needed. where is it built? [15:10] xivulon: I'm sure I said this before, but it's really not a good idea to lump everything into one like that [15:11] xhaker: its in progress. add the -propsed repo to your /etc/apt settings and you'll get notified when the build is complete. [15:11] s/-propsed/-proposed/ [15:11] xivulon: as I pointed out before, gfxboot offers a set of checkboxes each of which separately represents a common workaround people might need, and wubi should do similarly IMO === cradek_ is now known as cradek [15:12] cjwatson I have to use menu.lst [15:12] so I have a bit less flexibility [15:12] xhaker: since its an ABI bump you'll want to wait until LUM etc are also uploaded. [15:12] I can have one menu entry for each workaround though [15:12] xivulon: I'm concerned that just hammering them all in will break some systems in different ways [15:12] xforcevesa, for instance, produces severely degraded video performance if you don't need it [15:13] now I have Safe Grapghic Mode = xforceves and ACPI workarounds = acpi=off noapic nolapic [15:13] all_generic_ide will probably change the probe order of devices [15:13] I'm sure acpi=off breaks suspend [15:13] but I noticed a few users required irqpoll and/or all_generic_ide [15:13] and so on [15:14] so maybe I can do 3 groups [15:14] every one of those options will be detrimental in some way, otherwise it would be on by default [15:15] got that, shall we then keep Safe Graphic Mode and ACPI workardounds as they are and add a new one for irqpoll + floppy=off? [15:15] are irqpoll and floppy=off particularly connected to each other/ [15:15] ? [15:15] rtg: yep, i know the drill. but i didn't notice an upload to proposed yet [15:15] putting the 2 together since a couple of users apparently required them [15:15] I think you need to do the research to understand them before just dropping them in [15:15] let me fetch the post [15:16] that's why I am here :) [15:17] xhaker: yeah, the SRU process is kind of a pain for the kernel. I've got all of the bugs marked for SRU. I'll bug pitti soon to release the upload. [15:17] cjwatson: http://ubuntuforums.org/showthread.php?t=773759 [15:17] oh.. it's in the queue :D [15:18] xivulon: I have to say that I think it's much, much more important to gather information about what systems need those parameters and feed that information back to the kernel team, than to worry about producing a complete list [15:18] firstly, you're not going to be able to produce a complete list, period [15:19] secondly, if the kernel team get information about what systems are going wrong, they can add quirks to the kernel so that it uses those by default, which will help many more people [15:19] so the first response should ALWAYS be to get the necessary information (dmidecode, sometimes lspci -vvnn, etc.) so that quirks can be installed by default [15:20] absolutely [15:20] when you find yourself designing complicated systems to divide workarounds into different groups, that's a sign that it's gone too far [15:21] maybe some of these are just needed for a few systems, and we could clean those up for 8.04.1 [15:21] but sometimes people cannot boot at all so gathering that info becomes difficult [15:21] you aren't going to be able to assemble a complete list of why people can't boot, though [15:22] I'm just worried that the necessary followup might not be getting done here [15:22] all I aiming for is to provide a simpler avenue to get around the most common issues [15:22] 2 or 3 (xforcevesa, and acpi=off are already there, so it's a matter of deciding whether I should add something else for irqpoll & co) [15:22] right, but above you were talking about "a few users" - I'm just not sure there's enough data yet to indicate that these are really common, rather than just being something that affect a very small number of systems [15:23] http://ubuntuforums.org/search.php?searchid=40276200 (search wubi for irqpoll) [15:23] (which might be ones that lots of users use) [15:24] someone even added that as a recipe to the wubiguide, that is what really triggered my attention [15:26] none of this is specific to wubi surely, for that matter [15:26] absolutely [15:36] cjwatson: Well, ideally we fix the bug rather than blacklisting [15:36] sure, I just mean as an absolute minimum [15:38] cjwatson, I will leave it as it is then, If I observe more irqpoll or all_generic_ide recipes we will discuss whether it is worth to add an extra menu entry [15:39] * cjwatson nods [15:57] who is our best FS person to look at a unionfs bug that bites classmate and MIB atm ? https://bugs.launchpad.net/ubuntu/+source/linux-ubuntu-modules-2.6.24/+bug/224754 [16:00] I get this when I run the "check disk integrity" thing on the 64-bit desktop installation CD: "BUG: Soft lockup - CPU#1 stuck for 11s! [exe:2673]" [16:00] or release 8.04 [16:01] ogra: cking [16:01] thanks [16:01] googling that string suggests proprietary wireless drivers or something as the culprit, but surely none can be loaded at boot time off the live-cd? [16:01] seems to be caused by a hardlink in the sudo package [16:02] somehow unionfs doesnt like hardlinks atm [16:02] Zet: all hardware modules will get loaded off the live-cd [16:03] rtg: even by the integrity checker? [16:03] rtg: does the live-cd have proprietary drivers on it? [16:04] Zet: I can't remember if LRM is there, but I think not. [16:04] what is LRM [16:04] linux-restricted-modules (where all the binary blobs live) [16:05] well anyway, I checked the md5sum of the .iso I burned and it's ok === BenC__ is now known as BenC [16:05] I got Kubuntu installed but it completely freezes after a short while [16:05] Zet: are you getting a stack dump from the BUG message? [16:05] no [16:06] I haven't been able to check the system logs yet either [16:06] maybe I'll burn a Knoppix CD or something [16:06] Zet: what do you have that you can unplug in order to try and isolate the problem? [16:07] umm [16:07] a USB memory card reader, a wacom tablet, a digital tv card...? [16:07] ogra: OK, will look into it. What hardlinks are you referring to for this bug to occur? [16:08] but first things first: doing the dishes to reduce nagging [16:08] cking, the sudo package ships two ahrdliks [16:08] get back to you later [16:08] *hardlinks [16:09] cking, there is one in hardy-proposed which makes unionfs oops ... i bet you can try to pull it onto the liveCD to reproduce [16:09] ume as well as classmate use both a standard rofs/cow/unionfs setup similar to the CD [16:11] ogra: can you add some more details on how exactly to trip this on the bug report so that I can easily reproduce it? [16:12] ..if that's not too much of a problem. [16:13] i added: to reproduce use a system with unionfs underneath (liveCD might suffice) , enable the hardy-proposed repo and install the sudo package from there (it ships two hardlinks which i suspect to be the cause here) [16:13] does that suffice ? [16:14] cking, lool and me are both available for you if you need any tests etc [16:14] (lool in #ubuntu-devel) [16:16] * ogra tries with a live iso in virtualbox .... [16:16] ogra: that info is fine. It's a pity unionfs is being used - aufs is an alternative union fs [16:16] yeah [16:16] looking forward to intrepid for it [16:17] ..it's in hardy lum too [16:17] i already have the necessary patches for ltsp waiting where we use unionfs extensively ... [16:17] but nothing uses it yet in the ubuntu world [16:18] so i have to stick with the supported stuff for the release [16:19] ..the kernel oops may take some figuring out.. but I start digging early tomorrow if that's OK. [16:20] sure [16:21] i'll roll the updates into a new rofs anyway :) and automatic updates are disabled on the classmate ... not sure how urgent its for lool though [16:21] confirmed on the liveCD as well [16:23] ogra: just to confirm - which LiveCV are you using? [16:23] s/LiveCV/liveCD/ [16:24] the released hardy iso [16:24] in virtualbox [16:26] OK [16:53] rtg: Is it sufficient for SRU (Bug #224788) ? [16:55] abogani: hang on, lemme look. [16:56] Sorry if it isn't looks good but it is my first SRU request. :-( [16:58] abogani: what about the mutex fix? The changelog entry (from the git commit) implies that the toshiba_acpi module has always been enabled. [16:59] rtg: No BenC disabled it times ago. [17:00] abogani: ok, so the SRU is really 'enable toshiba_acpi'. The changelog entry (that I created) is misleading. [17:08] rtg, ping [17:09] dhaval: yo [17:09] rtg, how do i clone your kernel tree? [17:09] the stutters should not be happening [17:09] we should fix it. [17:10] dhaval: https://wiki.ubuntu.com/KernelMaintenance [17:11] git://kernel.ubuntu.com/ubuntu/ubuntu-hardy.git ? [17:11] yes [17:13] rtg, thanks! [17:13] i might ask you test out some patches to confirm if i manage to nail this one tonight [17:15] dhaval: what problem are you pursuing? [17:17] rtg, the stutters you mentioned in the last mail to the bz [17:18] dhaval: that was the original behavior with USER_SCHED. Isn't it an accurate description? [17:18] rtg, well, that is not expected behavior. it should not stutter. [17:18] i thought we had fixed it long ago [17:19] but we don't seem to have, so let's go and fix it. [17:19] I'm not sure what you mean. the bug report is describing _why_ I changed -generic from USER_SCHED to CGROUP_SCHED. [17:21] smb: what is going on with https://bugs.edge.launchpad.net/ubuntu/+source/module-init-tools/+bug/140511 ? [17:21] smb: is it against the wrong package? [17:23] smb: never mind. I found it: http://kernel.ubuntu.com/git?p=ubuntu/ubuntu-hardy.git;a=commit;h=77d6ec19785ff831150378a1073c21024e88fb05 [17:23] rtg: Why is that assigned to you? I fixed that in the kernel. init tools was wrong because blacklisting was no option. [17:23] smb: I'm going through the SRU process on these bug reports. its a real pain in the ass. [17:24] rtg: for hardy? [17:24] yep [17:24] rtg: I am confused. You need an sru for something that was in the release? [17:25] smb: as far as I can tell it was not in the release. [17:25] though it looks like it should have been. [17:26] rtg: hm. thought we had an upload for release after i put that in [17:26] smb: according to the log, you appear to be correct. [17:26] looks like 'debian/rules insertchanges' didn't do its job correctly. [17:27] rtg: must be something like that. i was a bit late but not that late [17:27] smb: it was committed Apr 9, I uploaded twice after that. [17:29] smb: hmm, maybe not. I'm gonna revert to the -16.30 tag and see just what is in the source. [17:29] rtg: so I thought. Well the last commit before was -16.30. Hm [17:30] rtg: No, not correct. It was the first -16 ABI [17:31] rtg: And you checked in -16.30 relase some time after that [17:32] smb: well, the blacklist code is in the -16.30 tag. damn, sure messed up m y changelog. [17:33] anyway, I gotta bolt for a couple of hours. back in a bit. [17:33] rtg, sorry missed your message. what i mean, is that even if a media player is getting lesser cpu time, it should still get scheduled quickly enough for stutters to not be noticable [18:08] Howdy folks, is there a wiki page/doc suggesting how to do a complete flavor build (kernel, modules, restricted modules, metapackages, etc) [18:08] I've automated my main kernel build process, but the auxiliary packages have me wondering if there is a better way [18:09] compbrain: not that I know of, but there's info in debian/binary-custom.d (for the kernel) [18:11] alex_joni: Yea, I've got that much. We're producing kernel iamges to our team ppa using that method for zumastor.org [18:11] but we've had requests for the other supporting packages to be built as well [18:13] compbrain: it's not very different for lum and lbm [18:13] pull them from git [18:13] then "NOEXTRAS=1 fakeroot debian/rules binary-modules-ppa && dpkg-buildpackage -S -I -i" [18:14] I also add DEB_BUILD_OPTIONS=parallel=2 in the front of that (so it goes a bit faster) [18:14] righto [18:14] lrm is not in git faik, so I usually get it with apt-get source linux-restricted... [18:15] Yea, right now i've got a py script to pull the latest kernel source package from our mirror, throw in the new flavor, flip a few bits, and output the modified source package for ppa upload [18:15] compbrain: maybe pulling from git will be easier [18:15] I'll try anything once ;) [18:16] heh, same here [18:16] but for lrm you can only get it with apt-get source afaik [18:16] same for meta [18:16] (for meta it's simply: fakeroot debian/rules binary && dpkg-buildpackage -S) === asac_ is now known as asac === dhaval is now known as dhaval_away === ogasawara__ is now known as ogasawara [20:21] rtg, do you think it would be possible to get the one digit fix from https://bugs.launchpad.net/ubuntu/+source/linux/+bug/115284 into one of the SRUs ? [20:21] it would really make a lot dvb users happy [20:22] ogra: lemme read through it. It won't go in until the next upload, but ought to make it for 8.04.1 [20:25] that would be cool :) [21:24] <_sourcemaker> how can i disable martian source and ll header checks? === smb_tp is now known as smb_away === johanbr_ is now known as johanbr [23:33] anyone awkae [23:35] wher'es tj