[01:25] hey guys i keep getting invalid module format and ive recompiled the kernel...turned loadable module support on...downloaded a new kernel and tried the same thing but nothing works [01:25] the module is in .ko format [01:30] zeroprog: is it for the same arch? [01:36] You'll get "invalid module format" if it's built for a different kernel version or configuration [01:36] Check dmesg for the actual error [01:42] hello: disagrees about version of symbol struct_module [01:43] and my kernel v is 2.6.28-11 and the kernel source is 2.6.28 === bjf is now known as bjf_afk [04:32] C [07:10] alright on kernel.org they dont have 2.6.28.11 so what linux source can i use [08:08] Keybuk, i had a conversation with the vfs union mount peeps last night and they sent me a rebase to 2.6.30 with a couple of race fixes, so i've make you a new kernel (again): https://edge.launchpad.net/~apw/+archive/green [09:20] cking, you had a machine (a dell i think) which had a problem with halt and reboot which only exhibited itself when splash was enabled but i think turned out to be like reboot=b was required and we added a quirk, can you remember? [09:21] apw, reboot=b forced it to work OK. Not sure about the quirk being added for the Dell Inspiron 6400 though. I cannot recall that (too much water under the bridge, or old age :-) [09:22] your memory matches mine [09:33] 请教大家个问题，编译内核之前要怎么配置才能在编译时把不需要的东西排除 [09:34] 怎么没人说话阿？ === cjwatson_ is now known as cjwatson [09:37] haitao0826: The language of this IRC channel is English. I'm afraid most people here don't read Japanese. [09:38] ok,sorry [09:38] bey [09:38] amitk, Its chinese [09:38] google says it says "Question to ask you, how to compile the kernel before the compile-time configuration in order to rule out things that do not need no one how to speak Albanian?" [09:41] <_ruben> all i saw was a bunch of ?'s ;) [09:43] _ruben, Iterestingly we saw the chinese letters, but the translation is still a bit weird :) [09:44] <_ruben> true ;) [09:49] smb: oh well. I thought 'tao' was more Japanese. I need to figure out how to tell Chinese, Korean and Japanese apart [09:50] eh, just someone speak in chinese? [09:52] apw, you are close [09:52] cooloney, google was close :) [09:54] apw: its quite compact (chinese) to encode all that information in such few 'characters' [09:55] yeah its very compact thats for sure [09:58] apw hi, would you have some time to investigate a swapon freeze during ubuntu install? [09:58] this happens when a swpa file is created and used within a fat fs [09:58] that sounds like a bad idea :) [09:58] (putting swap in fat) [09:59] do you get any panic output? [09:59] the alternative is swap at all [09:59] davmor2 has set up a machine with ssh access [09:59] one sec [10:07] xivulon, ok, so this swap file is it directly on the fat, or inside your ext4 loopback into a file crack ? [10:09] xivulon, is there a bug filed for this one? so i have somewhere to record stuff? [10:11] directly on fat [10:11] I guess 376120 is the same issue [10:11] the file is in /host/ubuntu/disks/swap.disk [10:11] where /host is the fat partition [10:12] davmor2 should provide you with instructions to ssh in [10:13] thaks [10:13] thanks [10:14] there should be a swapon and a couple of mkswap processes (from tests I assume) that cannot be killed [10:23] xivulon, yeah i think i know waht this might be [10:23] are we sure this bug# 376120 is the same bug or should i make a new one [10:24] and get that one dupped against it should it be the same cause, its not clear from the bug [10:26] apw I think it is the same bug, but feel free to open a new one with a more explicit title [10:27] thats ok, if you are pretty sure then i am happy [10:28] xivulon, if i get you a fixed kernel are you able to test it ? [10:32] apw: check with xivulon, but as far as I am aware wubi on windows side creates the drives and imports the cd image and then runs an automated install against that image on first run. [10:33] if it's a case of following instructions in this current environment I can test it. But only use little words :D I'm going to be away for a couple of hours so it would need to be after that, if that is okay [10:34] davmor2, i suspect it is complex as we need to replace the kernel somewhere in here, i'll let xivulon talk to that [10:36] apw: np's [10:42] xivulon, ? [10:50] hi apw [10:51] xivulon, hi ... i am building you some test kernels with what i believe is the fix for the issue applied [10:51] will let you know when they are built and pushed, and will update the bug also [10:51] apw thanks a lot [10:51] i am assuming you will be able to wave your hands over them and see if they fix things [10:52] hmm deployment would be complex though, as it will have no effect unless a new ISO is shipped [10:52] although users could replace the kernel inside of windows before rebooting [10:52] yeah indeed. but i am hoping you can at least frig it into a test install [10:52] to at least confirm this is the right fix [10:53] then we can make up some way to handle that [10:53] no module/initrd are involved correct? [10:54] hmm actually they are I guess [10:54] its actually the fat module which is updated, so yes [10:54] well vmlinuz/initrd is still ok [10:55] you build initrds on the machine normally [10:55] users can simply drop them into a directory, easy enough [10:56] as in the installation of an updated kernel image builds an initrd them [10:56] then [10:56] the initrd must be a live CD one (with casper and lupin-casper) [10:56] hrm, no idea how those get generated [10:57] once you have the patch ready I'll see to have one built [10:58] i'll let this test build run and then push up the patch and images, and you can use whichever works best for you :) [11:02] apw thanks a lot [11:02] np [11:04] apw: FYI I have cdimage/livecd-rootfs changes now to support using lzma compression for the live CD initramfs, once the next kernel lands; it saves 3MB so seems well worth it [11:05] cjwatson, nice. the size constraint in that context more than justifies any performance risk [11:05] xivulon: could you make wubi try each of /casper/initrd.gz, /casper/initrd.bz2, and /casper/initrd.lz? I see that it currently hardcodes initrd.gz [11:06] oh, and I'd better fix ubiquity too [11:06] * apw notes cjwatson is making much work for himself [11:06] oh, no, I don't need to fix ubiquity, never mind that [11:07] I did think about making it just initrd.img but I sort of like having the extension there [11:07] if it turns out to be a real hassle somewhere I'll reverse that choice [11:08] yeah ... its kinda nice to know what it is, just by the name as shell can do that check nice and easy [11:09] cjwatson, you might have some clever ideas on this, we have found what appears to be a kernel bug which affects the wubi installer and hangs it. the fix is likely a replacement kernel but it needs to be in a cdimage to be useful to wubi as i understand it [11:10] I don't have direct access to the buildds that produce the live filesystems [11:10] you could build updated initrd/filesystem combinations using the livecd-rootfs package though [11:10] xivulon: ^- [11:12] cjwatson, sure that is easy enough [11:13] but is that required for this jaunty update, or is it for karmic? [11:14] the lzma initrd would presumably be karmic onwards [11:14] in karmic I will try not to have an external boot dir at all and use grub2 loop module if possible [11:15] sounds sexy [11:15] will still need to edit the menu.lst (or whatever is called these days) [11:17] xivulon, things in /etc/grub.d/* which then make /boot/grub/grub.cfg [11:19] xivulon: what apw said [11:20] * apw muses that grub2 seems to have some kind of splash/background support [11:23] yep, will need to edit grub.cfg directly though, but should be easy [11:24] I will have to find partitions UUID from windows though [11:24] apw: weird, I can't even chroot into the union mount [11:24] bash just segfaults [11:24] in fact, now it just hangs [11:25] Keybuk, poop, so thats worse [11:25] which one you got? 7? [11:25] -8 [11:25] vfsunion6 [11:26] in fact, this looks like a very old one ;) [11:26] linux (2.6.30-8.10~vfsunion6) karmic; urgency=low [11:26] [ Andy Whitcroft ] [11:26] * forward port of the VFS union-mount patch set. [11:26] -- Andy Whitcroft < apw@canonical.com> Fri, 05 Jun 2009 11:42:53 +0100 [11:27] are you sure you uploaded a new one? [11:27] hrm [11:27] yeah ... one sec [11:28] there's a vfsunion7 in your "staging" PPA ;) [11:28] Keybuk, yeah forgot to copy it over, spanner [11:28] its copying now, take that one [11:28] ok [11:28] that was an update i got last night from val [11:28] and it is meant to fix a locking issue [11:29] so it might even fix what you are seeing [11:29] right, that sounds just like what I'm seeing ;) [11:29] they are being pretty responsive so far, keen to get some testing i suspect [11:30] i have tooo many ppas and no delete button, grrr [11:30] who would be a good person to pester about getting one of the kernel/ubuntu/misc/ modules updated for bug #387756 ? :) [11:30] Malone bug 387756 in linux "[karmic] please update tp_smapi" [Undecided,New] https://launchpad.net/bugs/387756 [11:31] isn't that tp_smapi thing the driver of questionable provenance? [11:32] apw: afaik it's been refused a merge into the vanilla kernel because the author insists on remaining anonymous [11:33] right, the definition of questionable provenance [11:33] apw: but we've been shipping it for ages, so pragmatic precedent wins the day and we can update it ;) [11:33] so lawyers may decide we remove it too :/ [11:33] lawyers schmawyers [11:33] easy [sic] for you to say [11:34] there is a blueprint for reviewing all those drivers [11:35] dammit [11:35] ? [11:35] oh, do you mean reviewing as in "with a view to removing anything we can't square away legally"? or "so we can get them all spruced up and fresh"? [11:36] a combination of the two, a review and decide what to do with each, with the result being either drop or update [11:36] would dropping mean entirely removing all trace from the archive, or demoting to a DKMS/modass package? [11:37] removing from a kernel point of view would be 'washing hands of' [11:37] i have no idea which way any particular item is going, or more specifically that item is marked RESEARCH, so its not been decided as yet [11:38] added links to both bug and spec (to the other) [11:39] ta [11:40] if I were a petulant person, I'd suggest that an anonymous driver is a drop in the ocean of legal questionability that pours forth from multiverse, but I'm not, so I'll just wait quietly and hope that I don't lose the excellent tp_smapi driver ;) [11:40] it is a worry that they won't tell people why they won't tell people who they are [11:41] probabally means they do not have the right to release the code, as it belongs to their employer or something [11:41] yeah I've always assumed it means they or someone they know works for ibm/lenovo and obtained the specs without permission [11:41] which is clearly not cool [11:42] and balanced against that obvious legal situation is the longevity of my battery ;) [11:42] anyway, thanks for the linkage :) [11:42] http://lkml.org/lkml/2008/11/17/355 [11:43] yep, if they had simply made up a believeable name and submitted it they would have gotten by [11:43] by showing they are anon and making it clear there is a _reason_ they cannot tell us [11:43] they taint the code [11:43] by implication [11:44] and the law sadly expects you to be reasonable, and not ignore hints like that [11:44] well they've said they are willing to show trustworthy people [11:44] which is why I linked that particular mail [11:44] no they are willing to prove they are a person [11:44] not explain why they cannot tell everyone who they are [11:45] which is the rub, its likely they have done something questionable [11:45] "willing to disclose is identity to [11:45] " [11:45] Assuming thats a typo of 'his' [11:45] that is just their identity [11:45] their need to hide their identity implies something about them or their working practice [11:45] or where they live [11:45] and one has to assume the worst, or open self to pain [11:46] well, thats clearly false. [11:46] right, there is a possibility its cause they live in china or north korea or something [11:46] everything is patented, so why do we write code at all? [11:46] the law tends to be kinder if you had no way to know in advance [11:46] this situation creates a reasonable expectation we s [11:46] should be suspicious, that is the issue [11:47] if they had called themselves something sane and we'd never known [11:47] life would have been fine [11:47] Could it not be that they work for am employeer that is competition to the Linux kernel, and no exclusive code clause in contract - but would appear bad if they were known. [11:47] I'd be much more suspicious of someone trying to pretend they didn't want to be anonymous [11:47] Daviey: ecactly. [11:48] its pretty unusual for an employer to not have the rights to everything one produces [11:48] its a standard clause in all contracts i have seen and worked under [11:48] There are many reasons that being known as the author of that code *without it being tainted* could be problematic. [11:48] yes, but the point is that the law is full of 'could reasonably expect to have assumed' type stuff in it [11:49] apw: works in both directions. [11:49] you don't get the benefit of the doubt most of the time [11:49] apw: Well i've avoided such contracts, but if that is the case - a HUGE amount of code in the Linux stack is therefore non-free, as the authors may not have had permission to release under a free licence [11:49] indeed so, and lawyers are notoriously 'assuming the worst' kind of people [11:49] Daviey, not necessarily i am employed, the code belongs to my work, i am specifically allowed to contribute it [11:50] lawyers are often extremely risk averse. You have to take that into consideration. [11:50] yes.. because you have your employeers permisson. I would imagine that a great deal of patches are written by pure hobbyists, away from dayjob [11:50] I wonder what our relationship with lenovo is like [11:51] they don't have to do anything further than agree not to sue anyone because of us distributing a driver that makes their hardware look better [11:51] in a previous job i had the right to do 'non-compete' things outside work in my own name [11:52] anyhow, its not my decision to accept it into or not into mainline, they made their reasons felt, and i can understand them [11:53] seemed rather inconclusive in the things I found, given Linus being willing to accept it :P [11:53] and if he does then we get it for free and all is well [12:09] and http://lkml.org/lkml/2008/11/18/272 says "SuSE asked, Lenovo said no" [12:09] (in the same thread lifeless linked to) [12:19] one does have to wonder exactly what suse asked though [12:20] apw: still a segfault with vfsunion7 [12:20] is weird [12:20] it's not even a segfault, the kernel seems to kill bash [12:20] hrm, well that might make sense [12:20] did you get any kernel stack traces on it [12:21] if not, what does strace say happened [12:21] strace just hangs [12:21] in the execve() [12:22] hrm [12:22] setuid perhaps? [12:22] just going to fire up upstart and see why it thinks bash exits ;) [12:23] heh nice [12:24] did you say there was nothing in dmesg at the time, most hard kills are reported [12:25] didn't look tbh [12:26] worth checking then after it happens [12:27] got one [12:28] BUG: unable to handle kernel NULL pointer deference at (null) [12:28] pastebin? [12:28] ouch [12:28] i'll have a look see if i can see what casues it if you paste the whole thing [12:28] somewhere [12:28] sure [12:28] one moment [12:28] it's inside inode permission things ;) [12:29] yeeks :( [12:31] * apw waits impatiently for the trace [12:33] Keybuk, poke [12:33] rebooting [12:33] ouch [12:33] got stuck in kernel and wouldn't shut down :p [12:33] it has that tendancy :/ [12:36] apw: http://pastebin.ubuntu.com/197021/ [12:37] eip == 0 ouch [12:47] useful stack trace? [12:55] Keybuk, i think its valid, an trying figure it out... what was your union of? [13:02] Keybuk, ?? === mdomsch is now known as mdomsch_bos === cjwatson_ is now known as cjwatson [13:18] Keybuk, think its worth writing up what you did and sending that with the panic to val and jblunk, cc: me [14:00] apw: back [14:00] had to run David to physio [14:00] it was a union of the filesystem.squashfs from today's daily-live [14:00] and a tmpfs [14:00] I ran chroot /mnt /bin/bash === sconklin-gone is now known as sconklin [15:23] smb, heads up i just pushed a patch to jaunty-lrm, looks like you are working in there too [15:24] apw, Thanks, though I must refresh my memory what I might do there [15:25] uh oh [15:27] apw, any chance you might kill that again for a sec? What version did you base on? [15:28] smb, ? i fetched origin and put it on the top [15:28] apw, Hm, yeah. It just looks to me I forgot to push a release [15:28] smb, then force it and i'll re do it [15:29] ok. thanks [15:29] done [15:30] aww, nobody's picked up my LKML patch :'( [15:30] Keybuk: you should have sent it to a more specific audience [15:30] rtg: there isn't one that I can tell [15:31] nothing in MAINTAINERS for cn_proc or connectors in general [15:31] Keybuk: I'm can;t remember who was involved in the last round of syscall discussions, but they'd bea good starting point [15:32] it isn't syscall though, no? [15:32] Keybuk: not really, but its up in the process control layer. Perhaps Ingo? [15:33] Keybuk, if all else fails always send it to akpm he knows everyone [15:33] true [15:33] will give it a few days ;) [15:34] Ingo, Andrew, Oleg, etc. tend to pick up things from LKML anyway [15:34] Keybuk: he might have missed it in the merge window storm [15:34] apw, Just for info, the patch for jaunty lrm is for whch bug? [15:35] BugLink: http://bugs.launchpad.net/bugs/305587 [15:35] Malone bug 305587 in linux-restricted-modules "[Jaunty] warning: missing LSB information " [Medium,Fix committed] [15:55] apw, jjohansen: is there a reason LGUEST_GUEST isn't enabled for i386? [15:56] in karmic, that is [15:56] * apw looks [15:59] LGUEST_GUEST implies we are building a paravirtualised kernel [15:59] so i'd not expect it to be on for a primary i386 kernel i don't think [15:59] rtg, ^^ [15:59] apw: right, it allows the 32 bit lguest kernel to boot under an lguest hyper visor. why wouldn't we enable it? [16:00] well it would depend on whether it is independant of being a real kernel at the same time [16:00] i would expect it enabled in a special kernel like the server virtual flavour [16:01] apw: all of the paravirt-op stuff is dynamic anyway. [16:01] i'll see if it is enableable in parallel [16:01] rtg then it should be fine. note it is only non-pae so we'd only get it in the new i386_nonpae flavour [16:01] LGUEST and PAE don't like each other. [16:01] I forget why. [16:02] apw: I think I remember someone at UDS requested it [16:02] Oh, apw already said that. [16:02] Never mind me. [16:02] shall i put it on the config review and spin a patch for it? [16:03] apw, soren: I'll get back to lguest in a bit. I'm currently making sure PAE works. [16:03] ack [16:03] apw: wait until I have the legacy flavour implemented [16:03] i was about to respin that config merge patch [16:03] if you are mid flavour mangle then it should wait [16:04] rtg ^^ [16:04] apw: it should be independent of the flavours, right? [16:04] yeah the code change is, but there is an associated fdr updateconfigs which isn't [16:05] apw: we could do them independently since the build process would pick up the config changes anyways [16:05] i.e., just the script patch with out the config file updates [16:05] yeah we can do that for sure [16:06] I can collapse them later [16:06] i'll send out the patch with a sample patch for the config change so peple can see what it does [16:06] but with a view to merging only the code change [16:06] right [16:06] ok plan [17:04] hi.. i am building kernel.. [17:04] how to make deb file out of it? [17:09] hi.. any body pls help.. [17:13] mohan_: start here: https://wiki.ubuntu.com/KernelTeam/KernelMaintenance [17:15] rtg: ok.. thanx.. [17:18] rtg: fully confused .. :( [17:18] i am right now done with make modules.. [17:19] mohan_: then you are doing the wrong thing. see 'https://wiki.ubuntu.com/KernelTeam/KernelMaintenance#Normal full build' [17:19] mohan_: you must use debian commands to build the kernel [17:19] the best way is to install devscripts and run 'debuild -b' [17:19] hmm,.. but before it worked.. [17:19] i installed RT kernel before in this way.. [17:20] mohan_: you may have installed it, but you didn't build it that way [17:20] oh.. ok.. [17:20] i didn't build a deb package though.. [17:20] but now i want.. [17:20] but doing this : fakeroot debian/rules binary-generic [17:21] mohan_: then do as I've indicated. plod through the steps [17:21] gave error: /usr/bin/fakeroot: line 176: debian/rules: No such file or directory [17:21] ok sir.. [17:21] mohan_: sudo apt-get install build-essential fakeroot devscripts [17:22] ok.. sir.. its installing.. [17:22] takes one min.. [17:23] i have done passing menu config and make and make modules command.. [17:23] opened package configuration.. [17:23] mohan_: then you've made a mess. do 'git checkout -f master;git clean -f -d' [17:23] thern 'debuild -b' [17:23] ok.. [17:24] now what should i set in this package configuration? [17:24] no mail? [17:24] huh? [17:24] postfix configuration dialogue box is asking [17:25] use the default [17:25] ok.. [17:25] what does git checkout do? [17:26] mohan_: uh, you must not be using the Ubuntu git repo? [17:26] no.. [17:26] i downloaded kernel manually from kernel.org and patched RT kernel to it.. [17:27] mohan_: then go back to https://wiki.ubuntu.com/KernelTeam/KernelMaintenance and read [17:55] The kernel team weekly meeting will start in #ubuntu-meeting in 5 min === bjf_afk is now known as bjf === lieb_ is now known as lieb === mcasadevall is now known as NCommander [21:21] hey guys how do u check MTU size in ubuntu [21:28] ip(8), ifconfig(8) [21:28] (same as in other Linux distributions) === Twigathy is now known as Twigaboob === Twigaboob is now known as Twigathy === stew is now known as HolieThePony [22:53] Does anyone know if the lack of provision of a linux-image-debug package is deliberate? [22:54] Or is it just an accident of someone thinking that it's redundant now we have ddebs..... except ddebs are effectively only available for the current development release of Ubuntu [22:59] maxb: rtg answered that some time last week [23:01] maxb: briefly, the omission is intentional, as they can be built on the servers but result in gigantic binary packages [23:10] It would be nice to have an official comment in LP 289087 [23:10] ubot3: ? [23:10] ubot3: bug 289087 [23:10] hrm [23:10] Malone bug 289087 in linux "Missing linux-image-debug packages and metapackages since Intrepid" [Undecided,Confirmed] https://launchpad.net/bugs/289087 [23:10] Factoid bug 289087 not found [23:12] maxb: feel free. https://lists.ubuntu.com/archives/kernel-team/2009-June/005931.html [23:20] "How to build them" is not the same question as "Why are they no longer downloadable" [23:22] maxb: a deliberate attempt to avoid bloating the archive has been raised. i actually asked this question some time ago at the most recent UDS Mtn View. [23:23] maxb: if you feel it is necessary to have the rationale written precisely in response to why the large debug packages are unavailable from cc.archive.ubuntu.com, then by all means, go for it. [23:32] Well, I feel it necessary that the users legitimately noting that a feature present in Hardy is no longer available have some sort of explanation why that is the case. === HolieThePony is now known as stew