[00:42] pgraner: pong, around until this horrible 'net connection dies (again) [03:23] dtchen: no worries, fixed the issue (sound as usual....) [05:22] 8/c [07:31] c [08:46] * hyperair scratches his head and wonders why his usb thumbdrive isn't detected [09:57] apw, did you and robbiew talk about his intel graphics issue with .31? [09:58] apw, I've upgraded but haven't booted into .31 yet, wondering if it's wise [10:08] hi all [11:20] hello all, I have ubuntu kernel source v2-6.27.7 , and I tried to build kernel like kernels frem kernel.org, (make xconfig, make) , but I got an compilation error ubuntu/misc/wireless/p80211/p80211netdev.c:968: error: ‘netdevice_t’ has no member named ‘wireless_handlers’ [11:21] sorry for a long line - is it correct to buill ubuntu kernel the same way as vanilla kernels? [11:22] rather, you should be using make-kpkg [11:23] but I do not need package, I need ti for other system where I can copy only vmlinux, initrd and /lib/modules [11:27] then i don't know =\ [11:28] rather, i think that if you're going to do something like that, you might as well compile from one of the kernels in kernel.org [11:28] the linux-source-* packages have ubuntu-specific changes included [11:31] Let us return to the compilation error - I do not have wireless defined in .config, why it got active ? [11:35] Correction - wireless was defined as ubuntu includes [11:41] so disable it [11:41] * hyperair stares at kernel compilation incredulously -- it's taking 3GB D= [11:47] hyperair, We have some help at https://wiki.ubuntu.com/KernelTeam/KernelMaintenance [11:47] hmm? [11:47] But in general, you do not do it as with a vanilla kernel but like debian packages [11:47] * hyperair clicks [11:48] fakeroot debian/rules binary-generic [11:48] ah [11:48] for a generic kernel build [11:48] That gives packages, but you also can find the objects in debian/build/build-generic [11:48] how do you prepare the debian/ directory? [11:49] ugh i'm going to run out of disk space at this rate X_X [11:49] Hm, did you not say you got the source? via apt-get source linux-image... [11:50] not me =\ [11:50] But I would have thoght [11:50] i was make-kpkg'ing a git kernel [11:50] any way would include the debian dir [11:50] using config-2.6.31-1-generic [11:50] except you did the evil mrproper [11:50] evil mrproper? [11:51] make mrproper just dumps the debian dir into nothingness [11:51] i see [11:52] just one of the pitfalls when coming from vanilla builds :) [11:52] by the way, does make-kpkg typically result in a kernel image nearly 1G in size? [11:52] the vmlinux is big [11:52] sounds like your kernel wasn't stripped [11:52] grr X_X [11:52] wait, stripped? [11:53] hmm it hasn't been stripped yet [11:53] i ^Z'd it [11:53] had to purge my apt cache to make space for it [11:53] and my pbuilder cache [11:53] vmlinux in the build dir is always big. It won't get packaged [11:53] But it is part of the build [11:54] i see [11:54] Doing kernel builds == buy large disk drive from my experinece :) [11:54] ouch X_x [11:54] it's a small partition [11:54] i should have done it in my home directory, but i did it in /usr/src instead because my home directory has even less free space [11:55] Karmic amd64 full build (with ccache) == 12G [11:55] woohoo X_X [11:55] thank god i'm not doing a full build [11:55] i'd trim down the config, but the last round ended up killing my usb drive support [11:55] O_o [11:55] for some reason or other [11:55] it just wouldn't mount [11:56] as in mount hung [11:56] and for another stick, the device didnt' even appear [11:56] damn weird =\ [11:56] dmesg showed it though [11:57] Once tried to unbind the ehci driver? [11:57] unbind? [11:59] echo the usb id into /module/uhci_hcd/drivers/pci:uhci_hcd/unbind ... or so [11:59] err actually, which kernel are we talking about? [12:01] .31 [12:01] from git rather [12:01] few days ago [12:01] Ah, ok. Not some older Jaunty which might have the problem with detection then [12:02] I thought I saw some older version number above [12:02] 2.6.27 [12:02] that wasn't me =p [12:02] that was levonshe [12:03] Ah, :-P all green and blurry [12:03] Well, same advice in general for him [12:04] irssi? =p [12:04] get nickcolor [12:06] xchat, which does some but I do not want to paint everyone differently. too much effort :P [12:06] heheh right [12:06] hmmmm [12:07] 1.2G /usr/src/debian/linux/linux-2.6/debian/linux-image-2.6.31-rc1-hyper3/ [12:07] something is seriously wrong about that ._. [12:07] it's not about an unstrippped vmlinux now is it? [12:10] after the build my build-generic is 4.9G alone [12:12] ._. [12:12] okay, but how did i end up wtih a linux-image deb 352M big? [12:14] That is slightly odd. The normal config ends up being around 30M [12:14] ._. [12:14] see [12:14] that's what i meant T_T [12:14] it's huge =( [12:14] and that is amd64, guess i384 might be a bit less [12:14] this is amd64. [12:16] the /lib/modules/`uname -r`/kernel/drivers directory is the largest [12:16] 773M [12:16] and you used "debian/rules editconfigs" to adapt configurations and the "fakeroot debian/rules binary-generic" to compile [12:16] ? [12:16] er no [12:16] i used make-kpkg clean [12:16] followed by make-kpkg kernel_image kernel_headers =\ [12:17] this is a git kernel [12:17] Would work with a git kernel, too [12:17] =\ [12:17] make that the upstream git tree [12:17] I am not sure about the make-kpkg. I know it was used once, but I never used it for anything after Hardy [12:18] including hardy [12:18] Ah, ok. the vanilla git tree.. [12:20] hyperair, maybe you would be interested in that http://kernel.ubuntu.com/~kernel-ppa/mainline/daily/2009-07-02/ ? [12:20] ooh i never knew that existed [12:21] i'd have to recompile it due to the whole acpi-cpufreq issue though [12:21] There are packages for all rc releases. Unfortunately broken for i386 currently. Hm, thought that was fixed... [12:21] http://kernel.ubuntu.com/~kernel-ppa/mainline/v2.6.31-rc1-fix1/ [12:21] maybe that is better [12:22] the rc release is broken for me as well [12:22] on amd64. [12:22] it panics [12:22] or otherwise blanks out my screen so i can't see the panic [12:22] oopses [12:22] Ah, no. Compile broken in that case [12:22] ? [12:22] The dailys do not seem to produce i386 packages [12:23] There was a incorrect div which cause a build failure on a hardy chroot (which is where the builds are done) [12:24] But that probably does not help you. Can you investigate which of the modules is so big? [12:28] the network modules [12:34] Any chance you compiled with or have all objects unstripped? With debugging and symbol information? [12:35] 2.8M cfg80211.ko [12:35] 88K cfg80211.ko.s [12:36] That is in the build directory (so not prepared for install) and the .s is after strip [12:37] you can do a "file " to see whether it is stripped, too [12:56] smb: i don't appear to have any .s files [12:56] smb: and none of the modules are stripped [12:56] hmm [12:56] hyperair, the .s was me [12:56] as a save file [12:56] oh [12:56] aha i see [12:59] hmmm [12:59] pkg-create-dbgsym at work perhaps? [13:03] I can't say. As I said, I actively do not use make-kpkg, so I can't say what things are at work. With our build the modules are unstripped before and will get striped on preparation for the packaging [13:04] make modules_install I expect does the same [13:06] INSTALL_MOD_STRIP controls that [13:07] i see [13:07] wiki.kubuntu.org says CONFIG_DEBUG_INFO=n works around the issue [13:10] Yeah, right. Without adding them those won't have to be stripped [13:14] And the default of y is useful to only need one go to get modules and vmlinux with debugging info, which can be stripped later. Some things tend to get forgotten when you just use the usual build command. :/ [13:17] heh i see =\ [13:17] ccache doesn't seem to like removing the debugging option [13:19] sounds a bit weird. that should care the least... [13:20] no i meant ccache is missing the cache completely =\ [13:21] Oh that. :) Well naturally [13:24] =( [13:26] Greetings everybody! [13:26] I've problems getting a ath9k-based PCMCIA wifi-nic working with kernel 2.6.30-10 under Karmic [13:27] I can't use 2.6.31 as that causes by i965-based laptop not to boot at all... probably due to currently broken KMS [13:28] rc1 is broken for i965. [13:28] compiling from git helps [13:28] I would believe apw rather sooner than later syncs up with it... [13:29] ? [13:29] MacSlow, Is there anywhere more info about the problem? [13:29] hyperair, sync from upstream git into the karmic tree [13:30] smb, well I can provide some info (dmesg, syslog... what ever you need.) [13:30] ah. [13:31] smb, the ath9k modules is loaded and shows up in dmesg without any hints to errors [13:31] smb, and the device is listed as wlan1 [13:31] Ok [13:32] smb, but iwlist scan only says "No scan results" [13:32] you read minds... [13:32] smb, I get a scan result only shortly after I plugged the ath9k PCMCIA-card in [13:33] smb, also nm-applet/NetworkManager lists the device, but has it's entries in the nm-applet's menu greyed out [13:33] Ok, was there a karmic kernel working? [13:33] I've no clue what that actually means [13:33] smb, briefly about 3 weeks ago when I asked asac for help [13:34] MacSlow, and just to rule out absolutely everythign. Any killswitch to twiddle? [13:34] smb, yes... either booting with it enabled or disabled does not change a thing [13:35] smb, btw... just wondering... does the kill-switch apply to all wifi-devices or just built-in ones? [13:35] It should only to the built in... [13:36] smb, since the built-in card is likely to be broken (that's why I use the ath9k-based PCMCIA one) I guess I can ignore the killswitch then [13:37] Oh, ok. Then I understood the "I get a scan result only shortly after I plugged the ath9k PCMCIA-card in" before wrongly [13:37] I assumed th internal one is broken but the external works... [13:37] But I should read the beginning :/ [13:39] So you get scan results for a short period of time and after that it stops... [13:39] hm [13:41] smb, yes [13:41] smb, when I pull the card out and back in [13:41] I can get scan results from "iwlist scan" for about 5 seconds [13:42] smb, by repeatedly issuing "iwlist scan" [13:44] smb, one thing I recognize in the output of "iwlist scan" is that the time for "Extra: Last beacon: XXXXms ago" [13:45] sound quite strange. especially without any hint of messages... [13:45] MacSlow, It gets bigger? [13:46] smb, continuesly grows until about 9000 ms [13:46] then it stops [13:46] yes [13:46] if that's any hint [13:47] A bit like something went down and it does not receive anything, then goes into sort of sleep. And the results received at the beginning are at some point invalidated [13:48] smb, any idea how to fix it [13:48] ? [13:48] or what I could try to make it "stay up" [13:48] Heh, no. First I would need an idea what is going on [13:49] smb, could you guess if it's a driver-issue of a NetworkManager issue? [13:50] Just checked whether there where recent changes in the module. Nothing from -8.9 to -10.12. Now how long is that in real time... [13:50] is somebody looking at pulling in the commit referenced in bug 392709? [13:50] Malone bug 392709 in linux "[karmic] linux-image 2.6.31-1.13 crashes on boot" [Undecided,Confirmed] https://launchpad.net/bugs/392709 [13:50] hello all. can you comment why i get via_chrome9: disagrees about version of symbol struct_module (I have kernel build fot Ubuntu 2-6.27.7, this driver is also 2.6.27-7) ?? [13:51] cjwatson, I look in a second [13:51] levonshe, Maybe external build pulling in the wrong headers... [13:53] MacSlow, 2.6.30-8.9 was done Jun-12 [13:53] MacSlow, So between then and -10.12 there were at least no changes to the module itself [14:00] cjwatson, I remember apw knowing of it and he was looking at some other things freshly commited this morning [14:00] ok, thanks [14:00] I put it on the release meeting agenda [14:04] hmm.. was .31 known to be booting? [14:07] boots but takes an eternity here [14:08] its oopsing all over the place [14:11] MacSlow, maybe you could go into the /sys/class/net/ directory (? name of the interface) and do a grep . * there [14:17] tjaalton: oopses on acpi_get_pci_dev for me in the find_video routine [14:18] amitk, That is the patch that was asked for by Colin [14:19] amitk: yep, same here [14:19] Fixed upstream [14:19] 412af978 ("ACPI: video: prevent NULL deref in acpi_get_pci_dev()" [14:19] nice [14:21] * hyperair wonders if anybody would be nice enough to help me get usb-storage working in 2.6.31 =\ [14:22] smb, I got full linux_source.tar, so the headers are consistent with the kernel, the module In question was taken from different place (VIA AREA), but it was said it prepared for Intrepid [14:24] levonshe, There would be several releases of Intrepid, the kernel version you had 2.6.27-7 is quite old the current one is 2.6.27-14 and the module might be prepared against that [14:25] smb, thanks. I will try patch to 2.6.27.14 [14:27] smb, sorry I was wrong, the module has vermagic=2.6.27-7-generic SMP mod_unload modversions 586 [14:29] levonshe, Do you compile parts of the module or do you only get the .ko? [14:29] You need to have the correct target kernel running and the matching headers installed for compiling [14:30] what does /proc/version_signature say? [14:36] smb, no I have development machine with different ubuntu versioin and kernel version, Unortunatel VIA does not provide source for module, so I have only binary .ko [14:40] levonshe, If the module is pre-compiled already and says 2.6.27-7-generic then it should in theory work if the machine you want to load the module is running the 2.6.27-7-generic kernel. If it is another Intrepid version you can try insmod -f, but without any guarantees [15:06] levonshe, No, it is 586 iirc, it means config options rather tuned for desktop, as opposed to -server === BenSee is now known as BenC [16:20] is anyone here using a .31 kernel who has a usb-storage device working? as in mounted and all? [16:39] hey guys, http://www.gossamer-threads.com/lists/engine?do=post_view_printable;post=1097001;list=linux affects 2.6.31-1 in karmic [16:39] tun interfaces dont work [16:42] its already merged in linus's tree, so does that mean just wait for the next build? [16:48] maco, basically. though it might be the one after the next [16:49] because the next is just prepared for the acpi_video null pointer [16:49] problem [16:51] ok [17:40] yo yo kernel nutjobs :p [17:40] seeing an issue with the 31 kernel [17:40] udevadm settle - timeout of 30 seconds reached,t he event queue contains: [17:41] and then it lists 2 devices under /syst/devices/pci0000:* [17:41] then I enter my encryption password [17:41] then I see device-mapper messages about "target device sda5 is misaligned" [17:41] I can only get this in recovery mode after adding "nomodeset" to the kernel line, otherwise it just black screens on me [17:43] you're using lvm? [17:43] (for the encryption i mean...as opposed to one of the other methods) [17:43] nixternal, ^ [17:44] not on the machine with issues [17:44] this machine, which is my ubuntu machine, is LVM and doesn't have those issues [17:46] given the "misaligned" error on the disk, i'd guess it's something to do with whatever encryption mechanism you're using [17:47] the one you use when installing from cd :) [17:48] i thought that was lvm *shrug* [18:44] has anyone seen something like this: http://pastebin.com/f535cca55 ? [18:49] Hello, I'm maintaining sl-modem package, I have a question about class_device_register, I understand that it's GPL: EXPORT_SYMBOL_GPL(class_device_register) [18:50] the code of sl-modem module has this license:MODULE_LICENSE("Dual BSD/GPL"); [18:52] will that be able to register a sysfs object to create the device automatically ? [19:47] hyperair: eek! [19:48] I tried to restore debconf and now something has gone wrong when trying to install the kernels? See http://pastebin.ca/1481697 [19:50] heh i got something similar to that once [19:50] i removed nvidia-common [19:51] hyperair: interesting! [19:51] =p [19:51] I'll do the same [19:51] assuming that it's only needed for nvidia :) [19:52] so.. sorry this is dumb but how do I go back and get those kernels reinstalled [19:52] now that I have removed nvidia-common [19:52] nixternal: which kernel specifically? [19:53] -13 [19:53] hyperair: grr http://pastebin.ca/1482194 [19:53] nixternal: -13? [19:54] unless i'm severely misparsing something, i have no idea what -13 refers to [19:54] * hyperair whistles [19:54] help! :) [19:54] * hyperair doesn't know anything about fglrx [19:55] is it really a fglrx problem? [19:55] I think it all came because I screwed debconf [19:55] by deleting one of its directories [19:55] ...you're on your own [19:55] =p [19:55] * hyperair is off to reboot [19:56] eek [19:58] anyone help?? [20:04] nixternal: if you're referring to karmic's 2.6.31-1-generic, your bug is already fixed (though it hasn't been merged into ubuntu-karmic.git) [20:05] nixternal: see drivers/md/dm-table.c:498, where the condition is incorrect [20:06] nixternal: changeset ea9df47cc92573b159ef3b4fda516c32cba9c4fd in linux-2.6.git if you're tracking it [20:47] dtchen: ya, I found that earlier when digging around. I saw where the guy fixed it upstream so I knew it was only a matter of time...it was found on the 25th and fixed the same day I think === mcasadevall is now known as NCommander [22:25] can anyone help with this ? http://pastebin.ca/1482194