UbuntuIRC / 2008 /10 /09 /#ubuntu-kernel.txt
niansa
Initial commit
4aa5fce
[01:24] <administrator_> hey all
[01:25] <administrator_> im thinking of grabbing 2.6.25 kernel for built in 8165 wlan support that i hear rumors of
[01:25] <administrator_> is there an official 2.6.25 (k)ubuntu kernel package?
[01:27] <crimsun> no. Do you mean 2.6.27?
[01:28] <administrator_> i didnt mean that
[01:28] <administrator_> crimsun: what if i mean that now
[01:28] <administrator_> ?
[01:30] <crimsun> (the only 2.6.25ish one is for -ports)
[01:41] <administrator_> crimsun: i didn't explain myself well. i dont care what version the knerel is, so long as it has a module for the 8165 wlan chipset
[01:41] <administrator_> but i found some ubuntu docs
[09:49] <mdz> cking: I had a git tree from earlier in Intrepid, and just did a 'git pull'
[09:49] <mdz> this gave me a bunch of conflicts, though I never modified by working tree. why is this?
[09:49] <cking> mdz: I suspect it's because it's been rebased.
[09:53] <cking> I see this occur when ever the latest rc patches are pulled in - if there are too many of these merge conflicts I just get the entire repo again from clean - which is probably not an efficient way of working buy my network connection is fast
[09:53] <cking> s/buy/but
[09:53] <cking> I am sure amit has the necessary runes to resolve this efficiently
[09:53] <amitk> mdz: during development, git pull will almost never work
[09:53] <amitk> mdz: because we rebase, as cking mentioned
[09:54] <mdz> amitk: how do you keep your tree up to date?
[09:54] <amitk> git fetch; git rebase origin
[09:54] <mdz> (I won't ask why we rebase...)
[09:54] <amitk> but wait
[09:54] <amitk> since your tree is probably in a mess right now...
[09:55] <amitk> mdz: do you have any local changes you care about? or do you just need the latest kernel tree?
[09:57] <amitk> mdz: 'git fetch; git reset --hard origin' will restore you to a pristine tree if you don't have (or care about) local changes.
[09:59] <mdz> amitk: thanks
[10:12] <munckfish> Hi I'm looking at bug 280451 raised by doko - is there any historic reason why the ports kernels builddep on gcc-4.1 specifically?
=== asac_ is now known as asac
=== thegodfather is now known as fabbione
=== thegodfather is now known as fabbione
[14:35] <munckfish> Hi I'm looking at bug 280451 raised by d o k o - is there any historic reason why the ports kernels builddep on gcc-4.1 specifically?
[14:36] <munckfish> I just have this recollection that I saw a check in <kernel src>/Makefile ages ago that enforced use of gcc 4.1 for a certain arch but I can't see it
[14:36] <munckfish> there now
[14:36] <rtg> munckfish: some of the arches used to have compiler dependencies. dunno if they are still true.
[14:37] <rtg> compiler version dependencies, that is.
[14:37] <munckfish> rtg: yep
[14:37] <rtg> if they no longer exist in Hardy, then they are probably not appropriate for the ports kernels.
[14:37] <munckfish> https://bugs.launchpad.net/ubuntu/+source/klibc/+bug/280451
[14:38] <rtg> munckfish: propose a patch and get smb_tp to test build
[14:38] <munckfish> rtg: yes will do
[14:42] <smb_tp> munckfish, best mail to the kernel team list
[14:43] <smb_tp> munckfish, if you are done and have something to pull
[14:43] <munckfish> smb_tp: yep usual procedure.
[15:39] <njpatel> is there some news on the e1000 for module regarding the eeprom-corruption?
[15:42] <rtg> njpatel: the EEPROM corruption is related to ICH{8,9,10} e1000e parts, and yes, it appears to have been resolved by marking mapped memory to be read-only.
=== BenC1 is now known as BenC
[15:44] <njpatel> rtg: anyway I can do that on my laptop right now? I am at the GNOME/GUI-hackfest without any network-access on my own laptop (just using a colleagues machine for the moment)
[15:47] <rtg> njpatel: you can do what on your laptop? update? 2.6.27-5 and later should be fine.
[15:48] <njpatel> rtg: oh... hm... I already updated to 2.6.27-6 (via usb-stick) and that still does not give me back a working eth0
[15:49] <njpatel> rtg: i am on a core 2 duo (using amd64 arch)
[15:49] <rtg> njpatel: what NIC? lspci -vvnn
[15:49] <njpatel> rtg: using a Ubuntu 7.10 live cd ethernet works so the hardware is ok
[15:51] <njpatel> 00:19.0 Ethernet controller [0200]: Intel Corporation 82566MM Gigabit Network Connection [8086:1049]
[15:52] <njpatel> Subsystem> Lenovo Device [17aa:20de]
[15:52] <rtg> njpatel: hang on a sec...
[15:54] <rtg> njpatel: ok, that is an ICH8 device. do you see any eth devices with 'ifconfig -a'
[15:55] <njpatel> rtg: I get lo and pan0
[15:55] <rtg> njpatel: look in dmesg, e.g., dmesg | grep -i e1000e
[15:56] <njpatel> oh one sec
[15:56] <rtg> njpatel: also, make sure its not still black listed in /etc/modprobe.d
[16:00] <njpatel> rtg: nothing shown from dmesg
[16:00] <rtg> njpatel: check for black listing?
[16:01] <njpatel> rtg: there is blacklist-e1000e in /etc/modprobe.d still
[16:01] <njpatel> just delete that? and try to restart networking?
[16:02] <rtg> njpatel: ok, then you're not completely up to date. 'sudo rm /etc/modprobe.d/ blacklist-e1000e; sudo update-initramfs -u'
[16:03] <njpatel> probably due to me just updating the package manually and not via apt-get or synaptic (since I have not network) so some scripts were skipped
[16:04] <njpatel> rtg: manually loading e1000e now should work I assume
[16:05] <rtg> njpatel: well, my first attempt at removing blacklist-e1000e in module-init-tools wasn't correct. Steve Langasek had to come along behind and clean up my mess.
[16:05] <rtg> njpatel: yep - you should be able to just modprobe e1000e (or reboot)
[16:06] <njpatel> hm.. looks good sofar... trying to plug in a cable an see NetworkManager hopefully do its magic
[16:07] <njpatel> rtg: sweet works again
[16:07] <njpatel> rtg: you don't also happen to know a bit about iwl3945
[16:07] <njpatel> :)
[16:08] <rtg> njpatel: a bit, though I'm not sure I'll of much help.
[16:08] <rtg> s/of/be of/
[16:14] <njpatel> rtg: ok huge thanks at least ethernet is working again
[18:36] <laga_> hum
[18:36] <laga_> aufs is broken *again*
[18:41] <amitk> laga_: in what way?
[18:41] <laga_> NFS branches
[18:41] <laga_> i'd like to blame kees cook commit (118463fa8236a123fdf6f81fb0c0176f7fd1caa7), but i can't tell for sure yet
[18:41] <laga_> i still wonder why i have to constantly explain funny changes to aufs to junjiro
[18:42] <laga_> kees cook re-introduced aufs_do_setattr which hasn't existed in a long time in aufs. although it looks like it's just a wrapper, so i'm not sure if it's to blame
[18:43] <laga_> amitk: http://launchpadlibrarian.net/18342118/DSC00183.JPG - this is what you get if you try to boot a mythbuntu-diskless client
[18:46] <laga_> i will set up a test system tomorrow and revert that commit. i'm glad someone noticed before the release :)
[19:02] <amitk> laga_: just talked to kees. If reverting that patch fixes the problem, could you possibly prepare a patch separating the vfs changes from the fsetattr changes?
[19:03] <laga_> yes, i think i could do that. i'll get back to you
[19:03] <ogra> do we have any way to support RT2700E wlan chipsets ?
[19:03] <ogra> apparently they are in the eee1000
[19:18] <mrxmike> ok
[19:19] <mrxmike> following -> Intrepid crashes on my Intel D945GCLF + Atom 230 (its a Intel bundled mobo+processor)
[19:19] <mrxmike> desktop / server (also tried alternative version)
[19:22] <ogra> you should mention that you try to run 64bit kernels on it, i doubt anyone will actually associate 64bit with atom (since thats the exact thing you ran into in the other channels ;) )
[19:22] <mrxmike> im currently installing the 32bit version of Intrepid beta server (32bit), the installation is nearly completed
[19:22] <mrxmike> ogra: i wasnt done typing yet :P
[19:22] <ogra> heh
[19:24] <mrxmike> So -> problem is, Intrepid (64) kernels do not work on the Atom 230 (probably not on the newer 330 either, i'll buy this as soon as possible)
[19:24] <mrxmike> but i cannot test that yet
[19:24] <mrxmike> funny detail, with acpi=off .. the kernel of the livecd's does work
[19:27] <mrxmike> What can i do, how can i help with adding support / fixing what is broken?
[19:28] <ogra> amitk, did you ever do any work with 64bit atoms yet ?
[19:29] <amitk> ogra: not yet... so what does 'crashed' mean here?
[19:31] <mrxmike> kernel segfaults, no CPU time for xxSeconds
[19:31] <mrxmike> errors
[19:33] <amitk> mrxmike: It will be hard to debug or look into without detailed info. dmesg, lspci -vvnn, /proc/version_signature attached to a bug report would help a lot.
[19:35] <amitk> but booting with acpi=off seems to suggest acpi bug in bios or kernel
[19:44] <mrxmike> amitk: i will provide all those details as good as i can, its easy ... as i can use the livecd for it...
[19:45] <mrxmike> but i just installed 32bit server, and i do want to use my server as well .. (so i keep that on)
[19:46] <amitk> mrxmike: so it works just fine with 32-bit?
[19:46] <mrxmike> i think it does yes
[19:46] <mrxmike> (well quiet sure actually)
[22:22] <lafeuil> Hi, on intrepid, why the current version of linux is not the 2.6.27 on powerpc ?
[22:32] <munckfish> lafeuil: because powerpc is now community supported and therefore uses the linux-ports kernel source not the standard one
[22:33] <munckfish> it's stuck 2.6.25 cause no one was motivated enough or had enough time to upgrade it
[22:33] <munckfish> I'm hoping that'll change in intrepid+1
[22:33] <munckfish> cause I'm hoping to do something about it myself
[22:35] <lafeuil> munckfish: Ok thanks for your answers
[22:35] <munckfish> lafeuil: np
[22:35] <lafeuil> When is intrepid+1 ?
[22:36] <munckfish> well pretty much 6.5 months from now I guess
[22:36] <munckfish> lafeuil: is there a specific feature you're missing?
[22:37] <lafeuil> munckfish: no, I understand now
[22:38] <lafeuil> munckfish: thank you
[22:38] <lafeuil> bye
[23:07] <Kano> hi, did anybody test gspca with 2.6.27? in worst case it does not even boot with webcam connected...
[23:09] <Kano> 2 ppl dont get the webcam working with it... with 2.6.26 old gspca it works
[23:14] <administrator_> hey all... im reading the ubuntu custom kernel docs... i dont want to use the git method on a production kernel, do i? i basically want a kernel built the same way as the default kernel, + the 8159 wlan driver...
[23:15] <Kano> just compile it externally,what the problem
[23:17] <administrator_> Kano: isnt there a certain kernel package out there just for (k)ubuntu?
[23:17] <Kano> you only need the headers + a driver that compiles against the kernel
[23:17] <administrator_> with whatever relevant patches that we use
[23:17] <administrator_> Kano: the driver is in the kernel id like to use
[23:18] <administrator_> so i just grab any linux-headers and kernel, and i dont worry about patches or anything like that?
[23:18] <Kano> which config option?
[23:18] <administrator_> Kano: its the wlan module for 8159 chipsets
[23:18] <Kano> config option?
[23:19] <Kano> there is no option with 8159 in the name..
[23:19] <administrator_> Kano: i dont know, i head someone say it was in there so i was gonna look through menuconfig
[23:19] <Kano> i think you are dreaming
[23:19] <administrator_> check 8150
[23:19] <administrator_> and pinch me
[23:19] <Kano> CONFIG_USB_RTL8150=m
[23:19] <Kano> is already set
[23:19] <administrator_> does it add support for 8159 in the newest kernel or something?
[23:20] <Kano> /sbin/modinfo rtl8150
[23:20] <Kano> shows it, what i find problematic is that rt2500usb and rt73usb share the same ids
[23:22] <administrator_> that doesnt look like it supports the 8159 chipset anywya
[23:22] <Kano> lsusb
[23:23] <Kano> then you can check the id
[23:24] <administrator_> its not a usb nic
[23:24] <Kano> then use lspci -nn
[23:24] <Kano> and show the line
[23:24] <administrator_> Kano: im beginning to think i was misinformed.. sure, one sec
[23:24] <Kano> 8150 IS usb
[23:25] <Kano> also wlan is very simple to use with ndiswrapper
[23:25] <administrator_> im a knuckle head, its 8185
[23:26] <administrator_> and i have serious problems with ndiswrapper, which im certain constitute a bug
[23:26] <administrator_> http://ubuntuforums.org/showthread.php?p=5931608
[23:26] <administrator_> which is why im here
[23:27] <Kano> administrator_: rtl8180 is already there, if you want to use ndiswrapper you have to blacklist it
[23:28] <administrator_> i dont have ndiswrapper, but i didnt have wlan0, so i assumed i needed it
[23:29] <Kano> administrator_: grep 10ec /lib/modules/$(uname -r)/modules.pcimap |grep 8185
[23:29] <administrator_> sorry, i meant to say 'i dont want to use ndiswrapper, im only using it because i didnt have an interface for my wlan, so google searching seemed to imply i needed it_
[23:29] <Kano> ndiswrapper is usually very good
[23:29] <Kano> but in that case you would have to blacklist rtl8180
[23:30] <administrator_> Kano: nothing returns from that grep
[23:30] <Kano> then your kernel is too old
[23:30] <administrator_> i have 8139 and 8169, which are obviously wired drivers
[23:30] <Kano> i use 2.6.27-6
[23:31] <administrator_> Kano: and you have it right?
[23:31] <Kano> of course
[23:31] <administrator_> great, sorry for the confusion, my fault
[23:32] <administrator_> so i can just grab linux kernel sources from anywhere right? there isnt any 2.6.27 *buntu specific-patched kernel that you would recommend i use
[23:32] <Kano> in 8.10?
[23:33] <Kano> it is
[23:33] <administrator_> these need to go into production, asap, or i would just use intrepid personally
[23:33] <administrator_> i need a kernel for use with 8.04.1
[23:51] <administrator__> Kano: thanks, those .debs worked great
[23:51] <Kano> fine
[23:51] <administrator__> lol