[12:58] <login_> hello
[01:00] <login_> I have been compiling kernel 2.6.22 and have run across a dilemna. My compiled kernel weighs in at about 400-500 mb while the ubuntu kernel weighs about 100-200 mb. How did you guys get the kernel to be so thin?While i was uncompressing the kernel i saw it uncompress folders such as sparc so is it possible that my kernel installed all architectures? if so , how can i make it only use 1386?
[01:03] <mjg59> Disable CONFIG_KALLSYMS_ALL
[01:04] <login_> thank you! i will try that when i recompile
[01:04] <login_> how do i disable it? do i edit something in the menuconfig?
[01:05] <mjg59> Yes
[01:05] <login_> thank you. the name will be exactly like that right?
[01:09] <login_> i am compiling right now but i cant find CONFIG_KALLSYMS_ALL in make xconfig window
[01:09] <mjg59> It's under debug options
[01:17] <login_> hmh i cant seem to find debug options. i feel very noobish to ask but could you give me a type of written map to where to find this option? like go togeneral setup and etc
[01:17] <login_> or is there a way to do this i nthe command line
[01:17] <login_> where i do disable CONFIG_KALLSYMS_ALL
[01:31] <login_> ?
[04:39] <yulin> Is there some good solution for promote the kernel compatibility about harddisk controller?
[05:24] <mikmorg> Could someone assist me as to where the debian post/pre inst/rm scripts can be found for linux-image, linux-headers, and linux-source? I figured they would be somewhere in git://kernel.ubuntu.com/ubuntu/ubuntu-feisty.git, but I can't find them.
[05:28] <crimsun> that's because make-kpkg (in the kernel-package binary package) handles it.
[05:29] <mikmorg> crimsun: Do you know where I should send a patch to those files?
[05:29] <mikmorg> crimsun: There is a problem with the kernel-img.conf parsing that needs to get fixed ASAP, hopefully before 16-30 comes out
[05:33] <crimsun> mikmorg: file a bug against the kernel-package source package, and attach the diff(s).
[05:36] <mikmorg> crimsun: I filed the patches against the files in /var/lib/dpkg/info/linux-*. I will change the target from linux-source to kernel-package. Thanks.
[12:07] <kraut> moin
[02:21] <zul> hey
[02:48] <zul> BenC: ping
[02:48] <zul> if your ae up
[02:57] <BenC> zul: hey
[02:57] <BenC> zul: Can you clean up your tree to remove the commit/revert?
[02:57] <zul> sure
[02:58] <BenC> should be able to branch from ubuntu-gutsy, and cherry pick the changes you want to keep
[02:58] <BenC> zul: thanks...so what did you need?
[02:58] <zul> that :)
[02:58] <BenC> heh
[02:58] <zul> ill clean it up tonight and request another pull
[03:02] <aquo> hey folks
[03:02] <aquo> i have some question
[03:02] <aquo> what is the status on https://bugs.launchpad.net/ubuntu/+source/network-manager/+bug/50214
[03:04] <aquo> aquo@cayley:/etc$ iwconfig --version
[03:04] <aquo> iwconfig  Wireless-Tools version 28
[03:04] <aquo>           Compatible with Wireless Extension v11 to v20.
[03:04] <aquo> Kernel    Currently compiled with Wireless Extension v21.
[03:05] <aquo> who is the person responsible for this?
[03:10] <zul> BenC: can you add to the kernel gitguide on the wiki on how to use cherrypick?
[03:11] <BenC> zul: git-cherrypick <sha>
[03:11] <zul> ah ok
[03:11] <zul> hey pkl
[03:11] <pkl__> hi zul
[03:13] <aquo> BenC: can you tell me who is responsible for bug what is the status on https://bugs.launchpad.net/ubuntu/+source/network-manager/+bug/50214
[03:14] <BenC> aquo: how about giving me a quick overview so I don't have to read the entire report, please?
[03:14] <aquo> kernel is compiled with wrong Wireless Extension
[03:15] <BenC> kernel is compiled with stock 2.6.22 wireless extensions
[03:15] <BenC> it's the only way it can be compiled
[03:15] <aquo> aquo@cayley:/etc$ iwconfig --version
[03:15] <aquo> iwconfig  Wireless-Tools version 28
[03:15] <aquo>           Compatible with Wireless Extension v11 to v20.
[03:15] <aquo> Kernel    Currently compiled with Wireless Extension v21.
[03:15] <aquo> ath0      Recommend Wireless Extension v13 or later,
[03:15] <aquo>           Currently compiled with Wireless Extension v21.
[03:15] <BenC> I don't see the problem there
[03:15] <BenC> v20 wireless-tools works fine with v21 kernel extensions
[03:15] <aquo> no, not everytime.
[03:16] <BenC> then wireless-tools needs updating
[03:16] <aquo> it is not possible to connect to hidden networks
[03:16] <BenC> it's not like I can tell the kernel to be v20
[03:16] <BenC> how can you be sure that v21 kernel extensions is what is causing that?
[03:16] <BenC> these extensions are meant to be compatible in cases like this
[03:16] <aquo> should be possible, the bug report states that people solved the problem by recompiling the kernel
[03:17] <BenC> v20 wireless-tools just wont use the v21 extensions
[03:17] <BenC> recompiling it how?
[03:17] <aquo> not sure, changing the version of Wireless Extension v21
[03:17] <BenC> I heard about this whole thing last week, and I'm just not convinced that v21/v20 compatibility is the problem
[03:18] <BenC> aquo: if you just change a line in the kernel to say it's v20, then you are actually breaking things
[03:18] <rtg> I think its more a problem with hidden SSID, not  WEXT versions.
[03:19] <rtg> There is alot of FUD in that report.
[03:19] <BenC> yeah, the version information is a red herring, IMO
[03:19] <aquo> i don't think so, i would be interesting to see if it worked with the original kernel on the media ...
[03:19] <aquo> 20-15
[03:19] <BenC> aquo: you can't compare 2.6.20 vs. 2.6.22 just based on v20/v21 kernel extensions
[03:20] <rtg> Yeah, there is too much change in the wireless subsystem.
[03:20] <aquo> i am just comparing 20-15 and 20-16
[03:20] <BenC> the bug could very well be somewhere else, or it could be in wireless-tools, and the fact that it breaks because it assuming something it shouldn't
[03:20] <aquo> i am still on feisty
[03:20] <rtg> aquo: Can you confirm it works if you enable broadcast SSID?
[03:21] <BenC> aquo: ok, then I'm getting ever more conflicting info on this bug report
[03:21] <aquo> yes, with broadcast it is working.
[03:21] <rtg> aquo: I think that implies the WEXT versions are fine. There might be something between wpa_supplicant and network manager. 
[03:22] <aquo> hoe, i think it is a major problem if i am not able to connect to hidden networks
[03:23] <BenC> aquo: I read a little more of the report, and it seems that the wireless-tools maintainer is agreeing with your assumption that v20 tools is breaking with v21 extensions
[03:24] <BenC> aquo: IMO, that clearly says that wireless-tools for v20 extensions is broken
[03:24] <aquo> "Hi Julie, upgrading only wireless tools won't make difference, because the kernel is recompiled with Wireless Extension version that is not supported by Wireless Tools. This is a simple mismatch and the kernel team can easily solve this by recompiling the kernel. I already did that on my system and it works."
[03:25] <BenC> aquo: I'd really like someone to tell me how to recompile the kernel with an older wext
[03:25] <BenC> that bit of info seems to be clearly absent from the report
[03:25] <aquo> don't know, but i will investigate.
[03:25] <BenC> it's not like it's a switch, or a config options...it's basically backing down the entire tree, subsystem and drivers, to older code
[03:26] <BenC> which we wont do
[03:26] <aquo> but first i want to check the situation on the original release media ...
[03:26] <BenC> so if newer wext tools fixes things, best bet is to find the patch that fixes this case
[03:26] <aquo> want to check if normal feisty has matching versions
[03:27] <BenC> we wont be making sweeping "downgrade wext" or "upgrade wireless-tools" changes in feisty, period
[03:29] <BenC> aquo: I'd also appreciate if you could update this bug report with what we discussed, to keep everyone informed of the status
[03:30] <aquo> ok
[03:32] <BenC> aquo: thanks
[04:41] <movi> i think i found a bug in the newest kernel in gutsy (and -7 too)
[04:41] <movi> the thing is, im using a Celeron 2.6
[04:42] <movi> and ksoftirqd/0 is using about 25-30% of the cpu all the time
[04:42] <movi> someone pointed me to look and /proc/interrupts
[04:42] <movi> and the thing is, timer only generates 62 interrupts, then stops
[04:43] <movi> i tried booting with noapic and lapic
[04:43] <movi> n
[04:43] <movi> noapic and nolapic
[04:43] <movi> and then the timer does generate interrupts, but the process eats up 35% of the cpu all the time
[04:44] <movi> 2.6.20 (on 7.04) works perfectly
[04:44] <kylem> while the timer "doesn't" generate interrupts, is "LOC" increasing?
[04:44] <movi> yes
[04:44] <movi> and lots it too
[04:44] <kylem> then it's fine.
[04:44] <kylem> LOC is the local APIC timer
[04:45] <movi> why the cpu usage then ?
[04:45] <movi> i tried disabling the usb controller and unloading the nvidia module - nothing
[04:45] <login_> Hi guys. I recompiled kernel 2.6.22 on my ubuntu box and comparing to ubuntu's kernel , it is ismply 5 times larger! my deb file is 194 mb and when i isntall it , it is 500 mb. From synaptic , i see that the ubuntu kernel is about 75 mb . how can i make my 2.6.22 compiled kernel thinner?
[04:45] <movi> i
[04:45] <movi> i booted in single mode too to rule-out any rigue processes - nothing
[04:45] <kylem> meh. i wouldn't worry about it.
[04:45] <pmjdebruijn> login_, how did you 'compile it'
[04:46] <pmjdebruijn> login_, anyway do you _need_ 2.6.22?
[04:46] <login_> someone suggested do disable CONFIG_KALLSYMS_ALL ya , i am trying to make an ubuntu remaster
[04:46] <login_> i compiled it by making it into  a deb package
[04:46] <login_> from this tutorial
[04:46] <login_> http://www.howtoforge.com/kernel_compilation_ubuntu
[04:46] <movi> login_: is this addressed to me ?
[04:47] <login_> to everyone that can help me
[04:47] <login_> u asked me how to ocmpile it so ya
[04:47] <login_> i made the kernel into  a eb package and the linux-image was 194 mb
[04:47] <pmjdebruijn> login_, that's the quick 'n dirty way to do it
[04:47] <movi> i didnt ;] 
[04:47] <login_> o didnt see it :P
[04:47] <pmjdebruijn> anyway it's weird that you get such large packages
[04:47] <login_> could yo utell me wich is the long and not dirty way to do it?
[04:48] <pmjdebruijn> it's on the wiki somewhere
[04:48] <login_> the ubuntu wiki?
[04:48] <pmjdebruijn> yeag
[04:48] <pmjdebruijn> yeah
[04:48] <login_> ok i will search for it
[04:49] <login_> man this is the largest ammount of people i have seen talking in this chat
[04:49] <login_> i am amazed
[04:50] <movi> so anyone know what should i do about my ksoftirqd problem ?
[04:50] <movi> (except going back to 2.6.20 and ubuntu 7.04) ?
[04:51] <aquo> BenC: https://bugs.launchpad.net/baltix/+source/knetworkmanager/+bug/50214/comments/38
[04:53] <login_> is the kernel tutorial this one?https://wiki.ubuntu.com/KernelCustomBuild?highlight=%28kernel%29. I am concerned that it told me to download the kernel source from the repos so it will be 2.6.20 rather then 2.6.22. Is it possible for me to replace that with the kernel 2.6.22 source . will it make a difference?
[05:01] <login_> aww you guys dissapeared again :(
[05:15] <login_> ill be waiting for you guys to reply :P
[06:08] <login_> ?
[06:21] <zul> BenC: I cleaned it up
[06:32] <login_> anyoe have an anwser for my second question?
[08:00] <login_> aynone?
[08:00] <login_> have an anwser to my question
[08:41] <login_> Could you guys tel lme what you did with kernel 2.6.22 to make it so light? my kernel com[pile comes to 193 mb
[08:41] <login_> !
[08:49] <infinity> login_: You've been asking this all day; this isn't a support channel.
[08:51] <login_> i know but no one has really and trly anwsered. I am sorry if i am being a bother but i dont know where to go since in the ubuntu forums if you ask anyhting and it is hard it gets ignored
[08:51] <BenC> login_: find /lib/modules/ -name \*.ko | xargs strip
[08:52] <login_> i type that when i am compiling?
[08:52] <login_> or do i isntall it and delete every .ko file that has xargs strip in it?
[08:56] <login_> plz anwser this will be my last question :'(
[08:59] <bdmurray> login_: you might find 'man strip' and 'man xargs' helpful
[08:59] <login_> ok
[09:02] <login_> i still dont get it .:(. O well. Does the gutsy kernel have any stamps on it saying it is an ubuntu kernel?
[09:03] <mjg59> login_: The Ubuntu kernel build process strips the modules before putting them in a package. If you do it by hand (without using the full kernel build process) you need to do that yourself.
[09:04] <login_> o so i run  find /lib/modules/ -name \*.ko | xargs strip while compiling
[09:05] <mjg59> No, when installed
[09:06] <mjg59> And I suspect you don't want to run strip without any arguments - that'll remove symbols needed for loading the modules
[09:06] <mjg59> But this really isn't the place to ask
[09:06] <mjg59> If you want to avoid it altogether, just disable CONFIG_KALLSYMS_ALL
[09:07] <login_> i disabled that but it only went down a couple of kb
[09:07] <login_> also , whe ni run the command it says this strip: Warning: '/lib/modules/2.6.22-custom/kernel/net/wanrouter' is not an ordinary file
[09:07] <login_> Again i am really sorry for asking noob questions here
[09:08] <mjg59> login_: Then don't?
[09:08] <login_> ok :( . bye
[09:28] <lamont> amitk said he was going to include jbailey's patch in -8.17, and yet it's still not in -8.18.... hrmpf
[09:29] <lamont> BenC: what do I need to do to get that hppa patch in before tribe3 freeze?
[09:30] <BenC> lamont: it will get in, we just didn't have the time to do it last week
[09:30] <BenC> parts of the patch were actually wrong, but we'll get it munged up properly
[09:30] <lamont> on the off chance that there are other issues, and because I still need to build debian-installer et al, it'd be nice if it hit the archive sooner rather than later....
[09:30] <lamont> thanks
[10:16] <bdmurray> I've started tagging bugs as kernel-oops BenC.  There are a few now.
[10:16] <kylem> do we have real tags yet?
[10:16] <kylem> or can people still tag anything however they want?
[10:22] <bdmurray> We have standardised tags but people can still add whatever they want.
[10:22] <bdmurray> The list of tags is at https://wiki.ubuntu.com/BugSquad/Tags
[11:07] <lamont> BenC: I have  the various patches in my tree now, I'll send email to kernel-team tonight.  must run now.
[11:08] <lamont> and I'll do a test build, too.