/srv/irclogs.ubuntu.com/2009/06/02/#ubuntu-kernel.txt

dtchenmaxb: the debug packages are generated from the source package; e.g., linux-image-debug-2.6.30-7-gene01:42
dtchenric01:42
dtchenmaxb: keep in mind that the tree is in flux while ports is rolled back in, so the packaging may well be different for -801:43
maxbNo, this is a state of affairs that has existed since Intrepid if the comments in malone are to be believed02:13
maxbIt looks like ddebs.u.c is keeping only one or two recent package versions, without regard for multiple distroseries02:14
mase_workhey guys , i  am running into the same symptoms as https://bugs.launchpad.net/ubuntu/hardy/+source/linux/+bug/240071 on a fully up to date 8.04.2 server. It says a fix has been released but I can't see anywhere what the fix actually was and if that means it's been released as part of the regular update cycle of it need a special repository03:06
ubot3Malone bug 240071 in linux "BUG: soft lockup - CPU#0 stuck for 11s!" [Undecided,Fix released] 03:06
mase_workI was hoping someone could guide me as to how i can determine if I am hitting this same issue or if I have something different that just presents the same03:07
cwillumase_work, that's an error, it's fixed in intrepid, somebody said they'd nominate it for hardy, but it looks like that didn't happen03:48
cwillumase_work, you might be able to get away with simply installing intrepid's kernel packages though03:48
cwilluactually, ignore that03:48
mase_workah ok. so i need to apply this patch normally03:50
mase_workmanually rather03:50
mase_workcwillu: sorry this might seem like a stupid question, is it possible in launchpad to get a patch or a git commit id which i could use to grab the patch 03:54
cwillumase_work, sometimes, but in this case, I don't think so03:55
cwilluit looks like it was a "here's a new version with lots of stuff fixed, is your bug fixed?", without a specific patch in mind03:55
=== mcasadevall is now known as NCommander
mase_workI see. Do you have any recommendations as to how best to deal with this ? 04:00
mase_workshould i revert to a previous kernel or move to a newer kernel ?04:00
cwillumove to a newer one is probably the best bet, you can just install the deb files for a later kernel image, they should Just Work.04:08
mase_workcwillu: so the intrepid-xen images are acceptable ?04:12
cwilluI would imagine s04:13
cwilluso04:13
mase_workok thanks. I am using pygrub so i should be able to do it. Do you know if the vpcu=1 work around that is listed is likely to actually work ?04:14
ZeroProghey guys im trying to learn some driver writing but am having trouble compiling and testing....is there an ubuntu doc for doing this?04:16
ZeroProganybody in here04:43
mase_workyeh there is04:47
mase_worki just don't know the answer to your question sorry04:48
mase_worki could google around but i presume you have already tried that approach and been unsuccessful04:48
ZeroProgexactly04:48
ZeroProgi guess the best place to start would just be ubuntu kernel programming but i cant seem to find a tutorial 04:49
lifelessits not an ubuntu specific thing04:50
lifelesslook for linux kernel docs/tutorials04:50
lifelessrusty russell has done several tutorials at LCA for instance04:50
ZeroProgi have but after talking to people in #kernelnewbies and they said ubuntu is different04:51
lifelessUbuntu takes the kernel and we add some patches and package it. Its still the same thing though in terms of writing code for it.04:52
ZeroProgyes but the source and headers and stuff is different is what i gather04:52
lifelessnews to me04:52
ZeroProgi followed 3 tutorials to the letter and i still get the same errors about not finding linux/init.h04:52
ZeroProgthe guys in the other channel said something about make-kpkg04:54
lifelessare you trying to build with the full kernel source locally?04:54
lifelessor using your installed kernel's headers?04:55
ZeroProgwhatever i need to do to write drivers heh04:55
lifelessdepends on what you're doin04:55
lifelessg04:55
lifelessbut I can't give you a useful pointer unless you tell me *what you were trying*04:55
ZeroProgu mean like what type of driver?04:56
lifelessno04:56
ZeroProgoh ok04:56
ZeroProgi understand....i read these tutorials about what was needed to write drivers and the consensus was to download and build a kernel source 04:56
ZeroProgso i followed the readme included with the source04:56
ZeroProgother tutorials said headers were sufficient but i still cant get ubuntu to recognize linux/init.h and the other includes 04:57
lifelessok, following the full source approach is easiest I think04:57
lifelessand the vanilla upstream kernel will build on Ubuntu04:57
lifelessin fact we use that for testing 04:57
ZeroProgok...is there perhaps a newb tutorial on downloading and installing that?04:58
lifelesssounds like you were following one04:58
ZeroProgyeah :/04:58
lifelessso, I'd follow it. Do the full source one, not the headers package - that way you are guaranteed to have everything basically the same.04:59
ZeroProgok04:59
ZeroProgthanks again04:59
ZeroProgwhere can i find that kernel05:04
lifelesshttp://kernel.org/05:05
ZeroProgohh ok i thought it was something specific my fault05:05
ZeroProgone last newb question for the night....how do u know its vanilla upstream05:06
lifelessif you get it from upstream :)05:11
ZeroProglol is there an upstream section on the site?05:11
mase_workthat is _the_ upstream05:13
ZeroProgjeez im confused...what makes it vanilla is that what its named?05:13
mase_workno 05:13
mase_workvanilla just means no added patches05:13
mase_workits the tree that linus holds05:14
mase_workUbuntu uses that then applies some added patches for various things05:15
mase_workand packages that up as the ubuntu kernel05:15
mase_workbut for the most part they are the same05:15
ZeroProgohhh ok05:15
mase_workthe version on kernel.org is the unpackaged source version.05:16
ZeroProgalright im gonna leave before i ask anything else dumb thanks again05:16
mase_workZeroProg: have you done much C programming before ?05:16
ZeroPrognot on linux but yes05:16
mase_workok cool. 05:16
mase_workhave fun05:17
ZeroProghaha ill try thanks05:17
=== fabbione is now known as fabb-rome
ZeroProgok i installed the new kernel....to use it it tells me what to do but after i do that i should be able to code correct?05:59
ikepanhchi smb12:01
* apw waves to ikepanhc 12:37
ikepanhchi apw12:47
ikepanhchave a question for you: bug 36096612:48
ubot3Malone bug 360966 in linux "bnx2x missing in initrd for install media" [Medium,Won't fix] https://launchpad.net/bugs/36096612:48
ikepanhccould I set affect to initramfs-tools?12:48
ograikepanhc, more likely debian-installer 12:49
ikepanhcogra: you mean debian-installer makes the initrd?12:50
ograinitramfs-tools is only used on installed systems or in conjunction with casper for the live images12:50
ograthe alternate installer uses udebs ... ogasawara mentioned ./debian/d-i/modules/nic-modules, thats a udeb spec file 12:51
ikepanhcoh, you are right12:51
ograso its actually a linux issue ... d-i only uses the produced udeb12:52
ikepanhcI know what I am wrong, I test with a new kernel install12:53
ikepanhcthanks ogra12:54
ikepanhcogra: but it is useless to send as Jaunty SRU, because we have released it12:55
ograi'm not sure, it would require a d-i SRU as well i think, cjwatson could answer that12:56
ikepanhcso, could we assume this issue only affect him because he use NFS as rootfs?12:58
ograwell, he doesnt talk anout nfsroot12:59
ograjust that there is no network available12:59
ikepanhcThe jaunty release kernel image contains the module he need, but he said he need the module when initrd.13:00
ograright13:00
apwthanks ogra 13:01
ograbut debian-installer uses the nic-modules udeb13:01
ograin the netboot iso the udebs are included, in the normal d-i the udebs are loaded from CD13:01
ograeither way the driver needs to be in the udeb, no matter if you do a local or netboot install13:02
ograindeed in case of nfsroot its more fatal :)13:03
ikepanhcThanks ogra, I will check the d-i and udeb13:03
ikepanhcI assume he use NFS rootfs because he will have the module after mounting local rootfs13:04
ikepanhcThen I have to check if there is another net module missing13:04
ograwell, how would he do an nfsroot install without network ? :)13:05
ikepanhcohoh, I forget this13:05
ikepanhcanyway, I re-open this bug :-D13:06
ograif its for SRU talk to cjwatson first about the d-i situation13:07
ograits pointless to add it to the udeb if d-i wont use it13:07
ikepanhcgot it13:10
ikepanhcthanks ogra13:11
apwsmb, you had a machine which worked, stopped working for a bit, and started working again on later kernels with no obvious changes in the area of the fault.   what was the symptoms there?13:24
smbapw, You probably mean the acer which crashed when chaging the kernel data ro13:25
apwyeah thats the one... thanks13:26
smbIt was the AA1, and crashed early on in the boot.13:26
apwyou remember nigelp's sata issue.  there his machine failed on -11.38 worked on either .39 or .40 and stopped again on .4113:26
apwand yet there is no sata changes in that sequence13:26
smbusing acpi=off made it working in most cases (strange enough)13:27
smbSounds weird indeed13:27
apwnnng... 13:27
apwanother wierdy for sure13:27
smbhow did nigelp's machine not work (mean the symptom there)?13:28
apwsata timeouts and panics (i believe he has flashing capslock syndrom at times)13:28
smbslightly different than the other as that was strictly locatable to the point of making the ro change (though sometimes you got killed by recursive stacktraces)13:30
smbDo we have any stactrace hint?13:30
apwnothing yet that i have seen.  he is testing -proposed and latest mainline for me now13:33
ckingapw: it's weird that we have not see this kind of issue on any other identical hardware13:48
apwi know ... very13:48
ckingapw, I suspect the H/W is borked13:54
apwok he has -proposed installed -12 kernel and is now happy13:54
apwexplain that one to me13:54
rimvishi, i have problem with aoetools who can help me?13:56
smbapw, random memory relocation and timing?13:56
apwsmb, its cirtainly looking like that kind of thing ... very annoying to debug13:56
smbapw, You think a fully debug (kernel hacking) kernel might help? Unfortunately this often removes the race13:58
apwyeah not sure ... cirtainly intereested in getting the latest mainline tested there13:58
apwto see if its going to come back when he upgrades to karmic13:59
apwnigelp is with us now :)13:59
apwnigelp, so you have no problems now with the -proposed kernel ?13:59
nigelpapw, none of those SATA issues that showed up that I had to do irqpoll and all_generic_ide for.14:01
apwvery perplexing14:01
nigelpthe test that triggered it was downloading a CD image.. I've just downloaded 10 in a row; it never used to make it past 100Mb before.14:02
smbIt will be interesting to see how the -13 kernel that will be in -proposed soon fares14:02
smbIf that fails it is definitely weirdness time14:02
apwnigelp, what was your disk controller?14:04
apwyeah a quick review of the changes between where it did not work before and the -proposed kernel isn't showing anything i would expect to help you14:07
nigelpapw, sorry, being ignorant.  How do I find out what my disk controller is? 14:09
apwlspci | grep -i sata 14:09
apwmight do the trick14:09
nigelpapw, is this what you're looking for?  ATI Technologies Inc IXP SB400 IDE Controller (rev 80)14:10
apwyep14:11
apwnigelp, we need to get all this data into the same bug.  if you have one with support lets get that out of them, else we need a new one14:12
apwi can imaging you want to get some work done now its working! but if you have a chance later to test the mainline kernel i pointed you to that would help.  and get that info into the bug too14:13
apwthen let me have the number when you have it14:14
nigelpapw, let me ask shang if he opened one.14:14
=== sconklin-gone is now known as sconklin
nigelpapw, support say its the same bug as: https://bugs.edge.launchpad.net/ubuntu/+source/linux/+bug/20663514:23
ubot3nigelp: Error: Could not parse data returned by Malone: timed out14:23
* apw pokes ubot314:24
apwbut that bug is duplicate of a bug which was fixed in intrepid and jaunty14:25
nigelpok.. so time for a new one?14:25
apwway back in feb.  so ... if you are seeing it now its either a new one, or that one was not fixed14:25
apwfor now i would open your own bug, and add your info there, if it _is_ dup we can dup it later14:26
apwubuntu-bug linux should do the trick14:27
=== BenC1 is now known as BenC
apwsmb, the karmic meta update should be in my sign directory if you could look it over14:35
smbapw, sure14:35
* ogra hugs BenC ... good to see you :)14:35
smbapw, uploaded14:43
apwsmb thanks14:43
* cking waves to BenC14:44
BenCogra, cking: Hey14:57
* BenC is out, packing for san fran14:57
apwnigelp, can you subscribe me to the bug when it exists15:06
nigelpapw, sure15:07
apwsmb, for hardy do we use linux-image-2.6.24 still?16:34
smb_tpapw, Hardy is 2.6.24, yes16:43
smb_tpor do you mean the naming?16:43
apwi mean do we use linux or something else for bugs there16:43
Kanohi apw , do you update the karmic kernel now and not rtg?16:44
apwour team is responsible for it, i am handling the rebases for a bit while rtg is off doing other things16:44
smb_tpapw, Hardy uses linux16:44
apwsmb_tp, thanksk ... and when did you go back to _tp?16:45
smb_tpapw, When I disconnected and irc still thinks me is there16:45
=== smb_tp is now known as smb
apwsmb, heh ... still not reset your router at midnight then16:46
smbapw, No, you should poke me :)16:46
smbapw, BTW, you can see what label to use in debian/changelog. It was linux-source-x.y.z before hardy16:48
apwahh before hardy.  good16:48
Kanoapw: are you interested in an aufs2 patch17:11
apwan updated one for karmic kernel?17:11
Kanoyes17:12
Kanoi tested it even with your live iso17:12
Kanocjwatson tries to use unionfs-fuse but his approach does not boot17:12
apwsend it over.  we are looking at vfs-union mount right now for the live cd's but should that not be possible the aufs patch would be a good fallback17:13
Kanoi needed to update it as you removed aufs completely17:14
Kanohttp://kanotix.com/files/kernel/unused-patches/2.6.30-ubuntu-aufs2.patch.bz217:14
Kanothats the new one17:14
Kanoalso i would like to see a patch which removes the usb ids from rt2500usb which are also in rt73usb17:15
Kanobecause i had 2 users with problems where rt2500usb was loaded and nothing worked17:15
Kano4 ids are in both drivers17:16
apwwe need to get those ids listed in a bug i recon17:17
Kanoit is easy to get em17:17
Kano(/sbin/modinfo -F alias rt73usb;/sbin/modinfo -F alias rt2500usb)|sort|uniq -d17:20
Kano/sbin/modinfo -F alias rt73usb rt2500usb|sort|uniq -d17:26
Kanothats even shorter...17:26
Kanoapw: also how about enabling the lzma options for smaller kernel size17:27
apwfor compressed kernels?17:28
Kanoyou could even update update-initramfs and use lzma if the kernel config has the needed setting17:28
apwhow well tested is that17:28
Kanosure17:28
Kanoif you dont enable nobody tests it17:28
Kanoit is one of the major points for 2.6.3017:28
Kanoi hope that squashfs will get lzma too because it is now inside the kernel17:29
apwlzma was the one which had poor rsync behaviour i think?  making it near useless for CD's17:30
apwiirc17:30
Kanonope17:30
Kanodid you never hear of lzma patches for squashfs17:30
Kanohttp://www.squashfs-lzma.org/17:30
Kanoslax used it17:31
amitkapw: ack re: lzma + rsync not being friends17:32
Kanotar.lzma is now the default for slackware17:32
Kanoso many use it now17:33
apwas i remember the size savings are impressive and would be desirable for our live images, but a problem for the QA process as its non-rsyncable17:33
Kanoi dont get it17:33
Kanorsync is independent of the data17:35
=== bjf_ is now known as bjf
bjf*** There will be an Ubuntu Kernel Team meeting on #ubuntu-meeting at the top of the hour.17:50
apwKano, right but the data needs to be similar between images to allow rsync to be efficient17:55
apwand lzma produces images which are completely different for a small change in the contents17:56
Kanowell squashfs stores files basically always the same17:56
Kanoshould not matter that way17:56
apwbjf, good idea to announce here17:57
Kanodid you really test squashfs-lzma or pure lzma?17:57
apwseems it does, from the testing that foundations has done17:57
apwso they tell me yes17:57
=== Kano32 is now known as Kano
=== Kano32 is now known as Kano
Notch-1hi all, should anybody tell me something about https://bugs.launchpad.net/ubuntu/+source/loop-aes-source/+bug/342902 ?21:00
ubot3Malone bug 342902 in loop-aes-source "Build error: ‘struct bio’ has no member named ‘bi_hw_front_size’" [Undecided,New] 21:00
Notch-1in a few words, we need CONFIG_BLK_DEV_LOOP back to "m", is it possible?21:02
=== sconklin is now known as sconklin-afk

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!