/srv/irclogs.ubuntu.com/2008/04/16/#ubuntu-kernel.txt

greearber, ok..trying again00:00
greearbthis is just a small little package?00:03
greearblinux-meta-.... ?00:03
amitkgreearb: you need to get 'apt-get source linux-image-2.6.24-16-generic'00:04
greearbok, sorry for my confusion...never used debian so much...downloading that now00:05
alex_joniamitk: you wouldn't be familiar with the livecd by any chance?00:07
amitkgreearb: it is confusing even to me sometimes :)00:07
alex_joniI'm looking for the list of packages which get installed by default.. and I can't seem to remember where that was00:08
amitkalex_joni: not particularly and definitely not at 2am :)00:08
alex_joniamitk: same TZ as here ;)00:08
amitkalex_joni: you want to catch cjwatson when he is online00:09
greearbok, that rules.d looks correct now.  It'll take me a few hours to port my patch forward probably...00:09
alex_jonigreearb: if you have a single patch file it's not that complicated00:10
alex_joniI would still create a custom flavour (call it greearb)00:10
greearbheh, I have a large number of patches to various parts of the kernel00:11
alex_jonithen look at debian/binary-custom.d00:11
greearbok00:11
alex_jonicreate a new folder called greearb ;) and put the patches in there..00:11
alex_jonithere's a README which will guide you further00:11
greearbok, will look at all that...I first need to get the patches so that they will apply to this kernel.00:12
alex_joni(not particularely sure it works with udebs for custom flavours.. but it should afaik)00:12
bigcx2hey all00:26
bigcx2i was in here this morning asking about squashfs00:26
bigcx2i've selected squashfs inside my kernel config00:26
bigcx2and i've run  make-kpkg --append-to-version -squashfs --revision r1 --initrd --config oldconfig kernel_image00:27
bigcx2but...00:27
bigcx2when i run update-initramfs -u -v i don't see that it's getting compiled into the kernel00:27
bigcx2any clues?00:27
cjwatsonalex_joni: given a live CD, the simplest way is to look at /casper/filesystem.manifest-desktop00:28
alex_jonicjwatson: I just figured that out.. thanks a lot anyway00:28
alex_jonibtw.. I read about 20 minutes of sources (ubiquity) until I reached that conclusion :D00:28
cjwatsonalex_joni: of course that's several steps down the chain; in general, it's the dependency-expansion of a bunch of "seeds"00:29
alex_jonicjwatson: short question though.. if I have a package I'm interested in.. I only need to put it in there, not it's dependencies too.. right?00:29
cjwatsonno, you need to put everything in there00:29
alex_joniis it me, or the older manifest-desktop didn't have versions?00:29
cjwatsonthere are ways to automate it (which we use in our own build processes) but they're probably overkill for a one-shot deal00:30
alex_joniyeah, sure.. I only have 1 package + dependencies00:30
cjwatsonalex_joni: it's you; the livefs manifests have had versions since they were created00:30
cjwatson(the keyword for the automation in question is "germinate")00:30
alex_jonihuh.. germinate? sounds fun00:30
cjwatsonsample dependency-expanded lists of packages for Ubuntu are at http://people.ubuntu.com/~ubuntu-archive/germinate-output/ubuntu.hardy/00:30
cjwatsongerminate: vt. to cause seeds to grow00:31
alex_jonicjwatson: cool.. probably extends my knowledge absorbing atm00:32
cjwatsonyeah, it's only if you care about how we maintain it00:32
cjwatsonvery much designed so that ordinary Ubuntu developers can forget about the process and not have to edit lists of packages every time a dependency changes00:32
alex_jonicjwatson: I do this once every couple of years/versions00:32
alex_joniand each time it's more complex (and automated)00:33
alex_joniso I'm not sure if anything I know now will be valid by 10.4 or whatever the next LTS will be :D00:33
alex_jonicjwatson: it looks like quite a lot of work00:34
cjwatsonthis particular bit of it has not changed a whole lot since 6.0600:35
cjwatsonalso, I documented the manifest handling in ubiquity/doc/README a while back00:35
cjwatsonyeah, manifest-desktop was created while ubiquity was still called espresso, so pre-release dapper00:36
alex_joniyeah, I noticed a couple of things are the same since dapper00:38
alex_joni(most of the customization part of the LiveCD - squashfs, etc)00:38
alex_jonicjwatson: I notice the docs on the wiki tell one to regenerate the filesystem.manifest then cp that as filesystem.manifest-desktop00:42
alex_joniso I guess that takes care of things00:43
greearbalex_joni, I think I have all of that documented, except for the new kernel...I'll be happy to email it to you...00:43
alex_jonigreearb: feel free if you think I can do anything with it :)00:45
cjwatsonalex_joni: unwise to cp, would need to cp and remove stuff00:46
cjwatsonI'm afraid I haven't been able to keep track of what people have been writing in the wiki - that didn't come from me00:46
greearbthere are about 10 wikis, at least, related to this :)00:46
alex_jonicjwatson: well.. they do remove ubiquity00:46
* alex_joni meant https://help.ubuntu.com/community/LiveCDCustomization00:47
greearbmost similar to others...and most slightly wrong it seems, depending on your version and what you are trying to do.00:47
cjwatsonalex_joni: I would be inclined to manually add the things you want to both, rather than trying to duplicate the list that's removed00:47
cjwatsonoccasionally I do try to dive in and clean stuff out, but labours of Heracles etc. :-)00:48
alex_jonicjwatson: how about after doing an apt-get update && upgrade.. which just pulled another 450 packages in?00:48
* alex_joni doesn't want to think about updating that list by hand00:49
alex_jonicjwatson: I'll do the cp right now (only temp. version for testing), then after the final hardy CD is released, I'll be careful to add only a couple of packages00:50
cjwatsonalex_joni: in that case diff the manifests beforehand and take account of what's removed00:51
cjwatsonalex_joni: ubiquity doesn't actually care about the versions in the manifest files - the versions there are just informational00:51
cjwatsonso if you literally did upgrade and not dist-upgrade, then it won't actually matter00:51
cjwatson(except cosmetically)00:51
alex_joniheh, ok.. good to know00:51
alex_jonibtw, whoever did parallel makesquashfs is my hero00:52
alex_joniParallel mksquashfs: Using 2 processors00:52
greearbyeah, it only takes maybe 10 minutes now :)00:53
alex_jonicjwatson: thanks again, I'm off to bed now00:58
=== smb_away is now known as smb_tp
jdongalex_joni: yeah, multithreaded squashing is my best friend too01:07
bigcx2anybody have any clues about the squashfs stuff?02:45
greearblike what?02:59
bigcx2i'm trying to compile squashfs into a 2.6.20 (ubuntu'ized) kernel03:05
bigcx2and it works but it never makes into the ramdisk?03:05
bigcx2doing an update-initramfs -u -v doesn't show it in there03:05
bigcx2this is for a livecd03:06
bigcx2so whenever i boot03:06
bigcx2it fails on mounting the root filesystem03:06
bigcx2which is obviously squashfs03:06
bigcx2i'm using make-kpkg with --initrd03:08
bigcx2if that helps at all03:08
bigcx2i'm not sure what else to try?03:09
greearbit seems a big pain to get a kernel updated on a live cd image...I'm trying to do the same myself.03:09
bigcx2apparently03:10
bigcx2have you gotten it to boot completely?03:10
greearbI'm hacking my patch set now..when it compiles, hopefully I'll get a bit further...  I'll write it up then and post it somewhere03:10
greearbnot with a custom kernel, but I did with an updated ubuntu kernel03:10
greearbie, something you can get with apt-get install ...03:11
bigcx2ahh -- what k version, what ubuntu version?03:11
greearb7.10 (er, current stable..whatever that is), and 8.04-beta is what I'm working on now.03:11
bigcx2i c03:12
bigcx2yea i'm hacking on feisty, which sucks -- the rest of the world has moved on lol03:12
greearbI'm guessing it will all be similar..but will be more sure when I actually get something working :)03:13
bigcx2hah yea03:14
greearbI got the debian/rules custom-binary-foo to build, but that didn't do udebs.  There is a binary-udebs target that seems to be compiling everything from scratch again...  Do I need to somehow tell it that it's part of the 'foo' build?04:33
bigcx2i talked to someone in here this morning that said udebs were only for the alternate install cd04:36
greearbwell, I'm trying to build an alternate install cd, so that's probably want I want04:36
bigcx2oh lol04:36
bigcx2in that case this might point you in the right direction04:37
bigcx2http://dsplabs.utt.ro/~juve/blog/index.cgi/0114755923204:37
greearbyep, found it already..it's for an older release, but some of it might still work04:38
bigcx2yea..isn't that what you were hoping for earlier :)04:39
=== imbrando1 is now known as imbrandon
=== asac_ is now known as asac
krautmoin08:08
=== doko_ is now known as doko
=== cradek_ is now known as cradek
TomJaegerHi.  What's the status on bug #124406?15:50
ubotuLaunchpad bug 124406 in linux "Keyboard keys get stuck and repeat (Feisty, Gutsy)" [Unknown,Confirmed] https://launchpad.net/bugs/12440615:50
TomJaegermore specifically, the issue Helge and I are seeing15:51
TomJaegerAlso, feel free to try and reproduce it, I don't think it's tied to any particular hardware15:53
alex_jonishort question on linux-meta.. trying to build debs for my custom flavour, but somehow the deps are twisted15:55
alex_jonidebian/control holds this dependency: linux-headers-$(kernel-abi-version)-rtai15:56
alex_jonihowever, after the package gets generated I see only linux-headers-2.6.24--rtai (missing the abi '16')15:57
alex_joni(looks the same for the traditional stuff I didn't touch.. like -rt)15:58
alex_jonihmm.. crap.. n/m me.. was missing gawk16:02
=== elmargol_ is now known as elmargol
TomJaegerbug #124406 anyone?16:27
ubotuLaunchpad bug 124406 in linux "Keyboard keys get stuck and repeat (Feisty, Gutsy)" [Unknown,Confirmed] https://launchpad.net/bugs/12440616:27
TomJaegerI should say that there is a patch and it just needs to be applied16:34
alex_joniwhat's a reason for an ABI bump?17:52
=== evand_ is now known as evand
thomthe ABI changing18:00
jdongthom: you have just made my day :D18:03
alex_jonithom: funny..18:08
alex_joniis it supposed to happen on released versions (e.g. hardy after it's shipped out) ?18:08
cjwatsonit can happen, yes18:09
cjwatsonsecurity fixes can change the module ABI from time to time18:10
cjwatsonthom's response is actually pretty accurate, but if you want something more detailed, the ABI in question is the interface between the kernel and modules; simplifying massively, if exported functions change their signatures such that modules need to be rebuilt to cope, that's an ABI change18:11
alex_jonicjwatson: I was vaguely familiar with what ABI stands for.. probably my question was bogus :)18:11
cjwatsonwe've had to change the ABI after release in almost every Ubuntu release so far18:12
cjwatsongutsy has been lucky so far18:12
cjwatsonbut all of dapper, edgy, feisty have had post-release ABI bumps18:12
alex_joniyeah, but in dapper it wasn't that problematic for me18:12
infinityalex_joni: What are you finding problematic about ABI bumps, post-dapper?18:13
alex_joniinfinity: having a custom flavour is what makes it problematic18:13
alex_joniand having to deal with lum, lrm, lbm and meta packages for all that18:13
alex_joniinfinity: don't understand me wrong.. what you guys have in place is a great infrastructure.. but it's a bit overengineered for my own needs...18:14
infinityAutomation is key. :)18:14
alex_joniinfinity: I understand.. but atm I'm doing the automation by hand.. step by step :)18:14
infinity"Oh look, new kernel"; fetch-sources; apply-local-patches; push-big-red-build-button; go-for-coffee.18:15
alex_joniinfinity: yeah, but the thing started as a livecd/install environment for our app (which contains kernel modules, and thus kernel-ver dependent)18:16
alex_joniso when ABI changes, and new kernels come out (which through git is quite easy to track), I still have to push a new release on our own software18:16
greearbok, back to trying to install a custom kernel on a live-cd.  Turns out the whole 'udeb' thing was a red herring, it doesn't build initrds for the live-cd.19:23
greearbI do have some good notes for anyone who wants to do that though....19:24
greearbnow, my problem is that I need to compile the headers for my custom kernel.19:24
greearbI got the source:  apt-get source linux-ubuntu-modules-2.6.2419:25
greearbthen attempted to build them:19:25
greearbakeroot debian/rules binary-arch arch=i386 flavours=foo19:25
greearbhowever, the auto-generated debian/control is still relating to generic and not foo.  I tried setting debian/d-i/kernel-versions to something with 'foo' in it, but it still does not fully work.19:26
greearbSince the debian/control file is regenerated each time, I can't even run 'sed' tricks to force it right.19:26
greearbsuggestions are welcome :)19:27
alex_jonigreearb: you need debian/rules debian/control first19:35
alex_jonibut that assumes you changed the needed places 19:36
greearbcan you be any more specific?  That doesn't make so much sense :)19:36
alex_jonicontrol.stub19:36
alex_joniyou need to add descriptions for you flav19:36
alex_jonithen you need to add you falvour to at least rules.d/i386.mk:custom_flavours19:37
alex_joniand to rules.d/0-common-vars.mk:all_custom_flavours19:37
greearbok, will try that in a sec..19:37
alex_jonibtw, back on breezy I needed two things19:38
alex_jonione is kernel + initrd (you can get those from /boot/ from you regular machine)19:38
alex_joniand the other is the udebs which get loaded lateron19:38
alex_jonibrb19:38
greearbdon't think I need the udebs, seems that is only for non-live CD builds19:39
alex_jonicjwatson: any ideas why I would get "/init: .:  159: Can't open /scripts/casper" ?19:47
alex_joni(I'll dig for it, but thought you might know without looking..)19:47
greearbseems control.stub might be auto-generated from control.stub.in ?19:48
alex_jonigreearb: you're getting warmer :P19:49
greearbno luck19:54
greearbbuilding with:  fakeroot debian/rules binary-arch arch=i386 flavours=foo19:55
greearbis that right?19:55
alex_joninope19:56
alex_joniI use fakeroot debian/rules custom-binary-arch=rtai iirc19:56
greearbfor the headers as well?19:56
greearbI'm not building the kernel here...just the headers package19:57
alex_joniI use fakeroot debian/rules custom-binary-rtai19:57
greearbok19:57
alex_jonithat builds image & headers too19:57
alex_jonibrb19:57
greearbbleh, I mean modules, not headers.  I need the cramfs modules and such19:57
greearbwill try poking some more19:57
=== ogasawara_ is now known as ogasawara
alex_jonicjwatson: changing boot option "boot=casper" to "boot=local" makes it go a bit faster .. but it hangs furtheron20:08
alex_jonicjwatson: I also get _tons_ of udevd messages so anything usefull is lost somewhere..20:08
greearbahh, by hacking debian/rules to add a fixup script after the debian/control is generated, I seem to be making progress20:15
greearbgot udebs for free, too :)20:16
greearbgot it to boot!  But, seems the gnome toolbar died, so no way I can find to open a bash window to check things in more detail.21:19
greearbnot sure if it's a qemu issue or something else21:19
alex_jonicjwatson: probably didn't update initramfs (mkinitramfs -o /initrd.gz 2.6.24-16-rtai)21:54
cjwatsonalex_joni: you forgot to install casper before running update-initramfs -u?22:23
cjwatson(or other mkinitramfs method)22:23
alex_jonicjwatson: probably just borked the initramfs (used the install one, not the live-cd generated one)22:28
cjwatsonoh, don't do that :)22:31
greearbany idea what the apt-get package for the 'install' desktop icon on the live-cd is called?  I managed to uninstall mine somehow...22:31
cjwatsongreearb: the desktop file is in ubiquity, but casper is what's responsible for getting it onto the desktop22:32
cjwatsonsee scripts/casper-bottom/10adduser in the casper source package22:32
greearbok, I removed a bunch of things...looks like I lost it to a dependency..will try re-adding, or if necessary, start over and be more careful.22:33
alex_jonicjwatson: I just booted the live cd I made :)22:36
alex_joni(this is from the live .. so it works :D)22:36
* alex_joni is happy22:37
greearbcongratz!22:37
greearbhere are my notes.  This worked for me, though I've yet to run it all from a clean install, so I might be missing a few things yet.22:40
greearbhttp://www.candelatech.com/oss/ubuntu-live-notes.txt22:40
greearbfeel free to link to them from any of the wiki pages and/or use this in any manner (including outright copying it somewhere else)22:40
alex_jonigreearb: you missed an 'e' on welcome.22:41
greearbfixed :)22:41
TomJaegerWhat does the kernel team think about the patch attached to bug #124406?23:07
ubotuLaunchpad bug 124406 in linux "Keyboard keys get stuck and repeat (Feisty, Gutsy)" [Unknown,Confirmed] https://launchpad.net/bugs/12440623:07
TomJaegerping23:20
smb_tpTomJaeger: Hi Tom, sorry I did not have much time today to look at it. I also would like to discuss this with the others.23:31
TomJaegercool.  I should mention that patch fixes a different (but nevertheless very annoying) problem than the original report was about.23:35
smb_tpTomJaeger: I am not sure I can promise much. If I look (not very thouroughly, I admit) at the patch it seems to change the scheduler and feels not like something to do that late in teh cycle. 23:41
smb_tps/and feels/and taht feels/23:42
smb_tps/taht/that/ (can't type any more)23:42
TomJaegerThat's why I bisected the issue right when I noticed how I could reproduce it.23:44
TomJaegerThis has been fixed upstream for almost two months now, and it's not like it makes some fundamental changes to the scheduler23:46
cjwatsoncan I suggest that it might be worth considering this for 8.04.1 rather than 8.04?23:47
cjwatsonthen there'll be time to find out if it breaks the world when applied to Ubuntu23:48
cjwatsonassuming that it goes in early in the 8.04.1 period23:48
smb_tpThat would be something I personally would feel better with. But surely I'd like to ask the others, too.23:49

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