[00:00] er, ok..trying again [00:03] this is just a small little package? [00:03] linux-meta-.... ? [00:04] greearb: you need to get 'apt-get source linux-image-2.6.24-16-generic' [00:05] ok, sorry for my confusion...never used debian so much...downloading that now [00:07] amitk: you wouldn't be familiar with the livecd by any chance? [00:07] greearb: it is confusing even to me sometimes :) [00:08] I'm looking for the list of packages which get installed by default.. and I can't seem to remember where that was [00:08] alex_joni: not particularly and definitely not at 2am :) [00:08] amitk: same TZ as here ;) [00:09] alex_joni: you want to catch cjwatson when he is online [00:09] ok, that rules.d looks correct now. It'll take me a few hours to port my patch forward probably... [00:10] greearb: if you have a single patch file it's not that complicated [00:10] I would still create a custom flavour (call it greearb) [00:11] heh, I have a large number of patches to various parts of the kernel [00:11] then look at debian/binary-custom.d [00:11] ok [00:11] create a new folder called greearb ;) and put the patches in there.. [00:11] there's a README which will guide you further [00:12] ok, will look at all that...I first need to get the patches so that they will apply to this kernel. [00:12] (not particularely sure it works with udebs for custom flavours.. but it should afaik) [00:26] hey all [00:26] i was in here this morning asking about squashfs [00:26] i've selected squashfs inside my kernel config [00:27] and i've run make-kpkg --append-to-version -squashfs --revision r1 --initrd --config oldconfig kernel_image [00:27] but... [00:27] when i run update-initramfs -u -v i don't see that it's getting compiled into the kernel [00:27] any clues? [00:28] alex_joni: given a live CD, the simplest way is to look at /casper/filesystem.manifest-desktop [00:28] cjwatson: I just figured that out.. thanks a lot anyway [00:28] btw.. I read about 20 minutes of sources (ubiquity) until I reached that conclusion :D [00:29] alex_joni: of course that's several steps down the chain; in general, it's the dependency-expansion of a bunch of "seeds" [00:29] cjwatson: 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] no, you need to put everything in there [00:29] is it me, or the older manifest-desktop didn't have versions? [00:30] there are ways to automate it (which we use in our own build processes) but they're probably overkill for a one-shot deal [00:30] yeah, sure.. I only have 1 package + dependencies [00:30] alex_joni: it's you; the livefs manifests have had versions since they were created [00:30] (the keyword for the automation in question is "germinate") [00:30] huh.. germinate? sounds fun [00:30] sample dependency-expanded lists of packages for Ubuntu are at http://people.ubuntu.com/~ubuntu-archive/germinate-output/ubuntu.hardy/ [00:31] germinate: vt. to cause seeds to grow [00:32] cjwatson: cool.. probably extends my knowledge absorbing atm [00:32] yeah, it's only if you care about how we maintain it [00:32] very much designed so that ordinary Ubuntu developers can forget about the process and not have to edit lists of packages every time a dependency changes [00:32] cjwatson: I do this once every couple of years/versions [00:33] and each time it's more complex (and automated) [00:33] so I'm not sure if anything I know now will be valid by 10.4 or whatever the next LTS will be :D [00:34] cjwatson: it looks like quite a lot of work [00:35] this particular bit of it has not changed a whole lot since 6.06 [00:35] also, I documented the manifest handling in ubiquity/doc/README a while back [00:36] yeah, manifest-desktop was created while ubiquity was still called espresso, so pre-release dapper [00:38] yeah, I noticed a couple of things are the same since dapper [00:38] (most of the customization part of the LiveCD - squashfs, etc) [00:42] cjwatson: I notice the docs on the wiki tell one to regenerate the filesystem.manifest then cp that as filesystem.manifest-desktop [00:43] so I guess that takes care of things [00:43] alex_joni, I think I have all of that documented, except for the new kernel...I'll be happy to email it to you... [00:45] greearb: feel free if you think I can do anything with it :) [00:46] alex_joni: unwise to cp, would need to cp and remove stuff [00:46] I'm afraid I haven't been able to keep track of what people have been writing in the wiki - that didn't come from me [00:46] there are about 10 wikis, at least, related to this :) [00:46] cjwatson: well.. they do remove ubiquity [00:47] * alex_joni meant https://help.ubuntu.com/community/LiveCDCustomization [00:47] most similar to others...and most slightly wrong it seems, depending on your version and what you are trying to do. [00:47] alex_joni: I would be inclined to manually add the things you want to both, rather than trying to duplicate the list that's removed [00:48] occasionally I do try to dive in and clean stuff out, but labours of Heracles etc. :-) [00:48] cjwatson: how about after doing an apt-get update && upgrade.. which just pulled another 450 packages in? [00:49] * alex_joni doesn't want to think about updating that list by hand [00:50] cjwatson: 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 packages [00:51] alex_joni: in that case diff the manifests beforehand and take account of what's removed [00:51] alex_joni: ubiquity doesn't actually care about the versions in the manifest files - the versions there are just informational [00:51] so if you literally did upgrade and not dist-upgrade, then it won't actually matter [00:51] (except cosmetically) [00:51] heh, ok.. good to know [00:52] btw, whoever did parallel makesquashfs is my hero [00:52] Parallel mksquashfs: Using 2 processors [00:53] yeah, it only takes maybe 10 minutes now :) [00:58] cjwatson: thanks again, I'm off to bed now === smb_away is now known as smb_tp [01:07] alex_joni: yeah, multithreaded squashing is my best friend too [02:45] anybody have any clues about the squashfs stuff? [02:59] like what? [03:05] i'm trying to compile squashfs into a 2.6.20 (ubuntu'ized) kernel [03:05] and it works but it never makes into the ramdisk? [03:05] doing an update-initramfs -u -v doesn't show it in there [03:06] this is for a livecd [03:06] so whenever i boot [03:06] it fails on mounting the root filesystem [03:06] which is obviously squashfs [03:08] i'm using make-kpkg with --initrd [03:08] if that helps at all [03:09] i'm not sure what else to try? [03:09] it seems a big pain to get a kernel updated on a live cd image...I'm trying to do the same myself. [03:10] apparently [03:10] have you gotten it to boot completely? [03:10] I'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 somewhere [03:10] not with a custom kernel, but I did with an updated ubuntu kernel [03:11] ie, something you can get with apt-get install ... [03:11] ahh -- what k version, what ubuntu version? [03:11] 7.10 (er, current stable..whatever that is), and 8.04-beta is what I'm working on now. [03:12] i c [03:12] yea i'm hacking on feisty, which sucks -- the rest of the world has moved on lol [03:13] I'm guessing it will all be similar..but will be more sure when I actually get something working :) [03:14] hah yea [04:33] I 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? [04:36] i talked to someone in here this morning that said udebs were only for the alternate install cd [04:36] well, I'm trying to build an alternate install cd, so that's probably want I want [04:36] oh lol [04:37] in that case this might point you in the right direction [04:37] http://dsplabs.utt.ro/~juve/blog/index.cgi/01147559232 [04:38] yep, found it already..it's for an older release, but some of it might still work [04:39] yea..isn't that what you were hoping for earlier :) === imbrando1 is now known as imbrandon === asac_ is now known as asac [08:08] moin === doko_ is now known as doko === cradek_ is now known as cradek [15:50] Hi. What's the status on bug #124406? [15:50] Launchpad bug 124406 in linux "Keyboard keys get stuck and repeat (Feisty, Gutsy)" [Unknown,Confirmed] https://launchpad.net/bugs/124406 [15:51] more specifically, the issue Helge and I are seeing [15:53] Also, feel free to try and reproduce it, I don't think it's tied to any particular hardware [15:55] short question on linux-meta.. trying to build debs for my custom flavour, but somehow the deps are twisted [15:56] debian/control holds this dependency: linux-headers-$(kernel-abi-version)-rtai [15:57] however, after the package gets generated I see only linux-headers-2.6.24--rtai (missing the abi '16') [15:58] (looks the same for the traditional stuff I didn't touch.. like -rt) [16:02] hmm.. crap.. n/m me.. was missing gawk === elmargol_ is now known as elmargol [16:27] bug #124406 anyone? [16:27] Launchpad bug 124406 in linux "Keyboard keys get stuck and repeat (Feisty, Gutsy)" [Unknown,Confirmed] https://launchpad.net/bugs/124406 [16:34] I should say that there is a patch and it just needs to be applied [17:52] what's a reason for an ABI bump? === evand_ is now known as evand [18:00] the ABI changing [18:03] thom: you have just made my day :D [18:08] thom: funny.. [18:08] is it supposed to happen on released versions (e.g. hardy after it's shipped out) ? [18:09] it can happen, yes [18:10] security fixes can change the module ABI from time to time [18:11] thom'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 change [18:11] cjwatson: I was vaguely familiar with what ABI stands for.. probably my question was bogus :) [18:12] we've had to change the ABI after release in almost every Ubuntu release so far [18:12] gutsy has been lucky so far [18:12] but all of dapper, edgy, feisty have had post-release ABI bumps [18:12] yeah, but in dapper it wasn't that problematic for me [18:13] alex_joni: What are you finding problematic about ABI bumps, post-dapper? [18:13] infinity: having a custom flavour is what makes it problematic [18:13] and having to deal with lum, lrm, lbm and meta packages for all that [18:14] infinity: 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] Automation is key. :) [18:14] infinity: I understand.. but atm I'm doing the automation by hand.. step by step :) [18:15] "Oh look, new kernel"; fetch-sources; apply-local-patches; push-big-red-build-button; go-for-coffee. [18:16] infinity: yeah, but the thing started as a livecd/install environment for our app (which contains kernel modules, and thus kernel-ver dependent) [18:16] so 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 software [19:23] ok, 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. [19:24] I do have some good notes for anyone who wants to do that though.... [19:24] now, my problem is that I need to compile the headers for my custom kernel. [19:25] I got the source: apt-get source linux-ubuntu-modules-2.6.24 [19:25] then attempted to build them: [19:25] akeroot debian/rules binary-arch arch=i386 flavours=foo [19:26] however, 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] Since the debian/control file is regenerated each time, I can't even run 'sed' tricks to force it right. [19:27] suggestions are welcome :) [19:35] greearb: you need debian/rules debian/control first [19:36] but that assumes you changed the needed places [19:36] can you be any more specific? That doesn't make so much sense :) [19:36] control.stub [19:36] you need to add descriptions for you flav [19:37] then you need to add you falvour to at least rules.d/i386.mk:custom_flavours [19:37] and to rules.d/0-common-vars.mk:all_custom_flavours [19:37] ok, will try that in a sec.. [19:38] btw, back on breezy I needed two things [19:38] one is kernel + initrd (you can get those from /boot/ from you regular machine) [19:38] and the other is the udebs which get loaded lateron [19:38] brb [19:39] don't think I need the udebs, seems that is only for non-live CD builds [19:47] cjwatson: any ideas why I would get "/init: .: 159: Can't open /scripts/casper" ? [19:47] (I'll dig for it, but thought you might know without looking..) [19:48] seems control.stub might be auto-generated from control.stub.in ? [19:49] greearb: you're getting warmer :P [19:54] no luck [19:55] building with: fakeroot debian/rules binary-arch arch=i386 flavours=foo [19:55] is that right? [19:56] nope [19:56] I use fakeroot debian/rules custom-binary-arch=rtai iirc [19:56] for the headers as well? [19:57] I'm not building the kernel here...just the headers package [19:57] I use fakeroot debian/rules custom-binary-rtai [19:57] ok [19:57] that builds image & headers too [19:57] brb [19:57] bleh, I mean modules, not headers. I need the cramfs modules and such [19:57] will try poking some more === ogasawara_ is now known as ogasawara [20:08] cjwatson: changing boot option "boot=casper" to "boot=local" makes it go a bit faster .. but it hangs furtheron [20:08] cjwatson: I also get _tons_ of udevd messages so anything usefull is lost somewhere.. [20:15] ahh, by hacking debian/rules to add a fixup script after the debian/control is generated, I seem to be making progress [20:16] got udebs for free, too :) [21:19] got 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] not sure if it's a qemu issue or something else [21:54] cjwatson: probably didn't update initramfs (mkinitramfs -o /initrd.gz 2.6.24-16-rtai) [22:23] alex_joni: you forgot to install casper before running update-initramfs -u? [22:23] (or other mkinitramfs method) [22:28] cjwatson: probably just borked the initramfs (used the install one, not the live-cd generated one) [22:31] oh, don't do that :) [22:31] any idea what the apt-get package for the 'install' desktop icon on the live-cd is called? I managed to uninstall mine somehow... [22:32] greearb: the desktop file is in ubiquity, but casper is what's responsible for getting it onto the desktop [22:32] see scripts/casper-bottom/10adduser in the casper source package [22:33] ok, 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. [22:36] cjwatson: I just booted the live cd I made :) [22:36] (this is from the live .. so it works :D) [22:37] * alex_joni is happy [22:37] congratz! [22:40] here 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] http://www.candelatech.com/oss/ubuntu-live-notes.txt [22:40] feel free to link to them from any of the wiki pages and/or use this in any manner (including outright copying it somewhere else) [22:41] greearb: you missed an 'e' on welcome. [22:41] fixed :) [23:07] What does the kernel team think about the patch attached to bug #124406? [23:07] Launchpad bug 124406 in linux "Keyboard keys get stuck and repeat (Feisty, Gutsy)" [Unknown,Confirmed] https://launchpad.net/bugs/124406 [23:20] ping [23:31] TomJaeger: Hi Tom, sorry I did not have much time today to look at it. I also would like to discuss this with the others. [23:35] cool. I should mention that patch fixes a different (but nevertheless very annoying) problem than the original report was about. [23:41] TomJaeger: 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. [23:42] s/and feels/and taht feels/ [23:42] s/taht/that/ (can't type any more) [23:44] That's why I bisected the issue right when I noticed how I could reproduce it. [23:46] This has been fixed upstream for almost two months now, and it's not like it makes some fundamental changes to the scheduler [23:47] can I suggest that it might be worth considering this for 8.04.1 rather than 8.04? [23:48] then there'll be time to find out if it breaks the world when applied to Ubuntu [23:48] assuming that it goes in early in the 8.04.1 period [23:49] That would be something I personally would feel better with. But surely I'd like to ask the others, too.