[12:05] zulcss@gmail.com [12:06] Forwarded. [12:06] merci [12:09] ABI has changed! Refusing to continue; please update the ABINAME accordingly. Differences [12:10] anyone have access to an amd64 box that i can borrow to test a build? [12:11] yeah [12:11] you need breezy chroot? [12:11] hoary [12:11] hm [12:11] I only have an x86 hoary chroot [12:11] meh.. [12:11] but I can make one for amd64 [12:12] thanks [12:12] i have my key on my launcpad page [12:13] bah, it has to fetch everything [12:13] typical [12:13] will be a few minutes :) [12:13] okie dokie [12:28] zul: I assume that was from a different log, since I see no ABI change in the one I sent you. :) [12:31] no that was in the log [12:32] thats the last line in the email you sent me [12:39] Err, then the mail was cut off... [12:40] ah ok.. [12:40] can you put it on the on public_html or something? [12:42] infinity: nm.. [12:42] Oh? [12:42] Learned how to scroll in gmail? :) [12:42] yes i did! wow.. [12:42] heh [12:42] i need a real email provider [12:43] I can hook you up with POP/IMAP. [12:43] that would be nice [12:43] Mial me (ha, ha, ha, the irony!), and I'll sort it for you when I'm less asleep. [12:44] s/Mial/Mail/ [12:44] ok will do [12:44] crappers [12:44] infinity: been up all night again? [12:44] ajmitch: Yeah. [12:49] zul: Which continent do you live on? [12:51] north america === ajmitch might try out keybuk's bzr ideas [12:51] infinity: about 2 hours away from jbailey [12:52] zul: Ahh, kay, the my mail services should be pretty good for you (Florida and/or Texas, depending on moon phase) [12:52] Better for you than they are for me anyway. ;) [12:52] depending on the moon phase? [12:52] IMAP from Australia to the US is kinda suck. [12:53] yeah...i wouldnt make fun of austrailia right now because im using your box ;) [12:53] heh [12:53] zul: Two colo boxes, in the process of failing over from one to the other and decomissioning the one in Florida. [12:53] feel free to make fun of NZ [12:53] ajmitch: It's hard not to. [12:54] I know, and I live here.. [12:54] infinity: ah i see [01:20] zul: Cheeky email for a man asking for a favour. :) [01:21] infinity: its not running drugs isnt it? :) [01:21] Wow, all those english words, and not a single english sentence to be found. [01:21] Congratulations! [01:21] heh [01:22] :P [01:22] infinity: nothing in my email [01:23] "So hook me up k thx" <-- That's what I was referring to as "cheeky". [01:23] ah...ok.. === jane_ [n=JaneW@dsl-146-167-152.telkomadsl.co.za] has joined #ubuntu-kernel [02:15] infinity: fixed [02:16] it's building ok now? [02:16] yeah its building ok now...keep fingers crossed thugh [02:17] & wait a few hours [02:18] I see it's only using 1 core [02:18] no -j ? [02:18] nah i forgot :( [02:18] ah [02:19] i could start over if you want.. [02:19] zul: A simple 1-liner, I assume? [02:19] zul: whatever works [02:19] infinity: well the first part was, then it was some funky asm shit [02:20] The security updates broke some inline assembly? [02:20] That's worrying... [02:20] no it was me, typos [02:20] Oh, phew. [02:21] i suck sometimes [02:21] We all do. [02:21] Anyone willing to do kernel security gets cut some slack. It's a thankless task. [02:22] i should check breezy as well === ajmitch thanks zul [02:22] you want a breezy chroot as well now? [02:22] please :) but ill let you know [02:23] Well, dapper built on all 6 arches, so BenC gets a gold star. === ajmitch debootstraps [02:24] rob schneider is a tool [02:28] k, got the 2.6.12 orig in place, just waiting for breezy debootstrap to go [02:28] ajmitch: thanks so much [02:29] I should put the breezy one on tmpfs & see how long the compile takes :) [02:29] As core-dev expands further, I think we're going to have to sort out a way to get some porter machines in the DC accessible by non-Canonical developers... [02:30] that would be good [02:30] (Obviously, on a seperate segment from any Canonical-senstive machines, etc, etc) [02:30] it is pretty necessary === ajmitch has access to most official archs, but I'm unusual [02:30] ajmitch: Well, it's all a learning and growth process. [02:30] When this all started, core-dev WAS Canonical staff. [02:30] As that shifts a bit, we need to figure out how to deal. [02:31] (Like the recent change to make the seeds accessible by all of core-dev, and not just staff) [02:31] Etc.. [02:31] ajmitch: well you are a debian dev...unlike us losers non canoical employees ;) [02:31] yeah, it's expected that things will change as you find you need them [02:31] zul: I mean non-debian machines, if I care to upgrade the ppc box downstairs :) [02:31] zul: I'm a Debian dev too, but I almost never use Debian machines. [02:31] ah.. [02:32] infinity: yeah but you are also a canoical empolyee :) [02:32] I use Debian boxes to debug arm, mips, and s390, since those are the only arches I don't have direct access to. [02:32] not many people have direct access to an s390 [02:32] zul: I don't use Canonical porting machines either. [02:32] zul: (Of course, I abuse the buildds directly sometimes, but let's not split hairs) [02:32] nyeah nyeah nyeah :) [02:33] infinity: you have sparc boxes? [02:33] ajmitch: I keep meaning to find one of the rare little deskside ones, but they're A) hard to acquire, and B) a bit pricey still. [02:33] i have sparc boxes though [02:33] (deskside s390, that is) [02:33] ajmitch: I have no sparc at home right now. It left when Daniel moved to Finland. I'll get another. [02:34] i was on the gentoo sparc team for a while === infinity revokes your key from the keyring. [02:34] heh [02:34] heh...i also maintained apache for gentoo :P [02:34] that was thankless [02:35] wasn't sabdfl one of the original debian apache maintainers? [02:36] check the changelogs :) [02:36] There are claims to that effect, though the changelog (which goes back to 1.1.1-1) is missing his name. [02:36] infinity: jelmer's estimate for samba4 & edgy - don't expect much more than a proof of concept [02:37] Hrm, but 1.1.1-1 wasn't the first upload. The source package was called apache-httpd before that. === infinity goes to find it. [02:37] google finds a messages to debian-changes from him [02:38] so the rumours seem true === infinity nods. [02:38] Which version? [02:39] he mentions 1.0.3-2 [02:39] & an old debian-devel thread, with iwj [02:39] Ah-ha. [02:40] That was before the source package format was sanitised. [02:40] Hence the lack of changelog. [02:40] I feel young [02:41] ajmitch: how old are you? [02:41] at the moment? 23 [02:41] you are young [02:42] 24ish next week, but yes, I am young [02:42] i so freaking old [02:42] Anyhow, the version of apache in Debian 1.1 (the first Debian release) was done by Miquel van Smoorenburg. [02:42] And by Debian 1.2, the new packaging was in effect. [02:42] So Mark's apache contributions were never in a Debian release. [02:43] cool.. [02:43] Sucks to be him. :) [02:43] ...this day in history [02:43] (My apache releases, OTOH, have been in a few distro releases now... Odd) [02:43] I guess this means that in ~10 years, I'll be a gazillionaire. [02:44] heh [02:46] http://cia.navi.cx/stats/author/zul?s_message=0 (whee) [02:47] yes at one point i didnt know any better [02:47] but it was just so easy to become a dev [02:49] So, you spent all that time maintaining apache, and you've not given me asingle bug report or patch for the apache packages in Debian/Ubuntu? [02:49] Shame on you! :) [02:49] now you're stuck doing kernel security stuff for ubuntu.. [02:50] i wouldnt say stuck :) [02:51] better you than me :) === zul_ [n=chuck@CPE0006258ec6c1-CM000a73655d0e.cpe.net.cable.rogers.com] has joined #ubuntu-kernel [02:57] zul_: did the kernel build finish, or did you forget to use screen? ;) [02:58] screen...its late ;) [02:58] ah well [04:24] im going to bed the compile is still going ill check in later [04:26] ok [04:27] you want to start the breezy compile now? [04:27] thanks for the helpi appreciate it === tuxmaniac [n=aanjhan@59.92.52.245] has joined #ubuntu-kernel === tuxmaniac [n=aanjhan@59.92.62.63] has joined #ubuntu-kernel === Lure [n=lure@ubuntu/member/lure] has joined #ubuntu-kernel === Lukketto [n=Lukketto@host147-92.pool8261.interbusiness.it] has joined #ubuntu-kernel === orient2000 [n=orient20@CPE0015e96c0cd3-CM0014f85e8fb0.cpe.net.cable.rogers.com] has joined #ubuntu-kernel === orient2000 [n=orient20@CPE0015e96c0cd3-CM0014f85e8fb0.cpe.net.cable.rogers.com] has left #ubuntu-kernel ["Konversation] === lloydinho [n=andreas@rosinante.egmont-kol.dk] has joined #ubuntu-kernel === doko_ [n=doko@dslb-088-073-106-012.pools.arcor-ip.net] has joined #ubuntu-kernel [01:12] hi all. can anybody enlighten me on how i can build the ubuntu kernel for a particular flavour to do some debugging? [01:13] if i cat the config and config.686 files into .config, and make-kpkg, i get a kernel which boots, but has serious module versioning issuse. [01:13] bernard_: move debian/config/$arch/$otherflavours out of the way [01:13] then you do fakeroot make -f debian/rules binary-deb [01:14] fabbione: ahh, ta. that won't do a clean first? [01:14] you can do: [01:14] then you do fakeroot make -f debian/rules clean binary-deb [01:14] (i also noticed the rules file borks if there's no -386 build done) [01:14] it breaks if you do a dpkg-buildpackage [01:15] that's why i didn't use it [01:15] and it "breaks [01:15] and it "breaks [01:15] fabbione: ah, np. i didn't want it to do a clean first. i'll give it a shot, thanks. [01:15] OH CRAP [01:15] and it "breaks" because 386 is expected to be there [01:15] bernard_: you will need to clean.. no matter what [01:16] oh? [01:16] what sources are you using? [01:16] apt-get source linux-source-2.6.15 ? or apt-get install linux-source-2.6.15 ? [01:16] 2.6.15-23 from dapper [01:16] apt-get source [01:17] bernard_: take the time and set up ccache [01:17] ok if you did create a .config and builded in the same source tree, you might as well kill that dir and unpack the source again [01:17] the build system makes a copy of the tree in debian/build/build-$flavour [01:18] bernard_: what alex_joni said is good too :)= [01:18] fabbione: ah, of course. [01:19] alex_joni: shall follow your advice. :) [01:20] it saves a lot of time later === Lathiat [n=lathiat@ubuntu/member/pdpc.basic.lathiat] has joined #ubuntu-kernel [01:23] any advantages when apt-get source vs. apt-get install linux-source ? === Lukketto [n=Lukketto@host147-92.pool8261.interbusiness.it] has joined #ubuntu-kernel [01:25] except the multi-platform stuff (I'm only interested for x86) [01:26] alex_joni: hmm, seems my MAKEFLAGS='"CC=ccache gcc"' isn't being passed to make-kpkg. how do you do it? [01:27] bernard_: use the envvars [01:27] if [ -d /usr/lib/ccache ] ; then [01:27] export PATH=/usr/lib/ccache:"${PATH}" [01:27] export CCACHE_DIR=/usr/src/.ccache [01:27] fi [01:27] i slam this one in .bashrc [01:28] yeah, what fabbione said [01:28] export CCACHE_NLEVELS=8 [01:28] and this one [01:28] still inside the if/fi block [01:28] fabbione: what's the basic difference between apt-get source and apt-get install linux-source? [01:28] but how do you get the kernel builds to use ccache? [01:29] bernard_: start compiling [01:29] bernard_: if you use my snippet, just logout and login again and build [01:29] bernard_: it's transparent [01:29] and it starts using ccache automatically [01:29] alex_joni: the source is the same, the build system isn't [01:29] bernard_: ccache will run the compiler itself, and cache the compiled code on disk, next time (the time you build the kernel again) it will use the cached stuff, it it didn't change [01:29] fabbione: ahh, $PATH... i see :) [01:30] fabbione: ty [01:30] fabbione: debuild vs. make-kpkg ? [01:30] alex_joni: well somehow yes.. [01:30] the install doesn't have a debian dir [01:30] source does [01:30] and you can do more tricks [01:30] right.. ok, thanks [01:30] like building an "official" source [01:31] or let say.. standard [01:31] enabling all the checks [01:31] checks? [01:31] like for the ABI compatibility (important) [01:31] yeah we do some build time checks [01:31] ahh.. ok, stumbled across that when I tried apt-get source [01:31] if you break the ABI you might not be able to load some external modules === Lukketto [n=Lukketto@host147-92.pool8261.interbusiness.it] has left #ubuntu-kernel [] [01:31] so you want to know that before installing the kernel [01:32] specially when you use out-of-tree modules [01:32] indeed.. ok, probably a bit much for me right now .. but good to know where to ask :D [01:34] thanks for your help alex_joni, fabbione. [01:34] no problem [01:36] bernard_: don't give up ;) it might feel like a bumpy road sometimes [01:39] alex_joni: ahh, i'm not easily put off :) [01:40] i did a lot of kernel hacking under debian and make-kpkg was great! i've just switched to ubuntu two days ago and i've been trying to nut out the fancy build system. [01:40] bernard_: doing it on dapper? [01:40] alex_joni: yup [01:40] oh, you mentioned 2.6.15 so probably yes [01:41] there's one small problem I encountered on dapper & make-kpkg [01:41] infinity helped me out on that [01:41] if you apt-get linux-source && make-kpkg then the usplash won't work [01:41] as in apt-get install? [01:42] bernard_: yeah [01:42] it seems that make-kpkg doesn't link vesafb.ko in the initrd dir, which appearantly causes fbcon not to get loaded, which means no usplash [01:43] ahhh, bizarre. [01:43] i found that too, but thought it was related to the symbol version issues i was seeing [01:43] for now I used a post-install script, placed by hand into the debian/ folder [01:44] let me blog the info [01:51] http://dsplabs.utt.ro/~juve/blog/index.cgi/01149939865 [01:52] ta :) [01:52] so it worked under debian/rules but not just make-kpkg? [01:52] np [01:53] I did use make-kpkg, but had to put the script there [01:53] make-kpkg includes the stuff from debian/ in the deb, so it will eventually get executed when you install the deb. [01:54] but there is no debian/ folder when you apt-get install linux-source, it gets created by make-kpkg, so you need to copy the script there while it's running [01:54] and beware that make-kpkg clean will delete debian/, so you'll need to recopy it later === bernard_ nods. === mjg59 [n=mjg59@cavan.codon.org.uk] has joined #ubuntu-kernel === Lukketto [n=Lukketto@host147-92.pool8261.interbusiness.it] has joined #ubuntu-kernel === Lukketto [n=Lukketto@host147-92.pool8261.interbusiness.it] has left #ubuntu-kernel [] === lloydinho [n=andreas@rosinante.egmont-kol.dk] has joined #ubuntu-kernel === tuxmaniac [n=aanjhan@59.92.63.163] has joined #ubuntu-kernel === aanjhan_ [n=aanjhan@59.92.54.180] has joined #ubuntu-kernel === tuxmaniac [n=aanjhan@59.92.54.180] has joined #ubuntu-kernel === lloydinho [n=andreas@rosinante.egmont-kol.dk] has joined #ubuntu-kernel === ivoks [n=ivoks@ubuntu/member/ivoks] has joined #ubuntu-kernel [08:40] next question... where do i bump the ABI version? and does it need to be completely numeric? [08:41] bernard_: if you're using the apt-get source then I think there is a rule to bump the ABI version automagically [08:41] but take this with a pinch of ??? .. I'm not very sure ;) [08:42] hehe. i gathered from the rules file that pulls it from the control file, but i was wondering if there was anything higher up that generated that [08:43] bernard_: I know BenC advised me to set a flag to ignore it once [08:43] ah hah. you're absolutely right - bumpabi :) [08:43] after asking smarter people (Ben Collins), it seems I have an ABI bump, and I needed to run: [08:43] $ echo "Yes" > debian/abi/i386.ignore [08:44] it seems that just ignores the fact they've changed [08:51] alex_joni: are you distributing your kernels? or are they mainly for you? [08:54] bernard_: not only for me [08:54] smallish number of users though [08:55] couple hundred tops I reckon [08:55] how do you version your packages? [08:56] i'm debating with myself the best way to do this still [08:56] make-kpkg --version=foo# [08:57] ahh, you don't use the abi verioning done by the rules file? [08:58] nope, not really [08:58] hmm. perhaps it's not worth the hassle for me either. [08:58] I also don't use the apt-get source package [09:01] zul: ping [09:02] bernard_: bet BenC knows better which way you should go :) [09:02] if you change the ABI, then people will likely need other packages rebuilt against your kernel (like linux-restricted-modules-2.6.15) [09:04] BenC: ah. if i'm building with debian/rules, i should get big fat warnings if the abi has in fact changed, yes? [09:05] bernard_: yeah, it's usually preceeded with a diff of the changed symbols [09:06] BenC: who might have some knowledge about the new dapper CD ? I wonder how hard it would be to change the kernel on it.. [09:06] alex_joni: there's probably a howto somewhere on the wiki [09:07] BenC: there wasn't for breezy, but I eventuall figured it out. but the dapper one looks nothing alike :) .. I'll keep looking [09:10] BenC: cool. is there any recommended way to maintain out-of-distro kernels with various patches (which alter the abi) ? eg, a different package name, or an abinum like 23.1? [09:10] bernard_: I'd got for something like 99.1 [09:10] like 2.6.15-99.1-386 [09:11] that way a new ubuntu kernel wont overwrite/override the custom one [09:13] ah k. will have a ponder. [09:13] thanks BenC [09:21] np