=== svenl_ [~luther@AStrasbourg-251-1-59-131.w82-126.abo.wanadoo.fr] has joined #ubuntu-toolchain [03:58] fabbione: I have the kernel-package change done here locally. I haven't a clue how to test it. =) [03:59] I've tried running the postinst and preinst with the right arguments locally, but I have a feeling that it gets attacked by sed or something on its way into the kernel package. [03:59] Leaving me with the impression that I don't smoke enough weed to do this hack without assistance. [04:05] fabbione: Can you proof the changes I've made? I've touched kernel/image.preinst, kernel/image.postinst and debian/changelog [04:05] fabbione: I've put the files up on chinstrap. If they're all good, can you please use this when uploading -rc6? [05:38] morning [05:38] You're up early. =) [05:38] yeah [05:39] Do you have brain space to look at the kernel-package stuff I put on chinstrap now? [05:40] Otherwise I'm headed off to bed in a moment. =) [05:40] yeah sure [05:40] you could have just uploaded dude :) [05:40] I don't like doing untested uploads. [05:41] it's not like i will get mad at you if you break breezy :) [05:41] True. [05:41] If you'd like, I can debsign and dput. =) [05:41] i am diffing now :) [05:43] 'k [05:43] I've signed my local copy, so if you like I'll dput it. [05:43] go ahead :) [05:43] i will test it to build 12rc6 [05:46] Whoosh, off she goes. [05:46] I've pinged md about module-init-tools, which needs an update for the modalias magic. [05:49] And I think it's now naptime. =) [05:49] g'n =) [05:50] good night jeff! [05:50] cya later === Seveas [~seveas@seveas.demon.nl] has joined #ubuntu-toolchain === chmj [~d3vic3@dumbledore.hbd.com] has joined #ubuntu-toolchain [10:06] elmo: ping? === ajmitch [~ajmitch@port162-41.ubs.maxnet.co.nz] has joined #ubuntu-toolchain [10:32] fabbione: sup [10:32] elmo: hey [10:33] if you have a few minutes, i would like to get the latest kernel-package in breezy-i386 on concordia and ccache on breezy/halley :) [10:33] be aware to install k-p only on i386 [10:33] i need to create images that i can test locally [10:34] halley done, b-i386 is trying... [10:35] yeah i guess i am spinning corcodia's CPUs a bit :) [10:37] thanks a lot dude :) [10:38] np [10:39] (finished) [10:39] perfect === \sh [~shermann@server3.servereyes.de] has joined #ubuntu-toolchain === Seveas [~seveas@ksl403-uva-154.wireless.uva.nl] has joined #ubuntu-toolchain [01:34] g'm [01:35] jeff :) [01:35] Heya Tollef. Happy Birthday. [01:36] hey jbailey [01:36] jbailey: i just finished the build on i386 with the new kernel-package [01:36] Cool! [01:36] downloading the image right now for testing :) [01:39] fabbione: If you could do two tests for me when you've got it? [01:39] Setting up linux-image-2.6.12-1-686 (2.6.11.94-1.1) ... [01:39] Aside from the regular install, try setting ramdisk = foo, and ramdisk = /usr/sbin/mkinitrd in kernel-img.conf [01:39] Failed to create initrd image. [01:39] dpkg: error processing linux-image-2.6.12-1-686 (--install): [01:39] subprocess post-installation script returned error exit status 2 [01:39] Errors were encountered while processing: [01:39] linux-image-2.6.12-1-686 [01:40] jbailey: thanks. :) [01:40] so there is something wrong in postinstall === doko_ [~doko___@dsl-084-059-045-032.arcor-ip.net] has joined #ubuntu-toolchain [01:40] fabbione: Ugh. [01:40] What's the perl equivalent of sh -x ? [01:40] jbailey: it doesn't have any, sadly. [01:40] Yar. [01:41] it has a built-in debugger, though. [01:41] I wish it'll grow a -x some day. [01:41] jbailey: i think i know what is wrong.. just a few secs [01:46] Mithrandir: yes it does [01:47] Kamion: it does? What's the magic for that? [01:48] PERLDB_OPTS='NonStop AutoTrace' perl -d [01:48] oooh, shiny. [01:48] LineInfo=foo to output trace to foo [01:48] see perldebug(1) [01:48] Creepy. =) [01:49] Denis Ritchie said the debugging was twice as hard as coding... And perl is the reference implementation! [01:49] jbailey: white spaces are not an opinion :))))) [01:49] my $ret = system($ramdisk . [01:49] this is your original line [01:49] but it calls a command that doesn't exist [01:49] - my $ret = system("mkinitrd " . [01:49] + my $ret = system($ramdisk . [01:50] note the space AFTER mkinitrd [01:50] my $ret = system($ramdisk . " " . [01:50] Oh, *lol* [01:50] this will ensure a space after the command :) [01:50] and works [01:50] at least... [01:50] let me reboot to see if it really works :) [01:50] or just use system(LIST) rather than system(EXPR) [01:50] fabbione: =) [01:50] although that wouldn't work if $ramdisk can have both command and arguments, I guess [01:51] Kamion: that's Manoj's code :))) [01:51] Kamion: and yes.. there are options after [01:51] ok [01:51] Kamion: It can't, thankfully. =) [01:52] The variable gets ( -e ) 'd in the preinst as a safety check. [01:52] jbailey: so what's the second test you want me to do? [01:52] fabbione: In kernel-img.conf try it with "ramdisk = foo", which should fail the preinst, and "ramdisk = /usr/sbin/mkinitrd" which is just overriding the default with the default but should work. [01:53] jbailey: sure [01:54] You are attempting to install an initrd kernel image (version [01:54] 2.6.12-1-686) but you do not seem to have a mkinitrd command in the [01:54] path. This will break the installation, unless initrd-tools are also [01:54] being installed right now. [01:54] Could not find /bin/foo in path. at /var/lib/dpkg/tmp.ci/preinst line 190. [01:54] The directory /lib/modules/2.6.12-1-686 still exists. Continuing as directed. [01:55] but it doesn't die at preinsta [01:55] clearly you might be bootstrapping the system [01:55] and since the kernel Depends: on mkinitrd [01:55] apt will ensure that mkinitrd is configured before configuring the kernel [01:55] What's the point of the check if it does die? [01:55] it will fail at a later stage in postinst [01:55] doesn't [01:56] that was ramdisk = /bin/foo [01:56] so the preinstall doesn't and actually can't fail [01:56] for the above reasons [01:57] i need to do another test... [02:01] jbailey: it seems to work fine [02:01] with ramdisk = /foo and ramdisk = /usr/sbin/mkinitrd [02:01] and no ramdisk [02:01] (given the little fix in the postinst) [02:01] i am upload kernel-package to fix that error [02:02] Nice, thanks. =) [02:05] * Bump Build-Deps on kernel-package 0.135ubuntu4 to include new postinst [02:05] script to support ramfs. [02:05] does it look sane? ;) [02:06] I'd say "ramfs variable in kernel-img.conf" [02:06] Otherwise I'd expect the gentoo kiddys to come demanding what this new ramfs thing we have is. [02:07] jbailey: perfect :) [02:07] that's what we want [02:07] people being curious of what we do [02:07] and take them out of the dark side of the force [02:08] a'ight. As you see fit. [02:08] At least your name is on the changlog entry. =) [02:08] ahah [02:13] "Add a simple cosmetic change"? dude. :P [02:14] Kamion: i didn't want to make jbailey look stupid for a missing "space" [02:14] come on.. he maintains glibc kernel-headers and half of the toolchain :) [02:15] Kamion: It is simple. It appears to work, versus not appearing to work. Isn't that cosmetic? =) [02:15] if it was somebody else i would have take out the sodomotron and larted him to death :))) [02:16] *lol* [02:16] I'm not sure "LART" is the right verb to use with the sodomotron [02:16] Kamion: today you really want to take out the best of me, don't you? ;) [02:16] unless you're using it in an unconventional way [02:16] i am just trying to be nice :) [02:16] :-) [02:17] My version of a lart has always been a baseball bat with a rusty nail in the end. [02:17] But it always could be a sodomotron, I guess. [02:17] Kamion: admit it! you love when i start yelling at people: "I WILL STICK MY ARM IN YOUR ASS AND SHAKE YOU A FLAG IN THE WIND!" [02:18] fabbione: Do you do that to get them excited or scare them? [02:18] I ask only for information. [02:18] haha [02:18] to scare them :) [02:19] elmo: i have the source for you... [02:19] elmo: my ~ on concordia amd64/linux-source-2.6.12_2.6.11.94-1.1_all.deb [02:20] you should be able to build your ppc64 custom kernel out of it [02:20] the configs are still there if you want to look at them [02:20] davis is still building, so i would appreciate if you can wait to reboot it [02:23] i think i killed Kamion [02:25] We need shirts for the. [02:25] "OMG! I kill Kamion!" [02:25] +ed [02:25] fabbione: love it> yes, it makes me horny baby [02:26] jbailey: I'm a traditional guy, I prefer a two-by-four [02:31] heheheh [03:07] fabbione: for ppc64? [03:09] elmo: yes [03:09] the source is _all :) [03:09] configs are on davis in my ~ if you need them [03:09] the debs will be there soon for the stock kernel === doko [~doko___@dsl-084-059-045-032.arcor-ip.net] has joined #ubuntu-toolchain === doko [~doko___@dsl-084-059-045-032.arcor-ip.net] has joined #ubuntu-toolchain [03:48] elmo: time to check nagios :) [03:54] jbailey: ping? [03:54] pong [03:54] Accepted linux-source-2.6.12 2.6.11.94-1.1 (source) [03:54] have fun :) [03:55] Nice. Have you reduced the ppc kernel flavours to something less than 10 hours of build time yet? =) [03:55] jbailey: nope... [03:55] i need a bit more people testing ppc64 before i can kill power3* power4* [03:56] i only need to give the debs to elmo [03:56] nad i am off :) [03:56] enjoy =) [04:19] elmo: linux-image-2.6.12-1-powerpc64-smp_2.6.11.94-1.1_powerpc.deb [04:20] it's in my home dir on davis [04:20] want me to test that? [04:20] jbailey: you wait for the one builded in the archive [04:20] elmo: yes please [04:20] fabbione: Yes, dear. [04:20] OH CRAP [04:20] no [04:20] i did build it in breezy [04:20] i am not sure you can have all the Depends: in hoary [04:21] jbailey: it's because i cheated on davis just to check the build. it has the old postinst [04:21] elmo: sorry.. i fucked that up.. i will prepare an image for you tomorrow [04:21] You are attempting to install an initrd kernel image (version [04:21] 2.6.12-1-powerpc64-smp) but you do not seem to have a mkinitrd command in the [04:21] path. This will break the installation, unless initrd-tools are also [04:21] being installed right now. [04:21] fabbione: ok, no prob [04:21] Could not find mkinitrd in path. at /var/lib/dpkg/tmp.ci/preinst line 189. [04:21] elmo: i guess you dist-upgraded all the chroot?? [04:21] hum? [04:22] no, I was installing that into base [04:22] because that's the error coming from kernel-package ubuntu3 [04:22] hmmm [04:22] it should have told you that you need initrd-tools installed [04:22] it did afterwards :) [04:23] linux-image-2.6.12-1-powerpc64-smp depends on initrd-tools (>= 0.1.78ubuntu1); however: [04:23] Package initrd-tools is not installed. [04:23] Depends: initrd-tools (>= 0.1.78ubuntu1), coreutils | fileutils (>= 4.0), module-init-tools (>= 0.9.13) [04:24] right, but that message comes from the preinst, I was using just plain dpkg [04:24] elmo: just purge and reinstall the kernel [04:24] yeah [04:24] that message should be there only if mkinitrd is not in path [04:24] don't worry I was always keeping the old good one ;) [04:24] and it should still recover if mkinitrd tools is installed later [04:27] elmo: did you manage to install it? [04:27] nah, needs >> hoary's initrd-tools [04:27] ok gimme a sec [04:30] dpkg-deb -b . linux-image-2.6.12-1-powerpc64-smp.deb [04:30] dpkg-deb: building package `linux-image-2.6.12-1-powerpc64-smp' in `linux-image-2.6.12-1-powerpc64-smp.deb'. [04:30] *DIRTY HACK ALLERT* [04:30] davis:~/blob/ [04:30] try that one [04:31] i hacked the control file to remove the versioned Depends: === lamont__ [~lamont@15.238.5.202] has joined #ubuntu-toolchain [04:32] i really really need to stop [04:32] hi lamont [04:33] cya later or tomorrow guys [04:33] elmo: good luck with that kernel :) [04:34] fabbione: hi [04:40] cat: /proc/modules: No such file or directory [04:40] cat: /proc/modules: No such file or directory [04:40] warning: gids truncated to 8 bits (this may be a security concern) [04:40] during postinst configure [04:58] duh [04:58] that's not happy [05:45] james@davis:~$ cat /proc/cpuinfo | grep -c ^processor [05:45] 2 [05:45] james@davis:~$ uname -a [05:45] Linux davis 2.6.12-1-powerpc64-smp #1 SMP Wed Jun 8 13:16:35 UTC 2005 ppc64 GNU/Linux [05:45] \o/ [05:46] |o| [05:46] /o/ [05:46] \o\ [05:46] what the fuck is the second one? [05:46] A tie fighter. [05:46] elmo: nice :-) [05:46] either you've been chopped off at the elbows, or that's a TIE fighter === daniels kicks infinity. [05:46] it's me waving my arms around, you freaks [05:46] elmo: pipe down, stumpy [05:47] WOAH [05:47] unfortunately somebody did remove powerpc64 support from the breezygoallist ... [05:47] time stamped dmesg, neat [05:47] daniels : Oh, it was a zucchini, by the way. I just remembered. Speaking of freaks. [05:48] Time stamps on dmesg?.. When did that happen? [05:48] ehh, davis is a dual processor machine? nice [05:48] [ 17.386107] PowerMac G5 Thermal control driver 1.2b2 [05:48] doko: yes, but smp didn't work with 32-bit :P [05:48] infinity: ^-- [05:49] infinity: uh, what? [05:49] (nevermind) [05:49] elmo : DO I get shiny new kernels on the ppc buildds at some point now? :) [05:49] time-stamped dmesg is in .11, IIRC [05:50] daniels : Is it a config option?.. Cause I'm running .11.11, with no timestamps. === infinity feels so un-1337. [05:50] infinity: sure, once fabbione uploads 2.6.12 final, I'll compile my own and upgrade them [05:50] infinity: 'IIRC' [05:51] elmo : I suppose we don't fear that anyone's abusing 'uname -m' and build will suddenly start to blow up? [05:51] s/build/builds/ [05:53] infinity: well, that's what breezy-test is for I guess [05:53] we can linux32 the builds, if you want [05:53] I just might. Warn me before you upgrade the kernels. [05:54] (I've run into issues where sparc64 hosts still needs to be linux32'd on some builds too) [05:54] Though, I'd hope those have all been fixed by now. [05:54] why? sparc buildds have linux32-ed since day 0 [05:54] the worst one is builds that break due to the presence of /lib64 [05:54] Yeah, I know the buildds do. [05:54] But I've done by-hand builds and forgot to do so. :) [05:54] Most turn out okay, some get hooped by lib64 and/or uname -m [05:57] elmo: what are the units on those timestamps? [05:58] jiffies, innit? [05:59] how ... helpful [06:00] a jiffie is like a bajillionteenth of a second [06:00] approximately one unit of doogie's attention span [06:01] *choke* [06:31] hey.. [06:31] would the ability to upload packages purely to be test built be helpful? [06:32] I thought that was a launchpad thing [06:32] the world is a launchpad thing [06:32] re [06:32] elmo: rocking!!! [06:33] but I'm doing breezy test anyway and we're not using it to rebuild the world, it'd be trivial to add yet another suite to the test archive [06:33] elmo: and yes.. there is an option in .12 for timestamp printk [06:33] or heck even just use breezy-test itself [06:33] Kamion: i think the timestamp is done seconds from boot [06:33] Kamion: still better than nothing tho :) [06:34] doko: you joking that ppc64 has been taken out of breezy goals???? [06:34] doko: we got toolchain and kernel.. we need 2 libs and it's there [06:35] elmo : Allowing random uploads to a test archive could be very handy, yes. [06:35] Linux davis 2.6.12-1-powerpc64-smp #1 SMP Wed Jun 8 13:16:35 UTC 2005 ppc64 GNU/Linux === fabbione goes and masturbates [06:36] BRAIN FLOSS [06:36] I doubt you'll ever get that one out. [06:36] NOW, DAMNIT [06:36] elmo: mind if i spin davis for a test??? [06:37] fabbione: well, it was a hoary goal, but maybe it didn't get added to breezy [06:37] doko: well we did it basically [06:37] so i mean [06:37] it's pointless to stop NOW [06:37] we just add it again ;) === fabbione gets ready to spin davis [06:39] hmm, what is wrong, if ifup/ifdown hangs forever? [06:40] elmo: btw .12 final should be out next week... so we won't have to wait for too long [06:40] fabbione, which libs are missing? [06:40] ncurses at least [06:41] doko: basically you can just check the sparc Packages and see which libs are both 23/64 bit [06:41] the same are missing on ppc64 :) [06:41] 23-bit? weird [06:41] yeah well :) [06:41] 32 [06:42] heh, s390 has 31 [06:42] elmo: breezy-test> yes, very [06:42] elmo: sshing into all the machines, screen'ing, dchroot'ing, etc, and then checking back is a pain in the arse [06:42] elmo: double points if all the build logs get emailed to the uploader automatically :) [06:42] elmo: since when davis has 3GB of RAM? [06:43] fabbione, elmo plugged another CPU as well ;) [06:43] fabbione: it should have 4 [06:43] where is the other one? [06:44] our porting machines are all meant to be dual proc, 4 Gb memory [06:44] doko: eehhe [06:44] fabbione: dunno [06:44] fabbione: it's trying to migrate to concordia [06:44] I can't remember what 2.6.10 did [06:44] because davis isn't cool enough [06:44] elmo: hmm it would be interesting to compare the 2 dmesg from 2.6.10 and 2.6.12 [06:44] elmo: because jbailey noticed a few interesting things in the diff [06:45] but it's definetely going MUCH FASTER [06:45] fabbione: ~james/kern.log* [06:45] kern.log.0 has a 2.6.10 one by the look of it [06:45] fabbione: but critically IS IT CRASHING? :) [06:46] Jun 8 16:17:21 davis kernel: [ 0.000000] Memory hole size: 2048MB [06:46] elmo: not yet.. i just started building the kernel... [06:46] Jun 8 16:17:21 davis kernel: [ 17.212083] Memory: 2972400k/5242880k available (2672k kernel code, 2269804k reserved, 1308k data, 278k bss, 212k [06:46] init) [06:46] Jun 8 16:17:21 davis kernel: [ 17.212238] Calibrating delay loop... 66.56 BogoMIPS (lpj=33280) [06:47] you got the same intersting values as jbailey [06:47] he has a 4GB machine [06:47] and it was reporting 4GB/6GB [06:47] lemme, go check the invoice, but I'm 99% sure it's got 4Gb [06:47] still 66.56 Bogo [06:47] only 66.56?? dude my laptop is faster than davis wtf === elmo hands daniels the glxgears-as-a-benchmark award [06:48] May 14 09:47:41 davis kernel: Memory: 2071296k available (1768k kernel code, 916k data, 364k init, 1310720k highmem) [06:48] i bet it scores higher on glxgears too!! [06:49] see 2.6.10 was able to see only 2GB === fabbione larts daniels with a sodomotron [06:49] actually, sorry my bad [06:49] it appears it really only has 2Gb [06:49] so, err. [06:49] wtf? [06:49] linux appears to be very confused [06:50] shall I down the box and check physically when you're done building? [06:50] are you at the datacenter? [06:50] yes :-( [06:50] IBM ARE TEH SUCK [06:50] i can stop now [06:50] it really fast :) [06:50] ok i am logged out [06:50] i don't want to keep you there forever [06:51] and i can run the build overnight :) [06:51] I can't imagine the kernel would just make up an extra gig of RAM. [06:51] infinity: exactly [06:51] Well, not without crashing pretty damn soon after boot. [06:51] (As soon as it tried to map something in the nonexistant RAM) [06:52] On the other hand, PPC kernels since the dawn of time have been infamous for not finding all the RAM lying around in a system. [06:52] So my bet would be on 4 gig real, 3 detected. [06:52] infinity: yes that's what i think [06:53] jbailey: you should really check the ram on your machine too [06:53] (It could just be 1024+1024+512+512, though, and we're all speculating on a bug that isn't there) [06:53] infinity: right [06:53] that's why elmo is shutting down the machine :) [06:54] Heh. [06:54] hum, how very confusing [06:55] it appears to have 1+1+.5+.5 [06:55] *infinity wins* === infinity ^5s fabbione. [06:56] gypped [06:56] Or, bonus. If the invoice only says 1+1. [06:57] oh, look, a bed [06:57] Yeah, I keep saying I'm heading off to one of those. [06:57] I'll try again. === infinity -> bed, again. [06:57] ehhehe [06:57] but, err, we paid for 2 [06:57] score. [06:57] ++ [06:58] elmo: you should scare the shit out of Mark [06:58] send him an email: "Fabbione's kernels add a GB of RAM on our PPC servers!!!" [06:58] haha [06:59] "... you slack-jawed hippy" [07:00] brb [07:07] elmo: i am going to spin davis to death :) [07:07] let's see how it behaves [07:09] let's start warming it up [07:09] -j100 [07:14] top - 18:14:22 up 13 min, 2 users, load average: 27.38, 16.02, 6.70 [07:14] it doesn't go higher than that! [07:14] the scheduler is really really nive [07:14] nice [07:25] elmo: i will let you know how it goes... [07:25] right now it seems as fast as concordia :) [07:25] i will have to time them [07:25]