=== ubuntulog [~warthylog@port49.ds1-van.adsl.cybercity.dk] has joined #ubuntu-kernel === Topic for #ubuntu-kernel: Ubuntu kernel development discussion | http://www.ubuntulinux.org/wiki/KernelTeam | There are no kernel bugs.. only broken hardware | http://people.u.c/~lamont/Archives/kernel-team@ubuntu.com--2005/ playground: kernel-debian--pre1,2--2.6.11.94 === Topic (#ubuntu-kernel): set by zul at Wed Jun 8 22:27:54 2005 === #ubuntu-kernel [freenode-info] please register your nickname...don't forget to auto-identify! http://freenode.net/faq.shtml#nicksetup === zul [~chuck@CPE0006258ec6c2-CM000a73655d0e.cpe.net.cable.rogers.com] has joined #ubuntu-kernel === sysrq [~sysrq@68-190-222-2.mpk-mres.charterpipeline.net] has joined #ubuntu-kernel [04:13] fabbione: Is there an easy way to generate the modules.alias file based on a subset of modules that I copy to the initramfs? [04:58] one down one more to go [04:58] Nice! [05:00] and it looks like im going to get a call back as well [05:03] Nice! === jbailey attempts to actually get sleep tonight now that the weather's finally broken. [05:09] We're down to 18 here. [05:10] how is the smog? :) [05:10] same here [05:10] Three days of reasonbly heavy rain at some point during the day have worked some magic. [05:11] But I could really wish that people would just stop driving their cars. [05:11] Oh well, sleep time. [05:11] then people in toronto would die [05:22] morning [05:23] mdz: i did ask elmo to remove 2.6.11 even before hoary release... [05:23] mdz: and asked after another few times [05:25] jbailey: i think you can do that using depmod with proper parameters... [05:25] jbailey: iirc it accepts a module dir to look into [05:28] hm interesting... busybox-cvs is FTBFS on sparc with a really nice error [05:29] /usr/lib/gcc/sparc-linux/4.0.1/../../../../lib/libc.a: could not read symbols: Bad value [05:29] this looks like either libc or gcc are doomed === zul [~chuck@CPE0006258ec6c2-CM000a73655d0e.cpe.net.cable.rogers.com] has joined #ubuntu-kernel === doko [~doko___@dsl-082-082-201-214.arcor-ip.net] has joined #ubuntu-kernel === lamont [~lamont@mix.mmjgroup.com] has joined #ubuntu-kernel === mjg59 [mjg59@cavan.codon.org.uk] has joined #ubuntu-kernel === calc [~ccheney@ip70-185-4-246.ma.dl.cox.net] has joined #ubuntu-kernel === Seveas [~seveas@ksl403-uva-153.wireless.uva.nl] has joined #ubuntu-kernel === amu [~amu@amu.developer.debian] has joined #ubuntu-kernel [12:04] hey amy [12:04] amu [12:05] hey === warthylog [~warthylog@port49.ds1-van.adsl.cybercity.dk] has joined #ubuntu-kernel === Topic for #ubuntu-kernel: Ubuntu kernel development discussion | http://www.ubuntulinux.org/wiki/KernelTeam | There are no kernel bugs.. only broken hardware | http://people.u.c/~lamont/Archives/kernel-team@ubuntu.com--2005/ playground: kernel-debian--pre1,2--2.6.11.94 === Topic (#ubuntu-kernel): set by zul at Wed Jun 8 22:27:54 2005 [12:33] c ya later [12:33] got another at 9 [12:35] zul: good luck dude === jbailey [~jbailey@CPE00501836c657-CM014260028338.cpe.net.cable.rogers.com] has joined #ubuntu-kernel === doko [~doko___@dsl-084-059-062-132.arcor-ip.net] has joined #ubuntu-kernel [01:07] fabbione: No, that's the error that is fixed with the newest glibc upload. [01:20] jbailey: sorry.. what error? [01:20] hm interesting... busybox-cvs is FTBFS on sparc with a really nice error [01:20] /usr/lib/gcc/sparc-linux/4.0.1/../../../../lib/libc.a: could not read symbols: Bad value [01:20] ah ok :) [01:20] But no, I'm mistaken [01:20] That's not the same error I was thinking of. [01:21] Could be solved with the same fix, though, who knows. [01:21] ehe ok [01:21] i will see once glibc are done [01:21] I should check on that. [01:22] we will see in the test chroot [01:22] with the new libc [01:31] Ayup [01:55] i so much hate wearing ties [01:58] Ahaha [01:58] Will you need it in the new job? [01:58] zul: how did it go? [01:59] just about to leave...yesterdays was pretty good they wanted a "portfolio" and definite second interview [02:00] nice [02:01] "portfolio"? === fabbione sighs [02:01] Are you applying to go painting? [02:01] nic-* udebs are driving me stupid === svenl_ [~luther@AStrasbourg-251-1-73-215.w82-126.abo.wanadoo.fr] has joined #ubuntu-kernel [03:22] hey lamont === JaneW [~JaneW@155.239.185.195] has joined #ubuntu-kernel === JaneW [~JaneW@155.239.185.195] has left #ubuntu-kernel ["Leaving"] [03:47] oh headache.. [03:55] zul: eh.... [03:55] it went ok...they asked for references [03:56] good [04:01] zul: what are you applying for? [04:01] desktop support/network admin [04:01] part time--junior, need a job [04:01] ah [04:01] thom: do you mind to do an ipw2100 test for me please? [04:01] sure [04:02] thom: remove the following modules from the stock kernel: [04:02] ipw2100 [04:02] i need a break..bbiab [04:02] (sec.. i can't find the full list :() [04:02] heh, no worries [04:02] i'll need to reboot to .12 anyway [04:03] -rw-r--r-- root/root 28937 2005-06-08 16:43:25 ./lib/modules/2.6.12-1-386/kernel/drivers/net/wireless/ieee80211/ieee80211.ko [04:03] -rw-r--r-- root/root 7872 2005-06-08 16:43:25 ./lib/modules/2.6.12-1-386/kernel/drivers/net/wireless/ieee80211/ieee80211_crypt.ko [04:03] -rw-r--r-- root/root 8561 2005-06-08 16:43:25 ./lib/modules/2.6.12-1-386/kernel/drivers/net/wireless/ieee80211/ieee80211_crypt_ccmp.ko [04:03] -rw-r--r-- root/root 11558 2005-06-08 16:43:25 ./lib/modules/2.6.12-1-386/kernel/drivers/net/wireless/ieee80211/ieee80211_crypt_tkip.ko [04:03] -rw-r--r-- root/root 6536 2005-06-08 16:43:25 ./lib/modules/2.6.12-1-386/kernel/drivers/net/wireless/ieee80211/ieee80211_crypt_wep.ko [04:03] drwxr-xr-x root/root 0 2005-06-08 16:43:25 ./lib/modules/2.6.12-1-386/kernel/drivers/net/wireless/ipw2100/ [04:03] -rw-r--r-- root/root 8522 2005-06-08 16:43:25 ./lib/modules/2.6.12-1-386/kernel/drivers/net/wireless/ipw2100/fsam7400.ko [04:03] there [04:03] remove this modules for real [04:03] and try to compile ipw2100 from upstream source [04:03] that will install the olds ieee stuff [04:03] ok [04:04] and the same ipw2100 driver [04:04] if you can test that and tell me if it works or not [04:04] i might be able to compile ipw2100 in a different way to make it working again [04:12] fabbione: no amd64 2.6.12 kernels in the archive? [04:12] doko: are you nuts? [04:12] they are in universe [04:12] like all the others :) [04:12] ok, don't expect dselect be up to date after an apt-get update ... [04:13] doko: dselect update != apt-get update === lamont__ [~lamont@15.238.7.83] has joined #ubuntu-kernel [04:32] fabbione: when you and i arent busy do you want to talk about the linux-non-supporte-modules thingy? === Seveas [~seveas@seveas.demon.nl] has joined #ubuntu-kernel [06:08] zul: you can bet on it.. we are just in a slightly different TZ [06:50] fabbione: it is starting to cause problems that 2.6.12 is not in main [06:51] fabbione: ndiswrapper-utils is uninstallable, and klibc is unbuildable [07:17] fabbione: maybe it's time to upload -1.2, eh? [07:23] anastacia has been wanting to pull it into main for a while now, and I've been ignoring it [07:25] find: warning: you have specified the -mindepth option after a non-option argument -type, but options are not positional (-mindepth affects tests specified before it as well as those specified after it). Please specify options before other arguments. === lamont__ wonders wth that actually means [07:35] fabbione: I'd like us to upload -1.2 sometime soon, not just because 94-1.1 is ftbfs on hppa [07:44] mdz: it's ok with me to move it main. I am almost done with cleaning the udeb mess. [07:44] mdz: and if you want you can move it now [07:44] lamont__: i would really love to complete the cleaning before 1.2. i miss only a few sets of udebs [07:44] (nic-* pcmcia-* scsi-*) [07:45] woot [07:45] lamont__: did you see the changelog? [07:45] didn't you? [07:45] fabbione: I was meaning sometime this week, not sometime this day... [07:45] i should be able to upload tomorrow [07:45] i need to check with pitti how much time i have to prepare another security update for hoary [07:46] mdz: or can we wait tomorrow after 1.2 is in? [07:46] mdz: otherwise mostlikely i will have to bump the abi at the first upload [07:46] mdz: that's kind of annoying for one day [07:46] fabbione: tomorrow is fine, just notify me when you are ready [07:46] mdz: as soon as i upload 1.2 [07:46] mdz: there is really a bit cleanup to make it ready for main :) [07:46] fabbione: if jbailey needs to make a klibc upload, you must answer to him though :-) === jbailey scrolls back. [07:47] mdz: ok :) [07:47] there are only 128 lines of almost detailed changelog :) [07:48] and i started to be less anal retentive than before [07:48] jbailey: klibc (main) build-deps on a 2.6.12 package (universe) [07:48] jbailey: btw.. klibc is FTBFS on sparc dude :) [07:48] Eh, when did klibc get promoted to main? =) [07:48] jbailey: if you want .12 in main.. we need to get that fixed :P [07:49] === jbailey ssh's [07:49] jbailey: i am setting up something for that... [07:49] that you will access ala people. [07:49] Yay! [07:50] Then I will start asking you guys to have them all in one place. =) [07:50] But first things first.. =) [07:50] let me ask elmo if he had any extra tought about it [07:51] fabbione: if you want to drive getting a process that elmo is happy with, I'd love to see it - any implementation changes I need to make to do the merge, etc. [07:51] err.. can't find umul. Unsigned multiply instruction? [07:53] lamont__: yup... [07:54] fabbione: thans [08:00] lamont__: did you read on #ubuntu-toolchain ? [09:35] damn im popular [09:37] jbailey: how goes the modalias stuff? [09:43] fabbione: i had a couple of ideas i was playing with this week but whenever you are free [10:06] zul: i will be free in 5 minutes [10:06] wife is in bed (finally) [10:12] zul: so.. what's up? [10:32] so i was thinking setup similar to d-i [10:33] i was looking through the drivers and some of them have ./Configure ./configure make and whatnot [10:33] so we would have a flat text file with a flag saying 0 for make 1 for whatever and 2 for whatever [10:33] or something like that [10:33] thoughts? [10:35] or we could script it in kernel-wedge [10:35] thats another option as well.. [10:36] we juat have to make as easy as possible to maintian [10:36] you already lost me :) [10:36] let me finish one thing i started while you were away ;) [10:36] heh i have a tendency to do that [10:37] bleah..:) [10:39] i just want it easy to maintain thats pretty much it right now [10:48] gotta pick up wife bbl [10:48] sorry [10:56] zul: don't worry.. [11:23] ok im back [11:31] ok [11:32] so explain what you are talking about... [11:32] step 1) [11:32] 1) figure out what drivers are we going to support [11:32] #1 we already agreed on that [11:32] 2) make sure that the modules build individually [11:33] #2 what do you mean individually? [11:33] make sure that there is no surprises when we are compiling them [11:33] ok they need to build... that's clear :) [11:34] yep [11:34] 3) add linux-non-supported-modues-* to debian/control debian/control.stub [11:34] yes [11:35] 4) add drivers-build, drivers-install to debian/rules [11:35] 5) profit $$$ [11:35] #4 ????? [11:35] what i don't understand is: [11:35] 1) why do you want to build them outside the kernel [11:36] 4) those are just arbirtary names [11:36] 2) adding an extra lever of complexity [11:36] what we need to do is way more simple than that [11:36] 1) how would you do that? [11:36] we already compile the drivers [11:36] they are already there [11:36] given your #1 create the list [11:37] match against this list and move the drivers from build-$flavour into debian/linux-non-supported-$ver-$abi-$foo-$bar-$mysyster-$yourdad [11:37] and create the debs... [11:38] (defined in debian/control and .stub) [11:38] it sounds to me you created a much bigger issue that what needs to be done [11:39] what lands in there is simple to control and doesn't need extra complexity [11:39] i was kind of thinking that so something like d-i right? [11:39] just one step before install [11:39] d-i is something else.. [11:39] ok [11:39] ill get d-i out of my head then [11:39] more levels we add, more complex it becomes to maintain [11:39] creating udebs is another story and needs other stuff [11:39] so something like the abi check? [11:39] that's already done at kernel build time [11:40] think in this way: [11:40] or better [11:40] look at debian/rules [11:40] more or less in sequence happens: [11:40] unpack [11:40] patch [11:40] configure [11:40] build [11:40] (****) [11:40] install [11:40] create debs [11:40] create udebs [11:41] now [11:41] build also includes the abi check [11:41] what we need to do is somehow: [11:41] add that (****) [11:42] that will move the non-supported-modules according to the list [11:42] from debian/build/build-$flavour/ [11:42] to debian/l-n-s-m-foo-etc/ [11:42] that's all the hook needs to do [11:42] ah i think i see [11:43] it's pointless to duplicate build mess [11:43] and stuff like that, when it's already there [11:43] so you would copy the source to debian/l-n-s-m/ [11:43] do what ever you need to do [11:43] no.. no source [11:43] i will mv directly hte builded binaries [11:44] where would you build the binaries from? [11:44] directly in the kernel, like we are doing now [11:44] ah...so we still maintain the patches? [11:44] yes [11:44] ok gotcha [11:45] but given that they are modules in l-n-s-m [11:45] we take the freedom to remove them if they don't build [11:45] now i understand :) [11:45] or upstream dies [11:45] or the patch doesn't apply anymore [11:45] they are 3rd class drivers [11:45] 1st class in vanilla kernel [11:45] 2nd class - external we must have [11:45] so you would have a list of modules per arch and a list of modules where the are suppose to go [11:46] 3rd class - nice to have if it builds [11:46] we need one list of modules per arch. you already know where they will go [11:46] heh i guess i should have gone to udu :) [11:47] cool...ill get started on the list tonight [11:47] eheheh cool :) [11:47] and i am off to sleep [11:47] i am tired [11:47] nighty night [11:47] im off to learn about iscsi as well