=== Seveas [n=seveas@seveas.demon.nl] has joined #ubuntu-kernel [05:12] morning [05:27] gah... fabbione: I'll do that patch when I wake up [05:29] lamont: don't worry.. if it's ok with you i can upload both in debian and here === jbailey [n=jbailey@modemcable139.249-203-24.mc.videotron.ca] has joined #ubuntu-kernel [05:55] hey jbailey [05:56] Heya Fabio [05:57] jbailey: we need to get klibc to build on sparc... [05:57] actually.. B-D can't even be satisfied [05:58] fabbione: Hmm, I thought it was building. [05:58] Which b-d? [05:58] on kernel headers? [05:58] you are still using the -1- abi reference... [05:58] we are up to 6 and soon to 7 :) [05:58] Ah, thought I had updated that. [05:58] Bloody abi bumps. [05:58] retarded klibc :) [05:59] *lol* [06:00] jbailey: the append to initr* patches... [06:00] i saw a bug about them... [06:00] I have a bug report for initramfs asking for vesafb and fbcon to be auto added on i386 [06:00] I thought the kernels had all the framebuffer stuff built in automatically. [06:00] eh it did with initrd... [06:00] fabbione: I need to hack more on it. The classic initrd case is OOPSing at startup, and I haven't troubleshot it yet. [06:01] placing the modules in /lib/modules/$ver/initrd [06:01] Sorry? I'm having trouble following. [06:01] (My bad for telling you something else at the same time as you told me something) [06:02] fabbione@gordian:/lib/modules/2.6.12-4-686/initrd$ ls [06:02] capability.ko vesafb.ko [06:02] the modules in that dir are automatically added to the initrd [06:02] Right, but I thought that you said that you inicluded all the necessary framebuffer things in the kernel for each arch. [06:02] So I didn't bother including them. [06:02] (There's alot of historical cruft in initrd-tools) [06:03] we build them as module.. and we ship them.. that's all [06:03] Ah, okay. [06:03] i don't include them automatically anywhere.. [06:03] If you didn't, I wouldn't have a console on ppc... [06:03] jbailey: i think some arches must have fb in the kernel [06:04] others don't or can't [06:04] Ah, okay. So it's a per-arch thing then. [06:04] yup [06:05] 'k [06:05] jbailey: the best thing you can do to see it, is to grab all -images- from jackass and dpkg -c them [06:05] to get an idea [06:05] something you can easily do from either chinstrap or rookery [06:06] i need to wake up my wife... brb [06:14] re [06:14] jbailey: klibc.. you said you were going or planned to change it to use l-k-h [06:15] if that doesn't work, i can still provide you with a meta package to B-D on [06:15] so you don't need to get too crazy about abi bumps [06:15] specially after release.. when we might bump ABI in a security update [06:15] and klibc won't build anymore [06:16] Right, it just hasn't been top on my priority list when sorting all the initramfs-tools stuff out. [06:16] Now that it's in use, I want to clean up things like this. [06:22] jbailey: ok.. === jbailey needs to go to sleep. [06:35] night jb [07:06] doko: ping? [08:12] fabbione: pong [08:14] doko: amd64 kernel on i386.... [08:14] what B-D do i need? [08:19] libc6-dev-amd64 [08:20] I'm not sure, if these headers really work at the moment, I didn't get feedback from jbailey yet. so you might try amd64-libs-dev instead [08:29] well i need something a bit more tested.. since i need to ask elmo to install these possible B-D on concordia [08:29] also.. what's the minimal version of gcc-3.4 that i need to use... [08:37] 3.4.4-6ubuntu4 [08:49] ok === JaneW [n=JaneW@wbs-146-185-138.telkomadsl.co.za] has joined #ubuntu-kernel === chmj [n=chmj@196.36.161.235] has joined #ubuntu-kernel === Mithrandir [n=tfheen@c5100BC63.inet.catch.no] has joined #ubuntu-kernel [10:17] I think perhaps there is something in the hoary kernel which leaks loads of memory. [10:18] that's called X and GNOME [10:19] on a server? [10:20] you should have specified it before :) [10:22] :-P [10:22] my slabinfo showed me those nice numbers: [10:23] biovec-1 5925600 5925825 16 225 1 : tunables 120 60 0 : slabdata 26337 26337 0 [10:23] bio 5925623 5925743 128 31 1 : tunables 120 60 0 : slabdata 191153 191153 0 [10:23] which means it has allocated almost 6M objects of size 16 and 128, a grand total of ~850MB. [10:24] that hurts a bit when the box runs SA and some other stuff. [10:26] compared to the box freshly booted: [10:26] biovec-1 1347 1800 16 225 1 : tunables 120 60 0 : slabdata 8 8 0 [10:26] bio 1341 1581 128 31 1 : tunables 120 60 0 : slabdata 51 51 0 [10:27] weird [10:28] nothing special here, runs apache2, random user processes, does have a raid5 with evms, but that's just devmapper. [10:35] hmmmm no idea really... [10:44] http://www.ussg.iu.edu/hypermail/linux/kernel/0506.3/0965.html === Seveas [n=seveas@ksl403-uva-141.wireless.uva.nl] has joined #ubuntu-kernel [10:56] ah, it seems to be related to the fact that my SATA controller generates some spurious errors which then causes memory leaks in the raid code. [10:59] ah neat bug [11:00] yeah [11:00] oh well, I'm off to hack the live cd. [11:12] JaneW: ping? [11:13] fabbione: pong [11:14] JaneW: i need a little help with the wiki update... [11:14] ok [11:14] we agreed to make a list of deferred points inside specs [11:14] (thought you needed a name) ;) [11:14] but i am not sure how you want me to do that [11:14] which spec? [11:14] LinuxKernelRoadmap is 90% implemented, shipped, tested... [11:14] JaneW: name is for after Colony 3 release ;) [11:15] fabbione: cool, what's deferred? [11:15] but there are still a few deferred points... [11:15] JaneW: the list of deferred points is in comments inside the spec itself [11:15] fabbione: once I create a deferred column and list the stuff there, you can make the LKR completed. [11:15] i did never change the specs, but only added comments on what happened/when and why [11:15] ok, I'll go look [11:15] oh in the spec... [11:16] same goes for InstallerVolumeManager... [11:16] JaneW: yes.. [11:16] the ubuntu wiki is much slower that the UDU one... [11:16] that was the easiest to keep detailed track of stuff [11:16] without "altering" the original specs.. [11:16] i did only add comments to them [11:17] sort of status tracker [11:17] fabbione: is it 2 items? [11:17] JaneW: ? [11:18] what do you mean by "2 items"? [11:18] fabbione: there's 2 things in the spec that says DEFERRED [11:18] checking.. i remember there were a bit more [11:19] there are more... [11:19] oh [11:19] i will update the page with "DEFERRED" [11:19] some of them are not capital letters :) [11:21] ok thanks [11:25] fabbione: can I make the rest of the goal completed (light green i.e. 100% done) ? [11:27] JaneW: yup... [11:27] YAY! *clap* [11:27] there.. page updated... [11:27] thanks [11:27] shew the wiki is SLOW [11:28] did i ever mention that i HATE wiki? [11:28] fabbione: hehe, ok I see 6 DEFERRED's now... [11:28] yes that's correct.. === fabbione updates InstallerVolumeManager [11:31] JaneW: InstallerVolumeManagement looks good enough for me [11:31] the last comment from 2005-07-14 is the same as reported to the meetingh [11:31] there are no updates since than [11:34] fabbione: has InstallerVolumeManagement had more testing yet? [11:34] fabbione: and do you think it will change before it can be marked completed? [11:35] JaneW: it won't change because we can't support our 3 arches [11:35] it will be completed for real in breezy+1 [11:36] there is nothing we can do more for breezy [11:36] can I make it complete then? for purposes of tracking what need to be done now [11:36] JaneW: it's ok with me or we can move it to implemented -> deferred [11:36] because the real target of the spec can't be achieved [11:36] it is useable at all? [11:37] on the other side the logic has been implemented and it works [11:37] is the spec "LVM/EVMS by default" or "possible to use LVM/EVMS"? [11:37] yes yes.. it works [11:37] Mithrandir: yes [11:37] fabbione: "yes" is a terrible answer to an "or" question [11:37] ok, then lets say complete? Although are portions deferred, in which case I;ll add it to deferred as well? [11:37] Mithrandir: LOL [11:37] damn.. i can't workaround you ;) [11:38] "This spec considers our options for setting up logical volume management in the installer by default." [11:38] fabbione: do you think we can get a fix for the slab issue into hoary, it's kinda critical that servers run out of memory after a month. [11:38] Mithrandir: i think so.. i can pass it at memory leak/ddos [11:39] fabbione: great, thanks. :-) [11:40] just tell me if you need somebody to test the update [11:42] Mithrandir: well if you can test it yourself that'd be great [11:42] fabbione: that patch doesn't apply to -10, so we need to backport it a bit. [11:42] a one liner that doesn't apply is an issue... === fabbione checks [11:43] the surrounding code has changed, but I think it's doable anyhow [11:49] Mithrandir: i can't find the code at all... [11:50] not even in .12 [11:51] http://lists.debian.org/debian-kernel/2005/07/msg00231.html ? [11:54] it's relative to http://marc.theaimsgroup.com/?l=linux-raid&m=111155701929593&w=2, I think [11:54] the first patch you pasted is in .13 git [11:55] and it applies to a function called [11:55] super_written [11:55] that's not in .10 or .12 [11:55] the missing link is in the theaimsgroup link, though [12:00] fabbione: ok InstallerVolumeManagement is completed in the main table and listed in deferred too, I hope that is ok. [12:00] Mithrandir: the patch at http://lists.debian.org/debian-kernel/2005/07/msg00231.html seems to be the correct one [12:00] JaneW: sure.. that's fine [12:01] Mithrandir: [12:01] static int sync_page_io(struct block_device *bdev, sector_t sector, int size, [12:01] fabbione: all your goals are completed - you can go on holiday again ;) [12:01] on line 332 has been splitted into 2 functions [12:01] that's why it doesn't apply [12:02] JaneW: i would love to.. but i guess somebody won't be happy about it :) [12:02] JaneW: i still need to push the SoC stuff in... [12:03] Mithrandir: if you can test that patch, i will add it as security thing for hoary [12:03] given that a memleak is NOT good [12:04] fabbione: ok, I'll do that. Can you prepare images or should I? [12:04] now.. time to prepare some food and to get a long break [12:04] Mithrandir: up to you.. if you can do it > * [12:05] fabbione: yes I hope the SoC stuff completes well, and I could give you some more goals if you are looking? ;) [12:05] otherwise give me time till tomorrow.. i guess i will have to prepare a hoary security update [12:05] JaneW: thanks but i have enough stuff to do :) [12:05] :) [12:06] fabbione: tomorrow is fine for me, I'm trying to find out wtf is up with the live cd atm [12:06] + give a head up to BenC [12:06] Mithrandir: ok [12:06] security... [12:06] update packages around... === fabbione is already tired at the second day of work [12:07] anyway.. long break time now [12:19] c [12:19] argh [12:19] sorry [01:26] === TheMuso [n=luke@dsl-202-173-132-131.nsw.westnet.com.au] has left #ubuntu-kernel [] [01:41] fabbione: ping [01:41] JaneW: pong? [01:41] fabbione: what's up with NFSRoot? (jbailey is not around), there are no comments listed at all [01:42] JaneW: it's part of ltsp/EarlyUserSpace... [01:42] it works :) [01:42] i will check with jbailey to update the wiki [01:42] ok thanks :) [01:46] fabbione: xen - is your student working on that whole goal? or only part of it? Is it justified in moving over to WIP? [01:46] JaneW: xen?? you just told me to forget about it yesterday :) [01:46] oh right, sorry that's Ed... [01:46] Mithrandir! ping === zul [n=chuck@CPE0006258ec6c2-CM000a73655d0e.cpe.net.cable.rogers.com] has joined #ubuntu-kernel [01:48] heylo [01:48] JaneW: pong === warthylog [n=warthylo@port49.ds1-van.adsl.cybercity.dk] has joined #ubuntu-kernel === Topic for #ubuntu-kernel: Ubuntu kernel development discussion ONLY | 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--preX,11--2.6.12 === Topic (#ubuntu-kernel): set by fabbione at Mon Aug 15 08:50:10 2005 [01:50] (JaneW/#ubuntu-kernel) Mithrandir: are we likely to get anothing completed there for Breezy, or will it need to be deferred? [01:50] (Mithrandir/#ubuntu-kernel) that depends on whether I should concentrate on getting it into ok-ish shape or not. [01:50] (Mithrandir/#ubuntu-kernel) I think it's doable for breezy [01:51] (Mithrandir/#ubuntu-kernel) but at the moment I'm not working on my goals but rather picking up various loose threads around. [01:53] JaneW: i doubt there is a chance to get Xen for breezy [01:53] that means a FeatureFreeze breakage [01:53] + ENOTIME to do a deep test [01:54] fabbione: it'd be universe for now. [01:54] and yes, mdz wouldn't be happy about the freeze exception [01:54] imho it's pointless to spend that many resources for a universe package [01:54] (I'd imagine) [01:54] given that the patch in 2.0.6 is for .11 [01:55] and porting it to .12 is a pain [01:55] fabbione: i have some stuff in my arch for you to merge [01:55] there is just too much work that needs to be done, without even be sure that it will work [01:55] zul: good boy :) [01:55] fabbione: do i get a cookie :) [01:55] fabbione: hmmm, don;t think mdz will allow that - so shall I defer? [01:55] JaneW: yes. let's get it done properly for breezy+1 [01:56] JaneW: unfortunatly upstream and us are way too much desynced [01:56] zul: with choccolate too ;) [01:56] wohoo! === fabbione baz gets zul [01:58] now if only dell would get off their butts === fabbione will soon go offline for brain crashing [02:00] i woke at 5am today! === fabbione needs some sleep [02:01] meh...toodles === fabbione goes offline [02:09] later fellas === chmj [n=chmj@196.36.161.235] has joined #ubuntu-kernel === doko_ [n=doko___@dsl-084-059-093-091.arcor-ip.net] has joined #ubuntu-kernel === JaneW [n=JaneW@wbs-146-185-138.telkomadsl.co.za] has joined #ubuntu-kernel === calc [n=ccheney@ip70-185-4-246.ma.dl.cox.net] has joined #ubuntu-kernel === lamont [n=lamont@15.238.5.97] has joined #ubuntu-kernel === jbailey [n=jbailey@modemcable139.249-203-24.mc.videotron.ca] has joined #ubuntu-kernel === Seveas [n=seveas@seveas.demon.nl] has joined #ubuntu-kernel [07:43] fabbione: building now [07:43] actually, there were some other issues that I want to look at for -6 [07:48] lamont: -6 ? [07:51] util-linux 2.12p-6 [07:51] and -6ubuntu1 [07:51] ah ok [07:51] fabbione: eh? i though you got stuff from my arch? [07:51] fabbione: but I'm also going to go to lunch, and email upstream before I upload. :-) [07:52] zul: not yet.. i did a checkout.. look at it and merge :) [07:52] lamont: ehhe sounds a good plan [07:52] zul: i meant.. i did a checkout.. now i need to look at them and merge :) [07:53] i understand fabbione speak.. [07:56] zul: die === fabbione goes for dinner [08:14] fabbione: it should be uploaded before reasonable people are awake in your area tomorrow. :-) === lamont lunches [08:14] lamont: ahaha ok :) [08:14] hmmm pasta today is going to be good... [08:14] i love to cook the tomato souce on very low fire... [08:15] * Use helper program in mount for guessed FS types too. Thanks to Manish [08:15] Singh and Fabio Massimo Di Nitto. Adds: 20guesshelper.dpatch [08:15] * Remove /usr/doc links on install. Closes: #322806, #322816 [08:15] * Fix /usr/bin/pg pager alternative. Closes: #323204 [08:15] it takes longer but it tastes 3 times more [08:15] super [08:19] hey fab [08:20] fabbione: please make RTC not be a module anymore in ubuntu kernels (assuming it still is) [08:20] http://bugs.debian.org/50572 [08:30] lamont: Hey - did that glibc update fix your hppa troubles? === jbailey_ [n=jbailey@modemcable139.249-203-24.mc.videotron.ca] has joined #ubuntu-kernel === calc [n=ccheney@ip70-185-4-246.ma.dl.cox.net] has joined #ubuntu-kernel [09:01] What populates /lib/modules/$(uname -r)/initrd? Is that an upstream decision or something the kernel team does? [09:42] hey BenC [09:42] lamont: we can't compile it for all arches [09:43] some of them don't have standard RTC [09:43] fabbione: Heya. [09:43] jbailey_: we populate it manually. [09:44] Okay, so I'll just trust that you guys will continue to handle that directory for now. [09:44] I had missed it in the initramfs-tools stuff. [09:45] jbailey_: i would love to kill it.. [09:45] if you want you can take it over via initramfs [09:45] it's only wasted space on duplicate modules [09:45] That would work too. How do you decide what goes in? [09:45] fabbione: It's hardlinked, so there's no wasted space. [09:46] jbailey_: i don't decide.. really. i found it from warty and ported forward.. [09:46] fabbione: Mmm.. Then let's leave it there for now and decide for 6.04 [09:46] It's the least amount of work for me to just pull it from there, and then at least we can handle it during feature hacking time. [09:47] sure [09:47] works for me [09:50] lamont: i will look at it again, but i remember that problem... [09:51] laster [09:56] jbailey: dunno - haven't gotten there yet [09:56] lamont: Well, even on your test build, though. [09:56] jbailey: libc seems to be broken on sparc [09:56] jbailey: I thought it was a gcc front end fix [09:56] it makes apt-utils -> apt-ftparchive segfault [09:56] fabbione: Just this most recent update? === lamont was offline from wed thru sunday, hopingto worry about things later this week, [09:57] jbailey: i did try both debian 2.3.5-3 and our latest [09:57] kyle was playing with the debian side of things, and planning to followup with me afterwards [09:57] jbailey: you're referring to the GOT issue? [09:58] lamont: Whatever the issue was that commenting out that patch was. [09:58] Was that the GOT issue, or was that a different one? [09:58] fabbione: Hmm. [09:58] jbailey: commenting out the patch fixed that issue. [09:58] lamont: 'k [09:58] the new issue is that we have GOT abs syms in shlibs, which makes many things unloadable. [09:59] jbailey: it actually leads to a "Bus Error" [09:59] lamont: Mmm, tasty. [09:59] fabbione: Right. But why do you think it's glibc? [09:59] because downgrading glibc make it working again? [09:59] gcc.gnu.org/PR23369 [10:00] fabbione: glibc doesn't generally do memory allocation, it may have had something in it before that was forcing alignment by doing an extra copy or something that doesn't exist now. [10:00] fabbione: We'll need to trace it to the call before it SIGBUSes and figure out what it's being called with. [10:00] jbailey: the bus error comes right during a pread [10:00] iir [10:00] iirc [10:01] Right, so is the buffer that's handed in correctly alligned? [10:01] i will need to do it again in a chroot [10:01] Sure, we can play dueling screen sessions again. =) [10:01] jbailey: yes.. but not before i finish to download a few torrents :) [10:01] i am not in the mood to stop them now [10:01] fabbione: I'm about full up for today anyway. [10:02] jbailey: dude.. it's like 10pm here :) [10:02] Exactly. =) [10:02] jbailey: did you install your sparc? [10:03] No, I've been away the past few days. [10:03] ah right... [10:03] forgot about that [10:03] Reminds me that I should ask Simon is I can borrow his null modem cable to get it setup. [10:04] jbailey: make one... [10:04] it's easy .. [10:05] I used to have a nice breakout box for doing things like that. [10:05] yeah i need to get one again [10:05] speaking of which... i need the console cable for the U60 [10:05] BenC: do you happen to have the schema of the connectors around? [10:06] U60 has the big fat female (DB25) [10:06] to connect to a normal serial port (9 pin or whatever it is now) [10:12] schema? [10:13] you just need a regular serial cable, and a NULL modem terminator on it (or a NULL cable) [10:14] regular 25-9 pin adapters work ok with the sparc serial too [10:14] BenC: hmm weird... i saw the SUN guy using a DB25 to RJ45 -> Cisco RJ45 (all crossed) -> RJ45 DB9 [10:15] the other solution didn't work.. [10:15] ugly [10:15] I have a 25pin cable, with a 9-25pin adapter and a null term [10:15] that's what i tested.. [10:15] i will try again [10:15] should work [10:55] http://edition.cnn.com/2005/US/08/16/computer.frenzy.ap/index.html [10:55] AHAHHAHA === fabbione goes to sleep [10:58] cya tomorrow fellas [11:17] g'n fabio === dmk [n=dmk@host81-155-36-227.range81-155.btcentralplus.com] has joined #ubuntu-kernel