/srv/irclogs.ubuntu.com/2005/08/20/#ubuntu-kernel.txt

=== mxpxpod [n=bryan@unaffiliated/mxpxpod] has left #ubuntu-kernel ["Leaving"]
fabbionemorning06:09
=== fabbione starts unfucking the baz repo
=== JaneW [n=JaneW@196.36.161.235] has joined #ubuntu-kernel
fabbioneMithrandir: so.. why do we need unionfs udebs?07:43
Mithrandirfor the live cd07:44
Mithrandirhttps://wiki.ubuntu.com/LiveCDFeatures07:44
fabbioneok07:45
fabbionei am unfucking the baz repo now07:45
fabbioneto resync with the last 2 kernel uploads07:45
fabbioneand after that we can open devel dances again07:46
=== fabbione powers up the sodomotron and points it towards baz --'s
=== JaneW [n=JaneW@196.36.161.235] has joined #ubuntu-kernel
=== Mithrandir [n=tfheen@c5100BC63.inet.catch.no] has joined #ubuntu-kernel
=== Mithrandir [n=tfheen@c5100BC63.inet.catch.no] has joined #ubuntu-kernel
Mithrandirfabbione: so you've added unionfs now?08:44
Mithrandiror should I?08:44
fabbioneMithrandir: i am still branching and merging...08:45
Mithrandirok08:45
fabbioneMithrandir: but i can do it.. so don't worry about it08:46
Mithrandircheers08:47
=== ..[topic/#ubuntu-kernel:fabbione] : 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
fabbionethere08:50
fabbionebaz resynced08:50
Mithrandiris there anything explaining what the logic in the version string is? :-)08:54
fabbionepre <- prerelease08:54
fabbioneX <- hidden abi number08:54
fabbione,11 <- next deb version08:55
Mithrandirprerelease as in "not yet uploaded", not "prerelease from kernel.org"?08:55
fabbionenot uploaded yet08:55
Mithrandirand what do you mean by hidden abi number?  Might break multiple times before next public breakage?08:55
fabbionehidden abi is a trick to avoid more than a devel branch...08:56
fabbioneexample:08:56
fabbionewhen we upload 6,10 it means ABI =6 ver 1008:56
fabbionebut during the devel cycle of 1108:56
fabbionewe don't know if the abi is going to break08:57
fabbioneso you can't name the pre branch as 6,1108:57
fabbioneso we hide it08:57
Mithrandirok08:57
fabbionebecause it might easily land as 7,1108:57
fabbionelook at preX,Y as HEAD08:57
fabbionewhere we all commit08:58
Mithrandirmm08:58
fabbioneonce it's ready we merge it mainline--2.6.1208:58
fabbionefrom there we branch the tag release and the next pre08:58
fabbionedon't ask me why this schema08:58
fabbionelamont did create it in early stage :)08:58
fabbionebut clearly merging from other repos is ok08:58
Mithrandirit makes some kind of sense, I guess.08:59
Mithrandirif I'm touching stuff, do you prefer me to commit into the kernel-team repo directly or work in my own space and then later merge it when it's ready?08:59
fabbioneMithrandir: i don't care either way09:00
fabbioneif you keep your repo, you need to ping me manually for merge or merge yourself09:00
Mithrandirbut kernel-team@ is supposed to be kept "working at all times" or are temporary breakages ok?  (I presume the former)09:01
fabbionebut if you commit directly, in most of the cases your changes will be test-builded by me each time i do a change09:01
fabbionewell the former is better, but if it breaks it's not a tragedy09:01
Mithrandirok09:02
fabbionelet say that i don't mind "it did break by mistake"09:02
fabbionebut breaking it on purpose i am a bit less happy09:02
Mithrandirok09:02
fabbioneor at least tell me in advance :)09:02
MithrandirI'm fine with doing on-purpose breakages in my own space. :-)09:03
fabbioneMithrandir: btw.. unionfs is/should be available on all arches09:04
fabbionewhen you have one module only in a list, mark it as ? is pointless09:04
fabbionebecause kernel-wedge will fail later with an empty udeb09:04
fabbioneso either is everywhere, or you need to create a per arch list09:05
Mithrandirok09:05
fabbioneif there is more than one module and one is available everywhere, than it makes sense to have other optionals09:05
fabbionebut iirc unionfs is available on all our kernel images09:05
Mithrandir'k09:05
fabbioneis there a bug assigned for unionfs udebs?09:11
Mithrandirnot to my knowledge09:16
fabbioneok09:16
fabbione* update pristine tree (kernel-team@ubuntu.com--2005/kernel-debian--preX,11--2.6.12--patch-1 => kernel-debian--preX,11--2.6.12--patch-2)09:17
fabbionethis is the commit you want to look at09:17
fabbioneJeg er s travlt og der kun i frst dag p arbejde09:21
=== fabbione shows that he is going to danish class now ;)
=== chmj [n=chmj@196.36.161.235] has joined #ubuntu-kernel
Mithrandirthanks09:25
Mithrandirshould (udeb) modules call depmod -a in their postinst?10:02
fabbioneno idea.. kernel-wedge takes care of creating that stuff10:02
Mithrandiroh well, apparently it doesn't.  I'll add a workaround in casper, then10:03
fabbionei think depmod -a is executed after all kernel modules have been downloaded10:04
fabbionebut i dunno about casper...10:04
fabbionethat's anna taking care of that10:04
fabbioneor at least it should10:04
MithrandirI've added a workaround and I'll speak with Colin once he's back again10:10
Mithrandirlive cd testing takes tiiiiime10:43
fabbioneeheheh10:43
fabbionei am doing a test build to create the unionfs udeb10:43
Mithrandiris unionfs supposed to be the slowest thing since sliced bread?10:44
fabbioneno idea10:45
MithrandirI think something I did made debconf unhappy10:46
fabbionethat's not difficult :)10:58
MithrandirI need to find something to generate Release files for me.10:59
Mithrandirapt-ftparchive release seems to work11:23
fabbioneyup.. i use it here :)11:27
fabbione(breezy-chroot)fabbione@davis:~ $ dpkg -c unionfs-modules-2.6.12-6-powerpc64-smp-di_2.6.12-6.11_powerpc.udeb11:48
fabbione-rw-r--r-- root/root   1718210 2005-08-15 09:41:53 ./lib/modules/2.6.12-6-powerpc64-smp/kernel/fs/unionfs/unionfs.ko11:49
Mithrandircheers11:57
Mithrandirfabbione: when can we have that in the archive?12:48
Mithrandirfabbione: I would really like to have it in quickly so we can get some unionfs testing12:49
Mithrandirsince I just uploaded a casper which needs it.12:50
fabbioneMithrandir: meh... in a few days :/01:01
fabbioneyou should have told me it was urgent :/01:02
Mithrandirfabbione: can I push a new version up with just that small change?01:02
fabbioneMithrandir: let me see a couple of things...01:03
fabbionei might be able to upload today...01:03
Mithrandirthat'd be great. :-)01:03
Mithrandirsorry about the miscommunication01:03
fabbioneyeah but it means pushing a big bunch of security fixes for the next upload...01:03
fabbionewe also need elmo/mdz for NEW love...01:08
fabbionethe new udeb will stall buildd -> archive movement01:08
Mithrandirargh, true :-(01:09
Mithrandircan we see when mdz gets around in a few hours?01:10
fabbionei have no problems with it, but i must leave around 3:30 pm01:10
fabbione(our time)01:10
fabbioneso i can upload.. (untested) but you need to follow up with mdz01:11
fabbioneanyway.. food now :)01:14
Mithrandirok, cheers01:16
mjg59fabbione: I'm going to have patches to feed you from laptop testing01:18
mjg59Need to do a bit more back and forth with upstream on a couple first, though01:18
fabbionemjg59: ok, but don't wait too long01:32
fabbionewe need to update heaps load of stuff in the kernel02:03
fabbioneTheMuso:02:03
fabbionebah02:03
fabbioneSO :02:03
fabbione1) i am going to upload -6,11 basically untested...02:03
fabbione2) NOBODY please plan to push URGENT stuff for a few days that needs an upload yesterday02:03
fabbionepush but wait...02:05
=== swarm [n=swarm@host124-43.pool8253.interbusiness.it] has joined #ubuntu-kernel
Mithrandirfabbione: actually, you don't need to hurry with the kernel, my casper upload got rejected.02:15
fabbioneMithrandir: ok!02:17
fabbioneperfect02:17
fabbionethat will give me the time to break it in unreasonable new ways :)02:17
=== zul [n=chuck@CPE0006258ec6c2-CM000a73655d0e.cpe.net.cable.rogers.com] has joined #ubuntu-kernel
zulmorning02:18
fabbionehey zul02:18
zulhey fabbione good vacation?02:20
Mithrandirfabbione: I'd  love it if you can get it in a couple of days, though.02:22
fabbioneMithrandir: 2/3 days should be fine.. 02:22
fabbionezul: more or less02:22
fabbionezul: i saw you missed me :)02:23
zulheh...02:31
zulat little...but not much02:31
chmjheh 02:31
fabbionezul: do you have anything that i need to merge?02:32
zulyeah a couple of bug fixes from bugzilla02:32
zulits in my arch under ,902:33
zuluh...we are at ,11 now?02:33
fabbionezul: mind to update to ,11 ?02:33
fabbioneyes.02:33
fabbione.1102:33
fabbionezul: i did import this morning the 2 uploads that have been done outside baz02:34
zulok02:34
zulsigh gimme a sec02:35
fabbionetake your time02:35
fabbionei will leave in less than one hour02:35
fabbioneso i will defenetely not upload today02:35
zulok good because im at work right now02:36
dokofabbione: the i386 biarch compilers are in the archive, hint, hint ...02:36
fabbionedoko: are you hinting me that you just offered volunteer to build the kernel?02:37
fabbionesince you got biarch done, you must have more spare time now :)02:37
Mithrandiris lrm in baz?02:39
fabbionedoko: more seriously.. i need to build an amd64 kernel on i386..02:39
fabbioneright?02:39
fabbioneMithrandir: nope.. lrm is crap.. we don't want it :)02:39
fabbionedoko: if so.. what is the usecase for it?02:39
Mithrandirfabbione: ok, since I need to fix it up a bit.  It's br0ken.02:40
fabbioneMithrandir: eheh ok02:41
zulfabbione: i just branced and merged my stuff in my arch so my changes are in X,9 in my arch02:41
fabbionezul: ok... can you just branch to ,11 ?02:41
fabbioneso we work on same versions?02:42
fabbioneit's already confusing enough :/02:42
zulok..will do on my break02:43
fabbionezul: thanks02:45
dokofabbione: use case is: install i386 on amd64 to have better 32bit compatibility, and be able to do things for 64bit as well, i.e. a chroot02:46
=== fabbione scratches his head...
fabbioneyou mean installing a 64bit kernel with a 32bit userland?02:47
dokoyes02:47
chmjis that possible  ?02:47
fabbionechmj: yes02:47
fabbionedoko: i *THINK* i can do it...02:48
fabbionebut i won't make more than one flavour...02:48
fabbioneonly a generic kernel02:48
fabbionethat also means bloating the i386 CD with udebs...02:48
dokofabbione: sure, that should be sufficient02:48
fabbionedid you mention that to Kamion?02:48
fabbionebecause it's another kernel to ship...02:48
dokono, not yet. how much is that in size?02:49
dokobut at least it's mentioned on our ToolchainRoadmap02:49
fabbionedoko: a bunch of MB02:49
fabbionechanges that needs to be done in the installer02:49
fabbioneand several other things...02:49
fabbioneit's not exactly trivial to do all of the above02:50
dokoouch02:50
fabbioneand for sure i am not going to do that in the installer while Kamion is away02:51
zuli...dislike...sybase02:51
=== JaneW [n=JaneW@196.36.161.235] has joined #ubuntu-kernel
fabbioneYAY for the ABI change!03:11
fabbione-+#define OCFS2_BUILD_VERSION "0.99.17"03:14
fabbione++#define OCFS2_BUILD_VERSION "1.1.0"03:14
fabbioneMithrandir: whatever fix you have for lrm, better you upload it asap03:17
fabbionenext kernel upload will bump the ABI03:17
Mithrandirfabbione: it's already uploaded.03:17
fabbioneah cool03:17
fabbionelet see in which wonderful ways the kernel is going to break :)03:19
fabbionei already noticed a target that goes banana with the debian version made of 2 digits :)03:19
Mithrandirheh03:20
fabbioneimpressive...03:21
fabbionethe force is strong in the bumpabi target...03:21
=== doko_ [n=doko___@dsl-084-059-072-034.arcor-ip.net] has joined #ubuntu-kernel
fabbionelater fellas...03:31
=== fabbione -> danish class
=== lamont-away returns
lamont-awayfabbione: I did it that way because I sometimes forget to tag...  But I almost always remember to branch for the new version04:43
lamont-awayor rather, I'm far less likely to release from non-mainline than I am to forget to tag from mainline04:43
=== Seveaz [n=seveas@seveas.demon.nl] has joined #ubuntu-kernel
=== BenC [n=bcollins@206.246.247.150] has joined #ubuntu-kernel
dokofabbione: "x86_64 kernel on i386 system" on u-u ml05:39
zulhey BenC 05:46
BenChey05:46
fabbionehey BenC09:03
fabbionewelcome to the team09:04
fabbioneBenC: i am going to have dinner and i will be back later (if wife will allow me ;))09:04
fabbioneso we can talk a bit09:04
zulyou are so whipped09:12
zulbut thats ok so am i09:12
BenCok09:15
=== lamont [n=lamont@15.238.5.9] has joined #ubuntu-kernel
zulhey lamont 09:21
lamonthey zul09:21
mjg59 /win goto #ubuntu-devel09:22
mjg59Hrngh.09:22
fabbionere10:19
fabbioneBenC: still around?10:21
fabbionelamont-away: ping?10:23
fabbionebah something is wrong with this client10:23
=== fabbione [n=fabbione@port49.ds1-van.adsl.cybercity.dk] has joined #ubuntu-kernel
lamontsup>10:26
fabbionelamont: did you read my email about util-linux patch?10:26
lamontnot yet10:27
=== lamont is still burried
fabbioneok10:27
lamontgenerally, I've tried to not deviate from upstream unless there was good cause.  I haven't really looked at the 200+ line patch, but it seems to be relatively isolated in it's changes... something as invasive as the hurd patch won't be in until upstream takes it.10:28
fabbionelamont: the patch is much smaller than it looks like10:29
lamontright10:29
fabbionedue to static int foo it needs to move a function a few lines above10:29
lamontit looks to be like it just calls the helper for the fs type if said beast exists10:29
lamontah, even better10:29
fabbioneexactly10:29
fabbioneso it's pretty simple in itself10:30
lamontlet me look at it tonight and I'll upload to both10:30
lamontand fire it upstream10:30
fabbionesure...10:30
fabbionei think Manish did try to get it upstream with no success10:30
fabbioneor no response from upstream....10:30
lamontright - upstream has bounced around a bit...10:31
fabbionewho is upstream? 10:31
fabbionevietse?10:31
lamontit might be semi-stable, esp since there's a new upstream version now as well10:31
lamontadrian bunk, iirc10:31
fabbioneah...10:31
lamontthe debian maintainer before me took it over upstream recently10:31
fabbioneok10:31
fabbionebtw.. we are bumping ABI for the next kernel upload10:32
fabbioneso if you have something intrusive to push, it's the right time10:32
fabbioneAH GREAT.. the second sparc buildd managed to die in less than 12 hours10:33
=== fabbione sighs
lamontmtg10:36
=== lamont needs to isolate the ia64 acpi issue, but that will be later this week at the earliest. likelihood of an ABI bump is >0
lamontjust because acpi seems to be intrusive10:40
fabbionelamont: if you build with ACPI disable, does it boot?10:40
fabbionebah i come back to talk with Ben and he disappeared :)10:40
=== fabbione will teach BenC the way of the dark side of the force
=== BenC [n=bcollins@206.246.247.150] has joined #ubuntu-kernel
BenCfabbione: ping10:48
fabbioneBenC: pong10:59
fabbionei tought you left...10:59
fabbioneand i was close to go away :)10:59
BenCsent you an email11:01
fabbioneyeah i just saw it...11:01
fabbionei think there are 2 groups of bugs we are interested in..11:01
fabbioneUSB and ACPI related11:01
fabbioneUSB seems to be a royal pain11:01
BenCyeah, I saw a lot of ACPI related bugs11:01
fabbioneACPI is well.. mjg59 crack11:01
fabbionebut it still needs extra love...11:02
fabbioneon the other side we are already in Feature Freeze for Breezy11:02
fabbioneso we need to avoid extra breakage if we can11:03
BenCright11:03
fabbione.10 was a crappy release11:03
fabbionetoo many patches coming from too many upstreams11:03
fabbione.12 is looking quite good compared11:03
BenCare there any show stoppers, or are the issued mainly non-regression things we want fixed?11:03
fabbioneshow stoppers none that i am aware of...11:03
BenCany serious regressions?11:03
fabbionethere is a Serial PATA patch that is sort of a problem11:04
BenCPATA?11:04
BenCor did you mean SATA?11:04
fabbionewe had a patch in hoary pulled from atadev for a mistake11:04
fabbione#1329811:04
fabbione(and other related bugs)11:04
fabbionePATA...11:05
fabbionezul: you need to learn to write Subjects in the email.. see.. i can't even copy paste without looking stupid :)11:05
BenChehe11:05
fabbionethe issue is that jgarzik almost crossburned me for inclusion of that patch11:06
fabbionebecause it can cause data corruption11:06
fabbionenow everybody is including it and we are not....11:06
BenChave the libata head patches been tested?11:06
fabbioneand tbh i have no fucking clue who to trust11:06
fabbionethey have been in hoary....11:06
fabbioneright now they are not in breezy11:07
fabbionebecause i did expected to find them in .1211:07
fabbionewhile they are not...11:07
BenCanyone besides the submitter able to test things?11:07
fabbionenot from our team11:07
BenChardware specific testing....the downside to all this :)11:08
BenCok, I'll put that on my list11:08
fabbioneehhe no shit!11:08
BenCanything else?11:08
BenCbtw, I have my G4 installed with ubuntu now11:08
BenCso I can test some ppc32 things11:08
fabbioneah nice11:09
fabbioneit looks like Jeff didn't make too much noise on redhat bugzilla...11:09
fabbionebut the bug explains that:11:09
BenCit's UP and only 512megs of ram, which is the stable type system that doesn't seem to show in most reports about ppc11:09
fabbionea) ghetto pulled the patch from jgarzik HEAD11:09
fabbioneb) we pulled from ghetto11:10
fabbionec) fedora and redhat did pull it from us11:10
fabbioned) jgarzik that works for RH got pissed11:10
fabbioneGO FEDORA!11:10
BenCwho/what is ghetto?11:10
fabbionegentoo11:10
BenCah, ok11:10
fabbionedude.. you should know that by amok ;)11:10
BenClol11:11
BenCAmok, the wiki of IRC11:11
fabbioneso true :)11:11
BenChaven't worked with ATA much, but libata is a kernel driver library right? (not userspace)11:12
fabbioneyup11:12
fabbionethe patch is much much cleaner than the old one11:12
fabbioneit seems safe to me11:13
fabbioneBenC: did you manage to strech your wings on baz?11:13
fabbione http://sourceware.org/ml/binutils/2005-08/msg00190.html <- (from doko11:14
fabbioneBenC: but if you feel lucky i have a few tons of security bugs for hoary :)11:15
BenCyou can dump as much on me right now as you like...haven't been here long enough to be "busy" :)11:17
fabbioneBenC: ahahah11:17
fabbioneenjoy it till you can11:17
fabbioneBenC: get a bit familiar with the kernel build.. we have a bunch of extra features that debian does not have 11:19
fabbionelike the ABI checker..11:19
fabbioneand some extra madness in debian/rules...11:20
fabbionelike we build all from one source...11:20
fabbioneincluding udebs11:20
dilingerfabbione: debian now builds all from one source, btw.. but not udebs.11:24
fabbionedilinger: uh yeah...11:25
fabbionesorry forgot about that11:25
BenCTo: bcollins@kernel.org11:33
BenCsweet11:33
fabbioneehehhe11:34
fabbionei have the BAD feeling i just managed to fuck up the sparc buildd....11:35
BenCimpossible, sparc is unfuckable11:35
fabbionei can manage..11:35
fabbione:)11:35
fabbioneah here it is11:36
fabbioneno shit it doesn't show anything on console..11:39
fabbionethe cable is disconnected....11:39
fabbione dists/breezy/main/binary-sparc/:Bus error11:42
fabbionei guess sid is sort of broken...11:42
BenCwhat command caused a bus error?11:47
fabbioneapt-ftparchive11:51
fabbioneafter a glibc upgrade...11:51
=== fabbione attempts a downgrade
BenCc++11:53
fabbionedpkg: /lib/libc.so.6: version `GLIBC_2.3.4' not found (required by dpkg)11:54
fabbioneAH CRAP11:54
fabbionei should have downgraded dpkg first....11:55
fabbioneSEE I CAN MANAGE TO DESTROY SPARC! :)11:55
=== fabbione tries to remember how to unfuck that
BenCar/tar11:58
BenCmaybe dpkg-deb will work11:58
fabbioneyeah..11:59
fabbionear x 11:59
=== fabbione heads to bed
fabbionenight guys12:02
fabbionecya tomorrow12:02

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!