[12:04] <mjg59> I don't think infinity has seen it though
[12:04] <cjb> No, not you.
[12:05] <mjg59> It's my concave chest he's thinking of
[12:05] <cjb> Oh!  I see.
[12:05] <cjb> I was imagining his concave chest enabled more build-up of sweat, resulting in said hotness and bother.
[12:09] <infinity> I think you need to stop imagining things altogether.
[12:10] <cjb> That works.  I could always leave for Go club like I've been meaning to for the last ten mins.
[12:23] <jbailey> BenC: For the X hang that I'm seeing, BenH is asking if I have the kernel side fix for ppc64.  Any idea if that's in the -8 kernels?
[12:23] <jbailey> -8.10, sorry
[12:23] <infinity> jbailey : You're SOOO yesterday, upgrade to -9.11.
[12:24] <jbailey> I will.  Just as soon as my ISP gets a !@#$ WORKING DNS SERVER DAMMIT
[12:26] <BenC> lol
[12:26] <BenC> ask benh when it was synced to linus
[12:26] <BenC> and I can tell you when we got it
[12:27] <BenC> I'm surprised it stayed broke, considering linus uses a G5 for his kernel work now
[12:27] <jbailey> Well, it's G5 + r300.
[12:27] <BenC> he has nvidia?
[12:27] <infinity> Plus CVS r300, to be exact.
[12:27] <jbailey> No idea.
[12:27] <BenC> ok
[12:27] <infinity> If you're not running Xorg pre-releases, you don't see the bug.
[12:27] <infinity> Or if you have DRI disabled.
[12:27] <infinity> Or, or...
 between rc5 and rc6
 a memory map fix
[12:32] <jbailey> BenC: ^^
[12:35] <BenC> -9.11 is -rc6
[12:35] <BenC> so that should fix it
[12:40] <sistpoty> hi 
[12:40] <BenC> sistopy: the thing that bothers me is that it worked once
[12:40] <sistpoty> yeah... I was quite surprised seeing that it doesn't work any longer :/
[12:40] <BenC> working once implies that something with the hardware changed
[12:40] <BenC> because obviously the kernel didn't between reboots :)
[12:40] <sistpoty> however if you look at the kernel messages, ide ctrl'ers come up in a different order
[12:41] <sistpoty> and btw.: it's working w.o. problems with 2.6.12 ;)
[12:42] <BenC> hmm
[12:43] <sistpoty> I did some more upgrades between reboots... I'll check what exactly they were
[12:45] <BenC> oh, wow, ide0 and ide1 were probed after 2,3,4,5
[12:45] <BenC> interesting
[12:45] <BenC> ide0 failed in your good system
[12:45] <BenC> ide0: I/O resource
[12:45] <BenC> 0x3F6-0x3F6 not free.
[12:46] <BenC> so it wasn't a success, just failed differently
[12:47] <sistpoty> he, k... ide0 is my cdrom drivers... I didn't actually use it 
[12:47] <sistpoty> s/drivers/drives/
[12:47] <BenC> have you tried any of the acpi=off pci=noacpi options?
[12:49] <sistpoty> not yet
[12:49] <sistpoty> I'll try that... brb
[12:51] <infinity> What bug number is this
[12:51] <infinity> ?
[12:51] <sistpoty> 20910
[12:51] <sistpoty> <- rebooting
[12:51] <infinity> BenC : Also, ide0: I/O resource 0x3F6-0x3F6 not free isn't an error, usually, it just means ide-generic got loaded after a better IDE driver already claimed the resources.
[12:52] <infinity> BenC : I was going to push you a change to make that printk less scary (and more obviously informative)
[12:52] <infinity> (We already dropped it from an error to a warning to get it off people's screens during boot, since most people will ALWAYS see it)
[12:55] <infinity> Oh, I see.
[12:55] <infinity> Nevermind.
[12:55] <infinity> This is... Interesting.
[12:57] <infinity> I'd bet on an ide-generic race, still.
[12:58] <infinity> (ie: ide-generic being loaded before, or during, the hptXXX load)
[01:02] <sistpoty> re
[01:03] <sistpoty> strange... adding acpi=off pci=noacpi did work, but again only once :(
[01:03] <infinity> Red herring, probably.
[01:04] <infinity> Can you try "echo 'blacklist ide-generic' > /etc/modprobe.d/blacklist-ide && update-initramfs -u"?
[01:05] <zul> heylo
[01:05] <sistpoty> infinity: sure
[01:06] <sistpoty> rebooting again
[01:06] <zul> hey infinity 
[01:08] <infinity> Yo.
[01:13] <sistpoty> infinity: works with ide-generic blacklisted (I rebooted twice, both times success)
[01:14] <infinity> Figured.
[01:14] <infinity> Can you update the bug to say so?
[01:14] <sistpoty> infinity: sure, will do
[01:15] <infinity> Then change the package to "udev", and reassign it to "scott-bugs@ubuntu.com"? :)
[01:15] <sistpoty> ok, will do
[01:15] <BenC> sweet
[01:16] <BenC> sorry, stepped out for a second
[01:16] <sistpoty> thx for your help BenC and infinity
[01:18] <BenC> jbailey: -9.11?
[01:18] <jbailey> BenC: Yes, I'm using it now.
[01:18] <BenC> if I boot straight to gdm, the keyboard only works for about 15 seconds and then stops
[01:19] <BenC> everything else works (mouse, UI, etc.)
[01:19] <BenC> if I boot to console, and then start gdm, everything is fine
[01:19] <jbailey> Weird.
[01:19] <infinity> That's pretty special.
[01:19] <jbailey> Not seeing that with my ppc64.
[01:19] <BenC> there's a bug report about it, so I know it isn't just me
[01:20] <zul> back later...hockey game
[01:20] <jbailey> zul: Who's playing
[01:20] <jbailey> ?
[01:20] <zul> ottawa/montreal
[01:20] <BenC> and I know it isn't a USB issue, since mouse is on the same bus
[01:20] <jbailey> zul: Oh.  I suppose I should thumb my chest and growl at you.
[01:20] <jbailey> zul: Consider it done. =)
[01:20] <zul> jbailey: wohoo :)
[01:20] <BenC> I think it's a timing issue between console keymap being loaded and gdm doing something with the keyboard aswell
[01:23] <infinity> The console keymap is loaded REALLY early (rcS.d/S05keymap.sh), so I don't see how it could conflict with gdm loading the X keymaps.
[01:24] <infinity> Oh look, jbailey crashed already.
[01:29] <jbailey> BenC: Sorry, X hung again.
[01:31] <BenC> infinity Oh look, jbailey crashed already.
[01:31] <BenC> :)
[01:32] <jbailey> =)
[01:32] <BenC> so rc6 doesn't fix it
[01:32] <BenC> infinity: well something after gdm is screwing it up
[01:33] <BenC> jbailey: and I have nothing special for ppc64 in my tree, it's all stock
[01:33] <BenC> but you could try a monolithic stock 2.6.15-rc6 just to be sure :/
[01:33] <BenC> who knows, maybe this is a gcc-4.0 issue
[01:55] <jbailey> BenC: You said that static klibc was working for you on ia64, right?
[01:55] <BenC> yes
[01:55] <BenC> I booted with it
[01:55] <jbailey> Cool.
[02:20] <zul> heyl
[02:25] <jbailey> zul: Did we win already?
[02:25] <zul> yes ottawa is winning 1-0
[02:26] <jbailey> Bah. =)
[02:26] <jbailey> Of course, I don't actually even know the name of the Montral hockey team. =)
[02:27] <infinity> Les Habitants.
[02:27] <jbailey> Err.
[02:28] <jbailey> "The people who live here"?
[02:28] <jbailey> Special. =)
[02:28] <infinity> (Some people will say it's the Montreal Canadians, but they're WRONG)
[02:28] <zul> canadiens
[02:29] <zul> they were always the canadiens in quebec
[02:30] <zul> heh...in ottawa we call them looser
[02:30] <jbailey> If we separate, I vote the name stays. =)
[02:30] <zul> loosers even..
[02:30] <jbailey> zul: It would certainly be hard to be tighter than someone from Ottawa...
[02:30] <jbailey> (Is this in violation of the CoC?)
[02:31] <zul> i am aghast that you said that
[02:31] <jbailey> terror-struck.
[02:31] <jbailey> Hmm
[02:31] <zul> impiling that im tight...which is true...but thats not the point
[02:32] <infinity> Ahh, they were always the Canadiens, the locals just CALL them Les Habitants.
[02:32] <infinity> Yay confusion.
[02:32] <zul> yay...back to the game
[02:32] <jbailey> infinity: Will you hate me if I move klibc-LiNeNoiSe.so to /lib?
[02:32] <jbailey> infinity: It seems to be where they're standardising putting it, even though all the apps are where we put them.
[02:32] <jbailey> It has the nice side effect of being able to test them on a running system.
[02:33] <infinity> jbailey : I don't care one iota, since we're removing it from initramfs anyway. ;P
[02:33] <jbailey> Lovely. =)
[02:33] <jbailey> I haven't gone through the effort of looking at that stuff yet.  Running out of time.
[02:33] <jbailey> But at least with my 1.1.8 upload it should work everywhere.
[02:33] <jbailey> Hell, even on alpha.
[02:34] <jbailey> Mithrandir: Around?
[02:34] <jbailey> ls
[02:37] <BenC> infinity: no more klibc in initramfs?
[02:38] <jbailey> BenC: He's not tight into my vision of getting rid of busybox eventually. =)
[02:38] <jbailey> infinity: Although I'm inclined to ask you to keep it in for now.
[02:38] <BenC> doesn't busybox have an option to compile against klibc?
[02:38] <jbailey> infinity: I've been thinking about buntu a bit more.
[02:38] <jbailey> Does it?
[02:38] <BenC> I saw that in the makefile
[02:38] <jbailey> Creepy.
[02:43] <infinity> Very creepy.
[02:44] <infinity> jbailey : I'm not tossing it tomorrow or anything.  It's not like it hurts US to have two libcs in there.
[02:44] <infinity> Our initramfs is fat, and who really cares?  No one.
[02:44] <jbailey> svenl. =)
[02:44] <infinity> But, eventually, a line needs to be drawn.  "reasonably small, and very functional", versus "really, really small, and not so functional"
[02:44] <infinity> And I have no qualms about having initramfs have two modes for that.
[02:45] <infinity> Basically, an "I detect that your setup will require glibc, so we'll copy that in, use the utils linked to glibc, and punt klibc out completely", or "Holy crap, you can build a tiny initramfs with just klibc, go you, let's do that then."
[02:46] <infinity> (The latter option being blacklisted on any arch where we think it may break, until all seem to be well-tested and reasonably maintained)
[02:46] <BenC> yeah, that sounds like the right plan
[02:47] <infinity> That means double-builds of anything we want in initramfs that supports klibc, but that's no big deal really.
[02:47] <BenC> just get rid of mkinitrd completely :)
[02:53] <BenC> maybe I was wrong about busybox
[02:53] <BenC> trying to remember what it was that could build with klibc
[02:53] <BenC> maybe it was udev
[02:53] <jbailey> Udev definetly.
[02:54] <jbailey> 'k.  klibc on ia64 hacked to work.
[02:54] <BenC> building static?
[02:54] <jbailey> Specifically installing static on ia64.
[02:55] <BenC> when will this be done?
[02:56] <jbailey> I have it in my tree right now.
[02:56] <BenC> just wondering when I can make the change to kernel-package :)
[02:57] <jbailey> The last thing I need to do before uploading is add some sort of test to it.
[02:58] <jbailey> I don't know how best to do that, though.  Hmm
[02:58] <jbailey> Suck program.
[02:59] <jbailey> I thin it's probably best to just build it everywhere and hope this time around.
[03:00] <jbailey> I'll get the patches in upstream to make testing not suck.
[03:05] <zul> 3-2
[03:19] <jbailey> BenC: ping
[03:19] <jbailey> ?
[03:19] <jbailey> It seems like asm-ppc doesn't contain errno,h
[03:19] <jbailey> Just want to confirm that asm-ppc is the new black.
[03:23] <BenC> asm-powerpc
[03:55] <fabbione> morning
[04:06] <zul> evening
[04:27] <BenC> hola
[04:29] <fabbione> hey Ben
[04:30] <fabbione> BenC: i have been checkign the amd64-generic kernel on i386
[04:30] <fabbione> we have two options to do it
[04:31] <fabbione> one is building it in the i386 upload
[04:31] <fabbione> the other is to do a little trick during amd64 build
[04:31] <fabbione> the latter takes about 5 secs of machine time
[04:32] <fabbione> the former is cleaner, but it introduce another flavour to build, with a i386 gcc crossbuilding to amd64 (that might introduce unwanted behaviour bugs)
[04:33] <fabbione> so i wanted to talk with you about what way you would prefer it done
[04:33] <fabbione> since i am doing other kernel stuff today, i might as well get around it
[04:35] <BenC> I see a problem with the first way
[04:36] <BenC> I looked at it, and it causes issues with kernel headers creation and such...we'd have to add some new handlers to the build for it to work
[04:37] <BenC> what is the benefit of it anyway?
[04:46] <fabbione> oh right.. kernel headers..
[04:46] <fabbione> i totally forgot about it
[04:46] <fabbione> i just did the image :/
[04:46] <fabbione> oh the benefit is that we skip one extra image to build on i386 and it takes about 5 secs to publish
[04:47] <fabbione> less config duplication
[04:47] <fabbione> but well you got a point i didn't think about
[04:47] <fabbione> so clearly it needs to be builded properly
[04:47] <BenC> well building it proper was the problem I was having :)
[04:48] <BenC> there's conflicts with cross building in our build system
[04:48] <fabbione> hmm ok
[04:48] <fabbione> i guess we need to get that fixed
[04:48] <BenC> it would take some new debian/rules crack to get it right
[04:49] <BenC> fabbione: did I tell you I ordered my 17" powerbook?
[04:49] <fabbione> BenC: ah cool.. no you didn't :)
[04:49] <BenC> should be here after the new year
[04:49] <fabbione> BenC: if the ppc installer in dapper doesn't work, i have breezy custom netboot images for it
[04:50] <BenC> ok
[04:50] <fabbione> it needs a modified .12 to go
[04:50] <BenC> have you tested flight livecd on your machine?
[04:50] <fabbione> somebody did kill halley..
[04:50] <BenC> flight2
[04:50] <fabbione> not yet
[04:50] <fabbione> i had no time to play with CD's yet
[04:50] <fabbione> i prefer netboot
[04:51] <fabbione> it's kind of a tradition that i test netinstalls while the others play with CD's
[04:52] <infinity> Oh, crap.
[04:52] <fabbione> GO HALLEY!
[04:52] <infinity> BenC : How far off is -9.12?
[04:52] <fabbione> WTF IS GOING ON WITH DC MACHINES?
[04:52] <infinity> BenC : I accidentally did a mass dep-wait clear, which freed up linux-meta to build on amd64. :)
[04:52] <fabbione> i can't fork anymore as i was used to
[04:53] <BenC> infinity: was going to upload within 30 minutes
[04:53] <infinity> BenC : Yay.  Then I won't feel too bad about it.
[04:53] <BenC> no ABI bump! :)
[04:53] <infinity> BenC : It's not like anyone canreliably dist-upgrade right now with dbus threatening to remove everything, so I doubt anyone will notice.
[04:54] <fabbione> speaking of dbus..
[04:54] <fabbione> i am going to lart somebody with that mono shit they did arch: all
[04:54] <fabbione> mono is not portable. mmmmmk?
[04:55] <infinity> Crazy talk, mono is the new Java, which means it runs EVERYWHERE*
[04:55] <infinity> * some conditions apply, check your retailer for details
[04:57] <jbailey> BenC: Is asm-ppc or asm-powerpc the right one?
[04:57] <BenC> asm-powerpc
[04:57] <infinity> powerpc
[04:57] <infinity> (why are we still shipping asm-ppc?)
[04:58] <BenC> for powerpc, some files are still in asm-ppc
[04:59] <BenC> ppc64 is totally converted to powerpc though
[04:59] <jbailey> $ ls -ald asm
[04:59] <jbailey> lrwxrwxrwx 1 root root 7 2005-12-20 21:10 asm -> asm-ppc
[04:59] <BenC> hmm
[04:59] <jbailey> BenC: Can you change the symlink pointer in linux-headers?
[04:59] <jbailey> I thought you had before, but I might be confused.
[04:59] <jbailey> But klibc will ftbfs on ppc from it.
[04:59] <BenC> is that linux-headers-2.6.15-9 or linux-headers-2.6.15-9-powerpc?
[05:00] <jbailey> no asm/errno.h
[05:00] <BenC> that's probably why
[05:00] <jbailey> jbailey@starshine:/usr/src/linux-headers-2.6.15-9/include$
[05:00] <BenC> linux-headers-2.6.15-9 doesn't use the right arch mapping
[05:00] <fabbione> it doesn't know about arch mapping iirc
[05:01] <fabbione> that's an arch: all pkg, isn't it?
[05:01] <jbailey> Filename: pool/main/l/linux-source-2.6.15/linux-headers-2.6.15-9_2.6.15-9.11_powerpc.deb
[05:01] <jbailey> Looks like no.
[05:01] <jbailey> Plus I count on the asm symlink being pointed right.
[05:01] <jbailey> Although I guess that could be set in the postinst
[05:02] <infinity> Ick.
[05:02] <fabbione> infinity: so what version of dbus i need to build to get out of that mess?
[05:02] <infinity> Real arch-mapping happens in the subflavour packages.
[05:02] <fabbione> ubuntu4?
[05:02] <infinity> The linux-headers generic packages should point the arch at the dpkg arch.
[05:02] <fabbione> or we need to wait a new one?
[05:02] <BenC> I see it
[05:03] <infinity> fabbione : Which mess?
[05:03] <fabbione> infinity: you said that dbus is attempting to deinstall *
[05:03] <fabbione> infinity BenC : It's not like anyone canreliably dist-upgrade right now with dbus threatening to remove everything, so I doubt anyone will notice.
[05:03] <infinity> fabbione : Oh.  Uhm.  Rebuild the whole archive against the current dbus?  mvo's been working on it.
[05:04] <jbailey> BenC: Do you want something in bugzilla from me for this?
[05:04] <infinity> Transitions are fun.
[05:04] <jbailey> Transitioning is fun.
[05:04] <jbailey> Err.
[05:04] <BenC> jbailey: nah, it's going to be fixed in -9.12, which will be uploaded in about 20 minutes
[05:04] <fabbione> infinity: amen.. the new d-bus is not even installable on sparc.. go figure
[05:04] <jbailey> BenC: Thanks.  I'll try klibc again tomorrow then,
[05:04] <infinity> jbailey : I like how that aligned nicely over here, due to our nicks being differing lengths.
 Transitions are fun.
 Transitioning is fun.
[05:05] <fabbione> i guess this is yet another xmas project..
[05:05] <infinity> Some sort of typographically-pretty jinx.
[05:05] <fabbione> get sparc to install again
[05:05] <jbailey> infinity: =)  I hadn't seen that.  x-chat start of line aligns.  Cool. =)
[05:06] <jbailey> Sleep time for now I guess.   demain!
[05:07] <fabbione> jbailey: night
[05:07] <infinity> 'Night.
[05:07] <BenC> good night
[05:08] <fabbione> time to build orgy the next security update
[05:09] <fabbione> oh weird http://bugzilla.ubuntu.com/show_bug.cgi?id=20771
[05:09] <fabbione> BenC: the guy did test the kernels with some reverted patches..
[05:10] <fabbione> i am curious to see if i revert all of them
[05:10] <fabbione> and just rebuild
[05:11] <BenC> odd results
[05:11] <fabbione> and the 2 removed patches are the most intrusive ABI killing ones
[05:11] <fabbione> exactly
[05:12] <BenC> fabbione: vtun is  up if you want to do any builds on the e3k
[05:15] <fabbione> BenC: cool thanks but i alredy did 9.11 here
[05:15] <BenC> 9.12 will be up soon, so you can do that (although, it's already built here :)
[05:16] <fabbione> ehhe
[05:16] <fabbione> ok
[05:16] <fabbione> if you can copy the sources you upload into /org/chroots i can build it in parallel
[05:16] <BenC> I do the sources from concordia
[05:17] <fabbione> oh ok
[05:17] <fabbione> i will wait it hits the archive
[05:17] <fabbione> i am in no hurry and woke up 2 hours before usual time
[05:17] <BenC> any chance I could get you to ^Z your build on concordia just till I get this source built?
[05:17] <fabbione> oh sure..
[05:18] <BenC> thanks
[05:18] <fabbione> i did slow it down :)
[05:18] <fabbione> but i have most of it in ccache
[05:18] <fabbione> so it shouldn't hurt you too much
[05:21] <infinity> mjg59 : I'll give you my firstborn if you fix acpi-support on ia64.
[05:22] <BenC> the only problem I have with acpi on ia64 is that it gets misrouted irq's and disables the acpi interrupt
[05:22] <BenC> well, that's my only current problem :)
[05:23] <BenC> if the irq worked, no telling what else might be wrong
[05:24] <fabbione> oh christ
[05:24] <fabbione> pitti did upload the 348t32 lang-packs
[05:33] <BenC> and with that, good night
[05:34] <fabbione> good night BenC 
[05:43] <infinity> BenC : Cheers, g'night.
[06:07] <lamont> BenC: if you want, I can give you a login on the hppa machine, where you can harvest abi files
[06:14] <lamont> BenC: new hppa kernel uploaded
[06:14] <lamont> well, 9.11
[06:15] <fabbione> that will be rejected since .12 is in
[06:16] <lamont> sigh
[06:17] <fabbione> i think i just crashed concordia
[06:17] <fabbione> or it's extremely slow
[06:18] <fabbione> the latter..
[06:23] <infinity> lamont : Also, "HA, HA!"
[06:23] <lamont> infinity: ??
[06:27] <infinity> (your kernel binaries being REJECTED)
[06:27] <lamont> PLBBBBTT
[06:29] <lamont> sigh.  if (a) forward-to-hp else if (b) forward to hp.
[06:29] <lamont> I think B wanted to be different.
[06:29] <lamont> next kernel should make it
[06:30] <fabbione> night lamont
[06:33] <infinity> Argh.
[06:33] <infinity> Guys, kill those builds. :)
[06:33] <cjb> !
[06:33] <fabbione> infinity: ?
[06:35] <fabbione> infinity: what's wrong with .12?
[06:36] <infinity> Missing /build/buildd/linux-source-2.6.15-2.6.15/debian/abi/2.6.15-9.11/amd64/amd64-server file.
[06:36] <infinity> make: *** [clean]  Error 1
[06:37] <fabbione> oh
[06:37] <infinity> Yeha, "oh".
[06:37] <fabbione> let me check if it's in git
[06:37] <infinity> One would assume not. :)
[06:37] <fabbione> never say never
[06:37] <infinity> Ben's script to auto-grab the ABI files probably wasn't updated to grab the ABI for the new flavour.
[06:39] <fabbione> right
[06:39] <fabbione> it's not there at all
[06:40] <fabbione> do you have a git tree handy?
[06:40] <infinity> Alright, well, assuming you have more bandwidth than I do (you do), you want to grab the amd64-server ABI stuff, update the source, and upload -9.13?
[06:40] <fabbione> infinity: i can do it, yes..
[06:40] <infinity> Well, I can do it just fine.
[06:40] <fabbione> but it's much simpler to do it on concordia anyway
[06:41] <fabbione> or chinstrap
[06:41] <infinity> It'll just take a while for me to upload. :)
[06:41] <fabbione> why?
[06:41] <fabbione> just sign the .dsc and .changes locally
[06:41] <infinity> Oh, true.  I could do it remotely, and just grab the .changes.
[06:41] <fabbione> do the work remote :)
[06:41] <fabbione> score
[06:41] <fabbione> :)
[06:41] <infinity> Alright, I'll do it then.
[06:41] <fabbione> ok
[08:39] <fabbione> doko: ping?
[08:56] <CataEnry> hi all
[10:08] <infinity> Oh, FFS.  I give up.  FTBFS on amd64 again.
[10:09] <infinity> I knew I should have tested the build.
[10:12] <fabbione> how does it fail?
[10:13] <infinity> It's the new bit Ben added for KPKG_ARCH during headers install.
[10:13] <infinity> export KPKG_ARCH=$(grep "^headers-amd64[[:space:] ] " debian/config/archmap | grep -v ^# | awk '{print $2}'); \
[10:13] <infinity>         cd debian/build/linux-source-2.6.15; \
[10:13] <infinity>                  HEADER_CLEAN_HOOK=/build/buildd/linux-source-2.6.15-2.6.15/debian/header-install \
[10:13] <infinity>                          make-kpkg --stem linux  --append_to_version -9 kernel-headers
[10:13] <infinity> unknown Debian architecture x86_64, you must specify GNU system type, too at /usr/bin/dpkg-architecture line 214.
[10:13] <infinity>  /usr/share/kernel-package/rules:1652: *** Error. I do not know where the kernel image goes to [kimagedest undefined]  The usual case for this is that I could not determine which arch or subarch tihs machine belongs to. Please specify a subarch, and try again..  Stop.
[10:13] <infinity> make[1] : Leaving directory `/build/buildd/linux-source-2.6.15-2.6.15/debian/build/linux-source-2.6.15'
[10:13] <infinity> make: *** [binary-debs]  Error 2
[10:14] <fabbione> ah yeah
[10:14] <fabbione> i know how to fix that
[10:15] <infinity> Yeah, as do I.  Just frustrating to see it fail during install.
[10:15] <fabbione> it's a missing line in debian/config/archmap
[10:15] <infinity> Too bad lamont's not around to feed me his ABI files, so I can include them in the next upload.
[10:15] <fabbione> i could build on hppa
[10:16] <fabbione> if you can give me the time to do so
[10:16] <infinity> Sure, go nuts.  Any of .11, .12, or .13 would be fine, so I can get the ABI file for -9
[10:16] <infinity> s/file/files/
[10:17] <fabbione> oh fuck
[10:17] <fabbione> i need to resync again the local mirror
[10:18] <fabbione> yeah i am on it..
[10:18] <fabbione> build test amd64 in the meanwhile

[10:18] <infinity> You're not abusing concordia, are you?
[10:18] <fabbione> i am only at -j10
[10:18] <fabbione> you can take the rest :)
[10:18] <infinity> (Zofia's using her amd64, so I can't)
[10:18] <fabbione> it's almost all in ccache
[10:19] <fabbione> so cpu is free
[10:19] <infinity> "only -j10"... Feh.
[10:22] <fabbione> building now
[10:42] <fabbione> ?
[10:42] <infinity> I think Ben was confusing kernel arch and KPKG_ARCH here.
[10:43] <infinity> KPKG_ARCH for x86_64 is amd64, right?
[10:43] <fabbione> yes
[10:43] <fabbione> look at the others :)
[10:43] <infinity> And KPKG_ARCH for parisc is hppa.
[10:44] <fabbione> hmmm
[10:44] <fabbione> no the other way around
[10:44] <infinity> Yay sanity.
[10:44] <fabbione> i think
[10:44] <infinity> No.
[10:44] <fabbione> i can't remember
[10:44] <infinity> Kernel arch != KPKG_ARCH.
[10:44] <infinity> At least, not the way they're used here.
[10:44] <fabbione> i know
[10:44] <infinity> parisc is definitely the kernel arch. :)
[10:44] <fabbione> right
[10:44] <infinity> (As is x86_64)
[10:45] <fabbione> amd64-server    amd64
[10:45] <fabbione> this is correct
[10:45] <infinity> So, yeah.  I think I'll fix up the generic headers bit to be 3 columns.
[10:46] <infinity> Actually, not needed, since the 3rd column would match $(arch) anyway.
[10:46] <infinity> But I guess, on the off chance more kernels get added where $(arch) != KPKG_ARCH, I should do it right.
[10:46] <infinity> Did sparc build?
[10:47] <infinity> It should fail in the same spot, unless both sparc and sparc64 are valid values for KPKG_ARCH.
[10:47] <fabbione> 9.11 is in the archive.. 
[10:47] <fabbione> the ABI files are in .12
[10:47] <infinity> 9.12 is where this "fix" was added.
[10:47] <fabbione> ah
[10:47] <infinity> So if 9.12 or 9.13 built on sparc, my hypothesis is shot.
[10:47] <infinity> But I think it should fail, same as it did on amd64.
[10:47] <fabbione> i dunno... i didn't even try to build it
[10:47] <infinity> Heh.
[10:47] <fabbione> since i know you are working on .14
[10:48] <infinity> How fast is the e3k?
[10:48] <fabbione> it depends..
[10:48] <fabbione> when it doesn't crash is fast
[10:48] <fabbione> but afaik ben did build .12 on the e3k
[10:49] <fabbione> infinity: may i suggest for the sake of mirrors.. to stop here and let ben fix .14?
[10:49] <fabbione> also because all these changes are not making it into git
[10:51] <infinity> I was going to git them just nowish.
[10:54] <infinity> I'm fine with holding off too, though.  Mainly cause I can't test sparc, though from my reading of kernel-package, I'm pretty sure of how this SHOULD be done.
[10:55] <fabbione> let's hold it
[10:55] <fabbione> it's not the end of the world
[10:56] <fabbione> but if you really want to build kernel pkgs.. i can send you -security :)
[10:56] <infinity> Heh.
[10:57] <infinity> Nah, the only reason I wanted to get it in was because -9 has still not built successfully on amd64, but -meta has (oops), so it's uninstallable.
[10:57] <infinity> Not really earth-shattering.
[10:57] <infinity> If you can finish off that hppa build, though, we can get the hppa ABI stuff in, and I'll tidy up some stuff locally here, and Ben can just pull.
[10:57] <fabbione> linux-image-2.6.15-9-amd64-server_2.6.15-9.11_amd64.deb
[10:57] <fabbione> dude..
[10:57] <fabbione> it's there
[10:58] <fabbione> yeah hppa is building
[10:58] <infinity> Oh, sorry.  Not "-9 hasn't built", rather ">= -9.12 hasn't built"... The headers in -9.11 were broken, so LRM couldn't build.
[10:58] <fabbione> ahh ok
[10:59] <fabbione> i should also go out to buy console cables
[10:59] <fabbione> i am waiting for warty/amd64 to finish to build
[10:59] <fabbione> so i can slam hoary on *
[11:34] <doko> fabbione: pong
[11:34] <fabbione> doko: i did try to add the amd64-generic kernel to i386
[11:35] <fabbione> but i am afraid it's not exactly doable
[11:35] <fabbione> and it's very very hackish
[11:35] <fabbione> i will give it another shot after i finish the security updates
[11:35] <fabbione> but i can tell you that it doesn't even compile the first .c
[11:38] <doko> hmm, strange.
[11:39] <fabbione> not really
[11:39] <fabbione> arch/x86_64/Makefile doesn't understand CROSSCOMPILE
[11:40] <fabbione> and forcing HOSTCC='gcc -m64
[11:40] <fabbione> makes everything cry
[11:40] <fabbione> again.. i need to check it more
[11:40] <fabbione> this was a fast test.. it might be other stuff that needs to be modified
[11:40] <fabbione> like make-kpkg
[11:48] <infinity> BenC : I have set myself up on rookery according to the wiki.  Pull at will.
[01:25] <doko> fabbione, BenC: is there a way to look at debian kernel changelog and say: this upload fixes remote exploits / local exploits (for the breezy-security uploads)?
[01:26] <fabbione> eh?
[01:26] <fabbione> isn't zcat /usr/share/doc/linux-image-$(uname -r)/changelog.Debian.gz enough?
[01:29] <doko> that's not the point. as a sysadmin I may know my local users and schedule a downtime for a reboot. if there are known remote exploits I should be a bit quicker. at a first glance from the changelog, I can't tell if the upload fixes remote exploits
[01:31] <infinity> Which is why, as a sysadmin, you sibscribe to ubuntu-security-announce, which gives you more detail on this (and other exploits that aren't the kernel)?
[01:33] <fabbione> doko: exactly. you look at the u-s-a
[01:34] <fabbione> last time we did try to alling changelog with USN it was a mess of confusion
[01:34] <fabbione> because USN is written after
[01:34] <fabbione> with very different working
[01:34] <fabbione> wording
[01:34] <fabbione> if you are a sysadm with that kind of paranoia, you also know where to find mitre
[01:37] <jbailey> BenC: ping?
[01:44] <BenC> jbailey: pong
[01:45] <fabbione> hey Ben
[01:45] <infinity> Morning, Ben.
[01:45] <infinity> BenC : Pull from my git repo on rookery to see how muhc I hate you.
[01:45] <BenC> good morning
[01:45] <BenC> heh, ok
[01:46] <infinity> (And how much I hate myself for not testing my first change..)
[01:49] <jbailey> g'morning Ben!
[01:49] <jbailey> BenC: The asm symlink still seems to be wrong.
[01:49] <jbailey> In -9.13
[01:49] <BenC> hmm
[01:50] <infinity> jbailey : Where does it point?
[01:50] <jbailey> asm-ppc
[01:50] <infinity> Woo.
[01:50] <jbailey> ii  linux-headers- 2.6.15-9.13    Header files related to Linux kernel version
[01:50] <jbailey> jbailey@starshine:/usr/src/linux-headers-2.6.15-9/include$ ls -ald asm lrwxrwxrwx 1 root root 7 2005-12-21 07:09 asm -> asm-ppc
[01:50] <infinity> So the part of -9.12 that broke (and I've fixed in git) is probably still broken anyway.
[01:51] <fabbione> SCORE
[01:51] <infinity> (broke, as in "failed to build on amd64")
[01:51] <BenC> figures my last minute change broke it
[01:51] <fabbione> you guys ROCK SO MUCH
[01:51] <fabbione> ;O
[01:51] <fabbione> BenC: i was wise not to burn your e3k to build .12 :)
[01:51] <fabbione> btw.. hppa ABI files are almost ready
[01:52] <infinity> Yay!
[01:52] <fabbione> i must sleep
[01:52] <fabbione> be back in 2 hours
[01:52] <infinity> Have a good nap.
[01:53] <fabbione> thanks
[01:54] <BenC> so I guess -9.14 will be uploaded in a few minutes :)
[01:54] <jbailey> BenC: \o/
[01:54] <BenC> later fabio
[01:55] <infinity> BenC : You found the symlink issue?
[01:55] <BenC> I thought your fixes did that?
[01:55] <infinity> No, jeff just told me it was wrong.  I just fixed your horrible FTBFS on amd64.
[01:55] <fabbione> BenC: if you can wait that i am back, i can give you the hppa abi files
[01:55] <infinity> My fix won't change a thing on powerpc.
[01:55] <fabbione> up to you
[01:55] <fabbione> they are almost done
[01:56] <BenC> I have the ones for .12
[01:56] <BenC> I saved my final builds this time
[01:56] <fabbione> ok
[01:56] <BenC> so I was just missing amd64
[01:56] <infinity> Oh, spiffy.
[01:56] <jbailey> I can cheerfully burn my morning on support calls and glibc if you need time on the kernel.
[01:56] <fabbione> and takes off
[01:56] <jbailey> I  just want to rid the world of initrd-tools.
[01:56] <infinity> You and me both.
[01:57] <jbailey> infinity: I don't consider this an exclusive club.. =)
[01:57] <fabbione> jbailey: i think i know why .15 did install fine on ia64
[01:57] <fabbione> jbailey: i will confirm my theory later
[01:57] <infinity> I suppose we could just violently set it manually, if we can't wokr out why it's being broken.
[01:57] <jbailey> fabbione: Cool. Don't burn any time on it.  I have a working klibc now.
[01:59] <BenC> infinity: I think your change will fix it
[02:00] <infinity> It shouldn't.
[02:00] <BenC> if it doesn't, it means that kernel-package needs fixing
[02:00] <infinity> powerpc should end up in precisely the same boat.
[02:00] <BenC> your change should fix amd64 failure
[02:01] <infinity> yes, it definitely will do that.
[02:01] <BenC> if we pass powerpc to kernel-package, it should DTRT
[02:01] <infinity> We already did.
[02:01] <infinity> So, it's broken. :)
[02:02] <infinity> Which means a kernel-package bughunt and upload before -9.14, if you don't want to see jbailey cry.
[02:07] <BenC> make prepare creates the symlink
[02:07] <BenC> which is in debian/rules:build of the linux-source build
[02:08] <infinity> Which means the KPKG_ARCH stuff later on was meaningless. :)
[02:10] <BenC> and the symlink in debian/tmp-headers is ok
[02:10] <infinity> !
[02:15] <BenC> the link in the .deb is right too
[02:15] <infinity> Uhm.
[02:16] <BenC> heh, the link in my 9.12 .deb is right
[02:16] <infinity> Oh ,christ, the postinst sets it.
[02:16] <infinity> $architecture = "ppc" if $architecture eq "powerpc";
[02:16] <infinity> There's the fix.
[02:17] <BenC> rofl
[02:17] <infinity> I'll upload a shiny new kernel-package right now.
[02:17] <infinity> May as well leave all the current hacks in now, if they work.
[02:17] <infinity> (and to save face) :)
[02:18] <BenC> yeah, I'll build-dep on this new kernel package
[02:19] <infinity> Does 'uname -m' now say powerpc as well?
[02:19] <infinity> Like, is ppc completely dead?
[02:19] <jbailey> jbailey@starshine:~$ linux32 uname -m
[02:19] <jbailey> ppc
[02:21] <infinity> Hrm.  Should I conditionalise this to a specific kernel version?
[02:21] <infinity> Or do we care about make-kpkg working on older kernels?
[02:22] <jbailey> I suspect it changes enough with initramfs tweaks, etc. that it's unlikely to work right on older kernels anyway.
[02:22] <infinity> Fair enough.
[02:22] <infinity> kernel-package AWAY.
[02:23] <infinity> BenC : 9.001ubuntu10 for your build-dep.
[02:24] <BenC> you uploaded?
[02:24] <infinity> Yup.
[02:24] <BenC> did you get the one for src.postinst aswell?
[02:24] <infinity> Yup.
[02:24] <BenC> ok
[02:24] <infinity> rgrep loves me.
[02:24] <BenC> it's a race then, because I uploaded too :)
[02:25] <infinity> Hah.
[02:25] <infinity> You probably beat me, due to my shitty upstream. :)
[02:25] <infinity> I started the upload 5 minutes ago, finished when I gave you the version number.
[02:26] <infinity> Oh, I win.
[02:26] <infinity> One REJECT for Ben.
[02:28] <infinity> Hrm, is there a way to avoid this "foo same as branch bar" on every "git pull"?
[02:28] <infinity> (Or am I not supposed to be using "pull" to keep up to date?... Yay, newbieism)
[02:28] <BenC> you can delete everything from .git/refs/heads except origin and master
[02:34] <BenC> damnit
[02:34] <BenC> concordia is buckling under fabio's build
[02:34] <BenC> gonna take a bit to get the upload ready
[02:34] <infinity> Big shock there.
[02:35] <infinity> If you use -j50, you'll fork hard enough to steal some of his -j10, probably.. <smirk>
[02:35] <BenC> only thing my build is doing is dpkg-deb --build :)
[02:35] <infinity> Anyhow, as long as you remember all the ABI files this time, it should be good to go.
[02:35] <BenC> he's doing -j150 :)
[02:36] <infinity> You're kidding..
[02:36] <BenC> no, load is 170
[02:36] <CataEnry> hi all :)
[02:36] <infinity> Fabio : concordia is a SHARED MACHINE, dude.
[02:37] <BenC> infinity: do you have the hppa 9.13 abi files?
[02:37] <BenC> guess I can use the .12 ones
[02:37] <infinity> BenC : Unless something when horribly wrong, cp -a 12 13
[02:37] <infinity> s/when/went/
[02:37] <infinity> I did that on the arches where 13 didn't build.
[02:37] <zul> heylo
[02:37] <infinity> (Actually, I did it for all arches, but diffed the ones that did build)
[02:47] <BenC> .12 only failed to build on amd64, correct?
[02:47] <BenC> or .13 I mean
[02:56] <infinity> Both, even.
[02:56] <infinity> For different reasons.
[02:58] <BenC> there she blows
[02:59] <BenC> hopefully this isn't a total flop like .12 :)
[02:59] <infinity> And 13...
[02:59] <infinity> Flopping is fun.
[02:59] <BenC> it will be dep-wait for awhile though
[02:59] <BenC> atleast kernel-package is arch-indep, so it doesn't have to build everywhere :)
[02:59] <infinity> It's already uploaded.
[03:00] <infinity> Will be in the :03 cron.daily.
[03:00] <infinity> Which is when linux-source gets in.
[03:00] <infinity> So no dep-wait.
[03:00] <BenC> cool
[03:00] <zul> oi...how many kernels are you going upload this week?
[03:00] <infinity> 37.
[03:00] <infinity> Ish.
[03:00] <BenC> I've done more uploads today for ubuntu than I've done for Debian in the past 3 years
[03:00] <zul> lol
[03:01] <zul> *sigh* i need openswan or freeswan now
[03:02] <zul> freaking work..
[03:30] <infinity> BenC : Either I didn't push correctly, or you didn't pull my debian/rules changes...
[03:31] <infinity> Oh, or I'm looking at the log for -9.13
[03:31] <infinity> Christ, that scared me.
[05:05] <zul> wheee...nagios you buggy pita
[05:42] <BenC> yay, git 1.0.0 released
[05:45] <zul> yahoo..
[05:48] <pappan> BenC: hi
[05:56] <jbailey> And now for my next magic trick:  Recovering my box now that it seems unwilling to load the USB keyboard drivers!
[06:03] <jbailey> Nope, it seems to be udev that's lost it.
[06:03] <jbailey> *sigh*
[06:03] <jbailey> No network card either.
[06:04] <BenC> is udev giving you an error?
[06:04] <BenC> somethng about get_message failing?
[06:07] <pappan> BenC: hi
[06:07] <BenC> hello
[06:08] <pappan> BenC: i am a newbie to kernel programming
[06:08] <jbailey> BenC: It's all hidden by usplash.
[06:08] <jbailey> BenC: I don't think I can turn that off from yaboot.
[06:09] <BenC> jbailey: debug might do ti
[06:09] <pappan> are there anything in the TODO list for which i may be eligible for
[06:09] <jbailey> BenC: 'kay, lemme try that.
[06:09] <jbailey> Right now it's sitting at an "X can't load" prompt. =)
[06:10] <jbailey> Hmm
[06:10] <BenC> pappan: most of the bug reports we have left are either waiting for more information (need to be verified against dapper flight2) or very complicated
[06:10] <jbailey> The other thing I could do is maybe use a recovery CD to force the network card to load so that I at least have ssh.
[06:10] <jbailey> BenC: Would that error go into a log file at all?
[06:10] <pappan> ok 
[06:11] <BenC> jbailey: it should, I can't remember
[06:11] <BenC> for my sparc64 it failed to moun the rootfs because it didn't load sd_mod
[06:11] <BenC> I think maybe scsi is builtin on ppc though
[06:11] <jbailey> 'kay.  I don't think debug will be enough to kill usplash.
[06:12] <jbailey> Oh, it's definetly making it as far as X almost loading.
[06:12] <jbailey> But the usual plug things: network, usb aren't getting loaded at all.
[06:19] <lamont> so what abi files do we need for hppa, anyway?
[06:20] <fabbione> lamont: none.. we managed to get them bulded in 3 places :)
[06:20] <lamont> sigh
[06:20] <lamont> at any rate, the signing should be dealt with for linux-*
[06:21] <lamont> if we find more excess-length binary: lines, I'll have to add them to the workaround
[06:21] <fabbione> i think the kernel is the only left
[06:21] <fabbione> X is not monocrappic anymore
[06:27] <jbailey> lamont!  I need your help sometime soon. =)
[06:28] <jbailey> BenC: 'kay.  I hardcoded the network driver in /etc/modules so I can poke at the machine in any way that you care for me to.
[06:28] <lamont> jbailey: sup
[06:28] <lamont> ?
[06:28] <BenC> jbailey: sounds like a udev issue, so I suggest checking that first
[06:28] <jbailey> lamont: Can you point me to a good resource on ia64 asm coding?
[06:29] <BenC> jbailey: stop udevd and start it in debug mode
[06:29] <jbailey> BenC: I agree.  I'll dig through there.
[06:29] <lamont> jbailey: I know they exist... I've been meaning to track me down a copy....
[06:29] <jbailey> lamont: There are some constructs that don't make any sense to me that appear to be loading a register with the contents of itself.
[06:32] <BenC> jbailey: maybe it is loading the registers with the contents of the memory where the register points to?
[06:32] <lamont> jbailey: Itanium Architecture for Programmers, by James S Evans and Gregory L Trimper
[06:32] <jbailey> BenC: Entirely possible.  I just hate guessing. =)
[06:33] <jbailey> lamont: Do you know if it's soft-copy available, or just DT format?
[06:33] <lamont> dunno
[06:33] <lamont> DT format?
[06:33] <jbailey> DeadTree
[06:33] <lamont> ah, ok
[06:34] <lamont> I imagine that digging around the intel and/or HP sites should uncover a few online docs on the subject, but I expect that much of the stuff is still someone's revenue generator, and therefore DT-only
[06:35] <jbailey> Mmm, yeah.
[06:35] <jbailey> Hmm, I could check safari, too.  I had forgotten about that.
[06:35] <jbailey> Since I dont' imagine that the demand for people who can code ia64 asm is all *that* high, I'm not inclined to spend alot of money to just fix klibc. =)
[06:36] <lamont> "intel will send him the reference manuals for free
[06:36] <lamont> "lamont: or you can get PDFs from the website
[06:36] <jbailey> Ooo, do they still do that?
[06:36] <jbailey> COol.
[06:36] <lamont> apparenlty
[06:38] <jbailey> lamont: Thanks. =)
[06:40] <jbailey> Interesting.  Something seems to be causing udevd to die.
[06:41] <jbailey> when I removed the /dev/.udev/queue directory, restarted udevd and reran udevplug, it all went fine.
[06:42] <jbailey> Given that my 'old' image is pointing to -8, which did work, I think it's not a kernel problem, but something got eaten in the power failure.
[06:43] <jbailey> Anyhow, will hack later.  Need food and then have customers to call.
[06:50] <fabbione> BenC: the diffstat is not difficult to get
[06:50] <zul> customers whats that?
[06:54] <BenC> fabbione: give me a few minutes
[06:54] <fabbione> yes
[06:54] <fabbione> i need to smoke and soon to stop
[06:54] <fabbione> it has been a way too long day
[06:55] <fabbione> oh
[06:55] <fabbione> there is a 4th point
[06:55] <fabbione> all of them are against breezy
[06:55] <fabbione> nobody reported against hoary/warty
[06:55] <fabbione> not that somebody is still using them.. but well
[07:10] <fabbione> BenC: http://people.ubuntu.com/~fabbione/diffstats
[07:10] <fabbione> from that diffstat is only missing the sparc fix from davem
[07:10] <fabbione> but that's sparc specific
[07:10] <fabbione> so it can't influence k7/k8
[07:11] <fabbione> and 2 most intrusive patches have been backed out for testing
[07:11] <fabbione> one at a time.. so it might also be a combinantion
[07:15] <fabbione> BenC: i will be back after dinner
[07:15] <fabbione> in an hour or two
[07:15] <BenC> ok
[07:15] <fabbione> i didn't spend 5 minutes with my wife all day
[07:17] <zul> she must be cranky
[07:51] <jbailey> \o/ the asm link is now correct. =)
[07:52] <BenC> thank goodness
[07:54] <jbailey> HATE KBUILD
[07:59] <jbailey> BenC: Do you know the people working on include/asm-powerpc?
[08:00] <jbailey> BenC: asm-powerpc/signal.h isn't usable for user application because types.h defines BITS_PER_WORD only ifdef KERNEL
[08:02] <jbailey> BenC: It looks like the simplest thing is to patch it to just define it as 32 or 64 depending on __powerpc64__
[08:04] <jbailey> The other option is that I could get around to fixing lkh to Just Work with klibc.  But that's a multi-day project.
[08:08] <fabbione> BenC: i just published 6 kernels on people
[08:09] <fabbione> each of them add one security patch on top of -9-
[08:09] <fabbione> i think that's the easiest
[08:09] <fabbione> 6 is basically -9.23
[08:09] <fabbione> 1 is -10.24 minus one patch
[08:33] <BenC> jbailey: if you have a patch to fix it up, I'll take it
[08:34] <BenC> fabbione: ok
[08:35] <jbailey> BenC: Well, if you know the people doing the work, it would be nice to ask them what they intend.  But otherwise, I can punt you a patch that will work.
[08:43] <BenC> benh I suspect
[10:21] <zul> laster