[02:43] <zul> heylo
[02:46] <zul> yummy...spam
[05:03] <fabbione> morning
[05:09] <jbailey> Eh... You're early...
[05:10] <fabbione> yes
[05:10] <fabbione> i couldn't sleep this night?
[05:10] <jbailey> Dont confuse me like that.  It's still an hour until bedtime. =)
[05:10] <fabbione> jbailey: any news about binutils?
[05:10] <fabbione> ehhehe
[05:10] <jbailey> fabbione: No, last I saw was the elmo told doko that he was working on an update for it.
[05:11] <fabbione> ok...
[05:11] <jbailey> That was last Friday or so, though.
[05:11] <fabbione> yes i recall that
[05:11] <jbailey> I was offline the whole weekend, pretty much.
[05:11] <fabbione> so was i
[02:16] <zul> i got a freaking song in my head
[02:25] <chmj> what song ?
[02:26] <jbailey> zul: Sean Paul? =)
[02:26] <jbailey> "let's get it on 'till the early morn..."
[02:33] <zul> hell no..
[02:34] <zul> its like "we are on drugs...we are alll on drugs...do you like drugs..." cant remember where i heard it though
[03:55] <fabbione> yo
[03:57] <zul> hey fabbione how goes it?
[03:58] <fabbione> zul: as usual :)
[03:58] <fabbione> you?
[03:58] <zul> good...laptop is suppose to arrive today
[04:00] <fabbione> ehehe lucky you
[04:05] <zul> the irony is that my package is probably the last one to be deliviered
[04:07] <fabbione> zul: why?
[04:07] <zul> just my luck and the way the day is going so far
[04:07] <fabbione> ahhh
[04:46] <BenC> fabbione? 
[04:51] <zul> its smaaaaaaallllll
[04:53] <mjg59> zul: Got it, then?
[04:54] <fabbione> BenC: ?
[04:59] <BenC> hey
[04:59] <BenC> how long will you be around?
[04:59] <BenC> also, is your OCFS2 stuff done and ready in the repo?
[05:00] <fabbione> BenC: the last update is in baz.
[05:01] <fabbione> i will be around for another 2/3 hours
[05:01] <fabbione> BenC: we will need another update later anyway
[05:02] <fabbione> BenC: http://oss.oracle.com/bugzilla/show_bug.cgi?id=544 and http://oss.oracle.com/bugzilla/show_bug.cgi?id=558
[05:02] <desrt> BenC; i just posted a patch to bugzilla... it'd be really awesome if it was in the next build
[05:02] <fabbione> BenC: otherwise we are fine imho
[05:02] <fabbione> BenC: upstream is working on it right now and we have a test setup at the DC.
[05:02] <desrt> ubuntu's inotify is broken
[05:02] <fabbione> BenC: there is also 2.6.12.6 out...
[05:02] <BenC> yeah
[05:03] <fabbione> desrt: what's broken with it?
[05:03] <desrt> fabbione; it reissues watch descriptors
[05:03] <desrt> fabbione; due to an off-by-one error (one-liner patch)
[05:03] <BenC> desrt: bug #?
[05:03] <desrt> http://bugzilla.ubuntu.com/show_bug.cgi?id=14364
[05:03] <fabbione> desrt: ah ok.. is the fix upstream already?
[05:03] <desrt> yes
[05:03] <desrt> in 2.6.13
[05:03] <fabbione> perfect
[05:04] <desrt> robert love also just personally looked over the bug report and told me it looks OK
[05:04] <BenC> fabbione: that's a question I had, how do we move to new kernels, if at all?
[05:04] <BenC> are we at a stage where breezy is just going to be 2.6.12?
[05:04] <fabbione> BenC: yes.. breezy is going to be .12
[05:04] <fabbione> BenC: Preview Freeze is in 2 days
[05:05] <fabbione> so i'd say we talk with mdz before uploading this kernel
[05:05] <desrt> spooky
[05:05] <BenC> before uploading 2.6.12-8.12?
[05:05] <BenC> yeah, very
[05:05] <fabbione> mdz: BenC and I were wondering if we can upload 2.6.12-8
[05:06] <fabbione> or do you prefer us to wait after Preview Freeze?
[05:06] <BenC> here's the thing, I want to get 2.6.12-8.12 uploaded because of the recent ABI changes
[05:06] <fabbione> BenC: we should get the inotify fixes and 2.6.12.6 in too
[05:06] <BenC> yeah
[05:06] <fabbione> since we break the ABI, it's a good idea to push the crack in
[05:07] <mdz> I would prefer that you do it now
[05:07] <BenC> I was thinking of holding off for some more fixes to go in (mjg sent me a lot yesterday)
[05:07] <mdz> do matthew's fixes change the ABI still further?
[05:07] <BenC> so do the 8.12 now, and then 8.13 in a few days will be ok?
[05:07] <BenC> no
[05:07] <mdz> yes, that's fine
[05:08] <mdz> though I'm considering doing a Colony on thursday
[05:08] <mdz> so not thursday
[05:08] <fabbione> BenC: if you commit everything in today i can do the ABI dance tomorrow as soon as i wake up
[05:08] <BenC> a lot of these fixes would be nice if you are doing a Colony
[05:08] <fabbione> that would make us gain 12 hours
[05:08] <fabbione> and make Colony for thursday
[05:08] <BenC> yeah, we can tag team it
[05:09] <fabbione> mdz: would that work for you?
[05:09] <BenC> what time do you wake up, I really want to do most of the next upload to get the system down, but have you around to answer questions
[05:10] <fabbione> BenC: it depends.. usually around 4 UTC
[05:10] <fabbione> but we need to get elmo or Kamion or mdz around to NEW the packages
[05:10] <fabbione> + we need to prebuild them everywhere to get the new ABI
[05:10] <fabbione> brb
[05:10] <BenC> so that's around midnight for me
[05:11] <BenC> I can do that
[05:11] <desrt> mmm.  thanks for the fast response, guys :)
[05:12] <BenC> desrt: I see the patch, and it looks good to me
[05:12] <mdz> fabbione: what time UTC would that be?
[05:12] <desrt> BenC; :)
[05:12] <fabbione> mdz: i wake up at 4 UTC....
[05:12] <desrt> cheers
[05:12] <fabbione> mdz: so if BenC will do everything and i only need to be around.. 
[05:13] <fabbione> mdz: i'd say we can finish in the usual 4/6 hours if there is an ftp-master around
[05:13] <fabbione> BenC: if you want to upload all yourself, we can gain quite a lot of time...
[05:13] <BenC> main thing I need help with is creating the source package, and make sure I have put everythign together right
[05:14] <fabbione> BenC: that's the easiest part :)
[05:14] <BenC> yeah, but do I need to kill off any baz stuff in the debian tree?
[05:14] <fabbione> BenC: that's even easier ;)
[05:15] <fabbione> BenC: you got the source package with the debian/ dir from baz...
[05:15] <BenC> and I do the dpkg-buildpackage -S, right?
[05:15] <fabbione> cd ..
[05:15] <fabbione> dpkg-source -b -i linux-source-2.6.12-2.6.12
[05:15] <fabbione> mkdir temp
[05:15] <fabbione> cd temp
[05:15] <fabbione> dpkg-source -x ../linux-source-2.6.12_2.6.12-8.XX.dsc
[05:15] <fabbione> TADA
[05:16] <fabbione> no more baz crap in the debian dir :)
[05:16] <fabbione> cd linux-source-2.6.12-2.6.12
[05:16] <BenC> ah, so dpkg-source does it
[05:16] <fabbione> dpkg-buildpackage -rfakeroot -S -m'Ben I rock Collins' -k$yourkeyid
[05:16] <fabbione> done ;)
[05:16] <fabbione> upload to jackass.ubuntu.com in ftp.. no need to use any fancy dir
[05:17] <fabbione> BenC: iirc there is a flag that you can pass to dpkg-buildpackage directly...
[05:17] <fabbione> but i can never remember it
[05:17] <fabbione> it's probably -i
[05:17] <BenC> what's the i386 dev machine name?
[05:17] <fabbione> concordia
[05:17] <BenC> yeah, I think it is
[05:17] <BenC> that's it, thanks
[05:17] <fabbione> use linux32 dchroot -c breezy-i386
[05:17] <fabbione> it's an amd64...
[05:17] <BenC> alright, I'll do these patches, test build, and then upload later today
[05:18] <fabbione> so you get both at the price of one ;)
[05:18] <fabbione> BenC: that's ok.. remember to commit everything in baz ;)
[05:18] <BenC> mdz: anyone be around, so 21:00 UTC?
[05:18] <BenC> say*
[05:19] <mdz> BenC: 2100 today, or 2100 tomorrow?
[05:19] <BenC> today
[05:19] <BenC> less than 6 hours
[05:19] <mdz> ok
[05:19] <mdz> I'll be around, certainly
[05:19] <fabbione> BenC: i won't be around at that time....
[05:19] <fabbione> so you will have to do it yourself...
[05:19] <BenC> fabbione: no problem, I think I got all I need
[05:19] <fabbione> but i am sure you can manage fine...
[05:19] <BenC> thanks
[05:19] <mdz> make sure elmo will be around to do the byhands
[05:20] <fabbione> mdz: will you take care of NEW?
[05:20] <fabbione> BenC: also.. ask Kamion (since he is around) to prepare d-i...
[05:20] <fabbione> BenC: if he can...
[05:21] <fabbione> otherwise i can do it for you now...
[05:21] <mdz> fabbione: I can process new binaries, but elmo should do the d-i processing
[05:21] <fabbione> mdz: yeps..
[05:21] <mdz> fabbione: but if we need elmo anyway, he can do all of it at once
[05:21] <mdz> so there is no point in splitting it between the two of us
[05:22] <fabbione> mdz: right :) but if both of you know.. it's better... we are not depended on one that can actually die from one sec to another because of a lightning stroke...
[05:22] <BenC> elmo will be around
[05:22] <fabbione> and still make Colony
[05:23] <fabbione> BenC: i did forward you a mail on how to bump the ABI in D-I
[05:23] <fabbione> did you check with elmo if you your upload privileges have been fixed?
[05:23] <fabbione> s/fixed/granted...
[05:23] <BenC> good question
[05:25] <fabbione> ARGH.. brb.. wife is in high bitching mode today
[05:25] <BenC> elmo claims no knowledge of me being ok'd to do uploads :)
[05:25] <BenC> so I guess that's a no
[05:26] <BenC> fabbione: raise the pimp hand, and then...do as your told like a good husband :)
[05:27] <BenC> he's going to take care of it though
[05:30] <fabbione> re
[05:30] <fabbione> eheheh
[05:39] <fabbione> BenC: ok.. so i am good to go that you will do all of it...
[05:39] <fabbione> if something breaks down horribly, i will be around at 4 UTC :)
[05:39] <fabbione> or on a plane to somewher without IP
[05:40] <BenC> hehe
[05:45] <zul> BenC: then carpet bomb his house
[05:45] <BenC> zul: how do you think I got him to act so quickly?
[05:45] <BenC> elmo's are very furry and flamable
[05:46] <zul> lol
[05:46] <fabbione> ahha
[05:46] <fabbione> zul: you got it???
[05:46] <fabbione> ahhaha
[05:49] <zul> fabbione: came like an hour ago
[05:50] <fabbione> zul: is it nice?
[05:50] <BenC> ok, wont be paying attention to IRC for the next few hours
[05:51] <fabbione> BenC: just answer my question
[05:51] <BenC> which one?
[05:51] <fabbione> if you are going to collect the ABI for all 6 arches, or do i need to take care of SCC
[05:51] <BenC> I'll collect them
[05:51] <fabbione> perfect!
[05:51] <BenC> or attempt to anyway
[05:51] <BenC> Kamion also wants me to do linux-meta
[05:51] <fabbione> BenC: well you have i386/amd64/ppc/ia64 at the DC
[05:52] <fabbione> you miss sparc/hppa
[05:52] <fabbione> BenC: yes. linux-meta and linux-restricted-modules are in the list of pkgs that we discussed on irc...
[05:52] <fabbione> kernel -> linux-restriceted-modules -> linux-meta -> d-i
[05:52] <zul> fabbione: very nice
[05:52] <fabbione> zul: eheh enjoy it :)
[05:52] <fabbione> zul: i am considering to buy a powerbook...
[05:52] <zul> oh i will...now i can code in the basement during the summer
[05:53] <BenC> I need to get a laptop before UBZ
[05:53] <fabbione> BenC: and a passport :)
[05:53] <BenC> donations accepted
[05:54] <fabbione> BenC: talk to mjg59 for one.. on what to get ;)
[05:54] <fabbione> BenC: same here...
[05:54] <zul> i dont need a passport i just need sponsorshipt
[05:55] <BenC> zul: send me your laptop, and consider it done :)
[05:55] <zul> :P
[05:55] <zul> im off for lunch...biab
[05:56] <fabbione> later zul
[06:40] <zul> that is a nice laptop but i have to put it away since i need to work
[08:09] <jbailey> BenC: Did you see the conversation in #ubuntu-devel yesterday about needing to find a way of asking the currently running kernel to regenerate its initramfs?
[08:10] <fabbione> jbailey: you will need to explain me tomorrow why you want something like that...
[08:11] <fabbione> i gotta run now..
[08:11] <jbailey> fabbione: By edit of my boss. =)
[08:11] <jbailey> edict, rather.
[08:11] <fabbione> yes ok.. to achieve what?
[08:12] <jbailey> So that new usplashes can update the initramfs.
[08:13] <fabbione> they can do it, even without asking the kernel to regenerate the initramfs
[08:13] <fabbione> they just have to call the correct mkinitramfs
[08:13] <jbailey> Right, except that they have no good way of knowing who's running, who's installed, etc.
[08:13] <jbailey> I think the right solution is initramfs bootloader magic, perhaps.
[08:14] <fabbione> i wonder why the kernel-team has been never asked to look at the usplash implementation...
[08:14] <fabbione> u <- userland :)
[08:14] <fabbione> fix it there :P
[08:15] <fabbione> now i don't want to sound silly, but people have been pushing crack into the kernel..
[08:15] <jbailey> Bzzt.  initramfs belongs to the kernel, you produce it so nothing else should touch it.
[08:15] <fabbione> without the kernel team being exactly informed
[08:15] <fabbione> jbailey: yes.. usplash is touching initramfs
[08:15] <jbailey> that's part of why I like the idea of bootloader assembly.
[08:15] <fabbione> and now the kernel team needs to find a solution for what i believe is a design error..
[08:15] <jbailey> That means that the kernel wouldn't have to do that assembly - takes responsibility for it away from you.
[08:16] <fabbione> anyway.. i think i am too tired to appreciate usplash, given that it doesn't even work on my machines :)
[08:16] <jbailey> Eh, really?
[08:16] <fabbione> better to go and get some food
[08:16] <jbailey> How does it fail?
[08:16] <jbailey> Enjoy, Fabio.
[08:16] <fabbione> jbailey: i don't get anything displayed
[08:16] <jbailey> And hello from Angie. =)
[08:17] <fabbione> give a hug to her :)
[08:17] <fabbione> cya
[08:40] <fabbione> jbailey: there is no way the kernel can address that problem, if not providing an extra info to usplash like: touch /lib/kernel/2.6.12-8.isanofficialubuntukernel
[08:40] <fabbione> at that point usplash can parse that dir and update the specific initramfs
[08:41] <fabbione> dir or configfile or whatever.. but i guess you get the idea
[08:41] <zul> jbailey: have you tried out those xen packages from ubuntu-devel?
[08:43] <jbailey> zul: I've not.
[08:44] <jbailey> fabbione: The kernel could take an md5sum at generation time, and regenerate the version of the initrd pointed to by /initrd.img iff the md5sum still matches.
[08:45] <fabbione> jbailey: i don't see how that can help.. the kernel is not supposed to touch stuff...
[08:45] <fabbione> and it's truely pointless to ask the kernel to do a job that userland can and in a much simpler way
[08:45] <jbailey> Eh?
[08:45] <jbailey> This is all packaging. =)
[08:46] <fabbione> jbailey: ok.. as i understand it:
[08:46] <fabbione> - kernel is installed
[08:46] <fabbione> - usplash is installed
[08:46] <fabbione> now..
[08:46] <fabbione> initramfs needs to be updated by usplash...
[08:46] <fabbione> right?
[08:46] <jbailey> Well, it needs updating because of a usplash update.
[08:46] <fabbione> because each time usplash is update, the user need to reconfigure the kernel as it is now
[08:47] <fabbione> perfect.. than we agree
[08:47] <jbailey> Right, except that the kernel owns the initramfs
[08:47] <jbailey> Or rather, the kernel package owns the initramfs. =)
[08:47] <fabbione> wrong...
[08:47] <fabbione> the kernel package generates the initramfs
[08:47] <fabbione> it doesn't own it
[08:47] <jbailey> Sure it does.  The 'touched it last' rule.
[08:47] <fabbione> it creates it :)
[08:47] <jbailey> What I think could happen is that as soon as you create it, you generate an md5sum hash of it.
[08:48] <fabbione> yes, but what is the problem you want to address?
[08:48] <jbailey> The problem that you don't know if the user has touched it in the meantime.
[08:48] <fabbione> you want to avoid usplash updating an initramfs
[08:49] <jbailey> The usplash has no way of knowing that information.  It doesn't touch the initramfs at all atm.
[08:49] <fabbione> hmmm jeee....
[08:50] <fabbione> well what kind of changes do we expect from user to initramfs?
[08:50] <fabbione> via /etc/mkinitramfs ?
[08:50] <jbailey> Dude, who knows? =)
[08:50] <jbailey> These are *users* we're talking about. =)
[08:50] <fabbione> yeah ok.. we will need to change this rule for breezy+1
[08:51] <jbailey> But this mechanism could be used to update lvm, evms, or *mumble*.
[08:51] <fabbione> we can't keep working around all possible combinations.. especially in front of a 3/5 years support
[08:51] <jbailey> Right.  I'd love to see it done as boot time assembly.
[08:51] <fabbione> you mean bulding the initramfs at boot time?
[08:52] <fabbione> i have a better idea :)
[08:52] <jbailey> Right, by the bootloaders.
[08:52] <jbailey> I have this vague idea that we could do something like:
[08:53] <jbailey> Have grub assemble the initramfs at boottime based on whatever modules seem obvious.
[08:53] <fabbione> jbailey: grub for i386.. what about the others?
[08:53] <jbailey> As part os S99make_it_possible_to_recover_my_system, then generate a static version of that initramfs.
[08:53] <jbailey> That way there's a recovery option that should be guaranteed to work.
[08:53] <fabbione> jbailey: i have a better idea...
[08:53] <jbailey> Sure. =)
[08:53] <fabbione> let's hack on http://fabrice.bellard.free.fr/tcc/tccboot.html
[08:54] <fabbione> so that it includes an lspci
[08:54] <fabbione> we will build the kernel from sources at boottime
[08:54] <fabbione> with the proper hw configuration
[08:54] <fabbione> we will build everything inline
[08:54] <fabbione> and all the devices that we know can't be inside a pc.. as modules
[08:55] <fabbione> :)
[09:06] <BenC> we should build the whole kernel as just objects and then do linking on install :)
[09:07] <fabbione> BenC: that's too easy :)
[09:07] <fabbione> it's more fun to build it from sources
[09:07] <BenC> what would be nice, actually, is if the device driver objects were the same whether they were module targets or built-in targets
[09:07] <BenC> just the linking would be different
[09:08] <fabbione> hmmmm
[09:08] <fabbione> you can probably achieve that...
[09:08] <BenC> yeah, would require a change in the module abi I'm sure
[09:08] <fabbione> nope...
[09:09] <fabbione> it's easier than you think...
[09:09] <fabbione> afaik the stuff you build inline are all the objects compiles at STAGE1 of the kernel
[09:09] <fabbione> modules get STAGE1 -> STAGE2
[09:10] <BenC> yeah, but module.h changes a lot of stuff (init/remove) depending on -DMODULE
[09:10] <fabbione> in stage2 from the object created in STAGE1, the .mod.c gets created, builded and linked...
[09:10] <BenC> I think the driver object is actually a bit different
[09:10] <BenC> it could be
[09:10] <BenC> they may have thought of this when they did the 2.6 kbuild system
[09:10] <fabbione> if you look in a mod.c that's created by that file we fixed for sparc...
[09:11] <fabbione> you see exactly the module interface...
[09:11] <fabbione> it's worth a try.. also because it's pretty easy
[09:11] <BenC> well if the entire module abi is in the .mod.c, then there you have it
[09:12] <fabbione> BenC: the only thing i am afraid is code that does:
[09:12] <BenC> someone just needs to write the tools and logic to do an install time linking of a static kernel, with modules just for things like hotpluggable devices
[09:12] <fabbione> ifdef CONFIG_FOO=m
[09:12] <fabbione> endif
[09:12] <fabbione> or something like that
[09:12] <BenC> yeah, I think ieee1394 may have (or maybe it's gone now) things like that
[09:13] <BenC> there's also the ones that have hardcoded "#ifdef MODULE"
[09:13] <fabbione> yeah i have seen a bunch of those
[09:14] <BenC> I think the scsi layer is littered with them
[09:14] <BenC> which defeats most of the advantage of it
[09:14] <fabbione> yeps
[09:48] <zul> later